Full text of "Bulletin of the Museum of Comparative Zoology at Harvard College"

See other formats

t\vs

HARVARD UNIVERSITY

Library of the

Museum of

Comparative Zoology

us ISSN 0027.4100

bulletin OF THE

Museum of

Comparative
Zoology

A Review of the North American
Fossil Amiid Fishes

JOHN R. BORESKE, JR.

HARVARD UNIVERSITY

CAMBRIDGE, AAASSACHUSEHS, U.S.A.

VOLUME 146, NUMBER 1
18 JANUARY 1974

PUBLICATIONS ISSUED

OR DISTRIBUTED BY THE

MUSEUM OF COMPARATIVE ZOOLOGY

HARVARD UNIVERSITY

Bulletin 1863-

Breviora 1952-

Memoibs 1864-1938

JoHNSONiA, Department of Molliisks, 1941-

OccAsioNAL Papers on Mollusks, 1945-

Other Publications.

Bigelow, H. B., and W. C. Schroeder, 1953. Fishes of tlie Gulf of Maine.
Reprint.

Bnies, C. T., A. L. Melander, and F. M. Carpenter, 1954. Classification of
Insects.

Creighton, W. S., 1950. The Ants of North America. Reprint.

Lyman, C. P., and A. R. Dawe (eds.), 1960. Symposium on Natural Mam-
malian Hibernation.

Peters'. Check-list of Birds of the World, vols. 2-7, 9, 10, 12-15.

Sprinkle, J., 1973. Morphology and Evolution of Blastozoan Echinoderms.

Turner, R. D., 1966. A Survey and Illustrated Catalogue of the Teredinidae
(Mollusca: Bivalvia).

Whittington, H. B., and W. D. I. Rolfe (eds.), 1963. Phylogeny and Evolu-
tion of Crustacea.

Proceedings of the New England Zoological Club 1899-1948. (Complete
sets only.)

Publications of the Boston Society of Natural Histor>'.

Authors preparing manuscripts for the Bulletin of the Museum of Comparative
Zoology or Breviora should send for the current Information and Instruction Sheet,
available from Editor, Publications OflBce, Museum of Comparative Zoology,
Harvaid University, Cambridge, Massachusetts 02138, U.S.A.

(g) The President and Fellow* of Harvard Colleee 1974

A REVIEW OF THE NORTH AMERICAN
FOSSIL AMIID FISHES

JOHN R. BORESKE, JR.i

CONTENTS

Abstract 1

Introduction 2

Acknowledgments 2

Abbreviations 3

A7?Jia calva 3

Nomenclature 3

Ecology 4

Geographic Distribution 4

Pleistocene Occurrences 4

Diagnosis 5

Morphometries 5

Methods 6

General Proportions and Growth 8

Comparisons with Fossil Forms 10

Discussion 17

Meristics 18

Supravertebral Scale Rows 18

Branchiostegal Rays 20

Fin Rays 20

Vertebral Elements 25

Vertebral Column of Amia calva 28

Vertebral Features 28

Vertebral Dimensions 33

Valid North American Fossil Genera and

Species 37

Amia fragosa 37

Amia uintaensis 47

Amia cf. uintaensis 64

Amia scutata 66

Amia cf. scutata 70

Amia cf. calva 72

Amiidae incertae sedis 72

Specimens Removed from the Amiidae 74

Summary and Conclusions 75

Literature Cited 81

Plates 84

Abstract. North American amiid fishes range

from Cretaceous ( Albian ) to Recent. Amiids are

1 Museum of Comparative Zoology, Harvard
University, Cambridge, Massachusetts 02138

Bull. Mus. Comp. Zool, 146(1): 1-87, January, 1974

common fossils in Late Cretaceous and Tertiary
freshwater deposits and apparently occupied a
habitat much like that of the Recent species Amia
calva. Morphometric, meristic, and cranial char-
acters of articulated specimens from the Fort
Union Fonnation (Paleocene), Green River For-
mation (Eocene), Florissant Fonnation (Oligo-
cene). Pawnee Creek Formation (Miocene), and
a Recent A. calva sample from Wisconsin have
been used here in an attempt to revise the taxon-
omy and evolutionary history of the group.

Whereas seven genera and twenty-three species
of fossil amiids have been described on the basis
of disarticulated, often isolated elements, only
three taxa have heen described from complete or
partially complete material. Amia fragosa (Late
Cretaceous to Middle Eocene), A. uintaensis (Pal-
eocene to Early Oligocene), A. scutata (Early to
Middle Oligocene), and A. calva (Middle Plio-
cene to Recent) are here considered the only
valid taxa. Amiid remains are first known in the
North American fossil record from the Early Cre-
taceous (Albian) Paluxy Formation of Texas.
This disarticulated material shows resemblances
both to A»ii« and to the Late Mesozoic European
genera Uroclcs and Arniopsis. Paramiatus gurleyi
(Romer and Fryxell, 1928) from the Green River
Formation of Wyoming is a synonym of A. frag-
osa. Tlie differences between Amia and the large
Early Cenozoic form Protamia are insufficient for
recognition of Protamia as a genus distinct from
Amia. The Eocene and Oligocene forms Protamia
media, Pappichthys medius, P. plicatus, P. scler-
ops, P. laevis, P. symphysis, P. corsonii, Amia
whiteavesiana, and A. macrospondyla are s\monyins
of A. uintaensis; they were based on undiagnostic
cranial and vertebral characters. Morphometric
and meristic similarities indicate that little evi-
dence exists for maintaining separate Oligocene
species A7nia scutata and A. dictyocephala. Amia
exilis is a synonym of Amia scutata; it was based
on undiagnostic vertebral characters. A. scutata
is morphometrically distinguishable from A. calva
only on the basis of a slightly larger head/stan-

Bulletin Museum of Comparative Zoology, Vol. 146, No. 1

dard-length ratio. The Eocene ta.xa Amia de-
pressus, A. newherrianus, A. gracilis, and Htjpamia
elegans are nomina diibia.

Comparison of the fossil forms with the Recent
Amia calva suggests the following ta.xonomic and
possible phylogenetic relationships: ( 1 ) Amia frag-
osa survived until the Middle or Late Eocene,
with no phylogenetic affinities with the modern
form; (2) Amia iiintaensis appears to be closer
than Amia fragosa to the ancestry of Amia calva,
which evolved through an intermediate fonn such
as Amia scutata; (3) establishment of the Recent
species Amia calva had begun at least by the be-
ginning of the Pliocene; and (4) diere are simi-
larities in the Paleocene and Eocene amiid fossil
record of North America and Europe.

INTRODUCTION

Aviia is a genus of freshwater fishes that
includes one of two extant representatives
of the holostean level of organization. It
includes a number of species of which only
Amia calva exists today; other forms of
Amia are found in the fossil record, and
extend from the Late Cretaceous to ap-
proximately the Middle Pliocene. This
study is an attempt to determine the taxon-
omic and phylogenetic relationships among
the various species of Amia. It is established
on osteology as well as on morphometric
and meristic data from both Recent and
fossil forms. This data is used to compare
the available features of the fossil forms
with Recent Amia calva and to detennine
the validity of previous descriptions based
on various osteological, morphometric, or
meristic character-states.

Until recently, a major difficulty in inter-
preting the taxonomy of fossil amiids has
been the paucity of articulated specimens.
Five genera and twenty-one species of fossil
forms have been described from disarticu-
lated, often isolated, elements (Table 19);
only two taxa have been described from
articulated specimens: Paramiatus ii^iirleyi
(Romer and Fryxell, 1928) and Amia
scutata (Osborn et al., 1875). Recent works
by Estes (1964) and Estes and Berberian
( 1969 ) , based on disarticulated elements
from the Late Cretaceous Lance and Hell
Creek formations, are the only published
studies of Amia fragosa, although O'Brien
(1969) completed an M.A. thesis on the

osteology of A. frap.osa, describing articu-
lated specimens from the Late Cretaceous
Edmonton Formation of Alberta.

Much more articulated material is now
available and provides more detailed in-
formation on the cranial and postcranial
anatomy of amiids. These specimens have
been useful in this revision of the taxonomy
as well as in the determination of possible
relationships to European and Asian forms.

In an attempt to understand the evolution
and interrelationships of the fossil and Re-
cent amiids, a growth-series study has been
made on a Recent A. calva sample from
Wisconsin, and is compared moi"phometri-
cally and meristically with the fossil forms.
A great number of fossil specimens, includ-
ing the holotypes and paratypes of all North
American amiid species, have been exam-
ined. Several European taxa have been
studied at the British Museum ( Natural
History), London; Museum National d'His-
toire Naturelle, Paris; and the Institut Royal
des Sciences de Belgique, Brussels.

ACKNOWLEDGMENTS

I am especially grateful to Professor
Richard Estes (University of California at
San Diego) for his advice and criticism in
the preparation of this manuscript. Cecile
Janot- Poplin and Sylvie Wenz ( Museum
National d'Histoire Naturelle, Paris), and
Karel Liem ( Museum of Comparative Zo-
ology) read the manuscript and offered
criticisms that substantially improved the
text.

Additional thanks are due to Donald
Baird ( Princeton University ) , Henry Booke
and Bany Cameron (Boston University),
William J. Hlavin (Cleveland Museum of
Natural History), Farish A. Jenkins, Jr.
(Museum of Comparative Zoology),
Charles Meehan ( Chamberlayne College),
Robert R. Miller (University of Michigan),
David Pariis ( New Jersey State Muse-
um), Colin Patterson (British Museum of
Natural History), Clayton Ray (National
Museum of Natural History), Bobb Schaef-
fer (American Musevun of Natural History),
Hans-Peter Schultze ( Geologisch-Paleon-

Fossil Amiids • Borcske

tologisches Institiit der Gcorg-Aiigust-Uni-
vcrsitiit, Gottingen), Keith Thomson (Yale
University), and Hainer Zangerl (Field
Museum of Natural History) for their help-
ful suggestions. I am also grateful to Leslie
Whone for preparation of tables, and to Siri
Falck-Pedersen Boreske, Laszlo Meszoly,
and Charles Chamberlain for illustrations.
This study was supported by grants from
Sigma Xi, Marsh Fund, and the Albion
Foundation.

ABBREVIATIONS

AMNH — American Museum of Natural
History, New York, New York.
ANSl^ — Academy of Natural Sciences of
Philadelphia, Philadelphia, Pennsylvania.
BMNH — British Museum (Natural
History), London, England.
CM — Carnegie Museum, Pittsburgh,
Pennsylvania.

F: AM — Frick- American Museum
Collection, New York, New York.
FHKSM— Fort Hays Kansas State Museum,
Hays, Kansas.

FMNH — Field Museum of Natural History,
Chicago, Illinois.

FSM — Florida State Museum, Gainesville,
Florida.

MCZ — Museum of Comparative Zoology,
Harvard University, Cambridge,
Massachusetts.

MNHN — Museum National d'Histoire
Naturelle, Paris, France.
NMC — National Museum of Canada,
Ottawa, Canada.

PU — Museum of Natural History,
Princeton University, Princeton, New
Jersey.

ROM — Royal Ontario Museum, Toronto,
Canada.

SMUSMP— Shuler Museum of Paleontol-
ogy, Southern Methodist University, Dallas,
Texas.

UA — University of Alberta Museum,
Edmonton, Canada.
UCMP — Museum of Paleontology,
University of California, Berkeley,
California.

UMM — West Texas Museum, University of

Texas, El Paso, Texas.

UMMP — l^ni\ersity of Michigan Museum

ot Paleontology, Ann Arbor. Michigan.

UMMZ — University of Michigan Museum

of Zoology, Ann Arbor, Michigan.

USNM— National Museum of Natmal

History, Wa.shington, D.C.

YPM — I'eabody Museum of Natural

History. Yale University, New Haven,

Connecticut.

AMI A CALVA LINNAEUS, 1766

Amid calvii is the only extant species of
the family Amiidae. It is a predaceous fish
that exclusively inhabits fresh waters of
the eastern LTnited States. Except for the
gar, Lepisosteus, Amia calva is the onl\'
other living representative of the holo.stean
fishes. Its common name, "bowfin," refers
to the long dorsal fin that arches in a bow
over most of the length of the fish's back.
Amia calva has previously been known as
the dogfish, marshfish, mudfish, grindle, or
lawyer.

The osteology of Amia calva has been
extensively described and discussed by
Schufeldt (18S5), Bridge (1S77), Allis
(1889, 1897), and Goodrich (1930). The
following discussion is limited only to the
nomenclatural problems, ecology, geo-
graphic distribution, and character-states
of Amia calva that are relevant to study of
the fossil forms.

Nomenclature

Jordan and Evermann (1896) noted that
although Linnaeus (1766) had applied the
binomial name Amia calva to the genus,
Gronow (1763) had earlier used Amia as
a nonbinomial name for fishes presently
classified as Apoiion Lacepede. They fur-
ther suggested that should Gronow's earlier
ipplication of the name be given prece-
dence and transferred to Apof^on, then
Ainiatus Rafinesque (1815) should replace
Amia Linnaeus. Jordon (1906) stated that
this transfer of names was a necessary com-
pliance with the rules of nomenclature, but
later (1919), although citing Opinion 20

Bulletin Museum of Comparative Zoology, Vol. 146, No. 1

( 1910 ) of the International Commission on
Zoological Nomenclature which favored
Gronovv's priority, Jordan found tlie trans-
fer of names inconvenient, for most autliors
had rejected Gronow's names as pre-
Linnaean. In 1925, Jordan recommended to
the Commission that certain names of
Gronow supported by Opinion 20 be re-
jected in favor of the more accepted
Linnaean terminology. The Commission's
Opinion 89 ( 1925 ) resolved ( among others )
the nomenclatural problem of Amia, by con-
curring with Jordan's recommendation that
". . . Amia Gronow be set aside in favor of
Amia Linnaeus, even if other names of
Gronow are allowed." Rafinesque's name
Amiatus is then a junior synonym of Aynia
Linnaeus.

Some later workers seem to have been
unaware of Opinion 89. Thus Hussakof
( 1932 ) accepted the validity of the transfer
of the name A7nia Gronow to the percoid
teleost Apo^on. Romer and Fryxell (1928)
named their fossil amiid from the Eocene
Green River Formation Paramiatus instead
of Paramia, and Whitley ( 1954 ) changed
the name of Lehman's (1951) fossil amiid
from the Eocene of Spitzbergen from Pseu-
damia to Pseudamiatus. The latter is invalid
as Pseudamia was a valid name in itself and
Pseudamiatus is its junior synonym regard-
less of the Amia- Amiatus controversy.

Ecology

Aside from notes regarding breeding, diet,
and zoogeographical occurrences, little has
been written in the past 50 years about the
ecology of Amia calva. Dean (1898) and
Reighard ( 1903 ) have made the only ex-
tensive published investigations of the
habits and habitat of the fish. A thorough
study of the biology of A. calva throughout
its range is long overdue.

Geographic Distribution

The distributional map of Amia calva
( Fig. 1 ) is based on information drawn
from Hubbs and Lagler (1967), and Blair
et al. ( 1968 ) , and from examinations of
unpublished records at the Ohio State Uni-

versity Museum of Zoology, Museum of
Comparative Zoology, and the University
of Michigan Museum of Zoology. The dis-
tribution limit is a flexible boundary allow-
ing for seasonal occurrences and other
natural variations. The known northern
limit of A. calva extends from the Missis-
sippi drainage system in Minnesota south of
Duluth, eastward through Lake Nipissing
and the Ottawa River to the St. Lawrence-
Champlain basin ( encompassing Quebec as
far north as Quebec City, and Vermont).
A. calva is distributed throughout the Great
Lakes region, but is not found in the Lake
Superior drainage basin, except in its outlet,
the St. Mary's River. Southward, it has
been recorded from the Hudson River to
western Connecticut ( recorded as the result
of introduction; Hubbs and Lagler, 1967);
Harrisburg, Pennsylvania, to the Susque-
hanna River; and along the Atlantic slope
to the Carolinas and Florida. Westward,
A. calva occurs along the Gulf Coast to
southern Texas as far south as Brownsville,
and northward, through eastern Texas,
southeastern Oklahoma, northwestern Ar-
kansas, eastern Missouri, and approximately
50 miles west of the Mississippi River to
Brainard, Minnesota.

Pleistocene Occurrences

Amia calva has been reported from only
three Pleistocene localities: (1) Chicago,
Illinois, (2) Vero Beach, Florida, and (3)
Itchtucknee River deposits, Columbia
County, Florida (MCZ 9524, 9529, 9542).
Hay (1911: 552) reported "Professor Frank
Baker (Chicago Academy of Science) has
shown me a considerable part of the skele-
ton and scales of a bowfin which he found
in the Pleistocene clay near Chicago." A
thorough search of the Chicago Academy
of Science collecrions failed to produce this
specimen. Hay (1917, 1923) listed Amia
calva among the fossil vertebrate remains
found in the Pleistocene sands at Vero
Beach, Florida. Swift ( 1968 ) , in his review
of fossil fishes of Florida, figured Hay's
Aiiiia specimens (left dentary and a gular
plate; FSM collections) and concluded that

Fossil Amiids • Borcskc

/ /«°«'m;;; L.._ ; ,

/■ \ r — ■'• ..—■•J^

! \ : NORTH DAKOTA ; ''w

, L . I \ I \minnesota

'""^GO- ^^ > • to ' ■»

^^^^^^■>.-,. / /o* ^1 !

I liJUH—^ m o fe'" oo ^— i

/ / 'l-^-.- i '^

V / ° /'^OlORAdS"— ^ — 1 •(__

\ / /■ j. VMlSbl) I

\ /' ; ; KANSAS ' ' "^

\ / J ! \

\ L .' ^ I o ,

\ .j^^'^oZ. f. ! I

\ I ; "^w MEx-6 T-«- -I

V; • Q *• .OKLAHOMA L

T / -TEXAS ^ 7;

> /• I ^

.. ^ / i '.

Fossil forms of "••v^ ' ! .

• A.fragosa \ s'"~-.

° A.uintaensis ''■^ '\

a A.c\.uintaensis \

' A. scuta to *•

» /J.cf. scuta fa \

■ Amiidae incertae sedis

X Pleistocene location of A.calva

Fig. 1. Distribution of km\a calva. Fossil occurrences of Amia spp. ore represented by symbols explained in the
legend.

A. calva was probably very common in the
Pleistocene fresh waters of the United
States. The pancity of Pleistocene material
does not necessarily mean the fish was not
common in the Pleistocene, but does indi-
cate that A77ua remains have not been
extensively collected or identified in exist-
iii'j; museum Pleistocene collections.

Diagnosis

Vertebral meristics similar to A. sctitata,
but head/standard-length proportion (0.271
mean) is smaller than in the fossil forms.
Extrascapular strap-shaped and relatively
wide at midlitie, as in A. scutata, but pos-
terior edge is curved so that it is proximally
convex, then concave toward the distal
corner, which results in a posterolateral
projection. Pterotic borders frontal pos-
teriorly rather than laterally; anterior end
is as wide as posterior end. Orbital excava-
tion is shallower than in other species, with

a mean depth-to-length ratio of 0.100. In-
fraorbital 4 is smaller than infraorbital 5,
less robust than in fossil Amia. Preopercu-
lum as wide dorsally as ventrally. Symphy-
seal incurx'ing of dentary relatively less than
ill A. fraii^osa, ])ut greater than in A. scutata
and A. uintacnsis: little or no overlapping
of dorsal coronoid articulation surface on
ventral siuface of ramus; deep Meckelian
groov^e. Ventropostcrior process of cleith-
rum less sculptured than in other species of
Amia. V^omerine teeth shaip, conical, num-
bering between 15-27, more anteriorly
placed than in A. uintacnsis or A. fraii.osa.
Bones less ossified than in fossil Amia.
Greatest known standard-length 650 mm.

MORPHOMETRICS

Comparison of morphometric and meris-
tic data of Recent and fossil Amia has
facilitated an e\'aluation of the taxonomy as
well as clarified anatomical trends. Many

6 Bulletin Museum of Comparative Zoology, Vol. 146, No. 1

generic or specific character-states for Amia
''dicfyocepJmla,'^ Amia scutata, "Paramiatus
gurleiji,^' and Amia fragosa have been pre-
viously estabhslied on osteological data
based primarily on gross anatomical propor-
tions (head/standard-lcngth ratio and posi-
tions of insertion of pelvic and anal fins/
standard-length ratios) and skull propor-
tions (parietal/frontal and operculum
width/length ratios). Meristic character-
states have also been used for A. "dictijo-
cephala" and A. scutata.

Altliough an age-growth analysis on Amia
calva was done by Cartier and Magnin
(1967), no moiphometric investigation of a
growth-series of Recent A. colva has yet
been completed or used for comparison
with fossil forms. Estes and Berberian
(1969: 10) suggest that knowledge of the
growth-series of A. calva would be of con-
siderable importance in tracing the ancestry
of the modern species.

Hammett and Hammett (1939) made a
moi-phometric study of the Recent Lepisos-
teus platijrhinciis, taking length dimensions
of a sample of live fish from Florida. Since
Lepisosteus, like Amia, is one of the two
extant holosteans, their analysis is poten-
tially useful in providing information on the
ancient species. However, they did not
actually compare the live material or data
with any fossil material.

According to Imbrie (1956), Simpson
et al. (1960), and Gould (1966), growth
studies offer excellent means with which to
clarify evolutionary and taxonomic prob-
lems in the fossil record. An interesting
model utilizing morphometric data for
synonymy of fossil forms was made by
Thomson and Hahn ( 1968 ) on the growth-
series patterns of Devonian rhipidistian
fishes, in which they showed that Thiirsius
clappi was actually a juvenile form of
Eusthenopteron foordi. In studying fossil
material, as Thomson and Hahn (196S:
201) indicate, there is a problem in deter-
mining the age, sexual matmity, and envi-
ronmental regime of the animal. Also, of
course, it is necessary to have sufficient fos-

sil material with which to erect an adequate
growth-series.

This present analysis is undertaken (1)
to determine w hether skull and axial skele-
tal proportions of amiids are isometric or
allometric with increasing size, (2) to
establish the variation in meristic charac-
ters of Recent A. calva, and ( 3 ) to compare
moiphometrics and meristics of Recent A.
calva with those of the fossil forms. This
study utilizes a small sample of 18 Recent
A. calva specimens from the St. Croix River,
Wisconsin. Measurements were taken from
a growth series that includes the size range
of most of the articulated fossil forms. The
largest A. calva specimen, from St. Joseph
County, Michigan (UMMZ 197683), was
analyzed to see whether the large specimen
would agree with the anatomical propor-
tions and meristic characters of the Wiscon-
sin specimens. Three smaller specimens
from Pewaukee, Wisconsin (MCZ 8970),
were also included. The fossil sample con-
tains six complete and ten partially com-
plete amiid specimens ranging in age from
Late Cretaceous to Late Miocene which,
although moiphometrically similar in vary-
ing degree, are too few to warrant conclu-
sions in themselves.

Methods

Measurements chosen for this study ( Fig.
2) are those of Hubbs and Lagler (1967:
20). In fossil forms, because of the lack of
preservation of internal soft anatomy as well
as the impossibility of determining their
interbreeding potential, these particular
measurements necessarily assume an in-
creased taxonomic significance, since they
often provide the only viable parameters
with which to designate genera and species.
Measurements taken on A. calva are limited
to those also represented in the fossil speci-
mens. Each of the A. calva measured was
X-rayed, except for three small specimens,
which were cleared and stained. The range
of error for aj] measmements taken on Re-
cent and fossil material is ±0.04 mm. The
range of error for the ratios is ±0.08 mm;

Fossil Aaiiids • Boreske

Fig. 2. Index to the measurements used, superimposed upon an outline drawing of Amia.

Key for body measurements:

TL = Total-Length

SL =z Standard-Length

H zi: Head-Length

C ^ Caudal-Length

Pf =3 Insertion of Pelvic Fin

P = Insertion of Anal Fin

HL =: Standard-Length minus Head-Length

ML := Standard-Length minus Mandible-Length

Key for abbreviations of cranial elements used in morphometric study:

Mandible

Gular

Infraorbital

Frontal

Par

Parietal

zi:

Operculum

Table 1. Length dimensions of 22 specimens of Amia calva L.: 21 from Wisconsin

(MCZ 8970'), 1 from Michigan ( UMMZ 197683)'*

Measurements in mm

Specimen

Class Range

Code

Total Length

No.

80.0

70.5

57.0

22.0

48.5

32.5

12.5

10.5

95.0-105.0

100.0

85.0

70.0

25.0

60.0

39.5

16.0

15.0

207.0-212.0

210.0

175.0

145.0

50.5

124.5

80.9

35.0

227.9-232.0

230.0

193.0

161.0

54.6

138.4

88.5

35.5

36.0

241.0

199.0

165.9

56.8

142.2

93.5

38.0

42.0

291.0

237.0

197.0

64.0

173.0

115.0

52.5

54.0

310.0

248.0

207.0

68.5

179.5

112.0

46.0

62.0

339.0

274.0

230.0

73.0

202.0

125.0

51.0

64.0

385.0

313.0

259.0

82.0

231.0

142.0

52.0

72.0

433.0

4.33.0

349.0

293.0

91.0

258.0

170.0

71.0

84.0

475.0

.399.0

335.0

103.0

296.0

181.0

82.0

76.0

507.0

423.0

359.5

109.0

317.0

192.0

93.0

81.0

13»»

756.0

648.0

545.3

164.0

480.0

299.0

138.0

102.0

See Figure 2 for abbreviations.

8 Bulletin Museum of Comparative Zoology, Vol. 146, No. 1

this margin of error is graphically inconse-
quential in this study. Specimens of A. calva
whose total length was between 207 mm
and 507 mm were selected because this
range of A. calva would provide the best
information for comparison with the fossil
species. Twenty-two specimens of A. calva
were measured (Table 1). Eighteen of
these are from the St. Croix River, Wiscon-
sin. These 18 specimens of A. calva fall
into ten categories arranged here by ap-
proximately 20-30-mm class range incre-
ments in total-length. Although these
categories represent arbitrary rather than
biological growth stages, they provide
suflBcient information on the morphologic
size changes of A. calva. Three smaller
specimens (MCZ 8970, also from Wiscon-
sin) witli a size range of 80-105 mm
total length (TL) were included to de-
termine whether they would follow the
predicted allometric effect on the growth-
series Hne, since, as Thomson and Hahn
(1968: 205) note, it is a common feature
for the early stages of juvenile animals to
have heads proportionately larger than the
bodies. Hay (1895) notes that an 80-mm
A. calva is beyond the embryonic stage and
is an early juvenile with most of its bones
at least partially ossified. The 80-mm speci-
men has a proportionately larger head to
standard-length ratio than the other mem-
bers of the growth-series (Table 3). Al-
though this ratio decreases slightly with
increasing size, the head /standard-length
ratio of 0.312 for the 80-mm specimen does
not deviate far from the growth-series line
(Figs. 3-4).

The largest specimen (UMMZ 197683)
was used as a size limit for the other end
of the growth-series continuum. It may be
assumed that this fish had already reached
the size or point of maturity at which fish
normally begin to decrease their rate of
growth. This specimen still retains the
morphological proportions of the smaller
specimens (Figs. 3-4) and, like them, falls
remarkably close to the constant relative
size-growth lines of the various proportions.
Although from Michigan, this specimen
does not appear to deviate from the growth-
series line established by the Wisconsin
specimens of A. calva. The Michigan speci-
men of A. calva, since it agrees with the
growth-series continuum established by the
Wisconsin specimens, is helpful in extend-
ing comparison to the larger fossil amiids:
"Paratniatus gurleiji" (FMNH 2201), Amia
fra^osa (MCZ 5341), and Amia uintaensis
(PU 13865), which are outside the size
range of the Wisconsin sample.

General Proportions and Growth

Allometric growth, according to Gould
(1966: 595), describes geometrically pro-
gressive change in shape or proportions
with size, and is generally represented by a
curvilinear line or, in certain cases, by a
straight line in which the Y-intercept is
significantly different from 0.

For the Amia calva growth series dis-
cussed here, the ordered pairs correspond-
ing to the proportions in each series have
been plotted on a graph, as well as the
straight line corresponding to the equation
y = a + bx (of the best fit computed ac-

Table 2.

Length

DrMENSIONS

; OF 6 ARTICULATED FOSSIL

AMIIDS

Measurements in

A. scutata PU 10172

'■«yo4.o

339.0

276.5

106.0

233.0

159.0

73.0

e«t65.0

A. scutata UMMP V-57431

—

388.0

313.8

121.0

267.0

183.0

83.0

A. kehreri BMNH P33480

249.0

191.0

160.8

59.2

131.8

89.0

38.5

58.0

"Paramiatus gurleiji"

FMNH 2201

702.0

510.0

430.0

157.0

353.0

<'«f265.0

78.0

192.0

A. fragosa MCZ 5341

575.0

455.0

383.0

142.0

313.0

210.0

75.0

115.0

A. uintaensis FV 13865

848.0

664.0

—

214.0

450.0

288.0

116.0

160.0

See Figure 2 for abbreviations.

Fossil Amhds • Boreske 9

320-

280-

240-

200-

160-

120-

80-

40-

—I 1 I I I

120 200 280

STANDARD

I I I

360
LENGTH mm

680

440

520

600

Fig. 3. Relative growth-lines of head-length (H), pelvic fin insertion (Pf), and ana! fin insertion (P) plotted arith-
metically against standard-length, for 18 specimens of Recent Ami'o calva (A = MCZ 8970 and H = UMMZ
197683 are included for comparison).

cording to the method of least squares); whicli nearly passes through the origin of

the results of these calculations appear in the graph. The coefficient of correlation is

Figures 3-4. Practically all the ratios in almost equal to 1.0 in each case, an indica-

Figure 3 fall onto straight lines, each of tion that the computed straight line provides

10 Bulletin Museum of Comparative Zoology, Vol. 146, No. 1

Table 3. Comparisox of length proportions in 22 specimens of Amio calva with fossil amiids

Specimen

Code

H/SL

Pf/SL

P/SL

0.312

0.461

0.177

0.294

0.464

0.188

0.289

0.462

0.200

0.283

0.459

0.184

0.285

0.470

0.191

0.270

0.485

0.222

0.276

0.452

0.185

0.266

0.456

0.186

0.262

0.455

0.198

0.261

0.487

0.203

0.258

0.454

0.206

0.258

0.454

0.220

0.259

0.461

0.213

0.258-0.289''

0.452-0.487*

0.184-0.222*

mean = (0.271)*

mean = (0.463)'

mean = (0.199)*

Oligocene

A. scutata PU

10172

0.313

0.469

0.215

A. scutata UMMP V-57431

0.312

0.472

0.214

Eocene

A. kehreri BMNH P33480
"Paramiatus gurleyi"

FMNH 2201
A. fragosa MCZ 5341
A. uintaensis PU 13865

0.310

0.466

0.201

0.308

*'sto.520

0.153

0.312

0.462

0.165

0.322

0.434

0.175

Range and mean exclude Specimen Codes 1 & 2 (MCZ 8970) and 13 (UMMZ 197683;

a very good fit for the ratio series, and that
the relative growth of these three dimen-
sions is essentially isometric rather than
allometric. The Wisconsin specimens (in-
cluding the 80-105-mm specimens ) and the
larger Michigan specimen all fall close to
the line calculated for each of the three
ratios (Fig. 3). The proportions of head-
length/standard-length, insertion of pelvic
fins/standard-length, and insertion of anal
fins /standard-length are shown in Table 3.
The head/standard-length ratio shows a
slight decrease with increasing size, but
this ratio series nonetheless has a very high
coefficient of correlation for the strength of
the linear relationship (Fig. 4).

The lengths of the mandible, parietal,
frontal, and operculum in Recent A. calva
appear in Table 4, and the proportional
ratios in Table 6. The relative growth rate
of each of these proportions is constant with
X and Y-intercepts of the straight line close
to the origin. The coefficient of correlation

for the variables in each of the proportions
is 0.997, 0.975, and 0.997, respectively ( Fig.
5). Combined, these two factors indicate
constant and therefore isometric relative
size-growth of the compared skull element.

Comparisons with Fossil Forms

Six articulated fossil specimens were
available for measurement and calculation
of head /standard-length and positions of
insertion of pelvic and anal fins/ standard-
length (Tables 2-3). The measurements
taken from the fossil forms are as exact as
conditions allow, although it must be
stressed that varying degrees of crushing
and distortion have occurred in fossilization,
and evaluation of the morphometries should
be qualified with this in mind.

Head /standard-length ratios (Fig. 4).
The fossil forms all show a slightly greater
head/ standard-length ratio than does the
Recent species (Table 3; Fig. 4). A.
uintaensis (PU 13865) is the largest of

Fossil Amiids • Boreske

Table 4. Length dimensions of Mandible (M), Gular (C), Frontal (F), Parietal (Par),
Infraorbital ^(I''), and Operculum (O) in 22 specimens of A. calva

Measurements in mm

Operci

ilum

Specimen

Dors. -Vent.

Ant.-Post.

Code

Par

(OL)

(OD)

13.5

8.0

10.0

5.0

8.1

7.5

14.9

9.4

11.4

6.1

5.8

9.0

8.4

30.0

19.0

18.0

9.7

12.0

14.0

12.9

32.0

21.0

20.0

9.9

13.5

15.0

14.2

33.5

20.5

19.0

10.0

14.2

16.0

14.9

39.0

26.0

25.0

11.0

17.0

16.9

16.5

42.0

28.0

25.5

11.5

18.0

18.1

17.5

45.0

27.0

26.2

13.5

21.0

19.8

18.5

53.0

31.0

14.7

22.5

22.8

21.8

56.0

32.5

32.2

17.0

26.5

22.6

22.5

63.0

39.0

38.4

18.5

30.5

28.1

27.8

66.5

42.5

39.0

19.5

31.5

27.8

28.5

13**

102.7

—

60.7

30.0

—

o MCZ 8970.
«<» UMMZ 197683.

all the fossil specimens, but nonetheless
has a greater head/ standard-length ratio
than any of the others. The head of this
form is so much more elongated than the
head in A. fra^oso (MCZ 5341), A. kehreri

(BMNH P33480), and "Paramiatus gurleiji"
(FMNH 2201) that it offsets the fact that
its vertebral column includes approximately
20 more vertebrae than do these three forms
(Table 9). Thus, although A. idntaensis

Table 5. Length dimensions of Head (H), Mandiiu^ (M), Gular (G), Frontal (F),
Parietal (Par), Infraorbital ^'{V'), and Operculum (O) in fossil amiids

Measurements in mm

Operculum

Dors.-Vent.

Ant.-Post.

Par

(OL)

(OD)

A. cf. scutata UCMP 38222

—

65.2

—

46.0

23.0

35.0

—

A. scutata PU 10172

106.0

62.5

31.2

35.0

16.0

—

29.0

28.0

A. scutata UMMP V-57431

121.0

74.2

44.3

20.0

29.1

27.9**

A. "dictt/ocephala"

AMNfl 2802

111.5

68.0

38.0

17.0

29.1

32.0

30.0

A. kchrcri BMNH P33480

59.2

30.2

—

20.0

8.4

15.3

20.5

19.0

"Paramiatus fiurletji"

FMNH 2201

157.0

80.0

—

58.0

23.6

25.0

40.0

37.0

A. fra^osa MCZ 5341

142.0

72.0

68.7

56.0

22.8

39.0

36.2

A. fragosa MCZ 9264

80.0

40.0

—

26.0

10.5

18.5

—

A. uintaensis PU 13865

214.0

—

— .

88.0

34.0

—

55.0

51.0

"Protamia" mongoliensis

AMNH 6372

—

81.0

—

54.0

52.0

A. uintaensis PU 16236

315.0**

220.0

158.0

160.0

60.0

—

95.0

A. fragosa MCZ 9291

— ■

—

27.0

25.0

A. fragosa AMNH 9315

—

29.0**

27.0

A. fragosa UA 5450*

—

26.0

10.0

—

A. fragosa UA 5458*

—

30.0

12.0

—

A. fragosa UA 5480*

—

20.0

26.0

24.0

•Data from O'Brien (1969).
•• Est.

12 Bulletin Museum of Comparative Zoology, Vol. 146, No. 1

240-

220-
200-

EQUATiONof STRAIGHT LINE- M= 7.268 + (0,234) (SLl

u
y

leo-

COEFFICIENT ol CORRELATION = 0.999

ISO-

MEAN ^ -. 0.271

STANDARD DEVIATION 5^ = 0.0117

COEFFICIENT of VARIATION =4.33%

E 140.

; 120-

S 100-

60-

xX'

SO'

40'

20-

320 400 460

STANDARD LENGTH mm

EQUATION of STRAIGHT LINE - PF = 1 438 + (0.454)(SL)

■^

288 '
240-

COEFFICIENT of CORRELATION .0.995
MEAN f^ '- 0.463
STANDARD DEVIATION |^ -■ 00131
COEFFICIENT of VARIATION ■ 2 83%

y '"

0 y

y
y
y
y

192-

144.

96-

48.

240 J20 400

STANDARD LENGTH mm

E
E
■z- 120'

EQUATION of STRAIGHT LINE ■■ P = -5.171 +(O.E27)(SL)

COEFFICIENT of CORRELATION = 0.987

MEAN ^ = 0.199

STANDARD DEVIATION^ .0.0135

COEFFICIENT ot VARIATION '6.78 %

240 320 400
STANDARD LENGTH mm

Fossil Ami ids • Boreske 13

has significantly more vertebrae than these
other forms, this feature is not reflected in
a comparison of head/ standard-length ra-
tios ( Table 3 ) . This is also true, to a lesser
extent, in both A. scutata specimens (PU
10172, UMMP V-57431) from the Ohgocene
Florissant Formation; these specimens fall
into the head/ standard-length range of the
three fore-mentioned forms, but like A.
uintaensis, possess vertebral columns having
nearly the same number of centra as those
in A. calva. Thus in themselves the head/
standard-length ratios are of little help in
comparing the fossil forms, but when
coupled with the corresponding lengths of
the vertebral column (based on number
of centra) they are informative. A. uintaen-
sis (PU 13865) and A. scutata (PU 10172,
UMMP V-57431) have relatively elongated
heads; A. kehreri (BMNH P33480), A.
fragosa (MCZ 5341), and "Paramiatiis g,ur-
leyi" (FMNH 2201) have relatively shorter
heads, since the head/standard-length ratio
is less than might otherwise be expected
considering the smaller total-number of
centra (only two-thirds the number of
centra of A. uintaensis, A. scutata, and A.
calva ) . A tentative growth-line ( also calcu-
lated by the best-fit method ) was included
for A. jragosa on the basis of three speci-
mens (Fig. 4). In comparison with the
growth-line of the Recent species (0.271
mean), it reflects the larger head/ standard-
length ratio of A. fragosa (0.310 mean).
The growth-line computed for A. jragosa
is linear and falls near the origin, indicating
that increase in head size/ standard -length
was isometric, as in A. calva.

Fin relations] lips. In the smaller fossil
forms, the ratio of the point of insertion
of the pelvic fin/ standard-length shows little
deviation from the modern species (Table
3; Fig. 4) except for two Eocene specimens,
"Paramiatus gurleyi" (FMNH 2201) and

A. uintaensis (PU 13865), which fall out-
side of the range on either side of tlie size-
growth line. The greater ratio for "Para-
miatus gurleyi," however, is probably the
result of distortion in its preservation. The
length of the pelvic fin insertion/ standard-
length does not appear to be a satisfactory
taxonomic index, distinguishing neither the
fossil forms from one another nor the fossil
forms from the Recent A. calva.

"Paramiatus gurleyi" (FMNH 2201), A.
uintaensis (PU 13865), and A. jragosa
(MCZ 5341) have a relatively shorter
dimension between the anal fin and the end
of the vertebral column than do A. calva,
A. scutata, and A. kehreri (Fig. 31). Any
attempt to inteipret the fossil data for this
ratio is complicated by the fact that con-
siderable overlap with the Recent species
occurs. Both long-bodied (A. scutata) and
short-bodied (A. kehreri) forms fall within
the range of A. calva, while other long-
bodied (A. uintaensis) and short-bodied
(A. jragosa, including "Paramiatus gur-
leyi") fonns fall below the range of the
Recent species (Table 3). Although the
ratio of anal fin/ standard-length may pos-
sibly be useful in distinguishing A. jragosa
(including "Paramiatus gurleyi") from A.
calva, A. scutata, and A. kehreri, it is
not useful in distinguishing either of the
two fossil fonns from one another or from
A. calva. The smaller dimension indicated
by the low ratios (0.153, 0.165) of A.
jragosa is doubtless a reflection of its
shorter axial column. The relatively small
(0.175) ratio for A. uintaensis is probably
in part the result of its longer head, wliich
increases its standard-length in relation to
the other forms; at any rate, the difference
between the A. uintaensis ratio and the
range for Recent A. calva is not very sig-
nificant.

Mandible I head ratios. A comparison of

Fig. 4. Relative growth-lines (broken-solid lines) of head-length, pelvic fin insertion, and anal fin insertion plotted
arithmetically against standard-length for Recent Amia calva (A = MCZ 8970 and ■ = U/AMZ 197683 are included
for comparison) with compared fossil forms: fl = A. hagosa (A. kehreri) BMNH P33480; f- = A. fragosa (Pararr^iafus
gurleyi) FMNH 2201; f'-^ = A. fragosa MCZ 5341; s^ = A. scufafa PU 10172; s^ = A. scutofo UMMP V-57431;
u = A. uintaensis PU 13865. The broken-dotted line is the "best fit" line for available specimens of A. fragosa.

14 Bulletin Museum of Comparative Zoology, Vol. 146, No. 1

100
90^

tOuATiON o( STRAIGHT LINE ■ M ■-! 010 + (0-636HM)

COEFFICIENT of CORRELATION ^ 0. 997

MEAN ^ ' 0 609

STANDARD DEVIATION ^ ■ 0 0171

COEFFICIENT of VARIATION -■ 2.81 %

55-

50.

45-

EQUATION 0) STRAIGHT LINE ^ PAR = 0 278 + (0 479HF) '^ y
COEFFICIENT ol CORRELATION • 0.975 /^ ^

40-

MEAN ^ . 0.495 / /"^

E
±

35-

STANDARD deviation""- 0 0333 ^^ /
COEFFICIENT of VARIATION- 6 73% / / "^ V

/ /

X
UJ

30-

f' y

/ /

25-

V A

/ /•'

20-

1!-

10.

/^'

5.
0.

30 40 50 60

SO 90 rOO 110 120

FRONTAL LENGTH mm

o-»r

EQUATION of STRAIGHT LINE : OD ■ - 1.559 + (l.079)(&LI

COEFFICIENT of CORRELATION - 0.997

MEAN g2 -0.964

STANDARD DEVIATION ^ - 0.0326

COEFFICIENT of VARIATION- 3.40%

20 30 40

OPERCULUIM LENGTH mni

Fossil Amiids • Boreske

Table 6. Cranial proportions in 22 specimens
OF A. calva

Specimen

Code

M/H

Par/F

OD/OL

0.614

0.500

0.926

0.596

0.535

0.933

0.594

0.539

0.921

0.586

0.495

0.947

0.590

0.526

0.931

0.609

0.440

0.976

0.613

0.451

0.967

0.616

0.515

0.934

0.646

0.474

0.956

0.615

0.528

0.996

0.611

0.482

0.989

0.610

0.500

1.025

0.626

0.497

0.586-

0.440-

0.921-

0.646*

0.539"

1.025"

mean

= (0.609)"

= (0.495)"

= (0.964)"

■* Range and mean exclude Sjiecimen Codes 1 & 2
(MCZ 8970) and 13 (UMMZ 197683).

the mandible/head ratios of Recent A. calva
with those of the fossil forms ( Table 7; Fig.
5) indicates that the A. scutata and A.
''dictyocephala" (AMNH 2802) ratios are
very close to those of A. calva. The A.
fra^osa .specimens (including "Taramiatus
^urleijr FMNH 2201 and A. kehreri BMNH
P33480) have a mean mandible/ head ratio
of 0.507, which, when compared to the
A. calva mean ratio of 0.609, indicates a
relatively smaller mandible to head size
(Table 7). Unfortunately, A. uintaensis
(PU 13865) cannot be used in this com-
parison, since the mandibles are buried in
matrix. A reconstruction of a disarticulated
A. uintaensis (PU 16236) specimen from
the Late Pal eocene has been made, and its
ratio is approximately 0.693. Thus man-
dible/head proportions may be valid for

distinguishing specimens of A. fra^usa and
A. uintaensis from one another as well as
from A. calva and A. scutata. This ratio,
however, caimot be used as a valid criterion
for distinguishing A. scutata from A. calva.
The 0.693 mandible/ head ratio of A.
uintaeims indicates that this form has the
largest mouth gape of the four valid species.
A tentative growth-line for the mandible/
head-length proportion of A. fra<i^osa, estab-
lished on four specimens, shows that its jaw
is 16 percent smaller than that of A. calva,
and in this respect confirms Romer and
Fryxell's (1928) reconstruction of "Paramia-
tus ^urleiji."

Parietal / frontal ratios. Only articulated
frontals and parietals were measured for
this study. The frontal-length was taken
from the anteriormost extent of the dermal
sculpture to the median point between the
most anterior and posterior extents of the
frontal-parietal suture; the parietals were
also measured by their midline anteropos-
terior length. The parietal/ frontal ratio of
the fossil forms as compared with that of
the Recent A. calva indicates that, in vary-
ing degree, the fossil species have relatively
shorter parietals and longer frontals ( Table
7; Fig. 5). The largest specimen of A.
uintaensis (PU 16236) is mostly disarticu-
lated, but fortunately the skull table is in-
tact. It has the smallest parietal/ frontal
ratio of all the fossil .species, 0.375. The
Eocene A. uintaensis specimen (PU 13865)
has a slightly larger ratio of 0.386, which
is nearly equal to the Edmonton A. jragosa
(UA 5450). All the other available A.
fragosa specimens, including "Paramiatus
gurleyi' ( FMNH 2201 ), have slightly larger
ratios and are quite consistent, ranging only

Fig. 5. Relative growth-lines (broken-solid lines) of mondible-length plotted arithmetically against head-length,
parietal-length plotted arithmetically against frontal-length, and operculum-depth (anteroposteriorly) plotted arith-
metically against operculum-length (dorsoventrally) for Recent Amia calva (A = MCZ 8970 and ^ ^^ '-"^'^'^^ 197683
are included for comparison when element measurements are available) with compared fossil forms: f =^ Amia
fragosa (A. kehreri) BMNH P33480; f- = A. fragosa (Paramiatus gurleyi) FMNH 2201; f* = A. frogosa MCZ 5341;
i^=A. fragosa MCZ 9264; f"' =: A. frogoso UA 5458; f*' = A. fragosa UA 5450; f = A. fragosa AMNH 9315;
f*^ = A. fragosa MCZ 9291; f'^ = A. fragosa UA 5480; si = A. scufaia PU 10172; s- = A. scufaia UMMP V-57431;
s-' = A. scufafa (A. dictyocephala) AMNH 2802; s'=A. cf. scufafa UC 38222; u^ = A. oin/oensis PU 13865; u- =
A. ointoensis PU 16236; m = A. mongo/zensis AMNH 6372. The broken-dotted line is the "best fit" line for available
specimens of Amia fragosa.

16 Bulletin Museum of Comparative Zoology, Vol. 146, No. 1

Table 7. Comparison of cranial proportions of recent and fossil amiids

M/H

Par/F

OD/OL

Recent

A. calva ( 18 ) ( Wise. )

Miocene
A. cf. scutata UCMP 38222

Oligocene

A. scutofa PU 10172
A. scufata UMMP V-57431
A. "dictyocephala"
AMMH 2802

Eocene

A. fce/jreh BMNH P33480
"Paramiatus gurleyi"

FMNH 2201
A. /ragosa MCZ 5341
A. /ragosa MCZ 9264
A. uintaensis PU 13865
"Pappichthys" mongoliensis

AMNH 6372

Paleocene

A. uintaensis PU 16236

Cretaceous
A. /ragosa MCZ 9291
A. fragosa AMNH 9315
A. fragosa UA 5450
A. fragosa UA 5458
A. fragosa UA 5480

0.586-0.646
mean = (0.609)

0.590
0.613

0.610

0.510

0.693*

0.440-0.539
mean = (0.495)

0.500

0.457
0.451

0.447
0.420

0.375

0.385
0.400

0.921-1.025
mean = (0.964)

0.965
0.959

0.937

0.927

0.509

0.407»

0.925

0.507

0.408

0.928

0.500

0.404

—

0.386

0.927

_^_

___

0.963

0.926
0.931''

0.923

Est

between 0.400-0.408. One of the specific
character-states that Estes and Berberian
(1969: 6) list for A. fragosa is a frontal-
length of approximately 2.8 times the length
of the parietals, which would give a ratio
of 0.357. This figure is smaller than that of
the known articulated forms, including the
specimens from the Edmonton and Will-
wood formations. Although they may have
placed too much emphasis on this specific
character-state, Estes (1964), Janot (1967),
and Estes and Berberian (1969) are justi-
fied in distinguishing A. fragosa from A.
calva on this basis since the ratio of the
fossil form is smaller than that of the Recent
A. calva, whose parietal /frontal proportions
have a mean ratio of 0.495, the frontals
being approximately twice the length of the
parietals. A tentative growth-line for
parietal/ frontal proportions, established on
six specimens of A. fragosa, including

"Paramiatus gurleyi" (FMNH 2201) and
A. kehreri (BMNH P33480), illustrates this
difference between the fossil form and the
Recent species (Fig. 5). A. kehreri dis-
plays slightly larger parietals, with a ratio
of 0.420, but considering the geographic
and temporal differences from the other
A. fragosa specimens, it is remarkably close
in this feature to its North American rela-
tives. The parietal /frontal proportions of
the specimens of A. "dictyocephala" and
A. scutata fall near the lower end of the
0.440-0.539 range of A. calva, and the
Miocene specimen of Amia ( UCMP 38222),
with its ratio of 0.500, is very near the mean
for A. calva. There is thus a definite trend
from the Cretaceous to the Miocene (and
Recent) toward an increase in parietal/
frontal ratio. The A. uintaensis and A.
fragosa specimens have parietal /frontal
ratios smaller than those of A. calva, while

Fossil Amiids • Boreske 17

the A. ^'dictyucepluila" and A. scutata speci-
mens are close to A. calva in this propor-
tion. There is, however, enough intra-
specific variation of parietal /frontal ratios
in the fossil species to cause an interspecific
overlap of the various forms, so that it is
impossible to detennine any definitive limits
between the consecutive fossil species and
A. calva.

Operculum-depth / operculurn-lengtJi ra-
tios. Although the fossil forms have a
slightly narrower operculum-depth relative
to their operculum-length, they all fall
within the operculum ratio range of 0.921-
1.02.5 for A. calva (Fig. 5). Table 6 indi-
cates that with increasing size in A. calva
there may be a trend from a narrower to a
slightly broader operculum. Romer and
Fryxell (1928: 521) describe the operculum
of "Paramiatus <!,urleiji" as being greater
dorsoventrally than anteroposteriorly. Al-
though Cretaceous and Eocene specimens
of A. fragosa have operculum ratios ( 0.923-
0.963) lower than the mean (0.964)
for A. calva, they still fall within the range
(0.921-1.025) of tlie Recent form. Thus the
variation of operculum shape within A. calva
contradicts Hussakof's ( 1932 ) supposition
that the operculum in "Pappichthys" mon-
goliensis (with a ratio of 0.963) is propor-
tionately narrower than that of A. calva, as
well as Estes and Berberian's ( 1969 ) diag-
nosis that A. fragosa has a relatively shorter
operculum-length as compared with height
than A. calva. Janot (1967) was also cau-
tious in assigning taxonomic importance to
the operculum proportions because of the
great variability within the Recent species.
The operculum width /length ratios in tlie
fossil specimens show little taxonomic sig-
nificance, although, as Estes and Berberian
(1969: 7) note, there does appear to have
been a slight temporal trend toward a
broader operculum.

Discussion

The six relative growth proportions
(Figs. 4-5) that were plotted for the A.
calva growth-series remained constant and
therefore isometric. This may be explained

by the fact that these ratios are derived
from external rather than internal dimen-
sions, and, as Gould (1966) points out,
it is usually the internal elements that must
increase at an allometric rate in order to
maintain the external surface area, whose
dimensions may be increasing at an iso-
metric rate (see meristic study). It may
be assumed that the relative growth for
these six proportions also maintained an
isometric rate in the fossil forms, since their
ratios invariably fall near the growth-lines
for the Wisconsin A. calva sample (Figs.
4-5). However, this assumption would
have to be confirmed with an actual
growth-series of the fossil forms.

The comparison of Recent with fossil
forms has also made it possible to determine
the taxonomic value of the skull/ body and
skull proportions. The moiphometric com-
parison of the fossil forms with Recent A.
calva suggests the following taxonomic and
phylogenetic trends :

1. All the fossil forms have slightly longer
heads relative to their standard-length
than does the Recent species ( Fig. 31 ) .
Unless the differences in vertebral meris-
tics are also considered, however, the
morphometric data for this feature are
not useful in comparing the various
fossil forms. A. uintaensis and, to a
lesser extent, A. scutata have relatively
longer heads with a vertebral column of
approximately 85 centra. A. fragosa, on
the other hand, is a short-bodied form
(approximately 65 centra) with a short
head. A. calva has a relatively long ver-
tebral column (81-90 centra) with a
shorter head than the other long-bodied
forms (A. uintaensis and A. scutata).

2. Pelvic fin insertion has no taxonomic sig-
nificance.

3. Anal fin insertion may have minor
ta.xonomic significance for the North
American specimens of A. fragosa wliich
are relatively shorter in this dimension
than in the other .species. There is too
much morphological overlap between

18 Bulletin Museum of Comparative Zoology, Vol. 146, No. 1

the species, however, to make this a
useful criterion.

4. A. uintaensis has a relatively longer
mandible/head ratio (0.693) than any
of the other species of Amia, while A.
frafiosa has a smaller ratio (0.507
mean). The mandible/head ratio of A.
scutata (0.604 mean) is close to that of
A. calva ( 0.609 mean ) .

5. There is a trend from the Late Creta-
ceous to Late Miocene in the lengthen-
ing of the parietals in relation to the
frontals. Although it is possible to dis-
cern groups that fall into categories of
smaller and larger ratios (Table 7), in-
terspecific moi-phological overlap makes
it difficult to separate any one of the
fossil species from the others on this
criterion.

6. All the fossil forms have operculum
depth /width proportions that fall into
the lower limits of the A. calva range
(0.921-1.025). These ratios show a
slight temporal trend towards increasing
width, but, while this is perceptible, it
is insufficient to indicate taxonomic sig-
nificance.

These trends suggest possible phylo-
genetic relationships between the various
amiid species. The moiphometric similari-
ties indicate that little evidence exists for
maintaining A. scutata and A. "dictijoce-
phala" as separate species. The Oligocene
A. scutata is distinguishable quantitatively
from A. calva only on the basis of a larger
head/ standard-length ratio, and in this fea-
ture it is intermediate between A. calva and
A. uintaensis. The moi-phometric evidence
indicates similarities between A. fra^,osa
(Cretaceous-Eocene), "Paramiatus ii^iirleiji"
( Eocene ) , and A. kehreri ( Eocene ) . Head/
standard-length ratio is approximately the
same among these three forms; insertion of
anal and pelvic fins /standard-length ratios
shows only minor differences. Mandible/
head size and parietal /frontal ratios are al-
most identical. Of all the .species, A.
uintaensis is the most morphometrically dis-
tinct. It has a relatively greater mandible/

head ratio and a smaller parietal /frontal
ratio than A. fra^osa (Table 7). Even
though it possesses approximately the same
total number of centia as A. calva and A.
scutata (Table 9), it still has a greater
head/ standard-length ratio than the two
latter species. Temporally, there are minor
trends in Amia towards lengthening of the
parietals in relation to the frontals, and in-
creasing operculum width to depth.

MERISTICS

Meristic elements have been used in
species diagnoses of various fossil amiids
by Cope (1875), Osborn et al. (1878),
Romer and Fryxell (1928), and Estes
( 1964 ) . A meristic study of both Recent
and fossil species of Amia was undertaken
to determine tlie relative value of such
diagnoses in the taxonomy of the amiids.
A comparison of the number of supraverte-
bral scale rows, the number of branchio-
stegal rays, and the number of pectoral,
pelvic, anal, dorsal, and caudal fin rays com-
prises the first part of the study, while com-
parative vertebral meristics comprise the
second part.

Supravertebral Scale Rows

Cope (1875) differentiated A. scutata from
A. calva and A. "dictyocephala" on the basis
of A. scutata s (USNM 5374) having seven
and a half longitudinal rows of large scales
above the vertebral column. Cope (1875)
described A. "dictyocephala" (USNM 3992)
as bearing ten to twelve rows of scales
above the vertebral column. A count of the
scale rows between dorsal fin distal pterygi-
ophores and the vertebral column in 20 Re-
cent A. calva (Table 8) gave a range of
seven to nine supravertebral scale rows.
Although the number of scale rows will
vary with the region of the trunk anatomy
from which the count might be taken, Cope
did not designate the point at which he
made his scale row count. Also, his speci-
men WAS so poorly preserved that his count
may have been affected by distortion of
the scales. The only way that a valid com-
parison of all the fonns could be made was

Fossil Amiius • Boreskc

TaULK fS. (loMl'AHlSON OF MKHISTIC: ELEMENTS IN HECENT AND FOSSIL NOHTH AMERICAN AMIIDS

Supra-

Branchi-

Pecf oral-Fin

Pelvic-Fin

Anal-Fin

Dorsal-Fin

C"aiidal-Fin

vertebral

ostegal

Lepido-

I.epido-

Lepido-

Lcpido-

Scale Rows

Rays

trichia

Irichia

trichia

tridiia

Amia calva (20)

7-9

10-13

16-19

7-8

8-11

45-49

23-27

Recent

7.5 av.

11.4 av.

16.8 av.

7.2 av.

10.5 av.

48.0 av.

25.7 av.

A. sciitata

YPM 6243"

USNM 4087*

PU 10172'

—

47""'

A. scutata

USNM 5374

7.5

—

A. scutata

YPM 6241

—

A. scutata

UMMP V-57431

46**

A. "dictijoccphala"

USNM 3992

7.8

—

48»»

—

A. "dicttjoccphaJa"

AMNH 2802

—

A. "dictijoccphala"

AMNH 2670

9*'

—

470 »

A. iiintacnsis

PU 13865

9"*

A. uintacnsis

AMNH 785

9«o

—

A. fragosa

MCZ 5341

19-20

"Paramiatus gu dcyi"

FMNH 2201

7-8

44-45**

A. fragosa

UA 5506

—

A. fragosa

UA 5425

—

" All one specimen.
"o Est.

to take the supravertebral scale row count
of both the USNM 3992 specimen and the
other fossil and Recent amiid specimens at
the same point. In this case, I took all
connts on a vertical line at the level of the
posterior pterygiophore of tlie anal fin. I
connted the nnmber of scale rows in speci-
mens of A. frcif^osa, A. uintaensis, and "Para-
miatus ^urleyi" in addition to those of
Cope's types of A. scutata (USNM 5374)
and A. "dictyocephala" (USNM 3992), as
well as referred specimens of A. scutata;
I then compared them with the supraverte-
bral scale row range in A. calva. The
supravertebral scale rows of fossil Amia
(Table 8) appear to fall within the supra-
vertebral scale row range of Recent A.
calva. Although Cope had described A.

"dictyocephakr (on the basis of USNM
3992) as having 10-12 scale rows, I believe
his count is too high. The supravertebral
scale rows in this and other fossil forms are
difficult to observe for several reasons.
Amiid scales are aligned diagonally to the
vertebral column rather than in parallel,
making it often difficult, particularly in
fossil material, to determine to which diag-
onal column the overlapping scale rows
belong. Also, the scales on the USNM 3992
specimen are broken into many parts, and
Cope may therefore have been counting
partial scales as whole ones. I believe that
I obtained a more reasonable estimate of
the supravertebral scale row count in this
specimen in the following manner: I
measured the average of the anteroventral

20 Bulletin Museum of Comparative Zoology, Vol. 146, No. 1

width of complete scales from the abdomi-
nal region (in which the scales are the
same size as in other places in the mid-body
region) and then divided that amount into
the distance between the midpoint of the
vertebral column and the dorsal fin distal
pterygiophore. In this case, the quotient
was 7.8, which is comparable with the
counts of approximately 7-9 in the other
Oligocene specimens and in A. calva ( Table
8). No taxonomic significance can thus be
applied to the number of scale rows above
the vertebral column since counts in Recent
and fossil Amia fall within a' relatively
narrow range.

Branchiostegal Rays

The number of branchiostegal rays was
included in the species diagnosis for A.
"dictyocephalo" (AMNH 2802), in which
Cope ( 1875) counted 12 rays. Osborn et al.
(1878) observed 13 branchiostegal rays in
A. scutata (PU 10172) and Romer and
Fryxell (1928) figured 12 such rays for
"Paramiatus ^urleyi" (FMNH 2201).
O'Brien (1969) counted 10 rays in A.
fragosa (UA 5506) from the Edmonton
Formation. On the basis of disarticulated
material from the Late Cretaceous Lance
Formation, Estes (1964) estimated that A.
fragosa would bear 14 branchiostegal rays,
like the Late Jurassic Sinamia zdanskyi de-
scribed from China by Stensio (1935; see
Liu et al, 1963 for range and distribution).
In the Recent A. calva sample (Table 8),
the number of branchiostegal rays ranges
from 10 to 13; the range among the few
known examples of fossil forms is from 10
to 13, an indication that the number of
branchiostegal rays has remained constant.

Fin Rays

Because of confusing duplication of ter-
minology used for fin description in the
literature, I will use that of Lagler et al.
( 1962 ) for the appendicular skeleton unless
I indicate otherwise.

All fin ray counts on Recent A. calva
were obtained from X-rays of 20 specimens
from Wisconsin and Michigan. The counts

taken from fossil forms are as accurate as
conditions allow, although a number of the
specimens are incomplete or show only
traces of the actual fin. The results of this
study must therefore be considered with
this in mind. I ol^tained these counts from
as close as possible to the internal fin sup-
ports rather than to the segmented and bi-
furcated distal lepidotrichia. There is a
one-to-one correspondence between the
number of lepidotrichia and the number of
pterygiophores in the anal and dorsal fins;
however, this is not the case in the pectoral,
pelvic, and caudal fins, which have more
lepidotrichia than fin supports (Fig. 31).

Pectoral fin. The number of pectoral fin
lepidotrichia has not been previously noted
in any of the original species descriptions
of fossil Amia. There are four .specimens in
which it is possible to make a pectoral fin
ray count (Table 8). A. scutata (1), A.
uintaensis (1), and A. fragosa (2) speci-
mens bear 16 to 18 pectoral lepidotrichia, a
number which is approximately the average
for 20 specimens of Recent A. calva which
displayed from 16 to 19 pectoral fin rays
(Table 8). O'Brien's (1969) analysis of A.
fragosa (Edmonton Formation) does not
include any quantitative comparison of its
pectoral fins with those of A. calva. He
does, however, observe that the pectoral
fins are qualitatively similar in the two
species. The pectoral fins of A. fragosa, A.
scutata, and A. uintaensis thus do not vary
meristically from those of A. calva. Lehman
(1951: 8), in his description of Pseudamia
heintzi from the Eocene of Spitzbergen,
notes that the pectoral fin has 13 complete
nonbifurcating lepidotrichia in the visible
portion of the fossil. This count is different
from that of both Recent and fossil North
American amiids, but as Lehman's plate 3
indicates, this difference may be caused by
matrix that overlies the ventral portion of
the pectoral fin, possibly covering addi-
tional lepidotrichia.

Pelvic fin. The number of lepidotrichia
of the pelvic fin was part of Cope's ( 1875 )
species diagnosis for A. "dictyocephala"
(USNM 3992) and that of Osborn et al.

Fossil Amiids • Boreske 21

(1878) for A. scutota (PU 10172). I
counted the lepidotrichia of these speci-
mens as well as those of on(> additional
Ohgocene specimen and compared them
with my sample of A. calva, which showed
between seven and eight pehic lepido-
trichia (Table 8). Although Osborn et al.
( 1878) counted ten pelvic lepidotrichia, my
recount of their A. scutata specimen (PU
10172) showed only seven (Plate 4). The
bifurcation of the fin rays might have been
inadvertently included in their original
count. The holotype of A. "dictyocepliala"
(USNM 3992) (Fig. 27) showed seven
rather than the six lepidotrichia that Cope
(1875) had diagnosed. A specimen of A.
scutata (UMMP V-57431) (Fig. 27A) also
has seven lepidotrichia; both of these are
within the range of Recent A. calva. Of the
remaining fossil forms, A. fraii^osa and
"Paramiatus ^urleyi" have eight, and A.
uintaensis nine, A. uintacnsis being the only
fossil form not to fall within the range of
Recent A. calva. This difference is insuf-
ficient to demonstrate any taxonomic value,
however, at least until more A. uintaensis
specimens are known.

Anal fin. Anal fin lepidotrichia have
been included in the diagnoses of A. "dic-
tyocephala" and A. scutata (Cope, 1875),
and also in the description of A. .scutata
(Osborn et al, 1878). Each of the original
counts of nine anal rays for each specimen
concurs with my recount and also falls
within the range of eight to eleven for Re-
cent A. calva (Table 8). A. jra<i,osa, "Para-
miatus pMrleyi" and A. uintaensis also fall
within the range of A. calva.

Dorsal fin. Although the number of
lepidotrichia in the dorsal fin has been
mentioned by several authors in their diag-
noses of fossil amiids, it is one of the more
difficult meristic counts to obtain, since a
complete dorsal region of the fossil is
required. Cope's type of A. "dictyocephala"
(USNM 3992) lacks a complete' dorsal fin,
so he counted only the 32 dorsal lepidotri-
chia between the beginning of the dorsal
fin and the posterior lepidotrichia of the
anal fin (Cope, 1875). Osborn et al.

(1878) reported 53 dorsal lepidotrichia for
A. scutata (PU 10172), but this must have
been an estimate, since the posterior por-
tion ot the dorsal fin as well as the entire
caudal fin is missing (Plate 4C). As the
two A. scutata specimens with complete
dorsal fins (AMNM 2670, UMMP V-
57431) have, respectively, 47 and 46 dorsal
lepidotrichia (Table 8), it seems that the
count of Osborn ct al. ( 1878) was high and
that the PU 10172 specimen would prob-
ably have corresponded with the other
Oligocene specimens.

Romer and Fryxell's ( 1928 ) diagnosis for
"Paramiatus fiurleyi" includes a dorsal fin
ray count of 45, which they note as being
slightly fewer than the count for A. calva.
O'Brien's ( 1969 ) discussion of Edmonton
Formation A. fra^osa does not include any
counts of dorsal lepidotrichia, although he
does note that the relative length of the
entire dorsal fin in A. fraiiosa is similar to
that of A. calva. In the A. calva specimens
I studied, the dorsal fin rays ranged be-
tween 45 and 49, the average approximately
48. Romer and Fryxell's diagnosis of "Para-
miatus fiurleyi" as having slightly fewer
dorsal lepidotrichia than A. calva is correct,
but this and all the related fossil forms fall
within the lower range of A. calva (Table
8). The number of dorsal fin rays appears
to have little taxonomic value.

It is interesting that the complete Amia
frafi:osa (MCZ 5341), "Paramiatus <^urleyi"
(FMNH 2201), and A. kehreri (BMNH
P33480) specimens have dorsal fins of
nearly the same length and contain the
same number of lepidotrichia as A. calva,
despite the fact that, on the basis of the
number of vertebrae, these species have a
much shorter body (Table 9). This con-
tributes to a proportional difference in the
body forms of these species, since the dorsal
fin in A. fra<^o.sa (including "Paramiatus
il,urleyi") terminates much closer to the
caudal fin than in A. calva (Plate 1; Fig.
31). However, as Shufeldt (1885) and Hay
(1895) implied, it is very doubtful that the
dorsal fin was fused into a continuous struc-
ture with the caudal fin in some ancestral

22 Bulletin Museum of Comparative Zoology, Vol. 146, No. 1

amiid. The Late Mesozoic European forms
of Urocles and Amiopsis have a much ab-
breviated dorsal fin that terminates more
anteriorly than does that of the species of
Amia. For Amiopsis dolloi, an Early Creta-
ceous (Wealden) amiid from Bernissart,
Belgium, Traquair (1911) figured 17 dorsal
fin supports, while Lange (1968) estab-
lishes a specific range of 17-25 for the Eu-
European Upper Jurassic Urocles. The
basis of Shufeldt's (1885: 8.5-86) model for
a primitive amiid with a continuous dorsal-
caudal fin was the presence in Recent Amia
calva specimens of what Shufeldt called a
"series of delicate little bones that continue
the interspinous bones of the dorsal fin as
far as the caudal fin." These five bones he
considered to be the continuation of the
dorsal interneural spines; Hay (1895), in
his discussion of Amia calva, refers to them
as "epural interspinous bones." These small
bones can also be seen in several of the
fossil amiids I have studied, especially A.
scutata (YPM 6241), A. fragosa (UA 5425),
and A. uintaensis (AMNH 785) (Fig. 8).
From these fossil forms, however, it is dif-
cult to determine whether the origin of
these bones is from the dorsal or caudal
neural spines. An examination of the caudal
fin of Urocles lepidotus ( Fig. 6; also
Nybelin, 1963: 506, fig. 17), which is
known to have an abbreviated, more an-
teriorly located dorsal fin, shows that these
epural bones are associated with the caudal

fin, which supports the upper caudal lepi-
dotrichia in much the same manner as the
hypurals in the ventral tail region. A
further indication that these epural inter-
spinous bones are not vestigial dorsal spine
supports is found in Traquair's ( 1911 ) plate
7 of Amiopsis dolloi and his plate 8 of
Amiopsis lata (both species from the Creta-
ceous [Wealden] of Belgium); these plates
show the bones to be clearly associated
with the caudal lepidotrichia (Fig. 7).

Fig. 6. Urocles lepidofus MCZ 8300, caudal fin.

Fig. 7. Restoration of Amiopsis dolloi, scales omitted
Early Cretaceous (Wealden), Bernissart, Belgium (after
Traquair, 1911).

Fossil Amiids • Boreske 23

There are no intermediate interspinous
bones between these bones in the caudal
region and those of the much more anteri-
orly situated dorsal spine. The fin of a 648-
mm SL Recent Amia calva (Fig. 10) does,
however, confirm tliat Shuf eldt ( 1885 ) was
correct in stating that the epurals are con-
tinuations of the interneural spines. Figure
10 shows three free interspinous epurals,
with a fourth that is either being fused onto
a neurd spine or is actually a single greatly
elongated neural spine. As only two of
these epurals are attached to lepidotrichia,
there is not a one-to-one correspondence
between the two elements, as in the hy-
purals in the main caudal region.

Caudal fin. With the exception of
Romer and Fryxell's (1928) diagnosis of
"Paramiatus g,urleyi," none of the original
descriptions of fossil Amia include counts of
the caudal fin rays. Although Romer and
Fryxell observed 20 caudal lepidotrichia, a
recount shows only 19 (Fig. 8E). Other
fossil forms tliat also show 19 caudal fin
rays are A. fragosa (UA 5425) from the
Edmonton Formation, A. fragosa (MCZ
5341 ) from the Green River Formation,
and A. kehreri from Messel (Andreae,
1895, plate 1, fig. 23). Another specimen
of A. kehreri from Messel (BMNH P33480)
has 18 lepidotrichia (Plate 2). Traquair's
( 1911 ) plate 7 of three specimens of Amiop-
sis dolloi shows between 15 and 17 caudal
lepidotrichia, while Urocles spp. have a
range between 12-18 caudal lepidotrichia
(Lange, 1968). The only Oligocene speci-
men with a complete caudal fin (YPM
6241) has 23 caudal lepidotrichia; the
Eocene specimens of Amia uintaensis show
23 to 24. Although my sample of Amia
calva displays caudal fins with a range of
23 to 27 lepidotrichia, the number of these
caudal fin rays is skewed toward the higher
limit of the range (Fig. 9). There is thus
a considerable difference between the num-
ber of lepidotrichia in Amia fragosa and
the majority of the A. calva specimens. A.
scutata is, however, within the range of the
Recent species, but occupies the lower
limits of the range.

Thus, of all the meristic elements so far
considered, it appears that the greatest dis-
parity between the fossil forms and the
Recent A. calva is in the number of caudal
fin rays. The number of caudal fin rays
therefore appears to have taxonomic impor-
tance and may have some functional as well
as morphological correspondence to the two
different amiid body types.

As discussed in the preceding section on
dorsal fins, there are two attachment bases
for the caudal lepidotrichia: epural inter-
spinous bones and the hypurals. The
epurals are usually attached to only two or
three of the caudal fin rays, while the
remainder of the lepidotrichia are sup-
ported by the hypurals. Nybelin {in 1963:
488) defines hypurals as "those haemal ele-
ments located to the rear of the emergence
of the caudal artery from the haemal canal"
(trans. Lund, 1967: 210) (Fig. lOB). Lund
(1967: 210) agrees instead with White-
house (1910: 592), who defines hypurals
as "any hypaxial elements that support
caudal fin rays" (Fig. lOA). Lund states
that the sole function of a hypural is to
support a caudal fin ray and therefore the
first hypural would be "the first haemal
spine in rem'ward progression to support a
caudal fin ray and the first ural centrum
is the centrum supporting the first hypural
element." Lund's definition is more practical
for paleontological use. Since there is an
intennediate joint (Figs. 8, 10). the major-
ity of the hypurals are not attached directly
to the urals. However, as Shufeldt (1885)
and Hay ( 1895 ) observed, the posterior-
most seven to nine hypurals are ankylosed
to the corresponding vertebrae ( Fig. IOC ) .
This same co-ossified condition of the last
hypurals is also evident in the fossil fonns,
so that the number of these fused hypurals
has remained constant throughout the evo-
lutionary history of Amia. Also, as Figure
10 shows, the seven or eight anteriormost
hypurals of Recent A. calva have a one-to-
two correspondence with the ventral lepi-
dotrichia. In most of the available fossil
amiid specimens, the ventral caudal portion
is poorly preserved, so that it is difficult to

24 Bulletin Museum of Comparative Zoology, Vol. 146, No. 1

Fig. 8. Caudal regions: A, Amia scufaia YPM 6241; B, A. scufafa AMNH 2671; C, A. uintaensis AMNH 785;
D, A. fragosa UA 5425; and E, A. fragosa FMNH 2201.

Fossil Amiids • Boreske 25

11 ^n ^li ^5 2r~

Caudal Fin Lepidotrichia no.

Fig. 9. Number of caudal lepidotrichia in 20 speci-
mens of Recent Ami'a calva.

arrive at an accurate count of the total
number of hypurals or to verify whether
this one-to-tvvo relationship exists in all the
amiid fossil forms. The only available fossil
form in which this one-to-two hypural-
lepidotrichia coiTCspondence in the ventral
caudal region can clearly be seen is in A.
scutata (YPM 6241; Fig. 8).

Vertebral Elements

Two regions of the vertebral column, the
trunk and the caudal regions, are defined by
their relationships to the ribs, neural arches,
and haemal arches. The trunk region con-
sists of monospondylous vertebrae that pos-
sess paired basapophyses having gradually
changing angles, dorsal neural facets, and
ventral aortal facets. The number of tnmk
vertebrae in my sample of Ainia calva varies
from 36 to 38. The caudal region consists
of three types of vertebrae, listed from an-
terior to posterior: regular monospondylous
centra bearing neural and haemal arches,
diplospondylous centra bearing neither
neural nor haemal arches (neural and
haemal facets still present ) , and ural centra.
Since the neural and haemal facets are still
present in the diplospondylous centra, there

is no way to differentiate the latter from
the monospondylous type in a disarticulated
state. In my sample of A. calva, the number
of regular caudal monospondylous centra
(24-26) fluctuates by two centra, that of
the diplospondylous caudal centra ( 14-17 )
by three (Table 9).

The posterior caudal region of A. calva
consists of two types of urals: centra with
hypurals attached by a layer of cartilage
(free urals), and centra that are fused
directly onto the hypurals, often lacking the
neural arches (fused urals). When dis-
articulated, the fused urals can often be
distinguished from the free urals, since
part of the hypural usually remains fused
to the ural, extending the posterior articular
surface downward. The nonfused (free)
urals cannot be distinguished in a disarticu-
lated state from the monospondylous or
diplospondylous caudal centra. The num-
ber of urals with fused hypurals is readily
counted, since they are distinguishable
from the remainder of the vertebrae. In
order to identify a free ural, it is necessary
to observe the relationship between the
ural and its conesponding hypural and
lepidotrichia. It is often difficult to make
this distinction between free and fused
urals, since the caudal region is seldom
complete in articulated fossil forms. In A.
calva the number of urals with ankylosed
hypurals ranges between seven and nine.
There are approximately seven principal
urals fused to hypurals, followed by one or
two small additional urals that do not ar-
ticulate with the preceding vertebrae but
lie dorsal to the upturned portion of the
vertebral column. Because it is difficult to
discern these urals in smaller specimens of
A. calva, the count may be slightly biased,
and a comparison of the fossil forms with
the range established for A. calva must be
made with this consideration in mind.

I counted the number of centra between
the anterior dorsal fin pterygiophore and
the posterior anal fin pterygiophore, since
Cope (1875) used the number of central
elements between these points as a specific
character for A. "dictyocephala" (USNM

26 Bulletin Museum of Comparative Zoology, Vol. 146, No. 1

Fig. 10. Am'ia calva (648 mm SL) caudal: A, Whitehouse (1910) and Lund's (1967) definition of first ural; B,
Nybelin's (1963) definition of first ural; C, first fused ural.

3992 ) . The range for the number of centra
in this region of Recent A. calva is 33 to 37.
There is considerable variation in total
number of centi'a {i.e., segments) in Recent
A. calva (81-90), which may pose a prob-
lem in comparing specific vertebrae. Thus
in two A. calva, for example, the eightieth
vertebral segment of one individual might
not correspond to the same position in the
vertebral column or even type of centrum
as the eightieth segment of the second indi-
vidual. This should be considered in any
comparisons of several A. calva individuals,
as well as in comparisons of the fossil forms,
which share this variation in vertebral seg-
ments (Table 9). Also, fusion of vertebral
elements may occur in Recent A. calva. In
some specimens, as many as five centra
were found fused together at points
throughout the vertebral column; this con-
dition was present to a lesser degree or
absent in other specimens (Tables 10-12).
These fused centra also occur in the fossil
forms, as in A. uintaensis (YPM 6244). The

actual number of such fused centra can
often be established only by counting ex-
ternal features such as basapophyses, neural
facets, aortal facets, or haemal facets.

Romer and Fryxell's ( 1928 ) study of
"Paramiatus gurleyi" is the only pubHshed
description of a complete articulated fossil
amiid. They distinguished this form from
the Recent species by the supposed pres-
ence of a deeper body, and also noted that
the number of centra was considerably less
than in A. calva. The vertebral column is
completely preserved, so that it is possible
to obtain an accurate count of the vertebrae
(Plate IB). "Paramiatus p,iirleyi" has 67
vertebral segments in contrast to the mean
of 86 in A. calva (Table 9). Osborn et al.
(1878) described A. sciitata (PU 10172) on
the basis of a specimen lacking a caudal
fin (Plate 4). Since the specimen is other-
wise complete, they were able to estimate
that their specimen had 82 vertebral seg-
ments.

Cope (1875) described A. "clictyoce-

Fossil Amiids • Boreske 27

Table 9.

Comparison of

VERTEBRAL

CHARACTERS IN

RECENT AND

FOSSIL AMUDS

Number of

Centra

between

Number

Dorsal-Fin

of Mono-

of Diplo-

of Ural

Pterygiophore

Total

Number of

spondylous

Centra

and Posterior

Number of

Trunk

Caudal

with Fused

Anal-Fin

Centra""

Centra

Hypurals

Pterygiophore"

Recent

Amia calva (20)

Wis. & Mich.

81-90

36-38

24-26

14-17

7-9

33-37

mean

= 85.8

= 37.3

= 25.2

= 16.2

= 8.3

= 35.5

Oligocene

A. scutata

PU 10172

83***

yooo

A. scutata

UMMP V-57431

81***

24***

15***

Y* 00

A. "dicttjoccphala"

USNM 3992*

—

Eocene

"Paramiatus ^urlcyi

•*

FMNH 2201*

Amia uintaensis

PU 13865

Amia uintaensis

AMNH 785

—

A. fragosa

MCZ 5341

A. kehreri

BMNH P33480

62***

6***

types.

o --

"" ^ including diplospondylous units (as one),
"o" Est.

pluild" from a specimen (USNM 3992) in
which only the mid-body region was pre-
served. He felt that the number of ver-
tebrae between the anterior dorsal fin
pterygiophore and the posterior anal fin
pterygiophore had ta.xonomic significance.
A comparison of this specimen with Recent
A. calva showed that the v'ertebral count of
this region is essentially the same in both
species. This character is therefore not
useful in distinguishing this species from
the Recent form or in characterizing it as
a specific taxon. The specimens of A.
scutata are within the range of A. calva in
total number of vertebrae as well as in the
number of vertebrae in the vimous cate-
gories (Table 9) . Based on the similarity of
number of vertebrae in A. scutata to that of
A. calva, it appears that the amiid vertebral

column has not changed meristically from
Oligocene to Recent.

Additional data from five undescribed
fossil amiid specimens with relatively com-
plete axial skeletons has been of consider-
able help in estimating vertebral counts
of the fossil forms. A complete specimen
of A. uintaensis from the Green River For-
mation (PU 13865) has a complete axial
skeleton (Plate 3). Interestingly, the total
number of centra (85) does not differ from
that of A. scutata or A. calva (Table 9).
The only variation is in the number of
trunk centra and the number of diplospon-
dylous caudal centra. There are fewer
trunk centra in this specimen of A. uintaen-
sis (31) than in A. scutata, which has a
mean of 36, or in A. calva, \\'hose trunk
centra are a mean of 37. A partially com-

Bulletin Museum of Comparative Zoology, Vol. 146, No. 1

plete A. uintaensis specimen (AMNH 785),
also from the Green River Formation, shows
almost the same number of diplospondylous
caudal centra as PU 13865 (20-21 respec-
tively). The lesser number of trunk centra
in both specimens of A. uintaensis is thus
offset by a greater number of diplospondy-
lous caudal centra. In comparing the verte-
bral column of A. uintaensis with that of
A. calva, A. scutata, and A. fragosa, it ap-
pears that although A. uintaensis shares the
same total number of vertebral segments
with A. scutata and A. calva, it does not
conform to their proportional division of
the column into trunk and caudal regions.
A. uintaensis has a trunk/ total-number ver-
tebral ratio of 0.365, while A. fragosa has a
ratio of 0.300 as compared to the A. calva
ratio of 0.440. Three complete specimens
referred to here as A. fragosa ("Paramiatus
gurleyi" FMNH 2022, A. kehreri BMNH
P33480, and A. fragosa MCZ 5341) have
vertebral columns that differ proportion-
ately and meristically from A. calva, A.
scutata, and A. uintaensis. A. fragosa has
significantly fewer centra than the other
fossil forms, with approximately 12 fewer
trunk vertebrae and 8 fewer monospondy-
lous caudal centra. It has approximately
the same number of diplospondylous caudal
centra as A. calva and A. scutata, with the
number of fused hypurals also essentially
the same (Table 9). Thus A. fragosa and
A. uintaensis are meristically distinct from
one another and also from A. scutata and
A. calva, suggesting that these two earlier
forms can be taxonomically separated on
vertebral meristic characters.

VERTEBRAL COLUMN OF AMIA CALVA

The existing taxonomy of many North
American fossil amiids is based primarily
on vertebral characters. Many of the spe-
cies of "Protamia," and the genus itself as
described by Leidy (1873a) from the
Bridger Formation, have been established
solely on height/ width proportions and
length (thickness), shape of the neural
and aortal facets, and various foramina of
isolated vertebrae. Fossil species of Amia

from the Bridger and Cypress Hills forma-
tions have also been defined on character-
states of isolated vertebrae. In order to
analyze this usage, variation in vertebral
character-states of A. calva has been studied.
The axial skeleton of Recent Amia calva
is relatively well known. It is one of the few
modem forms that have diplospondylous
vertebral centra posteriorly, a condition
that, according to Schaelfer (1967), func-
tionally increases the flexibility of the pos-
terior part of the body. Shufeldt (1885)
was one of the first to describe the verte-
brae of Amia, and Hay's ( 1895 ) well-
known work on the vertebral column of
Amia provides a relatively complete and
informative description of the axial skele-
ton, as well as one of the first discussions
of intracolumnar variation of the centra.
Hay observed some gradual changes in
centrum proportions, and in the position of
the neural and aortal facets.

Vertebral Features

Dorsal and ventral facets, basapophyses,
foramina, and ridges on the centra have
been used as diagnostic characters in the
taxonomy of fossil amiids. There are three
types of paired facets on the vertebrae:
dorsal neural facets for the neural arches,
ventral aortal facets for the aortal supports,
and haemal facets for the haemal arches.

Neural facets. The neural facets are
shallow depressions under the neural arch
bases, which in life are filled with cartilage.
Cartilage is present between the centrum
and its associated neural arch. Some speci-
mens of A. calva have much deeper facets,
with a small ossified ridge built up on the
borders. These neural facets occur in pairs
on the dorsal surface of both trunk and
caudal vertebrae, and between the two
facets lies a groove that partially receives
the spinal cord.

According to Hay (1895: 7-9), there is a
marked anteroposterior change in the posi-
tion of the neural facets. He contended
that at the anteriormost end of the vertebral
column the neural arch bases occur be-
tween two vertebrae and rest equally on

Fossil Amiids • Boraske

VENTRAL

dUJZ^

odZlL^

35
36

37
38

4II>

DORSAL

Fig. 11. Configuration of aortal facets (as) and neural facets (ns) on trunk and anterior caudal vertebrae of
Amio calva (339 mm SL).

both; going posteriorly the.se bases shift
gradually backward. He also observed that
there is a change in the spacing of the
neural iuches; they are close together in the
anterior trinik region and more widely
spaced posteriorly. Hay is correct in regard
to the change in spacing of the neural
arches, but he is not altogether correct in
his description of the change in position
of these arches in relation to the centra.
An examination of the Wisconsin A. calva
sample showed that, after the first few an-
teriormost centra and corresponding neural
arches, the middle of the neural arches is
situated at the juncture between the centra.
This placement continues along the axial
column until the first diplospondylous ver-

tebra occurs. At this point, the next five
to seven neural arches are found aligned to
the middle of each of the corresponding
centra, after which the arches appear to
move forward slightly and correspond ir-
regularly to the vertebral bodies.

The configuration of the neural facets
themselves varies in the trunk region of the
vertebral column of A. calva. The neural
arches in the anterior trunk region are
thicker and wider than those in the more
posterior trunk region which have become
more flattened and elongated. The shape
of the neural facets reflects this trend ( Fig.
11). After the first two ccMitra, the facets
assume an hourglass shape, being narrower
in the middle and broader at each end.

30 Bulletin Museum of Comparative Zoology, Vol. 146, No. 1

This can be related to the fact that the
neural arches are situated at the juncture of
two centra so that each neural facet sup-
ports the anterior and posterior halves of
two different neural arches, whose bases
are narrow at the extremity and thick in
the center. Although the neural facets in
any given specimen of A. calva conform to
this general trend, the individual configura-
tion of the facets varies slightly. Given this
variation in shape of the neural facets, it is
useless to attempt characterization of the
vertebral column of any amiid species based
on configuration of neural facets.

Aortal and haemal facets. On the ven-
tral side of the trunk vertebrae are two thin
cartilaginous projections that are located
on either side of the dorsal aortal supports.
When the skeleton is dried, these projec-
tions leave marked depressions, which, like
the neural facets, vary gradually from the
first anterior vertebra to the last few trunk
vertebrae; at this point the aortal facets
coalesce with the basapophyses (Fig. 11).
The point where these two elements are
completely merged marks the termination
of the trunk centra, and the next centrum
is that of the first caudal vertebra. These
structures, which were derived anteriorly
from the basapophyses and aortal facets,
here become the haemal facets.

The first pair of aortal facets is very small
and ovoid. The next few centra bear aortal
facets that, as Hay (1895) also observed,
are circular. The following aortal facets
become successively elongated, until, with
the tenth or twelfth vertebra, these facets
have evolved into a long pair of slits, usu-
ally narrower at the midpoint. Posteriorly,
these slitlike aortal facets remain basically
the same shape until, at the end of the
dorsal trunk region, they merge with the
basapophyses to form haemal facets. Hay
(1895: 54-57) states that the cartilaginous
aortal supports penetrate deeply into the
centra of younger individuals, while in
older specimens they rest superficially on
the centra. The aortal facets are deeper
and more distinct than the neural facets.

Beginning with approximately the tenth

or twelfth vertebra, the slit-shaped aortal
facets are vertically situated on either side
of an indentation that contains the aorta
(Fig. 11). The first four centra have
thicker and shorter supports with relatively
little or no space between them. The aorta
lies ventrally under the basioccipital,
which bears aortal supports whose facets
are of the same shape as the first four centra
(Estes and Berberian, 1969, fig. 2B for
A. frafi^osa). The aortal facets of the first
eight vertebrae are different from all other
trunk vertebrae, whose shape, as mentioned
above, is basically an elongated slit. These
aortal supports are thus helpful in dis-
tinguishing the first eight or so vertebrae
from the remainder of the trunk centra in
disarticulated specimens (Fig. 11).

Haemal facets. The haemal facets,
which contain a cartilaginous layer be-
tween the centrum and the haemal arches,
are nearly rectangular-shaped pairs that do
not vary along the caudal portion of the
vertebral column until the first fused urals.
The furrow or indentation that lies between
the aortal facets in the trunk centra con-
tinues in the caudal region between the
paired haemal facets, although it gradually
decreases in width and depth. Unlike the
neural facets, the haemal facets are out-
lined by an ossified border, which can be
helpful in distinguishing dorsal from ventrid
surfaces in disarticulated caudal vertebrae.

Since the ossified walls are tilted 20
degrees posteriorly to accommodate the
haemal arches, which articulate with the
cartilaginous layer diagonally rather than
laterally, those borders are also useful in
determining the anteroposterior orientation
of the centrum.

Basapophyses. Amia trunk centra are
distinguished from the caudal vertebrae by
their having prominent paired processes,
which ha\e been called transverse pro-
cesses, parapophyses, or diapophyses. I
follow the terminology of Bolk et al. ( 1936),
wherein they designate these structures,
which are the processes for pleural ribs, as
basapophyses ( "basalstiimpfe" ) . The first
centrum often lacks these basapophyses

Fossil Amiids • Boreske 31

<k^

Fig. 12. Shape of selected trunk and caudal vertebrae in Amia calva (339 mm SL).

(Tables 10-12), which are always present
on the succeeding centra and progressively
become longer until appro.ximately the
twelfth (Fig. 12). The basapophyses are
approximately the same length between
the twelfth and tlie thirty-second centra,
from which point they begin to diminish
gradually in length until the last trunk
centrum, where they coalesce with the
aortal facets. The lengths of the basapo-
physes were not individually measured;
this data would be of little practical use in
a comparison of Recent and fossil material
since these relatively fragile structures are
rarely preserved intact in fossils. The distal
end of each basapophysis is attached to a
pleural rib by means of cartilage. The

proximal ends of the basapophyses are
ankylosed to the ventral half of the verte-
bral body. These paired processes are solid
cyHnders (hollow at the tips) that are
slightly Battened dorsoventrally. Each pair
of basapophyses may not always be of equal
length or diameter, but they are extremely
regular in position. They form two con-
tinuous and symmetrical lines that gradu-
ally come closer together until the last
trunk centrum, where they are separated
only by aortal supports.

An important aspect of the basapophyses
in A. calva is the angle between each indi-
vidual pair which gradually decreases pos-
teriorly. Since the angl(> ])etween the
basapophyses is generally still available in

32 Bulletin Museum, of Comparative Zoology, Vol. 146, No. 1

fossil forms, even in those with broken basa-
pophyses, it is used here as a basis of
comparison between the Recent and fossi!
forms. Since the angles steadily decrease
posteriorly along the vertebral column
(Figs. 12, 14), they are also useful in
orienting disarticulated centra to approxi-
mate position along the column. Although
there is individual variation in these angles
(Tables 10-12), they are nevertheless con-
sistent enough to help in determining the
general position in the column of any
single trunk centrum. The range of angles
extends from approximately 180 degrees
anteriorly to 45 degrees posteriorly. Since
the three A. calva specimens studied were
of varying sizes ( 193 mm SL, 382 mm SL,
and 423 mm SL), it would appear that there
is no significant change in the angles with
increasing size or age of the fish (Fig. 14).
Although this transition is not perfectly
linear, the angles are always decreasing
posteriorly, and at least in the specimens I
measured, there was never an instance of
an angle's measurement being greater than
that of the preceding centrum. The angle
decrease occurs at a fairly constant rate
until approximately the thirtieth trunk ver-
tebra, at which point the rate of decrease
of the angles is much accelerated ( Fig. 14 ) .
The angle of the basapophyses is thus a
reliable parameter in identifying the gen-
eral position of isolated trunk centra.

Foramina, bone ridges, and first centrum.
The trunk centra of A. calva have lateral
foramina that, although lacking the uni-
formity of the neural and aortal facets,
occur in irregular, distinct paired linear
patterns. The foramina of the tnmk and
caudal vertebrae transmit numerous small
blood vessels.

On the lateral surfaces perpendicular to
the anterior and posterior articular surfaces
of the individual centra are prominent bone
ridges. These bone ridges add support to
the arch anlagen, and also help unify the
anlagen into a sturdy, functional vertebral
body (Schaeffer, 1967). Externally, these
bone ridges are not as regular as they are
internally, although they still lie antero-

posteriorly in the lateral and ventral regions
and extend vertically along the basapo-
physes. They are also quite prominent in
the notochordal furrow. Such bone ridges
are not a unique feature of A. calva, and are
common in teleosts.

The centra in A. calva are amphicoelous.
The first four to six centra differ from all
corresponding centra by having the anterior
articular surface more convex than concave.
The first centrum in nearly all specimens
observed lacked basapophyses, and should
therefore be considered a minor taxonomic
character since first centra do occasionally
occur with very small basapophyses. The

Angle of
basapophyses

Fig. 13. Index to the measurements used, superimposed
upon an outline drawing of an Amia calva vertebra.

Fossil Amiids • Boreske 33

^. co/yo verfebrol lengths

1( 24 n <0 <l Si (4 72 10 II

Verl. no.

A.calvo bosopophyseol angles

12 It 21

Vert. ■•.

---^

12 41 41

Vtrt. ■•.

v^i

•-v.'

A. uinfaensis

NEIGIT

mom

I i ii H W

—» — jr

Vtrt. M.

Ti n w

Fig. 14. Intracolumnar variation in the angle of basapophyses, length, height, and width of vertebrae in Re-
cent Ami'o co/vo (A = 423 mm SL; B ^ 382 mm SL; C =: 193 mm SL). Intracolumnar variation in height and width
of trunk and caudal vertebrae in A. uin/oensi's. Vertebral column model based on first six centra from PL) 10101
and fifty-nine centra from CM 25362; missing caudal centra have been interpolated and inferred based on PL) 13865.
The first anterior centra (PL) 10101) were larger specimens and thus the anterior region of the trunk vertebral column
model is "out-of-phase." Vertical lines = last trunk centrum.

ovoid shape of the aortal facets is a constant
feature of all first four to six centra ob-
served.

Vertebral Dimensions

A superficial but often-used character for
diagnosing fossil amiid species has been the
shape of the centrum. Descriptions for
Amia whiteavesiana, A. macrospondyla, A.
exilis, A. elegans, A. depressus, A. newher-
rianus, Protamia symphysis, P. media, P.
gracilis, P. uintaensis, P. plicatus, P. cor-
sonii, and P. laevis include centrum mea-
surements for height, width, and length
(thickness), as well as qualitative descrip-

tions of the fonii and proportions of the
centrum. Because isolated amiid vertebrae
have often been the only anatomical mate-
rial found in the fossil record, the original
diagnoses were obviovisly limited in that a
great deal of emphasis was placed on the
vertebral centrum. In considering a single
centrum shape as diagnostic for an amiid
species, early authors implicitly assumed
the vertebral column to be static, with no
physical change or variation among the
centra other than regional. Many new
species were therefore described solely on
variation in shape from other known amiid
types.

34 Bulletin Museum of Comparative Zoology, Vol. 146, No. 1

Table 10. iNiTtAcoLUMNAR variation in height, width, length, and angle of basapophyses

OF VERTEBRAE IN RECENT Amiu colva ( 193 mm SL)

Vert.
Cent.

Height
(mm)

Width
(mm )

Length
(mm)

Angle of
Basapophyses
( Degrees )

Vert.
Cent.

Height
(mm)

Width

(mm)

Length
(mm)

3.55

4.35

1.40

2.75

1.35

3.65

4.20

1.70

176

2.60

2.65

1.15

3.70

4.25

1.65

159

2.65

.90

3.40

4.15

1.70

159

2.55

2.60

1.30

3.25

3.90

2.05

153

2.50

2.45

1.35

3.15

3.65

2.00

145

2.50

1.20

3.00

3.65

1.90

140

2.45

1.15

3.05

3.60

2.00

138

2.50

2.45

1.20

2.95

3.55

2.20

134

2.30

2.40

1.15

3.00

3.55

2.30

130

2.40

2.30

1.15

3.05

3.40

2.10

122

2.30

2.20

1.15

2.90

3.65

2.25

118

2.30

2.05

1.10

2.75

3.25

2.10

103

2.25

2.15

1.00

3.00

3.50

2.15

2.30

2.15

.80

3.05

3.35

2.20

2.30

2.10

.85

3.00

2.90

2.25

2.05

1.00

2.95

3.05

2.15

2.05

1.90

.90

3.00

2.05

1.87

1.75

.90

3.10

3.08

2.25

1.65

.80'

3.15

3.00

2.15

1.50

.68

3.25

3.00

2.10

1.25

.53

2.80

2.90

2.25

.80

.50

2.75

2.90

1.50

" Fused.

Table 11. Intracolumnar variation in height, vi^idth, length, and angle of basapophyses

OF vertebrae in recent Amia calva (382 mm SL)

Vert.
Cent.

Height

(mm)

Width
(mm)

Length
(mm)

Angle of

Basapophyses

( Degrees )

Vert.
Cent.

Height

(mm)

Width

(mm)

Length
(mm)

6.30

7.21

1.90

180

5.50

5.05

3.50

6.25

7.41

2.45

173

5.15

4.65

2.40

6.00

7.50

2.50

161

5.00

4.80

2.35

5.95

7.25

2.85

161

4.90

4.60

2.37

5.82

7.10

2.95

159

4.70

4.45

2.50

5.56

6.90

3.05

154

4.70

4.45

1.90

5.50

6.65

3.45

143

4.65

4.55

2.00

5.50

6.25

3.45

140

4.70

4.45

2.10

5.50

6.25

3.45

133

4.80

4.10

1.90'

5.35

6.30

3.50

125

4.40

4.45

2.00

5.25

6.25

3.55

120

4.80

4.45

2.12'

5.30

6.25

3.55

116

4.50

4.10

2.12'

5.30

6.25

3.65

113

4.40

3.85

2.00

5.20

6.20

3.85

110

4.00

3.75

2.00

5.35

6.15

3.70

110

4.05

3.70

1.87

5.25

6.00

3.60

105

3.75

3.50

1.75

5.35

5.80

3.85

100

3.50

3.35

1.50

5.45

5.50

3.90

3.20

3.15

1.47

5.50

5.15

3.50

2.65

1.65

5.75

5.10

3.55

2.00

1.40

5.70

5.07

3.60

1.95

1.20

5.65

5.05

3.65

1.80

1.12

• Fused.

Fossil Amiids • Boreske 35

Table 12. Intracolumnar variation in height, width, length, and angle of basapophyses

OF vertebrae in recent Amia calva (423 mm SL)

Vert.
Cent.

Height
(mm)

Width

(mm)

Length
(mm)

Angle of

Basapophyses

( Degrees )

Vert.
Cent.

Height
(mm )

Width
(mm)

Length
(mm)

8.25

9.25

2.80

6.45

6.40

3.45

8.00

9.15

3.35

172

6.40

6.21

3.05

7.80

9.05

3.40

156

6.20

6.00

3.05

7.55

9.20

3.45

150

6.10

5.85

3.85

7.45

8.80

3.45

142

6.25

5.85

3.80

7.40

8.80

3.75

140

6.10

5.85

2.95

7.25

8.75

3.85

132

5.95

5.90

2.70

7.25

8.75

4.00

126

5.80

5.85

2.65

7.35

8.70

4.05

119

5.80

5.65

2.60

7.15

8.50

4.15

117

5.65

5.70

2.80

7.15

8.50

4.10

114

5.60

5.30

2.70

7.15

8.40

4.00

112

5.55

5.20

2.45

7.35

8.35

4.37

103

5.50

5.05

2.50

7.45

8.30

4.75

103

5.30

4.95

2.35

7.30

8.20

4.75

5.30

4.90

2.20

7.47

8.10

4.70

5.20

4.80

2.10

7.25

7.95

4.95

5.05

4.65

1.85

7.35

7.45

5.00

4.50

2.10*

7.35

6.90

5.00

4.75

4.45

2.25

7.50

6.70

4.95

4.15

1.55

7.65

6.50

4.90

3.25

1.50

7.60

6.50

4.50

2.50

1.50

7.40

6.20

4.35

2.00

1.45

<• Fused.

One of the detailed studies on intra-
columnar vertebral variation is Hoffstetter
and Case's ( 1969 ) work on the vertebral
column of snakes. Measuring individual
centra in sequence, they plotted this varia-
tion; similar graphs are used here (Fig.
14 ) . Three specimens of A. calva ( 193
mm SL, 382 mm SL, and 423 mm SL ) were
dissected and the individual vertebral
dimensions measured to determine verte-
bral variation.

Length. Centrum length (thickness)
was measured anteroposteriorly at the mid-
line, above the basapophyses (Fig. 13).
It was necessary to establish such a control
for this measurement because of the varia-
tion in thickness within each centrum. In
the trunk region, the centra are thickest
ventrally at the neural and dorsal facets.
The caudal vertebrae follow a similar pat-
tern, being slightly thicker ventrally and
dorsally, and thinner laterally. Every
second centrum was measured for length
(Tables 10-12). There is a distinct intra-

columnar variation for this measurement,
although the difference in length between
consecutive vertebrae is usually small.
There is also a general, if somewhat ir-
regular, pattern in centrum length in A.
calva (Fig. 14). The first two or three ver-
tebrae of each A. calva specimen are rela-
tively thin. These are followed by centra
that gradually increase in length until ap-
proximately the last trunk centrum at the
midbody. At this point there is a general
trend again towards thinner vertebrae, al-
though this pattern is erratic, particularly
between the fiftieth and the fifty-fourth
centra, where the thickness is suddenly in-
creased and then decreases. This sudden
change here in vertebral thickness occurs
directly above the midline of the anal fin.
The shortest vertebrae are the fused urals.
Height. Centrum height was taken dor-
soventrally at the midline, between the
aortal and neural facets (Fig. 13). Every
second centrum was measured up to the
fused urals, in which an accurate measure-

Bulletin Museum of Comparative Zoology, Vol. 146, No. 1

Table 13. Intracolumnar variation in height, width, length, and angle of basapophyses

OF VERTEBRAE IN Atnla uintaensis CM 25362

Vert.
Cent.

Height
(mm)

Width
(mm )

Length

( mm )

Angle of

Basapophyses

( Degrees )

Vert.
Cent.

Height
(mm)

Width

( mm )

Length
( mm )

24.5

32.8

9.8

179

19.5

7.5

25.0

32.5

9.8

178

19.8

18.8

7.0

26.8

32.0

9.5

177

19.8

18.5

6.9

26.5

32.1

9.5

176

20.0

17.5

6.5

25.5

32.8

9.0

174

—

6.5

26.0

31.0

9.1

172

19.5

18.0

6.5

25.4

31.1

9.5

164

18.5

17.5

6.0

27.5

31.0

9.5

160

18.5

17.5

6.2

25.5

30.0

10.0

156.5

17.5

6.0

26.0

30.5

9.8

154

17.2

16.0

6.0

26.5

31.5

9.5

153

17.0

16.0

6.0

27.0

30.5

9.8

149

16.5

16.0

5.5

25.1

29.5

9.5

143

16.5

—

6.0

24.5

29.5

9.5

132

16.2

15.5

5.5

24.0

30.0

9.5

122

16.0

15.0

5.8

24.5

28.2

9.0

117

16.0

15.5

5.0

24.0

29.0

9.5

110

15.5

14.0

5.5

24.0

28.5

9.0

102

15.2

14.5

5.2

24.0

26.5

8.5

102

15.5

13.2

6.0

24.5

27.0

10.0

15.0

12.2

5.5

23.5

25.0

11.0

14.5

12.0

6.5

23.5

25.0

11.0

14.5

12.5

5.5

23.2

25.0

12.0

14.0

11.5

5.2

23.0

22.5

11.0

14.0

5.0

25.0

22.5

11.0

13.5

11.5

5.0

—

11.0

13.5

10.5

4.5

—

10.0

4.5

23.0

19.5

9.5

12.0

9.0

4.5

22.5

9.5

8.5

7.0

4.2

21.0

20.0

7.5

ment would be obscured by the fusion of
the hypurals. This series of measurements
shows a basic pattern that is similar for
each of the individuals studied, although
there is less intracolumnar variation in the
height than in the length measurements
(Tables 10-12). The greatest height gen-
erally occurs at the anteriormost region of
the column, then decreases slightly until
the midtrunk region (Fig. 14). At this
point the height gradually increases until
it peaks at the end of the trunk region and
the beginning of the caudal section, after
which it decreases again toward the caudal
region (Fig. 14).

Width. The width measurements were
taken perpendicular to the height measure-
ments, at the widest section of the centrum
(Fig. 13). This dimension has a greater

linear slope than the length and height
dimensions ( Fig. 14 ) , which follow a more
bell-shaped curve. There is a greater varia-
tion within the vertebral column for width
dimensions (Fig. 14), as comparison of the
height and width slopes reveals. The great-
est intracolumnar width is always at the
anteriormost portion of the trunk region,
after which this dimension gradually de-
creases. There appear to be two areas
where the rate of decrease is greater, these
being at the terminus of the trunk region
and at the first fused ural.

Height/ width ratio. The centrum height/
width ratio has been a commonly used diag-
nostic character in amiid taxonomy. Hay
(1895: 7) correctly noted that the trunk
vertebrae are somewhat broader than high
( Fig. 14 ) , and at the terminus of the trunk

Fossil Amiids • Boreske

region the centra are nearly circular. The
proportions tend to be reversed in tlie
eandal region, however, with the height
generally exceeding the width, although to
a lesser degree than the proportional dif-
ference in the trunk region. The basic trend
in shape through the vertebral column is
tluis a marked horizontally elliptical cen-
trum approaching a progressively circular
one, which then again becomes slightly
vertically elliptical. Thus there is quite a
variation in the centrum shape throughout
the axiiil column, so that no one shape or
ratio of dimensions could reasonably be
considered diagnostic for all the centra of
the vertebral column.

VALID NORTH AMERICAN FOSSIL
GENERA AND SPECIES

Amia fragosa (Jordan, 1927)

Kitidleia fragosa Jordan, 1927: 145.
SUjlomylcodon lacus Ru.ssell, 1928a: 103.
Paramiatus giirleyi Ronier and Fryxell, 1928: 519.

Holotype. NMC 8533e, anterior portion
of right dentary.

Paratypes. NMC 8534a-d, f-n. (a), left
operculum fragment; ( b ) , cranial fragment;
(c), anterior portion of left dentary; (d),
anterior portion of right dentary; (f-g),
coronoid with two styliform teeth preserved;
(h-i), vomer without teeth preserved; (j),
styHform tooth fragment; (k-1), posterior
portion of right dentary; ( m ) , left maxilla;
(n), anterior portion of right dentary.

Ttjpe locality and horizon. Rumsey, Al-
berta. East half of section 31, T 34 S, R 21
W, Rumsey Quadrangle, Alberta; Edmon-
ton Formation.

Age range. Campanian (Late Creta-
ceous) to Bridgerian (Middle Eocene).

Hypodiii^m. Cretaceous. Oldman For-
mation, Alberta: AMNH 5934, palatal frag-
ments with styliform teeth; AMNH 5935,
operculum and dentary. "Mesaverde" For-
mation, Wyoming: AMNH 5932, dentary
and numerous coronoid teeth; AMNH 5933,
vertebrae. Judith River Formation, Mon-
tana: AMNH 10109, left vomer bearing
styHform teeth; AMNH 10110, dentary frag-

ments, vertebrae, and skull elements. Ed-
monton Formation, Alberta: ROM .3064,
coronoid teeth; ROM 3065, dentaries, verte-
brae, and cranial fragments; UA 5398-5507,
articulated and disarticulated specimens
(see O'Brien, 1969 for identifications).
Lance Formation, Wyoming: AMNH 9316,
pterotic; AMNH 9315, operculum; CM
25363, dentaries; PU 17013, dentaries;
UCMP 54013-54015, 54017, 54019, 54021-
54030, 54035-54038, 54040-54056, 54059-
54069, 54070-54120, 54141-54167, 54174-
54180, 54188-54198, 54260, 54262, dis-
articulated elements (see Estes, 1964 for
identifications). Hell Creek Formation,
Montana: PU 17016, 17048, dentaries; PU
17014, coronoid teeth; PU 20554, dentary
and vertebrae; MCZ 9286-9293, 9390-9432,
9559, disarticulated elements ( see Estes and
Berberian, 1969 for identifications ) .

Paleocene. Fort Union Formation, Wy-
oming: PU 17115, coronoid teeth; PU 17126,
coronoid teeth and vertebrae; PU 17117,
dentary and maxilla; PU 21525, portion of
cranial roof with associated dentaries; PU
20523, dentary and coronoid teeth; PU
21174, vertebrae. Paskapoo Formation, Al-
berta: UA 131, dentary, numerous tooth
plates, and vertebrae. Tongue River For-
mation, Montana: PU 20577, vertebrae,
premaxillary fragment, and coronoid teeth;
PU 20578, basioccipital and vertebrae; PU
17068, vertebra and dentary fragment.
Melville Formation, Montana: AMNH 2635,
cranial elements and associated dentaries.
Tullock Formation, Montana: PU 17069,
vomers.

Eocene. Will wood Formation, Wyo-
ming: MCZ 9264, nearly complete skull;
PU 18780, tooth plate; PU 21175, dentary
fragment and coronoid teeth; PU 16756,
dentary and cranial fragments; PU 17649,
anterior portion of skull; PU 21173, skull
fragments and vertebrae; PU 13261-13262,
cranial fragments and coronoid tc>eth.
Golden Valley Formation, North Dakota:
PU 18567, coronoid teeth and vertebrae.
Wasatch Formation, Wyoming: PU 13260,
tooth plates; PU 13259, cranial fragments
and dentaries. Bridger Formation Wyo-

Fig. 15. A, Amia calva. Recent, Wisconsin; above, lateral, and below, dorsal views of skull. B, Amia scutafa. Early
and Middle Oligocene; above, lateral, and below, dorsal views of skull (sensory canal system and pit-lines are not
known since skull elements are in articulation). C, Amio fragosa, Late Cretaceous to Middle Eocene; above, lateral,
and below, dorsal views of skull (sensory canal system and pit-lines after Estes, 1964). D, Amia uinfaensis, Paleocene
to Early Oligocene; above, lateral, and below, dorsal views of skull (sensory canal system is only known in the
mandible, operculum, nasal, lacrimal, antorbital, extrascapular, and suprascapular, all of which conform with those
of A. calva).

Abbreviations: a, angular; ao, antorbital; br, branchiostegal rays; d, dentary; ds, dermosphenotic; es, extrascapular;
fr, frontal; io, interoperculum; io- io'' io'* io'^', infraorbital series (suborbitals & postorbifals); la, lacrimal; m,
maxilla; n, nasal; op, operculum; p, preoperculum; pa, parietal; pt, pterotic; r, rostral (ethmoid); s, suprascapular; so,
surangular; sm, supramaxilla; so, suboperculum. Dotted lines indicate the sensory canal system; dashed lines indi-
cate pit-lines.

Fossil Amiids • Boreske

ining: YPM 6245, vomer and cranial frag-
mcMiVs; YPM 6246, vertebrae; YPM 6247,
dentary; YPM 6248, vertebra and cranial
fragments; YPM 6254, verte]:)rae, basioccip-
ital, vomer; YPM 6261, left opercnlnm;
ANSP 5630, vertebra. Green RivcT Forma-
tion, Wyoming: MCZ 5341, FMNH 2201,
complete specimens.

Known distribution. North Dakota,
Wyoming, Montana, and Alberta.

Revised diapwsis. Vertebral colimm
with significantly fewer total ctMitra (65
mean) than the other species, with approxi-
mately twelve fewer trnnk vertebrae (25
mean) and eight fewer monospondylous
caudal centra ( 17 mean ) . Distance be-
tween anal fin insertion and the end of the
vertebral column relatively short, with
dorsal fin terminating close to caudal fin.
Caudal lepidotrichia 19-20 rather than 23-
27. Ascending processes of parasphenoid
perpendicular to the main anteroposterior
parasphenoid axis; more posterior place-
ment of parasphenoid tooth-patch. Pari-
etals squared in outline. Marginal teeth
simple pointed cones, palatal teeth usually
stout styliform crushers. Supraorbital sen-
sory canal not entering parietal. Excava-
tion of orbital notch in frontal relatively
larger. Dentary with additional horizontal
shelf of coronoid articulation surface adja-
cent to lingual border of alveolar ridge;
coronoid articulation surface extensive,
overlapping ventral half of ramus; dentary
with pronoimced arch rather than gradual
curve in ventral outline. Greatest known
standard-length 510 mm.

Introduction

Jordan (1927) described Kindleia fra^osa
as a new genus of cichlid fish from the Late
Cretaceous Edmonton Formation of Al-
berta. This tentative placement of Kindleia
within the Cichlidae was largely the result
of his misinteipreting the splenial tooth
plates for fused lower pharyngeal bones
(Estes, 1964). One month later, Russell
( 1928a) independently published a descrip-
tion of Stylomyleodon lacus, a new fossil
amiid from the Late Paleocene Paskapoo

Formation of Alberta, and referred other
specimens from the Edmonton Formation
of Alberta to the same species. His descrip-
tion also included a dentary and palatal
teeth modified tor crushing. His relegation
of the genus to the Amiidae was based on
a correct inteipretation of the "splenial"
(= coronoid) tooth plates (Estes, 1964).
He suggested a relationship of Stylonujle-
odon to Platacodon nanus ( at that time
erroneously considered an amiid; see Estes,
1964) with the essential difference being
hemispherical rather than Hattened tooth
crowns.

Jordan later ( 1928 ) noted the similarity
of the two genera Kindleia and Stylomyle-
odon and asserted the prior claim of his
name Kindleia. Although he made no com-
ment on Russell's attributing Stylomyleodon
to the Amiidae, he rejected Russell's com-
parison of that genus with Platacodon on
the basis of Marsh's earlier conviction that
the latter was mammalian. In reply to
Jordan, Russell ( 1928b ) defended the valid-
ity of his genus on the supposition that its
dentary was distinct from that of Kindleia,
although he did agree on the similarity of
teeth and jaw fragments of the two genera.
Russell (1929) further attempted to vali-
date Stylomijleodon as a genus by com-
paring his type with new specimens
collected by Princeton University. This new
material confirmed his association of the
maxilla-dentary and palatine-coronoid den-
titions, and also substantiated his interpreta-
tion of Stylomyleodon as an amiid in which
the coronoid teeth were specialized for
crushing. He also admitted that there was
insufficient Platacodon material to deter-
mine any conclusive similariti(\s with Stylo-
myleodon, but, r(>f(Mring to Hatcher's (1900,
1901 ) work, did insist that Platacodon was
a fish. Simpson (1937) reported finding
additional specimens of Stylomyleodon Rus-
sell in the Fort Union Formation at Crazy
Mountain Field sites of Montana.

Estes ( 1964 ) , from his studies of amiid
material from the Lance Formation of Wyo-
ming, observed that whereas the type
dentary referred by Russell to Stylomyleo-

Bulletin Museum of Comparative Zoology, Vol. 146, No. 1

don was the pcsterior portion of an amiid
dentary, Jordon's type was the anterior
portion. From this fact he confirmed the
synonymy of Stylomtjleodon with the genus
Kindleia, at that time beUeving that it was
generically separated from Amia. Janot
( 1967 ) agreed with Estes on the synonymy
of Stylomijleodon with Kindleia, but did not
find sufficient cause to distinguish Kindleia
generically from A7?j/fl. Russell (1967) con-
tinued to leave the nomenclatural problem
of Stylomyleodon-Kindleia unsettled. Estes
and Berberian ( 1969) studied material from
the Late Cretaceous Hell Creek Formation
of Montana and confirmed Janot's proposi-
tion that Kindleia is a synonym of Amia.
They also suggested the possibility of
synonymy of A. fra<i,osa with A. keJireri
(Middle Eocene, Germany), A. russelli
(Late Paleocene, France), A. munieri
(Early Oligocene, France), and Paromiatus
gurleyi (Early Eocene, Wyoming), but
postponed formal synonymy of A. jrag,osa
with the prior name A. kehreri (Andreae,
1892), pending more detailed study of
Early and Middle Cenozoic specimens
from Europe. Estes and Berberian (1969:
10) concluded that the minor variations
that separated A. fra<^osa and its related
forms in Europe from A. calva are "super-
ficial and essentially primitive," and indi-
cated that the group might be close to the
ancestry of the Recent species A. calva.

A nearly complete skull from the Eocene
Willwood Formation of Wyoming (Fig.
16), two axial skeletons from the Eocene
Green River Formation of Wyoming ( Plate
1 ) , and a sample of disarticulated elements
from the Late Cretaceous, Paleocene, and
Early Eocene have yielded more informa-
tion on the osteology of Amia fragosa. Estes
(1964), O'Brien (1969), and Estes and
Berberian (1969) have studied this species
in detail, and I therefore discuss these speci-
mens only as they modify conclusions
reached by those studies.

Fossil Record

In addition to the stratigraphic list given
by Estes and Berberian (1969: 14, table 1)

of major freshwater deposits carrying A.
fragosa, three new localities are recorded
here: the Late Paleocene Silver Coulee
local fauna of the Fort Union Formation,
Wyoming, and the Early Eocene Willwood
and Wind River formations, Wyoming. The
luajor deposits in which remains of A.
fragosa have been found are summarized in
Table 18. Estes and Berberian (1969: 10)
state that the stratigraphic range of A.
fragosa extends from the Late Cretaceous
through at least Middle Eocene time in
North America. The earliest deposit in
which remains of A. fragosa have been
found is the Late Cretaceous (Campanian)
Oldman Formation of Alberta, and the
latest deposit is the Middle Eocene (Brid-
gerian) Bridger Formation of Wyoming.
Cavender (1968: 128), however, de-
scribes Amia scales from the Late Eocene
( Duchesnean ) Clarno Formation of Ore-
gon. Although these small scales (approxi-
mately 2 mm in length) are not as robust
as those of A. fragosa, they are more ossified
than A. scutata and A. calva scales. These
scales, along with the scales from the Horse-
fly River Beds of British Columbia (UMMP
collections) cannot at present be identified
as to species. They are best referable to
Amia sp., since no identifiable A. fragosa
elements have been found later than
Bridgerian and no specific character-states
for scales of A7nia have yet been deter-
mined.

Description

Neurocranium. Estes (1964: 29) stated
that the greater length of the basioccipital
and the presence of a second pair of aortal
supports in Amia calva indicated that the
basioccipital posterior to the spinal (inter-
vertebral) arterial foramina included only
one fused vertebra in Amia fragosa instead
of the two found in A. calva. O'Brien
(1969: 42) observed a similar condition in
two complete A. fragosa specimens from
the Edmonton Formation of Alberta. Estes
and Berberian (1969: 2-3) found nine
basioccipitals with one fused vertebra and
eleven with two fused vertebrae from the

Fossil Amiids • Boreske 41

Fig. 16. Amia fragosa MCZ 9264, Early Eocene, Willwood Formation, Wyoming; A, dorsal; B, ventrol.

Hell Creek Formation of Montana and in-
terpreted this as a variation in A. frci'^osa
not observed in the Lance sample. Janot
( 1967 ) noted that basioccipitals of the
European Late Paleocene Amia sp. also
showed this variation. Estes and Berberian

(1969: 2) suggested that a weak tendency
for fusion of vertebrae could be correlated
with increasing size, and that such varia-
tion might possibly e.xist in the Recent
species as well. Fifty Recent A. calvo skele-
tons examined, with a size range of 100-480

42 Bullctiti Museum of Comparative Zoology, Vol. 146, No. 1

mm SL, have the first two vertebrae fused
to the basioceipital. Three articulated and
twenty-two disarticulated Eocene and
Oligocene Amia uintaensis Ijasioccipitals
all have two vertebrae fused to the basioc-
eipital. Unfortunately, in specimens of A.
fni'^osa (MCZ 9264, PU 13261) having a
visible parasphenoid, the basioceipital re-
gions are poorly preserved. There is a pos-
sibility that the Lance sample by chance
contained only specimens with one fused
vertebra since there are only six specimens
known. Until more specimens of A. jra<iosa
and A. uintaensis with intact basioccipitals
become available, it is difficult to discuss
this point further.

Fig. 17. Comparison of parasphenoids of Amio spp.:
c, Amia calva. Recent, after Janot, 1967 (c'' := dorsal, c^'
= ventral); f, A. fragosa (ventral); u, A. uinfaensis
(ventral).

The length of the A. fra<j,osa parasphe-
noid posterior to the ascending processes is
10 percent shorter and slightly wider than
in A. calva, with the ascending processes
more posterior than in the Recent species
(Fig. 17). The proportion of the length
posterior to the processes to the length
anterior to these processes (0.780) is not
as small as in A. uintaensis (0.704) or as
great as in A. calva (0.900), and, on the
basis of this small sample, A fragosa is inter-
mediate among the three species for this
character. The region posterior to the
processes also appears more convex than in
A. calva, but not as convex as in A. uijitaen-
sis. The ascending processes are almost
perpendicular to the main anteroposterior
axis of the parasphenoid. Those of A.
fragosa form an approximately 85-degree
angle with the parasphenoid axis, while the
ascending processes of both A. calva and
A. uintaensis form approximately 70-75-
degree angles. The mid-ventral surface of
the parasphenoid bears small, sharp, con-
ical teeth. This tooth-bearing surface of A.
fragosa terminates anteriorly toward the
middle of the ascending processes, whereas
in A. calva this region narrows to a point
and extends to the posterior end of the
vomers (Fig. 17). In A. uintaensis, this
region also extends to the vomers, but
covers a wider surface area in the anterior
region than in A. calva. Nearly all the
tooth-bearing surface of A. frasj^osa lies in
the posterior half of the parasphenoid,
while in A. calva this surface is centered
between the posterior and anterior areas;
in A. uintaensis two-thirds of this surface
lie in the anterior region of the para-
sphenoid. The entire tooth-bearing surface
of A. fragosa is wider than that of A. calva,
since the anterior half of the tooth-bearing
surface tapers anteriorly in A. calva, while
in A. fragosa the anterior portion maintains
a more constant width. The basic outline of
the tooth-bearing surface in A. fragosa is
subrectangular; that of A. calva is more
tear-drop shaped, with the anterior apex
widened and extended to the vomers. The
two posterior parasphenoid flanges are

Fossil Amiids • Boreske 43

more splayed in A. jra<^osa tlian in A. caJva
or A. uintacnsis, and overlie three-fourths
of the basioecipital length. As Estes ( 1964:
29) notes, there is a relatively greater dor-
soventral parasphenoid thiekness in A.
fniiiosa than in A. calva. The parasphenoid
of A. iiiiitoensis is proportionately more
massive than that of A. frai!,osa: this mas-
siveness, however, is probably a function
of its greater size.

In A. fraiiosa, as in A. uintaensis, the
extrascapular is tear-drop shaped, being
narrow at the midline and expanded dis-
tally, while in A. calva and A. sciitata, it is
more strap-shaped and longer at the mid-
line. The proximal anterior corner is
squared off, as in A. scutata and A. calva.
The anterior edge is distally concave at
the pterotic-extrascapular suture, and the
posterior edge is convex, particularly
toward the distal end, which is straight
rather than curved as in the other species
of Amia. A. fragosa and A. uintaensis lack
the posterolateral projection displayed in
A. calva.

The suprascapular resembles that of A.
calva, except that the distal edge is rela-
tively straight, rather than incurved. The
posterior border is also straight, while in
A. calva there is generally a slight concavity
in the middle of this edge; in A. uintaensis
this border is convex.

The pterotic extends further anteriorly
than in A. calva, but not to the extent that
it does in A. uintaensis or A. scutata. The
dermosphenotic-pterotic suture is directed
posterolaterally in A. fragosa and antero-
laterally in A. uintaensis, A. scutata, and
A. calva. As in A. uintaensis and A. scutata,
this bone in A. fragosa is narrower an-
teriorly than posteriorly, whereas in A.
calva the widths of these ends are relatively
equal.

The dermosphenotic in A. fragosa is about
the same relative size as in the other species
of Amia. The anterior angle that forms the
posterior border of the orbit is slightly more
pronounced than in A. scutata and A. uin-
taensis, and considerably more so than in
A. calva (Fig. 28).

The parietal in A. fragosa is character-
istically square, whereas in A. calva, A.
scutata, and A. uintaensis it is longer than
wide. The length of the parietal relative
to that of the frontal is less than in A. calva
and A. scutata and about the same as in
A. uintaensis. The characteristic deep ex-
cavation in the frontal for the orbit is
displayed in all available specimens of
A. fragosa. This led Estes (1964: 36) to
postulate the presence of supraorbital bones,
but the articulated specimens figured by
O'Brien (1969) show that this was not the
case. As Figure 28 shows, the ratio of
orbital depth to length is greater in A.
fragosa than in tlie other Amia species. As
noted in the preceding section on the cranial
morphometries of the Recent A. calva, it is
difficult to assign a specific character-state
of parietal/frontal proportions to any of the
individual fossil Amia species because of
the similarity in parietal /frontal propor-
tions (Table 7). It is apparent, however,
that the frontals of the earlier species A.
fragosa and A. uintaensis are longer relative
to parietal-length than in the mid-Tertiary
A. scutata or Recent A. calva. This feature
is useful in comparing A. fragosa with these
two species, but ineffective in distinguishing
it from A. uintaensis.

As Estes and Berberian (1969) noted,
the nasal displays a bifurcation of the an-
terior border that is lacking in A. calva.
The bifurcation is also present in A. uintaen-
sis, and the bone has approximately the
same outline and size relative to head size
as the other forms. All available specimens
of A. fragosa show that the nasals lie much
closer to the frontals than in A. calva, A.
scutata, or A. uintaensis. Although Estes
( 1964 ) states that the lacrimal conforms
closely with that of A. calva, his restoration
lacks the small posterior notch in A. fragosa
which accommodates the anterior process
of infraorbital 2. The lacrimal in A. fragosa
is evenly tapered at the posterior end, and
is anteroposteriorly longer than in other
Amia. It is also more dorsoventrally convex
than in A. scutata and A. calva.

As in A. scutata, infraorbital 4 in A.

44 Bulletin Museum of Comparative Zoology, Vol. 146, No. 1

fragosa is much more dorsoventrally ex-
panded than in A. calva, with the antero-
posterior length extending almost to the
anterior edge of the preoperculum. Infra-
orbital 4 of A. fragosa and A. scutata is
more concave at the dorsal edge, and much
more convex ventrally. The pit-line marks
extend further ventrad than is indicated in
the reconstruction by Estes ( 1964 ) .

Branchiocranium. In A. fragosa, the
supramaxilla is relatively shorter than in
A. iiintaensis, A. scutata, and A. calva, with
a greater curve in the maxillo-supramaxil-
lary suture. The dorsoposterior corner in
A. fragosa tends to be angular, as in A.
scutata, whereas in A. calva and A. uintaen-
sis it is more rounded. The supramaxilla is
deeper and more truncated at the anterior
end than in other species of Ainia.

There is a dorsal shelf adjacent to the
lingual border of the alveolar ridge which
widens the anterodorsal surface of the den-

A. fragosa A. uintaensis A. scutata A. calva

A. fragosa

A. uintaensis

A. scutata

Fig. 18. Comparison of mandibles of
(transverse sections and ventral views).

spp.

tary (Fig. 18). This shelf is lacking in
A. calva and A. uintaensis, in which the
coronoid articulation surface slopes directly
downward from the alveolar ridge. This
region of the lingual dentary surface under-
lying the coronoids extends more ventrad
at the symphyseal edge than in A. calva,
and distinctly overlaps the ventral part of
the ramus. There is no such overlapping in
A. calva; the dorsal and venti'al halves of
this region separate to form Meckel's groove.
The anterodorsal section of the dentary in
A. uintaensis overlaps the ventral half, but
not to the extent that it does in A. fragosa,
and as the coronoid articulation surface is
thicker, this thickened area of bone forms
the dorsal wall of Meckel's groove as in
A. calva (Fig. 18). As Estes (1964: 36)
noted, the coronoid teeth are styliform and
extend almost to the ventral border of the
ramus at the anterior end; in contrast, the
coronoid teeth of A. calva, A. scutata, and
A. uintaensis are pointed and the coronoids
do not extend as far ventrally as in A.
fragosa. The anterior half of the dentary
length is more curved than in A. calva, A.
scutata, and A. uintaensis (Fig. 18). This
is displayed in the MCZ 9264 specimen
( Fig. 16 ) , in which this curve approximates
a 120-degree angle at the midpoint of the
alveolar ridge. The outline of the dentary
differs from that of A. calva and A. uintaen-
sis in that the anterior end maintains an
almost constant width up to the sharp
curve at the midpoint of the alveolar ridge,
at which point it widens noticeably. When
the outline and curvature of the anterior
end of the dentary of A. fragosa are com-
pared with those of other species, the result-
ing difference appears to be correlated with
A. fragosa's relatively smaller mandible/
head ratio (Table 7), smaller mouth gape,
and its wider cranial roof (Fig. 15).

Post-cranial Skeleton. On the basis of
specimens having only the lateral surface of
the vertebral column exposed, it was con-
cluded that centra of A. fragosa are indis-
tinguishable from those of A. scutata and
A. calva. Small disarticulated vertebrae are
also basically similar in morphology, there-

Fossil Amiids • Boreske 45

fore it is impossible to differentiate A.
frau^osa, A. sciitata, and A. calva. The mid-
trnnk vertel^rae of A. fra^osa, A. sciitata,
and A. calva differ from A. uintaensis mid-
trunk vertebrae, which are generally larger,
and snbtriangular rather than ovoid. A.
fragosa does, however, have a vertebral
column that differs proportionately and
meristically from that of other species ( Fig.
31). It has a significantly smaller total
number of centra than the other species,
with approximately 12 fewer trunk verte-
brae and 8 fewer monospondylous caudal
centra. It has the same number of diplo-
spondylous caudal centra as A. calva and
A. scutata; the number of fused hypurals is
also generally the same (Table 9). The
low number of total vertebrae in A. fragosa
is reflected by its shorter, deeper-bodied
shape. The distance between the anal fin
insertion and the end of the vertebral
column is relatively shorter than in the
other species. The dorsal fin also terminates
closer to the caudal fin than in any of the
other species of Amia (Plate 1; Estes and
Berberian, 1969: 10). A. fragosa has fewer
caudal lepidotrichia (19-20) than the other
species of Amia (23-27). The head/
standard-length ratio of A. fragosa is greater
than that of A. calva, but is not significantly
different from that of A. uintaensis or A.
scutata (Table 3). The latter case is true
despite the greater number of vertebral
centra in A. uintaensis and A. scutata; this
disparity may be explained largely by the
fact that the A. fragosa skull itself is rela-
tively shorter than that of the other two
forms, particularly A. uintaensis, which has
a greater head/ standard-length ratio than
A. fragosa. Thus head/ standard-length does
not significantly reflect the length of the
vertebral column, but may be used as a
character with this qualification in mind.
The known total-length of A. fragosa falls
within the range of A. calva and below that
of A. uintaensis (Tables 1-2).

Discussion

Marsh (1871: 105) described Amia new-
berrianus and Amia depressus on the basis

of disarticulated vertebrae and cranial ele-
ments from the Bridger Formation of Wyo-
ming. His main criteria for distinguishing
these forms from A. calva and from each
other were that the chordal foramen of A.
newherrianus was "considerably above the
center in the dorsal vertebrae," and that
A. depressus possessed broader vertebrae
than A. newherrianus and lacked the me-
dian groove on the lower surface of the
centra. The vertebrae indicated that both
species were approximately the size of
A. calva. Osborn et al. (1878: 102) noted
that since Marsh gave no measurements,
"the reference to Amia depressus cannot be
certain." Marsh further noted that these
specimens belonged to the Yale College
Museum, but the specimens now seem to
have been lost. Marsh had apparently as-
sumed that the characteristics of one verte-
bra represented those of the entire vertebral
column and was unaware of intracolumnar
variation in height/ width proportions, aortal
facet morphology, and position of chordal
foramen in the vertebral coliunn of Amia.
I infer from Marsh's report that the type
specimen of A. depressus is probably a first
to third trunk vertebra, since the aortal
grooves are lacking (Fig. 11) and vertebral
width exceeds height (Fig. 14). Using the
position of chordal foramen as a character
distinguishing A. neivberrianus is undiag-
nostic since the position of the chordal
foramen changes in relation to the relative
position of the vertebra along the column
(Fig. 12). Therefore, on the basis of
Marsh's undiagnostic characters and the
similarity in size and morphology of the
vertebrae to those of A. fragosa and A.
calva, I consider both A. depressus and A.
newherrianus as nomina duhia.

Leidy (1873a: 98) descrilied Amia graci-
lis from a single trunk vertebra, also from
the Bridger Formation of Wyoming. He
noted that the centrum has two "oblong
fossae" ( aortal facets ) instead of the charac-
teristic pair of v(>ntral ridges found in Amia
calva. The size of the centrum indicated to
Leidy that A. gracilis was a smaller species
than A. calva (Leidy, 1873b). The vertebra

46 Bulletin Museum of Comparative Zoology, Vol. 146, No. 1

(ANSP 5360) corresponds approximately
to the twelfth trunk vertebra in Ainia, since
the aortal facets are oblong and unridged
(Fig. 11). Although A. gracilis is small, it
falls well within the size range of A. fra^osa
and A. calva, and is considered as a nomen
duhium.

Estes and Berberian (1969: 10) suggested
the possibility of synonymy of Paramiatus
gurleyi (Plate 1) and Atnia fra<i,osa with
the European Amia kehreri (Plate 2) on
the basis of the close proximity of dorsal
and caudal fins for the former and similar-
ity of skull elements and teeth for the latter.
It was shown in the previous section that
Paramiatus <i,urleyi conforms not only to
A. kehreri, but also to North American spec-
imens of A. frafi,osa on the basis of body
morphometries and meristics. Cranial mor-
phometries were also shown to be similar.
In addition, an X-ray (FMNH X2201) of
the Paramiatus <^urleyi skull reveals infra-
orbitals 4 and 5 to be longer than Romer
and Fryxell (1928) and Estes (1964) had
noted. The two infraorbitals extend pos-
teriorly to the anterior edge of the pre-
operculum as they do in A. frau^osa (UA
5398) from the Late Cretaceous Edmonton
Formation of Alberta. The X-ray has also
revealed a displaced left vomer with 26
styliform teeth which has been rotated
through the skull roof and covered with
matrix. All other cranial elements conform
morphologically with other Cretaceous and
Eocene specimens of A. frafiosa. These
additional similarities of Paramiatus gurleyi
and A. fragosa confirm the synonymy of
these two species which was suggested by
Estes and Berberian (1969: 10), and I
therefore include Paratniatus gurleyi in the
synonymy of Amia fragosa.

Comments on Related European Forms

Amia kehreri was described by Andreae
(1892, 1895) from a caudal region, infra-
orbital 4, disarticulated trunk vertebrae,
and a left operculum from Middle Eocene
(Lutetian) deposits at Messel bei Darm-
stadt (specimens at the University of

Heidelberg, Andreae Collection). On the
basis of specimens from the same deposit
(BMNH P33480, Plate 2; P33488), it con-
forms morphometrically with A. fragosa in
head/ standard-length, pectoral fin insertion/
standard-length, mandible/ head-length, and
operculum-length/operculum-depth ( Ta-
bles 3, 7). The distance between the anal
fin and the end of the vertebral column
exceeds that of the North American A.
fragosa specimens, but is less than in A.
scutata or A. calva. The parietal/ frontal
ratio is also greater than in A. fragosa, but
smaller than in A. scutata or A. calva.
Meristics of vertebral elements as well as
the cranial characters discussed by Estes
and Berberian ( 1969 ) also conform with
those of A. fragosa. I agree with Estes and
Berberian ( 1969 : 10 ) that only differences
in temporal and geographical factors appear
to distinguish Amia kehreri from A. fragosa;
any osteological dissimilarities are of minor
significance.

The Middle Eocene European Geiseltal
deposits contain numerous amiid fossils;
according to Estes and Berberian ( 1969 )
some showed resemblances to A. kehreri.
This material is currently being described
by Anna Jerzmanska, Uniwersytet Wroclaw-
ski, Wroclaw, Poland.

Another related form is Amia valencien-
nesi from the Eocene of Puy-de-D6me,
France. Agassiz ( 1843 ) described the form
from one complete specimen and an an-
terior region of another (BMNH P446,
27736). Piton (1940) reviewed these .speci-
mens along with new material collected at
the same locality. A. valenciennesi also re-
sembles A. kehreri in its vertebral number
of 68 centra, close approximation of dorsal
and caudal fins, and an infraorbital 4 larger
than infraorbital 5; these similarities indi-
cate that synonymy with A. kehreri is in
order. The name A. valenciennesi precedes
A. kehreri, and thus has priority.

Estes and Berberian (1969: 7) showed
that Amia russelli Janot ( 1966, 1967 ) from
the Late Paleocene of France resembles A.
fragosa in (!) square parietals, (2) similar

Fossil Amiids • Boreske 47

parietal/frontal ratio, (3) largo orbital ex-
cavation in frontal, and (4) similar opc>r-
cnlnni height/width ratio. Thns A. russclli
conforms with many of the most distinct
characters of A. kehreri and A. valencien-
nesi, and should be considered a synonym
of the latter.

Estes (1964) re-evaluated Dechaseaux's
( 1937) redescription of the Early Oligocene
Amia munieri from France and noted simi-
larities with A. fraiS,osa which included ( 1 )
styliform vomerine teeth, (2) branchiostegal
rays rounded distally, ( 3 ) larger infraorbital
4 than infraorbital 5, and (4) similar
parietal /frontal proportions. The principal
difference between the forms is the small
excavation for orbits in A. munieri. Since
Dechaseaux's and Estes' studies, the speci-
men (MNHN R4632, skull and associated
cranial and postcranial elements) is being
further prepared to display the cranial roof
and palate more extensively. The frontal
lacks a prominent excavation for the orbits
as Estes (1964: 40) has noted, and in this
feature A. munieri resembles A. sctitata and
A. calva. A. mimieri is a very important
form because it represents the only com-
plete amiid specimen known from the Early
Oligocene, and, as noted, it displays inter-
mediate morphology of the cranial features
among the species of Amia. A. munieri
occurs very late in time in relation to the
last known occurrence of A. jragosa in
North America, and because there are no
complete specimens known from this age,
it represents a stage of evolution among the
amiids that is not found in North America.

Lehman ( 1951 ) described Fseudamia
lieintzi (Troms0 Museum Naturhistorisk
collections, Troms0, Norway) from a fairly
complete articulated sptx-imen and two
skulls from probable Eocene deposits in
Spitzbergen. He differentiated this form
from Amia on the basis of ( 1 ) Sinamia-\\ke
metapterygoid and (2) presence of a con-
cave notch on the dorsoposterior border of
the operculum. Estes ( 1964 ) noted that
Lehman was incorrect in his interpretation
of tlie nature of the metapterygoid and
operculum, and therefore suggested that

Fseudamia might be placed in the genus
Amia. From the examination of Lehman's
plates, it appears that this form resembles
A. fra^osa in its deep-bodied shape and
low parietal /frontal ratio (approximately
0.410), and that it may be synonymous with
A. valenciennesi and A. kehreri. I'urther
preparation would possibly be helpful in
uncovering palatal teeth, whose moiphology
would aid in a more definitive description.
Although the exact age of the Eocene
deposit in which the specimen occurred
is uncertain, this Spitzbergcni locality, if
Early Eocene, lies on the possible migration
route of amiids (and other vertebrates)
between North America and Europe.

Amia uintaensis (Leidy, 1873)

Protamia tiintacnsis Leidy, 1873a: 98.
Protamia media Leidy, 1873a: 98.
Pappichthy.s plicatus Cope, 1873: 635.
Pappichthtjs sclerops Cope, 1873: 635.
Pappichthy.s laevis Cope, 1873: 636.
Pappichthijs symphysis Cope, 1873: 636.
Pappichthys corsonii Cope, 1873: 636.
Pappichthys meditis Cope, 1884: pi. 4.
Amia ivhiteavcsiana Cope, 1891: 2.
Amia macrospondyla Cope, 1891: 2.

Holotype. ANSP 5558, anterior tiimk
vertebra.

Paratypes. ANSP 8044, first anterior
trunk vertebra; ANSP 3151, three posterior
trimk vertebrae; ANSP 5622, basioccipital.

Type locality and horizon. Henrv's Fork.
North half of section 5, T 12 N, R 111 W,
Sweetwater County, Wyoming; Bridger
Formation.

Age rouge. Torrejonian (Middle Paleo-
cene) to Chadronian (Early Oligocene).

Hypodi^m. Paleocene. Fort Union For-
mation, Wyoming and Montana: PU 17117,
maxillary; PU 17068, vertebrae and denta-
ries; PU 162.36, disarticulated skull and
trunk vertebrae; CM 25364, dentary; PU
17064, trunk vertebrae. Tongue River For-
mation, Montana: PU 20578, basioccipital
and vertebrae. Paskapoo Formation, Al-
berta: ROM 4653, vertebrae.

Eocene. Will wood Formation, Wyo-
ming: PU 21173, basioccipitals; PU 17227,
basioccipital and trunk vertebrae; PU 17649,

48 Bulletin Museum of Comparative Zoology, Vol. 146, No. 1

portion of cranium; PU 18760, skull frag-
ments, dentary, and vertebrae. Wasatch
Formation, Wyoming: AMNH 4635, dentaiy
and maxilla. Golden Valley Formation,
North Dakota: PU 18568, basioccipital.
Green River Formation, Wyoming: USNM
18147, skull fragments and vertebrae;
AMNH 785, complete caudal region; PU
13865, nearly complete specimen; MCZ
12916, disarticulated skull and associated
vertebrae. Wind River Formation, Wyo-
ming: AMNH 2437, dentary and skull frag-
ments. Bridger Formation, Wyoming: CM
25362, portion of cranium and vertebral
column; AMNH 4631, portion of cranium
with dentaries, gular, and basioccipital;
USNM 170976, maxilla; YPM 6238-6240,
6242, 6244, 6250-6253, 6257-6258, vertebrae
and basioccipitals; USNM 170973, 5450,
3962, 3963, 3966, PU 20523, 10101, ANSP
2337-2339, vertebrae; USNM 2181, ANSP
5632, trunk vertebrae; USNM 3959, trunk
and caudal vertebrae; ANSP 5580, mid-
trunk vertebra; AMNH 2539, anterior por-
tion of a left dentary, two premaxillae, right
quadrate, left epihyal, anterior portion of an
ectopterygoid, three trunk vertebrae, and
numerous fragments of angular; USNM
3965, left dentary; USNM 3968, anterior
dentary fragment; AMNH 2570, pre-maxil-
lary fragment, fragments of angular, left
quadrate fragment, trunk vertebra frag-
ment, and a caudal vertebra; USNM 3960,
PU 10099, 10110, vertebrae and a ural cen-
trum; USNM 5476, basioccipital; USNM
3961, left dentary fragment. Washakie For-
mation, Wyoming: FMNH 27465, 4509, ver-
tebrae. Uinta Formation, Utah: CM 2382,
maxillary fragment.

Oligocene. Cypress Hills Formation,
Saskatchewan: NMC 6197, trunk vertebra;
NMC 6198, caudal vertebra.

Known distribution. Montana, Wyo-
ming, Utah, North Dakota, Alberta, and
Saskatchewan.

Revised diagnosis. Vertebral column
with approximately 20 more vertebral seg-
ments in total number (85) than A. fragosa,
and five fewer trunk centra (31) and five
more diplospondylous caudal vertebrae

(21) than in the other long-bodied forms,
A. scutata and A. calvo. Mid-trunk verte-
brae subtriangular rather than ovoid. Pa-
latal teeth sharp, greatly curved inwardly.
Between 40-45 vomerine teeth as compared
with 15-17 in A. fragosa, A. scutata, and
A. calva. Hyomandibular more deeply
notched between opercular process and
extensor (dorsal) surface than in other
species; opercular process relatively larger.
Angle between alveolar ridge and exterior
surface of the dentary forms a more acute
angle than in the other species. Mandibular
ramus less curved than in other species, so
that angle between symphyseal ends of
dentaries is relatively narrow. Greater
mandible/head ratio (0.693) and head/
standard-length ratio (0.322) than any of
the other forms: A. uintaensis has a head
relatively longer and a mouth gape rela-
tively wider than do other species. Most
specimens are significantly larger than the
other species, with a relatively greater de-
gree of ossification of all bones. Greatest
known standard-length 800 mm.

Introduction

Leidy (1873a) reported numerous dis-
articulated vertebrae of a fossil fish related
to Amia from the Bridger Formation of
Wyoming. He distinguished a new genus
Protamia from Amia by its "two oval fossae"
( aortal facets ) on the ventral surface of the
centrum, and by large vertebrae character-
istically with a much greater width to
height proportion. Hijpamia, another new
genus from the same locality which Leidy
also related to Amia, was characterized by
also being larger than A. calva, and by
vertebrae whose sides converged into a
"medium prominence excavated into a pair
of oval fossae" deeper than those of Pro-
tamia. Later ( 1873b ) , Leidy published a
more complete and illustrated account of
the various species of the new genera
Protamia and Hypamia. In the same year
Cope ( 1873 ) described a new amiid genus,
also from the Bridger Formation, which he
named Pappichthys. He distinguished this
new genus from Amia by the "presence of

Fossil Amiids • Boreske

only one series of teeth, instead of several,
on the bones about the mouth." Osborn
et al. ( 1878 ) reported other finds of Pci})-
picJitJiys from the Bridger Formation which
seemed to fit Cope's description. Cope
(1884) further discussed his new genus,
and rejected Leidy's prior nomenclatiue
and description.

New ton ( 1899 ) discussed this nomencla-
tural controversy and asserted the validity
of Leidy's genius Protamia, since Cope's
later diagnosis \\'as no more effective in
characterizing the new genus than Leidy's
prior one. Newton bc^lieved that Cope's
description of PappicJitlitjs as having only
a single row of marginal teeth was taxo-
nomically undiagnostic, since this condition
would also include A. calva. Romer and
Fryxell ( 1928 ) accepted Leidy's earlier
description and genus as diagnostic, and
referred PappiclitJujs to Protamia. They
also mentioned Hypamia but found little
to distinguish it from Amia.

Hussakof (1932) continued to use Cope's
name, however, and reported large speci-
mens of Pappichthys from the Eocene of
Mongolia. He also noted Cope's error in
diagnosing the tooth characteristics of the
genus, since Pappichthys {Protamia) has
several rows of small teeth on the "splenial
bone." In comparison with Amia he noted
"points of difference in nearly every bone
available for comparison," and concluded
that Pappichthys was a valid genus, "not
merely a group of large-sized extinct species
of Amiatus."

Estes (1964), like Romer and Fryxell
(1928), referred Pappichthys to Protamia,
and reported several vertebrae and a maxil-
lary fragment from the Cretaceous Lance
Formation of Wyoming. He inteipreted the
increase in breadth over thickness of the
vertebrae as a po.ssible "function of in-
creased size," a condition that would also
allow for tlie comparatively more massive
nature of the maxillary fragment. He also
considered the retention of this genus as
arbitrary until enough materials were avail-
able. Janot (1967) did not consider this
single distinguishing characteristic of the

vertebrae as sufficient foundation for the
erecti(m of a new genus, and therefore sug-
gested relerring Protamia to Amia. Estes
et al. (1969) concurred with Janot in
synonymizing Protamia with Amia. The
present study confirms tliis synonymy;
Leidy's species (1873a) has priority and
the valid name of this fish is thus the oldest
specific name, Amia uintacnsis.

Revision of all forms referred now or in
the past to Protamia is much needed, for
these large amiids were diagnosed on char-
acters of isohited vertebrae and skull frag-
ments. This study gives more useful
diagnostic characters that provide a basis
on which the taxonomy of this group can be
established.

Fossil Record

The major deposits carrying remains of
Amia uintacnsis (Table 18) range in age
from Middle Paleocene to Early Oligocene.
Middle Paleocene specimens occur in the
Fort Union, Tongue River, and Paskapoo
formations and consist mostly of isolated
and broken centra, and dentary and maxil-
lary fragments. A nearly complete skull
(PU 162.36) with associated trunk and
caudal centra from the Bear Creek local
fauna of Montana (Fort Union Formation)
is the only articulated specimen from the
Late Paleocene. The Eocene material in-
cludes one complete articulated specimen
(PU 13865), one complete caudal region
(AMNH 785), and a disarticulated skull
(MCZ 12916) from the Creen River Forma-
tion. PU 13865 (Plate 3) has the axial
skeleton intact in matrix, with a dislocated
fifth centrum that is the only one available
for three-dimensional measurements. This
is also the only specimen in which a com-
plete vertebral count can be taken. AMNH
785 provides excellent meristic information
for the caudal region (Fig. 8C). CM 25362,
from the Bridger Formation, consists of a
left palatal and opercular series and an
almost complete, disarticulated vertebral
column that permitted the taking of a series
of centrum measurements. Other skull frag-
ments and vertebrae occur in many deposits

50 Bulletin Museum of Comparative Zoology, Vol. 146, No. 1

throughout the Eocene (Table 18). The
latest occurrence of A. uintaensis is repre-
sented by two isolated centra from the
Cypress Hills Formation (Oligocene, Chad-
ronian ) .

Description

Neurocranhan. Posterior to the spinal
arterial foramina the basioccipital includes
two fused vertebrae. As the basioccipitals
display great variation in the morphology of
the articular surface, it is difficult to char-
acterize this form on the basis of this
feature. However, the articular surface is
generally kidney-shaped, with dorsal in-
dentations bet\veen the neural facets, and
ventrally there is an indentation distal to
the aortal facets. In A. froii^osa and A.
calva the basioccipital has ovoid articular
surfaces with no dorsal indentations be-
tween the neural facets (Estes, 1964: 29,
fig. 15). In lateral view the distal articular
surface of the A. uintaensis basioccipital is
not perpendicular to the parasphenoid
flanges; the dorsal half of this surface is
more anteriorly directed than the ventral
half.

The parasphenoid is longer relative to
its width than it is in either A. calva or
A. fragosa, primarily in the region anterior
to the ascending processes. At the point
nearest the ascending processes, it lacks
the pronounced convexity and die accom-
panying anterior lateral notches found in
A. calva and A. fragosa. The ascending
processes are slightly less anteriorly oriented
in ventral view than in A. calva, but more
so than in A. fragosa ( Fig. 17 ) . The region
posterior to the ascending processes is rela-
tively shorter than in A. fragosa or A. calva;
it is also more massive and more ventrally
convex than in the other two forms. The
posterior parasphenoid flanges resemble
tliose of A. calva more tlian A. fragosa in
outline as well as juxtaposition; those of
A. fragosa are more laterally splayed than
in A. uintaensis or A. calva. The tooth-bear-
ing surface differs considerably from that of
A. fragosa and somewhat from A. calva in
outline and extent. As in A. calva, this

surface extends anteriorly to the vomers,
but its width is much greater and more
constant than in A. calva, which is narrowly
tapered anteriorly. Posteriorly, this surface
extends further than in A. calva, but not as
far as in A. fragosa. Approximately two-
thirds of the tooth-bearing surface lies
anterior to the ascending processes, while in
A. calva this area is anteroposteriorly cen-
tered, and in A. fragosa it is nearly all
posterior. The tooth-bearing surface covers
a greater portion of the ventral surface of
the parasphenoid than in A. fragosa or A.
calva; its basic outline is diamond-shaped,
with the anterior apex widened and ex-
tended to the vomers, while that of A.
fragosa is subrectangular and that of A.
calva is tear-drop shaped with the apex
sharply protracted anteriorly.

In A. uintaensis the distal edge of the
suprascapular is convex as in A. calva, while
in A. fragosa this edge is almost a straight
line. The posterior border is more rounded
distally than in A. calva and is convex rather
than concave.

In having the extrascapular rounded at
the distal border, A. uintaensis is the same
as A. calva and A. scutata, but differs from
both of them in that th(^ posterior border
is not concave, and from A. calva alone in
lacking the distal posterior process. The
anterior border is relatively straight, unlike
the condition in A. calva and A. scutata, in
which the lateral distal ends of the anterior
borders are directly posteriad. As in A.
fragosa the midline is shorter than in A.
scutata and A. calva.

As in A. scutata and in A. fragosa the
pterotic is narrower at the anterior than
posterior border, while in A. calva and, to
an extent, in A. scutata the ends are sub-
equal. As in A. fragosa they extend farther
anteriorly and adjoin the frontal s postero-
laterally. The dermosphenotic-pterotic su-
ture is anterolaterally directed, as in A.
scutata, but not as pronoimced as in A.
calva. The anterolateral edge of the pterotic
is indented and forms, witli the dermo-
sphcnotic, an additional concavity in the
outline of the cranial roof. Aside from this

Fossil Amiids • Boreske 51

anterior indentation, the lateral borders are
relatively straight, as eompared with the
smoothly coneave exterior sides of the
pteroties in A. scututa, A. calva, and A.
fruf!,osa. The posterior border forms a
smooth line, as in A. fragosa, and laeks the
small lappet that A. scutata and A. calva
display.

The dermosphenotie is similar to that ot
A. calva in relative size and outline, al-
though it does not jut as deeply into the
frontals. Its anterior border is rounded, as
in A. calva, rather than sharply angular, as
in A. fra^osa. The posterior half of the
outer lateral border is indented to form a
coneavity with the anterior tip of the
pteroties. The parietal in A. uintaensis is
elongated anteriorly, as in A. calva and A.
scutata, while that of A. jra^osa is relatively
square. The orbital excavation in the lateral
sides of the frontal is shallow as in A. calva
and A. scutata, while that of A. fragosa is
characteristically deep (Fig. 28). The sen-
sory canal cannot be determined. The
frontals are more elongated relative to
parietal length tlian in A. calva and A.
scutata; the parietal /frontal ratio is only
slightly smaller than that of A. fragosa
(Table 7). The distal lateral border tapers
anteromedially, and the anterior ends are
relatively pointed anteriorly, forming a deep
notch on the midline suture.

There is a slight bifurcation of the
anterior border of the nasal as in A. fragosa.
The nasal bones are relatively narrower
than in A. fragosa or A. calva, but are
otherwise similar in shape and relative size.
They are fairly well separated from the
frontals, as in A. calva and A. scutata, rather
than abutting them as in A. fragosa.

The lacrimal in A. uintaensis resembles
that of A. fragosa in general morphology,
although it lacks the posterior notch for the
anterior end of infraorbital 2 which is
present in the other species of Amia. The
lacrimal, like that in A. fragosa, is relatively
longer and more tapered posteriorly than in
A. scutata and A. calva. It is more dorsally
convex than in the other forms, but only
slightly more so than in A. fragosa.

The infraorbital 5 in A. uintai'nsis is
similar to that in A. fragosa and A. scutata,
being less robust posteriorly than in A.
calva. As in the other forms, it is narrower
anteriorly than posteriorly. The ventral
border is relatively straight, while that of
the other forms is posteriorly convex. Infra-
orbitals 2, 3, and 4 have not been identified.

The vomerine tooth patch in A. uintaen-
sis, as in A. fragosa, extends inore posteri-
orly than in A. calva (Fig. 19). The
\^omerine teeth are sharp and greatly curved
posteriorly; they exceed those of A. fragosa
and A. calva in number, each vomer bear-
ing between 40-50 teeth, as compared to
half that number in A. fragosa and A. calva.
The rostral and antorbital are identical to
that of the other species.

Brancliiocranium. The suture between
the anterior and posterior dermopalatine
cannot be discerned. In A. uintaensis the
dermopalatine has about twice the number
of teeth as in A. calva, and the tooth patch
extends more distad. The teeth are sharply
pointed, as are the vomerine teeth.

The hyomandibular is more deeply ex-
cavated between the opercular process and
the extensor ( dorsal ) surface, and the oper-
cular process is more massive and extends
further ventrad, forming a larger articula-
tion surface, as compared with the other
species of Amia. The articular surface of
the quadrate is more robust than in other
species of Amia and displays three cristae
ventrally rather tlian dorsally as in A. calva
and A. rolmsta (Janot, 1967: 144). The
ceratohyal resembles that of A. calva and
A. fragosa with the exception of its being
thicker at the neck of the proximal end.
The metapterygoid in A. uintaensis con-
forms very closely to that of A. calva in
outline and in the position of the anterior

Fig. 19. Comparison of vomers of A, Amia calva; B, A.
uintaensis; and C, A. fragosa.

52 Bulletin Museum of Comparative Zoology, Vol. 146, No. 1

basal process and the posterolateral otic
process.

The maxilla in A. uintaensis is more
robust and relatively longer, and its pos-
terior border is dorsoventrally wider than
in the other forms, particularly A. fragosa.
As in A. calva the small supramaxillary
notch occurs more anteriorly than in A.
fragoso. The dorsoposterior border is
rounded, as in A. calva and A. scutata,
rather than sharply angular, as in A. fragosa.
Anteriorly the maxilla is deeper and more
thickly ossified than in the other forms,
but this may be a function of greater size.
The supramaxilla resembles that of A. calva
in general morphology, being elongated and
narrowly tapered anteriorly, with a smooth-
ly rounded posterior end conforming to the
curve of the maxilla. The maxillo-supra-
maxillary suture is straight as in A. calva.
The premaxilla is identical to that of the
other species.

The dentary of A. uintaensis is similar to
that of A. calva and A. scutata in lacking
the dorsal shelf of the anterior lingual bor-
der of the alveolar ridge which occurs in
A. fragosa. The coronoids articulate more
or less vertically on the alveolar ridge, as in
A. scutata and A. calva. The anterodorsal
region of the dentary slightly overlaps the
ventral half, but not to the extent that it
does in A. fragosa; A. uintaensis seems to
be intermediate between A. fragosa and A.
calva in this feature, the latter having no
such ventral overlapping at the symphyseal
edge. The coronoid articulation surface of
the A. uintaensis dentary is thicker than in
A. fragosa and A. calva, but only slightly
more so than in A. scutata. At the termina-
tion of this surface, this thickened area of
bone forms the dorsal wall of the Meckelian
groove, as in A. calva. The ventral wall of
this groove is less well defined than in A.
calva, witli A. scutata being intermediate.
The anterior half of the dentary length in
A. uintaensis is evenly tapered to the
symphyseal edge; it is elongated and lacks
the sharp curve present in A. fragosa at the
midpoint of the alveolar ridge (Fig. 18).
There is only a trace of such a curve in the

dentaries of A. calva and A. scutata which
are also more elongated and evenly tapered
than in A. fragosa, although not to the ex-
tent that they are in A. uintaensis. Anteri-
orly, the bone is also relatively thicker than
in A. fragosa and A. calva; A. scutata also
displays this greater ossification at the
anterior end of the dentary. Posteriorly,
the dentary is very similar to that of A.
calva. The coronoid teeth are sharp and
conelike, extending to the midpoint of the
lingual surface, as in A. calva. As Janot
(1967) shows for A. robusta, the alveolar
ridge is more horizontal in A. uintaensis and
forms a more acute angle with the exterior
surface of the dentary than it does in A.
fragosa or A. calva; A. scutata is interme-
diate between A. uintaensis and A. calva in
this feature (Fig. 18). In A. uintaensis the
first coronoid (symphyseal) overlies only
the dorsal half of the anterior articular sur-
face of the dentary, as in A. calva and A.
scutata. The teeth are more sharply pointed
than in any of the other forms (Fig. 18).
The second coronoid is fragmentary, but
appears to resemble that of A. calva with
the exception of its having more sharply
pointed teeth. The prearticular specimens
available are fragmentary, but the lingual
surface possesses blunt-conical teeth similar
to those in A. calva and A. fragosa. Bor-
sally, however, these teeth are as sharply
pointed as the coronoid teeth. The angular
is slightly longer and higher than that of
A. calva. The posterior border is more ver-
tical, with the articular notch less pro-
nounced. It is more heavily ossified than
in A. calva, but this may be a function of
size. The surangular in A. uintaensis is
basically similar to that of A. calva, al-
though it is situated more dorsally and is
more rounded at the dorsal edge.

The gular is longer than that of A. calva
and A. fragosa (Fig. 20). It is also slightly
narrower at the posterior end than the
anterior end, while the reverse is generally
true in A. calva. Otherwise, the gular
strongly resembles that of A. calva. Despite
a few minor dissimilarities, the preopercu-
lum resembles that of A. calva. There is a

Fossil Amiids • Boreske 53

Fig. 20. Comparison of gulars of A, Amia calva; B,
A. uintaensis; and C, A. fragosa.

slightly more pronounced concavity in the
ventroposterior border than is exhibited in
A. calva; this concavity is altogether lacking
in A. fragosa. The line of curvature is about
the same as in A. calva; in A. fragosa the
preoperculum is more deeply curved. The
dorsal half is not quite as wide as the
ventral half, while in A. calva both ends are
fairly equal. In A. fragosa, however, the
dorsal half is much narrower and more
tapered than the ventral half, which is rela-
tively wider and bulbous. The operculum
in A. uintaensis is similar to that of A.
calva and A. scutata in operculum-depth/
operculum-length (Table 7). The suboper-
culum conforms in general morphology with
that of A. calva, although it is slightly more
robust, particularly in the posterior region.
The corners tend to be angular, as in A.
scutata and A. calva, rather than rounded,
as in A. fragosa. The interoperculum is
similar to that of A. calva, although more
robust. The anterodorsal border is more
convex than in A. calva, and is more deeply
impressed into the preoperculum. The
anteroventral border is narrowly tapered as
in A. calva, rather than smoothly rounded as
in A. fragosa. The first branchiostegal ray
conforms to that of the other species. Al-
though the lack of articulated material
makes any count of the rays difficult, in
MCZ 12916 there are 12 disarticulated
branchiostegal rays on the right side of the
cranial roof. As in A. fragosa the distal ends
of the rays are consistently rounded, rather
than squared as in A. calva.

Post-cranial skeleton. The supracleithrum

in A. uintaensis resembles that of A. calva
and A. fragosa, excepting the dorsal articu-
lation surface, which is rectilinear rather
than pointed as in A. calva. The distal
lateral border in the Paleocene specimens
lacks the notch that occurs in A. calva, but
this notch is present in the Eocene speci-
mens. The metacleithrum in A. uintaensis
is more elongated than in A. calva and A.
fragosa. The dorsal end is narrower than
in A. calva, and the ventral end is sciuared
off. The cleithrum in A. uintaensis is largely
similar to that of the other Ainia species,
but is more massive at the proximal end
than in A. calva, and the dermal sculpture
covers a greater area than in A. calva, ex-
tending to the distal border as in A. fragosa
and A. scutata (Fig. 21). The mid-distal
border is smoothly convex and lacks the
notch ventral to the metacleithrum which
is present in A. calva.

The preceding study of the vertebral
skeleton of A. calva revealed changes in
height/ width proportions, position of
chordal foramen, configuration of neural
and aortal facets, and in the basapophyscal
angles and length of basapophyses which
may be used here to discern similar trends
in A. uintaensis centra, for the fossil verte-
brae display the same features characteris-
tic of the Recent species even in disarticu-
lated state.

CM 25362 from the Bridger Formation is
the only specimen that has a relatively
complete, disarticulated, undistorted verte-

Fig. 21. Comparison of cleithra of A, Amia calva; B,
A. scufata; C, A. uintaensis; and D, A. fragosa.

54 Bulletin Museum of Comparative Zoology, Vol. 146, No. 1

bral column; as the centra are separable this
specimen is useful in comparisons with iso-
lated vertebrae. There are 59 centra pres-
ent: 25 trunk centra and 34 caudal centra,
including two fused urals. Many of the
preserved caudal centra are only fragments.
Since the articulated specimen (PU 13865)
has 85 vertebrae (see Table 9 for regional
numbers) it may be assumed that about 25
vertebrae are missing from CM 25362.
When comparing vertebrae from different
regions of the column in the two specimens,
it appears that CM 25362 lacks approxi-
mately six trunk and approximately twenty
caudal centra. The first anterior trunk cen-
trum present in the CM 25362 series pos-
sesses aortal facet configurations similar to
those of the seventh vertebra of the articu-
lated specimen ( PU 13865 ) . An articulated
but separable series of six uncrushed an-
terior trunk vertebrae (PU 10101), also
from the Bridger Formation, aids in the
reconstruction of the anterior region of the

A. uintaensis vertebral column (Figs. 22-
25). The basapophyseal angles of these six
PU 10101 vertebrae do not vary from 180
degrees. The first six anterior trunk verte-
brae from a partly disarticulated vertebral
column from the Paleocene specimen (PU
16236) resemble the six PU 10101 centra in
length and shape of aortal facets, even
though PU 16236 is a smaller individual.
The height/ width ratio of these latter centra
is difficult to determine, however, since the
specimen underwent postdepositional crush-
ing. The nearly complete vertebral series
of the CM 25362 specimen has been used
for the construction of the remaining trunk
and caudal region in the model of the A.
uintaensis vertebral column. The trunk
centra of CM 25362 have been arranged
according to basapophyseal angles that de-
crease from 180 to 46 degrees, as in A. calva.
Decreasing size was used to arrange the
caudal vertebrae.

Although A. uintaensis occurs much ear-

Table 14. Angle of basapophyses, length, height, and width of vertebrae of Amia
uintaensis compared with type specimens of synonymized taxa as illustrated

IN figure 22

Relative
Vertebral
Number

Specimen

Angle of

Basapophyses

( Degrees )

Length
(mm)

Height

(mm)

Width

(mm)

A. uintaensis PU 10101
A. uintaensis PU 16236
P. uintaensis ANSP 8044
P.'sp. USNM 170973

A. uintaensis PU 10101
A. uintaensis PU 16236
A. whiteavesiarm NMC 6197
P. sp. FMNH P27465

A. uintaensis PU 10101
A. uintaensis PU 16236
P. sp. USNM 3966
P. medius USNM 3959

A. uintaensis PU 10101
A. uintaensis PU 16236

A. uintaensis PU 10101
A. uintaensis PU 16236
A. uintaensis PU 13865

A. uintaensis PU 10101
A. uintaensis PU 16236
P. uintaensis ANSP 5558

ISO"

180°

180°
180°

180°

8.0

32.0

45.0

6.0

31.0

44.0

8.0

32.0

46.0

5.5

19.0

29.0

8.0

32.0

44.5

8.5

28.5

39.0

8.5

29.0

40.0

9.0

28.0

36.0

10.0

33.0

43.0

8.0

30.0

41.5

8.5

21.5

29.0

8.5

22.0

30.0

10.0

33.0

44.0

7.5

33.0

40.8

11.0

33.5

44.0

9.5

31.5

39.0

4.5

16.5

21.5

11.0

34.0

42.5

9.0

33.5

34.0

10.5

32.5

40.0

Fossil Amiids • Boreske 55

licr in time than A. calva and A. scutata, it liodicd form tlian its contemporary, A.
has approximately the same total number of fra^osa, which has a mean of 65 centra,
centra (85), and like them is a longer- Tlu> vertebral column of A. umfaenm does,

4. uintaensis
PU lOIOI

A. uintaensis
PU 16236

P. uintaensis
ANSP 8044

A . whiiteavesiana
NMC 6197

R sp.
USNM 3966

A. uintaensis
PU 13865

P. sp.
USNM 170973

P sp.
FMNH P27465

R uintaensis
ANSP 5558

P medius
USNM 3959

Fig. 22. First anterior trunk vertebrae (A,B) of Amia uinfaens'is compared with type specimens of synonymized taxa
(refer to Table 14 for data).

Bulletin Museum of Comparative Zoologij, Vol. 146, No. 1

R sp. P. sp.

USNM 170973 USNM 3962

P. Sp.
USNM 170973

"-^,/

P. plicotus
AMNH2539

P sp.
PU 20523

P Sp.
USNM 170973

P sp.
USNM 170973

12C

12d

Amia. sp.
ANSP2337

P. plicafus
USNM 170974

P. medius
USNM 3959

12a

p. medius P. sp.

YPM 6238 FMNH P27465

P. medius
USNM 3959

12b

Amia sp.
ANSP 2339

Fig. 23. Seventh through fourteenth mid-trunk vertebrae of Amia uinfaensis compared with type specimens of
synonymized taxa (refer to Table 15 for data).

Fossil Amiids • Borcske 57

however, differ meristieally from that of 36 (mean) in A. sctitata. The number of

A. calva and A. scutata in number of verte- diplospondylous vertebrae is 20-21, as eom-

brae in the various regions. There are 31 pared with 14-17 in A. calva and 15 in

trunk eentra in A. uintacnsis (PU 13865), A. scutata. This variation from A. ra/i;« and

as opposed to 37 (mean) in A. calva and A. scutata in the organization of the verte-

/? plicotus P. medius
USNM 3958 USNM 3959

R medius
YPM 6239

R sp.
USNM 3966

P symphysis
PU 10099

19a

19b

P medius
USNM 3959

R sp.
USNM 3966

R sp
USNM 3962

P sp. P sp.

USNM 3966 FMNH P27465

P medius
YPM 6240

P sp.
USNM 3966

19C

R sp. R sp-

FMNH P27465 USNM 3963

P. sp. P medius

USNM 3966 USNM 3959

R medius
USNM 3959

R loevis
USNM 3968

P uintaensis
ANSP 3151

Fig. 24. Fifteenth through twenty-second posterior trunk vertebrae of Amio uiniaemis compared with type speci-
mens of synonymized taxa (refer to Table 16 for data).

5(S Bulletin Museum of Comparative Zoology, Vol. 146, No. 1

Psp.
YPM 6242

/? uintaensis
ANSP3I5I

P. loevis
PU 10109

R laevis
PU 10109

P. media
ANSP5632

P Sp.
USNM 3963

Amia sp.
ANSP2338

P symphysis
PUIOIIO

A. mocrospondyla
NMC 6198

P. laevis
USNM 3968

P sp.
FMNH PF4509

P sp.
USNM 3966

P. medius
USNM 3959

P sp.
USNM 5450

P. medius
USNM 3959

P. corsoni
USNM 3961

Fig. 25. Posterior trunk and caudal vertebrae of Am/o umfoensis compared with type specimens of synonymized
taxa (refer to Table 17 for data).

Fossil Amiids • Boreske 59

Table 15. Angle of hasapophyses, length, height, and width ok vehtehuae of Amia
iiintacn.sis compared with type specimens of synonymized taxa as illustrated

IN figuhe 23

Relative
Vertebral
Number

Specimen

Angle of

Basapophyscs

( Def^rees )

Length
(mm )

Height
(mm)

Width
( mm )

7
8
9

12a

12b
12c
12d
14

P. sp. USNM 170973
P. sp. USNM 3962

P. sp. USNM 170973

P. plicatus AMNH 2539
P. sp. USNM 170973

P. .s7>. PU 20523

P. sp. USNM 170973

P. 77ic(lius YPM 6238
P. sp. FMNH P27465
P. mcdius USNM 3959

Aiiiiu sp. ANSP 2339

A7nia sp. ANSP 2337

P. ])licatus USNM 170974

P. mcdius USNM 3959

179°

8.5

19.5

25.5

7.0

20.0

28.0

178°

8.5

19.5"

26.0

177°

6.0"

18.0^

24.0

8.0

21.0

25.5

174°

7.5

25.0

30.0

7.5

22.0'

25.0

6.0

19.0

24.0

171°

11.0

30.0

35.0

9.0

22.0

28.0

167°

10.0

29.0"

40.0

166°

10.0

29.0

38.5

163°

8.0

24.0

30.0

160°

7.5

19.0

23.0

Est.

Taulk 16. Angle of hasapophyses, ijiingth, hkic;ht, and width of verteijuae of Amu/
uintactisis compared with type specimens of synonymized taxa as illustrated

IN figure 24

Relative
Vertebral
Number

19a
19b

19c

21
22

Sjiccimen

P. plicatus USNM 3958
P. mcdius IfSNM 3959
P. mcdius YPM 6239
P. sp. USNM 3966
P. sijmphysis PU 10099

P. mcdius USNM 3959
P. sp. USNM 3962

P. sp. USNM 3966

P. ,v/;. USNM 3966
P. sp. FMNII P27465

P. i7icdius YPM 6240

P. sp. USNM 3966

P. sp. FMNH P27465
P. sp. USNM 3963

P. sp. USNM 3966

P. Tticdius USNM 3959

P. mtY/tt/.v USNM 3959

P. /at't;i.v USNM 3968
P. uiiitacrisis ANSP 3151

Anglo of

Basapophyses

( Degrees )

Length
(mm)

Height
(mm)

Width
( mm )

156°

8.5
8.5
8.5
8.5
5.5

22.5
23.0
22.0
21.0
15.5

25.0
27.0
24.5
25.5
20.0

153°

9.0
9.0

24.0
23.0

29.0
28.0

149°

9.0

21.0

25.5

143°

8.5
11.0

23.0
28.0

27.8
33.0

139°

7.0

19.0

23.0

138°

—

136°

11.0
7.0

25.0
16.5

35.0
20.0

132°

9.0
9.0

21.0
22.0

27.0
26.0

122°

9.5

22.0

27.0

117°

10.0
12.0

26.0
28.0

30.0
29.0

Bulletin Museum of Coniparative Zoology, Vol. 146, No. 1

Table 17. Angle of basapophyses, length, height, and width of
uintaensis compared with type specimens of synonymized taxa

IN figure 25

vertebrae of Amia
as illustrated

Relative
Vertebral
Number

Specimen

Angle of

Basapophyses

( Degrees )

Length

( mm )

Height
( mm )

Width
( mm )

P. sp. YPM 6242

102°

9.0

17.0

22.0

P. tiintaensis ANSP 3151

97°

10.0

23.0

28.0

P. laevis PU 10109

90°

13.0

29.0

33.0

P. laevis PU 10109

83°

14.0

28.0

30.0

P. media ANSP 5632

80°

8.0

16.0

18.0

P. sp. USNM 3963

P. symphijsis PU 10110

62°

11.0
6.0

14.0

13.5

Amia sp. ANSP 2338

A. macrospondyla NMC 6198

46°

13.0
12.0

26.0
25.0

23.0
22.0

P. laevis USNM 3968

6.5

22.0

18.0

P. sp. FMNH PF 4509

7.0

19.0

18.0

P. s/;. USNM 3966

7.0

19.0

18.0

P. medins USNM 3959

6.0

16.0

18.0

P. 7ne(/ii« USNM 3959

5.5

17.5

18.0

P. s/;. USNM 5450

7.0

15.0

13.5

P. »!«//»« USNM 3959

5.0

17.0

11.0

P. corsonii USNM 3961

4.0

11.0

10.0

bral column into region.s and types of verte-
brae appears to be a useful taxonomic
character of A. nintaensi.^.

The neural, aortal, and haemal facets do
not appear to vary much from those of
A. calva. The first six ventral aortal facets
show basically the same pattern for both
species (Figs. 11, 22). The angle of basa-
pophyses in A. uintaensis differs from that
of A. calva in two ways. The first six verte-
brae all have basapophyseal angles of 180
degrees, and it is not until the seventh
vertebra that these angles gradually begin
to decrease. Because of this more posterior
beginning in the decrease of the angles and
because there are fewer tinmk vertebrae, the
rate of decrease of the basapophyseal angle
is greater. These angles range from ISO
degrees anteriorly to approximately 45 de-
grees posteriorly, about the same as the
range for A. calva.

The intracolumnar variation in centrum
shape seen in the vertebral column of Re-
cent A. calva also occurs in A. uintaensis
( Fig. 14). In some respects the latter shares
certain characteristics with A. calva. The
first centrum is broad and thin, and usually
lacks basapophyses (Fig. 22). However,
centra between the fourth and twentieth
vertebrae begin to acquire an almost sub-
triangular outline, as opposed to the sub-
elliptical form of the A. calva trunk centra
(Fig. 12). The subtriangular shape may be
a function of greater size of the centra.

The chordal foramen is open in all known
vertebrae of A. uintaensis from the Paleo-
cene. Eocene, and Oligocene, but is often
filled with detritus during fossilization.
Estes (1964: 42) observed that Cretaceous
specimens as well as the Late Faleocene
specimen (PU 16236) had the chordal
foramen smoothly closed with bone. A re-

Fossil Amiids • Boreske

examination of PU 16236 reveals that the
chordal foramen is actually filled with fine
sediment rather than bone, so that the
character of the closed foramen can only be
applied to the Cretaceous specimens.
Chordal foramen position in all specimens
shows slight intracolumnar variation along
the tiimk as in A. calva, although occurring
more dorsally. In the caudal region there is
virtually no difference between the two
forms.

Leidy characterized "Protamia" uintaensis
on the basis of five centra and one basioc-
cipital. His height/ width proportions were
described in relation to those in an un-
diagnostic intracolumnar standardization of
the centra of the A. calva vertebral column.
My measurements of the anterior trunk
centra reveal that the holotypc ANSP 5558
has a width 1.3 times the height, and para-
type ANSP 8044 has a width 1.6 times the
height. Other paratype centra are posterior
trunk centi-a with width/ height ratios of
approximately 1:1. Romer and Fryxell
(1928: 521) described a displaced posterior
trunk centrum as having a height of 10 mm
and a width of 12.5 mm, about the same as
in ANSP 5558. Estes (1964: 43), in his
discussion of the height/ width proportions
of A. uintaensis centra, misinterpreted
Leidy 's (1873a, 1873b) diagnosis of "Proto-
mia' uintaensis and Romer and Fryxell's
(1928) diagnosis of "Paramiatus ^wleyij'
indicating that vertebrae of tlie former were
three times as wide as deep, those of the
latter two times. Estes was correct, how-
ever, in his assumption that there is intra-
columnar variation in height/ width ratios.

The general pattern of intracolumnar
variation in the A. uintaensis vertebral
column is quite similar to that of A. calva;
there is the same trend from horizontally
elliptical centra to circular or vertically
elliptical centra ( Fig. 14 ) . Thus the earlier
diagnoses of A. uintaensis using height/
width ratios that attributed the proportions
of the anteriormost trunk vertebrae to the
entire column are undiagnostic.

On the basis of isolated centra and skull
material, the most commonly used character

in differentiating A. uintaensis from A. calva
has been the former's greater size. How-
ever, the articulated specimen (PU 13865),
which is the smallest known A. uintaensis,
is only 146 mm longer than A. fra^osa
(FMNH 2201) and 16 mm longer than the
largest A. calva known to me (UMMZ
197683). Estes (1964) suggested that the
widening of the A. uintaensis vertebrae
might be a function of its greater size;
Gould's (1966) statement that internal
elements generally increase at allometric
rates to provide sufficient surface area to
maintain the external surface area offers a
partial explanation as to why the large A.
uintaensis vertebrae have greater width in
proportion to height than they do in smaller
amiid vertebrae.

Discussion

Two species of Protamia, one species of
Ilypamia, six species of Pappichthys, and
three species of Amia have been described
on vertebral characters from isolated centra
and disarticulated cranial elements (Table
19). With the exception of Amia ichiteaves-
iana, A. selwyniana, and A. macrosponclyla
from the Oligoeene Cypress Hills Forma-
tion of Alberta, all these taxa are based on
material from the Bridger Basin, Bridger
Formation, of Wyoming. Each of these 12
taxa will be re-evaluated in the following
discussion. Of the twelve species and four
genera, 'Trotamia" uintaensis (Leidy,
1873a) is the oldest name. Leidy's type
specimens are all trunk vertebrae. The
holotypc ANSP 5558 (Fig. 22) is approxi-
mately the sixth anterior vertebra and
displays the characteristic subtriangular
outline of other specimens. The paratypes
include trunk vertebrae ( ANSP 8044, 3151 ),
and a large basioccipital (ANSP 5622). The
holotypc vertebrae and the basioccipital
are considered diagnostic for Amia uintaen-
sis, on the basis of their possessing the
characteristic subtriangular vertebral out-
line, and a kidney-shaped articular surface
of the basioccipital.

Leidy (1873a) described Protamia media
from two large trunk centra from the

62 Bulletin Museum of Comparative Zoology, Vol. 146, No. 1

Bridger Formation of Wyoming. His main
criterion for distinguishing this form from
A. calva and from the other species of
"Protarnia" was that the vertebrae were
twice the size of A. calva vertebrae and
"somewhat smaller than Protamia uintaen-
sis" (Leidy, 1873b). The holotype USNM
2181 appears to be from the anterior trunk
region ( approximately the seventh or eighth
centrum, as suggested by its proportions
and configurations of aortal facets). The
basapophyseal angle is approximately 178-
180 degrees. The paratype ANSP 5632 is
from the posterior trunk region, with an 80-
degree basapophyseal angle, which is ap-
proximately equivalent to the twenty-ninth
centrum in A. uintaensis (Table 17; Fig.
25). Cope ( 1884, plate 4, figs. 7-20) figured
"PappiclitJiys mcdius" on the basis of 14
disarticulated centra from the same locality
(USNM 3959). Eight of these are from the
trunk region and correspond to centra
within the anterior to mid-trunk region of
A. uintaensis (Tables 15-16; Figs. 23-24).
The remaining six centra correspond to cen-
tra in the caudal region (Tables 16-17;
Figs. 24-25). Cope gave no description, but
in figuring these specimens he allocated to
them his own genus, emending Leidy's
(1873a) prior nomenclature. Both Leidy
and Cope had apparently assumed that the
characteristics of one or a few vertebrae
represented those of the entire coliunn.
Both species fall well within the size range
of A. uintaensis (Tables 14-17), and are
here considered synonyms of the latter.

Leidy (1873a) describc-d Uypamia ele-
gans from one small trunk vertebra. He
characterized this form as possessing a cen-
trum that was characteristically "short in
proportion with its breadth, and it presents
sutural impressions for a contiguous pair of
neural arches" (Leidy, 1873b). ANSP 5580
appears to be from the mid-trunk region,
comparable to approximately the nineteenth
centrum as suggested by its proportions and
configuration of aortal facets. The basapo-
physeal angle is 138-139 degrees. These
character-states and the small size are not

unique, occurring as the do in all the other
species of Amia; Hypamia elegans is there-
fore a nomen duhium.

Cope (1873) described Pappiclithys
plicatus from the anterior portion of a large
left dentary (AMNH 2539). Other type
material included two premaxillae, a right
quadrate, a left epihyal, an anterior portion
of an ectopterygoid, three trunk vertebrae,
and numerous fragments of angulars. He
characterized this form primarily on the
basis of dermal sculpture of the "cranial
fragments being roughly grooved." The
angular in A. uintaensis is generally
marked by more pronounced dermal sculp-
tiu-e than the other mandible elements.
His diagnosis of the vertebrae (USNM
3958) is based on proportions and mor-
phology of neural and aortal facets, both
of which correspond to various trunk ver-
tebrae in A. uintaensis (Tables 15-16;
Figs. 23-24). The description of the re-
maining elements conforms with other
elements of A. uintaensis. Pappichthys
plicatus is tlierefore a synonym of the latter.

Cope (1873) described Pappichthys
sclerops from a large left dentary. He
characterized this form as possessing a
dentary "more compressed and deeper" than
that in A. calva and other species of "Pap-
pichthys." The dentary (USNM 3965) in
all respects greatly resembles all dentaries
that have been referred to A. uintaensis, and
I regard Pappichthys sclerops as a synonym
of the latter.

('ope (1873) described Pappichthys
laevis from a large anterior dentary frag-
ment (USNM 3968). Other type materials
include a premaxillary fragment, fragments
of angulars (AMNH 2570), a left quadrate
fragment, a trunk vertebra fragment, and a
caudal vertebra. Although Cope distin-
guished this taxon from other species of
Pappichthys on vertebral proportions, vari-
ances in dermal sculpture, dentary alveolar
count, and obliqueness of alveolar face,
these character-states occur in A. uintaensis.
PappicJitJujs laevis is therefore a synonym
of the latter.

Fossil Amiids • Boreske

Cope (1873) described Pappichthys sym-
pJu/sis from two large fragments of trunk-
vertebrae and a iiral (USNM 3960). His
diagnosis rests primarily on eonfiguration
of neural faeets and basapophyseal length.
Osborn et cil ( 1878: 104) later reported two
eaudal vertebrae as cotypes (PU 10099,
10110). Cope (1873) described Pappich-
thys corsonii from 12 centra (USNM 5475-
5476), a basioccipital (USNM 5476), and
a left dentary fragment (USNM 3961). He
distinguished this form from Pappiclithys
sympliysis on different neural facet mor-
phology, basapophyseal length, and height/
width proportions. Merrill (1907: 14) cites
^'PappicJitJujs sympliysis = Pappichthys cor-
sonii' without further discussion. The cen-
tra of both forms conform to centra in the
vertebral column of A. uintaensis (Table
17; Fig. 25) and the characters assigned to
the dentary and basioccipital of Pappich-
thys corsonii are also found in A. uintaensis;
thus both P. sympliysis and P. corsonii arc
synonyms of A. uintaensis.

From the Early Oligocene Cypress Hills
Formation, Saskatchewan, Cope (1891)
described Amia wliiteavesiana from an an-
terior vertebra (NMC 6197), and Amia
macrospondyJa from a caudal vertebra
(NMC 6198). Both these forms were
founded on variations of vertebral charac-
ters (height/ width proportions, lack of
basapophyses, and chordal foramen posi-
tion) that are also represented in the verte-
bral column of A. uintaensis. The type
centrum of A. tiJiiteavesiana corresponds
approximately to the second anterior verte-
bra in A. uintaensis (Table 14; Fig. 22),
that of the type centrum of A. macrospon-
(Jyla with the thirty-first centrimi in A.
uintaensis (Table 17; Fig. 25). Prior to
the appearance of Cope's ( 1891 ) publica-
tion. Ami (1891), in his review of the
Cypress Hill fauna, mistakenly listed A.
whiteavesiana under the name A. selwyni-
ana. A. macrospondyhi and A. ivhiteavesi-
ana are here considered synonyms of A.
uintaensis; A. sehcyniana is a iiornen
nudum.

Comments on European and
Asian Forms

Janot ( 1967 ) described a large amiid,
Amia rohusta, from the Late Paleocene of
France, on the basis of disarticulated
material. She distinguished this form from
A. calva and A. russeUi on the angle of the
ventral border of the dentary face, and on
morphology of the parasphenoid tooth-
bearing surface in addition to other minor
morphological differences. Many of the di-
agnostic elements or associations on which
A. uintaensis is based, such as coronoid and
vomerine teeth, regional vertebral counts
and dorsal cranial elements, are missing in
her material. The elements that she does
figiue, however, closely resemble the com-
parative bones in A. uintaensis. Simi-
larities include rounded distal ends of
branchiostegal rays (also in A. fragosa),
subtriangular morphology of trunk verte-
brae, extensive surface of parasphenoid
tooth-patch, and shallow orbital notch in
frontal (also in A. scutata and A. calva).
These marked similarities suggest that A.
rohusta is a synonym of A. uintaensis.
Current work on the relationship of the
North American and European continents
in the Early Cenozoic ( McKenna, 1972)
indicates that they were connected until tlie
Early Eocene and that there is great sim-
ilarity between the Paleocene and Early
Eocene mammalian taxa at that time.
There is thus no zoogeographic problem in-
herent in synonymizing these two species.

Hussakof (1932) described Pappichthys
mongoliensis from disarticulated elements
from the Late Eocene Ulan Shireh beds of
the Shara Murun region. Inner Mongolia
(collected by the American Museum Cen-
tral Asiatic Expeditions.) At the time of
Hussakof's description, this collection
(AMNH 6372) represented the most exten-
sive material of ''Pappichthys." The collec-
tion includes numerous dentaries, maxillae,
three gulars, three opercula, three cleithra,
an hyomandibular, a supracleithrum, a
vomer, and trunk and caudal vertebrae.

64 Bulletin Museum of Comparative Zoology, Vol. 146, No. 1

Hussakof distinguished this form from A.
calva by the length of the dentaries and
the morphology of the operculum, and
from species of "Pappichthys" and "Pro-
tamia" on the basis of comparison of verte-
bral size. A comparison of the Mongolian
material with A. uintaensis shows some dis-
similarities, but there is still a closer affinity
between this form and A. uintaensis than
with the other species of Amia. The vomer
bears numerous sharp vomerine teeth; the
hyomandibular is deeply arched, and the
lingual face of the dentaries conforms to
that of A. uintaensis. The dentary, however,
is quite elongated anteriorly, the supra-
cleithrum is narrower, and the dorsal
border of the operculum is short and
ascends at a 30-degree angle rather than
being horizontal as in A. uintaensis (and
in other Amia species). The extrascapular
is narrow and tapered to a point rather than
flattened medially. Thus, although Pappich-
thys mongoliensis is similar to A. uintaensis
in many features and is clearly related to it,
it also differs in some respects. It un-
doubtedly belongs to the genus Amia, and
retention of all the Mongolian specimens in
Amia mongoliensis seems the most practical
alternative at this time. The Mongolian
higher vertebrate taxa indicate that the
Turgai Straits at least partially isolated
Mongolia from Europe during at least part
of the Cretaceous, Paleocene, and Eocene,
and that probably little exchange took place
until the Late Eocene (Szalay and Mc-
Kenna, 1971: 280-281). It may be possible
that A. mongoliensis evolved from A.
uintaensis during this migration.

Amia cf. uintaensis

Hypodigm. Cretaceous. Lance Forma-
tion, Wyoming: CM 256, YPM 6311, trunk
vertebrae; UCMP 56276, two fragments of
a single vertebra; UCMP 56277, one com-
plete vertebra, one vertebral fragment, one
left maxillary fragment. Hell Creek Forma-
tion, Montana: AMNH 6385, trunk vertebra;
MCZ 9334, dentary tooth tips. Aguja For-
mation, Texas: UMM collections, maxillary

fragment. Ojo Alamo Formation, New
Mexico: USNM collections, trunk vertebra.

Discussion

Cretaceous specimens of large amiids
occur in both Lance and Hell Creek for-
mations and consist mostly of isolated
and broken centra, and teeth that have
been identified primarily on the basis of
size. The characteristic subtriangular out-
line of the trunk vertebrae is even more pro-
nounced in these Cretaceous specimens,
wherein the lateral centrum walls between
the basapophyses and the aortal facets are
concave ( Fig. 26 ) . The chordal foramen is,
as Estes ( 1964: 42) noted, closed with bone,
as are one-third of the vertebrae referred to
A. fragosa from the Lance Formation. How-
ever, Estes observed lateral concavities be-
tween the neural facets and basapophyses
in a large vertebral centrum (AMNH
6385) from the Hell Creek Formation
(mistakenly cited by him as AMNH 6835
from the Oldman Formation of Alberta).
Estes apparently confused neural with
aortal facets and thus figured the vertebra
upside down. Correct orientation of the
centrum (Fig. 26) shows concavities be-
tween the basapophyses and the aortal
facets. Thus, Estes was incorrect in con-
cluding that A. fragosa, A. calva, and the
Eocene specimens of A. uintaensis "also
seem to lack the concavity between the
'basapophysis' and neural arch present in
the large Cretaceous specimens." Two
other specimens from the Lance Formation
(YPM 6311, CM 256; Fig. 26) also show
the prominent concavities between the
basapophyses and aortal, rather than neu-
ral, facets. In addition to the vertebrae,
Estes described a maxillary fragment as
being larger and more robust than that of
A. fragosa, although "characteristically
amiid in tooth implantation and general
shape." A more complete maxillary frag-
ment (UMM collections) from the Aguja
Formation ( Big Bend National Park,
Brewster County, Texas) conforms with
Estes' (1964) description.

Fossil Amiids • Borcnke 65

Hlii

Fig. 26. Comparison of different Cretaceous vertebrae. Am'ia cf. uinfaensh: A, anterior trunk vertebra, CM 256,
Lance Formation, Wyoming; B, posterior trunk vertebra, AMNH 6385, Hell Creek Formation, Montana; C, mid-trunk
vertebra, YPM 6311, Lance Formation, Wyoming. Chondrichthyes: D, E, G (thin section), trunk vertebrae, FHKSCM
13024-9, Black Creek Formation, North Carolina; F, trunk vertebra, MCZ 12879, Peedee Formation, North Carolina.
Cetacean: H, caudal vertebra, FHKSCM 13025, Calvert Formation?, North Carolina.
1 rz dorsal, 2 ^ articular surface, 3 ::= ventral

Bulletin Museum of Comparative Zoology, Vol. 146, No. 1

Only three new centra and a maxillary
fragment have been identified since Estes'
(1964) study. The vertebrae, as noted
above, differ in certain minor respects from
the Paleocene and Eocene specimens.
Whether or not this material actually repre-
sents A. uintaensis or an earlier stage of
evolution can only be determined when
more complete Cretaceous material is avail-
able.

Amia scutata Cope, 1875

Amia dictycephala Cope, 1875: 3.
Amia exilis Lambe, 1908: 12.

Holotijpe. USNM 5374, incomplete spec-
imen lacking the head and body anterior to
the middle of the dorsal fin; anal and part
of dorsal and caudal fins well preserved.

Type locality arul horizon. Florissant,
Colorado. East half of section 2, T 13 S,
R 71 W, Teller County, Colorado; Flor-
risant Formation.

Age Range. Chadronian (Early Oligo-
cene) to Orellan (Middle Oligocene).

Hypodigm. Oligocene. Cypress Hills
Formation, Saskatchewan: NMC 6200, 6205,
vertebrae; NMC 6201, basioccipital. Chad-
ron Formation, South Dakota: PU 17172,
left dentary with posterior coronoid bearing
teeth, and a trunk vertebra. Lower Brule
Formation, South Dakota and Nebraska:
FMNH PF4508, PF4509, CM 3814, verte-
brae; FMNH PF4506, right vomer bearing
teeth. Florissant Formation, Colorado: PU
10172, nearly complete specimen (counter-
part = YPM 6243, anterior half; USNM
4087, caudal half); YPM 6241, complete
caudal region (with counterpart); UMMP
V-57431, nearly complete specimen; USNM
3992, partial specimen, lacking skull and
tail; AMNH 2802, nearly complete skiill;
AMNH 2670, partial specimen, lacking
skull and caudal region; AMNH 2671,
caudal region.

Known distribution. South Dakota, Ne-
braska, Colorado, and Saskatchewan.

Revised diagnosis. Vertebral meristics
similar to those of A. calva, but head/
standard-length proportion is intermediate
between that of A. uintaensis and A. calva.

Extrascapular thicker at distal end than in
A. calva, with concave posterior border.
Pterotic more similar to that of A. uintaen-
sis than of A. calva; anterior portion narrow
and extended laterally to the frontal. Or-
bital excavations more marked than in A.
calva, but not as deep as in A. uintaensis or
A. fragosa. Preoperculum resembles that of
A. uintaensis more than that of A. calva,
being narrower dorsally than ventrally.
Symphyseal incurving of the dentary less
than in A. calva, but greater than in A.
uintaensis. Ventroposterior process of
cleithrum heavily sculptured as in A. fragosa
and A. uintaensis. Infraorbital 4 larger
than infraorbital 5 as in A. fragosa and A.
uintaensis. Ossification of cranial bones ex-
tensive as in other fossil species, greater
than in A. calva. Greatest known standard-
length 390 mm.

Introduction

Cope's (1875: 3) description of Amia
scutata is based on a specimen lacking the
head and body anterior to the middle of
the dorsal fin, from the Middle Oligocene
Florissant Fomiation near Florissant,
Colorado. He distinguished this form
from Amia dictyocephala (found in the
same deposit; Cope, 1875) and Amia calva
by its larger scales "of which only seven
and a half longitudinal rows are visible
above the vertebral column." Cope de-
scribed A. dictyocephala from two partially
complete specimens lacking skidls and
caudal fins (USNM 3992, AMNH 2670),
two complete caudal regions (AMNH 2671,
USNM 4087), and a nearly complete skull
(AMNH 2802); Osborn et al. (1878) later
described another specimen of A. scutata
from the same deposit. Tliis specimen
was more complete, consisting of an axial
skeleton and a crushed skull. They believed
A. scutata to be a valid form, differing from
A. calva in having a proportionately larger
head.

Comparison of known specimens of A.
scutata revealed that the counterparts to
the specimen described by Osborn et al.
( PU 10172 ) were separated and sold to two

Fossil Amiids • Borcskr 67

Fig. 27. A, Amia scutata UMMP V-57431; B, A. scufafa PU 10172; C, A. "dictyocephala" USNM 3392; D, A.
"dicfyocepbala" AMNH 2670.

different miiseiims. The caudal portion of and is one of the paratypes used by Cope
the counterpart was found in the National (1875) in his description of A. dictijo-
Museum of Natural History (USNM 4087) cephala. The anterior region was found

Bulletin Museum of Comparative Zoologij, Vol. 146, No. 1

unlabeled at the Yale Peabody Museum
(YPM 6243; Plate 4).i

In 1967 another nearly eompk^te speci-
men was discovered from the same deposit
(Fig. 27A) and Cavender (1970: 42) re-
ported the specimen A. dictyocepliala as
differing from A. calva in having a larger
infraorbital 4, in the sculptin-e of cleithrum,
and "by its proportionately larger head and
orbit, and somewhat shorter body."

Fossil Record

Other than the Florissant Formation, the
only deposits from which elements of A.
scutata can be identified are the Cypress
Hills Formation of Saskatchewan, Chadron
Formation of South Dakota, and the Lower
Brule Formation of South Dakota and
Nebraska. Becker (1961: 38) reported
amiid scales (UMMP collections) from
the Late Oligocene Passamari Formation
and Middle Oligocene Grant Horse Prairie
Shale of Montana (Becker, 1962). Since
no specific characters for scales of Amia
have yet been determined, it is best to
allocate this material to Amia sp. Skinner
et al. (1968: 415) has reported Amia .sp.
vertebrae (F:AM 42947) from the Early
Miocene Turtle Butte Formation of South
Dakota. Only two specimens were found;
since the vertebrae of A. scutata and A.
calva are morphologically and meristieally
similar. Skinner et al.'s identification is the
only possible one at this time. The strati-
graphic range of A. scutata is therefore lim-
ited to the Early and Middle Oligocene.

Description

Neurocranium.. The basioccipital (PU
10172, NMC 6201 ) is similar to that of A.
calva. The only available parasphenoid
(PU 10172) is poorly preserved, but closely
resembles that of A. calva in length and
position of ascending processes.

The extrascapular in A. scutata differs
slightly from that of A. calva in that the
distal end is relatively thicker and the

1 The counterparts ( USNM 4087, YPM 6243) to
PU 10172 have been .subse(iuently accjuired by the
Museum of Natural History, Princeton University.

L< orbital length
D- orbital depth
^a^dermosphenotlc angle

A. fragosa D/L=O.I76mn.

A. uintaensis D/L=O.I55mn.

18°

Z^"! 40 = I34<'

A scutata D/L=O.I32mn.

I5«

A calva D/L"O.IOOmn.

15°

IB"

40=145°

Z4»r 4.0=135°

A. cf. scutata O/L'0.121

[■0=137°

Fig. 28. Orbital dimensions of ^m\a spp.

posterior lappet is less pronounced; also,
the posterior border is more convex (Fig.
15). As in A. calva, however, the proximal
anterior corner is squared off, and the
medial suture is relatively long. The
pterotic in Ax. scutata resembles that of A.
uintaensis more than that of A. calva in
general morphology, since the anterior half
is narrower than the posterior half; in the
Recent species the ends are nearly sym-
metrical. The anterior border extends fur-
ther laterally than in A. calva, and, as in
A. uintaensis, adjoins the distal lateral side
of the frontal, rather than the posterior
border as in A. calva. The dermosphenotic,
parietal, frontal, and nasal of A. .scutata
conform to these bones in A. calva. The
parietal/ frontal ratio is marginally within
the lower limit of the range of A. calva
(Table 7). The orbital excavation in the

Imxssil Amuus • liorcske

frontal (Fit:;. 2cS) is greater than in A. caha
bnt less than in A. jruii^osa or A. uintaensis.
Snprascapulars, antorbitals, and rostrals are
not preserved.

The laerinial is similar to that of A. calva,
b(>arint!; a posterior noteh for the reeeption
of infraorbital 2, but in A. scutata the laeri-
nial is more robust. Infraorbital 2 and infra-
orbital 3 are similar to these bones in A.
calva. Infraorbital 4 is more massive pos-
teriorly than in A. calva; it exeeeds infra-
orbital 5 in dorsoventral length, and the
posterodorsal corner, which in A. calva is
markedly acute is, in A. .scutata, more
squared off. This bone more closely re-
sembles that of A. jraii^osa: it is not avail-
ablc> for comparison in A. uintacnsis. Infra-
orbital 5 is less massive posteriorly than in
A. calva; in this feature it resembles that of
A. fra'^osa. It is also, as in A. fra^osa and A.
uintacnsis, dec>per anteriorly than in A.
calva.

Branchiocranium. The supramaxilla in
A. scutata is elongated and tapered to a
point anteriorly, with a relatively straight
ventral border as in A. calva. It is slightly
longer and more robust posteriorly, the
posterodorsal border being higher and less
obliquely curved than in A. calva, A. uin-
tacnsis, and A. jra<^osa. The preniaxilla
resembles that of A. calva. The maxilla is
wider posteriorly and more ossified anteri-
orly than that of A. calva, but otherwise
agrees with the bone in the Recent .species.

Dermopalatine, autopalatine, entoptery-
goid, ectopterygoid, metapterygoid, and
vomer are not pr(\served. Ho\v(^ver, conical
vomerine teeth are displayed on PU 10172,
and resemble those of A. calva rather than
those of A. fra<s,osa or A. uintacnsis. The
relative munber of teeth and their extent
on the vomer cannot be discerned.

As in A. calva and A. uintacnsis, the den-
tary in A. scutata lacks the dorsal shelf
adjacent to the lingual border of the alveo-
lar ridge seen in A. fra^osa (Fig. 18). The
bone is very thick, especially toward the
mid-lingual surface, where the dorsal and
ventral halves meet to form the Meckelian
groove. As in A. uintacnsis the upper wall

of this groove is primarily formed bv the
thickness of bone in dorsal half of the lin-
gual surface; the ventral half is barely
overlain by the dorsal half. The first coro-
noid does not extend past the Meckelian
groove, and bears sharp conical teeth, as in
A. calva (Fig. 18). The anterior half of the
dentary is more incurved than in A. uintacn-
sis, but not to the extent that it is in A.
calva. The anterior width of the dentary
also resembles that of A. calva and A. uin-
tacnsis in that it is evenly tapered almost to
the symphyseal edge. The angular and sur-
angular are similar to comparable bones in
A. calva, except that they, like the dentary,
are more extensively ossified. The mandi-
ble/head-length ratio in A. scutata is well
within this ratio range for A. calva (Table
7). The prearticular is not prc\served.

The preoperculuin is similar to that of
A. uintacnsis, being narrower dorsally dian
ventrally, radier than having both halves
rc^latively equal in widdi, as in A. calva.
The operculum resembles that of A. calva
in morphology and opercuhun-depth/oper-
culum-length (Table 7). The suboperculum
and interoperculum resemble those of A.
calva in general morphology, but the suture
between them is longer anteroposteriorly.
The branchiostegal rays are squared off
distally, as in A. calva.

Post-cranial skeleton. The supracleithnmi
and metacleithriun are not preserved. The
only part of the cleithrum available for
study is the ventroposterior process in
UMMP V-57431 which in A. calva is the
only area of this bone that is visible ex-
ternally. This region of the cleithrum in
A. scutata is heavily sculptured (Fig. 21),
and as in A. uintacnsis and A. fra^osa, this
d(Tmal ornauKMitation extends to the edge
of the bone. In A. calva, diis d(>rmal struc-
ture is limited to the cent(>r and dorsal re-
gion of this part of the cleithrum.

Th(> vertebral column of A. scutata re-
sembles that of A. calva both in number of
ccMitra (Table 9) and in general morphol-
ogy of the centra. The head /standard-
length proportion (0.312) is greater than in
A. calva (0.271 ), but less than in A. uintaen-

70 Bulletin Museuui of Comparative Zoology, Vol. 146, No. 1

sis (0.322). The insertion of peetoral fin/
standard-length and insertion of anal fin/
standard-length ratios nrv both within the
ranges of A. calva, althongh the latter pro-
portion for A. scutata is somewhat greater
than the mean for A. calva ( F'ig. 31 ).

Discussion

In the same paper as his deseription of
Aryiia scutata. Cope (1875: 3) described
Amia clictyocepliala, also from the Florissant
Formation. A. dictyocephala was distin-
guished from A. scutata by having 10 to 12
supravertebral scale rows, and 35 vertebrae
between the anterior dorsal fin pterygio-
phore and the posterior anal fin ptervgio-
phore (USNM 3992 AMNH 2670). "lie
further characterized this form from a skull
(AMNH 2S02) that "possesses twelve
branchiostegal rays, and a relatively smaller
orbit than in Amia calva." A re-examination
of these specimens in the previous section
on meristics showed that Cope's supra-
vertebral scale row count was in error, and
there is no perceptible difference in this
feature between Recent and fossil Amia
species (Table 8). In A. calva, the range
for the number of centra between the in-
sertion of the dorsal fin and the terminus
of the base of the anal fin is 33-37. In the
type specimen of A. dictyocephala (USNM
3992) the number of centra is 35, and the
mean number in specimens of A. scutata is
36; there is clearly no way that this fcnitiue
can be used to distinguish A. dictyocephala
from A. scutata and A. calva. Cope, on the
basis of AMNH 2802, thought that an orbit
in A. dictyocepJiala was smaller than one in
A. calva, but the small size was due largely
to the constriction of the orbit that resulted
from crushing of the dcrmosphenotic and
upward displacement of infraorbital 5. The
characters that Cope used to differentiate
A. dictyocephaki from A. scutata are un-
diagnostie, and my studies of the specimens
show no morphological or meristie differ-
ence; A. dictyocepJiala is here considered
to be a synonym of A. scutata.

Lambe (1908: 12-13) described Amia
exilis from a single basioccipital (NMC

6201 ) and two mid-trunk \ c>rtebrae ( NMC
6200, 6205) from the Farly Oligocene
Cypress Hills Formation of Saskatchewan.
The temporal occurrence of these elements
is equivalent to that of A. scutata. Lambe's
description of the basioccipital conforms to
that of A. scutata in being more extensively
o.ssified than in A. calva. His diagnosis of
the two centra is founded on height/ width
proportions, ehordal foramen position, basa-
pophyseal angle, and configuration of nc>ural
facets. Because A. scutata resembles A.
calva in vertebral morphology, the charac-
ters that Lambe uses to distinguish A. exilis
are undiagnostic; I therefore consider A.
exilis iis a synonym of A. scutata.

Amia cf. scutata

Hypodi^m. Miocene. Pawnee Creek
Formation, Colorado: UCMP 38222, nearly
complete cranial roof, infraorbitals 4 and 5,
nearly complete anterior portion of palate,
two branchiostegal rays, maxillae, and right
dentary.

Description

The general morphology of the cranial
roof resembles both A. scutata and A. calva
in parietal/ frontal ratio (Table 7), rectan-
gular parietals, and shape of dermosphen-
otic and nasal (Fig. 29). The extrascapular
more closely resembles that of A. scutata in
its greater width and less pronoimced distal
posterior lappets. The pterotic also resem-
bles that in A. scutata in its being narrower
anteriorly than posteriorly, and in bordering
die frontal laterally rather than posteriorly.
The size and depth of the orbital excavation
is intennediate between that of A. scutata
and A. calva (Fig. 28). The maxilla is
similar to that of A. calva, being less robust
posteriorly than that of A. scutata. The
branchiostegal rays are squared off distally,
as are tliose of both A. calva and A. scutata.
Infraorbital 4, although posteroventrally in-
complete, is clearly closer to that of A. scu-
tata than A. calva in being relatively larger
than infraorbital 5, and in the posterodorsal
corner being squared off rather than acute
as in A. calva. Infraorbital 5 resembles that

Fossil Amiids • Boreske 71

of A. scutata in size relative to infraorbital 4,
the anterior end bcnng narrower than in
A. scutata; this featnre eontributes to lessen-
ing the relative width of the orbit. The
dentary resembles that of A. scutata in
being wider anteriorly than in A. calva; the
dorsal lingual surfaee only slightly overlaps
the ventral lingual siuface as in A. scutata
(Fig. 18); Meckel's groove is thus similar
to that of A. scutata. There is no available
palate in A. scutata for comparison. The
number of vomerine teeth is 18 and 21,
which is bracketed by the range for A.
calva (Estes and Berberian, 1969: 5). As
Estes ( 1964 ) noted for this specimen, these
teetli are sharper and more incurved ex-
ternally than internally; this disparity is

more distinct in this form than in the extant
species. The hyomandibular, entopterygoid,
ectopterygoid, dermopalatine, and pre-
maxilla are poorly preserved, but appear
to resemble these bones in A. calva.

Discussion

Estes (1964: 36) and Estes and Tihen
(1964: 454) referred to this specimen as
Amia sp. (and in error gave the source as
White River Formation). The .specimen
resembles A. scutata in some elements, A.
calva in others, and is intermediate in
several character-states, notably bone thick-
ness and size of orbits. It does, however,
appear to show a stronger resemblance to
A. scutata than to A. calva, particularly in

Fig. 29. Amia cf. scutata DC 38222, Late Miocene, Pawnee Creek Formation, Colorado.

72 Bulletin Museum of Comparative Zoology, Vol. 146, No. 1

the morphology of the extrascapiilar, ptcro-
tic, dentary, and infraorbitals 4 and 5, and
I have thus compared it with the fossil
species. Since this is a form that is both
morphologically and temporally interme-
diate between A. scutata and A. calva, it is
difficult to determine whether or not this
specimen actually represents A. scutata or
a later stage of evolution leading to A. calva,
but it is at least of interest in documenting
the slow phyletic development toward A.
calva in mid-Cenozoic time.

Amia cf. calva

Hijpodi^m. Pliocene. Lower Valentine
Formation, Nebraska: UCMP 65851, an-
terior portion of left dentary and a trunk
vertebra; UMMP 521S7, right nasal, ectop-
terygoid fragment, unidentified cranial
fragments; UMMP 421S5, right dentary
fragment. Ogallala Formation, Kansas:
UMMP 55574-55578, three right and two
left dentary fragments; UMMP 55579, in-
complete right cleithrum; UMMP 55583, a
right extrascapular; UMMP 55580, a right
maxilla; UMMP 55585, a left premaxilla;
UMMP 55586, several scales.

Discussion

Smith (1962), and Estes and Tihen
(1964) described as Amia sp. a nasal and
dentary, and cranial fragments from the
Lower Valentine Formation, Nebraska.
Wilson (1968) described as Amia calva
denatry fragments, a premaxilla, a maxilla,
an extrascapular, an incomplete cleithrum,
and several scales from the Ogallala Forma-
tion, Kansas. This Early Pliocene material
resembles A. calva more closely than does
the Miocene A. cf. scutata specimen noted
above; the elements are very lightly ossified
as in the Recent species. The cleithrum is
distinctly A. calva-\ike in its lack of distal
marginal dermal sculpture. The dentary
fragments are also thinly ossified as in A.
calva, but are slightly wider relative to the
dentary in the Recent species, as in the
Miocene form. Temporally, this Pliocene
material is later than the Miocene form and
earlier than A. calva; morphologically, how-

ever, the available elements conform with
A. calva.

Amiidae incertae sedis

Hypodi^m. Cretaceous. Paluxy Forma-
tion, Texas: SMUSMP 62270, dentary frag-
ments, premaxillary fragment, vertebrae,
maxillary fragments, and an unidentified
palatal bone bearing teeth; FMNH 7050,
basioecipital; FMNH 7051, mid-trunk ver-
tebra; FMNH 7052, anterior trunk vertebra
fragment; FMNH 7053-7054, anterior trunk
vertebrae; FMNH 7055, caudal vertebra;
FMNH 7056, small vertebrae; FMNH 7049,
unidentified palatal bone bearing teeth.

Description

The dentaries are fragmentary (Fig. 30);
the only diagnostic features available for
comparison with other amiid forms are
related to the anterior region of the dentary.
The surfaces pits on the exterior side of the
dentary are relatively larger and deeper
than in any species of Amia. The dentaries
lack the dorsal shelf adjacent to the lingual
side of the alveolar ridge seen in A. fragosa.
The coronoid articulation surface descends
directly from the alveolar ridge, as in a Uro-
cles dentary from the Late Jurassic (Pur-
beck) of England (BMNH 48236). The
lingual surface above the Meckelian groove
is relatively short, even more so than in
Amia uintaensis, and the groove itself is
quite wide, more so than in BMNH 48236.
The anterior portions of the dentaries are
relatively straight, rather than incurved as in
Amia fragosa, and are evenly tapered to the
symphyseal edge. The dentary and pre-
maxilla teeth are broken, but in dorsal view
the interior surfaces of the broken teeth are
very even, lacking the serrated outline seen
in other species of Amia. Only the anterior
portion of the premaxilla is present; it bears
nine alveoli, conforming in this respect with
all Am,ia species. The premaxilla, although
incomplete, displays the anterior (ventral)
edge of the large foramen that is character-
istic of Amia. Only part of the anterior
maxilla is present in the specimens avail-
able, and since the more diagnostic aspects

Fossil A muds • Boreske 73

occur posteriorly, it is difficult to determine smaller fragment (SMUSMP 62270) bears

any affinities witli particular specic\s; the pillar-shaped teeth with nipple-like tips, as

anterior portions that are available gener- in the tooth-bearing palatal bones in species

ally conform with those of Amia. The of Amia. Posterior to the spinal arterial

specific bones to which the palatal frag- foramina the basioccipital includes one

ments belong cannot be identified. The fused vertebra. As in Amia fra^osa and

Fig. 30. Amiidae incerfae sedis, Early Cretaceous, Poluxy Formation, Texas: A'-A-, anterior portion of left den-
tary; B, premaxlllary fragment; C, anterior portion of rigfit maxilla; D, unidentified palatal fragment. XO-15

74 Bulletin Museum of Comparative Zoology, Vol. 146, No. 1

Amia caJva, the basioccipital has an ovoid
articular surface with no dorsal indentations
between the neural facets. The large verte-
brae are thickly ossified, as in the Creta-
ceous specimens of Amia cf. uintaensis.
The chordal foramina are closed and the
only available large mid-trunk centrum dis-
plays the pronounced triangular outline
characteristic of Amia uintaensis. None of
the large vertebrae display the character-
istic Amia aortal facets; they do, however,
possess neural facets, and the mid-trunk
centra bear basapophyses. The small verte-
brae are also thickly ossified and the
chordal foramina of the trunk vertebrae are
closed. As Traquair (1911: 39) noted for
Amiopsis cloUoi, the lateral sides of the
vertebrae are marked by a number of vari-
able excavations, or "oval fossae" (Fig. 7).
These smaller mid-trunk vertebrae, unlike
the large ones, display both aortal and
neural facets, as well as basapophyses and
lateral oval fossae.

Discussion

Thurmond ( 1969 : 88 ) reported "various
fragments of an undetermined amiid" from
the Paluxy Formation of Texas, which is
the earliest known occurrence of amiids in
North America. He further noted that
amiid material occurred both in freshwater
and marine zones and that a further descrip-
tion of this material would be the subject
of a later study. He was uncertain as
to whether the amiids occurring in the
marine zones were actually marine or were
freshwater forms secondarily deposited in
the marine areas. None of the material can
be referred to Atnia since it displays charac-
teristics of Amia uintaensis, Amia fragosa,
Urocles, and Amiopsis, as noted in the
above description. The vertebrae suggest
the possibility of more than one form: the
large vertebrae are subtriangular and re-
semble Am'a uintaensis in morphology,
with the exception of the lack of aortal
facets on the trunk vertel:)rae. The small
vertebrae are Amia frag,osa-\ike in morphol-
ogy; they possess aortal facets, but also
display lateral oval fossae characteristic of

Amiopsis. In reviewing the European Juras-
sic and Cretaceous Urocles, Lange (1968)
found little morphological justification to
warrant continued generic distinction be-
tween Urocles and species described by
Woodward ( 1916 ) as belonging to Amiop-
sis from the Purbeck Beds near Wevmouth,
Dorset. Lack of knowledge of the skull of
Amiopsis makes it impossible to compare
cranial elements with those of other amiids;
the singular postcranial feature charac-
terizing Amiopsis is the lateral o\'al fossae
of the v^ertebrae. Although Lange suggests
that both Amiopsis and Ainia evolved inde-
pendently from different Urocles species-
groups, it is premature to attempt to do
more than indicate morphological similar-
ities or dissimilarities since the phylogenetic
relationship of Amiopsis with Urocles or
Amia cannot be clearly defined until a
much-needed review of the taxon has been
completed, and until more Amiopsis mate-
rial is made available for study.

The Paluxy material shows resemblances
to two early Aryiia species, Amia uintaensis
and Amia fragosa, as well as to the Late
Mesozoic European amiids, Urocles and
Amiopsis. Whether the Paluxy material
represents one or more forms intermediate
between Atnia and Urocles (or Amiopsis)
or whether it belongs to some other group
of amiids that became extinct before the
end of the Cretaceous cannot be deter-
mined, since taxonomic evaluation of this
material is limited by the lack of articulated
specimens.

SPECIMENS REMOVED FROM
THE AMIIDAE

Miller (1968: 468-470, pi. 1, figs. 1, 3,
7-9) questionably identified as Protamia
sp. one large (FHKSCM 13025) and three
small centra (FHKSCM 13024-9) recovered
from a channel sandstone cut into the Up-
per Cretaceous Black Creek Formation,
Phoebus Landing, North Carolina. Since
all known Ainia are freshwater forms and
since these centra were associated with vari-
ous marine vertebrates. Miller (1968: 467)
concluded that the channel sandstone con-

Fossil Amiids • Boreske

tained a mixed fauna, "the ehannel sand-
stone formed in an estuarine or tidal en-
vironment."

My studies indicate that these specimens
are not amiid. The smaller vertebrae are
horizontally ovoid. A. uintaensis trunk cen-
tra have concavities between the basa-
pophyses and aortal facets ( Fig. 26, A-C ) .
A thin section (Fig. 26, G) through the
articular surface of one of the Nortii Caro-
lina specimens (FHKSCM 13924-9) has a
radial structure resembling that of S(juatimi
and other sharks (Hasse, 1882, tables 17-
18). All layers are laminated parallel to the
exterior surface and are crossed by various
perpendicular vascular foramina. Their
articular surfaces are slightly concave, while
those of Ainia are markedly so. Each of the
small vertebrae bear horizontal basapoph-
yses as in Recent Sqiialus, and are best
referred to the elasmobranchs.

The large vertebra is a cetacean caudal
(Fig. 26, H), possibly belonging to the
Cetotheriidae (Clayton Ray, 1971, personal
communication). The centrum is ovoid,
with very slightly concave articular surfaces,
and lacks a chordal foramen, as well as
ventral facets. The dorsal facets for the
accommodation of metapophyses are well
defined. Since this centrum is from a ma-
rine mammal, it is more probably from the
Miocene (Calvert Formation?) than from
the Cretaceous Black Creek Formation.

Eastman (1899) described Amiopsis dar-
toni from a partial opercular series, pectoral
fin, and associated cycloid scales from the
Late Jurassic marine Sundance Formation,
South Dakota. Eastman felt that the many
"stout ribs" associated with the pectoral fin
suggested a well-ossified Aniia-hkc verte-
bral column and the semicircular operculum
conformed with that of A. ctilvu. Since the
scalers are covered supt^rficially with ganoine
and appear elliptical, Eastman placed this
form among the Amiidae. He allocated the
generic name, Amiopsis, on a temporal
basis. According to Bobb Schaeffer, (1971,
personal communication) the holotype
(USNM 4792) and the paratypes (MCZ
9696, USNM 4793) are to be tentatively

referred to the Leptolepidae on the basis
of morphology of opercular series and pec-
toral fin lepidotrichia. Schaeffer is currently
studying the Late Jurassic North American
fishes and is including a more extensive
discussion of this material in his review.

SUMMARY AND CONCLUSIONS

This survey of the osteology, morpho-
metries, and meristics of the North Amer-
ican fossil amiids indicates that the extant
and fossil forms fall into foiu- groups
worthy of specific status: ( 1) Amia fragosa,
(2) A. uintaensis, (3) A. scutata, and (4)
A. colvci. All these forms, excepting A.
fragosa, have somewhat elongated bodies
(approximately 85 centra) and shai-p,
conical coronoid and palatal teeth. Al-
though the coronoid and palatal teeth of
A. uintaensis are more sharply curved in-
wardly, the teeth are closer in morphology
to those of A. scutata and A. calva than
to the styliform teeth of A. fragosa. A.
uintaensis, A. fragosa, and A. scutata all
have a larger infraorbital 4 than infraorbital
5, greater degree of ossification of cranial
elements, deeper orbital notch in the frontal,
greater head/ standard-length, and generally
larger parietal /frontal ratio. These charac-
ter-states clearly set the fossil species of
Amia apart from the Recent A. calva.

Articulated specimens have yielded more
information on the osteology of A. fragosa.
A. fragosa is a short-bodied form (approxi-
mately 65 centra) with a smaller number of
caudal lepidotiichia than in the other
species of Amia, styliform palatal and coro-
noid teeth, deeper orbital excavation in the
frontals, square parietals, and a short box-
like skull having relatively short mandibles
diat occupy about half the head-length. The
styliform crushing palatal teetlr of A. fragosa
suggest a durophagous habit, rather than
the more predaceous habit indicated by the
sharp palatal teeth of A. uintaensis, A.
scutata, and A. calva. Although it is known
that A. calva includes molluscs and crusta-
ceans in its diet, perhaps A. fragosa was
more exclusively adapted for shell crushing
than the Recent species.

76 Bulletiti Museuin of Comparative Zoologtj, Vol. 146, No. 1

Fig. 31. Skull and body structure of A, Amia calva; B, A. scufafa; C, A. uinfaensis; and D, A. fragosa.

Fossil Amiids • Boreske 77

PLEISTOCENE

PLIOCENE

POST-BLANCAN

BLANCAN-

HEMPHILLIAN

CLARENOONIAN

FOSSIL LAKE BEOS
(lOAHO FM.)

WAKEENEY It.
(OGALLALA FM.)

LOWER VALENTINE FM.

BARSTOVIAN

MIOCENE

HEMINGFORDIAN
ARIKAREEAN

EUBANKS l.f.
(PAWNEE CREEK FM.)

TURTLE BUTTE FM.

WHITNEYAN

OLIGOCENE

ORELLAN

CHAORONIAN

RUBY PAPER SHALE
(PASSAMARl FM.)

GRANT HORSE PRAIRIE SHALE

FLORISSANT FM.

ORELLA MEMBER
(BRULE FM.)

CHADRON FM.
CYPRESS HILLS FM.

OUCHESNEAN

EOCENE

UINTAN

BRIDGERIAN

WASATCHIAN

CLARNO FM.
HORSEFLY RIVER BEDS

UINTA FM.
WASHAKIE FM.

BRIDGER FM.
WIND RIVER FM.

FOSSIL LAKE BEOS
(GREEN RIVER FM.)

6OLOEN VALLEY FM.

WASATCH FM.

GRAYBULL BEDS
(WILLWOOO FM.)

CLARKFORKIAN

TIFFANIAN

PALEOCENE

TORREJONIAN

PUERCAN

MAASTRICHTIAN

CRETACEOUS

CAMPANIAN

ALBIAN

BEAR CREEK l.f,
(FORT UNION FM.)

SILVER COULEE l.f.
(FORT UNION FM.)

MELVILLE FM.
SAUNDERS CREEK l.f.
(PASKAPOO FM.)

CEDAR POINT QUARRY l.f.
(FORT UNION FM )

MEDICINE ROCKS l.f.
(TONGUE RIVER FM.)

ROCK BENCH l.f.
(FORT UNION FM4

TULLOCK FM.

MANTUA If.
(FORT UNION FM.)

HELL CREEK FM.
LANCE FM.
OJO ALAMO FM.
AGUJA FM.
EDMONTON FM.

JUDITH RIVER FM.
"MESAVERDE" FM.

OLDMAN FM.

BUTLER FARM l.f.
(PALUXY FM)

Table 18. Major deposits containing remains of Amia in the

WESTERN interior OF THE UnITEU StATES AND CaNAUA

78 Bulletin Museum of Comparative Zoology, Vol. 146, No. 1

Seven genera and twenty-three amiid
species (Table 19) have been described in
the literature. Estes (1964) synonymized
Stylomyleoclon lacus with Kindleia fra^osa,
and Estes and Berberian ( 1969 ) referred
the genus Kindleia to Amia, thereby con-
firming the suggestion of Janot (1969).
Paramiatus p,iirleyi (Romer and Fryxell,
1928) is unquestionably a synonym of A.
fragosa. Regardless of possible synonymy
with European taxa, the sti'atigraphic
range of A. fragosa is remarkably long,
extending as it does from the Late Creta-
ceous through the Middle Eocene. Al-
though A. fragosa is better known than the
other fossil species, and was extensively
described by Estes (1964), O'Brien (1969),
and Estes and Berberian (1969), its phylo-
genetic relationship to them and to A. calva
could not be understood without compara-
tive information on both the other fossil
forms and A. calva (Fig. 32).

A. newherrianus and A. depressiis
(Marsh, 1871), and A. gracilis (Leidy,
1873a), described from undiagnostic ver-
tebral characters, are considered here as
nomina duhia.

A. iiintaensis is a form having a relatively
greater body-length than the other species
of Amia. It has approximately the same
total number of vertebrae as A. calva and
A. scutata, but the arrangement of the
coliunn varies meristically from them. Its
head is more elongated tlian that of the
other forms, with the jaws occupying over
two-thirds of the head-length. The vouier-
ine teeth are sharp (as are the palatal and
coronoid teeth), as they are in A. scutata
and A. calva, but are more than twice as
numerous as in these later forms. The pres-
ent study confirms the opinions of Romer
and Fryxell ( 1928), Estes ( 1964), and Estes
and Berberian (1969) that the differences
between Amia and Protamia, Hypamia, and
Pappichthys are insufficient for the recogni-
tion of any of the latter as genera distinct
from Amia. Hypamia elegans (Leidy,
1873a) is considered a m)men duhium, be-
ing based on vertebral characters that can-
not be distinguished from those of the
other species. Protamia media (Leidy,

1873a), Pappichthys symphysis, P. corsonii,
P. medius, P. plicatus, P. sclerops, P. laevis
(all described by Cope, 1873), as well as
Atnia macrospondyla and A. whiteavesiana
(Cope, 1891), are all considered here as
synonyms of A. uintaensis; they were based
on undiagnostic vertebral characters and
morphology of the skull elements. Material
of large amiids from the Late Cretaceous
Lance and Hell Creek formations is referred
to A. cf. uintaensis, since the material differs
only in minor respects from the Paleocene
and Eocene specimens. It cannot be deter-
mined whether this material represents ac-
tual populations of A. uintaensis or an
earlier stage of its evolution. The strati-
graphic range of A. uintaensis extends from
the Paleocene to the Early Oligocene.

A. scutata, an Early to Middle Oligocene
long-bodied form, shares cranial characters
with both A. uintaensis and A. calva. Al-
though it has closer morphometric and
meristic affinities to the Recent form, it is
structually and temporally intermediate be-
tween A. uintaensis and A. calva; it resem-
bles the more primitive A. uintaensis in the
moi-phology of Meckel's groove and coro-
noid articulation surface of the dentary,
greater ossification, and in having an elon-
gated skull with a greater head/ standard-
length than in A. calva. A. dictyocephala
(Cope, 1875) is considered a synonym of
A. scutata; it was based on undiagnostic
meristic characters. In the evolutionarv con-
tinuum, A. scutata appears to be an inter-
mediate stage between A. uintaensis and
A. calva (Fig. 32). A more direct line of
evolution exists between A. scutata and A.
calva; this is supported by Miocene and
Pliocene amiid material that displays cra-
nial elements closely transitional between
the two species. Thus the Recent species of
A. calva had begun at least by the begin-
ning of the Pliocene, and A. calva was ap-
parently distinct from A. scutata by that
time. It appears that A. fragosa represents
an amiid population that survived until the
Middle or Late Eocene and had no phylo-
genetic affinities with the modern form be-
yond this time.

Fossil Amiids • Boreske 79

/Im/'o colva

RECENT

PLEISTOCENE

POST-BLANCAN

Rl ANT AM

HEMPHILLIAN

PLIOCENE

CLARENDONIAN

Ami a cf. calva

BARSTOVIAN

A mi a cf . scut at a

MIOCENE

HEMINGFORDIAN
ARIKAREEAN

WHITNEYAN

OLIGOCENE

ORELLAN
CHADRONIAN

Ami a scuta ta
- i

DUCHESNEAN

, J

EOCENE

UINTAN
BRIDGERIAN

WASATCHIAN

^ragosa Amia uintaensis

CLARKFORKIAN

'V"

PALEOCENE

TIFFANIAN

TORREJONIAN

PUERCAN

\ / -

MAASTRICHTIAN

\ Amia ct uintaensis

CRETACEOUS

CAMPANIAN

\ ;

ALBIAN Al

niidae

mcertae sedi

Fig. 32. Suggested phylogenetic relationships within the genus Am\a.

Bulletin Museum of Comparative Zoology, Vol. 146, No. 1

In the North American fossil record, fossil
remains unquestionably those of the family
Amiidae first occur in tlic Lower Cretaceous
(Albian) sediments of Texas. However,
none of the material can be referred to any
known species of Araia; it displays charac-
ter-states resembling those of Amia uintaen-
sis and Amia frcifi^osa, as well as the Euro-
pean Urocles. Some of the vertebrae re-
semble those of Amiopsis. The Paluxy mate-
rial may represent either one or more forms
transitional between Amia and the Late
Mesozoic European Urocles (or Amiopsis) ,
or an as yet undescribed line. The body-
length of Amia fra^osa appears to be a
primitive feature derived from the earlier
amiids Urocles, Siruimia, Ikechaoam,ia, and
Amiopsis. Despite thc>ir different vertebral
columns, Amia frafi^osa and A. uintaensis
show similar morphology of the cranial
elements, but the nature of the probable
common origin of these forms is still uncer-
tain in the absence of a more complete fossil
record.

Remains of amiids referable to or close to
Amia fragosa and A. uintaensis have been
described from the Paleocene, Eocene, and
Oligocene of Europe, and the Eocene of
Asia. Additional but still not definitive evi-
dence supports Estes' ( 1964 ) and Estes and
Berberian's (1969) suggested synonymy of
A. russelU (Late Paleocene, France), A.
kehreri (Middle Eocene, Germany), and
A. munieri (Early Oligocene, France) with
A. fragosa. Pseudamia Jieintzi (Eocene,
Spitzbergen ) and A. valenciennesi ( Eocene,
France) are also possible synonyms of A.
fragosa. A. valenciennesi is the oldest name
and would take precedence over A. fragosa.
Cranial similarities confirm the synonymy of
A. rohusta (Late Paleocene, France) with
A. uintaensis.

European and North American fossil
Amia occurred in freshwater deposits and
apparently occupied a habitat much like
that of the Recent species. According to
Westoll (1965: 19-20) the distribution of
freshwater vertebrates is a useful indica-
tion of "direct continental communication,"

Table 19. Amiid genera and species of various

AUTHORS discussed IN TEXT IN RELATION TO THE
REVISED lAXONOMY

Atnia calva

Kindlcia fragosa
Stijlomijleodon lactis
Amia fragosa
Paramiattts f^tirlcyi

Amia scutata
A7nia dictyocepliala
Aitiia cxilis

Protamia uintaensis
Protamia media
PappicJi th ys m edius
PappicJithys pJicatus
Pappichthys sclerops
Pappichthys lacvis
PappicJithys symphysis
PappicJithys corsonii
Atnia loJiiteavcsiana
Amia macrospondyla

Amia depressus
Amia newJwrriamis
Amia gracilis
Hypamia elcf^ans

Arnia sehvyniana

. . Amia calva
Amia fragosa

.Amia sctitata

Amia uintaensis

nomina dubia

.nomen nudum

since ". . . descendents of a common stock on
different modern continents must have used
essentially a terrestial route." The present
study further amplifies similarities in the
Paleocene and Early Eocene amiid fossil
record of North America and Europe. This
distribution of amiids adds to the similarity
of assemblages of Paleocene and Early
Eocene lower vertebrates (Estes et al.,
1967) and mammals (McKenna, 1972) on
the two continents. The occurrence of
Pseudamia Iwintzi in the Eocene deposits
of Spitzbergen may be additional evidence
for the existence of the De Geer migration
route (bridging Europe, Spitzbergen, and
North America during the Paleocene and
until the close of Sparnacian time), espe-
cially if suggested relationship to A. fragosa
could be demonstrated. The Asian form A.
mongoliensis resembles A. uintaensis in
minor respects but is sufficiently distinct in
itself to be maintained as a separate species.

Fossil Amiids • Boreske 8i

LITERATURE CITED

Agassiz, L. 1843. Recherches sur les poissons
fossiles. Neuchatel, tomes 1-5, atlases 1-5.

Allis, E. 1889. The anatomy and development
of the lateral line system of Amia calva. J.
Morphol., 2: 1-540.

. 1897. The cranial muscles and cranial

and first spinal nerves in Amia calva. J. Mor-
phol., 12(3): 1-814.

Ami, H. 1891. On some extinct Vertebrata from
the Miocene rocks of the north-west Territories
of Canada recently described by Professor
Cope. Science, 18: 53.

Andreae, a. 1892. Vorliiufige Mitteilung iiber
die Ganoiden ( Lepidostetis und Amia) des
Mainzer Beckens. Verb. Nat. med. Ver. Hei-
delberg (N.F.), 5: 7-15.

. 1895. Beitriige zur Kenntniss der fos-

silen Fische des Mainzer Beckens. Abh.
Senck. naturf. Ges., 18: 351-365.

Becker, H. 1961. Oligocene plants from the
Upper Ruby River Basin, southwestern Mon-
tana. Mem. Geol. Soc. Amer., 82: 1-127.

. 1962. Two new species of Mahonia

from the Grant-Horse Prairie Basin in south-
western Montana. Bull. Torrey Bot. Club,
89: 114-117.

Blair, W., A. Blair, P. Brodkorb, F. Cagle,
AND G. Moore. 1968. Vertebrates of the
United States. New York: McGraw-Hill
Press, 450 pp.

BoLK, L., E. Goppert, E. Kallius, and W. Lu-
BOSCH. 1936. Handbuch der vergleichen-
den Anatomie der Wirbeltiere 4. Berlin: Ur-
ban und Schwarzenberg, 1116 pp.

Boreske, J. 1972. Ta.xonomy and taphonomy of
the North American fossil amiid fishes.
(Abstr.) Bull. Geol. Soc. Amer., 4(1): 3-4.

Bridge, T. 1877. The cranial osteology of Amia
calva. J. Anat. Phys., 11: 605-622.

Cartier, D., and E. Magnin. 1967. La crois-
sance en longueur et en poids des Atnia calva
L. de la region de Montreal. Canad. J. Zool.,
45: 797-804.

Cavender, T. 1968. Freshwater fish remains
from the Clarno Formation Ochoco Moun-
tains of north-central Oregon. Ore Bin, 30
(7): 125-141.

. 1970. A new find of Amia dictyoce-

phala Cope from the Middle Tertiary Floris-
sant Lake Beds of western North America.
Abst. paper given at 50th Annual Meeting,
Amer. Soc. Ichth. Herp.: 42.

Cope, E. 1873. On the extinct Vertebrata of
the Eocene of Wyoming, observed by the ex-
pedition of 1872, with notes on the geology.
Ann. Rept. U.S. Geol. Surv. Terr. (1st series),
No. 6: 545-649.

. 1875. On the fishes of the Tertiary

shales of South Park. Bull. U.S. Geol. Surv.
Terr. (2nd series). No. 1: 3-5.

. 1884. The Vertebrata of the Tertiary

Formations of the West. Rept. U.S. Geol.
Surv. Terr. (Hayden), 3: 745-746.

. 1891. On Vertebrata from the Tertiary

and Cretaceous rocks of the northwest Terri-
tory. Book 1. The species from the Oligocene
or Lower Miocene of Cypress Hills. Contrib.
Geol. Surv. Canad. Paleo., 1: 2-4.

Dean, B. 1898. On the dogfi.sh (Amia calva),
its habits and breeding. Rept. Comm. Fish-
eries, Game, and Forests New York, No. 4:
1-10.

Dechaseaux, C. 1937. Le genre Amia, son his-
toire paleontologique. Ann. Paleont., 26: .3-
16.

Eastman, C. 1899. Jurassic fishes from the
Black Hills of South Dakota. Bull. Geol. Soc.
Amer., 10: 397-407.

Estes, R. 1964. Fossil vertebrates from the Late
Cretaceous Lance Formation, eastern Wyo-
ming. Univ. Calif. Publ. Geol. Sci., 49: 1-
180.

, AND P. Berberian. 1969. Amia ( =

Kimlleia) fragosa (Jordan), a Cretaceous
amiid fish, with notes on related European
forms. Breviora, Mus. Comp. Zool., No. 329:
1-14.

, , AND C. Meszoely. 1969. Lower

vertebrates from the Late Cretaceous Hell
Creek Formation, McCone County, Montana.
Breviora, Mus. Comp. Zool., No. 337: 1-33.

, M. HeCHT, AND R. HOFFSTETTER. 1967.

Paleocene amphibians from Cernay, France.
Amer. Mus. Novitates, No. 2295: 1-25.
-, AND J. TiHEN. 1964. Lower vertebrates

from the Valentine Formation of Nebraska.
Amer. Midi. Nat., 72(2): 453-472.

Goodrich, E. 1930. Studies on the Structure
and Development of Vertebrates. London:
Macmillan Co. Ltd., xxx + 837 pp.

Gould, S. 1966. Allometry and size in ontog-
eny and phylogeny. Biol. Reviews, 41: 587-
640.

Ha.mmett, F., AND R. Hammett. 1939. Pro-
portional length growth of gar ( Lepisosteus
platyrhincits DeKay). Growth, 3(2): 197-
209.

Hasse, C. 1882. Das natiirliche System der
Elasmobranchier. Jena: Verlag von Gu.stav
Fischer, pp. 77-284.

Hatcher, J. 1900. The Carnegie Mu.seum pale-
ontological expeditions of 1900. Science, 12
(.306): 718-720.

. 1901. Some new and little known fos-
sil vertebrates. Ann. Carnegie Mus. Pitts-
burgh, No. 1: 128-129.

Hay, O. 1895. On the structure and develop-

82 Bulletin Museum of Comparative Zoology, Vol. 146, No. 1

ment of the vertebral column of Amia. Publ.
Field Columb. Mus. Zool., 1(1): 1-54.

. 1911. Tlie Pleistocene of Indiana.

Ann. Kept. Ceol. Nat. Res. Indiana, .36: 552.

. 1917. Vertebrata mostly from stratum

No. 3, at Vero, Florida; together with de-
scriptions of new species. Rept. Geol. Surv.
Florida, 9: 43-68.

. 1923. The Pleistocene of North Amer-

ica and its vertebrated animals from the states
east of the Mississippi River and from the
Canadian Provinces east of longitude 95 de-
grees. Bull. Carnegie Inst., .322: 336-382.

HoFFSTETTER, R., AND J. Gasc. 1969. Verte-
brae and ribs of modem reptiles. In Biology
of the Reptilia. New York: Academic Press,
xvi -|- 468 pp.

HuBBS, C, AND K. Lagler 1967. Fishes of the
Great Lakes Region. Ann Arbor: University
of Michigan Press, viii -|- 289 pp.

HussAKOF, L. 1932. The fossil fishes collected
by the Central Asiatic Expeditions. Amer.
Mus. Novitates, No. 553: 1-16.

Imbrie, J. 1956. Biometrical methods in the
study of invertebrate fossils. Bull. Amer.
Mus. Nat. Hist., 108: 217-252.

Janot, C. 1966. Aiuia riisselli nov. sp., nouvel
Amiide ( Poisson holosteen ) du Thanetien de
Berru, pres de Reims. C. R. Soc. Geol.
France, 1966(3): 142.

. 1967. A propos des amiides actuels et

fossiles. Colloq. Intern. C. N. R. S., 163:
139-153.

Jordan, D. 1906. The fishes of Samoa. Bull.
Bureau Fisheries, 25: 2.37.

. 1919. The genera of fishes. Publ. Le-

land Stanford Junior University, 1919: 259.

. 1925. Opinion 89. In Opinions ren-
dered by the International Commission on
Zoological Nomenclature. Smithsonian Misc.
Coll., 73(3): 27-29.

. 1927. Kindleia, a new genus of cichlid

fishes from the Upper Cretaceous of Alberta.
Canad. Field Nat., 41: 145-147.

. 1928. Note on Kindleia and Stylomy-

leodon ( fossil fish from Cretaceous deposits
of Alberta). Canad. Field Nat., 42: 47.

, and B. Evermann. 1896. The fishes of

North and Middle America. Bull. U.S. Nat.

Mus., 47: 112-113.
Lagler, K., J. Bardach, and R. Miller. 1962.

Ichthyology. New York: John Wiley and

Sons, Inc., vii + 523 pp.
Lambe, L. 1904. Progress in vertebrate pale-
ontology in Canada. Trans. Roy. Soc. Canada,

No. 2: 27-43.
. 1908. The Vertebrata of the Oligocene

of the Cypress Hills, Saskatchewan. Contrib.

Paleo. Canada, 3(4): 12-13.
Lange, S. 1968. Zur Morphologic imd Tax-

onomie der Fischgattung Urocles aus Jura nnd

Kreide Europas. Palaeontographica, Abt. A,
131: 1-78.

Lehman, J. 1951. Un nouvel Amiide de
I'Eocene du Spitzberg, Pseudamia lieintzi.
Troms0 Mus. Arshefter, 70: 1-11.

Leidy, J. 1873a. Notice of remains of fishes in
the Bridger Tertiary Formation of Wyoming.
Proc. Acad. Nat. Sci. Philadelphia, 1873:
97-99.

. 1873b. Contributions to the extinct

vertebrate fauna of the western territories.
Rept. U.S. Geol. Surv. Terr. (Hayden), 1:
185-189.

Liu, T-S., H-T Liu, and T-T Su. 1963. The
discovery of Sinamia zdanskyi from the Ordos
Region and its stratigraphical significance.
Vert. Palasiatica, 7(1): 1-30.

Lund, R. 1967. An analysis of the propulsive
mechanisms of fishes, with reference to some
fossil actinopterygians. Ann. Carnegie Mus.,
39(15): 195-218.

Marsh, O. 1871. Communication on some new
reptiles and fishes from the Cretaceous and
Tertiary. Proc. Acad. Nat. Sci. Philadelphia,
1871: 103-105.

McKenna, M. 1972. Was Europe connected
directly to North America prior to the Mid-
dle Eocene? In Evolutionary Biology, Vol. 6.
New York: Appleton-Century-Crofts, pp. 179—
189.

Merrill, G. 1907. Catalogue of the figured
specimens of fossils, minerals, and ores in the
Department of Geology, LTnited States Na-
tional Museum. Bull. U.S. Nat. Mus., 53:
6-15.

Miller, H. 1968. Additions to the Upper Cre-
taceous vertebrate fauna of Phoebus Land-
ing, North Carolina. J. Elisha Mitchell Sci.
Soc, 84: 467^71.

Newton, E. 1899. On the remains of Amia,
from the Oligocene strata in the Isle of
Wight. Quart. J. Geol. Soc. London, 4: 1-
10.

Nybelin, O. 1963. Zur Morphologic und Ter-
minologie des Schwanzskelettes der Actinop-
teiygier. Arkiv. Zool, 15(35): 485-516.

O'Brien, D. 1969. Osteology of Kindleia frag-
osa Jordan (Holostei: Amiidae) from the
Edmonton Formation ( Maestrichtian ) of
Alberta. M.A. Thesis, University of Alberta:
vi -1- 118 pp.

Osborn, H., W. Scott, and F. Speir. 1878.
Paleontological report of the Princeton scien-
tific expedition of 1877. Contrib. Mus. Geol.
Arch. Princeton, No. 1: 102-104.

Piton, L. 1940. Paleontologie du gisement
Eocene de Menat ( Puy-de-Dome, flore et
faune). Mem. Soc. Sci. Nancy, 56: 1-303
pp.

Reighard, J. 1903. The natural history of Amia

Fossil A muds • Bote she

ralva Linnaeus. Mark Anni\cisaiy Vohinie

(art. t): 57-109.
RoMKH, A., AND F. FiiYXKLi,. 1 92S. Purdiuiatiis

^urlcyi, a ni-w (Iccp-hocliccl aiiiiid fish from

tlie Eocene of Wyoniinj,'. Aiiiei. |. Sei., 16

(90): 519-527.
|{us.st:LL, L. 1928a. A new fossil fish from the

Paskapoo Beds of Alberta. Aiiier. |. Sei., 15

(86): 103-107.
. 19281). The jfenera Kiiullcut and Sty-

loviylcodon. Anier. J. Sei., 15(87): 264.
. 1929. Tiie validity of the .ueniis Sttj-

IniiujUoiUm. Auier. J. Sei.. 17( 100): 369-;571.
. 1967. Paleontology of the Swan Hills

area, north-central Alheila. Roy. Ontario

Mus. C.'ontril)., No. 71: 1-31.
Sc:hakffeh, B. 1967. Osteichthyan vertebrae. |.

Linn. Soe. (Zool.), No. 47: 185-195.
Shufixut, R. 1885. The osteology of Ainiu

calva. Washington: Government Printing Of-
fice, 93 pp.
SiMi'.soN, (;. 1937. The Fort Ihiion of the

Crazy Mountain Field, Montana, and its

inannnalian faunas. Bull. U.S. Nat. Mus.,

169: \ + 287 pp.
, A. RoK, AND R. Lewontin. 1960.

Quantitative Zoology. New York: Harcourt,

Brace, and Co., iv + 440 pp.
Skinneh, M., S. Skinneh, and R. Gooius. 1968.

C-enozoic rocks and faunas of Turtle Butte,

south-central South Dakota. Bull. Anier. Mus.

Nat. Hist., 138: 381-436.
Smi'iii, C. 1962. Some Pliocene fishes from

Kansas, Oklahoma, and Nebraska. Copeia,

1962(3): 505-520.
Stensio, E. 1935. Smainia zdanskyi, a new

amiid from the Lower C^retaceoiis of Shan-
tung, China. Pahoiit. Sinica, .ser. C, No. 3:
1-48.

Swift, C. 1968. Fossil bony fishes from Flor-
ida. I'laster Jacket, No. 7: 2-11.

SzALAY, F., AND M. Mc:Kenna. 1971. Begin-
ning of the age of uiaiumals in Asia: tin- Late
Paleocene Cashato fauna, Mongolia. Bvill.
Amer. Mus. Nat. Hist., 144: 270-317.

Thomson, K., and K. IIaiin. 1968. (Jrowth ;nid
form in fossil rhipidistian fishes ( Cro.ssopter-
ygii). J. Zool., 156: 199-223.

Thuhmond, J. 1969. Lower vertebrates and
paleontology of the Trinity Croup. Ph.D.
Thesis, Southern Methodist Unixersity ( Libr.
Congr. No. 70-19259). 1.36 pp. Univ. micro.
Ann Arbor, Mich, (di.ss. ab.str. 31: 2156).

TnAQUAut, R. 1911. Les poissons wealdiens de
Bernissart. Mem. Mus. Roy. Hist. Nat. Belg.,
6: 1-65.

Westoll, T. 1965. ecological evidence bear-
ing upon continental drift. Symposium on
Continental Drift. Royal Soc. London, 1965:
12-25.

WiUTEHousE, R. 1910. The caudal fin of the
Teleostomi. Proc. Zool. Soc. London, 1910:
590-627.

Whitley, C. 1954. More new fish names and
records. Aust. Zool., 12: 57.

Wilson, R. 1968. Systematics and faunal anal-
ysis of a Lower Pliocene vertebrate assem-
blage from Trego County, Kansas. Contril).
Mus. Paleo. Univ. Mich., 22(7): 75-126.

Woodward, A. 1916. The fossil fi.shes of the
English Wealden and Purbeck formations.
Palaeontogr. Soc. London, Pt. 1, 1915: 1-48.

84 Bulletin Museum of Comparative Zoology, Vol. 146, No. 1

5 cm

afab.

.J<Vj,

Plate 1. A, "Paramiatus gurley'i" FMNH 2201, Early Eocene, Green River Formation, Wyoming; B, Amia iragosa
MCZ 5347, Early Eocene, Green River Formation, Wyoming.

Fossil Amiids • Boreske 85

Plate 2. Amia kebreri BMNH P33480, collected by Walter Kijhne in 1951 from Middle Eocene deposits at Messel
bei Darmstadt.

86 Bulletin Museum of Comparative Zoology, Vol. 146, No. 1

Plate 3. Amia uinfaensis PU 13865, Early Eocene, Green River Formation, Wyoming.

Fossil Amiids • Bnrrslr

?^" ■_;.■■•«;>;

.: '-^

-■

1 y

•^fe%i

' •

■•■-Tf^

•'i.

- . i

Plate 4. Amio scu/ofa. Middle Oligocene, Florissant Formation, Colorado: A, counterpart YPM 6243; B, counter-
part USNM 4087; C, PU 10172.

us ISSN 0O27.4100

Bulletin OF THE

Museum of

Comparative

Zoology

An Analysis of Variation in the Hispaniolan
Giant Anole, Anolis ricordi Dumeril

and Bibron

ALBERT SCHWARTZ

HARVARD UNIVERSITY

CAMBRIDGE, MASSACHUSETTS, U.S.A.

VOLUME 146, NUMBER 2
19 APRIL 1974

PUBLICATIONS ISSUED

OR DISTRIBUTED BY THE

MUSEUM OF COMPARATIVE ZOOLOGY

HARVARD UNIVERSITY

Bulletin 1863-

Breviora 1952-

Memoirs 1864-1938

JoHNSONLV, Department of MoUusks, 1941-

OccAsioNAL Papers on Mollusks, 1945-

Other Publications.

Bigelow, H. B., and W. C. Schroeder, 1953. Fishes of the Gulf of Maine.
Reprint

Brues, C. T., A. L. Melander, and F. M. Carpenter, 1954. Classification of
Insects.

Creighton, W. S., 1950. The Ants of North America. Reprint.

Lyman, C. P., and A. R. Dawe (eds.), 1960. Symposium on Natural Mam-
malian Hibernation.

Peters' Check-list of Birds of the World, vols. 2-7, 9, 10, 12-15.

Sprinkle, J., 1973. Morphology and Evolution of Blastozoan Echinoderms.

Turner, R. D., 1966. A Survey and Illustrated Catalogue of the Teredinidae
(MoUusca: Bivalvia).

Whittington, H. B., and W. D. I. Rolfe (eds.), 1963. Phylogeny and Evolu-
tion of Crustacea.

Proceedings of the New England Zoological Club 1899-1948. (Complete
sets only.)

Publications of the Boston Society of Natural History.

Authors preparing manuscripts for the Bulletin of the Museum of Comparative
Zoology or Breviora should send for the current Information and Instruction Sheet,
available from Editor, Publications OflBce, Museum of Comparative Zoology,
Harvard University, Cambridge, Massachusetts 02138, U.S.A.

AN ANALYSIS OF VARIATION IN THE HISPANIOLAN GIANT
ANOLE, ANOLIS RICORDI DUMERIL AND BIBRON

ALBERT SCHWARTZ'

Abstract. The nominal Hispaniolan species of
giant anole, Anolls ricordi, is considered to be in
actuality composed of three distinct allopatric spe-
cies: A. ricordi, A. barahonae, and A. baleatus.
Subspecies of all three species are described, but
only A. baleatus is well represented in collections.
A theoretical history of this species complex upon
Hispaniola is presented.

The Hispaniolan giant anole, Anolis ri-
cordi Dumeril and Bibron, 1837, has been
known to science for more than a century;
yet only in the hist 35 years has it become
evident that this species is not homoge-
neous in its characteristics throughout Haiti
and tlie Republica Dominicana. The spe-
cies was first named (as Anolis ricordii)
from Santo Domingo, as the entire island
was known at that historical period, but
specimens seem to have been rare in col-
lections thereafter. Schmidt (1921: 10) re-
ported four A. ricordi from two Dominican
localities. Cochran (1941: 133) Hsted 24
specimens (all but one of which were in
the National Museum of Natural History)
from 11 localities. Mertens (1939: 68-70)
studied 17 specimens in European collec-
tions and was the first to recognize that
there were two readily distinguishable pop-
ulations that he considered subspecies: A.
r. ricordi in Haiti, and A. r. baleatus Cope
in the Republica Dominicana. Williams
( 1962 ) reviewed the species in more detail
and examined 90 specimens. For this suite
of anoles, he described A. r. barahonae

1 Miami-Dade Community College, Miami,
Florida 33167.

Bull. Mus. Co

from tlie Sierra de Baoruco in the south-
western Republica Dominicana. Still later,
Williams (1965) studied an additional 80
specimens and named A. r. leberi from
Camp Perrin on the extreme distal portion
of the Haitian Tiburon Peninsula. Thus,
with increasing quantities of material from
more diverse localities, our knowledge of
the distribution and variation in this species
has increased accordingly.

A great many problems remain, however,
when one deals in detail with the variation
in A. ricordi. Williams (1962, 1965)
pointed out that records of the species were
of such a scattered nature (especially on
the Tiburon Peninsula but also elsewhere
on the island) that intergrades between
several of the subspecies remained unknown
and also that there were no specimens
available from large areas between named
populations. Williams and Rand (1969), in
their excellent summary of the geographic
differentiation in all species of Hispaniolan
anoles, pointed out (p. 15) that Anolis ri-
cordi was composed of "several described
subspecies, some of which are sharply
enough distinct to raise the question of pos-
sible species status." This is most especially
true of the taxa ricordi, baleatus, and bara-
honae, all of which are extremely well
characterized by both pigmental and struc-
tural details, but all of which occupy areas
(extensive in the cases of ricordi and ba-
leatus) without known intergradation be-
tween them or without close geographic
approximation. Thus, the closest ap-

mp. Zool., 146(2): 89-146, April, 1974 89

Bulletin Museum of Comparative Zoology, Vol. 146, No. 2

■s.

01
0

w»

•~

c
o
-c
0
i.
0

■<

'u
01

0
0

-J

3
CO

•V

-o

->t

-Q

1 1

c
01
vt
01

c
o

-c

■♦-

"a
E

■a

c
"a

'■*-

1 1

■D

01
0

-o

c
0

c
o

c
_0

'c
0
Q.

-a

c
1

c
D

c
o

k.
0

-C

*•

-o

3
0

■a

c
o

0
O

w»

*L»

(11
<u

(/)

•A

c
-><

<1)

o
o

^wi

«/)

—

0£

•A

-SI

k-
-0

«r

**-
o

lA
2

-a

0
-Q

k-'

O)
0

-c oo -Q ^ -^

HiSPANiOLAN Giant Angle • Schwartz 91

proaclu's of the ranges of ricordi and 1)0- material upon whieh A. r. leheri was based.

leatits (Copey and Peiia, both in the Our unique experience at Camp Perrin —

Republica Dominieana) are separated by namely, of many A. ricordi received from

some 115 kilometers airline. The subspe- local Haitians — showed that the species

cies harahomie and ricordi (Sierra de Bao- may not be necessarily rare. However, as

rueo and associated eastern coastal areas in A. equestris in Cuba, the cryptic greens

in the Republica Dominieana, and Saltrou of A. ricordi render the species virtually in-

in Haiti) are known from localities sepa- visible during the day except to all but the

rated by about 115 kilometers, and bara- most experienced observer. In 1963, Rich-

honae and balcatus by a gap of about 115 ard Thomas discovered that A. ricordi

kilometers (between the Sierra de Baoruco might be seemed at night, since individ-

and near Villa Altagracia, both in the Re- uals sleep quite exposed in a variety of ar-

publica Dominieana). boreal situations and are very conspicuous.

Schwartz and Garrido (1972) recently Thus, with the knowledge that the agroma
showed that the Cuban giant anole, Anolis (as the species is known in Haitian Creole)
equestris Merrem, is, in fact, a complex of or the saltacocote ( as the species is known
five species; they also suggested (p. 71) in Dominican Spanish) might be common
that, as Williams and Rand had pointed and thus easily secured by native collectors,
out, there was a good possibility that the and that individuals might be readily se-
Hispaniolan Anolis ricordi in time might cured at night while they slept, I had as
likewise be shown to be a complex of spe- one of my objectives to secure as many A.
cies. It is the purpose of the present paper ricordi as possible in order to clarify the
to discuss the variation in A. ricordi, based status of the named subspecies and in an
upon the examination of 403 specimens attempt to narrow the geographic gaps that
from a broad selection of geographic local- seemed to exist between ricordi, baleatus,
ities throughout Hispaniola. Despite my and barahonae. As more material accumu-
ha\'ing studied far more material than any lated, we were successful in the latter at-
previous investigator, there still remain tempt, but the range of variation in newly
many problems that cry out for solution, acquired material showed that the situation
As Schwartz and Garrido also pointed out was more complex than was supposed. In
in their analysis of Anolis equestris, the addition to specimens in the Albert
present paper in no way should be consid- Schwartz Field Series (ASFS), collected
ered as the final word on A. ricordi; rather by myself and field assistants, I have ex-
it is an attempt to comment in detail upon amined material in the American Museum
the known variation and distribution of of Natural History (AMNH), the Museum
this species in Hispaniola which may serve of Comparative Zoology (MCZ), and the
as a stepping stone for further work upon National Museum of Natural History
the species. ( USNM). For the loans of specimens I am

Between 1962 and 1971, I and my asso- grateful to Richard G. Zweifel, George W.

ciates collected extensively in both Haiti Foley, Ernest E. Williams, and George R.

and the Republica Dominieana. Latterly, Zug. In all of these collections there are

between 1968 and 1971, my work in His- other specimens that I have deliberately

paniola has been under the sponsorship of not elected to study, since many of them

two National Science Foundation grants, are from localities that are now well repre-

GB-7977 and B-023603. Specimens of Ano- sented by more recently collected lizards

lis ricordi collected in 1962-63 were avail- or that have poor locality data. Specimens

able to Williams and were reported upon in the collection of the Museum of Com-

by him (1965); in fact, the long series of parative Zoology have been collected un-

A. ricordi from Camp Perrin, Haiti, secured der NSF grant B-019(S01X and previous

for me by native collectors in 1962, was the grants to Dr. Williams. Most of the re-

92 Bulletin Museum of Comparative Zoology, Vol. 146, No. 2

cently taken ASFS A. ricordi have detailed
descriptions of color and pattern in life,
but, as in all such endeavors that span sev-
eral years, it is unfortimate that all details
of color and pattern have not been consis-
tently recorded as time has passed. Like-
wise, there are no color or pattern data on
most old specimens; thus, I feel less secure
in dealing with these older specimens or
those collected by others than myself and
parties than I am with those in the ASFS
which are carefully documented. However,
specimens from other localities must in
some way be dealt with, and I have done
so as carefully as possible, considering de-
tails of geography and what is known about
specimens of A. ricordi from adjacent local-
ities.

I wish to acknowledge with enthusiasm
the efforts on my behalf in the field of the
following men, without whose efforts the
quantity of A. ricordi presently available
to me would be far less: Jeffrey R. Buffett,
Carl Butterfield, James R. Dennis, Danny
C. Fowler, Ronald F. Klinikowski, David
C. Leber, James A. Rodgers, Jr., Bruce R.
Sheplan, and Richard Thomas. C. Rhea
Warren has given me a specimen of A. ri-
cordi from northern Haiti. My notes on
coloration and pattern of A. ricordi have
been greatly supplemented during the
present study by the color portraits exe-
cuted in the field by D. C. Leber; one of
these has been reproduced in black-and-
white in Williams ( 1965 ) , but the repro-
duction hardly does justice to the detailed
beauty of all the originals. I have been able
to examine the holotype of Eupristis ba-
leattis Cope through the courtesy of Alice
C. C. Grandison and A. F. Stimpson of the
British Museum (Natural History). Holo-
types and paratypes have been designated
or deposited in the above collections and
in the Caniegie Museum (CM).

THE PROBLEM

Mertens ( 1939 ) was the first to point out
that Haitian and Dominican A. ricordi dif-
fered from each other in one notable char-
acter— the height of the dorsal crest scales.

His figure 41 shows this character ex-
tremely clearly: in nominate ricordi from
Haiti, the nuchal crest scales are low and
inconspicuous, whereas in Dominican ba-
leatiis the nuchal crest scales are long and
attenuate. In addition, Mertens (1939: 69)
characterized ricordi as having 9 to 12
scales between the eyes; males of this sub-
species have one or more sharply defined
black blotches on the sides of the nape, the
occipital area flecked with black, and often
have black longitudinal stripes on the
flanks. On the other hand, baleatus has
from 6 to 8 scales between the eyes, and
males are without any black head, nape, or
lateral markings. Williams (1962) com-
pared these two taxa with barahonae in re-
gard to four characters: height and relative
length of nuchal and dorsal crest scales,
number of snout scales at the level of the
second canthal scale, and body pattern.
Later, when he described A. r. leberi, Wil-
liams (1965) employed these same charac-
ters to differentiate that subspecies.

The differences in these characters be-
tween the four recognized subspecies are
unequivocal: one can differentiate at a
glance between such distinctive animals as
leberi and barahonae or between ricordi
and baleattis, without recourse to micro-
scopic examination. The whole a.spects of
all four taxa are quite distinctive, whether
one is dealing with living or long-preserved
animals. What has been equivocal is the
relationships of these four taxa, since, as I
pointed out previovisly, they have been
known from rather isolated groups of lo-
calities, widely separated from each other.
In only one case (leberi-ricordi) have
specimens been regarded as intergradient
between two subspecies: these intergrades
are from a geographically plausible locality
that itself is widely removed from the two
"parent" populations.

As material has gradually accumulated,
it has become increasingly obvious that the
situation is even more complex than has
been previously recognized. For example,
in 1971, I had occasion to compare long se-
ries of living examples from the Peninsula

HisPANioLAN Giant Anoll. • Schwuiu

de Sainana and the adjacent "mainland" at
Cafio Abajo, and I was at once struck with
the differences between tliese two samples,
both of which have been regarded as ha-
leattis. In this case, the differences are not
particularly subtle but they do involve dif-
ferences in coloration and pattern which
often are evanescent after preservation.
The same statement may be made about A.
ricordi from the region near La Vega and
those from the Cordillera Septentrional. In
1971 I had occasion to collect specimens
from both these regions on two successive
days and thus was able to compare freshly
collected material directly. Again, the dif-
ferences are ones of pattern and color, but
they are so striking that it is misleading to
consider both these populations as being
identical genetically. I could multiply the
above examples but to no purpose; it is ob-
vious, when one sees living A. ricordi in the
field, that there are several populations
presently assigned to haleatus which are
quite distinctive.

On the basis of specimens collected by
Richard Thomas and myself in 1963, Wil-
liams ( 1965 ) reported A. r. ricordi for the
first time from the northwestern Republica
Dominicana in the region near Pepillo Sal-
cedo and Copey in Monte Cristi Province.
He noted, however, that, "Despite the new
collections one embarrassment remains. No
certain intergrades between the two strik-
ingly different forms ricordii and ])aleatiis
are yet known. . . . However, the area in
which intergrades may occur is being nar-
rowed: on the north coast of the Dominican
Republic between Monte Cristi and Santi-
ago and in the center of Hispaniola be-
tween Mirebalais ( MCZ 68479, 69404) and
Santiago. This still leaves a very wide area
of ignorance." Since the above was written,
I have secured specimens of the nominate
subspecies in four other regions: at Re-
stauracion, Dajabon Province, along the
Dominico-Haitian border and about 40 ki-
lometers airline south of the Monte Cristi
localities; on the southern slopes of the Cor-
dillera Central near Juan de Herrera in San
Juan Province; and throughout the Sierra

de Neiba between Hondo Valle and Valle-
juelo in La Lstrelleta and San Juan prov-
inces. These latter two regions (the south-
ern slopes of the Cordillera Central and the
Sierra de Neiba) are separated by the
rather xeric Valle de San Juan. Elsewhere
to the east and north, the Cordillera Cen-
tral harbors A. ricordi with long nuchal
crest scales and without black nape and
head markings in males (i.e., — haleatua),
as at San Jose de Ocoa, La Vega, and the
interior uplands near El Rio, and near Los
Montones on the Rio Bao. The fourth lo-
cality is perhaps the most significant; there
is one subadult male from Santiago Ro-
driguez Province near Los Quemados which
is clearly a ricordi. Of the haleatus locali-
ties, this one is closest to Los Montones, a
distance of 50 kilometers airline. Thus the
gap between ricordi and haleatus in north-
ern Republica Dominicana has been more
than halved, and there still is no genetic in-
fluence of one subspecies upon the other.

To the south, specimens of A. ricordi
from the Sierra Martin Garcia, and Azua
and Peravia provinces likewise narrow the
gap there between haleatus and ricordi on
one hand and between haleatus and hara-
honae on the other. In the former case, the
distance between ricordi and haleatus
(Vallejuelo and Sierra Martin Garcia) is
about 60 kilometers airline, without char-
acter dilution. In the instance of harahonae
and haleatus, only 20 kilometers separate
known localities (Barahona and Sierra
Martin Garcia) for these two taxa: how-
ever, the actual kilometrage is deceiving,
since, lying between these localities, is the
Valle de Neiba and the Bahia de Barahona.
Although this eastern extreme of the Valle
de Neiba is rather mesic and supports (or
did support ) large trees in many areas that
would presumably be suitable for A. ri-
cordi, the break between these two popu-
lations is sharp despite presumably suitable
ecology. I have little doubt that A. ricordi
occurs in this intervening region, and the
interaction of harahonae and haleatus
therein will be most interesting to ascertain.

Even more intriguing is the fact that the

94 Bulletin Museum of Comparative Zoology, Vol. 146, No. 2

Sierra Martin Garcia is an extreme eastern
isolate of the Sierra de Neiba, which is
elsewhere occupied by A. r. ricordi. This
small range, which reaches an elevation of
1350 meters, is completely surrounded by
extremely arid desert or low rolling xeric
hills, as well as by the Bahia de Barahona
on its southwestern edges. The heipeto-
fauna of the Sierra Martin Garcia is just
becoming knowti, and it supports a remark-
able fauna, including an endemic species
of Diplog,Iossus (Thomas, 1971) and a new
species of Sphaerodactylus, as well as other
unexpected novelties. Nevertheless, A. ri-
cordi seems to have reached this range
from the northeast (i.e., the southern slopes
of the Cordillera Central), since I regard
the Martin Garcia lizards as identical to
those from Azua and Peravia provinces.

Finally, the geographic relationship of
barohonae and ricordi has been to some ex-
tent clarified. A. r. harahonae has been
known only from the eastern uplands of
the Sierra de Baoruco and from three
southern lowland localities ( Enriquillo,
half-way between Enriquillo and Oviedo,
and Oviedo). Each of the latter localities
is represented by a single specimen. The
Enriquillo and Enriquillo-Oviedo speci-
mens are quite obviously harahonae, but,
as Williams (1965: 4) noted, the specimen
from close to Oviedo is quite different in
style of pattern and color from typical
harahonae. To the west, in Haiti, there has
been but a single specimen from Saltrou
which Williams (1965: 2) considered A. r.
ricordi and which "narrows the geographic
gap between ricordii ricordii and r. hara-
honae; however, it does nothing to narrow
the character gap." Two additional facts
are important. First, in 1971, we secured
a pair of A. ricordi from along the Domin-
ico-Haitian border north of Pedernales;
these individuals, although differing some-
what from typical harahonae and strongly
from the single Oviedo specimen, in no
way show any tendencies toward A. r. ri-
cordi. They are clearly related to hara-
honae, a rather surprising fact since they
are much closer (35 kilometers) to Saltrou

than they are to any harahonae locality (65
kilometers at Enriquillo). Secondly, Wil-
liams has recently received a fine se-
ries of A. ricordi from Source Carroye
near Thiotte; these lizards are quite obvi-
ously not A. r. ricordi but are closer in
many ways to the far-western A. r. leheri.
Thus the situation along the southern Hai-
tian coast between Saltiou and the eastern
coast of the Republica Dominicana at Ovi-
edo and its environs remains a true puzzle.
It seems likely that A. r. ricordi does not
cross the high Massif de la Selle, except
possibly by some circuitous route, and that
A. r. harahonae occurs up to the Dominico-
Haitian border, to within 11 kilometers of
a station ( Source Carroye ) where another
taxon occurs, without character dilution.

Interpretations of all these facts are seri-
ously hampered by the lack of specimens of
A. ricordi from throughout the Haitian Ti-
buron Peninsula. Material from the penin-
sula may be divided into four basic lots: a
short series from the region about Castillon
on the northern slopes of the Massif de la
Hotte near the tip of the peninsula; a very
long series of specimens from Camp Perrin
on the low southern slopes of the Massif de
la Hotte (the type series of A. r. leheri);
a short series from midway along the pen-
insula at Miragoane-Paillant; and a moder-
ate number of specimens from near the
base of the peninsula in the region
of Morne Decayette-Petionville-Port-au-
Prince. The lack of material from such
well-collected areas as Jeremie on the
northwestern coast or Jacmel and Les
Cayes on the southern coast is extremely
puzzling — we simply know nothing about
lowland A. ricordi throughout much of the
Tiburon Peninsula, except for the above
scattered records and the southern coast at
Saltrou near the Dominican border (and
the latter locality is not even on the penin-
sula proper).

To summarize the data from elsewhere
in Haiti and the Republica Dominicana,
there is a huge distributional hole in cen-
tral Haiti, with but two specimens {ri-
cordi) from Mirebalais, widely separated

HisPANiOLAN Giant Anoli

b.:,/Ui i:

from southern ricordi at and near Port-an-
Prince, and then a group of seattered local-
ities along the northern littoral of Haiti
from Port-de-Paix in the west to Terrier
Rouge in the east, and one specimen from
Marmelade in the interior Chaine de Mar-
melade. All these Haitian specimens have
low nuchal crest scales, and males variably
possess some black nape and side markings,
but there are differences between speci-
mens from the various sections which pres-
ently defy analysis, since the material is
too meagre and from too scattered locali-
ties.

The Republica Dominicana fares far bet-
ter as far as detailed coverage is con-
cerned. Aside from the material previously
noted as assigned to ricordi or harahonae,
there are now good series available from
the eastern half of the country, and, al-
though there are certain gaps even within
this region, they are not so appalling as are
those in Haiti. A. ricordi is rarer (or per-
haps less easily encountered) in arid re-
gions, and thus the most striking gaps in
the western half of the Republica Domini-
cana are those involving arid regions on
the one hand or high mountain masses on
the other. The latter situation, especially if
the slopes are pine-clad, does not appear
suitable for A. ricordi. and the species may
be truly absent from the uplands above
4000 feet (1220 meters), the highest eleva-
tion from which the species is known. In
arid regions, A. ricordi appears to be re-
stricted to riverine woods and forests; in
such situations, the species may not be un-
common, but it may require diligence to
secure even one specimen from a particular
region.

The detailed discussion above should
give the reader a background of both the
history and present knowledge of the dis-
tribution of A. ricordi against which the
following accounts can be most logically
followed. One further point is of interest.
A. ricordi is unknown by specimens from
any of the large Hispaniolan satellite is-
lands. In some cases (Isla Beata) the spe-
cies is not expected for a variety of reasons,

but in others (He de la Gonave, lie dc la
Tortuc, Ile-a-Vache) there seems no logical
reason for the absence of A. ricordi, dis-
counting the vagaries of overseas transpoit.
The species does occur on Isla Saona, but
remains uncollected there. Fowler and
Sheplan saw a sleeping A. ricordi on the
northern coast of Isla Saona in December
1971, but, after it had been shot, it was
lost in the undergrowth. The occurrence of
A. ricordi on any Hispaniolan satellite is
noteworthy, and it will be most interesting
to determine the status of the Isla Saona
population.

METHODOLOGY

The series of 403 A. ricordi was divided
into 14 samples on the basis of geography,
as follows: Republica Dominicana: 1) Pe-
ninsula de Samana (54 specimens); 2)
northeastern Republica Dominicana, from
Duarte and eastern La Vega provinces east
to the haitises region in northeastern San
Cristobal Province ( 37 ) ; 3 ) extreme east-
ern Hispaniola, Punta Cana-Juanillo, Boca
de Yuma, La Altagracia Province ( 16 ) ; 4 )
southeastern Republica Dominicana from
Higiiey and Las Lisas, La Altagracia Prov-
ince, west to Santo Domingo and Yamasa,
San Cristobal Province (61); 5) lowlands at
the northern base of the Cordillera Central
at Guaigui, La Vega Province (21); 6)
Cordillera Central at and above elevations
of 2000 feet (18); 7) Cordillera Septentri-
onal and north ( 15 ) ; 8 ) Sierra Martin Gar-
cia and Peravia and Azua provinces (6); 9)
Sierra de Baoruco and associated east coast
of the Peninsula de Barahona (33); 10)
Oviedo, Pedemales Province ( 1 ) ; Haiti:
11) Saltrou and vicinity, Dept. de I'Ouest
( 15 ) ; 12 ) region about Port-au-Prince,
Mirebalais, northern Haitian littoral,
Chaine de Marmelade, and (in the Repu-
blica Dominicana) region about Pepillo Sal-
cedo, Copey, Los Quemados, Restauracion,
and the southern slopes of the Cordillera
Central and the Sierra de Neiba (50); 13)
Camp Perrin and Marceline, Dept. du Sud
(54); and 14) vicinity of Castillon, Dept.
du Sud (6). This division into 14 regional

Bulletin Museum of Comparative Zoology, Vol. 146, No. 2

samples was completed after preliminary
examination of the material, scale counts
and detailed review of field notes on color-
ation and pattern were all made. In addi-
tion, two other small lots (eight specimens
from the vicinity of Miragoane, Dept. du
Sud, Haiti, and seven specimens from El
Seibo Province, Republica Dominicana)
were examined separately; in both cases,
these short series indicate intergradient
tendencies between pairs of the 14 major
samples noted above, and they were not in-
cluded with the latter.

The following data were taken on each
specimen:

1 ) Snout-vent length, in millimeters.

2) Number of snout scales across snout
at level of the second canthal scale, reck-
oned from the anterior corner of the orbit.

3) Number of vertical rows of loreal
scales.

4) Minimum number of scales between
supraorbital semicircles.

5) Number of scales between the inter-
parietal scale and the supraorbital semicir-
cles on each side, this datum written as a
fraction ( i.e., 5/5 = five scales in this posi-
tion on each side ) .

6) Number of scale rows between the
subocular scales and the supralabial scales.

7) Number of vertical rows of dorsal
scales in a distance equal to that from the
tip of the snout to the anterior bony wall
of the orbit, this distance measured by ver-
nier calipers, laid off on the back about
three rows below the median dorsal crest
scales, and the number of scales counted
under a binocular dissecting microscope.

8) Number of horizontal rows of dorsal
scales in the snout-eye distance, the scales
counted at midbody. This count was not
taken in most juveniles or on those speci-
mens that were shrunken or poorly pre-
served since, under the latter circumstances,
some smaller ventrolateral or ventral scales
will be included.

9) Number of transverse rows of ventral
scales in the snout-eye distance.

10) Number of lamellae on phalanges II
and III of the fourth toe.

11) Height of the nuchal crest scales,
categorized as very high, high, moderate,
or low.

12) Height of dorsal crest scales, cate-
gorized as high, moderate, or low.

13) Dorsal coloration and pattern of
males and females, separately.

14) Ventral coloration of males and fe-
males, separately.

15) Color of dewlap, in males and fe-
males, separately.

16 ) Color and pattern of chin and throat
in males and females, separately.

17) Color and pattern of upper surface
of head in males and females, separately.

18) Color of eyeskin.

19) Color and pattern of upper surfaces
of hindlimbs.

20) Color and pattern of juveniles and
subadults.

The above characteristics are variously
useful as far as delimiting the nameworthy
populations of A. ricordi. Detailed com-
ments on these characteristics are made be-
low, with especial attention to pitfalls in
their reliability and usage.

1) The snout-vent length of mature in-
dividuals of both sexes is remarkably uni-
form throughout the entire series. Males
are easily distinguished from females at
any age by the presence of one (occasion-
ally two) pairs of enlarged postanal scales.
Many ASFS specimens have the hemipenes
extruded. Males in general reach a larger
snout-vent length than females; the largest
male (ASFS V29284) has a snout-vent
length of 180 and is from sample (4).
whereas the largest female (ASFS V31397)
has a snout-vent length of 151 and is from
sample ( 12 ) . The mean difference in
snout-vent lengths between the two sexes
is about 10 mm in almost all samples with
the exception of maximally sized individ-
uals in both sexes in sample (2) where the
difference is 3 mm, sample (3) where the
difference is 27 mm, sample (4) where the

HisPANioLAN Giant Angle • Schwartz

difference is 32 mm, and sample (7) where
both sexes are of the same size. Whether
tliese exceptions to the 10-mm generahza-
tion are meaningful is debatable. At least
samples (2) and (4) include Icmg series
of specimens, and even samples (3) and
(7) include more than ten individuals.
Adults of the two sexes are readily distin-
guished moi-phologically, since males have
a high tail "fin" that is supported by the
bony extensions of the neural spines; this
feature is lacking in females, although they
may have a much reduced caudal crest in
the form of a low ridge. The terminal half
of the tails of many males is crestless; I at
first considered that this was due to break-
age with subsequent regeneration without
regeneration of the tail "fin." Many speci-
mens have this condition, however, without
any obvious change in basic caudal scale
shape and arrangement, and this is the nor-
mal condition in the tails of males. Often
the uncrested portion of the tail is quite
differently colored or patterned than the re-
mainder of the tail; such cases are due to
regeneration.

2) The number of snout scales at the
level of the second canthal has been em-
ployed as a characteristic to separate the
recognized subspecies. Williams ( 1962,
1965) recorded the following variation in
the four subspecies: ricordi, 7-9; Ijaleatus,
2-5; harahonae, 4-6; and leheri, 4-6 (3-6
on map, fig. 2, 1965: 7). It should be re-
called that Williams himself pointed out
that this count alone would not distinguish
all these taxa from each other. Certainly
overlap between haleatus and ricordi is
nonexistent, and between haleatus on one
hand, and harahoiiae or leheri on the other
hand, the overlap is small. Counts on 403
specimens made by myself do not extend
the parameters of snout scales at all: within
the entire lot of specimens, these scales
vary from 2 to 9, just as in Williams's data.
However, the variation within populations
may be much greater than Williams antici-
pated. For instance, in sample (13), the
counts vary between 2 and 7, and in series

(12) between 4 and 8. Most samples have
three or four categories of number of snout
scales. I am in no way implying that this
is an invalid or poor character for differen-
tiation of populations of A. ricordi, how-
ever; it is, rather, an extremely useful one
but requires amplification and interpreta-
tion.

If we examine only those samples (1-8)
which are assigned to haleatus, the amount
of variation in snout-scales is 2-5, exactly
that assigned to this taxon by Williams.
However, within the broad area covered by
haleatus, there are strong modalities of
snout-scales. In samples (1), (4), (5), and
(6), the mode is 2 scales, whereas in sam-
ples (2), (7) and (8), the mode is 4. Only
one sample, (3), has bimodes of 2 and 4
scales. In harahonae (9) the range is 2-
5 (mode 4). In those samples which are
associated with nominate ricordii (samples
11 and 12), leheri (sample 13) and adja-
cent Castillon material (sample 14), the
range is 2-9, thereby showing complete
overlap in range of this count with that of
both haleatus and harahonae. In fact, in
leheri ( 13 ) alone, the range of snout scales
(2-7) almost embraces that for all other
samples and thus the entire species. Mo-
dalities in this complex of samples are 5
(sample 11), 7 (sample 12), 4 (sample 13)
and 6 (sample 14). Sample (7) is nomi-
nate ricordi.

3 ) The number of vertical rows of loreal
scales ranges from 5 to 10. The greatest
variability is in samples (1), (2), (7), and
(12), where the row counts in each case
are 5-9, 6-10, 5-9, and 6-10. Most samples
have four categories of number of loreal
rows. The modes vary as follows: 5 (sam-
ple 11), 6 (samples 8, 9, 13), 7 (samples 1,
3, 4, 5, 6, 7, 12, and 14), and 8 (.sample 2).

4) The minimal number of scales be-
tween the supraorbital semicircles varies
between 1 and 5; no specimen has the semi-
circles in contact. Modes in general are
very strong, and the usual mode is 3 scales
(samples 1-7; sample 8 has a bimode of 2
and 3 scales); these are all samples that

Bulletin Museum of Comparative Zoology, Vol. 146, No. 2

are assigned to baleatiis. A mode of 3
scales occurs also in samples (11) and
( 13 ) , and of 4 in samples ( 12 ) and ( 14 ) .

5) The number of scales between the in-
terparietal scale and the supraorbital semi-
circles varies between 3/3 and 7/7. Modes
(which in some cases are quite strong and
in others less so) are: 4/4 (samples 1, 5, 6,
8, 9, 11, 13, 14) and 5/5 (samples 2, 4, 7,
12). Sample (3) is peculiar in having the
mode 4/5 (six of 16 lizards), with adjacent
counts of 4/4 (four lizards) and 5/5 (five
lizards). There is high variability in .this
count; it can be assessed in another fashion,
namely, the frequency with which any
scale count (i.e., 4, 5, 6, etc.) occurs within
the sample, regardless of its pairing with
another count on the other side of the head.
With the use of this technique, the fre-
quency of involvement of 4 scales in the
inteiparietal-semicircle contact varies be-
tween 43 percent (sample 1) and 67 per-
cent (sample 8), and of 5 scales between
52 percent (sample 2) and 64 percent
( sample 7 ) . Of the entire lot of specimens,
there is only one occurrence of 7/7 in this
position ( sample 12 ) but many occurrences
of 3/3 (samples 1, 5, 6, 9, 11, 12, 13, a total
of 12 incidences).

6) The number of scale rows between
the subocular scales and the supralabials is
fairly constantly 1, and this is a strong
mode or the exclusive category in all sam-
ples except sample ( 13 ) . In this lot ( which
is the type series, with a few new speci-
mens, of leberi), 48 percent of the lizards
have the suboculars and supralabials in
contact. Elsewhere, contact is absent in
samples (7) and (11). The frequency of
contact varies in all other samples between
3 percent (sample 9) and 17 percent (sam-
ples 6, 8, and 14). These three samples are
widely separated geographically and the
frequency in none of them even approaches
the very high incidence of contact in sam-
ple ( 13 ) . However, it is noteworthy that
samples (13) and (14) are adjacent geo-
graphically.

7), 8), 9) In reference to all counts in-
volving laying out the snout-orbit distance

on the body, I can do no better than once
more to reiterate the cautions previously
expressed ( Schwartz, 1964; Garrido and
Schwartz, 1968; Schwartz and Garrido,
1972) in reference to Anolis equestris and
members of the genus Chamaeleolis. For
these standard-distance counts, old and
poorly preserved, limp, curled, uninjected,
or otherwise distorted specimens are much
less useful and reliable than properly pre-
served, injected, and positioned lizards.
Luckily, by far the largest quantity of A.
ricordi under study are well preserved.
However, I have abandoned counts of hor-
izontal dorsals on young juveniles, even
well-preserved ones, or on any adults
whose condition precluded taking these
counts meaningfully. The juvenile situation
is peculiar in that invariably, despite the
relatively shorter snout of young specimens,
laying out this distance to count horizontal
rows involved including several rows of ex-
tremely lateral and ventral scales, which
are smaller than true dorsals and which
thus tend to increase the horizontal counts.
I have taken vertical dorsals and ventrals
on juveniles, however, and they do not
skew the data. Of the three standard-dis-
tance counts, those of vertical dorsals and
ventrals are much more reliable than are
those of horizontal dorsals.

Vertical dorsal scale rows vary between
12 and 26, with the low count in sample
( 4 ) and the high count in sample ( 12 ) .
Means range from 15.4 (sample 4) to 21.1
(sample 12). These two represent, respec-
tively, lots assignable to baleatus and ri-
cordi, sensu lato. No generalizations of
mean number of vertical dorsals in refer-
ence to samples and geography can be
made, since the range in samples now as-
sociated with baleatus varies between 15.4
and 19.2, with ricordi 17.3 and 21.1, bara-
honae 17.2, and leberi 16.5 and 16.7. Com-
parisons and significance of vertical dorsal
scale counts are shown in Table I.

Number of horizontal dorsal rows ranges
from 13 (sample 1) to 34 (sample 7).
Means vary between 17.1 (sample 11) to
24.6 (sample 12). The latter sample is that

HisPANioLAN Giant Angle • Schwartz

Table I.

Taxon

Comparison of number of vertical dorsal scales in popxjtlations
OF giant Hispaniolan angles

M (±2

standard

error of

mean)

c
a

c
c

■2

"a
£

-*-

—

ricordi 50

leheri 54

stihsolanus 15

harahonae 33

samanae 54

caendcolatus 37

litorisdva 16

scelestus 60

multistrtippus 20

sublimis 18

baleatus 15

21.1 ±

16.5 ±

17.3 ±

17.2 ±

16.6 ±

17.1 ±
15.9 ±

15.4 ±
18.6 ±

19.2 ±

.57
.46
.96
.67
.46
.72
.75
.45
1.06
.70

+ +

17.5 ± 1.16

A. r. victilus, A. b. alboceUatus, and A. b. fraiidator are not included. A "-)-" in the table indicates that the two
subspecies involved differ significantly ( non-overlap of two standard errors of mean ) ; a "— " indicates no statistical
difference. Note that the mean of A. r. ricordi differs significantly from the means of all other taxa; that of scelestus
differs significantly from those of all other taxa except litorisilva; and that of sublimis differs significantly from those
of all other taxa except multistruppus and baleatus.

of nominate ricordi, the former a peripheral
isolate of baleatus.

Ninnber of transverse rows of vential
scales varies between 15 (samples 1 and
13 ) and 34 ( sample 7 ) . Means range from
20.2 (sample 13) to 2.5.1 (sample 6); sam-
ple ( 13 ) is leheri.

10) The number of lamellae on pha-
langes II and III of the fourth toe varies
between 27 and 39. The variation in any
population is so great that this count is
meaningless as far as differentiation be-
tween any populations of A. ricordi.

11), 12) Williams (1962, 1965) de-
scribed the relative heights and lengths of
both the nuchal and dorsal body crest
scales. Certainly the differences between
baleatus and ricordi, for instance, are so
very obvious on casual examination that
one has no difficulty in ascertaining the
taxon involved. Williams also pointed out
(1962: 4-5) that in some specimens there
is "sometimes a regular alternation of rela-
tively high triangular single scales and
pairs of much lower, more quadrangular
scales," with the result, on some specimens,
of double crest scale rows on the neck; the
net effect of this condition is a rather in-
discriminate grouping or elongate patch of

multiple nuchal crest scales. Although tliis
condition occurs erraticallv, it seems to be
most predominant in specimens from the
Tiburon Peninsula, but it occurs elsewhere
in nominate ricordi and even occasionally
in specimens assigned now to baleatus.
Such a "hypertrophied" nuchal crest condi-
tion does not completely fit any logical geo-
graphical pattern nor is it totally consistent
within any sample, although there are ten-
dencies toward it as noted above. In any
event, it does not obscure the height of the
nuchal crest scales.

In my own analysis, I have used a
slightly different method in recording
height of crest scales. Nuchal scales were
recorded for each specimen as very high,
high, moderate, or low. Such a verbal quan-
tification is not totally satisfactory, since
the investigator's impressions may change
as the study progresses. To avoid this pit-
fall, I re-examined many specimens that
had been studied earlier in the work and
reconfirmed my own early impressions with
my later ones. Body crest scales were re-
corded as high, moderate, and low.

In nuchal crest scales, very high scales
are those which are very elongate, attenu-
ate, almost spinelike ( but of course flexible,

100

Bulletin Museum of Comparative Zoology, Vol. 146, No. 2

not stiff), with the base much shorter than
the height of the scale. High scales are
those which are shorter and less attenuate
than very high scales, but whose height is
still much greater than the base. Moderate
scales are lower and not attenuate, although
they may be pointed, with the height and
base about equal in length. Low scales are
lower than long. The same categories and
interpretations apply to body crest scales,
although no lizard has the body crest scales
so high as the nuchal scales.

Several other points are pertinent. I have
not used this datiun from juvenile and sub-
adult lizards ( all lizards below 100 mm
snout-vent length) since it is obvious that,
regardless of the taxon or sample, all young
A. ricordi have low nuchal and dorsal crest
scales, which, as the lizard matures, be-
come increasingly more specialized until
the adult condition is reached at about 110
mm snout-vent length. Thus young ricordi
and young baleatus, two taxa that are re-
markably distinct in this feature as adults,
are identical in crest development.

Secondly, it might be assumed that (es-
pecially) nuchal crest scales might be bet-
ter developed in adult males than in adult
females; this excessive development might
reasonably be assumed or construed as a
sexually dimorphic character, with hyper-
development in males. Such does not seem
to be the case, and many female baleatus,
for instance, have very high nuchal crest
scales, as high as those of males. In fact,
comparisons of males and females of indi-
vidual samples show that, within each sam-
ple, there is remarkable consistency be-
tween frequencies of the very high, high,
and moderate categories in both sexes.

Thirdly, as one might expect, there is a
sequence of crest scale heights in the nu-
chal-body series. If the nuchal scales are
very high, the body scales are high; if the
nuchal scales are moderate, the body scales
are low, etc. In no case have I recorded a
transition from very high nuchal crest
scales to moderate body scales, for exam-
ple. There is thus a definite correlation be-
tween height of nuchal scales and those of
the body crest.

13) -19) The color and pattern details of
Anolis ricordi throughout its range need
not be gone into in detail at this point. It is
now sufficient to acknowledge that these
lizards show metachrosis varying from
shades of green and green-gray to brown.
The pattern elements, however, are quite
stable, although the hues involved in the
pattern itself may change with changing
base colors. There is little evidence to in-
dicate that a lizard which is, for instance,
blotched in one color phase will become
crossbanded in another. Thus, despite
changes in hues and even in base colors,
patterns remain constant. It is of interest
to note that greens seem to be the colors
that predominate in the wild. The few A.
ricordi that I have seen during the day
have always been green. The many lizards
that I have seen and collected at night
were almost always green, and usually at
their maximum of pattern expression while
they slept. It is this nocturnal assumption
of the green phase coloration that renders
these lizards so very conspicuous at night
while they sleep on exposed branches,
vines, or among the arboreal greenery.
Even in those populations ( Boca de Yuma,
Sierra de Baoruco) in which the greens in-
volved are not bright, the lizards are still
quite obvious at night. It is only rarely
that one encounters a brown A. ricordi at
night. I have notes on only one instance of
this condition, a subadult lizard from the
Cordillera Septentrional.

15) The dewlap coloration in A. ricordi
is variable. In some populations, males
have a pale yellow to gray dewlap, whereas
in others the males have dewlaps that are
peach or vivid orange. It should be noted
that both sexes in A. ricordi have dewlaps
and that the general hue of the female
dewlap is like that of the males, except that
basally it is usually streaked with browns,
dark grays, or charcoal, and the ground
color is less bright than that of males. In
some regions, the female dewlap is quite
differently colored than that of males.

20) Juveniles and subadults present sev-
eral problems that are presently insoluble.
I suspect that much will be revealed once

HisPANioLAN Giant Angle • Schwartz

101

we know the repertory of pattern and color
in yoinig individuals, but these data are
not axailable nx)w. Although there are
many young specimens at hand, they are
inconsistent within populations or even
small samples. The juxenile color is nor-
mally some shade of green (or browns un-
der stress), most often with two to four
pale (cream, whitish, pale gray) cross-
bands. Many small specimens are a uni-
form green without any dorsal markings.
In other juveniles, the dorsum has many
conspicuous crossbands with two shades of
greens (or browns), separated by promi-
nent bands of pale greens (yellow-green,
pea-green), to give a very tigroid lizard;
this condition persists into the adults of one
population, as does the more simply
banded juvenile pattern noted above in
other populations. The juveniles usually
have dark gray to charcoal dewlaps, re-
gardless of their sexes, and often there are
charcoal or white markings on the neck or
aboN'c the forelimb insertion or somewhere
anteriorly. These variants are discussed un-
der each subspecies below, and there is no
need to go into the details here. However,
I do wish to point out that ju\'enile patterns
are more variable than are those of adults,
and that I do not know how to interpret
this situation.

SYSTEMATIC ACCOUNTS

AnoWs ricordi Dumeril and Bibron

Anolis ricordii Dumeril and Bibron, 1837. Erp.
gen., 4: 167.

Type locality. St.-Domingue; holotype.
Museum National d'Histoire Natmelle 1272.

Definition. A giant species of Hispan-
iolan Anolis characterized by the combina-
tion of moderate size (males to 160 mm,
females to 151 mm snout-vent length),
snout scales at level of second canthal
scales variable, between 2 and 9 (modally
4, 5, 6, or 7 by population) but usually 4
to 9 (97 percent), vertical loreal rows 5 to
10 (modes by population 5, 6, and 7),
scales between supraorbital semicircles 2 to
5 (modes 3 or 4 by population), inteipari-
etal scale separated from supraorbital semi-
circles modally by 4 or 5 scales, vertical

dorsal scale's generally small ( 14 to 26 in
standard-distance), ventral scales relatively
large (15 to 32 in standard-distance), nu-
chal crest scales in both sexes moderate to
low, never \'ery high or high, body crest
scales usually low, subocular scales usually
not in contact with supralabials but one
population is remarkably exceptional in
this character; dorsal body coloration and
pattern some shade of green, in some geo-
graphic regions flecked irregularly with
paler and darker green to give a beadwork
effect; male body pattern either of irregular
black to dark brown blotches on the neck,
occipital region of the head, and on sides
(often delimiting two pale longitudinal
bands) or with three longitudinal dark
brown stripes on each side or with dark
saddles and a bluish green flank stripe or
with a powdery pale blue-green lateral
stripe; females usually without dark dorsal
or lateral markings although in some areas
females have a darker brown reticulum,
three pale gray to yellow narrow cross-
bands, longitudinal black lines, or two pur-
ple flank stripes; dewlap in males variable,
from yellowish gray to peach, bright or-
ange, or deep yellow, in females from
peach to deep yellow or dull orange or
even inky brown or inky blue-black, chin
and throat dull yellow, yellow-green, or
pale blue-green in males, eyeskin dark
( charcoal, dark brown ) to light ( pale blue )
in males, charcoal to pale green in females,
and usually with a prominent pale subocu-
lar semicircle clearly delineated.

Distribution. The Tiburon Peninsula in
Haiti, east to the vicinity of Saltron, Dept.
de rOuest, thence north to the northern
Haitian coast from Port-de-Paix east to Ter-
rier Rouge and thence into the Republica
Dominicana east as far as the vicinity of
Los Quemados, Santiago Rodriguez Prov-
ince, south to Restauracion, Dajabon Prov-
ince; also extending from Haiti onto the
southwestern slopes of the Cordillera Cen-
tral in San Juan Province and in the Sierra
de Neiba in La Estrelleta and San Juan
provinces; altitudinal distribution from sea
level to elevations of about 4000 feet ( 1220
meters) south of Castillon, Dept. du Sud,

102 Bulletin Museum of Comparative Zoology, Vol. 146, No. 2

Haiti, and to 3500 feet ( 1068 meters ) west
of Marmelade in the Chaine de Marmelade,
Dept. de I'Artibonite, and 3400 feet ( 1037
meters) south of Elias Pifia in the Sierra de
Neiba, La Esti-elleta Province, RepubHca
Dominicana.

Anofis ricordi ricordi Dumeril and Bibron

Type locality. "St.-Domingue"; here re-
stricted to the vicinity of Port-au-Prince,
Dept. de I'Ouest, Haiti.

Definition. A subspecies of A. ricordi
characterized by the combination of mod-
ally 7 snout scales between second canthal
scales, 7 vertical rows of loreal scales, 4
scales between the supraorbital semicircles,
5/5 scales between the interparietal and the
supraorbital semicircles, high number of
vertical dorsal scales (17-26; mean 21.1),
high number of ventral scales ( 19-32; mean
24.7 ) , nuchal crest scales moderate ( rarely )
to low (usually), body crest scales moder-
ate (rarely) to low (usually), subocular
scales usually in contact with supralabial
scales; males usually with some black lat-
eral markings on the neck and thoracic re-
gion, and on the occipital region of the
head, but at times these markings are ab-
sent (see discussion below), females green
and without definite black lateral markings
but at times reticulate with brown, the re-
ticulum delimiting a pair of clear green lat-
eral stripes or with three pale gray to green
vertical narrow bars; dewlap variable, in
males from peach or pale peach to gray or
yellowish gray, and in females from peach
to blue-gray or inky blue or inky black ( see
discussion beyond).

Distribution. Northern Haiti from Port-
de-Paix east to Terrier Rouge and into the
Republica Dominicana as far east as the
vicinity of Los Quemados, Santiago Ro-
driguez Province, and as far south as Re-
stauracion, Dajabon Province, south in Haiti
to the Port-au-Prince region ( Morne De-
cayette, Diquini, Petionville), as well as
east into the Republica Dominicana in the
Sierra de Neiba and the southwestern
slopes of the Cordillera Central in La Es-
trelleta and San Juan provinces.

Discussion. I have little doubt that the
extensive range that I here ascribe to nom-
inate A. ricordi is incorrect. There are sev-
eral very obvious differences in coloration
and pattern between northern and south-
ern specimens of A. r. ricordi; thus the defi-
nition of the subspecies, in order to include
all pattern variants, is necessarily cumber-
some. The problem is presently unresolv-
able since, other than the series from near
Port-au-Prince and the specimens from
northern Haiti, there are huge areas in
Haiti whence specimens remain unknown.
A detailed discussion of the chromatic and
pattern features in the various segments of
A. r. ricordi is given below.

The series of 50 specimens assigned to
the nominate subspecies shows the follow-
ing variation. The largest male (ASFS
V31395) has a snout-vent length of 160,
the largest female (ASFS V31397) 151;
both are from 4.1 mi. NW Juan de Herrera,
San Juan Province, Republica Dominicana.
Snout scales at level of second canthals
vary between 4 and 8; the mode is 7 (22
specimens). The vertical loreal rows vary
between 6 and 10, with a mode of 7 (20
specimens). There are between 3 and 5
scales between the supraorbital semicircles
(mode 4). There are modally 5 scales be-
tween the interparietal and the semicircles;
5 scales are involved in 53 percent of the
combinations; actual counts are 3/3 (1),
4/4 (10), 4/5 (8), 5/5 (17), 5/6 (9), 6/6
(2), 7/7 (1), and 5/7(1). Vertical dorsals
range between 17 and 26 (mean 21.1), hor-
izontal dorsals between 19 and 30 (24.6),
and ventrals between 19 and 32 (24.7). Of
28 adult males, six have moderate nuchal
crest scales and 22 have these scales low;
of 11 females, all have the nuchal crest
scales low. Body crest scales are moderate
in two males and low in 26, and 11 females
have the body crest scales low. The sub-
oculars are separated from the supralabials
in 45 of 49 instances, and contact between
these scales occurs in four lizards (8 per-
cent ) .

The southern specimens from the Port-
au-Prince region and including two from

HisPANioLAN Giant Anole • Schwart::.

103

Mirebalais, consist of ten adult males, three
adult females, and two juvenile females
(MCZ 60013-14). The latter two speci-
mens (with snout-vent lengths of 89 and
92 mm) can be easily dismissed in that
they are presently patternless green. Color
notes in life on one southern male (ASFS
V9024) state that in the green phase, the
lizard had the dorsum a mixture of pale
green, brown, and yellow, with green the
predominant color, the net effect being one
of bead work. The upper surface of the
head was a mixture of pinkish and yellow
scales, the mental region and adjacent up-
per labials were dull yellow, and the venter
pale green. The dewlap was peach with
the dewlap scales yellow. All males (with
the exception of MCZ 69404, which is an
albino ) have some black to dark brown oc-
cipital blotching, usually extending onto
the neck and thence onto the area above
the forelimb insertion. The extent of the
dark anterior markings is variable, but they
are present in all males and quite vivid in
freshly taken specimens. A pale subocular
crescent is present, often extending posteri-
orly to form a pale preauricular blotch,
bounded above by a large dark temporal
blotch that may form an occipital chevron
by joining its mate on the other side.
Southern females were recorded in life as
pale to bright green without any dark pat-
tern, with a moderately well-defined pale
subocular crescent that may expand into a
pale preauricular blotch; the venters were
yellow-green. In one female, the dewlap
was reported as blue-gray with yellow-
green streaks. The hindlimbs are not prom-
inently banded. Neither sex has the throat
marked with any blotching or dotting, al-
though the females have the throat some-
what clouded with dull dark green.

The series for the Sierra de Neiba and
the Cordillera Central consists of six males,
four females, and eight juveniles and sub-
adults. The males were described in life
as being dark green with pale green cross-
bands, or pale green with four paler green
crossbands, or patternless green. The up-
per surface of the head was creamy tan.

black blotches occur in this series of males
also, but the occiput lacks clearly defined
black areas, and the black on the body is
much more extensive than it is in all south-
ern males, the extreme condition being that
shown by ASFS V31395, which has exten-
sive black blotching over two-thirds of the
back and sides. The pale subocular cres-
cent is very obscure, but there is a promi-
nent pale preauricular spot in most males.
The females from this region are plain
green, without dark markings, and there is
a prominent pale supralabial blotch in the
area which in males is occupied by the pale
preauricular blotch. Dewlaps in males and
females were invariably recorded as peach,
and both sexes had charcoal eyeskins. As
in southern specimens, the chins and
throats are pale green and without any defi-
nite markings, except that the throats of fe-
males are sufi^used with darker green. The
eight juveniles and subadults range in
snout-vent length from 68 mm to 92 mm.
The two smallest specimens, a male and a
female, were rich pea-green in life with
four, narrow, cream transverse crossbars,
and the smaller had in addition black
streaking in the green areas and a black
postauricular smudge. The ventral color
was rich pea-green and the dewlap skin
was blue-black. All juveniles and subadults
with snout-vent lengths of between 76 and
92 were bright emerald-green dorsally and
without any dorsal dark or pale markings;
one female juvenile ( snout-vent length 78 )
had a lateral black nuchal spot followed by
a bright yellow preaxillary bar, as well as
a bright yellow subocular mark. The dew-
lap was recorded as black in a juvenile fe-
male with a snout-vent length of 89. Of
the subadults, the most peculiar is a male
(ASFS V31323) with a snout-vent length
of 90 which shows, as preserved, a vague
series of vertical lateral pale and dark
areas, but as yet no black blotching typical
of adult males.

The northern specimens are six males and
one female from Haiti, five males and two
females from the Republica Dominicana,
and one Haitian subadult and two Domin-

104 Bulletin Museum of Comparative Zoology, \o\. 146, No. 2

ican subadult and juvenile lizards. Haitian
males are not only quite different from
northern Dominican males, but they are
also strikingly different from central and
southern males. In the Haitian males, the
dorsum is gray-green with yellow-green
flecking, or a beadwork mixture of dark and
light green scales. The upper surface of the
head is dark with light flecking, and in one
male the head was recorded as dark brown
with the centers of the scales pale purple.
No male has any occipital dark blotching,
and any body blotching, if it is at all pres-
ent, is extremely restricted and maximally
expressed as small black areas above the
forelimb insertion ( MCZ 66147). The pale
subocular crescent is obscure, but there is
a pale postlabial line leading to the auricu-
lar opening. Northern Dominican males,
on the other hand, are brightly colored and
have extensive black neck and side mark-
ings; 'n two males these latter extend far
posteriorly on the body and tend to delimit
two lateral stripes on each side. The upper
surfaces of the head are not mottled but
are pale uniform tan. In life, the pale sub-
ocular crescent is bold and pale blue to
white, and it may extend to the auricular
opening. In Haitian males, the dewlap is
grayish to yellowish peach (pi. 12C5; all
color designations from Maerz and Paul,
1950), pale gray-green (about pi. 19B2),
or yellowish gray (about pi. 20B1). In
northern Dominican males, the dewlap is
pale peach to pale yellow or grayish yellow,
and the dewlap may be speckled with
brown basally.

The single Haitian female is presently
unmarked green, with faint scattered cream
flecking. The larger of the two Dominican
females was pale green above with a darker
brown reticulum outlining a pair of green
lateral stripes on each side. There was a
postauricular brown smudge, followed by a
pale blue axillary smudge. The temples
were yellow-green, the lores pale blue and
brown, the eyeskin pale green, and there
was a pale blue subocular crescent that ex-
tended into a preauricular pale blotch. The
top of the head was marbled pale tan and

dark brown, and the venter was the same
color as the dorsum. The other Dominican
female was green without any dorsal mark-
ings.

The Haitian subadult (MCZ 66148) is a
female with a snout-vent length of 106. It
is speckled with pale scales on a dark
ground like Haitian males. The smaller of
the two Dominican males (ASFS V18008)
has a snout-\'ent length of 75 and was
bright yellow-green above with two cream
crossbands and a yellow subocular cres-
cent. The second Dominican male (ASFS
V32160) has a snout-vent length of 103
mm, and, like Dominican adult males, has
extensive black blotching on the head,
neck, and almost the entire dorsum. The
ground color was pale green, and the dew-
lap was dark brown.

To summarize all the above data, it is
obvious that I have included several popu-
lations in A. r. ricordi which differ rather
strikingly among themselves. Southern
Haitian males are marked with black on
the occiput, neck, and anterior sides, and
central Dominican and northern Domini-
can specimens increase this tendency to
show even more extensive black lateral
markings. On the other hand, northern
Haitian males as a group show very little
or no black markings and are basically
green-flecked green lizards. Northern Do-
minican males are much more colorful than
specimens from elsewhere, and much more
contrastingly marked. On the other hand,
all females are fairly similar, with the ex-
ception of the remarkably colored and pat-
terned female from the northern Republica
Dominicana. I suspect that it will ulti-
mately be shown that there are at least two
more nameworthy populations included in
A. r. ricordi as here defined by me: a north-
ern Haitian subspecies, a northern and cen-
tral Dominican population, as well as the
southern Haitian one. But the specimens
at this time are from such disjunct localities
and are so limited in number that I am un-
willing to make the suggested nomencla-
tural additions.

Remarks. A. r. ricordi occurs in a wide

HisPANioLAN Giant Anole • Schwartz

105

variety of situations but is of course always
associated with trees. Its altitudinal range
is from sea level' at many localities to eleva-
tions of 3500 feet (1068 meters) in the
Chaine de Marmelade in northern Haiti
and 3400 feet (1037 meters) in the Sierra
de Neiba. Almost all specimens taken by
myself and parties were secured sleeping
at night. \\^illiams (1965: 2-3) noted that
in the Monte Cristi region these lizards
sleep in viny tangles, especially where there
are dense "mats" or "curtains" of vines un-
der a canopy. Such a situation is ideal in
the xeric forests in the Monte Cristi area.
At Las Matas de Farfan, the lizards were
easily secured at night in a high-canopied
cafetal, sleeping on limbs, branches, or on
vines, either vertical or horizontal. A speci-
men from Morne Calvaire near Petionville
was seen during the late morning on a
mango tree in an open pasture, about 4 feet
(1.2 meters) above the base. Thomas com-
mented in his field notes upon a specimen
from Le Borgne which was observed 8 feet
( 2.4 meters ) above the ground on the trunk
of a tree; this male led the pursuers a merry
chase through a series of three trees and
finally sought refuge in dense grass on the
ground, where it was caught! The male
from Terrier Rouge was collected with a
slingshot while it rested head-down on the
main branch of a large tree 15 feet (4.6
meters) above the ground. South of Las
Matas de Farfan I secured a juvenile sleep-
ing on a horizontal vine in a tree-fern
thicket adjacent to a mountain brook. The
association of A. r. ricordi with rivers or
lakes is certainly fortuitous; the greatest
concentrations of these lizards occur in such
obviously mesic situations only because
there is often gallery forest restricted, or
limited by man, to streamsides. However,
such a situation is not a guarantee of secur-
ing specimens. In our effort to narrow the
previously existing gap between ricordi and
haleatus in the northwestern Republica
Dominicana, we questioned natives con-
cerning the occurrence of saltacocotes in
the region along the gallery-forested Rio
Yaque del Norte, which here passes

through cactus desert. W'e were assured
that the lizards indeed occurred there, and
we were fortimate in finding a superb area
of gallery forest in the steep-sided valley of
the Rio Guarabo, west of Los Qucmados.
The Guarabo is a southern affluent of the
Yaque, and we had no doubt that these
splendid hardwoods harbored A. ricordi.
But it was not until our fifth nocturnal visit
that a single subadult was secured, despite
the attentions of four collectors. The woods
here present a perfect aspect for A. ricordi
— dense and large trees connected and in-
terlaced with vines and lianas, all quite
rich and mesic; still, our experience indi-
cates that, at least at the time of our visit,
A. ricordi was distinctly uncommon or diffi-
cult to see in what elsewhere surely would
have been a typically simple area for col-
lection of these lizards. In this instance,
demands for at least one specimen from this
region forced persistence which ultimately
yielded the requisite animal. Such may
well be the case in many otherwise xeric
regions, where A. ricordi is restricted to
(and perhaps is rare in) more mesic river-
ine hardwood stands.

Specimens examined. HAITI: Dcpt. de
rOuest, Source Leclerc, Morne Decayette
(MCZ 65729-31); Diquini (MCZ 8619,
USNM 118902, USNM 123347, USNM
123988); Port-au-Prince (AMNH 49501);
Petionville (MCZ 60013-14); Morne Cal-
vaire, 1 mi. (1.6 km) SW Petionville, 2300
feet (702 meters) (ASFS X1711, ASFS
V8514, ASFS V9024); Mirebalais (MCZ
69404); Lancironelle, nr. Mirebalais (not
mapped) (MCZ 68479); Dcpt. de I'Arti-
honite, 8-9 km W Marmelade, 3500 feet
(1068 meters) (ASFS V9925); Dcpt. du
Nord Quest, Port-de-Paix (MCZ 63338);
Dcpt. du Nord, 3 mi. (4.8 km) SW Le
Borgne (ASFS V10005); 2 mi. (3.2 km)
SW Cap-Haitien (ASFS \'10766); Ti
Guinin, nr. Cap-Haitien (not mapped)
(MCZ 66147-49); 8 mi. (12.8 km) E Ter-
rier Rouge (ASFS V10169). RFPOBLICA
DOMINiCANA: Monte Cristi Province, 1
km W Copey (ASFS V1269, ASFS V1411-
12, ASFS V1470); Laguna de Salodillo, 7

106 Bulletin Museum of Comparative Zoology, Vol. 146, No. 2

km E Pepillo Salcedo (ASFS V1413);
Dajahon Province, Restauracion (ASFS
V18006-08); Santiago Rodriguez Province,
1.8 mi. (2.9 km) W Los Quemados, 500
feet (153 meters) (ASFS V32160); La Es-
trelleta Province, 6.7 mi. ( 10.7 km ) E
Hondo Valle, 2500 feet (763 meters)
(ASFS V31428); 11.0 mi. (17.6 km) S
Elias Pina, 3400 feet (1037 meters) (ASFS
V31509); San Juan Province, 4.9 mi. (7.8
km) NW Vallejuelo, 2400 feet (732 meters)
(ASFS V31305); 6.1 mi. (9.8 km) S Las
Matas de Farfan, 1800 feet (549 meters)
(ASFS V14562, ASFS V31469, ASFS
V31319-26); 4.1 mi. (6.6 km) NW Juan de
Herrera, 1600 feet (488 meters) (ASFS
V3139.5-99).

Anolis ricordi leberi Williams

AnoUs ricordii leberi Williams, 1965. Breviora,
Mus. Comp. Zool., No. 232: 4.

Tijpe locality. Camp Perrin, Departe-
ment du Slid, Haiti; holotype, MCZ 80935.

Definition. A subspecies of A. ricordi
characterized by the combination of mod-
ally 4 snout scales between second canthals,
6 vertical rows of loreal scales, 3 scales be-
tween the supraorbital semicircles, 4/4
scales between the interparietal and the
supraorbital semicircles, low number of
vertical dorsal scales (14-21; mean 16.5),
low number of ventral scales ( 15-28; mean
20.2), nuchal crest scales usually moderate,
rarely low in males, usually low, occasion-
ally moderate in females, subocular scales
in contact with supralabials in almost 50
percent of the specimens; males either pale
yellow-green with four dark saddles and a
bluish green flank stripe, or with about
three longitudinal dark brown lateral
stripes, or simply dark brown, females
bright green (much brighter than males),
with longitudinal black lines indicated and
at times a greenish tan middorsal wash;
dewlap bright orange or orange with an
anterior brown wash in males, and dull or-
ange, at times suffused or marbled with
brown, in females.

Distribution. Known only from the vi-
cinity of the type locality and Marceline,

on the southern slopes of the Massif de la
Hotte, between elevations of 1000 and 1220
feet (305 and 372 meters), Dept. du Sud,
Haiti.

Discussion. In contrast to the situation
in A. r. ricordi, A. r. leberi is known from
a long series of specimens all from the same
general area, at elevations between 1000
feet and 1220 feet (305 and 372 meters).
Williams (1965: 6) assigned a single juve-
nile (MCZ 38277) from Tardieu, near Pic
Macaya, Dept. du Sud, Haiti, to leberi
with some reservation. This locality is
northwest of Camp Perrin-Marceline, is on
the northern slopes of the Massif de la
Hotte, and is much closer to the known
distribution of the next subspecies to be
described below.

The series of 54 A. r. leberi shows the
following variation. The largest male
(ASFS X3034) has a snout-vent length of
147, the largest female (AMNH 98723)
153; both are from Camp Perrin. Snout
scales at level of the second canthal are ex-
tremely variable, and range between 2 and
7; the mode is 4 (23 specimens). The ver-
tical loreal rows vary between 5 and 8, with
a mode of 6 (26 specimens ) . There are be-
tween 1 and 4 scales between the supraor-
bital semicircles ( mode 3 ) . There are mod-
ally 4 scales between the intei-parietal and
the semicircles; 4 scales are involved in 64
percent of the combinations; actual counts
are 3/3 (3), 3/4 (7), 4/4 (25), 4/5 (10),
5/5 (2), 5/6 (2), 6/6 (1), 3/5 (1), and 5/7
( 1 ) . Vertical dorsals range between 14 and
21 (mean 16.5), horizontal dorsals between
15 and 24 (18.0), and ventrals between 15
and 28 (20.2). Of 39 adult males, 30 have
the nuchal crest scales moderate and nine
have them low; of 13 females, five have the
nuchal scales moderate and eight have them
low. Body crest scales are moderate in 12
males and low in 27 males, whereas only
one female has the dorsal crest scales mod-
erate and 12 have them low. The subocu-
lars are separated from the supralabials by
1 row of scales in 28 specimens and are in
contact with the supralabials in 26 speci-
mens. A. r. leberi is the only population

HisPANioLAN Giant Angle • Sclnvnrfz 1U7

that has sucli a high incidence (48 percent) from tlie type locahty. The smallest (MCZ

of subocular-supralabial contact. 83982 ) is a female with a snout-vent length

Males show three basic patterns: 1) dor- of 52. The body is longitudinally streaked,
sal ground color pale yellow-green with but there are as yet no definite longitudinal
four dark brown saddles and a bluish green lines. The subadults ( MCZ 80949-50, a
flank stripe that is complete; 2) about three male with a snout-vent length of 105, and
longitudinal dark brown stripes, the ecu- a female with a snout-vent length of 93)
tral one being the most prominent and both show indications of the longitudinal
complete; 3) and a uniform dark velvety stripes that are characteristic of adults, but
brown. In the two lighter phases, the eye- the stripes are better defined in the sub-
skin is pale blue, chin and throats are dull adult male than in the female. The two
yellow-green, and the subocular crescent is adult males and two adult females from
pale blue and very conspicuous. The dew- Marceline agree in all pattern details with
laps in males are orange (brighter than any the topotypical series; Marceline and Camp
Macrz and Paul designation) or orange Perrin are separated by about 4.5 kilome-
with an anterior brown wash. Females are ters airline.

bright green dorsally (much brighter than Comparisons. Although both A. r. ri-

males) with longitudinal black lines indi- cordi and A. r. leheri have several features

cated. There is a greenish tan wash on the in common, namely, the moderate to low

upper surface of the head, and there may nuchal and body crests, the presence of

be a greenish tan middorsal zone on the some sort of black body markings, and a

body. The dewlap in females is dull or- prominent pale subocular crescent, these

ange, often suffused or marbled with two subspecies are eminently distinct,

brown, and the eyeskin is green, paler than They differ in: modal number of second

that of the dorsum. In males, the venter is canthal snout scales (ricordi 7, leheri 4),

pale green and may be washed with brown modal number of loreal rows ( ricordi 7,

even in the green phase, and in females leheri 6), modal number of scales between

the venter is pale yellow-green, paler than the interparietal and supraocular semicir-

the bright green of the dorsum. cles {ricordi 5/5, leheri 4/4), higher means

In general aspect, male A. r. leheri are of vertical dorsal scales and ventrals (21.1,
lineate dorsally and laterally, the bold dark 24.7 in ricordi, 16.5, 20.2 in leheri, respec-
longitudinal lines usually interrupted by tively) and the very high incidence of con-
four irregular pale vertical crossbands, tact between the subocular scales and the
which are in tinn bordered with darker supralabials in leheri versus the rarity of
pigment. Although my field notes indicate this condition in ricordi. In addition, the
that there are about three longitudinal dark dewlap in male ricordi is most often some
stripes in males, these three stripes are the shade of peach (although the variation in
result of modification of two stripes, of dewlap shades and colors in ricordi is read-
which the more dorsal is the broader. In ily acknowledged), whereas in male leheri
many specimens, this upper flank stripe the dewlap is orange or orange with a
maintains its integrity, but in many others brown anterior wash. A ready hallmark be-
the upper stripe is hollowed centrally, re- tween the two subspecies is the presence of
suiting in three narrow dark stripes, rather a pale preauricular blotch in ricordi, a con-
than two stripes, of which the upper is very dition always absent in both sexes of leheri,
broad and the lower is narrow. Although with the result that instead of the pale sub-
females show some longitudinal striping, it ocular crescent's being incoi-poratcd into a
is much less conspicuous than in males, postlabial line or preauricular blotch as it
Male throats are immaculate, whereas fe- often is in ricordi, it is a bold and contrast-
male throats are suffused with dark green. ing pattern element.

There are three juveniles and subadults Remarks. All Camp Perrin specimens of

108 Bulletin Museum of Comparative Zoology, Vol. 146, No. 2

A. r. leheri were collected by natives and
thus I have no precise knowledge of the
habitat nor habits of this subspecies. Camp
Perrin lies in the lower southern foothills of
the high Massif de la Hotte, at about 1000
feet (305 meters), and the area in general
is very mesic and presumably was once
well forested, although now it supports
cafeieres with a high-canopy hardwood
shade cover. Williams (1962: 10) cited
field notes by A. S. Rand and J. D. Lazell,
Jr., on A. r. leheri at Camp Perrin and Mar-
celine; both accounts involve trees associ-
ated with coffee plantings.

Specimens examined. HAITI: Dept. du
Slid, Camp Perrin (ASFS X3033-,35, ASFS
X3038-39, ASFS X3041-42, ASFS X3182,
AMNH 93713-36, MCZ 80935-37, MCZ
S0939-42, MCZ 80944-53, MCZ 83982);
Marceline (MCZ 121115); Marceline area,
ca. 1000 feet (305 meters) (MCZ 122269,
MCZ 121779-80).

Anolis ricordi viculus new subspecies

Holotype. USNM 193974,^ an adult
male, from Castillon, 2500 feet (763 me-
ters), Departement du Sud, Haiti, taken by
native collector on 2.5-26 June 1971. Orig-
inal number ASFS V25059.

Parotypes. ASFS V25058, same data as
holotype; ASFS V25060, same locality and
collector as holotype, 27 June 1971; ASFS
V24801, ca. 2 km (airline) S Castillon,
3500-4000 feet (1068-1220 meters), Dept.
du Sud, Haiti, R. Thomas, 24 June 1971;
ASFS V9335, ca. 5 km (airline) SE Marche
Leon, 2200 feet (671 meters), Dept. du
Sud, Haiti, native collector, 15 March 1966;
MCZ 119035, Castillon, Dept. du Sud,
Haiti, T. P. Webster, A. R. Kiester, and na-
tive collectors, 31 August 1969.

Definition. A subspecies of A. ricordi
characterized by the combination of mod-
ally 6 snout scales between the second can-
thals, 7 vertical rows of loreal scales, 4
scales between the supraorbital semicircles,
4/4 scales between the interparietal and
the supraorbital semicircles, low number of
vertical dorsal scales (15-19; mean 16.7),
moderate number of ventral scales ( 19-24;

mean 21.8), nuchal crest scales usually
moderate but occasionally low in both
sexes, dorsal body crest scales low in both
sexes, subocular scales almost always sep-
arated by one row of scales from suprala-
bial scales; males bright green dorsally
with powdery pale blue-green lateral
stripes, throat pale green and unmarked,
venter pale green with pinkish and yellow-
ish suffusions, females dark olive-green to
bright green with two purple to powdery
blue-gray flank stripes edged with dark
brown, lower sides spotted bright green,
yellow-green, or bright green with four
bright yellow-green crossbands edged with
black, throat pale green; dewlap deep yel-
low to orange in males, dull orange (al-
most brown) to deep yellow with orange
streaking and bluish edge in females.

Distribution. Known only from the vi-
cinity of Castillon on the northern slopes
of the Massif de la Hotte at elevations be-
tween 2200 and 4000 feet (671 and 1220
meters) on the Tiburon Peninsula in Haiti;
probably the subspecies occurring at Tar-
dieu near Pic Macaya (see discussion).

Description of holotype. An adult male
with a snout-vent length of 143 and a tail
length of 165 (regenerated); snout scales
at level of second canthals 6, 7 vertical
rows of loreal scales, 3 scales between su-
praorbital semicircles, 6/5 scales between
interparietal and supraorbital semicircles,
vertical dorsals 15, horizontal dorsals 22,
ventrals 20, one row of scales between sub-
oculars and supralabials, fourth toe lamel-
lae on phalanges II and III 31, nuchal crest
scales moderate, body crest scales low; in
life, bright green above with a pair of lat-
eral stripes on each flank powdery pale
blue-green, the same color also on the face;
throat and neck pale blue-green; venter
pale green with pinkish and yellowish suf-
fusions; dewlap deep yellow, almost or-
ange.

Variation. The series of three males and
three females shows the following varia-
tion. The largest male (ASFS V25058) has
a snout-vent length of 148, the largest
female (ASFS V25060) 141; both are from

HisPANioLAN Giant Angle • Schwartz

109

Castillon. Snout scales at level of the sec-
ond canthal range between 5 and 9; the
mode is 6 (four, specimens). The vertical
loreal rows are 6 or 7, with a mode of 7
( five specimens ) . There are be^^veen 3 and
5 scales between the supraorbital semicir-
cles (mode 4). There are modally 4 scales
between the interparietal and the supraor-
bital semicircles; 4 scales are involved in 58
percent of the combinations; actual counts
arc 4/4 (3), 4/5 (l),5/5 (1), and 5/6 (1).
X'ertical dorsals range between 15 and 19
(mean 16.7), horizontal dorsals between 17
and 27 (20.0), and ventrals between 19
and 24 (21.8). Of three males, two have
the nuchal crest scales moderate and one
has them low; the same situation applies
to the three females. All specimens have
the body crest scales low. The suboculars
are usually separated from the supralabials
by one row of scales and are in contact
with the supralabials in one individual ( 17
percent ) .

Thomas's field notes on three males show
the variation in dorsal coloration and pat-
tern. The dorsum was bright green with
the flank stripes powdery pale blue- green,
this color occurring also on the face. The
throat and neck were also pale blue-green
and the venter was pale greenish with pink
and yellow suffusions. One male (ASFS
V9335) also had a white shoulder patch,
but other pattern details on this individual
were lacking since the specimen was badly
damaged. The dewlap in the males was re-
corded as deep yellow ( almost orange ) and
orange (PI. 11L6). One female was green
to dark olive-green dorsally with two pur-
ple flank stripes, edged with dark brown,
which were powdery blue-gray anteriorly.
The lower sides were spotted and suffused
with bright green or yellow-green. The
venter was pale green with a pinkish wash
in the pectoral region. The second female
was marked in quite a different fashion,
and the specimen still maintains the pattern
after preservation. The dorsum was bright
green with four bright yellow-green trans-
verse body bands with black edges; in this
specimen longitudinal stripes were also

present but only in the nuchal region, and
the venter, including the throat, was pale
green. In both females, the dewlaps were
recorded as "very dull orange" and "deep
yellow, almost brown, anteriorly, with or-
angish longitudinal striae, each edged with
dark gray-green, between striae pale gray-
green and most basal striae greenish; edge
of dewlap grayish (faintly blue)."

Comparisons. A. r. viculus is so very
different from A. r. ricordi in both color
and pattern that detailed comparisons are
hardly necessary. The black occipital, nu-
chal, and anterior body blotches of male
ricordi are absent in male viculus, and the
longitudinally striped pattern in both sexes
of viculus does not occur in ricordi. The
two subspecies differ also in scale counts,
as follows: modal number of snout scales
at second canthals ( ricordi 7, viculus 6 ) ,
scales between interparietal and supraor-
bital semicircles {ricordi 5/5, viculus 4/4),
and much lower means of vertical dorsal
and ventral scales (21.1, 24.7 in ricordi,
16.7, 21.8 in viculus, respectively). The
two taxa are similar in number of loreal
rows, number of scales between the semi-
circles, and in relative frequency of contact
between the subocular and supralabial
scales.

In every way, viculus is much closer to
leheri than to ricordi. The basic pattern
elements are comparable in these two sub-
species, since both are lineate; however,
the longitudinal flank stripes in leheri are
dark, whereas in viculus they are light; the
single banded female viculus is quite dif-
ferent in general aspect from banded le-
heri. As far as scale counts are concerned,
the two subspecies differ in the following
manner: modal number of snout scales at
second canthals {leheri 4, viculus 6), num-
ber of vertical loreal rows {leheri 6, vicu-
lus 7), and scales between supraorbital
semicircles {leheri 3, viculus 4). In mean
number of vertical dorsals and ventrals, the
two subspecies are very similar, and both
have the 4/4 condition as the mode for the
interparietal-semicircle relationship.

Discussion. Williams (1962: 7-8) con-

110 BuUetm Museum of Comparative Zoology, Vol. 146, No. 2

sidered the four specimens then available
from the central portion of the Tiburon
Peninsula as intergrades between ricordi
and leheri. A few more specimens have ac-
cumulated since that time; now there are
one adult male, one juvenile male, and six
adult females from this central region, as
follows: HAITI: Dept. du Sud, Pemel, nr.
Miragoane (not mapped) ( MCZ 66015-
16), PaiUant, 1800 feet (549 meters)
(ASFS V26535-37); Fond des Negres
(ASFS V26254, USNM 72631, USNM
72633). As preserved, the adult male
shows fairly obvious longitudinal streaking
of gray and dull green, a few scattered
dark flecks or small blotches above the
forelimb insertion, and a prominent suboc-
ular pale crescent. The adult male is an
almost ideal representation of extreme in-
tergradation between viculus and ricordi,
with both pale longitudinal stripes and
scattered remnants of the typical ricordi
extensive body blotching. Three recently
collected females in life were green with
longitudinal stripes, which were delimited
by absence of black flecking that occurs
elsewhere on the green ground. In the
brown phase, these longitudinal stripes had
a reddish wash. In all females, the pale
subocular crescent is obvious and bold, and
in one female (MCZ 66016) there is an ad-
ditional preauricular pale area that resem-
bles the condition in nominate ricordi. I
have no color data on the male dewlaps,
but that of one female (ASFS V26535) was
dull yellow distally and pale blue, smudged
with charcoal, basally; the dewlap scales
were yellow-green. The juvenile male
(snout-vent length 79) in life had a pat-
tern of longitudinal dorsolateral stripes and
dorsal crossbands, with a pale yellow sub-
ocular crescent. I interpret these lizards as
intergradient between ricordi and viculus.
The central Tiburon localities, however,
are far removed from the known localities
of viculus (110 km) on one hand and of
ricordi (70 km) on the other. Williams
(1965: 7) regarded the Fond des Negres
and Pemel specimens as ricordi X leheri
intergrades, and they could indeed be so

interpreted. Since, however, leheri occurs
on the southern slopes of the Massif de la
Hotte, and viculus on the northern slopes
of that range, and since all intergradient
specimens are from the northeastern re-
gions of the extreme eastern portion of the
Massif de la Hotte, it seems much more
likely that these central Tiburon specimens
are intergradient between ricordi and vic-
ulus on geographic grounds. They do not
disagree with my concepts of how inter-
grades between these two subspecies prob-
ably should appear.^

1 Since the above comments on the intergradi-
ent specimens were written, Williams has secured
a series of 28 lizards (MCZ 132302-29) from St.
Croix, 1 mi. (1.6 km) from Paillant, Dept. du
Sud, Haiti, from this same general region. There
are no color data on the specimens. The measure-
ments ( in mm ) and scale counts of these lizards,
combined with those from the eight previously
available soecimens, follow. Largest male ( MCZ
132325) 155, largest female (ASFS V26535) 148.
Snout scales at second canthals 4-9 (mode 6);
loreal rows 4-8 ( mode 6 ) . Modally 4/4 scales
between the interparietal and the supraorbital
semicircles; other counts: 3/3 (2), 3/4 (2), 4/5
(4), 5/5 (11), 5/6 (1), 3/5 (1); 4 scales are in-
volved with 50 percent of the combinations. Ver-
tical dorsals range between 14 and 21 (mean
17.0), horizontal dorsals between 15 and 24
(18.3), and ventrals between 16 and 31 (21.4).
Of the males, four have the nuchal crest scales
moderate and 15 have them low; of the females,
one has the nuchal crest scales moderate and 15
have them low. Body scales are low in all adult
specimens. The suboculars are modally separated
from the supralabials by one row of scales and
are in contact with the supralabials in seven liz-
ards ( 19 percent).

In scale counts, the entire series is much closer
to viculus than to ricordi; however, in some char-
acteristics, the series is closer to lebcri or to the
subspecies yet to be described from extreme south-
eastern Haiti. In fact, comparison of the scale
counts shows that there is little resemblance be-
tween the modes and means between these geo-
graphically intermediate specimens and nominate
ricordi, and as a whole they seem much more
closely allied to one of the other Tiburon subspe-
cies.

The males in the St. Croix series are variable
in pattern but none shows any clear-cut dark
blotching, typical of A. r. ricordi. Some males
are more or less unicolor (medium brown as pre-
served), whereas others have longitudinal stripes,
alternating light and dark, with usually one broad

HisPANioLAx Giant Angle • Schtvartz

111

There remains one other specimen from
the distal portion of the Tibnron Peninsnhi;
this is a jnvenile male ( MCZ 38277) with
a snont-vent length of 7(S, from Tardieu,
near Pic Macaya, collected by P. J. Darling-
ton. It is presently dull brown, but there
are clear indications of black-edged dorsal
crossbands that closely resemble the condi-
tion in one of the female paratypes of vi-
culus. Tardieu is presently unlocatable on
modern maps, but Darlington has indicated
to Williams that this place lies just to the
north of Pic Macaya, and thus rather close
to Castillon. Since there are no juvenile

dark stripe along the upper sides and most prom-
inent. One male has extensive dark brown body
markings, vertically oriented and alternating with
paler tannish areas to give a more-or-less verti-
cally barred appearance. The pale subociilar
crescent is very obvious in all males, and there is
no indication of a pale preauricular area.

The females are undistinguished. Most are
more or less solid green with some scattered paler
green scales to give a beadwork effect dorsally,
but there are also indications in some specimens
of longitudinal paler areas to give a somewhat
longitudinally lined appearance. As in the males,
the subocular pale crescent is obvious, and there
may be a weakly differentiated pale preauricular
area.

This newly collected series of A. ricordi is puz-
zling. The entire lot is so like, in general aspect,
specimens of leberi and viculus (and the yet-to-
be-named subspecies in southwestern Haiti ) and
shows so little tendency toward A. r. ricordi that
it is difficult to interpret them as intergradient be-
tween ricordi and viculus. The adult male (MCZ
66015) noted above in the body of the text is
from "Pemel, near Miragoane," a site that is un-
locatable on modern maps. Pemel may be "near
Miragoane" in only the very broadest sense. Spec-
imens that are known to have been taken in the
Miragoane-Paillant area show little evidence of
intergradation between vicidus and ricordi, and
are much closer to the former subspecies.

Everything considered, I strongly suspect that
with additional collecting on the central Tiburon
Peninsula it is probable that two taxa will be
found to occur here in sympatry and without
wide areas of intergradation, or that ricordi-styled
anoles occur close to ( but do not intergrade with )
leberi-styled anoles. The evidence at the moment
is far from unequivocal that ricordi and viculus
intergrade in this area. Only much additional
collecting along the Tiburon Peninsula will reveal
the actuality of the relationships of A. ricordi with
its southeastern relatixes.

viculus, I have no concepts of their appear-
ance; on geographic grounds, however, I
have little doubt that the Tardieu male is
representatixe of viculus rather than of
leberi.

Remarks. It may seem remarkable that
there should be two distinctive subspecies
of A. ricordi in such close geographical
proximity; Castillon and Marceline are sep-
arated by only 29 kilometers airline, and if
the Tardieu specimen is viculus, then the
distance between the localities for the two
subspecies is even shorter. However, be-
tween Castillon and Marceline lies the high
ridge of the La Hotte, including the cul-
minating peak of that range, Pic Macaya,
with an elevation of 7698 feet (2347 me-
ters). Such high and rugged country is
probably ecologically unsuitable for A. ri-
cordi, and the northern and southern pop-
ulations associated with the La Hotte have
differentiated because of isolation caused
by the intervening massif.

The known altitudinal range of A. r. vic-
ulus is between 2200 and 4000 feet (671
and 1220 meters). The Castillon area, ac-
cording to Richard Thomas, is generally
mesic but much of the original forest has
been cut. Still, enough trees and ravine
woods remain to offer haven for such a
tolerant and adaptable species as A. ricordi.
One female from south of Castillon was
taken by Thomas on the trunk of a large
tree about 5 feet (1.5 meters) above the
ground; all other ASFS specimens were se-
cured by natives.

The name vicidus is from the Latin for
"hamlet" or "small village" in allusion to
Castillon, the type locality.

Anol'is ricordi subsolanus new subspecies

Holotype. MCZ 130270, an adult male,
from Source Carroye, near Saltrou, Depart-
ement de I'Ouest, Haiti, one of a series col-
lected by George Whiteman in March 1972.

Paratypes. MCZ 130264-69, MCZ
130271-77, same data as holotype; MCZ
69405, nr. Saltrou, Dept. de I'Ouest, G.
Whiteman, summer 1962.

Definition. A subspecies of A. ricordi

112 Bulletin Museum of Comparative Zoology, Vol. 146, No. 2

characterized by the combination of mod-
ally 5 snout scales between second canthals,
5 vertical rows of loreal scales, 3 scales be-
tween the supraorbital semicircles, 4/4
scales between the interparietal and the
supraorbital semicircles, moderate number
of vertical dorsal scales (16-21; mean
17.3), moderate number of ventral scales
(18-27; mean 21.0), nuchal crest scales
rarely moderate, usually low in males, low
in females, subocular scales always sepa-
rated from supralabials scales by one row
of scales; males vaguely lineate dorsally
with two broad lateral grayish flank stripes,
or with three paler ( green in life? ) cross-
bands; females like males, or heavily
blotched with black laterally and on the
occiput, the black lateral markings in the
areas that are elsewhere occupied by
the gray lateral flank stripes; a pale suboc-
ular crescent present and prominent but no
pale preauricular blotch; dewlap color un-
known.

Distribution. Known only from the re-
gion about Saltrou, in extreme southeastern
Haiti, but see discussion below.

Description of holotype. An adult male
with a snout-vent length of 144 mm and a
tail length of 209 mm ( regenerated ) ; snout
scales at level of second canthals 4, 6 ver-
tical rows of loreal scales, 3 scales between
supraorbital semicircles, 3/4 scales between
the interparietal and the supraorbital semi-
circles, vertical dorsals 18, horizontal dor-
sals 17, ventrals 19, one row of scales be-
tween suboculars and supralabials, fourth
toe lamellae on phalanges II and III 33,
nuchal crest scales low, body crest scales
low; as preserved, dorsum dull dark brcwn
with three prominent blue-green cross-
bands, more or less confluent middorsally,
and outlined in dark brown to black; top
of head brown, paler than sides; throat
greenish, dewlap dull gray; belly dark
gray, underside of hindlimbs green; tail
brown.

Variation. The holotype and paratypic
series are composed of 10 males and five
females. The largest male (MCZ 130274)

has a snout-vent length of 152, the
largest female (MCZ 69405) 150; the male
is a topotype, the female is from near Sal-
trou. Snout scales at level of the second
canthals range between 4 and 7; the mode
is 5 (six specimens). The vertical loreal
rows vary between 5 and 7, with a mode of
5 (eight specimens). There are between 2
and 4 scales between the supraorbital semi-
circles (mode 3). There are modally 4
scales between the intei-parietal and the
supraorbital semicircles; 4 scales are in-
volved in 58 percent of the combination;
actual counts are 3/3 ( 1), 3/4 (3), 4/4 (4),
4/5 (3), 5/5 (2), and 5/6 (2). Vertical
dorsals range between 16 and 21 (mean
17.3), horizontal dorsals between 14 and 23
(17.1), and ventrals between 18 and 27
(21.0). Of 10 males, two have the nuchal
crest scales moderate and eight have these
scales low; all five females have the nuchal
crest scales low. All specimens have the
body crest scales low. In all specimens the
subocular scales are separated from the su-
pralabials by one row of scales.

I have no color notes in life nor have I
seen live specimens of A. r. suhsolanus.
Consequently, my comments on pattern in
this subspecies are based solely upon pre-
served material. In the series of males and
females, each sex shows two basic patterns.
The more common is a pair of longitudinal
flank stripes, the upper being broader, usu-
ally dull grayish in contrast to a greenish
ground color. In two specimens (one male
and one female; MCZ 130267 and MCZ
69405) these stripes are very prominent
and black; although they no longer have
their integrity in the female, they are still
very obvious. In addition, in the female
there is black pigment in the occipital re-
gion. A pale subocular crescent is present
in all specimens and is usually very con-
spicuous. In two specimens (the holotypic
male and a female— MCZ 130266) the dor-
sal pattern consists of three transverse
crossbands that are green, more or less
fused middorsally, and outlined with black
or dark brown. Many females show the lat-

HisPANioLAN Giant Angle • Schtvartz 113

eral flank stripes much less clearly than do
the males, but usually the stripes are at
least indicated. •

Comparisons. In general aspect, sub-
sola nus much more closely resembles far
western leheri and vicuhis than geographi-
cally closer ricordi. The latter subspecies,
however, occurs on the nortliern side of the
Massif de la Selle, whereas the localities for
suhsolamis are to the south of that range.
Since I do not know the coloration in life
of suhsolamis, I am unable to compare its
pigmentation with that of the other subspe-
cies. The presence of both longitudinally
striped and transversely barred specimens
in suhsolamis suggests its affinity with leheri
and vicuhis. A. r. suhsolamis differs from
A. r. ricordi in that the latter has (in its
southern populations) dark anterior mark-
ings on the occiput and above the forelimb
insertions, whereas these markings are ab-
sent in subsolanus. Additionally, southern
ricordi are patternless green, whereas sub-
solaiius females are longitudinally lined and
may have heavy dark anterior markings
(somewhat like male A. r. ricordi). At the
time of Williams's review of A. ricordi
(1965: 2), there was but a single A. ricordi
from the Saltrou region; by chance, this
specimen (MCZ 69405) is the heavily
marked female upon which I commented
above. Although Williams (loc. cit.) con-
sidered it a male, it lacks enlarged postanal
scales and a tail "fin," and it is a female.
Since female A. r. ricordi lack dark anterior
markings, this female is really quite differ-
ent from females of the northern subspe-
cies.

From the western subspecies leheri and
viculus, suhsolamis difi^ers meristically in
the following ways. From leheri, suhso-
lanus differs in having 5 versus 4 snout
scales at the second canthal, 5 versus 6 ver-
tical loreal rows, higher means in vertical
dorsal scales and ventral scales, and also
lacks specimens that have the suboculars in
contact with the supralabials ( leheri has 48
percent of the specimens with this condi-
tion). From viculus, suhsolanus differs in

having 5 versus 6 snout scales at second
canthal, 5 versus 7 vertical loreal rows, 3
versus 4 scales between the supraorbital
semicircles, and higher means of vertical
dorsal scales and ventral scales. In addition
to the pattern differences noted above
which differentiate suhsolanus from nomi-
nate ricordi, suhsolanus has 5 versus 7 snout
scales at the second canthals, 5 versus 7
vertical loreal rows, 3 versus 4 scales be-
tween the supraorbital semicircles, 4/4 ver-
sus 5/5 scales between the interparietal and
the supraorbital semicircles, and lower
means in vertical dorsal scales and ventral
scales.

Remarks. I am once more hampered in
my interpretation of suhsolanus by the
large distributional gap between its two
stations and any other stations for A. ricordi
to the west. The absence of specimens
from the southern coast, from such well-
known areas as Jacmel and Aquin, is truly
puzzling. The nearest locality to suhsolanus
along the Tiburon Peninsula is Fond des
Negres [ricordi X viculus), some 120 kilo-
meters to the west. Still further, the area
known to be occupied by A. r. leheri lies
some 205 kilometers to the west, near the
peninsula's tip. Known stations for A. r. ri-
cordi are very much closer (40 kilometers)
but lie to the north of the Massif de la
Selle. Closer even than' any of these is
harahonae; harahonae and suhsolanus are
known in this region for localities separated
by about 11 kilometers (see comments be-
low), but there is no evidence of intergra-
dation between these two taxa.

It is perhaps pertinent that leheri, vicu-
lus, and suhsolanus all seem closer in most
characteristics to each other than they do
to nominate ricordi. If it were not for the
specimens that I interpret as intergradient
between ricordi and viculus in the Mira-
goane-Paillant-Fond des Negres region, I
would be very tempted to consider these
three taxa as a species distinct from A. ri-
cordi. Much additional material from along
the Tiburon Peninsula will perhaps show
that my interpretation is wrong.

114 Bulletin Museum of Comparative Zoology, Vol. 146, No. 2

The name subsolamis is from the Latin
for "eastern," in alhision to the occurrence
of this subspecies in southeastern Haiti.
The precise areas where suhsolanus occurs
are a matter of question. I am unable to
locate Source Carroye on any modern map.
Williams advised me that Source Carroye
is very near Thiotte (according to the col-
lector, "Source Carroye is located northeast
direction and about V2 mile from the main
road after you leave the place of the
'marche'," that market being at Thiotte).
The elevation of Thiotte is about 900 me-
ters. The lone specimen from "near Sal-
trou" also poses the problem of just how
"near" this specimen was taken to Saltrou
itself. Any information on details of local-
ities or elevations of this and other speci-
mens taken along the Dominico-Haitian
border are mandatory. The distance be-
tween the Thiotte locality for suhsolanus
and the Pedernales specimens of harahonae
is about 11 kilometers. It is especially per-
tinent that harahonae is not known, along
the Dominico-Haitian border, from the low-
lands (where, incidentally, Anolis coeles-
tinus is called saltacocote by the natives),
but that harahonae occurs here as an in-
habitant of mesic riverine woods at an ele-
vation of 600 feet ( 183 meters ) .

Anolis barahonoe Williams

Anolis ricordii harahonae Williams, 1962. Brevi-
ora, Mus. Comp. Zool., No. 155: 8.

Ttjpe locality. Polo, Valle de Polo, Bara-
hona Province, Repiiblica Dominicana;
holotype, MCZ 43819.

Defiyiition. A giant species of Hispanio-
lan Anolis characterized by the combina-
tion of moderate size (males to 158 mm,
females to 148 mm snout-vent length),
snout scales at level of second canthal
scales 2 to 5 (mode 4), vertical loreal rows
2 to 5 (mode 6), scales between supraor-
bital semicircles 1 to 4 (mode 2), inteipa-
rietal scale separated from supraorbital
semicircles modally by 4 scales, vertical
dorsal scales generally small ( 15 to 34 in
standard-distance), ventral scales relatively
small (17 to 29 in standard-distance), nu-

chal crest scales in both sexes rarely high,
usually moderate to low, dorsal body crest
scales rarely moderate, usually low, suboc-
ular scales rarely in contact with suprala-
bial scales; dorsal body coloration basically
lichenate gray-green, grays, to browns and
black, giving a blotched effect that also oc-
curs in even the smallest juveniles, and
rarely (only in juveniles) with any indica-
tion of transverse crossbars, or solid brown
to grayish with faintly bluish white dark-
edged ocelli; dewlap pale yellow to peach
in males, pale yellow to pale peach in fe-
males; pale subocular crescent absent in
adults but indicated in juveniles by a pale
subocular spot.

Distrihution. The Sierra de Baoruco
and associated lowlands on the Peninsula
de Barahona, Repiiblica Dominicana, in-
cluding (probably) the semi-xeric forests
of the lowlands south of the Sierra de
Baoruco and southern Haiti; altitudinal dis-
tribution from sea level to 2600 feet (793
meters) northeast of Las Auyamas, Bara-
hona Province.

Anolis harahonae harahonae Williams
Type locality. Polo, Valle de Polo, Bara-
hona Province, Repiiblica Dominicana.

Definition. A subspecies of A. hara-
honae characterized by the combination of
modally 4 snout scales between second can-
thai scales, 4 vertical rows of loreal scales,
2 scales between the supraorbital semicir-
cles, 4/4 scales between the interparietal
and the supraorbital semicircles, relatively
low number of vertical dorsal scales ( 15-
23; mean 17.2), high number of ventral
scales (17-29; mean 22.1), nuchal crest
scales moderate to low, body crest scales
rarely moderate, usually low, subocular
scales usually separated from supralabial
scales by one row of scales, both sexes and
juveniles patterned with varying shades of
gray-green, grays, browns and black, giv-
ing a lichenate blotched effect; juveniles
with vague indications of three transverse
gray bands but that pattern only very
rarely even indicated in adults; dewlap
pale yellow to pale peach in both sexes, the

HisPANiOLAN Giant Angle • Schtvartz 115

female dewlap suffused with gray basally;
pale subocular crescent absent in adults but
indicated by a clear white subocular spot in
juveniles and subadults.

Discussion. A. b. harahonae has a rela-
tively circumscribed range in the Sierra de
Baoruco in the southeastern Republica Do-
minicana. Until our 1971 collections, the
taxon had been known only from the east-
ern portion of that massif, but two speci-
mens taken 13.0 mi. ( 20.8 km ) N of Peder-
nales along the Dominico-Haitian border
are unquestionably A. harahonae. These
individuals differ slightly from more east-
ern specimens of A. h. harahonae in colora-
tion, but they are so close to the nominate
subspecies that for the moment I have no
hesitancy in regarding them as that taxon.

The series of 33 specimens of A. h. hara-
honae shows the following variation. The
largest males (ASFS V29722, MCZ 125504)
have snout-vent lengths of 158, the largest
female (AMNH 50256) 148; the males are
from north of Pedernales and near Polo,
and the female is from Barahona. Snout
scales at level of second canthals vary be-
tween 2 and 5; the mode is 4 (18 speci-
mens). The vertical loreal rows vary be-
tween 5 and 8, with a mode of 6 (11
specimens). There are between 1 and 4
scales between the supraorbital semicir-
cles (mode 2). There are modally 4 scales
between the inteiparietal and the supraor-
bital semicircles; 4 scales are involved with
58 percent of the combinations; actual
counts are 3/3 (3), 3/4 (6), 4/4 ( 13), 4/5
(3), and 5/5 (5). Vertical dorsals range
between 15 and 23 (mean 17.2), horizontal
dorsals between 15 and 24 (18.2), and ven-
trals between 17 and 19 (22.1). Of 16
males, seven have the nuchal crest scales
moderate and nine have these scales low;
of 10 females, three have these scales mod-
erate and seven have them low. Body crest
scales are moderate in one male and low in
15, whereas all 10 females have the body
crest scales low. The subocular scales are
separated from the supralabial scales in 32
of 33 specimens (3 percent).

Exclusive of the male and female from

north of Pedernales, eastern specimens of
A. /;. harahonae are lichenate or blotclied
with gray-green, grays, browns, and black
in a random pattern, although occasional
individuals show remnants of the slightly
more obviously banded condition of the
juveniles. No specimen has been recorded
in the field as being bright green, and in
general the tones of green in harahonae
are dull and grayish. Some specimens (es-
pecially ASFS V30921, a male) were re-
corded as being gray, heavily blotched with
black, and thus without any green tints
whatsoever. The dewlap color in males
varies between pale peach and peach, and
in females between pale peach and yellow.

The Pedernales specimens were recorded
in life as being dark brown to gray dorsally,
obscurely banded with tannish. The heads
were tan above, the eyeskin pale gray, and
the female had the upper surfaces of all
limbs banded green and dark brown. The
most noteworthy difference between these
western specimens and those from the east-
ern uplands of the Sierra de Baoruco and
its associated lowlands is that the dewlaps
in both sexes were pale yellow, that of the
female suffused with gray basally.

Available juveniles and subadults vary in
length between 62 and 95. The juveniles
are colored and patterned essentially like
the adults, except that three pale grayish
crossbands are vaguely indicated in most
specimens. These bands are quite indis-
tinct and much obscured by the lichenate
harahonae pattern. Some juveniles were re-
corded as being banded and mottled pale
gray, dull pea-green, and black, witli a
black nuchal patch and a white nuchal
crescent on each side, whereas others were
recorded as crossbanded gray and dusky,
with some greenish on the lips, and the tails
banded gray and dusky to cream.

Remarks. Specimens secured by myself
and parties have all been taken in wooded
situations, between elevations of 600 feet
and 2600 feet ( 183 and 793 meters). River-
ine woods and the large shade trees in the
upland cafetales of the Sierra de Baoruco
offer optimum habitat for the species. Both

116 Bulletin Museum of Comparative Zoology, Vol. 146, No. 2

adults and juveniles were secured sleeping
at night; in general, the juveniles sleep
lower on shrubs and low trees, whereas
adults sleep higher (up to 15 feet — 4.6 me-
ters) on limbs, branches, and woody vines.
At night, despite the absence of bright
greens in the coloration, the lizards are
quite obvious because their pale grayish
hues contrast to the adjacent greenery. All
ages of A. h. harahonae sleep exposed, as
do other Hispaniolan giant anoles. The
pair from 13.0 mi. N Pedernales were se-
cured in rich riverine woods at an eleva-
tion of 600 feet ( 183 meters ) ; this is purely
a gallery forest situation, since in this re-
gion the open slopes are clad in Acacia for-
est or dry scrubby woodlands, whereas
rivers and creeks support much more luxu-
riant arboreal growth.

Almost all localities for A. h. harahonae
are in the highlands. However, the lizard
presumably occurs in coastal forested re-
gions as well. There are specimens from
the city of Barahona ( which is coastal ) and
from halfway between Enriquillo and
Oviedo, which is presumed to be coastal or
nearly so. A third specimen from Enri-
quillo likewise is presumably from a coastal
locality. However, in each of these cases,
it is possible that the lizards were secured
in the adjacent Sierra de Baoruco; this
mountain range comes abruptly to the coast
between Barahona and Enriquillo, and it
would be a simple matter to label speci-
mens from non-coastal localities as having
come from coastal populated areas. Al-
though negative evidence at best, we have
never oiuselves secured A. harahonae along
this coastal region, and residents of Bara-
hona responded negatively when ap-
proached to collect this lizard for us.

A. /;. harahonae is known from a locality
(13.0 mi. N Pedernales) that is only (pre-
sumably) 11 kilometers from a locality
(Thiotte) where A. r. suhsolanus occurs.
There are no other localities where these
two species approach each other, although,
since the northern slopes of the Sierra de
Baoruco are confluent with the northern
slopes of the Massif de la Selle and its affil-

iates, it is not unlikely that somewhere
along these northern reaches A. harahonae
comes into contact with A. r. ricordi. There
is no obvious reason for A. h. harahonae to
be promptly replaced by A. r. suhsolanus
at the Dominico-Haitian border; the polit-
ical boundary on these southern slopes is
the Rio Pedernales, a small stream that
surely offers no obstacle for these arboreal
lizards. It follows that A. h. harahonae
must occur in southeastern Haiti. Thus, as
previously noted, the accuracy of the suh-
solanus localities is more than academic. It
is possible that in southeastern Haiti, A.
harahonae is a more lowland lizard and A.
ricordi (suhsolanus) occurs on the higher
and better forested slopes of the Massif de
la Selle — the division may thus be altitu-
dinal as well as ecological. The precise re-
lationships between these two species re-
main to be determined; only further
detailed collecting in extreme southeastern
Haiti will reveal the siutation there. As far
as distinguishing A. r. suhsolanus from A.
h. harahonae, there is no problem, since the
styles of pattern (and presumably colora-
tion) are so very different as to preclude
confusion. If intergradation between suh-
solanus and harahomie occurs (and since I
here regard harahonae as a species distinct
from ricordi, I am obviously convinced that
it does not), then it must take place very
quickly, in a distance of some 11 kilome-
ters, since suhsolanus and the Pedernales
harahonae are completely different and
typical of their own populations, without
any indication of intergradation between
them.

Specimens examined. REPUBLIC A DO-
MINICAN A: Barahona Province, Barahona
(AMNH 50255-56); 14 km SW Barahona,
1200 feet (366 meters) (ASPS V23460-63,
ASFS V30263-70); Valle de Polo (MCZ
56141, AMNH 51235-37, AMNH 51240,
AMNH 51036); nr. Polo (MCZ 125504-06);
Las Auyamas (ASFS V30921); 8 km NE
Las Auyamas, 2600 feet (793 meters)
(ASFS X9676); Hermann's finca, nr. Par-
aiso (AMNH 51231-33); Enriquillo
(AMNH 51241); Pedernales Province, half-

HisPANiOLAN Giant Angle • Schwartz

117

way between Enriquillo and Oviedo
(AMNH 51230); 13.0 mi. (20.8 km) N
Pedernales, 600 feet ( 1S3 meters ) ( ASFS
V29722-23); locality unkno\\ai (AMNH
51229).

Anolis borahonoe olbocellotus

new subspecies

Holotype. MCZ 125611, an adult male,
from 13.1 mi. (21.0 km) SW Enriquillo,
Pedernales Province, Republiea Domini-
cana, taken by Richard Thomas on 10 De-
cember 1964. Original number ASFS
V4422.

Definition. A sub.species of A. hara-
honae characterized by the combination of
4 snout scales between second canthal
scales, 7 vertical rows of loreal scales, 3
scales between the supraorbital semicir-
cles, 4/4 scales between the interparietal
and the supraorbital semicircles, apparently
relatively high number of vertical dorsal
scales (19), high number of ventral scales
(26), nuchal crest scales high, body crest
scales low, subocular scales separated from
supralabial scales by one row of scales,
male (females unknown) dorsal ground
color nonlichenate brown to grayish with
white (faintly bluish) randomly placed
dark-edged ocelli, head light brown above,
dewlap pale yellow with a pink margin,
and a pale subocular spot.

DistriJnition. Known only from the type
locality, but presumably distributed
through the semi-arid forests of the Penin-
sula de Barahona south of the Sierra de
Baoruco (see discussion).

Description of holotype. An adult male
with a snout-vent length of 150 and a tail
length of 265; snout scales at level of sec-
ond canthal 4, 7 vertical rows of loreal
scales, 3 scales between interparietal and
supraorbital semicircles, vertical dorsals 19,
horizontal dorsals 23, ventrals 23, one row
of scales between suboculars and suprala-
bials, fourth toe lamellae on phalanges II
and III 34, nuchal crest scales high, body
crest scales low; in life, dorsum brown to
grayish, not lichenate, with randomly scat-
tered white (faintly bluish) dark-edged

ocelli involving from 1 to 4 scales; venter
white with gray mottling or stippling; dew-
lap pale yellow with pink along its outer
margin; upper surface of head light brown,
with large pale subcircular areas anterior
to the ear opening, and a conspicuous pale
blotch ])el()w the eye; soles of hands and
feet conspicuously pale yellow.

Comparisons. No mensural nor meris-
tic characters separate aU)ocellatns from
barahoiuie. On the other hand, the distinc-
tive coloration, pattern, and dc^wlap color
of aJhoceUatus are very different from those
of harahonae, and the presence of high nu-
chal crest scales likewise differentiates al-
hocellatus from the moderate to low scales
in Imrahonae. More detailed comparisons
are impossible, but certainly aIJ)OceIl(itus is
quite distinctive when compared with ])ara-
honue.

Discussion. It may seem foolhardy to
name a subspecies of A. harahonae from a
single specimen whose locality is only 11
kilometers from a presumed locality for A.
h. harahonae (half way between Enriquillo
and Oviedo). The holotype of A. h. alho-
cellatus is that lizard about which Williams
(1965: 4) commented, saying it "is typical
in squamation but peculiar in having very

distinct small Ufi,ht spots on the flanks

It will be recalled that it was a specimen
from Enriquillo (AMNH 51241) that
caused some hesitation when harahonae
was first described. In AMNH 51241 the
pattern was thought to be obscure banding;
the present specimen clearly shows spots
tending to be vertically aligned — a condi-
tion which is easily transformed into verti-
cal banding. It is possible that the ricordii
populations in the vicinity of Enriquillo
consistently show a distinctive pattern
though charact(>ristically harahoiuie in
squamation."

The specimen (AMNH 51230) from half-
way between Enriquillo and Oviedo is a
young male with a snout-vent length of
121. Since this lizard presumably came
from the lowlands of the Peninsula de
Barahona, it might logically be expected to
be alhocellatiis. However, the lizard is now

118 Bulletin Museum of Comparative Zoology, Vol. 146, No. 2

drab patternless brown, and there are no
indications that it was ever spotted. Pre-
sumably albocellatus and harahonae inter-
grade between Enriqiiillo (which Hes at
the extreme southeastern corner of the Si-
erra de Baoruco) and Oviedo (which Hes
well down on the Peninsula de Barahona).
Several facts have prompted my naming
this lonely specimen. First, I have exam-
ined the Enriquillo specimen noted by Wil-
liams, and, although it shows some indica-
tion of vertical crossbars, they are not any
more conspicuous than those in some more
recently taken A. /;. harahoruie from the
Baoruco highlands (Williams examined
only 17 harahoime at the time of its original
description; I have studied almost twice
this number). Secondly, the xeric to semi-
arid region south of the Sierra de Baoruco
has come to be known as an area of local
differentiation at the subspecific level for a
variety of reptiles; this alone is no reason
for naming albocellatus, of course. Thirdly,
although since 1964 when the holotype was
collected both I and others have spent con-
siderable time on the Peninsula de Bara-
hona and in the vicinity of Oviedo, we have
never seen or secured another A. harahonae
in this region. In September 1966, the very
severe hurricane Inez passed directly across
the Peninsula. What had once been high-
canopied semi-arid forest (as at Oviedo)
has been either totally destroyed or been
reduced (by 1969) to a landscape of bare
snags with some leafy growth just now be-
ginning to appear but at a much lower
canopy-level than previously. The changes
between the Oviedo area in 1964 and 1969
are so massive that, upon my first visit there
after Inez, I was unable to orient myself in
reference to our older collecting localities!
Certainly this entire region has suffered
greatly, and, with the destruction of trees,
it seems reasonable to assume that A. hara-
honae has suffered equally. The population
may never have been high, since such semi-
arid woods are not at all optimal habitat
for any of the Hispaniolan giant anoles, and
the destruction of the habitat must surely

have affected A. h. albocellatus adversely.
Since persistent visits to this area have
yielded no new material, and since the liz-
ard may presently be very rare, I have de-
cided upon the present course rather than
wait in hope for someone to secure a sec-
ond ( or more ) lizard.

Remarks. The Peninsula de Barahona
has been shown to have distinctive subspe-
cies (or even species) of a variety of rep-
tiles. Species that have described endemic
subspecies south of the Sierra de Baoruco
include: Sphaerodacttjhis difficilis Barbour,
Leiocephalus Imrahonensis Schmidt, Am-
eiva chrysolaema Cope, Ameiva lineolata
Dumeril and Bibron, Arnphisbaena gona-
vensis Cans and Alexander, and Dromicus
parvifrons Jan. Endemic Peninsula de
Barahona species are: Anolis longitibialis
Noble, Typhlops sijntherus Thomas, Lep-
totyphlops pyrites Thomas, and Uromacer
ivetmorei Cochran. Only one amphibian,
Eleutherodactylus alcoae Schwartz, is re-
stricted to the Peninsula. To the former
list can now be added Anolis harahonae.
The eastern half of the Peninsula, although
xeric, was originally clothed in dry forest,
much of it upon a series of limestone ter-
races, the highest point of which is the
Loma Gran Sabana, having an elevation of
1082 meters in the north and descending to
Cerro Caballo, and Loma de Chendo, hav-
ing elevations of 322 and 233 meters, re-
spectively, to the south. West of this ridge,
the land descends abruptly to Acacia-cac-
tus desert to the east of Cabo Rojo, and
this habitat continues to the Dominico-
Haitian border at Pedernales. Presumably,
A. b. albocellatus occurs throughout the
eastern half of the Peninsula in the for-
merly high-canopied forests of the lime-
stone terraces.

The holotype was secured by Richard
Thomas during the day in a viny tangle in
semi-xeric woods near Oviedo; the lizard
was in an edge situation, since beyond the
dense vine tangle the woods thinned to
more scrubby and cleared areas.

The name albocellatus is from the Latin

HisPANioLAN Giant Anole • Schivartz

119

"albus" for "white" and "ocellus" for "eye,"
in allusion to the white spots that are typi-
cal of the holotype.

Anolis baleatus Cope

Eti])ristis baleatus Cope, 1864, Pvoc. Acad. Nat.
Sci. Philadelphia, p. 168.

Type locality. Santo Domingo; holo-
type, British Museum (Natural History)
1946.8.29.22.

Definition. A giant species of Hispanio-
lan Anolis characterized by the combina-
tion of large size (males to 1<S0 mm,
females to 148 mm snout-vent length),
snout scales at level of second canthal
scales 2 to 5 (modally 2 or 4, by popula-
tion) but usually 2 or 3 (75 percent), ver-
tical loreal rows 5 to 10 ( modes by popula-
tion 6, 7 or 8), scales between supraorbital
semicircles 1 to 4 (modally 3), interpari-
etal scales separated from supraorbital
semicircles modally by 4 or 5 scales, verti-
cal dorsal scales generally small ( 12 to 24
in standard-distance), ventral scales rela-
tively small ( 15 to 34 in standard-distance ) ,
nuchal crest scales in both sexes very high
to high, rarely moderate, never low, body
crest scales usually high to moderate, rarely
low, subocular scales usually not in contact
with supralabial scales; dorsal body colora-
tion and pattern usually some shade of
green, varying from dull greenish brown to
bright emerald green, either conspicuously
crossbanded with few (3 or 4) to very
many crossbands, in the latter condition the
lizards appearing tigroid, or, on the other
hand, without crossbanding but blotched,
never stiiped or with dark occipital, nu-
chal, or lateral dark markings, dewlap in
males from pale yellow to vivid orange, in
females from brownish or very pale yellow
to orange or gray, often suffused with gray-
ish or brownish, or nearly white, chin and
throat yellowish, green, or orange, often
with a dark dotted or mottled or reticulate
pattern, and pale subocular crescent absent
in adults.

Distribution. The eastern two-thirds of
the Repiiblica Dominicana, from Puerto

Plata, Santiago, and La Vega provinces
south to San Cristobal Province and the
Distrito Nacional, and east to La Altagracia
Province; also in and near the Sierra Mar-
tin Garcia and the southern slopes of the
Cordillera Central and the Sierra de Ocoa
in Azua and Peravia provinces; occurs on
Isla Saona but unrepresented by specimens
from that satellite island.

Anolis boleotus baleatus Cope

Type locality. "Santo Domingo"; here
restricted to the vicinity of Puerto Plata,
Puerto Plata Province, Repiiblica Domini-
cana (see rationale for this restriction be-
low ) .

Definition. A subspecies of A. baleatus
characterized by the combination of mod-
ally 4 snout scales between second canthal
scales, 7 vertical rows of loreal scales, 3
scales between the supraorbital semicircles,
moderate number of vertical dorsal scales
(14-21; mean 17.5), high number of ven-
tral scales (19-34; mean 23.8), nuchal crest
scales very high (usually) to high or mod-
erate (rarely), body crest scales high
(rarely) to moderate (usually), subocular
scales always in contact with supralabial
scales, males from pale green or rich bluish
green to brown dorsally, with three bright
yellow to darker green or greenish brown
irregular crossbands, lower sides usually
bright yellow, females apparently with the
same body patterns and hues as the males
(see below), throat in males bright yellow
to bright orange, rarely mottled with
brown, dewlap in males always vivid to
brilliant orange, and the upper surfaces of
hindlimbs bluish green, conspicuously
barred with bright yellow.

Distribution. Known from the Cordil-
lera Septentrional and the northern coastal
plain of the Repiiblica Dominicana, from
Puerto Plata, Espaillat, and Santiago prov-
inces, but probably occurring elsewhere in
this range and to the north of it; specimens
from Los Bracitos, Duarte Province, should
also be included ( on geographical grounds )
with A. b. baleatus, since Los Bracitos lies

120 Bulletin Museum of Comparative Zoology, Vol. 146, No. 2

in the eastern extremity of the Cordillera locality of the name to the vicinity of a ma-
Septentrional, but the specimens are old jor city that is presumed to lie within the
and greatly discolored and I have not con- area to which I ascribe this boldly cross-
sidered them as pertaining to the nominate banded subspecies,
subspecies. The series of 15 A. b. haleatus shows the

Discussion. Eupristis haleatus Cope was following variation. The largest males
named from a single specimen from "Santo (ASFS V33558, ASFS V18123) have snout-
Domingo." I have examined the holotype, vent lengths of 148, and the largest fe-
collected by A. Salle, in the British Museum male ( MCZ 128380) has the same dimen-
( Natural History ) . Considering its length sion. These three lizards are all from the
of time in preservative, it is in excellent Cordillera Septentrional north of Puesto
condition and shows a striking pattern of Grande. Snout scales at level of the second
three bold pale body crossbands on a canthal vary between 2 and 4; the mode
darker dorsal ground color, contrastingly is 4 (eight specimens). The vertical loreal
banded hindlimbs and tail, and immaculate rows vary between 5 and 9, with a mode
throat. The specimen is a female, and, un- of 7 (six specimens). There are 3 scales
fortunately, I have only two adult females between the supraorbital semicircles in all
from the range ascribed above to A. h. ha- specimens. There are modally 5 scales be-
leatiis: both are without color data in life, tween the interparietal and the semicircles;
At least one of them (MCZ 57717) resem- 5 scales are involved in 63 percent of the
bles the pattern of the haleatus holotype combination; actual counts are 4/4 (1), 4/5
to a striking degree. (4), 5/5 (7), 5/6 (1), 6/6 (1), and 5/7

Through the courtesy of Ernest E. Wil- (1). Vertical dorsals range between 14 and
liams, I have a copy of a map prepared by 21 ( mean 17.5 ) , horizontal dorsals between
William J. Clench which shows the locali- 16 and 26 (19.7), and ventrals between 19
ties where A. Salle is known to have col- and 34 (mean 23.8). Of nine adult males,
lected. Considering the era of his travels six have the nuchal crest scales very high,
(the mid-1800's), Salle traveled widely two have these scales high, and one has
throughout the Repiiblica Dominicana, them moderate. Of three females, the nu-
from (in the north) Puerto Plata, Ponton, chal crest scales are very high in two and
Santiago, Moca, La Vega and Cotui, east high in one. The body crest scales are high
to Higiiey, Cabo Engafio and San Rafael in one male and moderate in eight males;
del Yuma, in the eastern interior to Hato in three females, the body crest scales are
Mayor and El Seibo, along the southern high in one and moderate in two. All spec-
coast from Santo Domingo to San Cristo- imens have the subocular scales in contact
bal, Bani, Azua, Barreras, and Barahona, with the supralabial scales,
and into the Valle de San Juan to the city Males are usually conspicuously cross-
of San Juan. He also ascended the south- banded. Specimens have been recorded as
ern slopes of the Cordillera Central near pale green with three irregular darker
San Jose de Ocoa. Although much of Sal- green crossbands, brown with three faint
le's Dominican travels was in territory of green-brown crossbands, or rich bluish
A. haleatus, he was also in the ranges of green with three bright yellow crossbands.
A. ricorcli and A. harahonae. The holotype. The lower sides are bright yellow (which
as V^illiams (1962: 2, footnote 1) pointed grades into a grayish venter), and this color
out, has elongate nuchal crest scales, and also occurs on the throat, which varies
there is no doubt that the name haleatus from yellowish to bright yellow or orange,
is applicable to some population that pos- occasionally mottled with brown. The dew-
sesses this character. Since Salle traveled lap is brightly colored; it has been recorded
within the range of the northern population as "vivid orange," "bright vivid orange,"
of A. healeatus, I have restricted the type "brilliant yellow-orange," and "very bright

HisPANiOLAN Giant Angle • Schtvmiz 121

orange." The upper siirfaee of the head is
reddish brown and the hindhnibs are green
to bhiish green,' barred with bright yellow.
In general, male A. b. haleatus are vividly
patterned and eolored lizards. I have eol-
lected no females myself and thus have no
notes on this sex from life; however, one
recently (1971) collected female ( MCZ
128380) still is dark green with several thin
vertical pale crossband remnants on the
sides and back, and another female ( MCZ
57717) is contrastingly patterned in dark
and pale green, the latter occurring as ver-
tical crossbands.

The series includes three subadults, with
snout-vent lengths between 73 and 83.
One of these (ASFS V33559; snout-vent
length 80) was medium brown dorsally
with a black postocular streak and an or-
range dewlap that was streaked with black
basally. None of the subadults as pre-
served shows any crossbanding or other
pattern elements. It is interesting that the
only Hispaniolan giant anole taken at night
sleeping in the brown phase is the above
mentioned subadult.

Remarks. All ASFS specimens collected
by myself and parties were secured at
night while the lizards were sleeping.
Typical situations are in gallery forest and
cafetales along mountain streams in the
Cordillera Septentrional. Favored sleep-
ing sites for these lizards in the region are
pendant and semi-pendant woody vines;
Fowler reported that one adult male se-
cured by him at night was not asleep and
was slowly ascending a tree trunk as
Fowler approached. It is possible that this
lizard had been disturbed by the bright
light from Fowler's flashlight or by unfa-
miliar movements and noises, since I doubt
that any of the Hispaniolan giant anoles
are normally active at night. However, all
these lizards waken quickly when dis-
turbed and unless promptly secured, grad-
ually wander away into the greenery and
are lost to view. One of the juveniles was
secured only 6 feet (1.8 meters) above the
ground, whereas one of the adults was
shot from a tree limb 35 feet ( 10.7 meters )

above a mountain stream. The specimen
from near Sosi'ia was taken in dense hard-
woods on a limestone substrate.

The altitudinal distribution of A. 1). ha-
leatus is from 1400 to 2200 feet ( 427 to 671
meters), but the ta.xon occurs much lower
than this, since the specimen from near
Sosua was in limestone hills near sea level.

Specimens examined. REPUBLICA DO-
MINICANA: Espaillat Province, 2km N Pu-
esto Grande, 1400 to 2200 feet (427 to 671
meters) (ASFS V18048, ASFS V33557-
59); 5 km N Puesto Grande (MCZ
128380); 11 km N Puesto Grande, 2100
feet (641 meters) (ASFS V18123, ASFS
V18292): Puerto Plata Province, 11 km SE
Sosua (ASFS V1717); Santiago Province,
Pena (MCZ 57713, MCZ 57715-19); no lo-
cality other than Santo Domingo — British
Museum (Natural History) 1946.9.28.22—
holotype of Eupristis haleatus.

Anolis haleatus multistruppus
new subspecies

Holotype. USNM 193975, an adult
male, from Guaigiii, 3 mi. (4.8 km) S La
Vega, La Vega Province, 300 feet (92 me-
ters), Repiiblica Dominicana, one of a se-
ries taken by Danny C. Fowler, Albert
Schwartz, and Bruce R. Sheplan on 9 No-
vember 1971. Original number ASFS
V33680.

Paratopes. ASFS V33681-86, MCZ
125612-15, CM 54107-12, same data as ho-
lotype; ASFS V18547-50, same locality as
holotype, J. R. Dennis, J. A. Rodgers, Jr.,
and A. Schwartz, 27 July 1969.

Definition. A subspecies of A. haleatus
characterized by the combination of mod-
ally 2 snout scales between second canthal
scales, 7 vertical rows of loreal scales, 3
scales between the supraorbital semicircles,
4/4 scales between the interparietal and the
supraorbital semicircles, high number of
vertical dorsal scales (14-24; mean 18.6),
moderate number of ventral scales ( 18-29;
mean 22.3), nuchal crest scales very high
to high (usually) to moderate (rarely),
body crest scales high (rarely) to moder-
ate (usually), subocular scales almost al-

122 Bulletin Museum of Comparative Zoology, Vol. 146, No. 2

ways separated from supralabial scales by
one row of scales, both sexes as adults re-
taining the complex juvenile pattern of
many fine dark green, green, and yellow
vertical bars, occasionally (in females)
bright pea green with three pale green
crossbars more prominent than any other
dorsal pattern elements, throat green to
yellow green, dewlap in males very pale
yellow to very pale peach, suffused basally
with pale gray, in females very pale yellow
to pale yellow, strongly suffused with pale
gray to entirely pale gray.

Distribution. Known only from the type
locality but presumed to occur on the
northern and probably eastern lower faces
of the Cordillera Central in proper habi-
tats; possibly extending as far west on the
northern face of this range as the Rio Bao
near Los Montones (see discussion below).

Description of holotype. An adult male
with a snout-vent length of 146 and a
tail length (broken) of 97; snout scales at
level of second canthals 2, 9 vertical rows
of loreal scales, 3 scales between the su-
praorbital semicircles, 4/4 scales between
the interparietal and the supraorbital semi-
circles, vertical dorsals 21, horizontal dor-
sals 25, ventrals 29, one row of scales be-
tween the suboculars and supralabials,
fourth toe lamellae on phalanges II and III
30, nuchal crest scales high, body crest
scales moderate; in life, dorsal body pat-
tern of many fine green, dark green, and
yellow crossbands, upper surface of head
grayish tan in contrast to the brighter dor-
sal colors, chin and throat very pale yellow
or yellow-green with no clearly delineated
darker green markings, and dewlap very
pale yellow, much suffused basally with
gray.

Variation. The series of 21 A. h. multi-
struppus is composed of eight males and 13
females. The largest male has a snout-vent
length of 146 and is the holotype. The
largest female (ASFS V33684) has a
snout-vent length of 136 and is a topo-
type. Snout scales at the level of the sec-
ond canthals range between 2 and 5; the
mode is 2 (15 specimens). The vertical

loreal rows vary between 6 and 9, with a
mode of 7 (nine specimens). There are 2
or 3 scales between the supraorbital semi-
circles (mode 3). There are modally 4
scales between the interparietal and the
supraorbital semicircles; 4 scales are in-
volved in 58 percent of the combinations;
actual counts are 3/3 (1), 3/4 (2), 4/4
(8), 4/5 (5), 5/5 (3), and 5/6 (1). Ver-
tical dorsals range between 14 and 24
(mean 18.6), horizontal dorsals between
17 and 25 (20.6), and ventrals between 18
and 29 (22.3). All three adult males have
the nuchal crest scales high, and of ten
females, three have these scales very high,
five have them high, and two have them
moderate. All three males have the body
crest scales moderate, whereas two females
have the body crest scales high, eight have
them moderate, and one has them low. In
all but one specimen (5 percent), the sub-
oculars are separated from the supralabials
by 1 scale.

Adults of both sexes retain the juvenile
multibanded pattern of dark greens, me-
dium greens, and yellow. One adult fe-
male was recorded as bright pea-green
with three pale green crossbands, which
are remnants of the hollowed yellow cen-
ters of the five or six dark brown to dark
green crossbands. In general, the total as-
pect of adults and juveniles is of a con-
trastingly tigroid lizard, the stripes varying
shades of greens, yellows, and ( in the dark
phase) browns. The upper surface of the
head is grayish tan in males and tannish
green in females, and the throat is un-
marked green to yellow-green. One of the
most striking features of A. b. multistrup-
pus is the faded dewlap coloration. In
males, the colors vary between very pale
yellow and very pale peach, basally suf-
fused with pale gray. In females, the dew-
lap is even more drab, with pale yellow
the basic color, but the gray suffusion may
be so extensive as to limit the yellow pig-
ment to the dewlap edge or to cause the
dewlap to be pale gray.

The type series includes seven juveniles
and subadults, with snout-vent lengths be-

HisPANioLAN Giant Angle • Schwartz

123

twecn 47 and 99. These prcstMit a imi-
fonii aspect of multiple dorsal bands as
described above, and even tlie largest of
the subadnlts clearly shows this condition.
In life, a small juvenile (snout-vent length
53) was recorded as pale gray with a
yellow-green head and about four reversed
chevrons between the neck and the hind-
limbs, these chevrons being the pale hol-
lowed remnants of the darker crossbands,
wliich, in this individual, arc obscure. The
small lizard also had a black postocular
line and a charcoal postangular smudge.
The juvenile and subadult dewlaps are
pale flesh to very pale yellow, somewhat
suffused basally with light to very dark
gray.

Comparisom. Meristically, inulfistrup-
pus differs from nominate haleatus in hav-
ing 2 (rather than 4) snout scales at the
level of the second canthals, 4/4 (rather
than 5/5) scales between the interparietal
and the supraorbital semicircles, and in
having slightly less ventrals (means 22.3
and 23.8). There is also a strong tendency
for both sexes of haleatus to have very
high nuchal crest scales, whereas these
scales are more often only high in midti-
struppus. It is in color and pattern that
these two subspecies differ most strikingly.
In the introduction to the present paper I
commented on my having collected speci-
mens from the Cordillera Septentrional and
Guaigiii on two succeeding days, and on
the color and pattern differences being at
once very apparent. The bright orange
throat and dewlap of haleatus contrast
quite obviously with the pale yellow to
gray dewlaps in multistruppus. The body
patterns of the two subspecies likewise are
quite different, with the finely and multi-
banded multistruppus in contrast to the ir-
regularly banded dorsum with only three
bands in ])aleatus.

Discussion. A. h. multistruppus is
known with certainty from only a single
locality, which lies at the foot of the Cor-
dillera Central at an elevation of 300 feet
(92 meters). The locality is unique in that
it represents an extensive stand of original

lowland forest in this region, hardwood
forest which abuts upon the lower pine-
clad slopes of the mountains. This locality,
Guaigiii, is separated from the known
range of A. h. haleatus by the Vallc de
Cibao, which here is a moderately arid
and broad valley presently very much un-
der cultivation. I have seen no specimens
from this intervening valley but surely the
lizards occur there, despite the cultivation.

One other specimen requires mention.
This is a subadult male (ASFS V33856)
with a snout-vent length of 55, from 3.4
mi. (5.4 km) SE Los Montones, Rio Bao,
1600 feet (488 meters). This locality is on
the northern slopes of the Cordillera Cen-
tral, some 45 kilometers to the west of
Guaigiii, but separated from Guaigiii by
intervening, moderately high spurs of the
Cordillera Central. The specimen was se-
cured by a local boy in an area of high-
canopied forest along the Rio Bao. A visit
by ourselves to this area at night yielded
no A. haleatus, despite exceptionally fine
conditions. The lizard in life was all green
except for a white preaxillary bar, and the
dewlap was dull brownish. This specimen
in no way resembles comparably sized ju-
venile multistruppus, in either color or pat-
tern. Its status remains uncertain.

To the east, multistruppus must come in
close contact or intergrade with the sub-
species that occurs throughout the north-
eastern portion of the Republica Domini-
cana; details of this contact will be
discussed under the account of the latter
subspecies. Likewise, to the south, multi-
struppus may come into contact with the
subspecies in the high Cordillera Central;
details of this association will be discussed
under the description of the Central sub-
species.

Remarks. All specimens of A. h. multi-
struppus were collected on two occasions,
while the lizards slept at night. Young in-
dividuals were taken from generally low
situations on shrubs and the lower
branches of trees, whereas adults were ob-
served sleeping in the higher canopy; the
total range of heights was between 5 feet

124 Bulletin Museum of Comparative Zoology, Vol. 146, No. 2

and 25 feet (1.5 and 7.6 meters). The Rio scales always high, subocular scales al-

Camu flows through the Guaigiii woods, most always separated from supralabial

and many individuals were taken from scales by one row of scales, both sexes

tree limbs that overhang the river. either marbled or blotched with varying

The name multistruppus is from the shades of greens or browns, or dark brown

Latin "multus" for "many" and "struppus" banded with dull cream, never with many

for "thong, strap," in allusion to the many fine crossbars, venter in males pale green,

dorsal crossbands in this subspecies. flecked with darker green, and male dew-
laps pale yellow-orange to orange, gray

Anolis baleatus sublimis new subspecies basally and marbled green anteriorly, fe-

Holotijpe. CM 54104, an adult male, male dewlaps irregularly yellow-orange

from 0.3 mi. (0.5 km) E El Rio, 3800 feet with brown spotting.

(1159 meters). La Vega Province, Repub- Distribution. The uplands of the Do-

lica Dominicana, taken by Richard Thomas minican Cordillera Central at elevations

on 26 June 1963. Original number ASPS between 2000 and 4000 feet, in the area

X8114. between El Rio, La Palma, and Jarabacoa.

Paratypes. ASFS X8558, 4 km SW El Description of holotype. An adult male

Rio, 4000 feet ( 1220 meters ) , La Vega with a snout-vent length of 143 and a tail

Province, Republica Dominicana, R. F. length of 167 (regenerated); snout scales

Klinikowski, 2 July 1963; USNM 62104-05, at level of second canthals 3, 7 vertical

El Rio, La Vega Province, Republica Do- rows of loreal scales, 2 scales between the

minicana, W. L. Abbott, 19 May 1919; supraorbital semicircles, 4/4 scales be-

AMNH 41294, El Rio, La Vega Province, tween the interparietal and the supraor-

Repiiblica Dominicana, G. K. Noble, 31 bital semicircles, vertical dorsals 20, hori-

August 1922; ASFS V18594, La Palma, 14 zontal dorsals 21, ventrals 29, one row of

km E El Rio, 3500 feet (1068 meters). Re- scales between the suboculars and supra-

publica Dominicana, J. A. Rodgers, Jr., 30 labials, fourth toe lamellae on phalanges II

July 1969; MCZ 107019-21, La Palma, 14 and III 31, nuchal crest scales and dorsal

km E El Rio, 3500 feet ( 1068 meters ) , La body crest scales high; in life, dorsum

Vega Province, Republica Dominicana, na- dark brown banded with dull cream, this

tive collectors for E. E. Williams and A. S. pattern extending onto the tail, eyeskin

Rand, 25-31 July 1968; MCZ 128397, La grayish with a pale yellow eyering, venter

Palma, 14 km E El Rio, 3500 feet ( 1068 pale green, flecked with darker green, chin

meters). La Vega Province, Republica Do- and throat concolor with and patterned

minicana, T. P. Webster and R. B. Huey, like venter, dewlap pale yellow-orange,

6 July 1971; ASFS V18363-69, 8 km W grayish basally and marbled with green an-

Jarabacoa, 2000 feet (610 meters). La teriorly.

Vega Province, Republica Dominicana, J. Variation. The series of 18 sublimis is

A. Rodgers, Jr., 19 July 1969. composed of nine males and nine females.

Definition. A subspecies of A. baleatus The largest male (USNM 62104) has a

characterized by the combination of mod- snout-vent length of 150, the largest

ally 2 snout scales between second canthal female (MCZ 107021) 141; the male is

scales, 7 vertical rows of loreal scales, 3 from El Rio, the female from La Palma.

scales between the supraorbital semicir- Snout scales at the level of the second

cles, 4/4 scales between the interparietal canthals range between 2 and 5; the mode

and the supraorbital semicircles, high num- is 2 (10 specimens ) . The vertical loreal

ber of vertical dorsal scales (17-21; mean rows vary between 6 and 9, with a mode of

19.2 ) , high number of ventral scales ( 19- 7 ( eight specimens ) . There are 2 to 4

32; mean 25.1), nuchal crest scales very scales between the supraorbital semicir-

high (usually) to high (rarely), body crest cles (mode 3). There are modally 4 scales

HisPANioLAN Giant Angle • Schwartz 125

between the interparietal and the supraor-
bital semieircles; 4 scales are involved in
65 percent of the combination; the actual
counts are 3/3 ( 1), 3/4 ( 1), 4/4 ( 10), 4/5
(1), 5/5 (3), and 5/6 (1). Vertical dor-
sals range between 17 and 21 (mean 19.2),
horizontal dorsals between 17 and 24
(20.4), and ventrals between 19 and 32
(25.1). Of the six adult males, three have
the nuchal crest scales very high and three
ha\e them high, whereas all five adult fe-
males have these scales very high. All
adults of both sexes have the dorsal body
crest scales high. Three lizards ( 17 per-
cent) have the subocular scales in contact
with the supralabials.

In the green phase, adults of both sexes
are irregularly marbled or blotched with
varying shades of green or browns,
whereas in the brown phase, the body is
dark brown with three cream crossbands.
In males the venter and the chin and
throat are pale green, flecked or mottled
with darker green, the flecking or mottling
variably expressed in the series. The dew-
lap in males is pale yellow-orange to or-
ange, gray basally and often with marbled
green markings anteriorly, these markings
being a continuation of the dark green
throat markings. In females, the dewlap is
irregularly mottled with yellow-orange and
has some brown spotting. As preserved,
the series is remarkably uniform in show-
ing vague pale-and-dark marblings or mot-
tlings, and no adult shows any indication
of crossbands.

The series of paratypes includes six ju-
veniles and subadults, with snout-vent
lengths between 49 and 94. The three
smallest of these (49-70) are presently
patternless, as is also a specimen with a
snout-vent length of 73. Two other sub-
adults (snout-vent lengths 75 and 94)
show vague indications of mottled dorsum
with (in the larger) three slightly paler
dorsal crossbands. The larger of these two
specimens was recorded in life as dark
green dorsally with pale green crossbands,
and the interbars are mottled or marbled
with greens. The smallest juvenile noted

above was bright yellow-green in life and
had the venter slightly paler yellow-green;
the concealed surfaces of the thighs were
lead-gray, bordered above by bufl^y. The
absence of pale crossbars in very young
specimens of S'tihli)nis is noteworthy

Comparisons. There are no meristic
counts that separate sul)Iimis from adja-
cent imiltisfruppus; the means of ventral
scales in the two subspecies difl^er slightly,
however (22.3 in sul)li77iis, 25.1 in multi-
struppus). There is also a tendency for
su])limis to have more consistently very
high to high nuchal crest scales. The two
subspecies diff^er abundantly in body pat-
tern, however, with miilfistruppus having
many fine dorsal crossbands and sublimis
having basically a blotched dorsal pattern
with three bars present in some instances.
The juveniles of these two subspecies are
equally as distinct as the adults are in dor-
sal body pattern. The dewlaps are brighter
in male su])limis than in male muUistnip-
pus, the latter tending toward pale yellow
and yellow-grays, whereas in the former
the dewlaps are yellow-orange to orange,
although there is a gray basal suff^usion.
The ventral and throat flecking or mot-
tling in siihlimis differs from the unmarked
condition in mtiltistruppiis.

A. h. sublimis differs from A. h. hcileatiis
in having 2 (rather than 4) snout scales at
the level of the second canthals, 4/4
(rather than 5/5) scales between the in-
terparietal and supraorbital semicircles,
higher means of vertical dorsal scales ( 19.2
versus 17.5) and ventrals (25.1 versus
23.8). Both subspecies have very high to
high nuchal crest scales. In color, haleaius
is much the brighter, with an immaculate
bright yellow to orange throat and bright
yellow to orange dewlap in males, whereas
the dewlaps in sublimis are as bright as
those in haleatus but have a gray basal
wash. The patterned throat and venter in
sublimis differ from the immaculate throat
in buleatus. The dorsal patterns of these
two subspecies likewise are quite different,
that of baleatus regularly consisting of

126 Bulletin Museum of Comparative Zoology, Vol. 146, No. 2

three pale crossbands, whereas that of sub-
limis is mottled or blotched.

Discussion. A. h. suhlimis is closest
geographically to miiltistruppus; the two
subspecies are known from localities sep-
arated by only 20 kilometers airline
(Guaigi^ii and 8 km W Jarabacoa), but
minimally by a 1700-foot (519 meters) dif-
ference in elevation and by extensive
stands of pine forest, a habitat which no
Hispaniolan giant anole occupies. All spec-
imens of suhlimis were collected in mon-
tane gallery forest along streams, and the
subspecies appears to be restricted to this
sort of situation. Rand and Williams
(1969: 9) noted that they collected one
juvenile about 10 feet (3 meters) up on a
small branch of a forest tree at La Palma,
and that two adults were brought to them
by natives from a large tree in a nearby
agricultural area. A. h. suhlimis is thus
not known to come into contact with multi-
struppus on the northern slopes of the Cor-
dillera Central nor with the yet-to-be-de-
scribed subspecies to the east in the
Dominican lowlands. Likewise, it should
be recalled that the southwestern slopes of
this range are occupied by A. r. ricordi;
the nearest localities for ricordi and sul)-
lirnis (Juan de Herrera; south of El Rio)
are separated by about 70 kilometers air-
line, but this intei^vening area is composed
of the rugged and very high massif of the
Cordillera Central whose upper elevations
are covered by pine. It seems unlikely that
ricordi and suhlimis come into contact di-
rectly across the Cordillera.

The juvenile (ASFS 33856) from near
Los Montones upon which I commented
in the discussion of A. h. miiltistruppus
may be correctly assigned to suhlimis, since
the habitat and elevation for that specimen
is much more like that for suhlimis than
multistruppus. In color and lack of pattern
it agrees quite well with small suhlimis,
but until adults have been collected in the
Los Montones region (which lies some 30
kilometers to the northwest of Jarabacoa,
the closest suhlim.is locality) I am reluc-
tant to extend the known range of suhlim,is

into that area. It is this Los Montones A.
haleatus which is closest geographically
(50 kilometers) to an A. ricordi locality
( Los Quemados ) in the northwestern por-
tion of the Republica Dominicana.

Remarks. All ASFS A. h. suhlimis were
taken at night while asleep. All situations,
as noted above, were stream-associated
hardwood forest and cafetales, and the liz-
ards slept on vines and branches in their
customary fashion. The restriction of suh-
limis to riverine gallery forest is doubtless
artificial, since it is only along rivers and
streams in this area that any of the original
montane hardwood forests still remain. In
one case (west of Jarabacoa) the stream
was extremely steep, whereas in others the
streams were level. The altitudinal distri-
bution ( to which the name suhlimis refers )
is high. Only A. r. viculus reaches as high
an elevation in the Massif de la Hotte in
southwestern Haiti.

Anolis baleofus caeruleolatus
new subspecies

Holotype. USNM 193976, an adult
male, from 1.0 mi. (1.6 km) S Caiio Abajo,
>/Iaria Trinidad Sanchez Province, Repub-
lica Dominicana, one of a series collected
by native collectors on 28 November 1971.
Original number ASFS V34486.

Paratypes. CM 54119-26, MCZ 125628-
33, ASFS V34502-13, same data as holo-
type; AMNH 6017, Villa Riva, Duarte
Province, Republica Dominicana, C. R.
Halter, May-July 1915.

Associated specimens. REPCBLICA
DOMINICANA: Duarte Province, Los
Bracitos (AMNH 41465-66); ca. 4 km NE
Ponton (Rio Cuaba) (ASFS V2987); San-
chez Ramirez Province, 1 km SE La Mata
(ASFS V33650-51); La Vega Province,
12.8 km NW Bonao, 1200 feet (366 meters)
(ASFS V4317); 71 km NW Santo Domingo
(= near La Cumbre) (MCZ 128369); San
Cristohal Province, 5.0 mi. (8.0 km) NE
Gonzalo, 1000 feet (305 meters) (ASFS
V29420-21).

Definition. A subspecies of A. haleatus
characterized by the combination of mod-

HisPANioLAN Giant Anole • Schwartz

127

ally 4 scales between second canthal scales,
8 vertical rows of loreal scales, 3 scales be-
tween the supraorbital semicircles, 5/5
scales between the interparietal and the
supraorbital semicircles, moderate number
of vertical dorsal scales (14-22; mean
17.1), moderate number of ventral scales
(15-32; mean 22.4), nuchal crest scales
very high to high (usually) to moderate
or even low (rarely) in both sexes, body
crest scales extremely variable, modally
moderate in both sexes, but with some oc-
cinrences of high and many occurrences
of low body crest scales, subocular scales
almost always separated from supralabial
scales by one row of scales, both sexes
some shade of green (usually dark) with
foiu- pale green crossbars and with bright
sky-bhie blotches along the junction of the
green dorsal color and the paler venter
(less prominent in females than in males),
dewlap in males pale yellow to orange, in
females pale yellow to orange but with
much dark brown to grayish streaking or
smudging, throat in males deep yellow-or-
ange and immaculate or with very faint
greenish dots, in females yellow-green to
bright yellow, always with some darker
green dots, rarely marbled with dark green,
but never streaked with that color.

Distribution. Northeastern Republica
Dominicana, from Duarte, Sanchez Rami-
rez, La Vega, and northern and eastern
San Cristobal provinces, to the base of the
Peninsula de Samana (Caiio Abajo); in-
tergrades with the subspecies to the south
and east in the region of El Seibo Province.

Description of holotijpe. An adult male
with a snout-vent length of 137 and a tail
length of 250; snout scales between second
canthals 4, 7 vertical rows of loreal scales,
3 scales between the supraorbital semicir-
cles, 6/6 scales between the interparietal
and supraorbital semicircles, vertical dor-
sals 16, horizontal dorsals 23, ventrals 26,
one row of scales between the suboculars
and supralabials, fourth toe lamellae on
phalanges II and III 30, nuchal crest scales
very high, body crest scales moderate; in
life, dorsum dark green with four pale

green crossbars, the dark green color
blending fjuickly at the junction of the dor-
sal and ventral color into a series of diag-
onally directed sky-blue areas that give a
ragged appearance to the jmiction of the
dorsal and ventral colors; dorsal crossbands
continue onto the tail; cascjue gray-green,
eyeskin pale pea-green; dewlap pale yel-
low-orange, chin slightly deeper yellow-or-
ange, throat yellow-orange, immaculate
except for some vague pale greenish
smudges posterolaterally.

Variation. The series of A. b. caeruleo-
lattis consists of 20 males and 17 females.
The largest male (ASFS V34505) has a
snout-vent length of 148, the largest fe-
male (AMNh' 6017) 145. The male is
from the type locality, the female from
Villa Riva. Snout scales at the level of the
second canthals range between 2 and 5;
the mode is 4 ( 14 specimens ) . The verti-
cal loreal rows vary between 6 and 10; the
mode is 8 ( 15 specimens ) . There are 2 or
3 scales between the supraorbital semicir-
cles (mode 3). There are modally 5 scales
between the interparietal and the supraor-
bital semicircles; 5 scales are involved in
52 percent of the combinations; actual
counts are 4/4 (3), 4/5 (6), 5/5 (10), 5/6
(7), 6/6 (4), 6/7 (1), 4/6 (1), and 5/7
( 1 ) . Vertical dorsals range between 14
and 22 (mean 17.1), horizontal dorsals be-
tween 15 and 25 ( 19.9 ) , and ventrals be-
tween 15 and 32 ( 22.4 ) . Of 16 adult males,
four have the nuchal crest scales very high,
11 have them high, and one has them mod-
erate. Of 17 adult females, four have the
nuchal crest scales very high, ten have
them high, and three have them moderate.
In the adult males, the body crest scales
are high in six males, moderate in eight,
and low in two, whereas in the adult fe-
males, these scales are high in five, mod-
erate in six, and low in six. All but two
lizards (6 percent) have the suboculars
separated by one row of scales from the
supralabials.

In a series of 12 adult male topotypes,
the dorsal ground color was recorded as
some shade of green (usually dark green)

128 Bulletin Museum of Comparative Zoology, Vol. 146, No. 2

with four pale pea-green crossbands. The patterned hke adults except that the sky-
dorsal green color blends quickly ventro- blue lower edges to the dorsal color were
laterally into a series of irregular sky-blue absent and the dewlap was streaked brown
patches or blotches that mark the border and gray basally. The chin and throat
between the dorsal green and the pale yel- were immaculate pale green. There are no
low to cream venter. These sky-blue color data on the other juveniles, and none
patches are often prominently extended of them presently shows any pattern,
onto the lateral margins of the venter as a Comparisons. A. h. caeruleolatus dif-
series of diagonal, posteriorly directed fers from all previously described subspe-
areas, which, upon preservation, are still cies in».having the sky-blug patching along
prominent features of the lower sides. The the lower sides. In having four dorsal pale
upper surface of the head was gray-green green body bands, caeruleolatus differs
to brown, the eyeskin pale pea-green. The strikingly from multistruppus with its mul-
dorsal banded pattern of dark and light tiple banding; in addition, the dewlap of
green continues onto the tail. The dewlap multistruppus is pale and often grayish, in
is pale yellow-orange, yellow, or orange, contrast to the generally brighter dewlaps
and the chin is slightly deeper yellow-or- of caeruleolatus. From nominate baleatus,
ange, concolor with the throat, which is caeruleolatus differs in having the throat
either immaculate (usually) or with very yellow to yellow-green rather than bright
faint greenish dots or smudges. Eleven fe- yellow to orange, and female caeruleolatus
male topotypes were colored and patterned have the throat with dark green markings,
dorsally like the males, with the pattern From high upland sublitnis, caeruleolatus
extending onto the tail, but there is only a differs in having the sky-blue blotches ven-
vague indication of the ventrolateral sky- trolaterally and in lacking ventral mark-
blue pigmentation. The necks of females ings, and whereas caeruleolatus has com-
were often streaked with dark and pale parably pigmented dewlaps^ those in
greens. The chin and throat were yellow sublimis are generally paler and often suf-
to yellow-green, regularly with some fus^d at least basally with gray. The dor-
darker green dots, blotches, or occasionally sal patterns of both sublimis and caeruleo-
marbled with dark green. The female dew- latus are comparable, since both are
lap was yellow to pale orange, streaked crossbanded.
with dark brown or grayish. As far as meristic counts are concerned,

Two females from the haitises region caeruleolatus differs from the named sub-
near Gonzalo were deep to emerald green species in the following ways. Compared
in life with yellow dewlaps having varying with baleatus, caeruleolatus has modally 8
amounts of brown streaking or smudging; (rather than 7) vertical loreal rows, and a
the limbs were contrastingly banded dark lower mean number of ventral scales (22.4
and pale green. The throats were bright versus 23.8). There is also a strong ten-
yellow to bright green, with scattered dency for adult caeruleolatus to have mod-
deeper green spots in each case. In a pair erate to low body crest scales, whereas in
from La Mata, the dorsa were bright baleatus the tendency is toward high to
green, somewhat marbled with yellow and moderate body crest scales. Compared
yellow-green, the upper surfaces of the with multistruppus, caeruleolatus has mod-
heads were pale fawn, the eyeskin pale ally 4 (rather than 2) snout scales at the
grayish green, and the dewlaps orange in level of the second canthals, 8 rather than
both sexes. 7 vertical rows of loreals, 5/5 rather than

The series of A. b. caeruleolatus includes 4/4 scales between the interparietal and

four juveniles and subadults with snout- the supraorbital semicircles, and a lower

vent lengths from 60 to 91; the largest of mean of vertical dorsal scales (17.1 versus

these is a topotype that was colored and 18.6). With regard to body crest scales,

HisPANioLAN Giant Angle • Schwartz 129

these two subspecies show the same situa-
tion as caeruleolatus and baleatus. Com-
pared with .suJ)limis, caeruleolatus has 4
(rather than 2) snout scales at the level of
the second canthals, 8 (rather than 7) ver-
tical rows of loreals, 5/5 (rather than 4/4)
scales between the interparietal and the
supraorbital semicircles, and lower means
of both vertical dorsals (17.1 versus 19.2)
and ventrals (22.4 versus 25.1). A. h. suh-
limis has not been recorded as having the
dorsal body crest scales other than high,
in contrast to the strong tendency in cae-
ruleolatus of having these scales moderate
to low.

Discussion. A. h. caeruleolatus centers
in the extremely mesic eastern portion of
the Valle de Cibao in that area that has
the most rainfall in the Republica Domin-
icana. I have already commented on the
specimens from Los Bracitos, Duarte Prov-
ince; these specimens are old and pattern-
less and are from a locality in the Cordillera
Septentrional which is, farther west, occu-
pied by A. h. baleatus; I include them with
caeruleolatus provisionally. The specimen
from Ponton, Duarte Province, is a juve-
nile (ASFS V2987; snout-vent 60) and is
presently patternless; no color data are
available. It too I only provisionally re-
gard as caeruleolatus. The two specimens
from La Vega Province (ASFS V4317,
MCZ 128379) are also without color data
in life, and the former is a patternless ju-
venile (snout-vent 69). Specimens from
these last two localities also require verifi-
cation as to subspecfic status.

A. h. caeruleolatus presumably inter-
grades with four subspecies: baleatus, mul-
tistruppus, the subspecies on the Peninsula
de Samana, and subspecies to the south-
east. Only in the last case are specimens
that I interpret as intergradient known,
and they will be discussed under the de-
scription of the southeastern subspecies.
No intergrades are known between the
Samana subspecies, baleatus, or multistrup-
pus. Distance between caeruleolatus and
the nearest localities for these subspecies
are: Samana subspecies — 13 kilometers

(Caiio Abajo and 5 km NW Sanchez);
baleatus — 50 kilometers (Los Bracitos and
Pena); multistruppus — 12 kilometers (12.8
km NW Bonao and Guaigiii). Of these
presumed areas of contact, that between
caeruleolatus and the Samana subspecies is
not unexpected; the area between the two
known localities is very open and relatively
barren and devoid of trees and appears al-
ways to have been so. There are fine high
swamp-forests in the western part of this
intervening region, and it is possible that
intergrades between these two distinctive
subspecies will be encountered in these
forests. Most puzzling is the absence of
intergradation between caeruleolatus and
multistruppus. The specimen from north-
west of Bonao is a juvenile, but it does not
show the characteristic multiple crossbands
of both young and adult multistruppus. It
may be that multistruppus occupies only
the foothills of the Cordillera Central and
that the zone of intergradation between
multistruppus and caeruleolatus is very
abrupt.

Remarks. A. b. caeruleolatus is known
from sea level to an elevation of 1000 feet
(305 meters) in the haitises region near
Gonzalo and 1200 feet ( 366 meters ) north-
west of Bonao. Specimens were secured
primarily from native collectors; the long
series of topotypes is due to the industry
of the inhabitants of Cafio Abajo. The
Cafio Abajo area is one of cafetales and
cacaotales with high canopied shade-trees,
and the lizards apparently are extremely
abundant in this optimal habitat. The
pair of lizards from La Mata were secured
by me while they were copulating on the
side of a large shade-tree in a cafetal about
4 feet (1.2 meters) above the ground at
1225 hours. The two females from Gon-
zalo were taken during the day on large
trees adjacent to a small spring in the
haitises; the surrounding area was under
heavy cultivation, but the doline slopes
were covered locally with undisturbed for-
est.

The name caeruleolatus is from the
Latin "caeruleus" for "blue" and "latus"

130 Bulletin Museum of Comparative Zoology, Vol. 146, No. 2

for "side," in allusion to the sky-blue lower
sides of this subspecies.

Anolis baleatus samanae
new subspecies

Holotype. CM 54105, an adult male,
from 7.6 mi. (12.2 km) NE Sanchez, 1000
feet (305 meters), Samana Province, Re-
publica Dominicana, one of a series col-
lected by native collectors on 28 November
1971. Original number ASFS V34474 .

Paratypes. ASFS V34475-79, same data
as holotype; USNM 193990-92, same local-
ity as holotype, native collectors, 27 No-
vember 1971; MCZ 125634, 5.0 mi. (8.0
km) NW Sanchez Province, Republica Do-
minicana, J. Aria, 27 November 1971; ASFS
V34495-96, 5.0 mi. (8.0 km) NW Sanchez,
Samana Province, Republica Dominicana,
J. Aria, 28 November 1971; CM 54127-30,
5.0 mi. (8.0 km) NW Sanchez, Samana
Province, Republica Dominicana, J. Aria,
30 November 1971; MCZ 12563.5-39,
USNM 193993-4001, 5.0 mi. (8.0 km) NW
Sanchez, Samana Province, Republica Do-
minicana, J. Aria, 1 December 1971; ASFS
V34514, ASFS V34836-38, Las Terrenas,
Samana Province, Republica Dominicana,
native collector, 28 November 1971; ASFS
V1904, 6 km E Sanchez, Samana Province,
Republica Dominicana, R. Thomas, 30 Oc-
tober 1963; AMNH 28651, Samana, Sa-
mana Province, Republica Dominicana, J.
King, August 1924; AMNH 39817-23,
AMNH 42285, Laguna, Samana Province,
Republica Dominicana, W. G. Hassler, Oc-
tober-December 1929; USNM 61928, Cayo
Hondo, Samana Province, Republica Do-
minicana, W. L. Abbott, February 1919.

Definition. A subspecies of A. baleatus
characterized by the combination of mod-
ally 2 snout scales at level of second can-
thai scales, 7 vertical rows of loreal scales,
3 scales between the interorbital semicir-
cles, 4/4 scales between the inteiparietal
and the supraorbital semicircles, moderate
number of vertical dorsal scales ( 13-20;
mean 16.6), moderate number of ventral
scales (16-29; mean 22.1), nuchal crest
scales very high to high (usually) to mod-

erate or low (rarely) in both sexes, body
crest scales high to moderate but often low
in both sexes, subocular scales almost al-
ways separated from supralabial scales by
one (rarely 2) row of scales; dorsum in
both sexes in life blotched dark green,
greenish, dull gray-green, brown, or black-
ish, dewlaps in males dull yellow to pale
yellowish orange, in females very pale yel-
low to pale yellowish orange, streaked with
blackish or brown basally, and chin and
throat in males cream to yellowish or yel-
low-orange, mottled with black or gray, in
females pale green to greenish yellow with
dark green to brown streaking or even re-
ticulate.

Distribution. The Peninsula de Samana
in the northeastern Republica Dominicana,
and apparently islets in the Bahia de Sa-
mana.

Description of holotype. An adult male
with a snout-vent length of 145 and a tail
length of 222 (regenerated); snout scales
between second canthals 3, 6 vertical rows
of loreal scales, 3 scales between the supra-
orbital semicircles, 4/4 scales between the
interparietal and the supraorbital semicir-
cles, vertical dorsals 18, horizontal dorsals
19, ventrals 21, one row of scales between
the suboculars and supralabials, fourth toe
lamellae on phalanges II and III 30, nuchal
crest scales very high, body crest scales
high; in life, dorsum mottled dull greens
and gray-brown with whitish (almost
cream but suffused with pale gray); upper
surface of head mixed dark brown and
gray, venter dull greenish, dewlap orange,
chin and throat creamy to yellowish, not
marked with green.

Variation. The series of 54 A. b. sama-
nae consists of 32 males and 22 females.
The largest male (AMNH 39807) has a
snout-vent length of 157; the largest fe-
males (CM 54130, USNM 193994) have
snout-vent lengths of 145. The male is
from Laguna, the females from 5.0 mi. NW
Sanchez. Snout scales at the level of the
second canthals range between 2 and 5;
the mode is 2 (24 specimens). The verti-
cal loreal rows vary between 5 and 9; the

HisPANioLAN Giant Angle • Scliwartz 131

mode is 7 (25 specimens). There are 2 or
3 scales between the supraorbital semicir-
cles (mode 3). There are modally 4 scales
between the interparietal and the supraor-
bital semicircles; 4 scales are involved in
43 percent of the combinations: actual
counts are 3/3 (2), 3/4 (3), 4/4 ( 17), 4/5
(8), 5/5 (13), 5/6 (5), 6/6 (1), 6/7 (2),
4/6 (1), and 3/5 (1). Vertical dorsals
range between 13 and 20 (mean 16.6),
horizontal dorsals between 13 and 27
(19.3), and ventrals 16-29 (22.1). Of 30
adult males, 14 have the nuchal crest scales
very high, 15 have them high, and one has
them moderate; in 20 adult females, nine
have the nuchal crest scales very high,
nine have them high, one has them mod-
erate, and one has them low. Body crest
scales in males are high in three lizards,
moderate in 16, and low in ten; in females,
11 have these scales moderate and ten have
them low. The suboculars are separated
from the supralabials by one row of scales
in all but four specimens (7 percent),
which have them in contact, and one spec-
imen (2 percent), which has 2 rows of
scales in this position.

A. ]). samanae is basically a blotched liz-
ard, and no adults show any indication of
crossbanding. The body is irregularly
blotched with blackish, dark green, dull
green, gray-brown, and occasionally there
are sky-blue areas along the ventrolateral
margin of the dorsal coloration in males,
but these areas are not so prominent as in
caeruleolatus. Regardless of the dorsal
shades, the upper surface of the head is
mixed dark brown and shades of gray in
both sexes. The hindlimbs are finely
barred with pale and dark green. The
venter is dull greenish in both sexes. The
dewlap in males varies from dull yellow or
pale yellowish orange to orange, and the
chin and throat are yellowish, cream, or
yellow-orange, mottled with black or gray.
In females, the dewlaps are very pale yel-
low, pale yellow-orange, or grayish orange,
at times streaked with blackish or brown
basally, and the chin and throat ground
color is pale green, marbled, streaked, or

even reticulate with dark green to (rarely)
brown.

There arc one juvenile (AMNH 28651;
snout-vent length 40) and two subadult
(snout-vent lengths 92 and 97) A. 1). sa-
manae. The subadults are old and discolored
but their patterns seem not to differ from
those of full adults. The juvenile on the
other hand, has fom- bold pale crossbars on
the dorsum, the pattern continuing onto
the tail. This young individual has the
umbilicus still present and is presumably
near hatchling size.

Comparisons. Since samanae and cae-
ruleolatus are adjacent geographically, the
most pertinent comparisons are between
them. Examples of these two populations,
as noted in the introduction to the present
paper, were available to me simultaneously
and I was struck with their differences in
life. A. 1). samanae is a blotched lizard
whereas caeruleolatus is a crossbanded
one; the latter subspecies also typically has
sky-blue ventrolateral blotches, a feature
absent (or occasionally poorly expressed)
in male samanae. Male dewlap colors are
similar in both subspecies, although fe-
male dewlap colors in samanae seem some-
what paler than those of caeruleolatus.
The chin and throat markings of the two
subspecies are quite distinct; in male cae-
ruleolatus, the throat is deep yellow to
yellow-orange, at best with very faint gray-
ish dots or smudges, whereas in male
samanae the throat is yellowish or cream
to yellow-orange, mottled with black or
gray. In female caeruleolatus, the throat
is yellow to yellow-green, always with
some dark green dots, blotching, or mar-
bling, whereas in samanae females, the
throat is pale green, greenish yellow, or
yellow-green, with dark green to brown
streaking or reticulum.

The only subspecies thus far described
which is blotched like samanae is the Cor-
dillera Central suhlitnis, although caerul-
eolatus may show a marbled dorsum in
some areas. No pigmental or pattern dif-
ferences separate samanae and sublimis,
since in both dorsal coloration and color of

132 Bulletin Museum of Comparative Zoology, Vol. 146, No. 2

the dewlap the major color involved is
green. However, the throat in male sub-
limis is pale green, whereas in samanae it
is cream to yellow-orange. Certainly rnulti-
struppus and samanae are easily distin-
guished in the field by their very different
dorsal patterns, for example, and haleatus,
with its very, very bright chin and thioat,
both of which are immaculate, is quite dis-
tinctive from samanae.

In meristic data, samanae differs from
caeruleolatus in having 2 (rather than 4)
snout scales, 7 (rather than 8) vertical
rows of loreals, and 4/4 (rather than 5/5)
scales between the inteiparietal and the
supraorbital semicircles. From multistrup-
pus, samanae differs in having a lower
mean of vertical dorsal scales ( 16.6 versus
1S.6), and the same difference occurs be-
tween samanae and su])Umis ( 16.6 versus
19.2) and in ventrals (22.1 versus 25.1).
From haleatus, samaiuie differs in having
2 (rather than 4) snout scales, 4/4 (rather
than 5/5) scales between the inteiparietal
and the supraorbital semicircles, and lower
means in both vertical dorsals ( 16.6 versus
17.5) and ventrals (22.1 versus 23.8). The
nuchal crest scales in samanae are more
consistently very high to high than they
are in any of the other subspecies of A.
haleatus.

Discussion. As pointed out in the dis-
cussion of A. h. caeruleolatus, there are no
intergrades known between that subspe-
cies and samanae. The isthmus of the Pe-
ninsula de Samana is much cleared and
locally even barren, but there are large
western swampy areas that support mag-
nificent hardwood forests toward the land-
ward side. These forests may well support
intermediates between samanae and cae-
ruleolatus, or, because of their proximity
to the mainland, they may be inhabited by
caeruleolatus. Specimens from 5.0 mi. NW
Sanchez, that locality for samanae which
is closest to a known locality for caeruleo-
latus ( 18 kilometers ) , show no tendencies
toward the crossbanded condition of cae-
ruleolatus.

A. h. samanae is the only Hispaniolan
giant anole known by specimens from any
off-shore island or islet. The specimen
from Cayo Hondo, taken by W. L. Abbott,
constitutes this record, although I am un-
able to locate this islet. I assume it is one
of the archipelago within the Bahia de Sa-
mana.

Remarks. All but one A. h. samanae se-
cured by myself and parties were native-
collected. The exception is a lizard taken
by Richard Thomas, one of two seen on a
small tree and in a vine tangle in a steep
limestone ravine east of Sanchez. The area
of the type locality is in the uplands of the
Sierra de Samana on the road between
Sanchez and Las Terrenas. Thus newly
constructed road passes through superb
mesic high-canopied forest, and much of
the area is not yet seriously disturbed. Ob-
viously from the number of lizards secured
by natives in this region, A. /;. samanae is
common. The range is not high, with a
maximum elevation of 1673 feet (510 me-
ters) in Monte Las Caiiitas; this mountain
lies between Sanchez and Las Terrenas.
Specimens from Las Terrenas itself were
secured by natives from near-coastal mesic
cafetales and cacaotales, and lizards from
northwest of Sanchez were in similar situ-
ations.

Only three other reptiles (Diplo^lossus
sternurus alloeides Schwartz, Leiocephalus
personatus pyrrholaemus Schwartz, and
Dromicus parvifrons niger Dunn) are known
to have differentiated at the subspecific
level on the Peninsula de Samana. Sphaero-
clactylus clenchi Shreve and Sphaeroclac-
tijlus samanensis Cochran both occur there
and have as yet unnamed populations, one
of which in each case is limited to the pen-
insula. It is also of interest to note that in
Anolis clisticJius Cope, the Samana popula-
tion is identical to the population on the
soutliern shores of the Bahia de Samana
(ignigularis Mertens), but that the range of
this subspecies is interrupted at the head
of the Bahia de Samana by A. cl. domini-
censis Reinhardt and Liitken (see Schwartz,
1968: 280-81, for details).

HisPANioLAN Giant Angle • Sclucmiz 133

Anolis baleatus litorisilva new subspecies
Holotype. USNM 193977, an adult
male, from 1.2 km SSW Piinta Cana, La
Altagracia Province, Rcpviblica Domini-
cana, one of a series collected by Danny
C. Fowler and Bruce R. Sheplan, on 24
November 1971. Original number ASFS
V35095.

Paratypes. ASFS \'35096-100, same
data as holotvpe; CM 54113-14, MCZ
125616-17, 5.5 km SSW Punta Cana, La
Altagracia Province, Republica Domini-
cana, D. C. Fowler, 27 November 1971;
ASFS V29090, Juanillo, La Altagracia
Province, Republica Dominicana, native
collector, 24 July 1971; ASFS V961-62, 0.5
mi. NW Boca de Yuma, La Altagracia
Province, Republica Dominicana, R. F.
Klinikowski, R. Thomas, 2 September 1963;
ASFS VI 136, 2.5 km NW Boca de Yuma,
La Altagracia Province, Republica Domin-
icana, native collector, 4 September 1963;
ASFS V17573, 4 km NW Boca de Yuma,
La Altagracia Province, Republica Domi-
nicana, A. Schwartz, 13 June 1969; ASFS
V17616, 2 km NW Boca de Yuma, La Al-
tagracia Province, Republica Dominicana,
J. B. Strong, 15 June 1969.

Definition. A subspecies of A. I)(ileatus
characterized by the combination of 2 or 4
scales at level of the second canthal scales,
7 vertical rows of loreal scales, 3 scales be-
tween the interorbital semicircles, 4/5
scales between the interparietal and the
supraorbital semicircles, low number of
vertical dorsals (13-19; mean 15.9), low
number of ventral scales ( 18-26; mean
21.3), nuchal crest scales always very high
to high in both sexes, body crest scales
high (rarely) to moderate or low, suboc-
ular scales usually separated from supra-
labial scales by one row of scales; dorsum
in life varying from light blue-brown to
light greenish brown in males, dull brown
to olive-brown in females, blotched with
creamy to gray, dewlap in males bright
orange, brownish in females, and chin and
throat (including lips) bright orange in
males, pale yellow-green in females.

Distri])ution. Extreme eastern Repub-
lica Dominicana in La Altagracia Province,
from Punta Cana to the \'icinity of Boca de
Yinna.

Description of holotype. An adult male
with a snout-vent length of 136 and a tail
length of 183 (regenerated); snout scales
between sc^cond canthals 2; 6 vertical rows
of loreal scales, 3 scales between the supra-
orbital semicircles, 5/5 scales between the
interparietal and the supraorbital semicir-
cles, vertical dorsals 15, horizontal dorsals
16, ventrals 25, subocular scales in contact
witli the supralabial scales, fourth toe la-
mellae on phalanges II and III 31, nuchal
crest scales very high, dorsal body crest
scales high; in life, dorsum blotched light
blue-brown and light green-brown; venter
pale gray-green; chin, lips, and dewlap
bright orange.

Variation. The series of 16 A. h. litori-
silva is composed of six males and ten fe-
males. The largest male (MCZ 125616)
has a snout-vent length of 158, the largest
female (ASFS V961) 131. The male is
from 5.5 km SSW Punta Cana, the female
from 0.5 mi. NW Boca de Yuma. Snout
scales at the level of the second canthals
range between 2 and 5; there are t\vo
modes, 2 and 4, each with five individuals.
The vertical loreal rows vary betwe(Mi 6
and 9; the mode is 7 (nine specimens).
There are 2 to 4 scales between the supra-
orbital semicircles (mode 3). There are
modally 4/5 scales between the interpari-
etal and the supraorbital semicircles; 5
scales are involved with 59 percent of the
combinations; actual counts are 4/4 (4),
4/5 (6), 5/5 (5), and 5/6 (1). Vertical
dorsals range between 13 and 19 (mean
15.9), horizcmtal dorsals between 14 and
22 (18.5), and ventrals between 18 and 26
(21.3). Of four adult males, three have
the nuchal crest scales very high and one
has them high; of five adult females, two
have these scales very high and three have
them high. In the males, the body crest
scales are high in one and moderate in
three, and in the females, these scales are

134 Bulletin Museum of Comparative Zoology, Vol. 146, No. 2

moderate in two and low in three. The
siiboculars are separated from the siipra-
labials by one row of scales in all but one
specimen (6 percent).

A. h. litorisilva is essentially a blotched
lizard whose colors do not include bright
or even medium greens. The color notes
on the holotype apply equally well to the
other adult males — the dorsum is blotched
with bluish browns and light greenish
browns, without any clear greens, and
the blotching is often more pronounced on
the head than on the body. In females, the
dorsum is dull brown to olive-brown with
only occasional slight remnants of a lighter
green pattern on the head; the blotching
in the female involves creamy to gray pig-
mentation. The venter is pale gray-green
or whitish green in males, pale greenish
gray in females. The dewlap in all adult
males was recorded as bright orange, and
brownish in females. In males, the chin
(including the lips) is bright orange, and
pale yellow-green in females. The upper
surface of the head in males is blotched
like the body and is dark chocolate in fe-
males. In females, the upper surfaces of
the hindlimbs were recorded as olive-
brown, blotched with cream to gray like
the dorsum.

The series of A. h. litorisilva contains
seven juveniles and subadults (snout-vent
lengths 45 to 88). The smallest juvenile
(ASFS V17573, female) was bright green
in life with four pale buffy crossbands and
dark green shadow-bars between the cross-
bands; the tail was ringed cream and dark
gray, and the venter was pale green. The
dewlap was yellow-green and gray. A
slightly larger female (ASFS V17616) with
a snout-vent length of 57 was yellow-green
dorsally and without bands, the head was
brown; the eyeskin was green, and the
venter yellow-green. The tail was banded
black and yellow-green, and the dewlap
was mainly brown with the scale rows yel-
low-green. A still larger female (ASFS
V1136) with a snout-vent length of 67 was
green, faintly crossbarred with grayish
green, and there were charcoal smudges on

the neck. Two male subadults with snout-
vent lengths of 71 and 83 (ASFS V35099-
100) from the type locality were recorded
by Fowler as follows: "One with a strong
vertical banding pattern alternating brown-
green and white-gray, which extends from
tip of tail to the head where it becomes
slightly more diffuse; on the other, the dor-
sal groimd color is dull brown with rem-
nants of banding pattern only around head;
the ventral ground color of the first is gray-
green with brown mottling, the second is
dull gray-brown; in both juveniles, the
dewlap is orange-green and the chin and
lips are green." The largest subadult
(ASFS V29090) was patternless green
above, and the dewlap was orange with
charcoal stripes; the specimen is a female.
Comparisons. Because of its blotched
(rather than crossbanded) pattern, litori-
silva requires comparison with samanae
and suhlimis. The general effect of the
dorsa of all three subspecies is quite sim-
ilar, but samanae and suhlimis are much
the brighter lizards, with greens predomi-
nant in the dorsal pigmentation. On the
other hand, litorisilva is a much more drab
lizard, without clear greens in the adults,
the tendency being toward more sombre
hues, primarily shades of browns. From
all other described subspecies, litorisilva
differs in being blotched rather than cross-
banded and also in having much less
gaudy dorsal colors. In meristic counts,
litorisilva differs from the remaining sub-
species in the following ways. From cae-
ruleolattis, litorisilva differs in having 7
(rather than 8) vertical loreal rows, and
lower means of vertical dorsals (15.9 ver-
sus 17.1) and ventrals (21.3 versus 22.4).
From rnultistruppus, litorisilva differs in
lower means of vertical dorsals ( 15.9 ver-
sus 18.6) and ventrals (21.3 and 22.3).
From stihlitiiis, litorisilva differs in having
lower means of vertical dorsals ( 15.9 ver-
sus 19.2) and ventrals (21.3 versus 25.1).
From haleatus, litorisilva differs in having
lower means of vertical dorsals ( 15.9 ver-
sus 17.5) and ventrals (21.3 versus 23.8).
Meaningful comparisons of litorisilva with

HisPANiOLAN Giant Anole • ScJiwartz 135

other subspecies in counts of snout scales,
and scales between the interparietal and
the supraorbital semicircles, are impossible
since litorisilva has a bimodal condition in
the former (and the bimodes are 2 and 4,
those counts which occur singly as the
mode in the other subspecies) and has a
mode of 4/5 in the latter (whereas all
other species have either 4/4 or 5/5).
Considering the fairly large series of litori-
silva (16 specimens), these two "abnormal"
conditions are puzzling. At least in the
case of 4/5 counts, the absence of 3/3 or
3/4 counts in litorisilva suggests that this
subspecies tends toward a 5/5 count.

Discussion. A. /;. litorisilva appears to
be the extreme eastern isolate of the more
widespread A. haleatus stock. It occupies
semi-arid forests on and near the coast (as
at Juanillo and Punta Cana) and on the
limestone ridge behind Boca de Yuma.
Both situations are far more xeric than is
customary for A. haleatus, and the faded
nongreen coloration of the adults is doubt-
less a response to the dry and open to
dense forest conditions of this region.
Nevertheless, individuals are quite con-
spicuous at night as they sleep exposed. A.
b. litorisilva presumably comes into con-
tact with the subspecies to the north and
west (named below) but intergrades are
presently unknown; in the vicinity of Hig-
iiey (the closest locality for the adjacent
subspecies) the lizards are more brightly
colored and crossbanded and quite unlike
litorisilva.

Remarks. All but one specimen of li-
torisilva were collected by myself and par-
ties. Individuals were found sleeping in
primarily coastal forest (to which the
name, from "litus" for "shore" and "silva"
for "forest," refers in Latin) at elevations
from 4 to 15 feet (1.2 to 4.6 meters) above
the ground. Generally, juveniles sleep
closer to the groimd and in more dense
situations than adults. One juvenile was
taken from a roadside Acacia, a most un-
usual situation (since Acacia is a distinct
xerophyte) for any giant anole. Several
adults were taken in dense viny tangles,

sleeping on the woody vines; the advan-
tage of this situation was made (piite ob-
vious wIkmi I attempted to catch a large
adult at night by hand. The light from my
flashlight wakened the lizard almost im-
mediately, and although 1 was extremely
careful not to jar any of the vines, this was
a vain endeavor. At the first jostling, the
lizard jumped to the ground and escaped
in the dry leaf litter and understory.

Anol'is haleatus scelestus new subspecies

Holotype. CM 54106, an adult male,
from 5.1 mi. (8.2 km) E Santo Domingo
(from Rio Ozama), Distrito Nacional, Re-
publica Dominicana, one of three collected
by David C. Leber and Richard Thomas
on 18 June 1964. Original number ASFS
V2460.

Paratopes. ASFS V2461-62, same data as
holotype; MCZ 125618-27, 8.4 mi. (13.4
km) NE La Romana, 100 feet (31 meters).
La Romana Province, Republica Domini-
cana, B. R. Sheplan, 22 November 1971;
CM 54115-18, USNM 193981-89, 8.4 mi.
(13.4 km) NE La Romana, 100 feet (31
meters). La Romana Province, Republica
Dominicana, D. C. Fowler, A. Schwartz,
17 July 1971; MCZ 16321, La Romana, La
Romana Province, Republica Dominicana,
E. Leider, 1922; ASFS V29284-300, 0.2 mi.
(0.3 km) N Otra Banda, 350 feet (107 me-
ters). La Altagracia Province, Republica
Dominicana, D. C. Fowler, A. Schwartz,
26 July 1971; ASFS V21699-700, 1 km NE
Higiiey, La Altagracia Province, Republica
Dominicana, J. R. Dennis, R. Thomas, 16
August 1969; USNM 193979-80, 0.7 mi.
(1.1 km) W Higiiey, La Altagracia Prov-
ince, Republica Dominicana, R. Thomas,
29 August 1963; ASFS V1038, 1 mi. (1.6
km) W Higiiey, La Altagracia Province,
Republica Dominicana, R. Thomas, 3 Sep-
tember 1963; ASFS V28757, 15.5 mi. (24.8
km) E San Pedro de Macoris, Rio Cumay-
asa. La Romana Province, D. C. Fowler,
12 July 1971; ASFS V28910-16, 15.5 mi.
(24.8 km) E San Pedro de Macoris, Rio
Cumayasa, San Pedro de Macoris Province,
Republica Dominicana, D. C. Fowler, A.

136 Bulletin Musewyi of Comparative Zoology, Vol. 146, No. 2

Schwartz, 16 July 1971; ASFS V28847, 15.5
mi. (24.8 km) E San Pedro de Macoris, La
Romana Province, Repiiblica Dominicana,
A. Schwartz, 15 July 1971.

Associated specimens. REPUBLICA
DOMINICANA: La Altagracia Province, 1
km SE Las Lisas (ASFS V17434-35); San
Cristobal Province, 8 km N Yamasa, 200
feet (61 meters) (ASFS V28656).

Definition. A subspecies of A. haleatus
characterized by the combination of mod-
ally 2 scales at level of the second canthal
scales, 7 vertical rows of loreal scales, 3
scales between the supraorbital semicircles,
5/5 scales between the interparietal and
the supraorbital semicircles, low number of
vertical dorsals ( 12-20; mean 15.4 ) , low
number of ventral scales ( 17-28; mean
21.1), nuchal and body crest scales always
very high to high in both sexes, subocular
scales usually separated from supralabial
scales by one (occasionally two) row of
scales; dorsum in both sexes either green
with three pastel green crossbands or dark
green flecked with light green, cream with
some greenish to brownish green smudges,
dewlap in males deep yellow to deep or-
ange, streaked or smudged with dark
brown to charcoal, and throat in females
dark green marbled with yellow and pale
green (males unrecorded).

Distribution. Southeastern Republica
Dominicana, from the Sierra de Yamasa
and the vicinity of Santo Domingo in the
west, east to the region about Higiiey and
Las Lisas in La Altagracia Province.

Description of Jiolotype. An adult male
with a snout-vent length of 152 and a tail
length of 267; snout scales between second
canthals 4; 8 vertical rows of loreal scales,
2 scales between the supraorbital semicir-
cles, 4/5 scales between inteiparietal and
supraorbital semicircles, vertical dorsals
16, horizontal dorsals 16, ventrals 22, sub-
ocular scales separated from supralabial
scales by one row of scales, fourth toe la-
mellae on phalanges II and III 34, nuchal
crest scales high, body crest scales moder-
ate; in life, dorsum olive-green with six

pastel green crossbands, tail and venter
light green; dewlap dark yellow.

Variation. The series of 61 A. b. sceles-
ttis consists of 27 males and 34 females; a
large number of the specimens are juve-
niles and subadults. The largest male
(ASFS V29284) has a snout-vent length
of 180, the largest female (ASFS V29286)
147; both are from near Otra Banda. Snout
scales at the level of the second canthals
range between 2 and 4; the mode is 2 (32
specimens). The vertical loreal rows vary
between 5 and 8, with a mode of 7 (25
specimens). There are 1 to 4 scales be-
tween the supraorbital semicircles (mode
3). There are modally 5/5 scales between
the interparietal and the supraorbital semi-
circles; 5 scales are involved in 49 percent
of the combinations; actual counts are 3/4
(2), 4/4 (14), 4/5 (14), 5/5 (17), 5/6
(11), 6/6 (1) and 4/6 (1). Vertical dor-
sals range between 12 and 20 (mean 15.4),
horizontal dorsals between 15 and 25
(18.8), and ventrals between 17 and 28
(21.1). Of 11 adult males, nine have the
nuchal crest scales very high and two have
them high. Of 16 adult females, nine have
these scales very high and seven have them
high. Body crest scales in males are high
in two lizards, moderate in eight, and low
in three; in females, the body crest scales
are high in two, moderate in eight, and low
in six. Fifty-three specimens have the sub-
oculars separated from the supralabials by
one row of scales, whereas in four lizards
( 7 percent ) these scales are in contact, and
in two lizards (3 percent) they are sepa-
rated by tv/o rows of scales.

In general, both sexes of A. b. scelestiis
show a pattern of about six or seven fine
crossbands that are often obscured by dor-
sal blotching. Colors are shades of greens,
with brighter green the base color and the
blotching tending toward darker shades.
The crossbands are lighter pastel shades of
green, and in some lizards the dorsal
ground color is olivaceous. Another vari-
ant, which is somewhat more prevalent in
females, is an olive green to dark green

HisPANioLAN Giant Angle • Schwaiiz

137

dorsum, flocked with pale green. Two fe-
males from near Higiiey showed still an-
other style of body pattern and color, with
the dorsal ground color cream with some
dark green to brownish green snuidges,
and the neck with alternating pale blue
and charcoal markings, the pale blue mark-
ings persisting onto the cheeks. In males
the upper surface of the head is brown,
and in females it is mixed brown and
green, with the snout and supraocular
scales deep green in some lizards. In fe-
males, the chin and throat are dark green,
marbled with yellow and pale green. The
dewlap is rather \'ariable; in males it has
been recorded as dark yellow or deep yel-
low to orange or dark orange, whereas in
females the dewlap varies from yellow to
dark orange with dark brown, olivaceous,
or charcoal streaking, marbling, or smudg-
ing. Although there are no color notes in
life, in the preserved lizards the eyeskin is
regularly pale gray, and I presume that in
life the eyeskin is set off from the rest of
the head color in some pigmental fashion.
Many specimens of both sexes have the
lower sides tigroid with "stripes" extend-
ing conspicuously onto the lateral sides of
the abdomen.

There are 34 juvenile and subadult A. b.
sceJestus, with snout-vent lengths between
46 (USNM 193989) and 94 (ASFS
V21699-700). Three juveniles (snout-vent
lengths 46-61) have umbilici still present.
This entire suite of young lizards shows a
remarkable diversity in dorsal pattern.
Even small specimens may be either uni-
color green (usually with a vertical nuchal
white crescent and a white subocular
spot ) , green with three or four yellow body
bands, or there may be many more bands
resulting from the inteiposition of pale
body bands between the primary pale
body bands. One specimen (ASFS \'29296;
snout-vent, length 70, male ) has both pale
body bands and interstitial pale blotching,
whereas another lizard (MCZ 125621;
snout-vent length 86, female) already
shows the adult pattern of several fine pale

crossbands on a green ground. The largest
subadults, however, (ASFS V21699-700;
snout-vent lengths 94, male and female)
are both presently unicolor and show no
indications of the adult body banding.
That a single juvenile may demonstrate a
pattern change is shown by the following
notes on ASFS V28757, a female with a
snout-vent length of 54: "Alive, emerald
green with about foiu' pale yellow cross-
bands on body; dead — seven narrow brown
body bands which are hollowed, and the
dorsal groimd color now pale yellow-
green." The dewlap in young males is or-
ange, in young females from dull yellow
streaked with charcoal to charcoal.

Comparisons. In color and pattern, A.
h. scelestus differs from all other subspe-
cies. No other named population has six
or seven narrow dorsal crossbands; even
multistnippus is much more conspicuously
banded than scelestus and lacks any sort
of dorsal blotching. A. h. scelestus is
known to intergrade with more northern
caeruleolatus and is presumed to meet li-
torisilva. In each case, there is no difficulty
distinguishing the adjacent forms chromat-
ically. A. h. caeruleolatus typically has (in
males) sky-blue blotches along the junc-
tion of the dorsal and ventral colors, and
is prominently crossbanded with three dor-
sal crossbands. A. h. litorisilva is a blotched
lizard, the dorsal colors much more drab
than those of scelestus, tending toward
browns and brownish greens. Perhaps
scelestus most closely resembles multi-
stnippus, but, although both are banded,
the bands in multistruppus are much finer
and much more numerous than the six or
seven pale dorsal crossbars in scelestus.

A. h. scelestus, with modally 2 snout
scales, differs from caeruleolatus, which
has 4 snout scales. In having 7 vertical lor-
eal rows, scelestus differs from caeruleola-
tus, which has 8 rows. In having 5/5 scales
between the interparietal and supraorbital
semicircles, scelestus differs from samanae,
multistruppus, and .mhlimis, all of which
have 4/4. A. h. scelestus has the lowest

138 Bulletin Museum of Comparative Zoology, Vol. 146, No. 2

mean of vertical dorsals ( 15.4 ) of all
named subspecies, being most closely ap-
proached by litorisilva (15.9). A. b. sceles-
tus males are larger than those of any other
subspecies ( 180 in scelestus, 158 in litori-
silva, which is second largest) and in fact
this subspecies exceeds all other Hispanio-
lan giant anoles in size, being most closely
approached by male A. r. ricordi, which
reach a snout-vent length of 160.

Discussion. I am uncertain that all
specimens included in scelestus should be
so associated. This is especially true of the
specimen from near Yamasa (ASFS
V28656); this is a juvenile male and its
taxonomic status remains somewhat in
doubt, since it is young. It is also possible
that specimens from Santo Domingo like-
wise are not identical with more eastern
lizards, although the two samples agree
fairly well.

A. b. scelestus and A. b. caeruleolatus in-
tergrade in the region of El Seibo Prov-
ince; I have examined the following mate-
rial from El Seibo which I consider
intergradient: 3.5 mi. (5.6 km) S Sabana
de la Mar (ASFS X7877); 2.1 mi. (3.4 km)
N El Valle (ASFS X7861-62); 3 km N El
Valle (ASFS V3157-58); 10.5 km N Hato
Mayor (ASFS V35329-30). This series
consists of three juveniles and four young
adults (with snout-vent lengths between
112 and 127). The single adult male
(ASFS X7877) was tannish gray in life
with darker brown blotches, a pale green
venter, and an orange dewlap. Two adult
females (ASFS X7861-62) were pale pea-
green with vertical gray bars, the upper
surface of the head grayish tan, venter
green, and dewlap grayish orange. The
lower jaw and throat were green mottled
with darker green. In general this series
seems closer to caeruleolatus than to scel-
estus, but the male lacks sky-blue ventro-
lateral markings. On the other hand, the
vertical gray bars, recorded for the female,
resemble the pattern of scelestus rather
than that of female caeruleolatus. It seems
likely that caeruleolatus and scelestus in-
tergrade in this region.

Remarks. Almost all ASFS scelestus
were secured while the lizards were asleep
at night. Typical situations are lowland
cacaotales and cafetales with their high
canopied shade-trees, along lowland
streams (as at Otra Banda and Yamasa),
and in woods associated with limestone
cliffs (east of Santo Domingo). The long
series from the Rio Cumayasa is from the
high riverine woods along that stream; re-
markably, we secured only juveniles and
subadults at this locality, despite three
nocturnal visits. One juvenile from this lo-
cality was taken on the exposed branch of
an Acacia tree along an open road. Per-
haps the most remarkable place whence A.
b. scelestus has been taken is the locality
northeast of La Romana. This place is a
deep and well-wooded ravine through
which flows a clear stream; however, the
ravine is completely surrounded on all
sides by cane fields, and the ravine woods
are completely isolated at the ravine rim
from other such ecologies, if they even still
exist in this area. A. b. scelestus was ex-
ceptionally abundant in this particular and
very restricted strip of riverine gallery for-
est. Elevations above ground recorded for
sleeping scelestus range from 2 to 20 feet
(0.6 to 6.1 meters), with juveniles usually
sleeping much lower than adults. The alti-
tudinal distribution of A. b. scelestus is in
general low, with recorded elevations from
sea level to 200 feet. It is likely that this
subspecies also occurs in the uplands of
the Cordillera Oriental, but as yet there
are no specimens from areas within that
rather low-lying but mesic and well-for-
ested massif.

The name scelestus is from the Latin for
"unlucky, wretched," in allusion to the dif-
ficulties involved with collecting this sub-
species at the La Romana ravine noted
above.

The transition betu'cen scelestus and //-
torisilva must be very abrupt; the two sub-
species are known from localities separated
by only 28 kilometers. The habitats of the
two subspecies are quite different, with
scelestus inhabiting very mesic situations

HisPANiOLAN Giant Angle • Schwartz 139

and litorisilva xeiic coastal woods. Inter-
estingly, this same eastern region of the
Repiiblica Dominicana is also an area of
abrupt changes in subspecies of Anolis di-
stichus, where the subspecies ifi,ni<iularis
Mertens and properus Schwartz have
ranges which coincide rather closely with
those of scelestus and litorisilva (see
Schwartz, 1968: 275, map). The question
of intcrgradation between scelestus and the
southwestern subspcx-ies next to be named
will be discussed under that taxon.

Perhaps more so than any other subspe-
cies, scelestus seems to show a very spotty
distribution. Two instances are worthy of
mention. There are excellent extensive
coastal forests at Cabo Caucedo south of
the Aeropuerto Internacional de las Ame-
ricas on the southern Dominican coast. Re-
peated diurnal and nocturnal visits to these
splendid woods yielded no A. baleatus, de-
spite what seems to be more than adequate
habitat. A second locality, east of Boca
Chica along the same coast, likewise sup-
ports extensive fine stands of lowland hard-
wood forests, and there also, despite many
diurnal and nocturnal visits, we have never
encountered A. baleatus. It is possible that
these two instances of fairly dry coastal
woods are not suitable for scelestus
(whereas they surely would be for litori-
silva) and that scelestus simply does not
occur there.

Anolis baleatus fraudator
new subspecies

Holotype. USNM 193978, an adult fe-
male, from 4 km W, 6 km N Azua, Azua
Province, Repiiblica Dominicana, one of
two taken by Richard Thomas, on 23 July
1969. Original number V21384.

Paratypes. ASFS V21385, same data as
holotype; ASFS V21433, Barreras, Azua
Province, Repiiblica Dominicana, native
collector, 25 July 1969; ASFS V723, 1.1 mi.
(1.8 km) S San Jose de Ocoa, 1400 feet
(427 meters), Peravia Province, Repiiblica
Dominicana, R. F. Klinikowski, 24 August
1958; ASFS V21203, Sierra Martin Garcia,
about 3000 feet (915 meters), above Bar-

reras, Azua Province, Repiiblica Domini-
cana, R. Thomas, 20 July 1969; ASFS
V31207, Sierra Martin Garcia, above Bar-
reras, between 2000 and 2800 feet (610
and 854 meters), west slope, Mt. Biisi'i,
Barahona Province, Repiiblica Dominicana,
B. R. Sheplan, 15-17 September 1971.'

Definition. A subspecies of A. baleatus
characterized by the combination of mod-
ally 4 scales at level of the second canthal
scales, 6 vertical rows of loreal scales, 2 or
3 scales between the supraorbital semicir-
cles, 4/4 scales between the interparietal
and the supraorbital semicircles, high num-
ber of vertical dorsal scales (17-21; mean
18.8), low number of ventral scales (18-
26; mean 20.7), nuchal scales high, body
crest scales moderate in only adult female,
subocular scales usually separated from
supralabial scales by one row of scales;
dorsum (in female) mottled pale and
darker gray, with three irregular white
crossbands, and blotched with yellow-
green, top of snout and lores straw, labials
dull yellow, and dewlap nearly white with
a yellowish or cream wash.

Distribution. The Sierra Martin Garcia
in Barahona and Azua provinces, and
along the southern slopes of the Cordillera
Central and the Sierra de Ocoa in Azua
and Peravia provinces.

Description of holotype. An adult fe-
male with a snout-vent length of 133 and
tail length of 244; snout scales between
second canthals 4; 6 vertical rows of loreal
scales, 2 scales between supraorbital semi-

1 Since the present manuscript was completed,
a juvenile female ( MCZ 132301) with a snout-
vent length of 57 mm, was secured by E. E. Wil-
liams and J. Roughgarden at a locality south of
La Honna, Peravia Province, on 19 July 1972.
This lizard is to be considered a paratype. It has
3 snout scales at the level of the second canthals,
6 loreal rows, 3 scales between the supraorbital
semicircles, 5/5 scales between the interparietal
and the semicircles, 16 vertical rows of dorsal
scales and 20 rows of ventral scales, and 1 scale
between the suboculars and the supralabials. Both
nuchal and body crest scales are low. As pre-
served, the lizard is dull greenish with indications
of dark dorsal crossbars, and it lacks any pale
dorsal markings.

140 Bulletin Museum of Comparative Zoology, Vol. 146, No. 2

circles, 4/3 scales between the interpari-
etal and the supraorbital semicircles, verti-
cal dorsals 17, horizontal dorsals 24,
ventrals 21, subocular scales separated
from supralabial scales by one row of
scales, fourth toe lamellae on phalanges II
and III 33, nuchal crest scales high, body
crest scales moderate; in life, dorsum mot-
tled pale and dark gray, blotched with yel-
low-green and with three irregular white
crossbands, labials dull yellow, top of snout
and lores straw, and dewlap nearly white
with a yellowish or cream wash.

Variation. The only adult is the holo-
type; the remainder of the paratypic series
is composed of juveniles and subadults
with snout-vent lengths between 74 and 96
(three males, two females). Snout scales
at the level of the second canthals range
between 2 and 4; the mode is 4 ( four spec-
imens). The vertical loreal rows vary be-
tween 5 and 7, with a mode of 6 (three
specimens). There are 2 or 3 scales be-
tween the supraorbital semicircles; both
categories have the same frequency. There
are modally 4/4 scales between the inter-
parietal and the supraorbital semicircles; 4
scales are involved in 67 percent of the
combinations; actual counts are 3/4 (2),
4/4 (3), and 5/6 (1). Vertical dorsals
range between 17 and 21 (mean 18.8),
horizontal dorsals between 20 and 24
(21.4), and ventrals between 18 and 26
(20.7). The only adult specimen (a fe-
male) has the nuchal crest scales high and
the dorsal body crest scales moderate. Five
specimens have the suboculars separated
from the supralabials by one row of scales
and one lizard has these scales in contact
( 17 percent ) .

The details of the color and pattern of
the only adult, the female holotype, have
already been given. The juveniles and
subadults show the same general pattern
configuration as does the adult. The small-
est juvenile (snout-vent length 72), a fe-
male topotype, was gray with yellowish
mottling and a pat*"ern of three irregular
crossbands, a faint white scapular stripe,
and black postauricular and postorbital

spots. The dewlap was charcoal with
white scales. The next largest individual
(snout-vent length 74), a male, had the
dorsum pale green with irregular trans-
verse barring; the upper surfaces of the
limbs were pale green and gray-green, and
the tail was banded pale green and gray-
green. The venter was whitish. The dew-
lap was very dark yellowish with an or-
ange wash posteriorly. A slightly larger
male (snout-vent length 85) was pale
green, much marbled and shaded with tan
to gray and with some faint evidence of
transverse crossbands; the chin and throat
were gray-green, and the dewlap pale gray-
ish orange. A female from the Sierra Mar-
tin Garcia (snout-vent length 88) was
green and brown dorsally and without pale
markings; the dewlap was marbled with
charcoal. Finally, the largest subadult
(snout-vent length 76), a male, had the
dewlap dirty yellow with orange streaking.
In the case of fraudator, the very pale ( al-
most white) adult fem.ale dewlap appears
to be preceded ontogenetically by brighter
and more typically A. haleatus hues.

Comparisons. No other subspecies of A.
haleatus approaches the pale colors of
fraudator, nor does any other subspecies
have such a pale dewlap. Although frau-
dator combines the blotching and trans-
verse crossbands in the same fashion as
does scelestus, fraudator is in all ways a
paler lizard. Comparisons in details of
color and pattern with all other subspecies
of A. haleatus are unnecessary. A. h. frau-
dator differs from samaime, scelestus, multi-
struppus, and suhliinis in having 4 rather
than 2 snout scales at the second canthals,
and only fraudator has a mode of 6 verti-
cal rows of loreals (7 or 8 in all other sub-
species). In having 4/4 scales between the
inteiparietal and the supraorbital semicir-
cles, fraudator differs from caeruleolatus,
scelestus, and haleatus, all of which have
5/5. Although fraudator has a high mean
(18.8) of vertical dorsals, in which it is
exceeded only by suhlimis (mean 19.2),
fraudator has the lowest mean (20.7) of

HisPANiOLAN Giant Angle • Schwartz

141

ventrals of all subspecies, being approached
most closely by scelestus (21.1).

Discussion. Apparently A. h. fraudator
is a pale subspecies that is restricted to fa-
vored situations in the xeric regions asso-
ciated with the Llanos de Azua along the
southern slopes of the Cordillera Central
and the Sierra de Ocoa, a southern affiliate
of the former range. The subspecies ap-
parently also occurs in the Sierra Martin
Garcia, an eastern isolate of the Sierra de
Neiba (which, it will be recalled, is else-
where occupied by A. r. ricordi) and sur-
rounded by extreme desert. The specimen
from Barreras, which lies at the foot of the
Sierra Martin Garcia, is interesting in that
it seems a most unlikely locality for any
giant anole; however, I assume that the
specimen, which was native-collected, was
taken either in nearby Cocos groves or on
the lower wooded slopes of the range it-
self. Two specimens from the higher eleva-
tions of the Martin Garcia are from dense
woods, and the specimen from San Jose
de Ocoa was taken from a large tree at the
edge of a pasture. The type locality is
semi-xeric woods with vine tangles and
mango trees in an otherwise cultivated but
xeric region. Probably A. b. fraudutor is
widely distributed in suitable situations
through much of this region, but the lizard
appears to be rare; Buffett and I collected
in semi-mesic riverine woods at a locality
4 km W and 17 km N Azua at an elevation
of about 500 feet ( 153 meters ) , both dur-
ing the day and at night, without seeing
any giant anoles. Natives just south of San
Jose de Ocoa at an elevation of 1400 feet
(427 meters) did not secure specimens for
us in semi-mesic woodlands. Since the al-
titudinal distribution of fraudator extends
from sea level to about 3000 feet (915 me-
ters) in the Sierra Martin Garcia, the ele-
vations of the above-mentioned localities
are within the known altitudinal range of
the subspecies, and indeed our San Jose de
Ocoa locality was quite close to where
Klinikowski secured one of the paratypes.

Remarks. The name fraudator is from
the Latin for "deceiver" in reference to the

resemblances between this sui)species and
A. harahonae. In fact, my decision to re-
gard fraudator as a subspecies of baleatus
rather than harahonae is based more upon
the juveniles than the adults of fraudator;
this is not exclusively due to the fact that
there are more juveniles of fraudator than
adults but rather that the patterns shown
by juvenile fraudator are more typically
those of A. Jjaleatus than of A. harahonae.
A. J), harahomie and A. h. fraudator are
alike in modal numbers of scales at the
level of the second canthals (4), vertical
loreal rows (6), and scales between the in-
terparietal and the supraorbital semicircles
(4/4), and they do not differ strikingly in
means of body scales ( 17.2, 18.8 in vertical
dorsals; 18.2, 21.4 in horizontal dorsals;
22.1, 20.7 in ventrals). In these means,
harahonae is lower in dorsal body counts,
but higher in ventral coimts. The moder-
ate nuchal crest scales of fraudator occur
also in harahonae, but most female hara-
honae have these scales low. No female
harahonae has moderate dorsal body crest
scales as does the female fraudator,
whereas moderate body crest scales occur
in females of most subspecies of A. halea-
ttis ( only female suhhniis lack them ) . Tak-
ing all evidence into consideration, I have
elected to consider fraudator a subspecies
of A. ])aleatus, but its resemblances to A.
harahonae are acknowledged. The distance
separating these two species in this area
is only 20 kilometers (see introduction),
and it is not unlikely that A. harahomie has
been derived from fraudator across the
strait that is now the Valle de Neiba (see
discussion). On the other hand, A. h.
fraudator is removed by some 60 kilometers
from the nearest A. ricordi locality in the
nearby Sierra de Neiba. There is no ques-
tion that fraudator is not correctly associ-
ated nomenclatorially with A. ricordi.

The apparent geographic isolation of
fraudator in relation to other subspecies of
A. haleatus is probably artificial. The near-
est records for other subspecies are: cae-
ruleolatus — 38 kilometers ( San Jose de
Ocoa and La Cumbre); scelestus — 55 kilo-

142 Bulletin Museum of Comparaiive Zoology, Vol. 146, No. 2

meters (San Jose de Ocoa and Yamasa);
and sublimis — 50 kilometers (San Jose de
Ocoa and south of EI Rio). There are
suitable habitats for giant anoles between
caertileolatus, scelestus, and fraudator, but
specimens are lacking. The intervening
high Cordillera Central between the ranges
of sublimis and fraudator probably acts as
a barrier to prevent contact between these
two subspecies.

DISCUSSION

My decision to consider Anolis ricordi
as three species rather than one has some
precedent in the Schwartz and Garrido
(1972) treatment of the Cuban Anolis
equestris, wherein that species was divided
into five species. However, the two situa-
tions, although comparable, are far from
identical. In the A. equestris complex,
there are at least a few incidences of sym-
patry between members of the species-
complex which give clues to the facts of
the situation; there are sti'ong differences
in size of dorsal scales; there are some
strong differences between details of pat-
tern and coloration of the axillary stripe
and the dewlap which likewise suggest
that we are there dealing with more than
one species. But on the other hand, the
Hispaniolan giant anoles show absolute
differences in the nuchal and body crest
scales and differences in the pattern of the
body itself, as well as modal differences in
scutellar details. In addition, there are no
cases as yet known in Hispaniola of sym-
patry between the three entities that I re-
gard as full species. The gaps between
them are narrow, however, and I feel
strongly that it is merely a matter of get-
ting into the intermediate areas and, once
there, being fortunate enough to encounter
giant anoles.

It should be obvious from my systematic
treatment that I am convinced that we are
dealing in Hispaniola with three distinct
species — ricordi, barahonae, and baleatus.
Surely the differences between ricordi and
baleatus are such that, when taken in sum,
one has no doubts that he is involved with

two very different animals. The differences
here are much greater, for instance, than
between Anolis distichus Cope and Anolis
hrevirostris Bocourt, two species that were
long confused and that resemble each
other moiphologically to a very great de-
gree. Yet once one learns what the char-
acters are for separating them, he experi-
ences little difficulty in dealing with both
populations or individuals, either alio- or
sympatric, of these two species. The dif-
ferences in life, as far as pattern and color
are concerned, are not particularly subtle,
and the details of scutellation are not di-
chotomous, but the modal differences are
so well correlated with the pigmental and
pattern traits that we now recognize these
two species with assurance.

An even more obvious parallel is Anolis
carolinensis Voigt and Anolis allisoni Bar-
bour in Cuba. These two species of green
anoles, long confused as A. porcatus Gray
(or A. c. porcatus), were shown by Ruibal
and Williams (1961) to be a sibling pair,
fairly allopatric but both widely distrib-
uted throughout much of Cuba, and to
differ structurally by the condition of the
postauricular area. The presence (allisoni)
or absence (carolinensis) of a deep and
elongate postauricular groove in these two
species is correlated with very striking dif-
ferences in adult pattern and coloration
and other details of scutellation.

The same situation, that of two species
masquerading under a single name, can
also be demonstrated in Anolis alutaceus
Cope and Anolis clivicola Barbour and
Shreve (Schwartz and Garrido, 1971), and
the two species recently confused under
Anolis spectrum Peters; both these situa-
tions pertain to Cuban species. Sr. Gar-
rido also advises me that he has much
evidence to indicate that Anolis cyanopleu-
rus cupeyalensis Peters is in fact a sym-
patric sibling, rather than a subspecies, of
A. cyanopleurus Cope.

I could cite other examples in Antillean
iguanids (Leiocephalus) and anguids
(Diploglossus) which demonstrate quite
clearly the above trend. As more material

HisPAXioLAN Giant Angle • Schwartz

143

from more diverse localities becomes avail-
able, and as this material is subjected to
re-evaluation with differing and more mod-
ern philosophies, our impressions of rela-
tionships among Antillean anolines have
been modified or changed. A major factor
in such revisions has invariably been a
great quantity of new material from areas
that had previously been unsampled, cou-
pled with pigmental, ecological, and etho-
logical data from the living specimens. A
second general line of evidence, equal to
or possibly suipassing morphological and
distributional data in importance, is karyo-
typic and electrophoretic information. One
or both of these areas of investigation are
increasing our knowledge of the complex-
ities within such a genus as Anolis. When
these two areas of research — morphologi-
cal and biochemical — can be brought to
bear simultaneously upon a single species
or species complex, the results may be even
more meaningful than either is alone. As
yet this has not been done in any of the
Antillean giant anoles, so that my conclu-
sions, based upon morphology and distri-
bution, remain to be verified by other evi-
dence. Yet I feel as secure as any
systematist can be when he is dealing with
data that are incomplete.

As pointed out in the introduction to the
present paper, the taxa ricordi, haleatus,
harahonae, and leberi are, on inspection,
unequivocally distinct. But the degree or
level of differentiation of these four taxa
seems to be two-fold. On one hand (ri-
cordi and haleatus), the two populations
are easily separable on the basis of a struc-
tural feature (the nuchal crest scales), a
character that is strongly correlated with
obvious pigmental and pattern traits. On
the other hand, the differences between
harahonae or leheri and ricordi are pri-
marily ones of pigmentation and pattern,
with morphological differences much less
trenchant than between ricordi and ha-
leatus. At the outset such a dichotomy sug-
gests that it might be more proper to con-
sider "A. ricordi" as a complex of full
species than as one species with four (or

more) subspecies. Appar(Mitly Williams
and Rand (1969) had the same inclina-
tions, since they indicated that the differ-
ences between some of the then-named
populations of A. ricordi were such as to
suggest that there might be more than one
species involved.

Once the above assumption has been
made — namely, that A. ricordi is composed
of more than one species — then the prob-
lem first becomes one of differentiating and
delimiting the component species. There
is no difficulty here in separating A. ricordi
and A. haleatus on the basis of crest scales.
None of the populations of A. haleatus has
the moderate (rarely) to low (usually)
nuchal crest scales of A. r. ricordi. In ad-
dition, the narrow geographical gaps that
exist between A. ricordi and A. haleatus
also suggest that these two taxa may be
either allopatric or may meet and occur
sympatrically without intergradation.

The status of the Tiburon populations
that I associate nomenclatorially with A.
ricordi and that of A. harahonae as a dis-
tinct species are less clear than the ricordi-
haleatus relationship. First, the named
populations leheri, viculus, and suhsolamis
have in common a suite of pattern and
color features that ally them more closely
to each other than to A. r. ricordi. The
only evidence for this relationship is the
occurrence of presumed viculus X ricordi
intergrades in the Miragoane-Paillant re-
gion. Were it not for these specimens, I
would be strongly tempted to consider the
three Tiburon taxa as comprising a sepa-
rate species. Any interpretation of the re-
lationships of the Tiburon taxa suffers
from paucity of material from a variety of
localities.

The situation with A. harahomie is in
some ways puzzling. Although there is no
question that it is distinct from A. ricordi,
its relationships to A. haleatus are much
less certain. This uncertainty is caused by
A. h. fraudator, that population assigned to
A. haleatus which is closest geographically
to A. harahonae. It is particularly unfortu-
nate that fraudator is known from only

144 Bulletin Museum of Coinparative Zoologij, Yo]. 146, No. 2

one adult and several juvenile and sub-
adult specimens, since adult males ( primar-
ily) would be most instructive in compar-
ing frattdator with barahonae. On the
other hand, the closeness of fraiidator and
barahonae in characteristics may be rather
a reflection of the ancestry of A. bara-
honae— namely, that it is a south island
(sensu Williams, 1961) invader from the
north, and that the parent population has
been fraudator rather than any other sub-
species of A. baleatus or A. ricordi from the
west.

It might be more proper either to con-
sider A. barahonae as conspecific with A.
baleatus (the two taxa linked through
fraudator) , or to consider fraudator a sub-
species of A. barahonae; either interpreta-
tion has merit. The course that I have
taken seems satisfactory at the moment but
surely is subject to reinterpretation with
the acquisition of more material from this
critical geographic area.

The history of the Hispaniolan giant
anoles appears to be correlated with the
two palaeo-islands that have been fused at
the level of the Cul de Sac-Valle de Neiba
plain with lowering Pleistocene sea levels.
I suggest the following history for the com-
plex; the reader should keep in mind that
such a history is based upon taxonomic
premises that are inductive, and the cau-
tions and uncertainties that I expressed
above have special application here.

Distributional evidence suggests that the
giant Hispaniolan anole stock was origi-
nally restricted to the north island (north
of the Cul de Sac-Valle de Neiba plain).
In this region, two distinctive species arose,
ricordi in the west and baleatus in the east.
There apparently has been local differen-
tiation on the north island at a subspecific
level in both these species, but that in ri-
cordi remains unanalyzed because of too
few specimens. On the other hand, differ-
entiation in A. baleatus is now fairly well
known and documented. This species oc-
curs east of the Cordillera Central and on
the southern slopes of that range and in
the Sierra Martin Garcia. There have been

local population differentiations in response
to the various ecologies within the area oc-
cupied, with two major integumental
trends (coloration and pattern) and details
of scutellation of the head and body (al-
though the latter is not so clear as the for-
mer ) .

There seem to have been two subse-
quent invasions of the south island. To the
west, a (presumably) early invasion of the
A. ricordi stock crossed what is now the
Cul de Sac Plain into the Port-au-Prince
area. It is pertinent that many north island
species have made this same crossing and
have extended their ranges but little fur-
ther. These species with more restricted
ranges have been handicapped either by
competition with already established spe-
cies, improper ecological situations, or rela-
tively recent arrival. A. ricordi seems to
have been an early arrival, without local
competitors, and with abundant proper
ecology (mesic forests). The species has
thus expanded its range after the original
crossing to cover the entire Tiburon Penin-
sula, having somewhere succeeded in cross-
ing the mountainous spine of the Massif de
la Hotte-Massif de la Selle. Local differ-
entiation along the Tiburon in response to
lack of genetic contact across the interior
mountains has also taken place. Further
speculations on details of the history of A.
ricordi on the Tiburon Peninsula are point-
less, since the specimens upon which any
generalisations may be made do not as yet
exist in collections.

A second invasion to the east occurred
presumably at a later date, after the estab-
lishment of A. ricordi on the Tiburon Pen-
insula. This latter invasion resulted in the
differentiation of A. barahonae (from a
fraudator or pre-fraudator stock on the
southern portion of the north island ) in the
Sierra de Baoruco and its subsequent ex-
pansion onto the southern portion of the
Peninsula de Barahona and east along the
southern slopes of the Sierra de Baoruco.
With the previous establishment of A. ri-
cordi to the west (as at Thiotte), the
western movement of A. barahonae was

HisPANiOLAN Giant Angle • Schwartz

145

halted by the presence of the related spe-
cies. I have no doubt that both A. ricordi
and A. barahonae will be found to be
closely allopatric or synipatric in extreme
southeastern Haiti between the Dominico-
Haitian border and Saltrou, and also that
these two species meet and interact along
the northern slopes of the Sierra de Bao-
ruco and the Morne des Enfants Perdus.

One other distributional detail requires
comment. The occurrence of A. r. ricordi
in the main mass of the Sierra de Neiba on
the northern side of the Valle de Neiba
and of A. h. fraudator in the Sierra Martin
Garcia, an extreme eastern isolate of the
Sierra de Neiba, has already been noted.
The Martin Garcia seems to have been
long isolated from not only the Sierra de
Neiba but also from all other Hispaniolan
mountain masses; it is ideally a montane
island in a sea of desert. It seems likely
that this range was unoccupied by giant
anoles of either species (A. ricordi or A.
baleatus), despite the fact that the range
forms a portion of the Neiba uplift. Inva-
sion of the Martin Garcia was possible
from either the northwest (ricordi) or the
northeast (baleatus). Of the two species,
A. baleatus was the more vagile and
reached the Sierra Martin Garcia from the
relatively more mesic southern slopes of
the Cordillera Central before A. ricordi
reached it across the deserts and xeric hills
between the Sierra de Neiba and the Mar-
tin Garcia. This upland population in
turn was responsible for the invasion of the
Sierra de Baoruco across the better for-
ested and more mesic eastern end of the
Valle de Neiba.

Wetmore and Swales (1931: 235) re-
ported the finding of recent Anolis ricordi
skeletal material in Barn Owl (Tyto alba)
pellets from L'Acul, Dept. du Sud, Haiti,
on the Tiburon Peninsula, and Hecht
(1951: 23) noted the abimdant remains of
the species from deposits in "Deep Cave,"
near St. Michel de I'Atalaye, Dept. de I'Ar-
tibonite, Haiti. Etheridge (1965: 101) re-
ported A. ricordi remains from recent owl
pellets near the mouth of a cave near Boca

de Yuma, La Altagracia Province, Repu-
blica Dominicana. Etheridge (op. cit.: 87-
88) also noted pre-Columbian giant anole
remains from a cave at Cerro de San Fran-
cisco near Pedro Santana, La Estrelleta
Province, Repiiblica Dominicana. From the
suite of about 80 cranial elements and
eight pelves, Etheridge extrapolated that
the maximally sized individuals in the cave
deposits had a snout-vent length of 190-
192 mm, some 30 mm larger than any liv-
ing A. ricordi recorded ( 159 mm, fide Eth-
eridge, op. cit.: 88). The maximally sized
Hispaniolan giant anole recorded in the
present paper reaches a length of 180 mm
(male A. b. scelestus from Otra Banda, La
Altagracia Province, Republica Domini-
cana). The difference between this mod-
ern living lizard and the maximally sized
pre-Columbian lizards is not so great as
Etheridge's data suggest. Intriguingly, the
Cerro de San Francisco area lies within the
known range of A. r. ricordi, and the larg-
est specimens of this subspecies ( male with
a snout-vent length of 160 mm, female 151
mm) are from the southern slopes of the
Cordillera Central, veiy close to the Cerro
de San Francisco area. Although there
seems to have been some change in maxi-
mum size in Hispaniolan giant anoles with
the passage of time, these changes have
not been of the magnitude that previous
data suggested.

LITERATURE CITED

Cochran, D. M. 1941. The herpetology of
Hispaniola. Bull. U.S. Natl. Mus., 177: 398,
120 figs., 12 pis.

Cope, E. D. 1864. Contributions to the herpe-
tology of tropical America. Proc. Acad. Nat.
Sci. Philadelphia, pp. 166-181.

DUMERIL, A. M. C, AND G. BiBRO.N. 1837.

Erpetologie gcnerale ou histoire naturelle
complete des reptiles, vol. 4. 571 pp., 14 pis.

Etheridge, R. E. 1965. Fossil lizards from the
Dominican Republic. Quart. Jour. Florida
Acad. Sci., 28(1): 83-195, 3 figs.

Garrido, O. H., AND A. Schwartz. 1968. Cu-
ban lizards of the genus Chamaeleolis. Quart.
Jour. Florida Acad. Sci., 30(3): 197-220, 2
figs.

Hecht, M. K. 1951. Fossil lizards of the West

146 Bulletin Museum of Comparative Zoology, Vol. 146, No. 2

Indian genus Aristelliger ( Gekkonidae ) .
Amer. Mus. Novitates, No. 1538: 1-33, 8 figs.

Maerz, a., and M. R. Paul. 1950. A Diction-
ary of Color. New York: McGraw-Hill Book
Co., pp. vii + 1-23, 137-108, 56 pis.

Mertens, R. 1939. Herpetologische Ergebnisse
einer Reise nach der Insel Hispaniola, West-
indien. Abh. senckenberg. naturf. Ges., 449:
1-84, 10 pis.

Rand, A. S., and E. E. Williams. 1969. The
anoles of La Palma; aspects of their ecolog-
ical relationships. Breviora, Mus. Comp.
Zool., No. 327: 1-18, 1 fig.

RuiBAL, R., AND E. E. Williams. 1961. Two
sympatric Cuban anoles of the carolinensis
group. Bull. Mus. Comp. Zool., 125(7):
183-208, 11 figs.

Schmidt, K. P. 1921. Notes on the herpetology
of Santo Domingo. Bull. Amer. Mus. Nat.
Hist., 44(11): 7-20, 12 figs.

Schwartz, A. 1964. Anolis equestris in Ori-
ente Province, Cuba. Bull. Mus. Comp.
Zool., 131(12): 407-428, 7 figs.

. 1968. Geographic variation in Anolis

distichus Cope ( Lacertilia, Iguanidae ) in the
Bahama Islands and Hispaniola. Bull. Mus.
Comp. Zool., 137(2): 255-309, 4 figs., 2 pis.

, and O. H. Garrido. 1971. The status

of Anolis alutaceus clivicolus Barbour and

Shreve. Caribbean Jour. Sci., 11(1-2): 11-
15.

. 1972. The lizards of the

AND

Anolis equestris complex in Cuba. Stud.
Fauna Curasao and Caribbean Is., 39(134):
1-86, 8 figs.

Thomas, R. 1971. A new species of Diploglos-
sus (Sauria: Anguidae) from Hispaniola.
Occ. Papers Mus. Zool., Louisiana State Univ.,
40: 1-9, 4 figs.

Wetmore, A., and B. H. Swales. 1931. The
birds of Haiti and the Dominican Republic.
Bull. U.S. Natl. Mus., 155: 1-483, 2 figs., 26
pis.

Williams, E. E. 1961. The evolution and rela-
tionships of the Anolis semilineatus group.
Breviora, Mus. Comp. Zool., No. 138: 1-8, 1
pi.

. 1962. Notes on Hispaniolan herpetol-
ogy. 6. The giant anoles. Breviora, Mus.
Comp. Zool., No. 155: 1-15, 1 fig.

. 1965. Hispaniolan giant anoles (Sau-
ria, Iguanidae ) : new data and a new subspe-
cies. Breviora, Mus. Comp. Zool., No. 232:
1-7, 2 figs.

, and a. S. Rand. 1969. Anolis insolitus, a

new dwarf anole of zoogeographic importance
from the mountains of the Dominican Repub-
lic. Breviora, Mus. Comp. Zool., No. 326:
1-21, 6 figs.

us ISSN 0027-4100

BulLetln OF TH

seum

Comparative
Zoology

A Revision of the Cardinalfish Genus

Epigonus (Perciformes, Apogonidae)^

with Descriptions of Two New Species

GARRY F. MAYER

HARVARD UNIVERSITY

CAMBRIDGE, MASSACHUSETTS, U.S.A.

VOLUME 146, NUMBER 3
19 SEPTEMBER 1974

PUBLICATIONS ISSUED

OR DISTRIBUTED BY THE

MUSEUM OF COMPARATIVE ZOOLOGY

HARVARD UNIVERSITY

Breviora. 1952-
BULLETIN 1863-

Memoirs 1864-1938

JoHNsoNiA, Department of Mollusks, 1941-

OccAsioNAL Papers on Mollusks, 1945-

SPECIAL PUBLICATIONS.

1. Whittington, H. B., and E. D. I. Rolfe (eds.), 1963. Phylogeny and
Evolution of Crustacea. 192 pp.

2. Turner, R. D., 1966. A Survey and Illustrated Catalogue of the Teredini-
dae (Mollusca: Bivalvia). 265 pp.

3. Sprinkle, J., 1973. Morphology and Evolution of Blastozoan Echinoderms.
284 pp.

4. Eaton, R. J. E., 1974. A Flora of Concord. 211 pp.

Other Publications.

Bigelow, H. B., and W. C. Schroeder, 1953. Fishes of the Gulf of Maine.
Reprint.

Brues, C. T., A. L. Melander, and F. M. Carpenter, 1954. Classification of
Insects.

Creighton, W. S., 1950. The Ants of North America. Reprint.

Lyman, C. P., and A. R. Dawe (eds.), 1960. Symposium on Natural
Mammalian Hibernation.

Peters' Check-list of Birds of the World, vols. 2-7, 9, 10, 12-15.

Proceedings of the New England Zoological Club 1899-1948. (Complete
sets only.)

Publications of the Boston Society of Natural History.

Price list and catalog of MCZ publications may be obtained from Publications
Office, Museum of Comparative Zoology, Harvard University, Cambridge, Massa-
chusetts, 02138, U.S.A.

A REVISION OF THE CARDINALFISH GENUS EPIGONUS
(PERCIFORMES, APOGONIDAE), WITH DESCRIPTIONS
OF TWO NEW SPECIES^

GARRY F. MAYERS

TABLE OF CONTENTS resim-ected and redescribed on the basis of new

. ,^Y material, and Htjnnodus atherinoides Gilbert and

:^ j*^ " , ,« H. megalops Smith and Radcliffe are synonymized

Introduction 14^ ^^^^^j^ ^ occidentaUs Goode and Bean.

Metnods Species descriptions include discussions of dis-

Systematics """r " ici tribution, geographic variation, ontogenetic

Genus Epigonus Rafmesque 151 ^^^^^^^ a„j taxonomic problems. An in%'estiga-

Diagnosis 151 ^^^^ ^^ ^j^^ ^^p^^ ^^j ^ lenimen (Whitley) reveals

Description 151 ^^^^^ ^^ holotype and paratypes are not conspe-

Key to the Species of Epigonus 152 ^.jjj^ Instead, the paratypes are members of E.

Epigonus telescopus 152 denticulatus Dieuzeide. A key to the species of

Epigomis macrops 159 Epigonus is provided at the beginning of the

E))igonus ))andionis 163 paper.

Epigonus fragilis 169

Epigonus occidentaUs 170 INTRODUCTION

Epigonus denticulatus 1^5

Epigonus oligolepis 179 Selected species of Epigomis have been

Epigonus tretvavasae 183 j^^^^^^^^^ 1^ ^^^.^^^ f^^ at least two hundred

Epigonus pectimfcr loo J , , onr, r>^\ i j

Epigonus wbustus 1S9 ti^ty Y^ars. Vaillant (1888: 25) remarked

Epigonus lenimen 193 that E. telescopus was recognized in ancient

Epigonus crassicaudus 197 times, and Risso (1810: 303) reported that

Species Incertae Hedis 199 j-j-jj^ gpecies was prized for its firm, deli-

Aek„S^!Sr°!!!' ::;:::::::::::::::::::::::::: loE! cious-tasti„g „,«,, although it was rardy

Literature Cited 200 taken. The presence of common names tor

Appendix 203 E. telescopus in vocabularies of western

Mediterranean and North Atlantic fishing

Abstr.\ct. a study of the deep-sea Apogonidae communities ( Doderlein, 1889 ) provides
results in a revision of the genus Epigonus additional evidence of man's long-term
Rafinesque. Twelve species are recognized in- awareness of the species. £. telescopus is
eluding two new forms — E. oligolepis and £...,, • n i i • . i i i. „r
pectirUfer. E. fragilis (Jordan and Jordan) is still occasionally sold in the markets of
southwestern Europe.

' This paper is based on a portion of a thesis Two other species of EpigOnus are cap-
presented to Harvard University in partial ful- tured by commercial fishermen. E. denti-
fillment of the requirements for the Ph.D. in culcitUS is edible ( Dieuzeide et al., 1953:

^i?'"g>- ^ r -1^) and is taken in the Mediterranean.

- Department of Marine Science, University of y-r .., .i ,i . ,■ ^^t■^^■ >., f^^

c u \ri J c^ u . I 171 • 1 oo-m 1 Uutil recently this form was mistaken tor

South Florida, St. Petersburg, Florida 33/01 and ^■'^_, , _, , ,

Museum of Comparative Zoology, Harvard Uni- the yOUng of E. teUsCOpUS. E. crassicaudus

\ersity, Cambridge, Massachusetts 02138. is caught by Chilean fishermen. Like E.

Bull. Mus. Comp. Zool., 146(3): 147-203, September, 1974 147

148 Bulletin Museum of Comparative Zoology, Vol. 146, No. 3

telescopus, it is not taken in sufficient
numbers to support a separate fishery but is
captured by fishermen trawhng for more
abundant deepwater organisms.

Although African Epigonus are not pres-
ently exploited, tropical eastern Atlantic
stocks may represent future sources of pro-
tein for mankind. Surveys sponsored by
the Organisation of African Unity and the
U.S. Agency for International Development
revealed these fishes are "of possible po-
tential importance ( not necessarily by pres-
ent marketing standards) [Williams, 1968:
79]." The same may be true for Caribbean
and Gulf of Mexico Epigonus; however,
complete data have not been compiled for
the latter areas.

A major hindrance to the evaluation of
deep-sea cardinalfish stocks has been taxo-
nomic confusion. The systematic history of
Epigonus began in 1810 with Risso's de-
scription of Pomatomus telescopus and
Rafinesque's account of its synonym Epi-
gonus niacrophthahnus. During the fol-
lowing seventy-one years, work on the
genus was primarily limited to re-descrip-
tions of E. telescopus and discussions of its
biology (e.g., Cuvier, 1828; Valenciennes,
1830; Capello, 1868; Moreau, 1881).

The surge in oceanographic exploration
during the last quarter of the nineteenth
and beginning of the twentieth centuries
rapidly increased the number of nominal
Epigonus AAke species. Among the forms
described between 1881 and 1920 were
Apogon pandionis Goode and Bean, 1881;
E. occidentalis Goode and Bean, 1896;
Hynnodus atherinoides Gilbert, 1905; Oxij-
odon macrops Brauer, 1906; and Hynnodus
me galops Smith and Radcliffe, 1912. In
the following decade, three new species
and two new genera appeared in the liter-
ature.

Much of the confusion associated with
the taxonomy of Epigonus stems from ma-
terial described prior to 1930. Early taxa
were based on small samples. Because
many nations participated in oceanographic
research, specimens were deposited in

scattered institutions and descriptions ap-
peared in diverse publications. Conse-
quently, it was difficult for workers to
obtain either comparative material or a
broad overview of the group's systematics.
These shortcomings were aggravated by
inaccurate, under-illustrated descriptions
based on ill-considered characters. It was
common, for example, to use dentition
patterns to define generic boundaries, yet
tooth arrangements are difficult to observe,
easily damaged, and subject to ontogenetic
and geographic variation. As a result, an
inordinately large number of Epigonu.s-\ike
forms was recognized by the end of the
1920's.

Although generic taxonomy was stream-
lined by Fowler and Bean in 1930 and
Matsubara in 1936, species-level taxonomy
became increasingly complex. New forms
were described in 1935, 1950, 1954, and
1959. In addition, misidentifications of
Epigonus were published in several widely
circulated works on regional faunas (e.g..
Smith, 1949b and 1961; Gosline and Brock,
1960).

The aim of the present study is to clarify
the species-level systematics of the Epi-
gonus-\ike fishes. Data from traditional
characters are evaluated and augmented
by information from characters not pre-
viously examined for this group. A special
effort is made to discuss features such as
dentition patterns that caused taxonomic
confusion in the past. The ecology, func-
tional anatomy, zoogeography, and evolu-
tion of Epigonus will be discussed in future
works on the genus.

METHODS

Measurements were made to the nearest
tenth of a millimeter by tlie use of Helios
needlepoint dial calipers; characters larger
than 190 mm were measured with a meter
rule or GPM Anthropometer. Measure-
ments routinely taken include:

Standard length (SL) — from tip of snout
to base of caudal fin.

EriaoNus Systematics • Mayer 149

Head length (HL) — from tip of snout to
tip of opercular spine.

Body depth — between dorsal and v(Mitral
surfaces of body at level of peKic fin
base.

Head height — from quadratomandibular
joint vertically to bony rim above eye.

Eye diameter — between anterior and pos-
terior margins of orbit as defined by
first and sixth suborbitals.

Snout length — from tip of snout to an-
terior margin of orbit.

Interorbital width — shortest distance be-
tween bony rims above eyes.

Maxillary length — from tip of snout to
posterior margin of maxilla.

Lower jaw length — from tip of mandible
to quadratomandibular joint.

Caudal peduncle depth — shortest dis-
tance loetween dorsal and ventral sur-
faces of caudal peduncle.

Caudal peduncle length — from posterior-
most anal fin ray to caudal fin base.

First spine length (first spine of first
dorsal fin, D,I; first spine of second
dorsal fin, DJ; second spine of anal
fin, AH; pelvic fin spine, PJ) — from
base to tip of spine along anterior edge.

Counts were made under a dissecting
microscope with the use of dissecting
needles or insect pins. A Fibre-Lite High
Intensity Illuminator proved invaluable for
examinations of oral, branchial, and visceral
structures. Gill raker and branchiostegal
counts were made on the left side of speci-
mens; remaining counts and measurements
were made on the right side whenever
possible. Counts made include: fin spines
(indicated by Roman numerals), fin rays
(indicated by Arabic numerals), branchi-
ostegal rays, rakers on first gill arch, lateral
line scales, pyloric caeca, vertebrae (pre-
caudal + caudal, including hypural fan),
pleural and epipleural ribs, and basal
ptervgiophores between neural spines 9
and 10.

Osteological data were obtained from
radiographs taken at the Woods Hole
Oceanographic Institution, the Museum of

Comparative Zoology, and the Harvard
University School of Public Health. Holo-
types of Oxyodon iiiacrops and Scepterias
Icninwn were radiographed at the Zoolo-
gisches Museum der Humboldt Universitat
and Australian Museum, respectively. More
comprehensive osteological studies were
based on cleared and stained specimens
prepared by trypsin digestion (Taylor,
1967). Osteological terminology follows
that presented by Gosline ( 1961 ) and
Mead and Bradbury (1963). Suborbital
bones are numbered from 1 to 8 beginning
with the rostralmost element (lacrimal).

Statistical data were analyzed with the
use of the Harvard Computation Labo-
ratory's IBM 360/65 digital computer.
Standard techniques described by Mayr
(1969: 189-193) and Simpson et al. (1960:
65-68, 83-88) were employed for analyzing
meristic data. Morphometric characters
were examined with the aid of regression
techniques specified by Simpson et al.
(1960: 215-233, 238) and Bailey (1959:
91-99).

Before undertaking regression analyses,
morphometric data were plotted against
SL. Graphs were drawn according to a
BMD 05D plotting routine (Dixon, 1967:
71 ) and served as visual tests for linearity
of scatter. Only characters exhibiting linear
scatters were analyzed by regression tech-
niques. As a second precaution against
nonlinearity, subadult specimens ( < 40
mm SL) were excluded from statistical
samples.

Data from several morphometric char-
acters are presented both as ratios (i.e.,
percent of SL or HL) and as regression
parameters. The former are intended only
as identification aids. As Royce (1957: 17)
points out, heterogenic growth makes the
use of ratios in fish taxonomy inefficient
and may lead to erroneous conclusions.

Collection and institution names are ab-
breviated as follows in this paper:

ABE —Collection of Dr. T. Abe,

Tokyo
AM — Australian Museum, Sydney

150 Bulletin Museum of Comparative Zoology, Vol. 146, No. 3

BMNH — British Museum (Natural
History), London

BPBM — Bernice P. Bishop Museum,
Honohihi

CM — Carnegie Museum; collections

presently housed in FMNH,
Chicago

DM — Dominion Museum, Welling-

ton

EBM — Estacion de Biologia Marina,
Universidad de Chile, Viiia
del Mar

FMNH —Field Museum of Natural
History, Chicago

IRSN — Institut Royal des Sciences
Naturelles de Belgique, Brus-
sels

ISH — Institut fiir Seefischerei, Ham-

burg

LACM — Los Angeles County Museum
of Natural History, Los Ange-
les

MCZ — Museum of Camparative Zo-
ology, Harvard University,
Cambridge

MNHN —Museum National d'Histoii-e
Naturelle, Paris

MZF — Museo Zoologico di Firenze,
Florence

RUSI — J.L.B. Smith Institute of Ich-
thyology, Rhodes University,
Grahamstown

SAM — South African Museum, Cape
Town

SMF — Natur-Museum Senckenberg,
Frankfurt am Main

SU — Stanford University; collec-

tions presently housed in the
California Academy of Sci-
ences, San Francisco

TABL — Tropical Atlantic Biological
Laboratory, Miami

UMML — Rosenstiel School of Marine
and Atmospheric Science,
University of Miami, Miami

USNM — National Museum of Natural
History, Washington, D.C.

UZM — Universitetets Zoologiske Mu-
seum, Copenhagen

ZMB — Zoologisches Museum der
Humboldt Universitiit, Berlin

Descriptions are based on material listed
by Mayer (1972: Appendix II). Additional
data were obtained from examinations of
the seventeen specimens listed below. All
seventeen fishes were radiographed.

E. robustus: ISH 1132/66, 3 specimens,
121.1-142.5 mm, WALTHER HER-
WIG Sta. 237/66, 36°00S, 52°58'W,
800 m. ISH 189/71, 9 specimens,
147.0-198.0 mm, WALTHER HER-
WIG Sta. 121/71, 37°44S, 54°43'W,
800 m. ISH 269/71, 1 specimen, 147.5
mm, WALTHER HERWIG Sta. 340/
71, 38°50'S, 54°25'W, 1000 m. ISH
430/71, 1 specimen, 124.1 mm,
WALTHER HERWIG Sta. 348/71,
38°20'S, 54°33 W, 997-1040 m.

E. fmgilis: LACM 32668-6, 1 specimen,
72.5 mm, 2 mi. off Haleiwa, Oahu,
Hawaii, 65 fms. SU 32262, 2 speci-
mens, 90.0-93.9 mm, Honolulu, Hawaii.

Distributions were determined from
material examined and from published
accounts. Because of the confusion in
Epigomis taxonomy, published data were
used only if species identifications could
be verified from included descriptions,
illustrations, etc. Data from specimens of
doubtful identity were not considered. A
complete list of station data taken from the
literature is provided by Mayer (1972:
Appendix II ) .

No attempt has been made to provide
exhaustive synonymies for Epigonus spe-
cies. References are cited only if they (1)
are taxonomically or zoogeographically im-
portant; (2) provide outstanding descrip-
tions, illustrations, or synonymies; or (3)
represent verifiable misidentifications. Non-
taxonomic accounts have been omitted, as
liave references to cruise summaries and
faunal lists.

SYSTEMATICS

Statistical data are presented in tables
accompanying species descriptions; meristic

Epigonus Systematics • Mayer 151

characters witli low variability are reported
ill the text as value, followed in parentheses
by number of specimens exiiibiting that
v^alue. Meristic and mensural data from
holotypes of new species are presented in
the Appendix.

Genus Epigonus Rafinesque, 1810

Epifiomis Rafinesque, 1810: 64. (Type .species:
Epigonus macrophthahnus Rafinesque, 1810
by in()n()t>p\'. A synonym of Pomatomus
telescopus Risso, 1810.)

Tt'h'scops Bleeker, 1876: 261. (Type .species:
Poiuatonniti tclescopiuiu [sic!] Risso, 1810 by
original designation. Pomatomus deemed in-
applicable. )

Pomatomichthijs Ciglioli, 1880: 20. (Type species:
Pomatomiclithys constanciae Giglioli, 1880 by
monotypy. A synonym of Pomatomus teles-
copus Risso, 1810.)

Hynnodiis Gilbert, 1905: 217. (Type species:
Hijnnodus athcrinoides Gilbert, 1905 by mono-
typy. A synonym of Epigonus occidentaJis
Goode and Bean, 1896.)

Oxyodon Brauer, 1906: 287. (Type species:
Oxyodou Diacwps Brauer, 1906 by monotypy.)

Xystramia Jordan, 1917: 46. (Type species:
Glossamia pandionis Goode and Bean, 1881
by original designation. Glossamia deemed
inapplicable. )

Scepterias Jordan and Jordan, 1922: 44. (Type
species: Scepterias fragilis Jordan and Jordan,
1922 by monotypy.)

Paraliynnodus Barnard, 1927: 525. (Type species:
Parahynnodus robustus Barnard, 1927 by mono-

typy- )

Diapiosis. Epip^onus is distinguished
from other lower perciform genera by a
mosaic of characters including 8 suborbital
bones, all lacking subocular shelves; large,
thin-walled swimbladders with postero-
dorsal ovals; VII or VIII first dorsal fin
spines; 1,9 or 1,10 second dorsal fin ele-
ments; 11,9 anal fin elements; 15-23
pectoral fin rays; and 17-35 gill rakers. No
member of the genus exhibits fang-like
conical teeth, such as are found in Cheilo-
dipterus, or anteriorly projecting teeth,
such as are found in Rosenblattia.

Description. Body elongate, fusiform;
dorsal and ventral profiles slightly convex,
similar. Mouth oblicfue, terminal; upper
jaw protrusile; maxilla excluded from gape.

sheathed l)y lacrimal anteriorly, free pos-
teriorly; supramaxilla absent. Eye large,
round or oval. Nostrils paired, rounded or
slit-like, two on each side of head.

Premaxillae, mandibles, vomer, and pahi-
tines edentulous or bearing conical teeth;
tongue and endopterygoids rarely dc>ntiger-
ous; ectopterygoids edentulous. Gill rakers
moderate to long, 17-35; branchiostegal
rays 7 (3 + 4); pseudobranchiae present.

Opercular .spine either weak, flattened
and poorly ossified, or pungent and bony;
spine ventral to one or more horny or mem-
branous spinelets. Preopercle with double
edge; angle frequently produced.

Dorsal fins VII-1,9, VII-1,10, or VIII-
1,10, separated by distinct interdorsal
space; rudimentary subcutaneous eighth
spine present in seven-spined forms. Anal
fin 11,9; pectoral fins 15-23; peKic fins 1,5:
caudal fin forked, 9 + S principal rays,
upper- and lowermost rays unbranched.

Scales large, deciduous, ctenoid. Lateral
line complete, extending parallel to dorsal
profile on dorsolateral surface of trunk,
descending to midline on posterior portion
of caudal peduncle, continuing on tail;
lateral line scales 33-51; canal simple,
broadening into deltoid or Y-shaped tube
at rear edge of scale. Scale pockets cover-
ing most of body including occiput, soft
dorsal, anal, and caudal fins; scales absent
from snout; no axillary scale at base of
PJ spine.

Suborbitals 8, all lacking subocular
shelves. Vertebrae 25; basapophyses on
vertebrae 3 or 4. Predorsals 3, first and
second interdigitating between neural
spines 2 and 3, third located behind neural
spine 3. Caudal skeleton with 2 autogenous
haemal .spines, 6 hypurals (hypural 1 =
parhypural sensu Monod, 1968), 3 (>purals,
2 (rarely 1) pairs of uroneurals. Actinosts
4, 3VL' borne by scapula.

Swimbladder large, thin-walled, lacking
anterior or posterior projections to cranium
and neural arches; diaphragm absent; oval
posterodorsal; retia mirabilia well devel-
oped. Stomach U- or Y-shaped; pyloric

152 Bulletin Museum of Comparative Zoology, Vol. 146, No. 3

caeca .5-34, may be modified into lu-
minescent organs; intestines simple, folded
into three segments. Specimens dioecious;
no evidence of hermaphroditism or oral
brooding.

Habitat: Engybenthic; continental slope
between approximately 200 and 1200
meters.

Key to Species of Epigonvs

la Opercular spine weak, poorly ossified, or
absent (opercular spine refers to the
ventralniost reinforced projection from the

posterodorsal edge of the opercle) 2

b Opercular spine pungent, bony 7

2a Lateral line scales 46-51; tongue eden-
tulous or bearing scattered tooth patches

b Lateral line scales 33-36; tongue cov-
ered with tooth patches (Fig. lA)
E. oligolepis

3a Gill rakers 23-34; premaxillary teeth short,
conical or villifonn, not visible when

mouth closed 4

b Gill rakers 17-21; premaxillary teetli
elongate, thin, inwardly recurved, visible
when mouth closed E. macrops

4a Pyloric caeca 7-14; first dorsal fin VII,
rarely VIII; vertebral count 10 + 15;
specimens not exceeding 220 mm SL 5

b Pyloric caeca 21-34; first dorsal fin VIII,
rarely VII (DiVIII often small or rudi-
mentary); vertebral count 11 -|- 14;
specimens to 550 mm SL E. telescopus

5a Body long, shallow; depth 15.8-23.6% SL;
peduncle length 25.4-32.2% SL; caudal

peduncle ring absent 6

b Body short, deep; depth 22.4-29.6% SL;
peduncle length 22.0-26.3% SL; caudal
peduncle ring present on specimens
shorter than 110-120 mm SL (Fig. IB)
E. pandionis

6a Gill rakers 28-34; pyloric caeca 10-14;
head length 31.2-38.6% SL; 2 pterygio-
phores between neural spines 9 and 10,

rarely 1 E. denticulatus

b Gill rakers 25-26, pyloric caeca 7-8;
head length 30.0-34.0% SL; 1 pterygio-
phore between neural spines 9 and 10
E. fiagilis

7a Body moderate to deep, 20.0-32.0% SL;
dorsal fins VII-1,9, rarely VII-1,10; gill

rakers 26-35 8

b Body shallow, 14.0-19.5% SL; dorsal fins
VII-1,10, rarely VII-1,9; gill rakers 22-27
— . E. occidentalis

8a Gill rakers of lower arch simple, awl-
shaped 9

b Gill rakers of lower arch pectinate ( Fig.

IC) - E. pectinifer

9a Tongue edentulous 10

h Tongue covered with tooth patches

E. trewavasac

10a Head length 28.0-36.6% SL; head height
14.7-18.8% SL; gill filaments moderate

or short 11

b Head length 36.8-41.9% SL; head height
18.9-21.1% SL; gill filaments long

,_ E. crassicaiidiis

11a Fin spines long, DJ 14.8-18.7% SL,
All 13.0-20.8% SL; interorbital width
8.7-10.2% SL; eyes large, 40.0-51.1% HL

E. leinmen

b Fin spines moderate, D2I 10.0-12.6% SL,
All 9.2-13.3% SL; interorbital width
6.5-8.2% SL; eyes moderate to small,
37.4-42.2% HL E. robustus

Epigonus telescopus (Risso, 1810)
Figure 2

Pomatomus telescopus Risso, 1810: 301, plate IX,
fig. 31 (original description; Nice; holotype
examined, MNHN B862); Lowe, 1841: 173;
Capello, 1868: 160; Moreau, 1881: 386, fig.
125; Vaillant (in part), 1888: 376.

Epigonus macwphthalmus Rafinesque, 1810: 64
( original description; no type locality; holotype
lost).

Pomatomus telescopium Cuvier, 1828: 171 (in-
correct emendation of Pomatomus telescopus
Risso, 1810); Valenciennes, 1830: 495;
Valenciennes, 1837-1844: 6, plate I; Giinther,
1859: 250; Cocco, 1885: 85; Holt and Calder-
wood, 1895: 405, plate LXIl.

Pomatomus cuvieri Cocco, 1829: 143 (original
description; seas of Messina; holotype not
examined ) .

Pomatotnus cuvicrii \'alenciennes, 1830: 501 (in-
correct emendation of Pomatomus cuvieri
Cocco, 1829).

?Pomatomichthys constanciae Giglioli, 1880: 20
( original description; Straits of Messina; holo-
type not examined, MZF 3089); Goode and
Bean, 1896: 234.

Epigonus telescopus Goode and Bean, 1896: 232;
Cligny, 1903: 9; Barnard, 1927: 523; Gall,
1931: 1, fig. 1; Fowler, 1936: 736, fig. 326;
Smith, 1949b: 206, fig. 474.

Scepterias lenimen, Whitley ( in part ) ( not Whit-
ley, 1935), 1968: 56.

Diagnosis. E. telescopus is the largest
species of the genus, growing to over 550
mm SL. Specimens are characterized by
21-34 pyloric caeca and eight first dorsal

EriaoNvs Systematics • Mayer 153

j^i^'

^^.^fffi^^'^h'"**'^^'

Figure 1. A. Tongue of E. oligolepis. Stippled areas
indicate tooth patches; shape and size of tooth patches
may vary among specimens. B. Caudal peduncle of
young E. pandionis showing anterior ring and posterior
band. C. Gill raker of E. pectinifer showing nub-like
processes.

fin spines. The opercular spine is blunt and
poorly ossified and distinguishes the species
from E. occidentalis, E. trewavosae, E.
pectinifer, E. rohiistus, E. lenimen, and E.
crassicaudus, which have pungent oper-

Y12yc

cular spines. Unlike remaining congeners,
E. telescopus possesses 11 + 14 vertebrae.

Description. Meristic data presented in
Table 1; regression data for morpho metric
traits presented in Table 2.

Body thickset, shortened; anterodorsal
profile slightly convex, rising most steeply
from tip of snout to interorbital region;
body moderate to deep, 21.2-26.3%^ SL;
caudal peduncle short, 19.0-26.5% SL.

Head moderate to deep, height 13.3-
SL; length 30..5-37.9% SL; snout
blunt; angle of gape moderate to large;
lower jaw equalling or protruding slightly
beyond upper jaw. Maxilla rarely exceed-
ing %-% eye length, posterior margin of
maxilla broad, posteriormost point near
ventral surface of bone; maxillae of large
specimens scaled. Eye round, 49.5-58.9%
HL; circumorbital tissues scaled, scale
pockets particularly apparent in large spec-
imens; anterodorsal rim of orbit projecting
into profile in small forms, reaching profile
in larger forms; interorbital width 9.0-
10.9% SL.

Dentition variable with age (see Onto-
genetic change); premaxillae, mandibles,
vomer, and palatines dentigerous; tongue
edentulous.

Opercle bearing short, poorly ossified
spine ventral to 1-8 membranous or poorly
ossified spinelets; spine and spinelets sepa-
rated by shallow gap; spinelets occasionally
obscured by underlying membranes. Pre-
opercle variable with age; angle rounded,
slightly produced in specimens shorter than

Figure 2. Epigonus telescopus, 220.0 mm SL, ISH 70/63.

154 Bulletin Museum of Comparative Zoology, Vol. 146, No. 3

Table 1. Epigonus telescopus meristic data. X = mean; SD = stan-
dard DEVIATION; n = NUMBER OF SPECIMENS.

Range

Pectoral fin rays 20.85 19-23 0.71 54

Gill rakers 24.40 23-26 0.85 52

Lateral line scales 48.14 46-50 1.09 50

Pyloric caeca 25.25 21-34 3.59 16

200 mm SL, broadly produced in larger
forms; minute serrations on angle and ven-
tral surface of bone, rarely along posterior
surface dorsal to angle; striations radiating
from inner edge of angle. Interopercles
and subopercles without stiiations, occa-
sionally bearing minute serrations on pos-
tero ventral surfaces. Gill rakers simple,
awl-like.

First dorsal fin VII (7), VIII (46),
eighth spine small or rudimentary, lack-
ing membranous connection to preceding
spines; second dorsal fin 1,9 (1), 1,10 (52),
1,11 (1); anal fin 11,9 (56); D,I long,
3.5-6.3% SL; DJ, All short, 5.3-9.5%,
5.7-10.6%o SL respectively; Pol moderate,
6.5-11.9% SL.

Vertebrae 11 + 14 (18); epipleural ribs

Table 2. Epigonus telescopus regression data, b = regression coeffi-
cient ± 95%o CONFIDENCE INTERVAL; a = Y INTERCEPT; n = NUMBER OF

specimens. All regressions on SL.

Body depth

-2,

Head height

Eye diameter

Snout length

-2.

Interorbital width

-0,

Maxillary length

Lower jaw length

-0.

Caudal peduncle de

pth

NONLINEAR

Caudal peduncle length

2 1

D2I

1 1

All

P2I

NONLINEAR

Ei'iaoNus Systematics • Mayer 155

Table 3. Ontogenetic changes in the dentition of E. telescopus.

A. PREMAXILLARY DENTITION

< 200 mm SL

Extent 1/2-2/3 of ventral

surfoce

Pottern

I row

200-400 mm SL

2/3-7/8 of ventrol
surface

1-2 irregular rows
tapering to I row

> 400 mm SL

2/3-7/8 of ventral
surface

Multiple irregular rows

B. MANDIBULAR DENTITION

Extent

< 150 mm SL
Entire coronoid surface

150-250 mm SL
Entire coronoid surface

> 250 mm SL
Entire coronoid surface

Pottern

row

2-3 irregular rows
tapering to |-2 rows

3, 4, or 5 irregular rows

C. VOMERINE DENTITION

Extent

< I 75 mm SL
Center of vomer

> I 75 mm SL
Entire face of vomer

Pattern

Scattered teeth in few
irregular rows

Numerous teeth in multiple
irregular rows

D. PALATINE DENTITION

Extent

< 150 mm SL
Length of ventral surface

> I 50 mm SL
Length of ventral surface

Pottern

1-2 irregular rows
tapering to I row

2-5 irregular rows
tapering to I row

7 (11), 8 (2), inserting on vertebrae 1-7
or 1-8 respectively; pleural ribs 9 (17),
inserting on vertebrae 3-11.

Large specimens black or brown-violet,
iridescent in life ( Risso, 1810; Steindachner,
1891; Dons, 1938). Color in alcohol vari-
able with mode of collection and preser-
vation; skin often abraded, revealing under-
lying white-orange tissue; scale pockets
mottled with black or brown, melanophores
more densely packed near caudal edges;
pigment darker in larger fish; skin oily,
cutaneous fat deposits adding rust-colored
tint; opercular area black. Guanine de-
posits occasionally occurring on opercular,
tlioracic, and abdominal regions; iris black

with silver highlights; branchial membranes
black; mouth darkening with age (see
Ontogenetic change ) .

Description based on 54 specimens 68.1-
553 mm SL.

Ontogenetic change. Several marked
ontogenetic changes occur in E. telescopus,
the most noticeable involving dentition
patterns. Tooth-bearing bones of young
specimens exhibit relatively prominent con-
ical teeth. Teeth become more numerous
witli growth but appear smaller and form
weak conical or villiform bands. As Table
3 illustrates, older specimens have more
complex tooth patches with larger numbers
of tooth rows.

156 Bulletin Museum of Comparative Zoology, Vol. 146, No. 3

Epigonus Systematics • Mayer 157

A second change involves oral pigmenta-
tion. Young individuals have white or pale
yellow mouths; melanin is present only in
the vicinity of the pharynx. By the time
specimens reach 175-225 mm SL, black
pigment extends anteriorly to cover the
entire tongue. Shortly thereafter, the palate
becomes totally blackened, and by 300
mm SL, the entire mouth is dark.

The above changes are associated with
alterations in intestinal length. Measure-
ments of fourteen specimens ranging from
90.7-553 mm SL indicate that intestinal
length increases from 66-737^ SL in small
specimens (90.7-128.5 mm) to 98-108%
SL in moderate-sized individuals (220-250
mm). Thereafter, intestines grow more
slowly, reaching 110-115% SL in the largest
specimens. The coincidence of rapid in-
testinal growth, dentition changes, and de-
velopment of oral pigment suggests that
E. telescopus modifies its feeding habits
with growth.

Distribution. E. telescopus has an anti-
tropical distribution in the Atlantic, oc-
curring from Iceland to the Canary Lslands
and reappearing along the western coast
of South Africa (Fig. 3). Specimens have
also been taken in the Subtropical Con-
vergence region east of New Zealand. The
species is well known in the western Med-
iterranean and has been captured once
off the eastern coast of North America. A
single specimen is known from shallow
water off Norway (Dons, 1938).

Adults are taken by bottom trawl or long-
line and are most abundant from 300 to
800 meters; however, specimens have been
captured from water as shallow as 75 to
80 meters to as deep as 1000 to 1200 meters.
Koefoed (1952) reports four pelagic ju-
veniles from the Azores; Bertolini (1933)
mentions the presence of juveniles in the
Tyrrhenian Sea.

Earlier workers reported the range of E.
telescopus to include St. Helena (Val-
enciennes, 1837-1844; Giinther, 1868;
Bauchot and Blanc, 1961), tropical west
Africa (Osorio, 1898; Poll, 1954; Bauchot

and Blanc, 1961), and the Indian Ocean
(Steindachner, 1907; Fowler, 1935). These
accounts are based on misidentified or
tenuously identified material. The .speci-
mens described by Giinther, Poll, and
Bauchot and Blanc are E. pandionis, while
that examined by Fowler is Scomhrops-
like. Valenciennes' identification is ba.sed
on an impublished description and figure
by a St. Helena resident and must be re-
garded with suspicion. Reports by Stein-
dachner and Osorio could not be evaluated,
because neither includes a description or
figure of the material studied.

Geographic variation. The scarcity of
material from South Africa and New
Zealand makes it difficult to judge the
degree to which Northern and Southern
Hemisphere populations of E. telescopus
have diverged. Comparisons of dorsal and
pectoral fins, pleural and epipleural ribs,
lateral line scale counts, gill rakers, and
pyloric caeca reveal no subspecific dif-
ferences (coefficients of difference ^ 0.44).
Moiphometric characters, on the other
hand, exhibit greater variability. Of eight
traits successfully analyzed, three are sig-
nificantly different at both the 95%, 98%,
and 99% levels of confidence (Table 4).
These differences suggest that northern
and southern populations represent gemi-
nate subspecies; however, additional ma-
terial must be collected, especially from the
Southern Hemisphere, before definitive
statements can be made on intraspecific
variability.

Ta.xonomic notes. Pomatomichthtjs con-
stanciae Giglioli, 1880 is pro\'isionaily con-
sidered a junior synonym of E. telescopus
on the basis of work by Tortonese and
Queirolo (1970). These authors re-exam-
ined and, for the first time, figured the
holotype of P. constanciae. The latter
species is known only from the type speci-
men. The original description (Giglioli,
1880) is incomplete; no adequate rede-
scription has ever been published.

Data from the papers mentioned above
indicate a similarit\' between P. constanciae

158 Bulletin Museum of Comparative Zoology, Vol. 146, No. 3

Table 4. Comparison of regression coefficients from Northern and Southern
Hemisphere populations of E. telescopus. Data evaluated at the 95%, 98%,
and 99% levels of confidence. df = degrees of freedom; nb = regression
coefficients of Northern Hemisphere specimens; Sb = regression coefficients
OF Southern Hemisphere specimens; SD = significant difference between tab-
ular AND calculated VALUES OF t; t ^ CALCtJLATED VALUES OF t.

Significance

irt

CT>

1. 54

Body depth

0. 82

Head height

3. 88

Eye diameter

1. 75

Snout length

3. 19

Interorbital width

1. 05

Maxillary length

. 16

0 . 16

Lower jaw len

gth

. 19

0. 71

Caudal pedunc
depth

NONLINEAR

Caudal pedunc
length

. 22

0. 58

^2^

INSUFFICIENT

DATA

All

. 04

3. 80

^2^

NONLINEAR

Table 5. Comparison of dorsal and pectoral fin counts from E. telescopus, P.

CONSTANCIAE, AND E. TREWAVASAE. DaTA FOR P. CONSTANCIAE FROM GiGLIOLI (1880)
AND TORTONESE AND QuEIROLO (1970); REMAINING DATA FROM PRESENT STUDY.

E. telescopus P. constanciae E. trewavasae

VII VII

I, 9.
1,9 rarely

I, 10

18 16-18

First

VIII,

dorsal

rarely

fin

VII

Second

I, 10,

dors a 1

r ar ely

fin

I, 9

Pectoral
fin

19 - 2:

Ei'iGONus Systematics • Mayer 159

Figure 4. Epigonus macrops, 154.6 mm SL, USNM 207679.

and E. telescopus but also suggest an af-
finity between P. constanciae and E. tre-
wavasae Poll, 1954. As is shown in Table
5, dorsal and pectoral fin counts fall within
the range of E. treicavasae rather than E.
telescopus. Tortonese and Queirolo's figure
similarly shows the holotype to possess a
sharp opercular spine, short DJ, and long
PJ — all characteristics of E. treicavasae.
Mensural data fail to differentiate P.
constanciae from either species. Unlike E.
treicavasae but like E. telescopus, the holo-
type lacks lingual teeth (Giglioli, 1880).

In view of the uncertainty surrounding
P. constanciae, a closer study of this form
must be undertaken. The problem is all
the more pressing, because E. treicavasae
is recorded from the Mediterranean for the
first time in this paper.

Common names. Comprehensive lists of
common names for E. telescopus are pro-
vided by Doderlein (1889), Nobre (1935),
and Bini (1968). Three names not re-
corded in these works are "Mejluza" — Gran
Canaria ( Steindachner, 1891), "Devil-fish"
— North Sea area (Ehrenbaum, 1928), and
"Big-eyed cardinal fish" — New Zealand
( Anonymous, 1961 ) .

Epigonus macrops (Brauer, 1906)
Figure 4

Oxijdon macrops Brauer, 1906: 288, fig. 172
(original description; Indian Ocean, land-locked
sea on west coast of Sumatra, VALDIVIA Sta.
186, 03°21'01"S, 101°11'05"E, 903 m; syntype
examined, ZMB 17678); Weber and de Beau-
fort, 1929: 351, fig. 81; Nomian, 1939: 60.

Diagnosis. E. macrops may be distin-
guished from all congeners by its low gill
raker counts ( 17-21 ) . It is further char-
acterized by eight fully developed first
dorsal fin spines and eight pyloric caeca,
one of which may function as a lumin-
escent organ.

Description. Meristic values presented
in Table 6; regression data for morpho-
metric traits presented in Table 7.

Body elongate; anterodorsal profile rising
steeply to occipital area; thereafter, weakly
convex, almost horizontal to first dorsal fin;
body depth 19.7-24.1% SL; caudal pe-
duncle length 22.0-26.7% SL.

Head length 34.1-38.5% SL; head licight
17.2-21.9% SL; snout blunt; angle of gape
large; lower jaw protruding beyond upper
jaw. Maxilla rarely exceeding Vs-% eye
length; posterior margin of maxilla broad,
bearing posteriormost point at ventral sur-
face of bone. Eye round to oval, 39.7-
48.3%' HL; anterodorsal rim of orbit pro-
jecting strongly into dorsal profile; inter-
orbital region wide, 9.5-11.7% SL.

Teeth conical, frequently recurved. Pre-
maxillary and mandibular teeth prominent,
needle-like, arranged in single row along
length of jaws; mandibular teeth occa-
sionally forming double row near sym-
physis; vomerine teeth few, moderate,
arranged in 2-4 irregular rows or in a
triangular or diamond-shaped patch; pala-
tin(\s bearing 2-6 teeth, arranged in single
row covering anterior half or second quar-
ter of bone; tongue edentulous.

Opercular spine short, weak, bony, ven-

160 Bulletin Museum of Comparative Zoology, Vol. 146, No. 3

Table 6. Epigonus macrops mebistic data. X = mean; SD = standard

DEVIATION; n = NLTNIBER OF SPECIMENS.

Range

Pectoral fin rays 18.87 18-19 0.35 30

Gill rakers 18.63 17-21 0.87 32

Lateral line scales 48.61 46-50 0.83 28

Pyloric caeca 8.00 8 0.00 15

tral to 3-10 spinelets; spine and spinelets
separated by shallow, occasionally narrow
gap. Preopercular angle weakly produced,
rounded, serrate; serrations occasionally ex-
tending to posterior and ventral surfaces of
bone, rarely absent; striations radiating
from inner edge of angle. Subopercle and
interopercle generally serrated, occasionally
striated. Gill rakers short, awl-like.

First dorsal fin VII (1), VIII (29);
second dorsal fin 1,9 (1), 1,10 (31); anal
fin 11,9 (30), 11,10 (1). D^I, DJ, All
short, equalhng 1.2-2.9%, 5.3-7.7%, 5.9-
9.9% SL respectively; PJ moderate, 11.7-
14.1% SL.

Vertebrae 10 + 15 (25); epipleural ribs
6 (23), inserting on vertebrae 1-6; pleural
ribs 8 (24), inserting on vertebrae 3-10.

Table 7. Epigonus macrops regression data, b = regression coeffi-
cient ± 95% confidence interval; a = Y intercept; n =: number of
specimens. All regressions on SL.

± •

Body depth

Head height

Eye diameter

Snout length

-0.

Interorbital wi

dth

Maxillary leng

Lower jaw leu

gth

Caudal peduncl

e dep

- 1.

Caudal peduncl

e len

gth

D2I

All

P2I

Ei'icoNus Systematics • Mayer

161

Figure 5. Caudal peduncle of young E. macrops bear-
ing anterodorsally canted ring.

Specimens probably black in life. Color
in alcohol variable with preservation; scale
pockets covered with black melanophores
near posterior edges; skin trecjiiently
abraded, revealing pink-yellow muscnla-
ture; opercular bones transparent, colored
black by underlying branchial membranes;
iris black; mouth black in adults. Young
bearing anterodorsally canted caudal pe-
duncle ring (see Ontogenetic change).
First pyloric caecum modified into lumin-
escent organ ( see Remarks ) .

Description based on 32 specimens 77.8-
206.0 mm SL.

Ontogenetic change. The transition from
juvenile to adult in E. macrops is marked
by changes in pigmentation and body
shape. Pelagic juveniles 15-37.9 mm SL
and young demersal forms 77.8-79.8 mm
SL bear a thin, black, anterodorsally tilted
ring circling the center portion of the
caudal peduncle (Fig. 5). Specimens
larger than 90 mm SL lack this marking.
Melanophores forming the rings are deeply
embedded in the peduncle musculature
and cannot be obliterated by abrading the
surface of the fish.

Adult E. macrops arc characterized by
black oral and branchial membranes. Al-
though these areas are colorless or poorly
pigmented in specimens smaller than 40
mm SL, the former surfaces darken and the
latter become covered with brown melano-
phores by the time fish reach 80 nun SL.

Juvenile E. macrops appear longer
and shallower than adults. Ratio-on-size
diagrams for interorbital width (i.e., in-
terorbital width/ SL vs. SL) indicate al-
lometric growth takes place in small
specimens. Similar statements are probably-

true for head height, eye length, and
caudal peduncle measurements but could
not be tested because of damage to juvenile
specimens.

DistriJnition. E. macrops adults are taken
exclusively by bottom trawls between 550
and 1100 meters in the Lidian Ocean,
Gulf of Mexico, Caribbean Sea, and West-
ern Atlantic. Specimens are most abundant
between 640 and 920 meters. Pelagic ju-
veniles are known from the Caribbean at
depths of 120 to 550 meters (Fig. 6).

GcograpJiic variation. No investigation
made because of inadec^uate Indian Ocean
samples.

Taxonomic notes. Brauer's description of
Oxyodon macrops ( 1906 ) is based on two
syntypes from the eastern Indian Ocean
( 172 and 212 mm total length ) . Of these,
only the larger is in the Zoologisches Mu-
seum der Humboldt Universitiit; the smal-
ler has been lost. The misplaced type may
have been deposited in the Zoologisches
Institut der Universitiit Leipzig and may
reappear when portions of this collection,
presently stored in Berlin, are sorted and
catalogued (Karrer, personal communica-
tion ) .

Remarks. Specimens of E. macrops bear
eight pyloric caeca; one of these appears
modified into a bioluminescent organ. The
luminescent caecum arises from the mid-
ventral surface of the pylorus just before
the duodenum and main body of pyloric
appendages (Fig. 7). It extends ventrally
until it reaches the floor of the abdominal
cavity, bends anteriorly and inserts into a
pouch formed by the black peritoneal
lining of the body cavity. At the posterior
edge of the pelvic girdle, the caecal pouch
lies over a thin, translucent portion of the
body wall which may function as a biolu-
minescent window. Externally the biolumi-
nescent window is covered by a single
large scale. The caecal pouch is lined with
silver or silver-gray pigment. Guanine
deposits appear most concentrated anter-
odorsally.

.\lthouah there is no direct evidence to

162 Bulletin Museum of Coiiiparative Zoology, Vol. 146, No. 3

m
in

^—

T3
O

3
CO

■D

H—

"cO

If)

•

c
fl)

Q.
O

CD
Q.

■n

■^-

■^

(D O
U. TO

Epiconus Systematics • Mayer 163

Figure 7. Luminescent organ of E. macrops. BW, body wall; D, duodenum; LPC, luminescent pyloric caecum;
LW, luminescent window; PC, nonluminescent pyloric caeca; PER, peritoneum; R, reflector; S, stomach.

support tlie claim that E. macrops is lu-
minescent, the modifications described
above are similar to those found in several
luminescent perciforms. Pernpheris klun-
zingeri and Parapriacanthus ransonneti
(Pempheridae) have luminescent organs
embedded in the thoracic ventral muscula-
ture formed from, or directly associated
with, the first pair of pyloric caeca ( Haneda
et al., 1966). Luminescent shallow-water
apog(jnids such as Apogon ellioti and
Siphamia nwiirnai also have luminescent
organs associated with the alimentary canal.
In both of the latter forms, anal and/ or

thoracic organs are connected by duct to
the intestine. As in E. nuicrops, tissue
above the luminescent structures may serve
as a reflector (Iwai, 1959; Haneda et al.,
1966).

Common names. None.

Epigonus pandionis (Goode
and Bean, 1881)
Figure 8

Apogon pandionis Goode and Bean, 1881: 160
( original description; off entrance to Cliesa-
peake Bay; holotype examined, USNM 26228);
Jordan and Gilbert, 1882: 564.

Figure 8. Epigonus pandionis, 141.7 mm SL, TABL uncatalogued.

164 Bulletin Museum of Comparative Zoology, Vol. 146, No. 3

Table 8. Epigonus pandionis meristic data. X = mean; SD = stan-
dard deviation; n = number of specimens.

Range

Pectoral fin r ays
Gill rakers
Lateral line scales
Pyloric caeca

17. 81

17- 19

0. 57

27. 84

26- 30

0. 88

101

47. 63

46- 49

0. 66

10. 81

10- 13

0. 74

Glossamia pandionis Goode and Bean, 1896: 231.

Xystramia pandionis Jordan, 1917: 46.

Epigonus telescopus. Poll (not Risso, 1810),

1954: 89, fig. 26; Bauchot and Blanc (in part),

1961: 70.

Diagnosis. E. pandionis is the most ro-
bust species of the genus. Specimens
shorter than 110-125 mm SL are dis-
tinguished by a posterodorsally canted ring
circhng the caudal peduncle.

E. pandionis differs from E. macrops and
E. telescopus by bearing VII (rarely VIII)
spines in the first dorsal fin and 10-13
pyloric caeca. It is unlike E. oligolepis
because it has 46-49 lateral line scales and
may be distinguished from E. treioavasae,
E. pectinifer, E. robustus, E. lenimen, E.
crassicaudus, and E. occidentalis because
it lacks a pungent, bony opercular spine.
E. pandionis most closely resembles E.
fragilis and E. denticulatus but is differ-
entiated by its short caudal peduncle ( 22.0-
26.3% SL) and deep body (22.4-29.6%
SL). It further differs from E. dejiticulatiis
by exhibiting gill raker counts of 26—30
and a single basal pterygiophore between
neural spines 9 and 10.

Description. Meristic values presented in
Table 8; regression data for morphometric
traits presented in Table 9.

Body shortened, robust; anterodorsal pro-
file convex, particularly between occiput
and first dorsal fin; body deep, 22.4-29.6%
SL; caudal peduncle short, broad, length
22.0-26.3% SL.

Head length 33.0-39.0% SL; head height

19.0-22.2% SL; snout blunt; angle of gape
large; upper jaw subequal to lower jaw.
Maxilla reaching %-y-2 eye length; pos-
terior margin broad, posteriormost point at
ventral edge of bone. Eye round or slightly
oval, 37.4-48.7% HL; anterodorsal rim of
orbit projecting into profile in smaller
specimens, reaching profile in larger forms;
interorbital region wide, 9.2-11.5% SL.

Dentition variable with age (see Onto-
genetic change); premaxillae, mandibles,
vomer, and palatines dentigerous, bearing
conical, occasionally recurved teeth; tongue
edentulous.

Opercular spine short, horny, ventral to
2-5 (usually 3-4) poorly ossified spinelets;
spine separated from spinelets by gap;
spinelets occasionally obscured by un-
derlying membranes. Preopercular angle
broad, rounded, moderately produced;
striations radiating from inner ridge to
edges of angle; serrations along posterior
and ventral surfaces of bone. Subopercle
and interopercle bearing scattered serra-
tions. Gill rakers awl-like.

First dorsal fin VII (95), VIII (5); sec-
ond dorsal fin 1,9 (2), 1,10 (97), 1,11 (1),
11,10 (1); anal fin 11,8 (2), 11,9 (96), 11,10
(1); DJ long, 3.6-8.1% SL; DJ, All, PJ
short, equalling 5.0-8.6%, 5.0-8.7%, 8.9-
12.7% SL respectively.

Vertebrae 10 + 15 (36); epipleural ribs
6 ( 25 ) , 7 ( 1 ) , inserting on vertebrae 1-6
and 1-7 respectively; pleural ribs 8 (36),
inserting on vertebrae 3-10.

Epigonus Systematics • Mayer 165

Table 9. Epigonus pandionis regression data, b = regression coefft-

CIENT ± 95% CONFIDENCE INTERVAL; a = Y INTERCEIT; 11 = NUMBER OF

SPECIMENS. All regressions ON SL.

Body depth

-2.

Head height

-0.

Eye diameter

-0.

Snout length

I n t e r 0 r b i t a 1 width

1 1

Maxillary length

-0.

Lower jaw length

-0,

Caudal peduncle dep

-0.

Caudal peduncle len

gth

D2I

All

P2 I

Pigmentation variable with age (see
Ontogenetic change ) ; scale pockets mottled
with black; fin membranes black; opercular
region of adults black-slate gray; mouth
primarily light; iris black. Specimens fre-
quently abraded, underlying tissue pale
yellow-rust brown; guanine deposits rare,
if present occurring on opercular complex,
isthmus, thorax, or abdomen; silvered forms
generally from old collections.

Descriptions based on 104 specimens
45.7-194.0 mm SL.

Onto<ienetic chan2,e. Maturation in E.
pandionis is accompanied by changes in
pigmentation and dentition. The most
striking transformation involves caudal pe-
duncle markings. Specimens smaller than S5
mm SL bear a thin, black, posterodorsally
sloped ring circling the central portion of
the caudal peduncle. Melanophores form-
ing the ring are deeply embedded in pe-
duncle musculature and are not easily
abraded. A broader, more superficial band

of pigment circles the caudal peduncle at
the base of the caudal fin (Fig. IB). As
specimens grow beyond 85 mm, the rings
become fainter and begin to disappear.
Fish larger than 110 mm SL may com-
pletely lack peduncle markings, and by
125 mm SL, rings are absent from \irtually
all specimens. Since E. pandionis becomes
sexually mature at approximately 110 mm
SL, altered markings may reflect changes
in habit or behavior associated with repro-
ductive individuals.

Gill rakers and branchial membranes are
converted from pale yellow to black. Spec-
imens smaller than 55 mm SL bear scat-
tered black melanophores on gill rakers
but lack opercular pigmentation. B\- 60
mm SL rakers have become totalK' dark,
and traces of melanin ha\'e appeared on
membranes lining the opercle. Pigment
becomes denser with growth and spreads
ventrally. By 100 mm SL the opercle is
completely lined with dark tissue. Since

166 Bulletin Museum of Comparative Zoology, Vol. 146, No. 3

Ei'iGONus Systematics • Mayer 167

opercular bones are translucent, the process
appears outwardly as a darkening of the
opercle.

Ontogenetic changes in dentition involve
the production of increasingly complex
tooth patches. Specimens smaller than 80
mm SL bear single rows of teeth on the
premaxillae and palatines. Mandibular teeth
are arranged in patterns analogous to those
found on the premaxillae or in double rows
that taper to a single row posteriorly.
Vomerine teeth occur in 1-2 chevron-
shaped clumps. As growth takes place,
teeth are added to all dentigerous surfaces.
Large specimens (>130 mm SL) have as
many as 3-4 tooth rows on palatines and
anterior segments of dentaries and premax-
illae. Vomerine teeth may become suf-
ficiently numerous to cover the entire face
of the bone.

Distribution. E. pandionis is amphi-At-
lantic, occurring primarily in the Caribbean,
Gulf of Mexico, and Gulf of Guinea (Fig.
9). The species has been taken as far
north as New Jersey and as far south as
French Guiana in the western Atlantic.
It occurs between Portuguese Guinea and
Angola in the eastern Atlantic. Adults are
captured exclusively by bottom trawls be-
tween 210 and 600 meters. American forms
are most numerous from 300 to 500 meters,
while African populations are most abun-
dant between 260 and 450 meters. A single
pelagic juvenile (35.5 mm SL, MCZ 48839)
was taken at 275 to 300 meters in the
Caribbean.

Geo<:,raphic variation. Statistical analyses
provide conflicting assessments of the
similarity of African and American popula-
tions. Meristic characters reveal little vari-
ability. Coefficients of difference calculated
for standard counts are always less than
or equal to 0.49 — far below conventional
levels of subspecies recognition. Mensural
data, on the other hand, suggest there are
considerable differences between the pop-
ulations. Of thirteen traits analyzed, seven
separate eastern and western populations
at the 95% level of confidence, five separ-

ate them at the 98% level, and two separate
them at the 99% level (Table 10).

A closer examination of the characters
exhibiting signilicant differences reveals
that regression coefficients of American
E. pandionis are always greater than those
of African forms. Since regression coef-
ficients are a measure of relative growth,
observed intraspecific variation may reflect
environmental factors.

Water temperature is a major parameter
determining growth rates in fishes. If other
factors are conti'olled, rates of growth in-
crease proportionally with temperature
(Brown, 1957: 391). With this in mind, it
is interesting that temperatures are gener-
ally higher and superficial warm-water
layers thicker in the western tropical At-
lantic (Ekman, 1953). At 300 meters Gulf
of Mexico and Caribbean temperatures vary
from 10 to 18° C while west African tem-
peratures range between 9 and 11° C. At
500 meters the difference is slightly less
pronounced — 8-13° C as opposed to 6-8°
C (from temperatin-e profiles in Fuglister,
1960; Wiist, 1964; and Nowlin and
McLellan, 1967). One would therefore
expect western Atlantic E. pandionus to
grow more rapidly and exhibit larger re-
gression coefficients than eastern Atlantic
forms. In view of these findings, the two
morphs are not considered to represent
separate subspecies.

Remarks. See E. trewavasae: Remarks
for discussion of E. pandionis .sensu Lozano
(1934), Navarro et al. (1943), and Maurin
(1968).

Specimens of doubtful identity. Five spec-
imens were examined that resembled E.
pandionis but could not, with certainty,
be placed in the species. Four were taken
in the Atlantic, the fifth in the Gulf of
Oman (see Mayer, 1972: Appendix II for
complete data). These fishes were not
considered when preparing the description
of E. pandionis, nor were they used in
morphometric, meristic, or distribution
analyses.

Tlie Atlantic specimens include three

168 Bulletin Museum of Comparative Zoology, Vol. 146, No. 3

Table 10. Comparison of regression coefficients from eastern and %vestern
Atlantic populations of E. pandionis. Data evaluated at the 95%, 98%, and

99% LE\rELS OF confidence. DF =: DEGREES OF FREEDOM; Eb = REGRESSION COEFFI-
CIENTS OF EASTERN ATLANTIC SPECIMENS; SD = SIGNIFICANT DIFFERENCE BETWEEN
TABULAR AND CALCULATED VALUES OF t; t=3 CALCULATED VALUES OF t; Wb = REGRES-
SION COEFFICIENTS OF WTESTERN ATLANTIC SPECIMENS.

Significance
Wb Eb DF t S S S

Body depth

Head height

Eye diameter

Snout length

Interorbital width

Maxillary leng

^th

Lower jaw len

gth

Caudal pedunc
depth

Caudal pedunc
length

D2I

All

P2I

fishes from St. Helena. The most recently for only the Caribbean form, which was

collected (UZM P45148) was incorrectly taken at relatively shallow depths. Exact

identified as E. telescopus by Banchot and determination of the variants' status must

Blanc (1961). The two older fonns (BMNH await the capture of additional material.

1868.3.11.14/15) are probably the fish The Indian Ocean form (BMNH

discussed by Giinther (1868). The re- 1889.4.15.24) is distinguished from E.

maining specimen (USNM 207703) was pandionis by its shallow body (22.5% SL),

taken in the Caribbean. narrow interorbital region (8.3% SL), den-

The four Atlantic individuals are basi- tigerous glossohyal, numerous weak oper-

cally similar to E. pandionis but exhibit cular spinelets, and elongate gill filaments,

shallower heads ( 17.4-19.8% SL), narrower The last trait suggests the fish may have

interorbital regions (8.6-9.4% SL), fewer inhabited an oxygen minimum layer. As

pyloric caeca (8-9), and fewer gill rakers with the Atlantic variants, additional ma-

( 25-27 ) . In these respects they resemble terial must be collected before the status of

E. fragilis. Little is known about the habits the form can be determined,

of the variants; station data are available Common names. None.

Epigonus Systematics • Mayer 169

Z^A

^
y

x»

Figure 10. Epigonus fragilis, HOLOTYPE, 89.1 mm SL, CM 3900/FMNH 55204 (from Jordan and Jordan, 1922).

Epigonus fragilis (Jordan and
Jordan, 1922)
Figure 10

Scepterias fragilis Jordan and Jordan, 1922: 45,
plate II, fig. 2 (original description; Honolulu
market; holotype examined, CM 3900/FMNH
55204).

?Hynnodus fragilis Pietschmann, 1930: 13.

Diagnosis. E. fragilis most closely re-
sembles E. pandionis but may be dis-
tinguished by its shallow body (18.8-21.1%
SL) and short, shallow head (length 31.7-
34.0% SL, height 16.0-17.4% SL). Unlike
£. pandionis, E. fragilis lacks peduncle
rings on specimens smaller than 100-120
mm SL.

In the past E. fragilis has been confused
with Hijnnodus atherinoides, a junior syn-
onym of E. occidentalis. E. fragilis may be
distinguished on the basis of body depth
( see above ) , pectoral fin counts ( 16-17 ) ,
and the absence of a pungent, bony oper-
cular spine. Weak opercular armor, to-
gether with second dorsal fin counts of 1,10
differentiate E. fragilis from E. treivavasae,
E. pectinifer, E. rohustus, E. lenimen, and
E. crassicaudus. Gill raker counts of 25-26
separate E. fragilis from all remaining con-
geners except E. telescopiis. E. fragilis may
be distinguished from the latter by the
presence of 7-8 pyloric caeca.

Description. E. fragilis is known from
only five specimens. Of these, the holotype
is of little descriptive value. The specimen
is severely dehydrated and has become
discolored, brittle, and shrunken. The fol-

lowing account is based primarily on two
recently captured specimens of E. fragilis
(LACM 32668-6 and USNM 207704) and
two forms collected by D. S. Jordan in 1921
(SU 23246). The latter are mentioned in
the original description of E. fragilis but
are not designated as types.

All meristic and mensural data are pre-
sented in the text. Detailed statistical
analyses were not undertaken because of
small sample size.

Body elongate; anterodorsal profile con-
vex, rising without interruption from tip
of snout to first dorsal fin. Body depth
18.8-21.1% SL; caudal peduncle length
25.4-26.9% SL.

Head short, 31.7-34.0% SL; head height
16.0-17.4% SL; snout blunt, 7.2-7.9% SL;
angle of gape moderate; jaws equal. Max-
illa reaching % eye length; posteriormost
point of maxilla at ventral edge of bone.
Eye round, 38.1—41.5% HL; anterodorsal
rim of orbit reaching profile; interorbital
width 8.8-9.4% SL.

Dentition variable with age. Teeth con-
ical; premaxillary teeth in irregular double
rows anteriorly, tapering to single row
posteriorly, occupying anterior %-% of
bone. Mandibular dentition more promin-
ent than that of premaxilla; teeth recurved,
occupying from % to entire length of
dentary, arranged in single or double rows
near symphysis and single row posteriorly.
Vomerine teeth recurved, arranged in oval
or diamond-shaped patch, covering entire
face of bone in adults. Palatine teeth

170 Bulletiti Museum of Comparative Zoology, Vol. 146, No. 3

Figure 11. Epigonus occidentalis, 152.7 mm SL, MCZ 48840.

medially recurved, arranged in single-triple
rows anteriorly, tapering to single row
posteriorly; tongue edentulous.

Opercular spine weak, ventral to 7-9
small serrae; angle of preopercle produced,
rounded, ornamented with striations and
weak serrations; subopercle and inter-
opercle unornamented. Gill rakers 25 (3),
26 (1), simple, awl-like. Pyloric caeca
7(1), 8(2).

First dorsal fin VII (4), VIII (1);
second dorsal fin 1,10 (5); anal fin 11,9
(5); pectoral fin 16 (1), 17 (3); DJ
moderate to long, 5.9-8.9% SL; DJ short,
6.9% SL; PJ long, 10.1-10.2% SL; All
broken.

Vertebrae 10 + 15 (4); epipleural ribs
not visible on radiographs; pleural ribs
8 (4), inserting on vertebrae 3-10. Pored
lateral line scales 49 ( 2 ) .

Color in alcohol yellow-brown; fin mem-
branes dark; iris silver-black; mouth light;
branchial membranes light, darkening with
age.

Distribution. E. fragilis is endemic to
the Hawaiian Islands (Fig. 12). The spe-
cies is demersal and has been taken between
120 and 125 meters.

Taxonomic notes. Six years after E.
fragilis was described, Fowler ( 1928 ) syn-
onymized the species with a second Ha-
waiian apogonid, Hijnnodus atherinoides
Gilbert, 1905. The synonymy achieved
moderate acceptance and appeared in sev-
eral publications (e.g., Matsubara, 1936;
Tinker, 1944; Gosline and Brock, 1960).
Fowler's conclusions were based on a

33-mm specimen (BPBM 3914) obtained
by the Tanager Expedition. The specimen
is in extremely poor condition. All colora-
tion has been lost, most of the muscle
tissue has decomposed, and much of the
skeleton has become decalcified. Although
it is impossible to identify the fish because
of its condition, the following traits suggest
it is neither E. fragilis nor H. atherinoides:
dorsal fin elements — VIII-1,8; anal fin ele-
ments— 11,6; vertebrae — 11 + 14. These
data differ from Fowler's report of VI-I,8
dorsal elements, no anal spines, and 7 anal
rays.

As was discussed in the diagnosis, E.
fragilis is distinct from H. atherinoides.
Fowler's synonymy appears to have been
based on inaccurate data taken from an
incorrectly identified fish.

Common names. None.

Epigonus occidentalis Goode
and Bean, 1896
Figure 11

Epigonus occidentalis Goode and Bean, 1896: 233,
plate LXVI, fig. 236 (original description;
Steamer BLAKE, off Barbados, 237 fms.; holo-
type examined, MCZ 28032 ) .

Hijnnodus atherinoides Gilbert, 1905: 618, plate
79 (original description; ALBATROSS Sta.
3867, Pailolo Channel, Hawaii, 284-290 fms.;
holotype examined, USNM 51601); Jordan
and Jordan, 1922: 44; Fowler and Bean, 1930:
121.

Hijnnodus megalops Smith and Radcliffe, 1912
{in Radcliffe, 1912): 445, plate 38, fig. 3
(original description; ALBATROSS Sta. 5388,
12°51'30"N, 123°26T5"E, between Bnrias and
Luzon, Philippines, 226 fms.; holotype ex-
amined, USNM 70255).

Ei'iGONus Systematics • Mayer 171

Table 11. Epigonus occidentalis meristic data. X = mean; SD

STANDARP nEVIMION; n = NUMBER OF SPECIMENS.

Range

Pectoral fin rays
Gill r aker s
Lateral line scales
Pyloric caeca

20.

19-2 1

0. 59

24.

22-27

1. 08

48.

46-51

0. 97

8-13

1. 05

Table 12. Epigonus occidentalis regression data, b = regression

COEFFICIENT ± 95% CONFIDENCE INTERVAL; a = Y INTERCEPT; n = NUMBER
OF SPECIMENS. AlL REGRESSIONS ON SL.

0. 72

Body depth

-1. 72

Head height

0. 53

Eye diameter

0, 66

Snout length

0. 06

Interorbital width

0. 83

Maxillary length

0. 88

Lower jaw length

1. 26

Caudal peduncle de

pth

-0. 90

Caudal peduncle len

igth

1. 50

D2 I

1. 42

All

2. 18

P2I

0. 67

Diagnosis. E. occidentalis is distin-
guished from all other congeners by the
combination of shallow body depth (14.1-
19.57t SL), reduced gill raker counts (22-
27), and the presence of a pungent, bony
opercular .spine. It is frequently confused
with E. denticulatus.

Description. Meristic values presented

in Table 11; regression data for morpho-
metric traits presented in Table 12.

Body elongate, cigar-shaped; anterodor-
sal profile weakly convex, flattened, rising
gradually from tip of snout to interorbital
region, leveling off toward occipital region,
and rising gradually to base of first dorsal
fin. Body depth 14.1-19.5% SL, body

172 Bulletin Museum of Comparative Zoology, Vol. 146, No. 3

width subequal to or greater than body
depth; caudal peduncle narrow, length
22.4-28.1% SL.

Head length 30.5-37.9% SL; head height
13.3-17.2% SL; angle of gape moderate to
small; lower jaw equalling or protruding
slightly beyond upper jaw. Maxilla reach-
ing Vi-% eye length; posterior margin
of maxilla moderate to narrow, posterior-
most point at ventral edge of bone. Eye
long, oval, 40.6-52.3% HL; anterodorsal
rim of orbit reaching or projecting into
dorsal profile; interorbital region narrow,
5.6-8.5% SL.

Teeth conical; premaxillary and man-
dibular teeth frequently recurved, arranged
in simple single row or single row widening
to double or triple rows near symphysis;
teeth covering % to entire length of pre-
maxilla and % to entire length of dentary;
vomerine teeth arranged in 1-4 irregular
rows; palatines rarely edentulous, teeth
1-10, arranged in single row, covering
anterior Vi-V^ of bone; tongue edentulous.

Opercular spine pungent, bony, ventral
to 1-3 poorly ossified spinelets; spine sep-
arated from spinelets by shallow indentation.
Preopercular angle produced, rounded or
pointed, bearing serrations and striations;
subopercle serrate, occasionally striate;
interopercle variable, frequently serrate.
Gill rakers short, awl-like.

First dorsal fin VII (59); second dorsal
fin 1,10 (59); anal fin 11,8 (1), 11,9 (59);
DJ, DJ, All, PJ short, equahing 1.1^.2%,
4.8-7.8%, 4.8-9.2%, and 8.0-11.3% SL re-
spectively.

Vertebrae 10 -I- 15 (35); epipleural ribs
6 (19), 7 (5), inserting on vertebrae 1-6
or 1-7 respectively; pleural ribs 7 (31),
8(1), inserting on vertebrae 2-9 or 3-9.

Color in alcohol variable with preser-
vation; skin frequently removed by trawl-
ing; underlying tissue pale yellow, yellow-
pink, occasionally marked with rust brown;
scale pockets and fin membranes black;
opercular area black-slate gray, occasion-
ally tinged with silver; lower jaw, bran-
chiostegal membranes, and thoracic and

abdominal regions occasionally silvered;
guanine most prevalent on specimens from
old collections. Mouth color variable with
age (see Ontogenetic change); iris and
branchial region dark.

Description based on 62 specimens 58.2-
178.9 mm SL.

Ontogenetic change. The most striking
age-related change in E. occidentalis is
the development of oral pigmentation. As
in E. telescopus and E. macro'ps, immature
forms bear pigmentless or slightly pig-
mented mouths, while adults have black-
ened oral membranes. Pigmentation first
appears in specimens 80-110 mm SL.
Melanophores develop just anterior to the
pharynx and spread rostrally, covering a
third of the roof and floor of the mouth and
half of the tongue by the time specimens
reach 115-130 mm SL. By 150 mm SL the
tongue is completely black, and by 175-
180 mm the entire mouth is dark. Branchial
membranes undergo an analogous trans-
formation before specimens reach 58 mm
SL.

A faint black ring circling the middle of
the caudal peduncle was observed on
three small E. occidentalis (< 65 mm SL).
Similar markings were absent from larger
individuals. The rings are reminiscent of
markings observed on young E. macrops
and E. pandionis and probably represent
a juvenile feature that is lost with growth.

Distribution. E. occidentalis has been
taken in the Caribbean, Gulf of Mexico,
and western tropical Atlantic. It is also
known from the Philippine and Hawaiian
Islands (Fig. 12). The species is caught
by bottom trawls between 360 and 735
meters. Adults are most abundant in the
Caribbean from 500 to 550 meters.

Geographic variation. E. occidentalis, as
here defined, includes two nominal species
— Hijnnodus atherinoides Gilbert, 1905 and
H. megalops Smith and Radcliffe, 1912.
The former originally represented a Ha-
waiian endemic; the latter represented a
Philippine form. In 1930 Fowler and Bean
synonymized the Pacific morphs. In the

Epigonus Systematics • Maxjer 173

Figure 12. Distribution of E. fragilis and E. occidentalis. Map A shows localities
in the Caribbean and Gulf of Mexico. Map B shows localities in the western
Pacific. J^ E. fragilis, individual haul of demersal adults; % E. occidentalis, indi-
vidual haul of demersal adults; cross-hatching indicates areas where E. occiden-
talis are frequently taken.

174 Bulletin Museum of Comparative Zoology, Vol. 146, No. 3

3 cm

Figure 13. Epigonus denticulatus, 115.1 mm SL, UMML 12463.

authors' opinions, characters separating the
two forms were "simply minor discrepancies
of portraiture and should never have been
credited as specific distinctions [p. 122] ."

Although descriptions and illustrations
of H. atherinoides and H. megalops suggest
a link with E. occidentalis, detailed com-
parisons of the three forms were never
made. To a large extent this was the re-
sult of inadequate sampling. Until the
initiation of the OREGON cruises in 1950,
few E. occidentalis were available for study.
Pacific forms are still poorly represented;
only seven specimens have been collected.
Reports of additional material by Fowler
(1928), Matsubara (1936), Smith ( 1949a,b,
1961), Kamohara (1952), and Moreland
( 1957) are based on misidentifications.

Comparisons of E. occidentalis and the
H. atherinoides-H . megalops complex pro-
vide no evidence to support their status
as separate species. Analyses of head
length, body depth, head height, eye di-
ameter, snout length, interorbital and max-
illary widths, caudal peduncle length and
depth, and All and PJ lengths reveal no
significant differences between the pop-
ulations at either the 95%, 98%, or 99%
levels of confidence. Meristic data also
show considerable overlap for most char-
acters; however, the coefficients of dif-
ference for pyloric caeca and gill raker
counts are above conventional levels of
subspecies recognition (1.68 and 1.99, re-
spectively). In addition, Atlantic and

Pacific populations may be distinguished
by minor qualitative characters such as:

(1) short, rounded preopercular angles
in Atlantic forms; longer, pointed
angles in Pacific specimens;

(2) fusion of uroneurals 1 and 2 in
Atlantic forms (based on 3 alizarin
preparations ) ; separate occurrence
in Pacific forms (based on 1 alizarin
preparation ) .

On the basis of the above information,
Atlantic and Pacific forms are placed in
the same species but considered members
of separate subspecies. Formal description
of the subspecies must await the capture
of additional Pacific specimens.

Remarks. A single unripe female E.
occidentalis (USNM 197353, 172.1 mm SL)
was found carrying small egg masses in the
anterior portion of its mouth (anterior to
the tongue and vomer). The masses con-
tained 125 oval eggs 0.40-0.55 mm in
diameter. The presence of eggs in the
mouth of an Epigonus is of interest, be-
cause several shallow-water apogonids ex-
hibit oral brooding. No such activity has
ever been reported for deep-sea forms.

Although it is difficult to say with
certainty, the E. occidentalis eggs are prob-
ably not incubating clutches, but rather
non-apogonid ova ingested during trawling.
Unlike the egg masses of typical oral
brooding apogonids, those found in E. oc-
cidentalis are broken, disrupted, and con-
tain very few eggs. An 84.9-mm specimen
of Cheilodipterus affinis was reported in-

Epigonus Systematics • Mayer 175

Table 13. Epigonus denticulatus meristic data. X = mean; SD

STANDARD DEVIATION; n = NUMBER OF SPECIMENS.

X Range SD n

Pectoral fin rays

19.

18-20

0. 56

Gill rakers

30.

28- 34

1. 10

Lateral line scales

48.

46- 49

0. 76

Pyloric caeca

11.

10- 14

0. 85

ciibating 21,000 eggs 0.35-0.4 mm in di-
ameter (Smith et al., 1971: 8-9). The ova
fully occupied the oral and branchial
chambers and extensively distended the
head. These conditions were not observed
in E. occidentalis.

It is possible that the eggs represent the
remnants of a larger mass that was spit
out and partially reingested. However,
were this the case, one might expect to
find eggs in the stomach (Sakomoto, 1930)
or gill rakers. No eggs were found in
either region.

Finally, Breder and Rosen (1966) state
that eggs of oral brooding apogonids are
lield together by fibers attaching to one
pole. The eggs of E. occidentalis are
loosely embedded in an open matrix of
fibers. Grape-like egg clusters character-
istic of Apogon semilineatus (Ebina, 1931:
20 ) were not observed.

Common names. None.

Epigonus denticulatus Dieuzeide, 1950
Figure 13

Pomatomus telescoptis, Vaillant (in part) (not

Risso, 1810), 1888: 376.
Scepterias lenimcn, Whitley (in part) (not Whit-
ley, 1935), 1935: 230; Whitley (in part),

1940: 420.
Epigonus atherinoides, Matsubara (not Gilbert,

1905), 1936: 120, fig. lA; Smith, 1961: 378,

fig. 3; Kamohara, 1952: 37, fig. 31.
Hynnodus atherinoides. Smith (not Gilbert, 1905),

1949a: 101; Smith, 1949b: 210, fig. 495A.
Epigonus denticulatus Dieuzeide, 1950: 89, figs.

1-2 (original description; Algerian Coast at

200-500 m; holotype not examined); Tortonese,
1952: 72, 1 fig.; Dieuzeide et al., 1953: 216,
2 figs.; Tortonese and Queirolo, 1970: 33,
fig. 6.

Diagnosis. E. denticulatus lacks a fully
ossified opercular spine, bearing instead
3-7 membranous projections. This feature
distinguishes it from E. occidentalis, E.
treivavasae, E. pectinifer, E. rohustus, E.
lenimen, and E. crassicaudtis, which have
pungent, bony opercular spines. E. denti-
culatus is differentiated from E. telescopus,
E. macrops, and E. fragilis by the presence
of 10-14 pyloric caeca and 28-34 gill
rakers. It differs from E. oligolepis by
bearing 46-51 lateral line scales. E. denti-
culatus closely resembles E. pandionis but
may be distinguished on the basis of the
former's shallow body (15.8-23.67r SL),
long caudal peduncle (25.9-32.2% SL),
and short DJ (2.4-3.7% SL).

Description. Meristic values presented
in Table 13; regression data for morpho-
metric traits presented in Table 14.

Body fusiform, slightly compressed; an-
terodorsal profile rising gradually above
snout, becoming steeper and slightly con-
vex over eyes, thereafter rising gradually
to first dorsal fin; body moderate to shal-
low, depth 15.8-23.6% SL; caudal peduncle
narrow, length 25.9-32.2%o SL.

Head moderate to short, 31.2-38.6% SL;
head height 16.0-19.8% SL; snout short,
blunt; angle of gap(> moderate to large;
lower jaw protruding slightly beyond up-
per jaw. Maxilla reaching %-% eye length,

176 Biilletifi Museum of Comparative Zoology, Vol. 146, No. 3

Table 14. Epigonus denticulatus regression data, b = regression

COEFFICIENT ± 95% CONFIDENCE INTERVAL; a = Y INTERCEPT; 11 = NUMBER
OF SPECIMENS. AlL REGRESSIONS ON SL.

Body depth

-4.

Head height

Eye diameter

Snout length

Interorbital width

Maxillary length

Lower jaw length

Caudal peduncle d

epth

, 01

-0.

Caudal peduncle li

ength

, 01

D2I

, 01

All

, 06

, 01

, 43

P2I

, 08

, 01

, 92

posteriormost point near ventral surface
of bone. Eye round or slightly oval, 40.3-
48.0% HL; anterodorsal rim of orbit
reaching dorsal profile, projecting into pro-
file in smaller specimens; interorbital
width 8.2-10.4% SL.

Teeth small, conical, occasionally re-
curved; premaxilla bearing single row of
teeth along anterior Vs-% (usually %) of
bone. Mandibular teeth arranged along
length of dentary in irregular single row,
occasionally double near symphysis; larger
specimens with 3-4 rows near symphysis.
Vomerine teeth variable, arranged in 1-4
irregular rows. Palatine dentition occupy-
ing length of bone, arranged in simple
single row or double row tapering to single
row posteriorly; large specimens bearing
3-4 rows of teeth anteriorly. Tongue gen-
erally edentulous, rarely Ijearing isolated
tooth patches on glossohyal or edges of
tongue.

Opercle lacking bony spine, bearing in-

stead 3-7 (usually 5-6) jagged, mem-
branous projections; projections often ob-
scured by underlying tissues. Peropercular
angle produced, broadly rounded, striations
radiating from inner edge, angle occasion-
ally serrate; subopercle and interopercle
occasionally serrate. Gill rakers simple,
awl-like.

First dorsal fin VII (53); second dorsal
fin 1,9 (1), 1,10 (56), 10 (1); anal fin 11,8
(1), 11,9 (57). DJ moderate, 2.4-3.7%
SL; Dol, All, P,I short, 5.2-8.0%, 6.0-8.2%,
7.9-10.0% SL respectively.

Vertebrae 10 + 15 (44); epipleural
ribs 6 ( 32 ) , 7 ( 1 ) , inserting on vertebrae
1-6 or 1-7 respectively; pleural ribs 8 ( 44 ) ,
inserting on vertebrae 3-10.

Color in alcohol variable with preserva-
tion; skin frequently removed by trawling,
underlying tissue pink-brown or yellow;
scale pockets mottled with numerous
brown-l)lack melanophores, dorsal surfaces
of body and head more heavily pigmented.

Epigonus Systematics • Mayer 177

Guanine deposits frequently occurring on
gill cover, ventral surface of mandible,
isthmus, thoracic region, and abdomen to
anus; iris black; mouth light; branchial
region dark.

Description based on 58 specimens 57.0-
187.5 mm SL.

Ontogenetic change. Two young spec-
imens of E. dcnfictihifus (29.2 mm SL,
MCZ 48846, and 49.7 mm SL, MCZ 48847)
were examined in the course of this in-
vestigation. These specimens were taken
by midwater trawls made in the central
North Atlantic and Gulf of Mexico and
reveal that the life cycle of E. denticulatus
includes a pelagic juvenile stage.

The pelagic young resemble adults in
most respects. For example, the juveniles
bear diagnostic gill raker counts and
opercular ornamentation. However, slight
changes in body shape are associated with
growth. The 29.2 mm specimen has a more
shallow body, shorter head, narrower inter-
orbital region, and smaller eyes than
demersal adults. Similar trends are present
but less apparent in the larger juvenile.

Juvenile dentition patterns are basically
like those of adults but involve fewer and
relatively larger recurved teeth. Oral and
branchial regions are light in young speci-
mens. The latter areas darken with age.

Distribution. E. denticulatus is the only
cosmopolitan species in the genus (Fig.
14). Specimens have been taken from the
southwest coast of Japan, the Gulf of
Mexico, and the Caribbean. In addition,
the species occurs continuously from the
western Mediterranean, south along the
western coast of Africa to the tip of the
continent. It reappears south of the Great
Australian Bight and southeast of New
Zealand.

Adults are generally taken by bottom
trawls between 300 and 600 meters, al-
though specimens have been captured from
as shallow as 200 meters and as deep as 830
meters. Pelagic juveniles have been taken
by IKMT between 130 to 145 meters and
350 to 425 meters.

Geographic variation. E. denticuhitus
may be divided into North Atlantic, South-
ern Hemisphere, and Japanese populations.
North Atlantic forms include material from
the Mediterranean, northeast Atlantic,
Caribbean, and Gulf of Mexico. Southern
Hemisphere populations contain specimens
from the southeast Atlantic, Australia, and
New Zealand.

Statistical analyses reveal surprisingly
little divergence between North Atlantic
and Southern Hemisphere specimens. Co-
efficients of difference for standard meristic
characters are far below accepted levels
for subspecies recognition (all are ^ 0.53),
and regression coefficients for mensural
data are virtually identical. Only maxil-
lary lengths differ significantly at the 95%
level of confidence. It is clear from the
data that North Atlantic and Southern
Hemisphere E. denticulatus do not repre-
sent separate subspecies.

Detailed analyses of the Japanese pop-
ulation could not be undertaken because
of inadequate sampling. Only one speci-
men was available from the area. On the
basis of this fish, the Japanese population
appears closely allied to the rest of the
species. With the exception of eye di-
ameter, standard counts and measm-ements
made on the Japanese morph fall within
the 95% and 99% confidence intervals of
remaining E. denticulatus. Eye diameter
falls outside the 95% confidence interval
but within the 99% confidence interval.

The similarity of E. denficuhiius pop-
ulations, despite the wide rangc> of the
species, suggests ( 1 ) there may be con-
siderable gene flow between populations,
(2) the present distribution may have been
achieved only recently, or (3) evolution
is occurring very slowly. Discovery of a
pelagic juvenile in the mid-North Atlantic
gives credence to the first hypothesis
and proN'ides a mechanism for the dis-
persal of a species with demersal adults
such as E. denticulatus.

Common names. "Castagnera briina" in
Monaco (Bini, 1968).

178 Bulletin Museum of Comparative Zoology, Vol. 146, No. 3

E
o

O)
Q.

3
CO

3
T3

3
CO

•a
'>

■D

c
o

ii.

Epigonus Systematics • Mayer 179

Epigonus oligolepis sp. nov.
Figure 15

llolotype: One specimen, 90.8 nun SL, taken
from the Straits of Florida bv M/V COMBAT,
Sta. 436: 21 July 1957, 1319 to 1530 hrs.;
24°13'N, 81°42'W; 300 fnis., 10' flat trawl.
USNM 207718.

Parat>pes: One specimen, 126.7 mm SL, M/V
OREGON, Sta. 4731: 27 February 1964;
27°35'N, 92°32'W; 250-300 fms.; 40' flat
trawl. MCZ 48848.

Three specimens, 52.7-72.7 mm SL, Steamer
ALBATROSS, Sta. 2643: 9 April 1886;
25°25'00"N, 79°55'15"W; 211 fms. USNM
109430.

Three specimens, 53.7-84.2 mm SL, M/V
OREGON, Sta. 5043: 26 September 1964;
12°01'N, 6P53.5'W; 210-250 fms.; 40' shrimp
trawl. USNM 207719.

One specimen (cleared and stained), 62.0
mm SL, locality data identical with those of
preceding lot. USNM 207720.

One specimen, 117.1 mm SL, M/V OREGON,
Sta. 3741: 26 August 1962; 29°10'N, 88°01.5'
W; 300-340 fms.; 100' flat trawl. USNM
207721.

Diagnosis. E. oligolepis is distinguished
from all congeners by lateral line scale
counts of 33-36 and the presence of lingual
and endopterygoid teeth.

Description. Meristic values presented in
Table 15; regression data for morphometric
traits presented in Table 16.

Body elongate, moderately compressed;
anterodorsal profile rising gradually from
tip of snout to interorbital region, rising
more steeply and becoming slightly convex
to occiput, thereafter rising gradually to
base of first dorsal fin; body depth 19.8-
24.5% SL; caudal peduncle length 23.9-
27.2% SL.

Head moderate to long, 34.4-43.0% SL;
head height 16.6-18.8% SL; snout pointed;
angle of gape moderate; lower jaw pro-
truding beyond upper jaw. Maxilla reach-
ing %-% eye length; posterior margin
of maxilla rounded, posteriormost point
between midline and ventral margin of
bone. Eye round to slightly oval, 40.1-
43.77o HL; anterodorsal rim of orbit reach-
ing or projecting into dorsal profile; inter-
orbital width 8.5-9.6% SL.

Teeth small, conical; premaxilla edent-
ulous or bearing few teeth on anterior Vi-
-f. of bone; mandibular teeth arranged in
single or double row antericnly, single row
posteriorly; teeth covering anterior half
of bone and occasionally extending along
length of dentaiy. Vomer covered with
irregular tooth patches, teeth extending
posteriorly along midline of palate; pala-
tine teeth arranged in single or multiple
rows anteriorly, single row posteriorly,
covering from half to entire length of bone;
endopterygoid dentigerous; auxiliary tooth
patches occurring between vomer, pala-
tines, and endopterygoids; tongue den-
tigerous, bearing lateral and glossohyal
tooth patches (Fig. lA).

Opercular spine weak, poorly ossified,
ventral to 2-6 membranous spinelets; spine
and spinelets separated by moderate gap;
spinelets occasionally obscured by under-
lying membranes. Preopercular angle rec-
tangular or slightly produced; preopercle,
subopercle and interopercle unserrated.
Gill rakers simple, awl-like.

First dorsal fin VII (10); second dorsal
fin 1,10 (10); anal fin 11,8 (1), 11,9 (9).
Fin spines moderate; DJ 2.7-4.0% SL;
D,I 10.9-12.1% SL; All 10.3-12.2% SL;
PJ 11.0-13.6% SL.

Vertebrae 10 + 15 ( 10 ) , epipleural ribs
7 (4), 8 (1), inserting on vertebrae 1-7
or 1-8 respectively; pleural ribs 7 (10),
inserting on vertebrae 3-9.

Color in alcohol variable with preserva-
tion; specimens frequently abraded reveal-
ing underlying pale yellow or pink-purple
tissue. Recently collected specimens bear
scale pockets mottled with numerous
melanophores; dorsal surfaces of head and
trunk more heavily pigmented; iris black.
Specimens from old collections devoid of
melanin, bearing silver on opercular region,
isthmus, thoracic region, and abdomen to
anus; iris silver. Mouth light, dotted with
brown or black melanophores; l)ranchial
region light in small specimens, darkening
with age.

180 Bulletin Museum of Comparative Zoology, Vol 146, No. 3

r-
o

E
E

o
c>

>
\-
O

Q.
O

3
C
O

Epigonus Systematics • Mayer 181

Table 15. Epigonus oligolepis meristic data. X = mean; SD = stan-
dard DEVIATION; n = NUMBER OF SPECIMENS.

Range

Pectoral fin rays 17.20 16-18 0.79 10

Gill rakers 30.50 29-31 0.71 10

Lateral line scales 34.70 33-36 1.06 10

Pyloric caeca 8.83 8-10 0.75 6

Table 16. Epigonus oligolepis regression data, b = regression coef-
ficient ± 95% confidence interval; a = Y intercept; n = number of
specimens. All regressions on SL.

1 1

Body depth

-2.

Head height

-2.

Eye diameter

Snout length

Interorbital width

-0.

Maxillary length

-0.

Lower jaw length

Caudal peduncle d

epth

-1.

Caudal peduncle L

e n g t li

D2I

-1.

All

-0.

P2I

Description based on 10 specimens 53.7-
126.7 mm SL.

Ontogenetic change. Two juvenile E.
oligolepis (32.0-32.2 mm SL, USNM 207722)
were taken by bottom trawls from
the Gulf of Mexico. These specimens
exhibit many traits characteristic of adult
forms but differ in head shape, meristics.

and dentition. Unlike adults, young E.
oligolepis have smaller eyes (38.2-39.4%
HL) and wider interorbital regions (10.4%
SL). Dorsal fin and gill raker counts are
reduced to VI-I,10 and 26 respectively.
Premaxillary, mandibular, and lingual tooth
patterns are similar to those of mature
individuals, but dentition associated with

182 Bulletin Museum of Comparative Zoology, Vol. 146, No. 3

Figure 16. Distribution of E. oligolepis. ■ individual haul of demersal adults; □ individual
haul of demersal juveniles.

the roof of the mouth is strongly reduced.
Vomers and palatines are edentulous or
bear 1-4 teeth; auxiliary tooth patches have
not developed. Endopterygoid teeth are
present but few in number, relatively long,
and medially recurved.

Distribution. E. oligolepis is endemic to
the Gulf of Mexico-Caribbean region (Fig.
16). Specimens have been taken by bot-
tom trawls between 380 and 660 meters.

Remarks. The type specimens of E.
oligolepis exhibit two seemingly disparate
color patterns. One lot, taken in 1886 by
the ALBATROSS, is devoid of melanin but
bears extensive guanine deposits. Remain-
ing fish, all more recently collected, bear
no silver but are dotted with numerous
melanophores. These differences are arti-
facts of preservation.

Specimens collected by early workers
were generally placed directly into ethanol,
while material obtained today is fixed in
10 percent formalin (Hubbs and Lagler,

1958: 16-17). When ethanol is used as a
fixative, it leaches out melanins but does
not affect guanine deposits. Specimens
become pale, but silver pigment is retained.
Formalin has the opposite effect; it
blackens melanophores but destroys gua-
nine crystals. The appearance of preserved
specimens is thus dependent on fixative
composition, concentration, and immersion
time. An alcohol-formalin mixture con-
taining one tablespoon of full strength
formalin per two gallons of 6.5-75 percent
ethanol might be used instead of conven-
tional fixatives to preserve both guanine
and melanin deposits (Myers, personal
communication ) .

Etymology. Oligolepis (Greek), few
scales, from oligos, few, and lepis, scale;
a noun in apposition, refers to the reduced
number of lateral line scales characterizing
the species.

Common names. None.

Epigonvs Systematics • Mayer 183

Figure 17. Epigonus trewavasae, 98.6 mm SL, USNM 207723.

Epigonus trewavasae Poll, 1954
Figure 17

Glossamia pandionis, Lozano (not Goode and

Bean, 1881), 1934: 89; Navarro, 1942: 202;

Navarro et al., 1943: 136, plate XXII, fig. A.
Epigonus trewavasae Poll, 1954: 91, fig. 27

(original description; NOORDENDE III Sta.

52, 06°08'S, 11°30'E, 280-290 m; holotype

examined, IRSN 209).
Epigojitis pandionis, Maurin (not Goode and

Bean, 1881), 1968: 69, fig. 36.

Diagnosis. E. trewavasae is most likely
to be confused with E. robustus, E. leni-
men, E. crassicaudus, and E. pectinifer. It
is distinguished from the first three species
by vertebral counts of 10 + 15 and the
presence of glossohyal and lateral lingual
teeth. The fourth form, E. pectinifer, bears
only glossohyal teeth or a totally eden-
tulous tongue. E. trewavasae may be
further differentiated from E. pectinifer on
the basis of the former's 30-35 awl-like
gill rakers and long, pungent Dol and All
(12.7-16.5% SL, 13.8-16.8% SL respec-
tively). E. trewavasae is unlike remaining
congeners because it bears a pungent, bony
opercular spine, second dorsal fin counts
of 1,9, and pectoral fin counts of 16-18.

Description. Meristic values presented
in Table 17; regression data for morpho-
mctric traits presented in Table 18.

Body elongate; anterodorsal profile flat,
rising without interruption from snout to
base of first dorsal fin; body moderate to
deep, 23.1-27.0% SL; caudal peduncle
length 24.3-27.5% SL.

Head length 33.7-38.1% SL; head height
16.6-18.7%^' SL; snout pointed; angle of
gape small to moderate; lower jaw pro-
truding beyond upper jaw, bearing two
nubs on anterior surface of mandible.
Maxilla reaching slightly less than % eye
length; posterior margin of maxilla narrow,
rounded, or bearing posteriormost point
near midline of bone; short, pungent mus-
tache-like process projecting from postero-
ventral surface of maxillary head. Eye
round, slightly oval in younger specimens,
41.1-49.1%f HL; anterodorsal rim of orbit
reaching profile; interorbital width 8.8-
10.8% SL.

Dentition variable with age (see Onto-
genetic change); teeth conical, small, fre-
quently microscopic, present on premaxiL
lae, mandibles, and vomer; palatines
occasionally edentulous; tongue bearing
lateral and glossohyal tooth patches.

Opercular spine pungent, bony, sur-
mounted by 2-3 horny spinelets; spine and
spinelets separated by large gap; spinelets
often obscured by underlying opercular
membranes. Preopercular angle narrowly
produced, unserrated or bearing serrations
on angle and ventral surface of bone; in-
teropercle and subopercle unserrated or
weakly serrated. Gill rakers simple, awl-
like.

First dorsal fin VII (14); .second dorsal
fin 1,9 (13), 1,10 (1); anal fin 11,9 (14);
DJ moderate, 2.4-3.2% SL; DJ, All, PJ,

184 Bulletin Museum of Comparative Zoology, Vol. 146, No. 3

Table 17. Epigonus trewavasae meristic data. X = mean; SD

STANDARD DEVIATION; n =: NUMBER OF SPECIMENS.

Range SD

Pectoral fin rays
Gill rakers
Lateral line scales
Pyloric caeca

17. 54 16- 18 0. 66 13

33. 15 30- 35 1. 46 13

47. 69 47- 49 0. 75 13

7. 00 6- 8 0. 60 12

Table 18. Epigonus trewavasae regression data, b = regression co-
efficient ±95% confidence interval; a ^ Y intercept; n =: number of
specimens. All regressions on SL.

0. 38

0. 03

-2. 03

Body depth

0. 29

0. 02

-4. 43

Head height

0. 19

0. 01

-1, 26

Eye diameter

0. 17

0. 02

-0, 49

Snout length

0. 07

0. 02

1. 19

Interorbital width

0. 09

0, 01

1. 41

Maxillary length

0. 15

0. 02

0. 69

Lower jaw length

0. 16

0. 01

0. 39

Caudal peduncle de

pth

0. 13

0. 01

-1. 79

Caudal peduncle len

igth

0, 26

0. 02

-0. 08

D2 I

0, 15

0. 03

-0. 32

All

0. 18

0. 03

-0. 65

P2I

0, 14

0. 01

0. 36

long, pungent, 12.7-16.5%, 13.(8-16.8%,
13.8-16.27^ SL respectively.

Vertebrae 10 + 15 (12); epipleural ribs
6 (9), 7 (2), inserting on vertebrae 1-6
or 1-7 respectively; pleural ribs 7 (8), 8
(4), inserting on vertebrae 3-9 or 3-10
respectively.

Color variable with presei'vation; speci-
mens abraded, revealing underlying yel-

low to yellow-pink tissue; fin membranes
dark; scale pockets covered with dense
brown or black melanophores; dorsal sur-
face of trunk more heavily pigmented than
ventral; opercles brown, black, or slate
gray; guanine deposits occurring occasion-
ally on opercular region and from isthmus
to bases of paired fins; iris black with sil-
ver highlights; mouth light; branchial re-

Epigonus Systematics • Mayer 185

JO' SO"

Figure 18. Distributions of E. trewavasae and E. pectinifer. Large map shows localities in the Atlantic; insert
shows localities off Japan. E. trewavasae: ^ individual haul of adults; Q individual haul of juveniles; cross-
hatching indicates areas of capture cited in the literature. £. pectinifer: ■ individual haul of adults; □ individ-
ual haul of juveniles; A report from the literature.

gion light in smiill specimens, becoming
l)lack with age.

Description based on 13 specimens 70.9-
153.9 mm SL.

Ontogenetic change. The most striking
ontogenetic changes in E. trewavasae are
associated with the development of adult
tooth patterns. Large specimens bear ir-
regular double or triple rows of premaxil-
lary and mandibular teeth that taper to a
single row posteriorly. Vomers are covered
with minute conical teeth, while palatines
are either edentulous or bear single to
double rows of teeth.

Dentition patterns are simple in small
specimens but become more complex as
teeth are added during growth. A 29.8-
mm juvenile lacks both premaxillary and
mandibular teeth. By 70-75 mm SL teeth
are present in single rows on the jaws, and
by 145 mm SL adult tooth patterns pre-
vail. As premaxillary tooth patches widen,
they extend posteriorly and eventually

cover the first half of the bone. Analogous
expansion occurs in vomerine tooth
patches.

Distribution. E. tretcavasae is known
from equatorial west Africa, northwest
Africa, and the western Mediterranean
(Fig. 18). It has been taken by bottom
trawls between 200 and 600 meters.

Geographic variation. Statistical com-
parisons of African and Mediterranean E.
trewavasae were not made because of
small sample size. As additional material
is collected, the following intraspecific
differences should be examined:

( 1 ) vomerine and palatine teeth more
strongly developed in Mediterra-
nean forms;

(2) chin nubs more strongly developed
in African forms;

(3) preopercular serrations more
strongly developed in Mediterra-
nean forms.

Although the significance of thc\se features

186 Bulletin Museum of Comparative Zoology, Vol. 146, No. 3

is unknown, they suggest that African and
Mediterranean forms may represent sepa-
rate subspecies.

Taxonomic notes. Pomatomichthys con-
stonciae GigHoh, 1880 may be a synonym
of E. trewavasae Poll, 1954. See E. tele-
scopus: Taxonomic notes, for a discussion
of this possibility.

Remarks. Dieuzeide (1950: 104-105)
reported that specimens designated as
Glossamia pandionis hy hozano (1934) and
NavaiTO et al. (1943) were actually mis-
identified E. denticulatus. This is incor-
rect. Lozano's report is based on a single
specimen (131 mm total length) taken
from the Catillian coast. Among the char-
acters cited for this fish are dorsal fin
counts of VII-1,9, pectoral counts of 16,
and an All subequal to the eye diameter
(p. 89). All of these are characters diag-
nostic of E. trewavasae. E. denticulatus
bears 10 rays in the second dorsal fin, 18-
20 pectoral rays, and an All equalling half
the eye diameter.

Navarro et al.'s specimens also appear
to be E. trewavasae. Altliough no descrip-
tion is provided, the account includes a
photograph (plate XXII, fig. A) that shows
the fish have deep bodies, pungent oper-
cular spines, and long D2I, All, and P2rs.
All of these features are characteristic of
E. trewavasae.

More recently, Maurin (1968) mistook
E. trewavasae for E. pandionis. Propor-
tional measurements of body depth, head
height. All, and P-I made on Maurin's
figure 36 (p. 69) fall within ranges char-
acteristic of E. treioavasae; however, pub-
lished gill raker counts of 28-30 (p. 70)
are lower than expected.

Common names. None.

Epigonus pectin! fer sp. nov.
Figure 19

Ilolotype: A 114.3-mm SL specimen taken from
the Caribbean west of Grenada by M/V
OREGON, Sta. 5043: 26 September 1964,
12°01'N, 61°53.5'W, 210-250 fms., 40' shrimp
trawl. USNM 207725.

Paratypes: One specimen, 97.4 mm SL, 16
September 1964, Suruga Bay, commercial trawl.
ABE 64-2085.

One specimen, 100.6 mm SL, 14-31 October
1964, Suruga Bay, commercial trawl. ABE
64-2245.

One specimen, 99.8 mm SL, 14-31 October
1964, Suruga Bay, commercial trawl. ABE
64-2248.

Two specimens, 95.2-117.1 mm SL, station
data identical with those of holotype. MCZ
48850.

One specimen (cleared and stained), 108.1
mm SL, station data identical with those of
holotype. MCZ 48851.

One specimen, 94.8 mm SL, R/V PILLS-
BURY, Sta. P-582: 23 May 1967; 21°10'N,
86°18'W; 250-155 fms.; 10' otter trawl. UMML
30378.

One specimen, 111.2 mm SL, M/V OREGON,
Sta. 4405: 27 September 1963; 11°53'N,
69°28"W; 215 fms.; 40' flat trawl. USNM
207726.

Ten specimens, 101.8-120.6 mm SL, station
data identical with those of holotype. USNM
207727.

Nine specimens 81.5-118.9 mm SL, station
data identical with those of holotype. USNM
207728.

Two specimens (cleared and stained), 94.8-
98 mm SL, station data identical with those
of holotype. USNM 207729.

Epigonus rohiistiis, Matsubara (not Barnard,
1927), 1936: 121, fig. IB; Kamohara, 1952:
37.

Diagnosis. E. pectinifer is characterized
by comb-like gill rakers on the lower half
of the first gill arch. This feature, together
with glossohyal dentition (present in most
specimens) and vertebral counts of 10 +
15, differentiate E. pectinifer from E. ro-
htistus, E. lenimen, and E. crassicaudus.
E. pectinifer most closely resembles E. tre-
ioavasae but is distinguished by less exten-
sive lingual dentition, fewer gill rakers
(26-30), and shorter DJ and All (11.2-
12.7% SL and 11.9-14.0% SL respectively).
E. pectinifer may be separated from re-
maining congeners by its pungent, bony
opercular spine, second dorsal fin counts
of 1,9, and pectoral fin counts of 15-18.

Description. Meristic values presented
in Table 19; regression data for morpho-
metric traits presented in Table 20.

Epigonus Systematics • Mayer 187

E
E

UJ
Q.

O
V>
Q.

C
O

188 Bulletin Museum of Comparative Zoology, Vol. 146, No. 3

Table 19. Epigonus pectinifer meristic data. X = mean; SD = stan-
dard DEVIATION; n = NUMBER OF SPECIMENS.

Range SD

Pectoral fin rays 16.03 15-18 0.57 29

Gill rakers 27.59 26-30 0.98 29

Lateral line scales 48.14 47-49 0.58 29

Pyloric caeca 6, 10 5- 7 0. 41 29

Table 20. Epigonus pectinifer regression data, b = regression coef-
ficient ± 95% confidence interval; a = Y intercept; n = number of
specimens. All regressions on SL.

-2.

Body depth

-4.

Head height

-2.

Eye diameter

-3.

Snout length

Interorbital width

-1,

Maxillary length

-2.

Lower jaw length

Caudal peduncle de

pth

C audal peduncle length

-0.

D2I

All

P2I

Body elongate; anterodor.sal profile flat
or slighdy convex, rising withont interrup-
tion from snout to base of first dorsal fin;
body depth 21.1-24.6% SL; caudal pedun-
cle narrow, length 2.5.1-28.7% SL.

Head short to moderate, 31.3-35.7% SL,
shallow, 14.2-16.9% SL; snout wide,
pointed; angle of gape small; lower jaw
proti-uding slightly beyond upper jaw;

nubs at anterior end of mandible paired,
barely discernible, or absent. Maxilla
reacliing %-y2 eye length, posterior margin
narrow, rounded, or bearing posteriormost
point near midline of bone; short, pun-
gent, mustache-like process projecting
from posteroventral surface of maxillary
head. Eye round or slightly oval, 38.7-
45.4% HL; anteiodorsal rim of orbit not

Epigonus Systematics • Maijer 189

reaching profile; intcrorl)ital widtli 7.7-
9.4% SL.

Teeth small, conical; premaxilla edentii-
lons or bearing teeth anteriorly; when
present, teeth 1-15, arranged in single
row. Mandibular teeth covering all or part
oi dentary, arranged in single row. Vomer-
ine teeth strong, arranged in tightly packed
()\'al patch. Palatines edentulous or bear-
ing teeth anteriorly; when present, teeth
1-6, arranged in single row; tongue with
glossohyal teeth, rarely edentulous.

Opercular spine pungent, bony, ventral
to 2-3 horny spinelets; spine and spinelets
separated by large gap; spinelets occasion-
ally obscured by underlying membranes.
Preopercular angle narrowly produced,
serrated; subopercle and interopercle un-
serrated or weakly serrated. Gill rakers
pectinate, bearing nub-like projections
proximally along mesial surfaces (Fig.
IC); pectinate structure variable in extent,
most prominent on ventral portions of gill
arch.

First dorsal fin VII (28); second dorsal
fin 1,9 (29); anal fin 11,9 (29); D,I
short, 1.6-2.8% SL; D,I moderate, 11.2-
12.7% SL; All, PJ, 11.9-14.0% SL.

Vertebrae 10 + 15 (29); epipleural ribs
6 (17), 7 (13), inserting on vertebrae 1-6
or 1-7 respectively; pleural ribs 8 (29),
inserting on vertebrae 3-10.

Color in alcohol brown-black; fin mem-
branes black; scale pockets covered with
densely packed melanophores; skin often
abraded, revealing underlying yellow-pink
tissue; iris black; branchial region white to
dark gray; mouth light.

Description based on 30 specimens 81.5-
120.6 mm SL.

OntO(!,enetic change. A 33.8-mm E. pec-
tinifer was taken by bottom trawl in the
Gulf of Mexico (USNM 207731). The
specimen appears similar to adults and pro-
vides little evidence of ontogenetic change.
The major difference is the presence of six
rather than seven first dorsal fin spines.

Distribution. E. pectinifer is known
from the Caribbean Sea, Gulf of Mexico,

and eastern coast of Japan (Fig. 18).
Specimens were taken between 280 and
550 meters.

GeograpJiic variation. Definitive com-
parisons of Japanese and American E. pec-
tinifer were not undertaken, because only
three oriental specimens were available for
study. The latter forms were, however, in-
dividually compared with Amcnican fish.
The analyses revealed virtually no differ-
ences between the populations aside from
a slight tendency toward broader caudal
peduncles and shorter maxillae and man-
dibles by the Japanese specimens.

Remarks. A teratological specimen of
E. pectinifer was taken from the Yucatan
Channel (109 mm SL, UMML 30379). The
fish was captured at depths characteristic
of E. pectinifer and bears diagnostic traits
such as 27 gill rakers ( many are pectinate ) ,
VII + I dorsal fin spines, 16 pectoral fin
rays, and 10 + 15 vertebrae. The tongue
is edentulous. Unlike the condition in
typical forms, opercles are not fully ossified
and lack spines and spinelets. Similarly,
the lateral line is incomplete on the right
side and bears only 43 pored scales on the
left. Other differences include enlarged
teeth and chin nubs, 10 rather than 9 dor-
sal rays, and 8 rather than 5-7 pyloric
caeca.

The aberrant specimen was not consid-
ered in preparing the species description.

Etymology. Pectinifer (Latin), comb-
bearer, from pecten, comb, and ferare, to
bear; a noun in apposition, refers to the
comb-like gill rakers characterizing this
species.

Common names. None.

Epigonus robustus (Barnard, 1927)
Figure 20

Epigonus macrops Gilchrist and von Bonde, 1924:
14, plate I, fig. 3 (oiij^inal description; S..S.
PICKLE Sta. 344, 30°12'00"S, 14°25'()()"E,
510 fms.; Sta. 347, 31°58'00"S, 16°00'00"E,
670 fms.; syntype examined, RUSI 669; name
suppressed, junior homonym of Oxi/odot} macrops
Brauer, 1906); Barnard, 1927:' 523; Smith,
1961: 377, fig. 2.

190 Bulletin Museum of Comparative Zoology, Vol. 146, No. 3

Figure 20. Epigonus robustus, 154.6 mm SL, LACM 11449-7.

Parahynnodiis robustus Barnard, 1927: 525, plate
XXII, fig. 4 (original description; off Cape
Point, 460 fms.; holotype in poor condition, not
examined, SAM 13080).

Hymwdus robustus Smith, 1949b: 210, fig. 495.

Diagnosis. E. robustus sti-ongly resem-
bles E. pectinifer, E. trewavasae, and E.
lenimen. It may be distinguished from
the former two species by the absence of
hngual teeth. In addition, mihke E. pec-
tinifer, it has awl-Hke gill rakers. E. ro-
hu^us differs from E. lenimen by having
a narrow interorbital region (6.5-8.2%
SL), short DJ (10.0-12.6% SL) and short
All (9.2-13.3% SL). E. robustus may be
distinguished from E. crassicaudus by the
former's short head (28.0-34.0% SL) and
shallow body (20.3-24.6% SL). It differs
from remaining congeners by bearing a
pungent, bony opercular spine, vertebral
count of 11 + 14, and nine rays in the
second dorsal fin.

Description. Meristic values presented
in Table 21; regression data for morpho-
metric traits presented in Table 22.

Body elongate, moderately compressed;
anterodorsal profile weakly convex, rising
without interruption from tip of snout to
base of first dorsal fin; body depth 20.3-
24.6% SL; caudal peduncle moderate to
long, 25.3-30.7% SL.

Head short, shallow, length 28.0-34.0%
SL, height 14.8-16.3% SL; snout short,
pointed; angle of gape moderate to large;
lower jaw protruding beyond upper jaw,
bearing two nubs of variable prominence

on anterior surface of mandible. Maxilla
reaching ¥3-^/4 eye length; posterior margin
of maxilla narrow, rounded or bearing
posteriormost point near midline of bone;
small, weak mustache-like process project-
ing from j)osteroventral surface of maxil-
lary head. Eye round to oval, small, 37.4-
42.4% HL; anterodorsal rim of orbit not
reaching dorsal profile; interorbital region
narrow, 6.5-8.2% SL.

Teeth small, conical; premaxilla edentu-
lous or bearing single row of teeth on an-
terior half of bone; mandibular dentition
covering all or part of dentary, arranged
in double row anteriorly, tapering to single
row posteriorly; vomer bearing 1-6 irregu-
lar rows of teeth; palatines edentulous or
bearing teeth on anterior half of bone;
tongue edentulous.

Opercular spine pungent, bony, ventral
to 2-3 membranous or horny spinelets;
spine separated from spinelets by wide
gap; spinelets often obscured by underly-
ing membranes. Preopercular angle not
produced, serrations on posterior and/ or
ventral surfaces of bone rarely absent; sub-
opercle and interopercle serrated. Gill
rakers simple, awl-like.

First dorsal fin VI (1), VII (27), VIII
(1); second dorsiil fin I, 9 (28), II, 8 (1);
anal fin II, 9 (29). DJ short, 1.4-2.5%
SL; DJ, All, PJ moderate to long, 10.0-
12.6%, 9.2-13.3%, 11.7-15.3% SL respec-
tively.

Vertebrae 11 + 14 (29); epipleural ribs

Epigonus Systematics • Mayer 191

Table 21. Epigonus robustus meristic data. X = mean; SD = stan-
dard DEVIATION; n = NUMBER OF SPECrMENS.

X Range SD n

Pectoral fin rays 16.79 16-18 0.55 29

Gill rakers 31.68 30-33 0.93 29

Lateral line scales 48.76 47-50 0.91 29

Pyloric caeca 6.36 5- 8 0.78 28

Table 22. Epigonus robustus regression data, b = regression coef-
ficient ± 95% confidence interval; a = Y intercept; n = number of
specimens. All regressions on SL.

Body depth

-7.

Head height

- 1.

Eye diameter

Snout length

Interorbital widtl:

- 1.

Maxillary length

Lower jaw length

Caudal peduncle dep

ith

- 1.

Caudal peduncle len

gth

D2I

o„

2 1

All

8„

P2I

6 (2), 7 (8), inserting on vertebrae 1-6 brown or black niclanopliores; brancliial

or 1-7 respectively; pleural ribs 9 (29), in- region black. Body very oily; body cavity

serting on vertebrae 3-11. filled with rust brown fat globules; viscera

Color variable with preservation, pale and swinibladder often completely envel-

yellow to rust brown; scale pockets out- oped in fat.

lined by small black or brown melano- Description based on 29 .specimens

phores; opercular region tinged with 121.1-198.0 mm SL.

black; iris black; mouth light, mottled with Distribution. Most specimens of E. ro-

192 Bulletin Museum of Comparative Zoology, Vol. 146, No. 3

EriGONus Systematics • Mayer 193

Figure 22. Epigonus lenimen, 139.0 mm SL, UZM P45165.

biistus liave been taken by bottom trawLs
between 800 and 1225 meters off south-
eastern South Ameriea, South Africa, and
AustraHa (sec Fig. 21). One specimen
(ISH 430/71) was taken by a deep pelagic
trawl.

Geographic variation. No investigation
was undertaken because insufficient ma-
terial was available from South Africa and
Australia.

Taxonomic notes. Epigonus macrops
Gilchrist and von Bonde, 1924 was des-
cribed from two syntypes; the larger was
19(S mm (SL?). These specimens, together
with manv others collected bv the Fish-
cries and Marine Biological Survey, were
lost while being transferred to the South
African Museum. A portion of the ma-
terial was subsequently rediscovered at
Rhodes University, Grahamstown. From
the contents Smith (1961: 378) described
a specimen that he believed to be "Gil-
christ and \'on Bonde's type of macrops
from 600 fathoms off St. Helena Bay." This
fish was re-examined during the present
study.

Smith's specimen measures 162.2 mm SL
and thus cannot be the larger syntype;
however, it conforms to the descriptions
and proportions supplied by Gilchrist and
von Bonde and probably represents the
smaller type for which no length was pub-
lished.

An unusual aspect of the syntypes is that

the locations at which they were captured
will never be precisely known. The speci-
mens were taken at different stations. Al-
though these are recorded in both the orig-
inal description of E. macrops and in the
1921 report of the Fisheries and Marine
Biological Survey (Gilchrist, 1922), neither
account specifies which data are associated
with which syntype.
Common names. None.

Epigonus lenimen (Whitley, 1935)
Figure 22

Scepterias lenimen WHiitley (in part), 1935: 230
(original description; Great Australian Biglit:
south from Eucla, 350—450 fnis.; holot\-pe
examined, AM E3368); Whitley, 1940: 420,
fig. 33; Wliitley (in part), 1968: 56.

Epigonus lenimen Scott, 1962: 191, 1 fig.

Diagnosis. E. lenitnen is distinguished
from E. robustus and E. crassicaudus by
its broad interorbital region (8.7-10.2%
SL), long DJ (14.9-18.7% SL), and large
eyes (40.0-51.1% HL). It is further dif-
ferentiated from E. crassicaudus by shorter
head lengths (32.7-36.67^ SL) and .shal-
lower head heights (16.2-18.8% SL). E.
lenimen lacks lingual teeth but has 11 + 14
vertebrae and thus may be distinguished
from E. trewavasae and E. pectinifer. Un-
like remaining congeners, E. lenimen bears
a pungent, bony opercular spine, nine
second dorsal fin rays, and 16-18 pectoral
fin rays.

194 Bulletin Museum of Comparative Zoology, Vol. 146, No. 3

Table 23. Epigonus lenimen meristic data. X = mean; SD = standard
deviation; n = nxtmber of specimens.

Range

Pectoral fin rays 16.96 16-18 0.58 28

Gill rakers 30.29 28-34 1.27 24

Lateral line scales 48.12 47-50 0.91 26

Pyloric caeca 7.33 7- 9 0.56 24

Table 24. Epigonus lenimen regression data, b = regression coeffi-
cient ± 95% confidence interval; a = Y intercept; n = nltmber of
specimens. All regressions on SL.

-0.

Body depth

-3.

Head height

-1.

Eye diameter

-1,

Snout length

Interorbital width

-0.

Maxillary length

Lower jaw length

Caudal peduncle dep

-0.

Caudal peduncle len

gth

D2I

-0.

All

-2.

P2I

-2,

Description. MerLstic values presented in
Table 23; regression data for morphometric
traits presented in Table 24.

Body elongate; anterodorsal profile flat
or weakly concave, rising without inter-
ruption to first dorsal fin, more steeply
inclined behind occiput in large specimens;
body moderate to deep, 21.5-27.5% SL;

caudal peduncle moderate to long, 23.6-
29.3% SL. Head length 32.7-36.6% SL;
head height 16.2-18.8% SL; snout moder-
ately pointed; angle of gape moderate,
variable with age; lower jaw protruding
slightly or not at all; no prominent nubs on
anterior surface of mandible. Maxilla
reaching Vs-V2 eye length; posterior margin

Epigonus Systematics • Mayer 195

Table 25. Comparison of E. LENiMEy paratypes with specimens of E. lknimen and E.

DENTICULATVS. PaRATYPE MERISTICS REPORTED AS VALUE, FOLLOWED IN PARENTHESES HY NUM-
BER OF SPECLMENS EXHIBITING THAT VALUE. RATIOS ARE EXPRESSED AS PERCENTAGES.

E . 1 e n i m e n
E_. 1 e n i m e n paratypes E^. denticulatus

Dorsal fin rays 8-9 9(1), 10(11) 10

Pectoral fin rays 16—18 19(6), 20(6) 18—20

Vertebrae 11+14 10+15 10+15

10(1), 11(3) , „ , ,

Pyloric caeca 7 — 9 12(7) —

BH/SL 21.5—27.5 18.4 — 21.7 15.8—23.6

D2 I/SL 14.9—18.7 6.0—7.6 5.3—8.0

AII/SL 13.0 — 20.8 6.2—7.1 6.0—8.2

P2 I/SL 12.5 — 18.7 8.5—9.9 7.9—10.0

of maxilla narrow, rounded, or bearing First dorsal fin VII (29); second dorsal
posteriormost point near midline of bone; fin 1,8 (1), 1,9 (28); anal fin 11,8 (2), 11,9
weak mustache-like process projecting from (26); DJ moderate, 2.0-4. 17^ SL; DJ,
posteroventral surface of maxillary head, All long, 14.9-18.7%, 13.0-20.8% SL re-
process occasionally absent. Eye large, spectively; PJ moderate to long, 12.3-
oval, 40.0-51.1% HL; anterodorsal rim of 18.7% SL.

orbit reaching dorsal profile; interorbital Vertebrae 11 + 14 (29); epipleural ribs

width 8.7-10.2% SL. 6 (6), 7 (12), 8 (2), inserting on vertebrae

Teeth small, conical; premaxilla eden- 1-6, 1-7, or 1-8 respectively; pleural ribs
tulous or bearing single row of teeth oc- 9 (28), inserting on vertebrae 3-11.
cupying anterior half of bone. Mandible Color in alcohol variable; skin often
edentulous or bearing single row of teeth abraded, revealing underlying pale pink-
occupying up to % of dentary; tooth row yellow tissue; fin membranes and scale
occasionally double near symphysis. Vo- pockets mottled with numerous black
mer edentulous or bearing up to seventeen melanophores; head, opercular region, and
teeth arranged in diamond-shaped patch fin bases deep rust brown. Guanine de-
or in 1-3 irregular rows; palatines edentvi- posits variable, occurring on ventral por-
lous or bearing 1-2 teeth anteriorly; tongue tions of opercular region, isthmus, pectoral
edentulous. and pelvic fin bases, and abdomen to anus;

Opercular spine pungent, bony, ventral silver chromatophores on dorsal, anal, pee-
to 1-5 (usually 2) membranous or horny toral, or pelvic fin rays; iris black with
spinelets; spine and spinelets separated by silver highlights; moutli light, dotted with
wide gap; spinelets frequently obscured by melanophores; branchial region light in
underlying membranes. Preopercular angle small specimcMis, blackening with age.
narrowly produced, occasionally serrated; Description based on 32 specimens 40.0-
subopercle and interopercle unserrated or 147.8 mm SL.

weakly serrated. Gill rakers simple, awl- Distrihution. E. lenimen is known from

like. three localities (Fig. 21). The liolotype

196 Bulletin Museum of Comparative Zoology, Vol. 146, No. 3

0 I 2 3 cm

Figure 23. Epigonus crassicaudus, 259.0 mm SL, MCZ 48855.

was taken south of Austi-alia between 622
and 823 meters. Remaining specimens
were taken off New Zealand between 530
and 660 meters.

Taxonomic notes. Although G. P. Whit-
ley was the first to describe E. lenimen
(1935), inaccuracies in his papers have
produced several problems. The most
serious involve the type series and type
locality of the species.

The original description of E. lenimen
designates a holotype and nine paratypes.
A figure of the new form was not included
but was published in a subsequent paper
(Whidcy, 1940: fig. 33). Both the de-
scription and the illustration were based
exclusively on the holotype. Whitley re-
alized that the paratypes were different
from the holotype but considered them
to be poorly preserved specimens ( Whitley,
1935: 320).

A re-examination of the type series re-
veals that the paratypes are not conspecific
with the holotype. They are, instead, mem-
bers of E. denticulattis. As is shown in
Table 25, counts and measurements from
the paratypes always fall within ranges
characteristics of E. denticulatus. Pectoral
fin counts, vertebral number, pyloric
caecum counts, and fin spine lengths are
particularly noteworthy in this respect.
The paratypes further resemble E. denti-
culatus by bearing dentigerous palatines
and weak opercular armor, E. lenimen, on

the other hand, is characterized by eden-
tulous palatines and pungent opercular
spines.

Confusion over the type locality stems
from Whitley's 1940 paper. The locality
is cited as "from 190-320 fathoms, S.W.
from Eucla, Great Australian Bight [p.
420] ." This contradicts the data presented
in the original description: "Great Aus-
ti-alian Bight; south from Eucla, 350 to 450
fathoms [p. 231]." The 1940 citation is
extremely similar to station data listed for
paratypes AM E3581-3582 in 1935 ("Great
Australian Bight; SW from Eucla, 190-320
fathoms. 126° 451/2'E long. [p. 231]"). In
the absence of other information, it must be
concluded that erroneous locality data were
inserted in the 1940 publication through an
editorial oversight.

The most recent taxonomic questions
arise from Whitley's check list of New
Zealand fishes ( 1968 ) . This work includes
two incorrect citations in the synonymy
of E. lenimen. The first is based on a fish
taken off the Chatham Islands and tenta-
tively identified as Hynnodiis atherinoides
(Moreland, 1957). This specimen was later
re-identified as Grahamichthijs radiatus
(Moreland, personal communication). The
second misidentified .specimen is a "Big-
eyed Cardinal Fish" captured off Cape
Palliser, New Zealand (Anonymous, 1961).
This fish is actually a specimen of E.
telescopus and is presently in the collec-

Ei'ic.oNus Systematics • Mayer 197

tions of the Dominion Museum (DM 3072,
examined ) .

Common names. None.

Epigonus crassicaudus de Buen, 1959
Figure 23

Epiguinis crcis.'iicattdii.s de Buen, 1959: 196
(original description; preabysnial zone off
Valparaiso, Chile; holotype not examined, EBM
10.183).

Diagnosis. E. crassicaudus is strongly
compressed. It reaches 260-270 mm SL
and is the second largest species in the
genus. E. crassicaudus may be distin-
guished from E. trewavasae, E. pectinifer,
E. rohustus, and E. lenimen by its deep
head (18.9-21.2% SL) and deep body
(24.3-32.0% SL). It differs from remain-
ing congeners by bearing 9 rays in the
second dorsal fin and 6-7 pyloric caeca.

Description. Meristic values presented
in Table 26; regression data for morpho-
metric traits presented in Table 27.

Body elongate, compressed; anterodorsal
profile rising from tip of snout to occiput,
becoming moderately convex from occiput
to base of first dorsal fin. Body deep, 24.3-
32.07f SL; caudal peduncle broad, moderate
to short, 21.6-26.4% SL.

Head long, deep, postorbital portion
greatly expanded, length 36.8-41.9% SL;
height 18.9-21.27r SL; snout moderately
pointed in small specimens, blunt in adults;
angle of gape moderate to small; mandible
long, strongly protuberant, young bearing
two weak nubs on anterior surface of lower
jaw. Maxilla reaching %-% eye length;
posterior margin of maxilla broad, rounded
or bearing posteriormost point between
midline and ventral surface of bone. Eye
round, small, 34.2-39.6% IlL; surrounded
by numerous small scale pockets; antero-
dorsal rim of orbit reaching dorsal profile^;
interorbital region narrow, 6.2-8.5% SL.

Teeth small, conical, occasionally villi-
form, larger in small specimens; premax-
illary teeth arranged in irregular single
or double rows tapering to single row

posteriorly and covering from % to entire
length of bone; mandibular teeth arranged
in multiple rows, tapering to single row
posteriorly, covering from V2 to entire
IcMigtli of dentary; vomer edentulous or
bearing up to six irregular rows of minute
teeth; palatines edentulous or bearing 1-3
teeth anteriorly; tongue edentulous.

Opercular spine pungent, bony, ventral
to 3-5 flat, horny spinelets; spine separated
from spinelets by narrow gap; spinelets
often obscured by underlying membranes.
Preopercular angle slightly produced, pos-
terior and/ or venti'al surfaces serrated;
subopercles and interopercles serrated.
Gill rakers awl-like, short; gill filaments
long.

First dorsal fin VII (22); second dorsal
fin 1,9 (20), 1,10 (2); anal fin II.8 (1),
11,9 (21); DJ 2.0-3.6% SL; DJ 9.-8-13.2%
SL; All 10.3-14.0% SL; PJ 13.0-15.5%
SL.

Vertebrae 11 + 14 (25); epipleural ribs
6 (2), 7 (16), inserting on vertebrae 1-6
or 1-7 respectively; pleural ribs 9 (25),
inserting on vertebrae 3-11.

Color in alcohol variable with preserva-
tion; skin frequently abraded, exposing
underlying pink tissue and orange-rust fat
deposits; skin exti'emely oily; fin membranes
black; scale pockets mottled with numerous
black melanophores; dorsal portion of body
darker than ventral; forehead, snout, an-
terior half of mandible, and circumorbital
area heavily invested with black pigment;
opercles black or slate gray. Guanine de-
posits occasionally on opercles, isthmus,
pectoral and pelvic fin bases, and al)do-
men to anal fin; iris variable — black, siher,
or black with silver highlights; mouth and
branchial region light, darkening with age.

Description based on 27 specimens 80.3-
262.5 mm SL.

Ontogenetic change. Two juvenile E.
crassicaudus (12.2 mm SL, MCZ 48857,
and 21.7 mm SL, MCZ 48858) were taken
off the Chilean coast by midwater trawl.
Although these forms bear characteristics
diagnostic of the species, they differ con-

198 Bulletin Museum of Comparative Zoology, Vol. 146, No. 3

Table 26. Epigonus crassicaudus meristic data. X = mean; SD

STANfDARD DEVIATION; n = NUMBER OF SPECIMENS.

Range

Pectoral fin rays 18.05 17-19 0.58 22

Gill rakers 32.27 31-34 0.70 22

Lateral line scales 47.86 46-49 0.85 21

Pyloric caeca 6.87 6- 7 0.35 15

Table 27. Epigonus crassicaudus regression data, b = regression

COEFFICIENT ± 95% CONFIDENCE INTERVAL; a = Y INTERCEPT; n = NUMBER
OF SPECIMENS. AlL REGRESSIONS ON SL.

Body depth

-3.

Head height

-1.

Eye diameter

Snout length

Interorbital width

NONLINEAR

Maxillary length

-0.

Lower jaw length

-1,

Caudal peduncle dep

- 0,

Caudal peduncle len

gth

D2 I

All

P2I

sideral)ly in appearance and habit from
adnlts.

Most .striking is the juvenile pigment
pattern. Pelagic specimens are basically
pale yellow with large, brown patches
covering most of the caudal peduncle.
Caudal peduncle rings, like those found on
E. pandionis young, are absent, although
myotomes are outlined by thin brown

bands. Brown pigment extends anteriorly
as a band from the caudal peduncle to the
frontal region of the head. A poorly de-
fined black stripe extends across the snout
to the anterior rim of the orbit. In general,
juvenile E. crassicaudus resemble E. teles-
copiis young figured by Koefoed (1952:
plate IIA).

The midwater capture of E. crassicaudus

Epigonus Systematics • Mayer 199

juveniles suggests that tlie life cycle of tlK>
specit^s includes a pelagic stage. Unfortu-
nately, the data available are not sufficient
to determine the duration of this stage.

Distribution. E. crassicaudus is endemic
to the waters off central Chile (Fig. 21).
Adults have been captured by bottom
trawls made between 200 and 400 meters;
juveniles were taken by midwater trawls
fishing from 200 to 270 meters.

Common mimes. None.

Species Incertae Sedis

Micwichtlnjs coccoi Riippell, 1852: 1 (original
description; "Mare siculum"; holotype not ex-
amined, SMF 1069).

The original description of M. coccoi
provides only a superficial account of the
holotype. Subsequent papers either para-
phrase Riippell's work {e.g., Canestrini,
1860; Doderlein, 1889) or are based on
material not compared to the holotype
(i.e., Facciola, 1900; Caporaicco, 1926;
Gonzales, 1946). It is questionable whether
the latter specimens are conspecific with
the holotype.

Most recent revisers (e.g., Schultz, 1940;
Norman, 1957) have synonymized Micro-
ichthys with Apogon; however, the data
are inconclusive and also suggest an affinity
with Epigonus (Eraser, 1972: 5). A re-
examination of the holotype must be under-
taken to clarify the status of M. coccoi.

A second species of Microichthys — M.
sonzoi Sparta, 1950 — does not appear to be
an Epigonus on the basis of vertebral and
dorsal fin counts. The only known speci-
men of this species has been lost (Torton-
ese, personal communication).

ACKNOWLEDGMENTS

This work would not have been possible
without the assistance and support of
niunerous people. I wish to thank the
following scientists and institutions for
material used in this study: J. R. Paxton
and D. Hoese, Australian Museum; A. W.
Wheeler and G. Palmer, British Museum

(Natural History); J. Randall, Bemice P.
Bishop Museum; W. Eschmeyer, California
Academy of Sciences; E. Bertclsen, Carls-
bergfondets; J. Moreland, Dominion Mu-
seum; L. P. Woods, Eield Museum of
Natural History; P. Struhsaker, National
Marine Fisheries Service, Honolulu; R.
Raymond, Instituto de Fomento Pesquero;
X. Missonne, Institut Royal des Sciences
Naturelles de Belgique; G. Krefft, Institut
fiir Seefischerei; M. M. Smith, J. L. B.
Smith Institute of Ichthyology; I. Naka-
mura, Kyoto University; R. J. Lavenberg,
Los Angeles County Museum of Natural
History; M. Bauchot, Museum National
d'Histoire Naturelle; M. Poll, Musee Royal
de I'Afrique Centrale; E. A. Lachner and
T. H. Eraser, National Museum of Natural
History; M.-L. Penrith, South African Mu-
seum; George R. Vliller, Tropical Atlantic
Biological Laboratory; R. S. Gaille, Texas
Parks and Wildlife Department; M. Leible,
Universidad Catolica de Chile; C. R.
Robins, University of Miami; J. Nielsen,
Universitetets Zoologiske Museum; B.
Nafpaktitis, University of Southern Cali-
fornia; T. Abe, University of Tokyo; R.
Backus and J. Craddock, Woods Hole
Oceanographic Institution; and C. Karrer,
Zoologisches Museum, Berlin. W. Klause-
witz of the Natur-Museimi Senckenberg
provided invaluable information on the
holotype of Microichthys coccoi, and E.
Tortonese of the Museo Civico di Storia
Naturale, Genoa, answered numerous ques-
tions about problematical forms such as
Pomatomichthys constanciae and Micro-
icJithys sanzoi.

I am greatly indebted to Ernst Mayr,
Giles W. Mead and Karel F. Liem for
their guidance, criticism, and support of
my work. I am also grateful to Richard L.
Haedrich for reading the manuscript and
assisting in the planning of this research.
Special thanks are extended to G. S. Myers
for assistance with taxonomic problems.

I wish to thank the staffs of the Fish
Department, Museum of Comparative
Zoology, and Department of Natinal Sci-

200 Bulletin Museum of Comparative Zoology, Vol. 146, No. 3

ences, Boston University, for their useful
comments and practical help. Karen Green-
leaf and Pat Allen typed the final draft of
this manuscript.

Illustrations of eleven of the twelve
species of Epigonus were prepared by L.
Laszlo Meszoly. Jordan and Jordan's il-
lustration of E. fragilis (Fig. 10) was made
available through the courtesy of the
Carnegie Museum.

Finally, a hearty vielen Dank to my wife
for her patience, encouragement, and ed-
itorial assistance.

Support for this work was provided by
NSF Graduate Fellowships during 1966
to 1971 and a grant from Harvard Uni-
versity's Committee on Evolutionary Bi-
ology (GB7346).

LITERATURE CITED

Anonymous. 1961. Caught only 3 times before.

Wellington Evening Post.
Bailey, N. T. 1959. Statistical Methods in

Biology. New York: John Wiley and Sons,

Inc. 200 pp.
Barnard, K. H. 1927. A monograph of the

marine fishes of South Africa, Part II. Ann.

South African Mus., 12(2): 419-1065.
Bauchot, M. L., and M. Blanc. 1961. Poissons

marins de I'Est Atlantique Tropical, II.

Percoidei (Teleosteens Perciformes) lere

partie. Atlantide Rep. No. 6: 65-100.
Bertolini, F. 1933. Apogonidae. Fauna e flora

del Golfo di Napoli. Uova, larve e stadi

giovanili di Teleostei. Pubbl. Staz. Zool.

Napoli, 38: 306-309.
BiNi, G. 1968. Atlante dei Pesci delle Coste

Italiane, Vol. IV. Roma: Mondo Sommerso

Editrice. 163 pp.
Bleeker, p. 1876. Systema Percarum revisum.

Pars I et II. Arch. Neerl. Sci., 11: 274-340.
Braxjer, a. 1906. Die Tiefsee-Fische. I. Syste-

matischer Teil. Wiss. Ergeb. Deut. Tiefsee-

Exped. "Valdivia," 1898-1899, 15: 1-420.
Breder, C. M., and D. E. Rosen. 1966. Modes

of Reproduction in Fishes. Garden City,

N. Y.: Natural History Press. 941 pp.
Brown, M. E. 1957. Experimental studies on

growtli. In The Physiology of Fishes, Vol.

I, M. E. Brown (ed. ). New York: Academic

Press, Inc., pp. 361-400.
Buen, F. de. 1959. Notas preliminares sobre

la fauna marina preabismal de Chile, con

descripcion de una familia de rayas, dos

generos y siete especies nuevos. Bol. Mus.

Nac. Hist. Natur., 27(3): 171-201.
Canestrini, J. 1860. Zur Systematik der Per-

coiden. Verb. Zool. Bot. Ver. Wien, 10:

291-314.
Capello, F. de B. 1868. Peixes novos de

Portugal e da Africa occidental e caracteres

distinctivos d'outras especies ja conhecidas.

J. Sci. Math. Phys. Natur. Acad. Real Sci.

Lisboa, 1(2): 154-169.
Caporaicco, L. dl 1926. II cranio de "Micro-

ichthvs coccoi" Riipp. Monit. Zool. Ital.,

37(6): 127-132.
Cligny, a. 1903. Poissons des Cotes d'Espagne

et de Portugal ( Ocean Atlantique ) . Premiere

partie. Boulogne. 30 pp.
Cocco, A. 1829. Su di alcuni nuovi pesci de'

mari di Messina. Giom. Sci. Lett. Art. Sicil.,

7(77): 138-147.
. 1885. Indice Ittiologico del Mare di

Messina. Nat. Sicil., 4(4): 85-88.
CuvrER, G. 1828. Des pomatomes. In Histoire

Naturelle des Poissons, Tome Second, G.

Cuvier and A. Valenciennes. Paris: F. G.

Levrault, pp. 169-174.
Dieuzeide, R. 1950. Sur un Epigonus nouveau

de la Mediterranee (Epigonus denticulatus,

nov. sp.). Bull. Sta. Aquic. Peche CastigUone,

N.S., No. 2: 89-105.
, M. Novella, and J. Roland. 1953.

Catalogue des poissons des Cotes algeriennes.

II. Osteopterygiens. Bull. Sta. Aquic. Peche

Castiglione, N.S., No. 5: 1-258.
Dlxon, W. J. (ed.). 1967. BMD Biomedical

Computer Programs. Univ. Calif. Publ. Auto-
matic Computation No. 2. Berkeley: Univ.

Cahf. Press. 600 pp.
DoDERLEiN, p. 1889. Manuale Ittiologico del

Mediterraneo. Fascicolo IV. Palermo: Giornali

di Sicilia. 188 pp.
Dons, C. 1938. Notes on fishes III. Epigonus

telescopiis (Risso), new to Norway. Det

Kgl. Nor. Vidensk. Selsk. Forh., 11(35):

141-142.
Eblna, K. 1931. Buccal incubation in the two

sexes of a percoid fish, Apogon semiUneatus

T. & S. T. Imp. Fish. Inst. Tokyo, 27(1):

19-21.
Ehrenbaum, E. 1928. Rare fishes in the North

Sea. Nature (London), 121(3053): 709.
Ekman, S. 1953. Zoogeography of the Sea.

London: Sidgwick and Jackson, Ltd. 417 pp.
Facciola, L. 1900. Sul Microichthys coccoi

Rapp. Monit. Zool. Ital., 11(5): 188-194.
Fowler, H. W. 1928. The fishes of Oceania.

Mem. Bernice P. Bishop Mus., 10: 1-540.
. 1935. South African fishes received

from Mr. H. W. Bell-Marley in 1935. Proc.

Acad. Nat. Sci. Philadelphia, 87: 361-408.
. 1936. The marine fishes of West Africa

Ei'iGONus Systematics • Mayer 201

based on tlie collection of the American
Musemn Congo Expedition, 1909-1915. Bull.
Amer. Mus. Nat. Hist., 70(2): 607-1493.
— , AND B. A. Bean. 1930. The fishes of

the British Museum, Vol. I. London. .524 pp.
1868. Report on a c(jllection of fishes

the families Amiidae, Chandidae, Duleidae,
and Serranidae obtained by the United States
Bureau of Fisheries Steamer "Albatross" in
1907 to 1910, chiefly in the Philippine
Islands and adjacent seas. Bull. U.S. Nat.
Nhis., 10(100): 1-334.

FuASER, T. H. 1972. Comparative osteology of
the shallow water cardinal fishes IPerci-
formes: Apogonidae] with reference to the
s>'steniatics and exolution of the family.
Ichthvol. Bull. J. L. B. Smith Inst. Ichthyol.,
Rhodes Univ., No. 34: 1-105.

FucJLLSTER, F. C. 1960. Atlantic Ocean atlas of
temperature and salinity profiles and data
from the International Geophysical Year of
1957-1958. Woods Hole Oceanogr. Inst.
Atlas Ser., 1 : 1-209.

Gall, J- le. 1931. Epigomts telescopus Risso
1810. In Faune Ichthyologique. Copenhagen,
Conseil Permanent International pour 1' Ex-
ploration de la Mer, 2 pp.

GiGLiOLi, E. H. 1880. Elenco dei Mammiferi,
degli Uccelli e dei Rettili Ittiofagi Ap-
partenenti alia Fauna Italica e Catalogo degli
Anfibi e dei Pesci Italiani. Firenze: Staniperia
Reale. 85 pp.

Gilbert, C. H. 1905. The deep-sea fishes of
the Hawaiian Islands. In The Aquatic Re-
sources of the Hawaiian Islands. II, D. S.
Jordan and B. W. Evermann (eds. ). Bull.
U. S. Fish. Comm., 23: 575-716.

Gilchrist, J. D. F. 1922. Report No. 2 for the
year 1921. Fish. Mar. Biol. Surv. South
Africa, Rep. No. 2: 1-84.

, AND C. vox BoxDE. 1924. Deep-sea

fishes procured by the S. S. "Pickle" (Part
II). Fish. Mar. Biol. Surv. South Africa,
Rep. No. 3, Spec. Rep. VII: 1-24.

GoxzALES, T. 1946. Contributo alia conoscenza
dello s\iluppo post-embrionale in Micro-
ichtJit/s coccoi, Riippell. Boll. Pesca Piscicolt.
Idrobiol., Ser. 22, 1 ( 1 ) : 39-46.

GooDE, G. B., AXD T. H. Beax. 1881. De-
scription of a new species of fish, Apogon
))andionis, from deep water off the mouth
of Chesapeake Bav. Proc. U. S. Nat. Mus.,
4: 160-161.

, AXD . 1896. Oceanic Ichthy-
ology. Wa.shingtou, D. C: Smithsonian In-
stitution. 553 pp.

GosLixE, \V. A. 1961. The perciform caudal
skeleton. Copeia, 1961(3): 265-270.

, AXD \'. E. Brock. 1960. Handbook of

Hawaiian Fishes. Honolulu: Univ. Hawaii
Press. 372 pp.

Giinther, A. 1859. Catalogue of the Fishes in

made at St. Helena by J. C. Melli.ss, Esq.
Proc. Zool. Soc. London, pp. 22.5-228.

Haxeda, Y., F. H. Johxsox, and O. Shi.mo.mura.
1966. The origin of luciferin in the luminous
ducts of Pampriacmithus ransonncti, Pem-
plicris kltiiizingcii, and Apogon cllioti. In
Bioluminescence in Progress, F. H. Johnson
and Y. Haneda (eds.). Princeton: Princeton
Univ. Press, pp. 533-545.

Holt, E. W., AND W. L. Calderwood. 1895.
Survey of fishing grounds, west coast of
Ireland, 1890-1891. Report on the rarer
fishes. Scient. Trans. R. Dublin Soc, Ser.
2, 5: 361-524.

HuHBS, C. L., AND K. F. Lacler. 1958. Fishes
of the Great Lakes Region. Ann Arbor:
Univ. Michigan Press. 213 pp.

I\\Ai, T. 1959. Notes on the luminous organ
of the apogonid fish, Siphaiiiia incijiinoi.
Ann. Mag. Natur. Hist., Ser. 13, 2(21):
545-550.

Jordan, D. S. 1917. Notes on Glossamia and
related genera of cardinal fishes. Copeia,
No. 44: 46-47.

, AND C. H. Gilbert. 1882. Synopsis

of the fishes of North America. Bull. U. S.
Nat. Mus., 16: 1-1018.

, AND E. K. Jordan. 1922. A hst of

the fishes of Hawaii, with notes and de-
scriptions of new species. Mem. Carnegie
Mus., 10(1): 1-92.

Kamohara, T. 1952. Revised descriptions of
the offshore bottom-fishes of Prov. Tosa,
Shikoku, Japan. Rep. Kochi Univ. Nat. Sci.,
No. 3: 1-22.

Koefoed, E. 1952. Zeomorphi, Percomorphi,
Plectognathi. Rep. Sci. Res. Michael Sars
N. Atl. Deep-Sea Exped. 1910, 4, Pt. 2(2):
1-26.

Lowe, R. T. 1841. A synopsis of the fishes of
Madeira; with principal synonyms, Portu-
guese names, and characters of the new
genera and species. Trans. Zool. Soc. London,
2(14): 173-200.

LozANO, L. 1934. Algunos peces pelagicos o
de profundidad procedentes del Mediterraneo
occidental. Bol. Soc. E.span. Hist. Natur.,
34: 85-92.

Matsubara, K. 1936. Biometry of two species
of Japanese cardinal-fishes, with special
reference to their taxononn. J. Imp. Fish.
In.st. Tokyo, 31: 119-1.30.

Maurin, C. 1968. Ecologie ichthyologiques des
fonds Chalutables atlantiques (de la baie
ibero-marocaine a la Mamitanie) et de la
Mediterranee occidentale. Theses presentees
a la Faculte des Sciences de I'Universite de
Nancy. No. d'Ordre: A. O. 2 182. 145 pp.

202 Bulletin Museum of Comparative Zoology, Vol. 146, No. 3

Mayer, G. F. 1972. Systematics, functional
anatomy, and ecology of the caidinalfish
genus Epigonus ( Apogonidae ) . Ph. D.
Thesis, Hanard University. 190 pp.

Mayr, E. 1969. Principles of Systematic Zo-
ology. New York: McGraw-Hill, Inc. 428 pp.

Mead, G. W., and M. G. Bradbury. 1963.
Names of bones. In Fishes of the Western
North Atlantic, H. B. Bigelow and W. C.
Schroeder (eds.). Mem. Sears Found. Mar.
Res., 1(3): 20-23.

MoNOD, T. 1968. Le complexe urophore des
poissons teleosteens. Mem. Inst. Fondam.
Afr. Noire, No. 81. 705 pp.

MoREAU, E. 1881. Histoire Naturelle des Pois-
sons de la France. Vol. II. Paris: G. Masson.
572 pp.

MoRELAND, J. 1957. Appendix 6: Report on
the fishes. New Zealand Dep. Sci. Ind. Res.
Bull., 122: 34.

Navarro, F. de P. 1942. Nota preliminar sobre
los paces de la costa de Africa, desde el Cabo
Bojador a la Bahia de Tanit ( Resultados de
una compana industrial de pesca d'arrastre).
Bol. Real Soc. Espan. Hist. Natur., 40(5-6):
189-214.

, F. LozANO, J. M. NovAz, E. Otero,

AND J. Sainz Pardo. 1943. La pesca d'
arrastre en los fondos del Cabo Blanco y
del Banco Arguin (Africa Sahariana). Trab.
Inst. Espaii. Oceanogr. Numero 18. 225 pp.

Nohre, a. 1935. Descricao dos Peixes de
Portugal. Vertebrados (mamiferos, reptis e
peixes). I. Fauna Marinha de Portugal.
Porto. 574 pp.

Norman, J. R. 1939. Fishes. Sci. Rep. John
Murray Exped., 7(1): 1-116.

. 1957. A Draft Synopsis of the Orders,

Families and Genera of Recent Fish and
Fish-like Vertebrates. London: British Mu-
seum (Natural History). 649 pp.

Nowlin, W. D., jr., and H. J. McLellan.
1967. A characterization of the Gulf of
Mexico waters in winter. ]. Mar. Res., 25(1):
29-59.

Osomo, B. 1898. Da distribviicao geographica
dos peixes e crustaceos colhidos nas pos-
sessoes Portuguezos d'Afrique occidental e
existentes no Museu Nacional de Lisboa.
J. Sci. Math. Phys. Natur. Acad. Real Sci.
Lisboa, Ser. 2, 5: 185-202.

PiETSCHMANN, V. 1930. Rcuiarks on Pacific
fishes. Bull. Bernice P. Bi.shop Mus., 73:
1-24.

Poll, M. 1954. Poissons IV. Teleosteens
acanthopterygiens (premiere partie). Re-
sidtats scientifiques. Expedition Oceano-
graphique Beige dans les Eaux Cotieres
Africaines de I'Atlantique Sud (1918-1949),
4(3A): 1-390.

Radcliffe, L. 1912. Descriptions of fifteen
new fishes of the family Gheilodipteridae,
from the Pliilippine Islands and contiguous
waters. Proc. U. S. Nat. Mus., 41(1868):
431-446.

Rafinesque, G. S. 1810. Indice d' Ittiologia
Siciliana; Ossia, Gatalogo Metodico dei Nomi
Latini Italiani, e Siciliani dei Pesci, che si
Rinvengono in Sicilia: Disposti Secondo un
Metodo Naturale e Seguito da un Appendice
che Gontiene la Descrizione di Alcuni Nuovi
Pesci Siciliani. Messina: Presso Giovanni
del Nobolo. 70 pp.

Risso, A. 1810. Ichthyologie de Nice ou Histoire
naturelle des poissons du department des
Alpes Maritimes. Paris: Chez F. Schoell.
388 pp.

Royce, W. R. 1957. Statistical comparison of
morphological data. In Contributions to the
Study of Subpopulations of Fishes, J. G.
Marr (coordinator). U. S. Fish. Wildl. Serv.
Spec. Sci. Rep. Fish., No. 208: 7-28.

RuppELL, W. 1852. \'erzeichniss der in dem
Museum der Senckenbergisclien naturfor-
schenden Gesellschaft aufgestellten Samm-
lungen. Vierte Abteilung. Fische und deren
Skelette. Frankfurt-a.M.: J. D. Sauerliinder.
40 pp.

Sakamoto, K. 1930. Buccal incubation in the
percoid fish, Apogon lineatus T. & S. J. Imp.
Fish. Inst. Tokyo, 26( 1 ) : 9-10.

ScHULTZ, L. p. 1940. Two new genera and
three new species of cheilodipterid fishes,
with notes on the other genera of the familv.
Proc. U. S. Nat. Mus., 88(3085): 403-423.

Scott, T. D. 1962. The Marine and Fresh
Water Fishes of South Australia. Adelaide:
South Australian Branch of the British Sci-
ence Guild. 338 pp.

Simpson, G. G., A. Roe, and R. C. Lewontin.
1960. Quantitative Zoology. New York: Har-
court, Brace and World, Inc. 440 pp.

Smith, G. L., E. H. Atz, and J. C. Tyler. 1971.
Aspects of oral brooding in the cardinalfish
Cheilodiptcius af finis Poey (Apogonidae).
Amer. Mus. Novitates, No. 2456: 1-11.

Smith, J. L. B. 1949a. Forty-two fishes new
to South Africa, with notes on others. Ann.
Mag. Natur. Hist., Ser. 12, 2: 97-111.

— . 1949b. The Sea Fishes of Southern

Africa. South Africa: Central News Agenc>-,
Ltd. 550 pp.

1961. Fishes of the family Apogonidae

of the western Indian Ocean and the Red
Sea. Ichth%ol. Bull. Rhodes Univ., No. 22:
373-418.
Sparta, A. 1950. Su di una nuova specie di
Micwichthi/s: M. sanzoi (n. sp.). Boll. Pesca,
Piscicolt. Idrobiol., Ser. 26, 5(2): 202-206.

EricoNus Systematics • Mayer 203

Stein'dachner, F. 1891. lehthNologischc Bei-
triiKt' (X\^). i'her ciiiige scltenc iiiul ucnc
Fisc'luiitc'ii aus ck'in caiuiiisclu'ii Aicliipcl.
Sitzber. Akad. VViss. Wien, 100(1): 343-374.

. 1907. Fi.sche aiis Siidaiabicn imd

Sokotra. Denkschr. Akad. Wiss. Wien, 71:
123-168.

Taylok, W. R. 1967. An enz\-matic method of
elearing and staining small \eitel)iates. Proe.
U.S. Nat. Mns., 122(3596): 1-17.

Tinker, S. W. 1944. Hawaiian Fishes. Ilono-
kdii: Tongg PnbHshing Company. 404 pp.

ToRTONESE, E. 1952. Un percoide marino e
batifilo nno\o per Tittiofauna itaHana (Epi-
^oinis clciilicuUitiis Dienz.). Boll. Pesca
Pi.scicolt. Idrobiol. Ser. 28, 7(1): 72-74.

, AND L. C. QuEiROLO. 1970. Contribute

alio studio dell ittiofauna del Mar Ligure
orientale. Ann. Mus. Civ. St. Natur. Genova,
78: 21-46.

Vaillaxt, L. 1888. Poissons. E.xpeditions Sci-
entifiques du Travailleur et du Talisman
pendant les Annees 1880, 1881, 1882, 1883.
Paris: G. Masson. 406 pp.

V^ALENCiENNES, A. 1830. Additions et correc-
tions anx tomes II, III, IV, et V. In Historic
Naturelle des Poissons, Tome Sixieme, G.
Cuvier and A. Valenciennes. Paris: F. G.
Levrault, pp. 495-559.

. 1837-1844. Ichthyologie des iles Can-
aries, on Histoire naturelle des poissons
raportes par M. M. Webb et Berthelot. In
Histoire Naturelle des lies Canaries, P. B.
Webb and S. Berthelot (eds.). Vol. 2, Pt. 2.
Paris, pp. 1-109. [pp. 1-8 published Dec.
1842; plate 1 published May 1837.]

Weuer, M., and L. F. de Beaufort. 1929. The
Fishes of the Indo-Australian Archipelago.
y. Anacanthini, Allotriognathi, Heterosomata,
Berycomorphi, Percomorplii: Families Kuhli-
idae, Apogonidae, Plesiopidae, Pseudo-
plesiopidae, Priacanthidae, Centropomidae.
Leiden: E. J. Brill, Ltd. 458 pp.

Whitley, G. P. 1935. Studies in ichthyology.
No. 9. Rec. Aust. Mus. Sydney, 19(4):
215-250.

-. 1940. Illustrations of some Australian

fishes. Aust. Zool., 9(4): 397-428.

. 1968. A check-list of the fishes re-
corded from the New Zealand region. Aust.
Zool., 15(1): 1-102.

Williams, F. 1968. Report on the Gnincan
Trawling Survey, \'ol. I. General Report.

OAU Scientific, Technical and Research
('oninn'ssion, Ful)licali<)n 99. 828 pp.
W'iisi, (;. 1961. Slratitication and Circulati(m
in the Antillean-Caribbcan Basins. New York:
C]()luml)ia l'ni\ . Press. 201 pp.

APPENDIX

The following chart lists all meristic and
niorphometric data for the holotypcs of
E. oUiiolepis sp. nov. and E. pectinifer sp.
nov. Measurements are given in milli-
meters.

oligolepis

E. pectinifer

USNM 2 077 18

USNM 207725

MERISTIC DATA

Dorsal fin

VII-I, 10

VII -I, 9

Anal fin

II, 9

Pectoral fin

Pelvic fin

I, 5

Lateral line
sc ale s

Gill r ake r s

Pyloric caeca

Ve r tebr ae

10 + 15

Pleural ribs

E p i p 1 e u r a 1 rib

MORPHOMETRIC

DATA

90. 8

114. 3

33. 2

40. 5

Body depth

21.2

28. 1

Head height

17. 1

18. 8

Eye diameter

14. 5

16. 1

Snout 1 e n g t li

7. 8

10. 9

I n t e r 0 r b i t a 1 w i

idth

8. 4

10. 0

Maxillary leng

15. 6

18. 2

Lower jaw len

gth

16. 3

18. 5

Caudal p c <l u n c :
d e p t h

8. 9

11. 7

Caudal p e d u n c :
length

23. 6

32. 1

D2l

10. 9

13. 7

All

11.1

13. 7

P2I

12. 3

14. 4

us ISSN 0027-4100

BulLetln OF THE

Museum of

Comparative

Zoology

The Spider Family Anyphaenidae
in America North of Mexico

NORMAN PLATNICK

HARVARD UNIVERSITY

CAMBRIDGE, MASSACHUSETTS, U.S.A.

VOLUME 146, NUMBER 4
19 SEPTEMBER 1974

PUBLICATIONS ISSUED

OR DISTRIBUTED BY THE

MUSEUM OF COMPARATIVE ZOOLOGY

HARVARD UNIVERSITY

Breviora 1952-

BuLLETiN 186a-

Memoirs 1864-1938

JoHNSONiA, Department of Mollusks, 1941-

OccAsiONAL Papers on Mollusks, 1945-

SPECIAL PUBLICATIONS.

1. Whittington, H. B., and E. D. I. Rolfe (eds.), 1963. Phylogeny and
Evolution of Crustacea. 192 pp.

2. Turner, R. D., 1966. A Survey and Illustrated Catalogue of the Teredini-
dae (Mollusca: Bivalvia). 265 pp.

3. Sprinkle, J., 1973. Morphology and Evolution of Blastozoan Echinoderms.
284 pp.

4. Eaton, R. J. E., 1974. A Flora of Concord. 211 pp.

Other Publications.

Bigelow, H. B., and W. C. Schroeder, 1953. Fishes of the Gulf of Maine.
Reprint.

Brues, C. T., A. L. Melander, and F. M. Carpenter, 1954. Classification of
Insects.

Creighton, W. S., 1950. The Ants of North America. Reprint.

Lyman, C. P., and A. R. Dawe (eds.), 1960. Symposium on Natural
Mammalian Hibernation.

Peters' Check-list of Birds of the World, vols. 2-7, 9, 10, 12-15.

Proceedings of the Nevi' England Zoological Club 1899-1948. (Complete
sets only.)

Publications of the Boston Society of Natural History.

Price list and catalog of MCZ publications may be obtained from Publications
Office, Museum of Comparative Zoology, Harvard University, Cambridge, Massa-
chusetts, 02138, U.S.A.

THE SPIDER FAMILY ANYPHAENIDAE IN AMERICA
NORTH OF MEXICO'

NORMAN PLATNICK-

Abstract. E\'idence from the tracheal system,
claw tufts and courtship behavior is used to
justify the family status of Anyphaenidae. Sug-
gested relationships between Anyphaenidae and
Clubionidae, Amaurobiidae and Argyronetidae
are disclaimed. The faniih' Amaurobioididae is
newly synonymized with Anyphaenidae. Generic
problems within the family are discussed. The
thirty-six species occurring north of Mexico are
described, their diagnostic characters pointed out
and illustrated, their distributions mapped, notes
on their habits given, and keys to genera,
species groups and species provided. The genera
AuiipliaencUa and Ciagus are newly synonymized
witli Wulfila. Thirteen species are described as
new: Amjpliacna aJachua, A. arhida, A. uutumna,
A. catalina, A. cochise, A. gertschi, A. fiiJjIwides,
A. hespar, A. lacka, A. rita, Aijslia aninda,
Wulfila hnjantac and W. wiinda. Nineteen new
synonymies are recognized.

INTRODUCTION

This study had three objectives: to de-
termine whether or not the anyphaenids
should be treated as a distinct family; to
examine the relationships between the any-
phaenids and the other groups of spiders
with which they have been associated in
the past; and to reclassify the species oc-

' This study was presented to the Department
of Biology at Harvard University in partial ful-
fillment of the requirements for the degree of
Doctor of Pliilosophy.

"Present address: Department of Entomology,
The American Nhiseum of Natural History, Cen-
tral Park West at 79th Street, New York, New
York 10024.

curring north of Mexico on generic and
specific levels.

The anyphaenids are a diverse group
with perhaps five hundred species. Thirty-
six species are known to occur in America
north of Mexico and are included here.
About 375 species have been described
from the Neotropic region, as well as
around ten from the Palearctic and five
from the Oriental. The South American
species show the widest spectrum of body
forms; they range from 2-25 mm in length
and are often intricately colored or have
peculiarly elongate chelicerae or legs.

As in most spiders, little is known of the
ecology or behavior of anyphaenids. They
are wandering hunters. In the eastern
United States, where long-legged species
predominate, they are most often collected
by sweeping foliage in fields and meadows,
and seem to be primarily noctiu-nal. How-
ever, in the western United States, where
most species have shorter legs, they are usu-
ally found in forests by sifting through
litter and turning logs and stones. They
feed on various groups of insects, and
though they have been observed to prey
heavily on such Lepidoptera as the fall
webworm, IlijpJiantria cunea (Warren et
al., 1967), they are proliably not very se-
lective. In captivity they will consume
Drosophila eagerly. Their principal ene-
mies in nature are the mud-dauber wasps
of the family Sphecidae, as evidenced from
the hundreds of individuals, particularly

Bull. Nhis. Comp. Zool, 146(4) : 205-266, September, 1974 205

206 Bulletin Museum of Comparative Zoology, Vol. 146, No. 4

of the diurnally active genus Aijsha, that
are frequently collected from wasp nests.
Krombein ( 1967 ) cites especially the wasp
genus Trypargilum in this respect. Like
most nearctic spiders, males and females
usually mature in early spring, with males
living through early summer and females
living through the summer. In some south-
ern species, however, both sexes are found
matm-e year-round. Also, some species in
the Amjphaena celer group are matiu-e
throughout the winter. Anyphaenids make
little use of silk, other than in building re-
treats under leaves or stones and of course
in building egg sacs, which are usually
round, made of soft white silk, not leathery
or papery, and contain between 50 and
150 eggs.

The North American species are 2-9 mm
long; the largest species belong to the
genus Aijsha, the smallest to Wulfih.
There are always eight eyes in two rows;
the median eyes are usually closer to the
laterals than to each other; unlike many
gnaphosids, the eyes are always round, and
unlike many clubionids, the anterior me-
dian eyes are usually smaller than the
others. Other than the genitalia, the main
structural differences between males and
females are the sternal and coxal modifi-
cations (pointed spurs, rounded knobs, or
clumps of short thick setae) found on
males in some groups.

In many species groups it would be im-
possible to distinguish the species without
using genitalic characters. The palpus
(Text-fig. 3) usually has a large median
apophysis (the shape of which is often
species-specific), a small conductor and a
conspicuous embolus. Besides the retrolat-
eral tibial apophysis (almost always of
great diagnostic value) a ventral tibial
apophysis (some Aijsha) or a retrolateral
patellar apophysis (some Teudis) may be
present. The female epigyna and internal
genitalia are extremely diverse and diffi-
cult to characterize. The two epigynal
openings are located posteriorly and are
extremely difficult to see unless a portion

of the male embolus has been left behind
after mating. Many species have an addi-
tional anterior median epigynal opening
into which the retrolateral tibial apophysis
or median apophysis fits during mating.
The genitalia of anyphaenids, particularly
of the South American species, are more
complex than those of clubionids and
gnaphosids. Among the clubionids, only
Chiracanthium has genitalia that seem in
any way close to those of anyphaenids.

For the area treated here, only three im-
portant papers have been published on
anyphaenids. Bryant (1931) summarized
the very sparse data then available on the
group in the United States, while Chicker-
ing (1937, 1940) described many of the
species occurring in Panama and the Canal
Zone, a number of which also occur in the
United States.

ACKNOWLEDGEMENTS

I would like to thank first Herbert W.
Levi for his painstaking and patient help
with all aspects of this project. Willis
Gertsch contributed much of his knowl-
edge of the group as well as the drawings,
done by the late Wilson Ivie, of the genus
Oxijsorna.

This investigation was supported in part
by Public Health Service Research Grant
AI-01944 from the National Institutes of
Allergy and Infectious Diseases, H. W.
Levi, principal investigator; by Grant GB-
36161 from the National Science Founda-
tion, H. W\ Levi, principal investigator;
and by Grant GB-19922 from the National
Science Foundation, R. C. Rollins, princi-
pal investigator. The Department of Biol-
ogy, Harvard University, by means of
Summer Research Grants in Evolutionary
Biology in 1971 and 1972 and a Richmond
Fellowship in 1973, afforded me much of
the time and field work necessary to com-
plete this work. Miss Suzanne Barbier of
Radcliffe Gollege assisted greatly with the
examination of tracheal systems and her
work is deeply appreciated.

Finally, the following people loaned

SpTDKn FA\rii,v ANVPiiAEXinAK • Platnirk

207

specimens from their private collections
or from their cited institntions: Paul II.
Aniaud, Jr. (California Academy of Sci-
ences), Joseph A. Beatty, Jr., James E.
Carico, John A. L. Cooke (American Mu-
seum of Natural History), Charles D.
Dondale (Canadian National Collections),
R. R. Forster (Otago Museum), Willis J.
Gertsch (American Museum of Natural
History), Al Jung, B. J. Kaston, Robin
Leech, William B. Peck, Vince Roth, Rich-
ard J. Sauer (Michigan State University),
William A. Shear, Bea Vogel, H. K. Wal-
lace, Fred R. Wanless (British Museum,
Natiu-al History), and Howard V. Weems
(Florida State Collection of Arthropods).

THE FAMILY STATUS OF
ANYPHAENIDAE

Simon considered tlie anyphaenids to be
a subfamily of the large family Clubionidae
and used as the key character for distin-
guishing the anyphaenids the advanced
placement of the tracheal spiracle. Later
authors, notably Petrunkevitch and Bris-
towe, thought this character so significant
that they gave the anyphaenids family sta-
tus, though still believing the group to be
closely related to the Clubionidae. The
comparatively recent discovery that in
some families closely related, congeneric
species sometimes have very different res-
piratory systems (see Levi, 1967) has led
most arachnologists to denigrate the im-
portance of respiratory structm-es as macro-
taxonomic characters. Thus most modern
arachnological works still treat the any-
phaenids as a subfamily of Clubionidae.
A notable exception, however, is Lehtinen
( 1967 ) , who maintains ( correctly, I be-
lieve) that the classical family Clubionidae
is a highly polyphyletic assemblage of un-
related two-clawed spiders that lack any
noticeable modifications of the body. Leh-
tinen splits the clubionids into several fam-
ilies, largely but not strictly along the lines
of the old subfamily divisions, and accords
the anyphaenids full status as a family.

Forster (1970) agrees with this assessment
of the anyphaenids.

To check on the validity of this classifi-
cation, a variety of clubionid genera were
examined and compared with anyphaenids,
with the result that the anyphaenids arc
here considered a distinct family, for two
major reasons. One is the classical reason
— the tracheal system. Examination of the
tracheae of males and females of the club-
ionids CAuhiona o])esa Hentz, Chiracan-
thium mildei L. Koch, Trachelus tratujuil-
his- (Hentz), Castkineira cin<i,uluta (C. L.
Koch), Agroeca pratensis Emerton, Phniro-
timpus alarius (Hentz), and the any-
phaenids Am/pJmena celer (Hentz), Amj-
phaena pectorosa L. Koch, Amjphaena
calif ornica ( Banks ) and Aijsha <i,racilis
(Hentz) disclosed three major differences
between anyphaenid and clubionid tra-
cheae (see Methods for the technique
used). First, anyphaenid tracheae extend
through the pedicel of the spider into the
cephalothorax and legs, while those of
clubionids are restricted to the abdomen
(see Figs. 47 and 50). Associated with this
is the externally observable advanced
placement of the tracheal spiracle in any-
phaenids. Second, the tiacheae are rela-
tively much larger in anyphaenids. In all
the clubionids examined, even the main
tracheal tubes are very thin and narrow;
anyphaenid tracheae are three to four
times as wide. Third, none of the clubionid
species examined showed any sexual di-
morphism in the tracheal system, whereas
male anyphaenids have considerably larger
tracheae than do the females. The size
of the tracheae may be correlated with the
high activity levels of anyphaenids: my
collecting experience indicates that they
can run extremely rapidly when disturbed.

The larger size of the tracheae in males
may be associated with the increased respi-
ration necessary for the extra activity re-
(juired to locate, court and copulate with
a female. Anyphaenid courtship is (.ex-
tremely active; films of the courtship of
Anijphacna accentiiata show that the abdo-

208 Bulletin Museum of Comparative Zoology, Vol. 146, No. 4

embolus

median apophysis

conductor

retroloteral tibial
apophysis (RTA)

Text-Figures 1-3. Claw tuft of Clubiona obesa Hentz, lateral view, diagrammatic. 2. Claw tuft of Aystia gracilis
(Hentz), lateral view, diagrammatic. 3. Generalized palpal structure of Anyphaena.

men of the male is vibrated up and down
so rapidly that only a blur is visible
(Thompson, G. H., and E. R. Skinner,
Courtship in Spiders, Oxford Scientific
Films). Although the mating behavior of
very few species in either group has been
studied in detail, the vast difference be-
tween anyphaenid courtship and the rather
sluggish courtship behavior of Clubiona
and related genera wovild seem to provide
additional evidence for separating the two
groups (Platnick, 1971).

Evidence that is probably just as impor-
tant as the tracheae for considering Any-
phaenidae a distinct group is provided by
the claw tufts. Clubionids have claw tufts
that are composed of numerous straight
simple setae densely clumped together
( Text-fig. 1 ) . Anyphaenid claw tufts, how-
ever, are composed of two rows of large,
lamelliform setae that are greatly expanded
at their distal ends (Text-fig. 2). All the
anyphaenids examined have these peculiar,

easily recognizable claw tufts, but so far
as known, no clubionids do, though some
phruroliths have superficially similar claw
tufts.

For these reasons, Anyphaenidae is here
considered a distinct family not very
closely related to any of the groups cur-
rently included in the Clubionidae.

RELATIONSHIPS OF THE FAMILY
ANYPHAENIDAE

In addition to the clubionids, the any-
phaenids have been associated with three
other families of spiders: Amaurobiidae,
Argyronetidae and Amaurobioididae. Leh-
tinen (1967) placed the anyphaenids in
his branch Amaurobiides and stated that
they are probably derived from Amauro-
biidae: Macrobuninae and therefore lack
ecribellate, two-clawed relatives. Forster
( 1970 ) agreed with the placement of Any-
phaenidae in Amaurobiides (and specifi-

Spider Family Axypiiaenidae • Plalnick

209

cally included the family in his supeifainily
Dictynoidca) but cited the families Ar-
gyionetidae and Amaurobioididae as close
relatives. Representatives of all three fami-
lies were examined to determine the degree
of their relationship, if any, to the Any-
phaenidae.

Lehtinen gave no evidence for his state-
ment that the anyphaenids are probably
deri\'atives of Amaurobiidae: Macrobuni-
nae, presumably because there seems to be
none. An examination of specimens of
one genus in this subfamily, Arctohius
Lehtinen, indicates that it would be diffi-
cult to find araneomorph spiders less likely
to have given rise to the anyphaenids. The
species of Arctohius are cribellate, three-
clawed spiders that lack claw tufts and
possess an unelaborated tracheal system.
Further, the genitalia show no similarities
to those of anyphaenids.

Likewise, Forster gave no evidence for
associating the family Argyronetidae with
tlie an\q3haenids; his decision to do so was
based, I believe, on the similarities in the
tracheal systems of the two groups. The
elaboration of the tracheal system in Ar-
gyroneta, however, is probably associated
with their invasion of an aquatic habitat
and the resultant demands on the respira-
tory system. All the other characters, in-
cluding the three claws, lack of claw tufts
and the characteristic pattern of ti'icho-
bothria distribution, indicate that Argt/ro-
neta is, as it is usually regarded, a close
relative (if not actually a member) of the
family Agelenidae.

The family Amaurobioididae was cre-
ated by Hickman ( 1949 ) for the single
genus Ammirohioides O. P. -Cambridge,
which has at various times been included
in the families Drassidae ( = Gnaphosi-
dae), Ctenidac, Clubionidae and Miturgi-
dae. The genus is known from New Zea-
land, Tasmania, southern Chile and South
Africa. The spiders live in rock crevices
in the tidal zone, where they build tubular
silk retrtnits and are regularly submerged
at high tide (Lamoral, 1968).

Specimens of this rare genus provided by
R. R. Forster revealed not only a typically
anyphac>nid-]ike tracheal system, but also
the lamelliform claw tufts so characteristic
of anyphaenids. Further, the genitalia are
close to those of the anyphaenid genus
Oxysouui, and the body form is similar to
that of several species of anyphaenids
known from Chile, Peru, and Argentina.
For these reasons, the family Amauro-
bioididae is newly synonymized with
Anyphaenidae in the taxonomic section of
this paper.

Thus the problem of the correct macro-
taxonomic placement of Anyphaenidae has
been clarified but not solved by this study
of the groups with which the family has
been associated in the past. Futiu-o work
should start with an examination of the
family Miturgidae (as construed by Leht-
inen ) .

Although it was necessary to limit the
scope of the detailed revision to the man-
ageable number of species occurring north
of Mexico, all available specimens from
other areas were examined to gain an
overview of the family. Preliminary im-
pressions indicate that the family probably
originated in the southern half of South
America with subsequent radiations north-
ward. As indicated by the ability of
Amaurohioides to withstand prolonged sub-
mersion, it is likely that early anyphaenids
were able to survive hydrochore dispersal
by rafting, etc., across considerable ex-
panses of water.

GENERIC PROBLEMS IN THE
ANYPHAENIDAE

The generic taxonomy of anyphaenids
is currently chaotic. Every author who has
worked with the group, including Petrunke-
vitch (1930), Bryant (1931) and Chicker-
ing (1937), has expres.sed frustration at
the confusion and ambiguity in the use of
many of the most common generic names.
One of the principal causes of this con-
fusion is the interesting e\olutionary pat-
tern encountered time and again witliin

210 Bulletin Museum of Comparative Zoology, Vol. 146, No. 4

this family: species tend to occur in groups
that are remarkably homogeneovis in
genitalic structure but quite distinct from
other such groups. Often many of the
species in these groups are sympatric, are
found in a rather limited area and are
clearly the result of radiation within that
area. An excellent example of this is the
occurrence of nine closely related species
of the Anyphaena celer species group in
the mountains of southeastern Arizona. It
is tempting to consider each of these
groups a genus, as unambiguous key char-
acters are then available to distinguish
genera. Such an approach would at least
double the number of genera found in the
United States, and, if applied to the Central
and South American fauna, would neces-
sitate the creation of a vast number of
new genera. If, instead, characters refer-
ring to the general body form are used, a
more workable classification in terms of
both number and size of genera results.
Unfortunately, this makes the unambiguous
definition of genera much more difficult
and makes keys to genera awkward and
cumbersome. With either approach, how-
ever, reliable genera composed of mo-
nophyletic groups of species can be estab-
lished.

The second approach to anyphaenid clas-
sification has been taken by the majority
of former authors, and is continued in this
work. Thus the European genus Any-
phaena is used for the bulk of the any-
phaenids occurring in the United States,
even though only one of our species, Any-
phaena aperta, is actually a close relative
of the European Anyphaena accentuata,
type species of the genvis. Nonetheless,
all the species here included in Anyphaena
share a basic body form. The neotropical
genus Wulfila is used for all the pale,
long-legged species, even though they are
genitalically quite diverse; the other genera
used here are similarly construed. Although
this system is not wholly satisfactory, it
seems decidedly better than creating a
host of new generic names that are likely

to fall into synonymy when a detailed ge-
neric revision of the group as a whole can
be carried out.

METHODS

Tracheae were examined by dissecting
away the dorsal cuticle of the abdomen
and boiling the spider in ten percent
sodium hydroxide for ten minutes. By this
method, all the soft structures in the
abdomen are digested away, leaving the
tracheae intact.

Types of the new species are being
deposited in the American Museum of
Natural History, New York City, and the
Museum of Comparative Zoology, Harvard
University. Type depositories are abbrevi-
ated as follows: AMNH — American Mu-
seum of Natural History, BMNH — British
Museum, Natural History, MCZ — Museum
of Comparative Zoology.

Measurements and drawings were made
with a standard ocular grid. Measurements
of gross morphological featiu-es are ac-
curate to ~ 0.04 mm; measurements of
ocular featiu'es are accurate to ~ 0.01 mm.
Rather than selecting a small number of
measurements and providing means and
standard deviations for these on the basis
of a small series of specimens, one male
and one female of each species were
measured in detail. As only one of the
species included here shows any significant
variation in size, this procedure was deemed
more informative. Actual measurements
are given rather than ratios since in many
cases (e.g., Anyphaena catalina and A.
arhida) closely related species differ sig-
nificantly in size but not in their relative
proportions. Most of the measurements
taken are self-explanatory, though a few
need furdier comment. Cephahc width
refers to the width of the carapace at a
point just behind the posterior median
eyes, and thus provides an indication of
the degree to which the carapace is nar-
rowed in front.

The difficult problem of accurately de-
scribing die eye relationships has been

Spider Family Anyphaenidae • Platnick 211

solved by providing a set of measurements
from wliich it is possible to reconstrnct,
using grapli paper, the exact eye arrange-
ment. Diameters are given using the con-
ventional abbreviations (AME = anterior
median eye, ALE = anterior lateral eye,
PME = posterior median eye, PLE =
posterior lateral eye). The length of each
eye row is measured from the lateral edge
of one lateral eye to the lateral edge of
tlie other lateral eye. Curvature of the eye
rows is described as viewed frontally, not
dorsally. This was accomplished by posi-
tioning the spider in sand, a technique
found most useful for making all the
measurements. The dimensions of the
median ocular quadrangle (MOQ) are
given, as well as the distances between
each of the eyes. The latter measurements
extend between the edges of the lenses of
the eyes under consideration (not just
between the dark circles surrounding each
eye).

The relative length and thickness of each
leg is indicated by the tibial length index —
the tibial width divided by the tibial
length, with the result multiplied by 100
to obtain a whole number. All tibial
measurements were taken from a dorsal
view and refer to the maximum lengths
and widths. The lower the tibial index,
the longer and thinner the leg; conversely,
the higher the index, the shorter and thicker
the leg. In practice the index varies from
around 3 to 35.

Ventral spination of the leg segments is
indicated by the standard formula in which
the number of spines on the proximal,
median and distal thirds of the leg segment
are given. Only ventral spines, not lateral
ones, are included, and any even number
in the formula may be taken to represent
a pair of spines. Unless the last number is
followed by an asterisk, the last pair of
spines is terminally located. Thus, for
example, the formula 2-2-2* indicates that
the segment bears three pairs of ventral
spines, the last pair of which is not termi-
nally located. The term "spine" is used in

its conventional arachnological sense and
rc'fers to the moxable macrosetae found
on the legs. Similarly, the term "clypeus"
is used to refer to the area between the
anterior eye row and the anterior edge of
tlie carapace and not to the small sclerite
folded under the carapace. Since neither
usage of tht> term reflects certain knowledge
of homology with the insect clypeus, the
old and established usage should be main-
tained.

Scale lines for the drawings always equal
0.1 mm. Each scale line applies to all
consecutively numbered drawings imtil a
new scale line appears. Exceptions are
noted in the captions.

TAXONOMY
Anyphaenidae

Anyphaenidae Bertkau, 1878, Arch. Naturg., 44:
358, 379. Tvpe genus Anyphaena Sundevall,
1833.

Amaurobioididae Hickman, 1949, Pap. Proc. Roy.
Soc. Tasmania, 1948: 31. Tvpe genus A77iauro-
hioides O.P.-Cambridge, 1883. NEW SYN-
ONYMY.

Diagnosis. The combination of the ad-
vanced tracheal spiracle and the lamelli-
form claw tufts will serve to distinguish the
anyphaenids from all other families.

Description. Chelicerae diaxial, not fused
together at base. Labium free. Without
cribellum or calamistrum. With one pair
of book lungs and a tracheal spiracle lo-
cated considerably anterior to the spin-
nerets, most often midway between spin-
nerets and epigastric furrow, sometimes
closer to one or the other. Eight eyes in
two rows. Six spinnerets, anterior spin-
nerets approximate, colulus represented
only by hairs, anal tubercle unmodified.
Legs prograde, metatarsi and tarsi I and
II scopulate, tarsi with two toothed claws
and claw tufts composed of lamelliform
setae.

Key to Genera
IN America north of Mexico

la. Tracheal spiracle luucli closer to epigastric
furrow tlian to spinnerets ..Aysha

212 Bulletin Museum of Comparative Zoology, Vol. 146, No. 4

lb. Tracheal spiracle roughly midway between
epigastric furrow and spinnerets 2

2a. Legs very long and thin. Leg I greatly
elongated, tibial index (width/length X
100) usually 5 or less Wulfila

2b. Legs normal, tibial index of leg I usually
8 or more 3

3a. Chelicerae with 2 retromarginal teeth

Oxysoma

3b. Chelicerae with 4-9 retromarginal denticles
4

4a. Carapace usually with two dark paramedian
longitudinal bands; chelicerae not produced
forward; femora not much darker than
other leg segments Anijphaena

4b. Carapace without dark paramedian longi-
tudinal bands; either chelicerae produced
forward or femora much darker than other
leg segments Teudis

Anyphaena Sundevall

Amjphaena Sundevall, 1833, Conspectus Arachn.,
28. Type species by monotypy Aranea ac-
centuata Walckenaer, 1802.

Diagnosis. The combination of the fol-
lowing characters will serve to distinguish
the genus in America north of Mexico:
trachael spiracle roughly midway between
epigastric furrow and spinnerets, leg I not
greatly elongated, chelicerae with 4-9
retromarginal denticles and not produced
forward, femora not much darker than
other leg segments. The carapace usually
has two dark paramedian longitudinal
bands. The genus is used here in a very
broad sense; this prevents simple diagnosis,
and makes detailed descriptions of each
species group more meaningful than a
description of the whole genus.

Uncertain names. Types of the follow-
ing species were unavailable and are too
poorly described to permit identification:
Cluhiona agresiis Hentz, 1847, type de-
stroyed; Chihiona fallens Hentz, 1847, type
destroyed, Cluhiomi suhlurida Hentz, 1847,
type destroyed; Amjphaena argentata
Becker, 1879, type lost; and Amjphaena
striata Becker, 1879, type lost. The three
Hentz Cluhiona species were transferred to
Anijphaeiui by Marx (1890), but there is
little justification for this in the vague

descriptions. All the above names are
regarded as nomina cluhia.

Species groups. Although there seem to
be several species groups of Amjphaena in
the Neotropic region, only four occur north
of Mexico. The celer group is the largest;
it has representatives at least as far south
as Panama and probably contains over
thirty species. The pectorosa and pacifica
groups are closely related and occur com-
monly in Mexico as well as the United
States; it is difficult to place females in
one group or the other unless the male is
also known; they probably contain together
at least twenty species. The accentuata
group is predominantly Palearctic and prob-
ably contains at least five species.

Key to Species Groups

la. Metatarsi I and II with one pair of ventral
spines accentuata group

lb. Metatarsi I and II with two pairs of ventral
spines 2

2a. Retrolateral tibial apophysis of males bifid,
with ventral prong elongated ( Figs. 18-20,
25-32). Epigynum of females with a hood
(Figs. 21, 23, 33, 36, 37, 39-42) ._ celer group

2b. Retrolateral tibial apophysis of males not
bifid or elongated (Figs. 55-58, 69-71).
Epigynum without a hood (Figs. 66, 67,

72, 74, 77, 79) 3

3a. Eastern United States. Coxae III and IV

of males with pointed spins (Figs. 59-62).
Female epigyna on broad sclerotized plates
(Figs. 74, 77, 79); internal genitalia lacking

long ducts (Figs. 75, 78, 80)

pectorosa group

3b. Western United States. Coxae III and IV
of males without pointed spurs, though
rounded knobs may be present. Female
epig>'na not on broad sclerotized plates
(Figs. 66, 67, 72); internal genitalia with
long, sometimes coiling, ducts (Figs. 68,

73, 76) _._. pacifica group

Anyphaena celer Group

Diagnosis. Males of the celer group may
be recognized by their retrolateral tibial
apophysis, which is usually bifid with an
elongated ventral prong (Figs. 18, 20, 26).
Females have a characteristic epigynum
consisting of a hood, two sidepieces and a
midpiece (Figs. 9, 33), though the mid-

SpiDEii Family Axyphakxidae • Pkilnick 213

piece is reduced in A. crebrispimi ;ind A.
(lixiana (Figs. 21, 23).

Description. Total length 3-7 nnn, with
males of most species between 3.3-4.6 mm,
females of most species between 4.1-5.9
mm. Carapace longer than wide, narrowed
in front to less than half its maximum
w idth. Clypens height greater than anterior
median eye diameter. Posterior median,
posterior lateral and anterior lateral eyes
subeqiial in size, larger than anterior
medians. Procurved posterior eye row
longer than recin'\'ed anterior row. Median
ocular ({nadrangle longer than wide in
front, wider than long in back. Anterior
median eyes separated by their diameter,
by their radius from anterior laterals.
Posterior medians separated by their diam-
eter, slightly closer to posterior laterals
than to each other. Anterior laterals
separated by their radius from posterior
laterals. Sternum longer than wide, un-
modified. Chelicerae with 4-5 promarginal
teeth and 6-9 reti-omarginal denticles.
Abdomen longer than wide, tracheal
spiracle midway between epigastric fur-
row and base of spinnerets. Leg formula
1423. Metatarsi I and II with two pairs
of \'entral spines. Males often with femur
III thickened distally, set with stiff short
setae ventrally; tibia III ventral spines
thickened, cone-like; coxae set with clumps
of stiff short setae. Palpus with an elon-
gated median apophysis, retrolateral teg-
ular apophysis, conspicuous curving em-
bolus and conductor. Retrolateral tibial
apophysis bifid, with dorsal prong reduced
in some species. Epigynum with hood,
two sidepieces and midpiece; two simple
spermathecae.

Variation. None of the species in this
group show any significant individual or
geographic intraspecific variation in struc-
tin-e, size or coloration.

Key to Species

la. Dorsal and \entral prongs of retrolateral
tibial apophysis ( RTA ) roughly equal in
length (Figs. 18, 19); epigynal hood wide,
more than four times the minimum width

of epigynal sidepiece ( Figs. 9, 11); east-
ern U.S 2

II). Ventral prong of retrolateral tibial apophy-
sis ( RTA ) nuieh longer than dorsal prong
(as in Figs. 26, 27); epigynal hood
narrow, less than four times the minimum
width of epigynal sidepiece ( as in Figs.
33, 36); western U.S. 3

2a. Dorsal prong of RTA broad, with a trans-
lucent ridge (Fig. 18); epigynal hood a
thick oval, sidepieces straight (Fig. 9)
— - - — _.. celer

21). Dorsal prong of RTA narrow, without a
translucent ridge (Fig. 19); epigynal hood
a thin oval, sidepieces rounded ( Fig. 1 1 )
- - _ inaculata

3a. Base of RTA expanded into a broad
triangle (Fig. 20); retrolateral tegular
apophysis prolonged medially (Fig. 3);
epigynal sidepieces more than three times
the width of epigynal hood (Fig. 21)
- crebrispina

3b. Base of RTA not expanded; retrolateral
tegular apophysis not prolonged medialK ;
epigynal sidepieces less than three times
the width of epigynal hood 4

4a. Dorsal prong of RTA bearing a sharp spur
(Fig. 25); epigynal midpiece greatly re-
duced, sidepieces widely separated pos-
teriorly (Fig. 23) dixiana

4b. Dorsal prong of RTA without a spur;
epigynal midpiece conspicuous, sidepieces
approximate posteriorly 5

5a. Males '. 6

5b. Females 14

6a. Dorsal prong of RTA with two triangular
processes separated by a concave notch
( Fig. 26 ) judicata

6b. Dorsal prong of RTA witliout triangular
processes „. 7

7a. Dorsal prong of RTA with a long recurved
hook (Fig. 29) autumna

7b. Dorsal prong of RTA without a long re-
curved hook 8

8a. Dorsal prong of RTA witli a basal hook
(Figs. 31, 38) _.-.. 9

8b. Dorsal prong of RTA without a basal
hook 10

9a. Conductor and retrolateral tegular apophy-
sis recurved (Fig. 15) catalina

9b. Conductor and retrolateral tegular apoph-
ysis not recurved ( Fig. 17 ) arhida

10a. Dorsal prong of RTA a shaiply pointed

spike ( Fig. 32 ) liespar

101). Dorsal prong of RTA not a sharply

pointed spike 1 1

11a. Fmbolus with a conspicuous enlargement

(Figs. 7, 13) .12

111). Embolus without a con.spicuous enlarge-
ment (Figs. 21, 33) 13

214 Bulletin Museum of Comparative Zoolog,ij, Vol. 146, No. 4

I ^J'', ,^-^ m-p.

Anyphaeno maculota \ q\ —>>

I -*

\ \A-

^~« ',

>-:-:

Anyphoena crebrispino

/—

\\v

Anyphaena dixiana j»- \

v^\ \

Anyphaena rita '

V f

1 ~T
1 1

\h~^

' 1

Anyphaena

cochise V

Map 1. Distributions of Anyphaena arbida, A. autumna, A. catalina, A. celer, A. cochise, A. crebrispina, A. dixi-
ana, A. gibboides, A. hespar, A. judicata, A. maculata, A. marginalis and A. rita.

12a. Dorsal prong of RTA more than half the

length of ventral prong ( Fig. 35 )

cochise

12b. Dorsal prong of RTA less than half the

length of ventral prong (Fig. 28)

rita

13a. Median apophysis sharply pointed; con-
ductor short, bent (Fig. 14); Oregon and
Utah gibboides

13b. Median apophysis rounded; conductor
long, straight (Fig. 6); Arizona and New
Mexico marginalis

14a. Epigynal hood wider than long; midpiece
not wider than hood, without constric-
tions; sidepieces very wide (Fig. 40);
Oregon and Utah gibboides

14b. Epigynal hood as long as wide or midpiece
wider than hood or sidepieces narrow;
Arizona and New Mexico 15

15a. Epigynal midpiece a very broad triangle
( Fig. 37 ) rita

15b. Epigynal midpiece othenvise 16

16a. Spermathecae much further apart pos-
teriorly than anteriorly ( Fig. 49 ) _-, hespar

16b. Spemiathecae as far apart anteriorly as
posteriorly 17

17a. Epigynal hood much wider than long ( Fig.
39 ) _ autumna

17b. Epigynal hood as long as wide —18

18a. Epigynal midpiece less than twice the

length of epigynal hood (Fig. 41)

catalina

18b. Epigynal midpiece more than twice the
length of epigynal hood ...19

19a. Epigynal midpiece a short triangle (Fig.
33 ) judicata

19b. Epigynal midpiece an elongate triangle
(Fig. 36) marginalis

Anyphaena celer (Hentz)
Map 1; Figures 1, 9, 10, 18

Chibiona celer Hentz, 1847, J. Boston Soc. Natur.
Hist., 5: 452, pi. 23, fig. 20 ( 9 ). Male holo-
type, female allotype from Alabama and North
Carolina in the Boston Soc. Natur. Hist.
(Boston Museum of Science), destroyed by
beetles.

Anyphaena incerta Keyserling, 1887, Verb. zool.
hot. Ces. Wien, .37: 452, pi. 6, fig. 22 ( $ ).
Female holotype from Cambridge, Massachu-
setts, in MCZ, examined. Emerton, 1890, Trans.
Connecticut Acad. Sci., 8: 186, pi. 6, figs.
2-2d, $,9.

Anyphaena celer, Simon, 1897, Hist. Natur.
Araign., 2: 96. Bryant, 1931, Psyche, 38: 111,
pi. 6, fig. 9, pi. 8, figs. 25, 28, $, 9. Chickering,
1939, Pap. Michigan Acad. Sci., 24: 51, figs.

Spider Family Axyphakxidae • PJaliiick 215

Plate 1
Figures 1-8. Left palpi, ventral view. Figures 9. 11. Epigyna, ventral view. Figures 10, 12. Internal genitalia,
dorsal view. 1,9,10. /Anyphaena ce/er (Hentz). 2,11,12. Anyphaena maculata (Banks). 3. Anyphaena crebri-
sp'ma Chamberlin. 4. Anyphaena dixiana (Chamberlin and Woodbury). 5. Anyphaena judicata O. P. -Cambridge.
6. Anyphaena marginalis (Banks). 7. Anyphaena rita new species. 8. Anyphaena autumna new species.

216 Bulletin Museum of Comparative Zoology, Vol. 146, No. 4

t>'

1-4, $, 9. Comstock, 1940, Spicier Book,
rev. ed., p. 577, figs. 634-6.35, $,9. Kaston,
1948, Bull. Connecticut Geol. Natur. Hist.
Surv., 70: 407, figs. 1471-1476, $,9. Roewer,

1954, Katalog der Araneae, 2:528. Bonnet,

1955, Bibliographia Araneorum, 2: 343.
Gatjenna celer, Comstock, 1912, Spider Book, p.

563, figs. 634-635, $,9.

Dia<ino.sis. Anyphaena celer is most
closely related to A. maculata. Males of
both species have dorsal and venti-al RTA
prongs roughly equal in length, but A. celer
males may be distinguished by the trans-
lucent ridge on their dorsal prong (Fig.
18). Females may be separated by the
straight epigynal sidepieces and widely
oval epigynal hood of A. celer (Fig. 9).

Male (Jackson Co., Illinois). Total
length 4.54 mm. Carapace 2.12 mm long
1.58 mm wide, cephalic width 0.83 mm
clypeus height 0.07 mm, pale yellow with
thin dark broken border and two dark
paramedian longitudinal bands. Eyes:
diameters (mm): AME 0.06, ALE 0.12,
PME 0.11, PLE 0.12; anterior eye row
0.44 mm long, slightly recurved; pos-
terior eye row 0.60 mm long, procurved;
MOQ length 0.24 mm, front width 0.19
mm, back width 0.31 mm; eye interdis-
tances (mm): AME-AME 0.06, AME-
ALE 0.03, PME-PME 0.10, PME-PLE
0.10, ALE-PLE 0.05.

Sternum 1.06 mm long, 0.94 mm wide,
pale yellow with dark markings opposite
coxae, translucent border and darkened
extensions to coxae. Chelicerae 0.79 mm
long with 4 promarginal teeth and 8 retro-
marginal denticles, pale yellow with boss
outlined in gray. Labium and endites pale
yellow, darkest proximally; endites not
invaginated.

Abdomen 2.30 mm long, 1.62 mm wide,
pale white with transverse rows of dark
markings; venter with scattered dark mark-
ings. Epigastric furrow 0.85 mm from
tracheal spiracle, spiracle 0.88 mm from
base of spinnerets.

Legs pale yellow with scattered dark
markings. Tibial lengths (mm) and
indices: I 1.98, 12; II 1.82, 14; III 1.17, 21;

IV 1.73, 16. Ventral spination: tibiae I-IV
2-2-2; metatarsi I, II 2-2-0, III 2-0-2, IV
2-2-2. Femur III thickened distally with
clump of short thick setae ventrally. Tibia
III ventral spines 1, 2 on retrolateral side
thickened, cone-like. Coxae III, IV pro-
lateral ventral surface with clump of short
thick setae.

Palpus as in Figures 1, 18.

Female (Wayne Co., Ohio). Coloration
as in male. Total length 5.87 mm. Carapace
2.07 mm long, 1.39 mm wide, cephalic
width 0.86 mm, clypeus height 0.05 mm.
Eyes: diameters (mm): AME 0.05, ALE
0.10, PME 0.10, PLE 0.10; anterior eye row
0.42 mm long, recurved; posterior eye row
0.58 mm long, procurved; MOQ length
0.30 mm, front width 0.19 mm, back width
0.32 mm; eye interdistances (mm): AME-
AxME 0.07, AME-ALE 0.05, PME-PME
0.12, PME-PLE 0.07, ALE-PLE 0.08.

Sternum 0.99 mm long, 0.88 mm wide.
Chelicerae 0.76 mm long with teeth as in
male.

Abdomen 4.10 mm long, 2.13 mm wide.
Epigastric furrow 1.57 mm from tracheal
spiracle, spiracle 1.37 mm from base of
spinnerets.

Legs unmodified. Tibial lengths (mm)
and indices: I 1.58, 15; II 1.42, 16; III
1.01, 24; IV 1.60, 15. Ventral spination:
tibiae I, II 2-2-2, III, IV 1-1-2; meta-
tarsi I, II 2-2-0, III 2-0-2, IV 2-2-2.

Epigynum as in Figure 9, internal geni-
talia as in Figure 10.

Natural history. Mature males have been
taken every month except June, mature
females year-round. Specimens have been
taken in houses, deciduous forests, on
leaves, flowers, ti-eesides, in pitfalls and
footprints in snow.

Distribution. Eastern United States from
southern New England west to Wisconsin,
south to Florida and Texas (Map 1).

Anyphaena maculata (Banks)
Map 1; Figures 2, 11, 12, 19

Caijcnna maculata Banks, 1896, Trans. Amer.
Ent. Soc, 23: 64. Male holotype from Wash-
ington, D.C., in MCZ, examined. Bishop and

Spider Family Anyphaenidak • Plat nick 217

N^^

Figures 13-17. Left palpi, ventral view,
ventral view. Figures 22, 24. Interna
Anyphaena gibboides new species. 15.
17. Anyphaena arbida new species. 18
22. Anyphaena crebrispina Channberlin.

Plate 2
Figures 18-20. Left palpi, retrolateral view. Figures 21, 23. Epigyna,

genitalia, dorsal view. 13. Anyphaena cochise new species. 14.
Anyphaena catalina new species. 16. Anyphaena hespar new species.

Anyphaena celer (Hentz). 19. Anyphaena maculata (Banks). 20. 21,
23, 24. Anyphaena dixlana (Chamberlin and Woodbury).

218 Bulletin Museum of Comparative Zoology, Vol. 146, No. 4

Plate 3
Figures 25-32. Left palpi, retrolateral view. Figure 33. Epigynum, ventral view. Figure 34. Internal genitalia,
dorsal view. 25. Anyphaena dixiana (Chamberlin and Woodbury). 26, 33, 34. Anyphaena judicata O. P. -Cam-
bridge. 27. Anyphaena marginalis (Banks). 28. Anyphaena rita new species. 29. Anyphaena autumna new
species. 30. Anyphaena gibboides new species. 31. Anyphaena catalina new species. 32. Anyphaena hespar
new species.

Spider Family Anyphaenidae • Platnick 219

Crosby, 1926, T. Elislia Mitclioll Sci. Soc, 41:
189, pi. 24, figs. 37, 38, $, 9.
Amjphacna nuiciihita, Simon, 1897, Hist. Natiir.
Araign., 2: 96. Brvant, 1931, Psyche, 38: 111,
pi. 6, fig. 8, pi. 8, fig. 31, $, 9. Kaston, 1948,
Bull. Connecticut Geol. Natur. Hist. Surv.,
70: 409, figs. 1457-1458, $,9. Roewer, 1954,
Katalog der Araneae, 2: 529. Bonnet, 1955,
Bibliographia Araneoruni, 2: 345.

Diagnosis. Anyphoena inaculata is most
closely related to A. celer. Males may be
distinguished by the short dorsal prong of
the RTA, which lacks a translucent ridge
(Figure 19); females by their rounded
epigynal sidepieces and narrowly oval
epigynal hood (Figure 11).

Male (Durham Co., North Carolina).
Coloration as in Amjphaena celer. Total
length 3.74 mm. Carapace 2.09 mm long,
1.54 mm wide, cephalic width 0.77 mm,
clvpeus height 0.08 mm. Eyes: diameters
(mm): AME 0.07, ALE 0.11, PME 0.10,
PLE 0.10; anterior eye row 0.44 mm long,
recurved; posterior eye row 0.58 mm long,
procurved; MOQ length 0.23 mm, front
width 0.20 mm, back width 0.31 mm; eye
interdistances (mm): AME-AME 0.05,
AME-ALE 0.03, PME-PME 0.10, PME-
PLE 0.10, ALE-PLE 0.04.

Sternum 1.08 mm long, 0.79 mm wide.
Chelicerae 0.63 mm long with 4 promar-
ginal teeth and 8 retromarginal denticles.

Abdomen 2.02 mm long, 1.08 mm wide.
Epigastric furrow 0.31 mm from tracheal
spiracle, spiracle 0.41 mm from base of
spinnerets.

Tibial lengths (mm) and indices: I 2.00,
11; II 1.75, 13; III 1.12, 24; IV 1.82, 15.
Ventral spination: tibiae I, II 2-2-2*, III,
IV 2-2-2; metatarsi I 2-1-0, II 2-2-0, III
2-0-2, IV 2-2-2. Modifications of third
leg as in A. celer.

Palpus as in Figures 2, 19.

Female (Pope Co., Illinois). Coloration
as in male of A. celer.

Total length 4.68 mm. Carapace 2.07
mm long, 1.60 mm wide, cephalic width
0.97 mm, clypeus height 0.08 mm. Eyes:
diameters (mm): AME 0.08, ALE 0.10,
PME 0.10, PLE 0.11; anterior eye row 0.48

mm long, recurved; posterior eye row 0.63
mm long, procur\'ed; MOQ length 0.30
mm, front width 0.22 mm, back width 0.33
nun; eye interdistances (mm): AME-
AME 0.07, AME-ALE 0.04, PME-PME
0.14, PME-PLE 0.11, ALE-PLE 0.05.

Sternum 1.15 mm long and 0.95 mm
wide. Chelicerae 0.71 mm long with teeth
as in male.

Abdomen 3.02 mm long, 2.11 nun wide.
Epigastric furrow 0.85 mm from tracheal
spiracle, spiracle 0.90 mm from base of
spinnerets.

Legs unmodified. Tibial lengths (nun)
and indices: I 1.42, 18; II 1.48, 18; III 0.99,
25; IV 1.58, 16. Ventral spination: tibiae
I, II 2-2-2', III 1-1-2, IV 2-1-2; metatarsi
I, II 2-2-0, III 2-0-2, IV 2-2-2.

Epigynum as in Figure 11, internal geni-
talia as in Figure 12.

Natural history. Mature males have been
taken from late September through early
February, mature females from mid-Octo-
ber through mid- April. Specimens have
been taken from Spanish moss, by sweep-
ing in bottomland pine and hardwood
forests, by sifting leaves and by Malaise
trap.

Distribution. Mid-eastern states from
Long Lsland south to North Carolina, wost
to southern Illinois, eastern Missouri and
northern Alabama ( Map 1 ) .

Anyphaena crebrispina Chamberlin
Map 1; Figures 3, 20, 21, 22

Aiufpliacna crebrispina Chamberlin, 1919, Pomona
Coll. J. Ent. Zool., 12: 10, pi. 4, fig. 4 {$).
Male holotype from Clareniont, California, in
MCZ, examined. Bryant, 1931, Psyche. 38: 113,
pi. 6, fig. 11, $. Roewer, 1954, Katalog der
Araneae, 2: 528. Bonnet, 1955, Bibliographia
Araneoruni, 2: 343.

Amjphaena zina Chamberlin, 1919, Pomona Coll.
J. Ent. Zool., 12:11, pi. 4, fig. 5(9). Female
holotype from Clareniont, California, in MCZ,
examined. Roewer, 1954, Katalog der Araneae,
2: 530. Bonnet, 1955, Bibliographia Araneorum,
2: 349. NEW SYNONYMY.

Diagnosis. Anyphaena crebrispina is the
most aberrant member of the celer group,
but is most closely related to A. dixiatm.

220 Bulletin Museum of Comparative Zoology, Vol. 146, No. 4

Plate 4
Figures 35, 38. Left palpi, retrolateral view. Figures 36, 37, 39-42. Epigyna, ventral view. Figures 43-46, 48,
49. Internal genitalia, dorsal view. Figure 47. Anyphaenid tracheae, diagrammatic. Figure 50. Clubionid
tracheae, diagrammatic. 35. Anyphaena cochise new species. 36, 43. Anyphaena marginalis (Banks). 37, 44.
Anyphaena rita new species. 38. Anyphaena arbida new species. 39, 45. Anyphaena autumna new species.
40,46. Anyphaena gibboides new species. 41,48. /Anyp/7aeA?a ca?a//na new species. 42,49. Anyphaena hespar
new species.

Spidkh I-'amily Anyphaexidak • I'latnick 221

Males of A. cre])ris-}nna may be readily dis-
tinguished by the greatly expanded base
of the RTA (Fig. 20). If' this speeies were
known solely from the female, it would be
impossible to place it in the celer group:
the epigynum, with its greatly expanded
sidepieces and its hick of an c>xternally visi-
ble midpiece, is totally unlike that of any
other species in this group ( Fig. 21 ) .

Male (Los Angeles Co., California).
Coloration as in AnypJiaena celer. Total
length 4.61 mm. Carapace 2.00 mm long,
1.57 mm wide, cephalic width 0.74 mm,
clypeus height 0.10 mm. Eyes: diameters
(mm): AME 0.07, ALE 0.10, PME 0.10,
PLE 0.10; anterior eye row 0.43 mm long,
recurved; posterior eye row 0.56 mm long,
prociuved; MOQ length 0.25 mm, front
width 0.18 mm, back width 0.29 mm; eye
interdistances (mm): AME-AME 0.04,
AME-ALE 0.02, PME-PME 0.09, PME-
PLE 0.09, ALE-PLE 0.04.

Sternum 1.10 mm long, 0.S8 mm wide.
Chelicerae 0.55 mm long with 5 promar-
ginal teeth and 8 retromarginal denticles.

Abdomen 2.65 mm long, 1.58 mm wide.
Epigastric furrow 0.79 mm from tracheal
spiracle, spiracle 0.68 mm from base of
spinnerets.

Tibial lengths (mm) and indices: I L69,
13; II 1.51, 15; III 1.06, 22; IV 1.69, 14.
Ventral spination: tibiae I 2-2-2*, II 1-2-
2\ III 1-2-2, IV 2-2-2; metatarsi I, II 2-
2-0, III 2-0-2, IV 2-2-2. Modifications of
third leg as in A. celer.

Palpus as in Figures 3, 20.

Female (Los Angeles Co., California).
Coloration as in male of A. celer. Total
length 4.39 mm. Carapace 1.85 mm long,
1.37 mm wide, cephalic width 0.77 mm,
clypeus height 0.08 mm. Eyes: diameters
(mm): AME 0.07, ALE 0.09, PME 0.09,
PLE 0.09; anterior eye row 0.41 mm long,
recurved; posterior eye row 0.56 mm long,
procurved; MOQ length 0.23 mm, front
width 0.18 mm, back width 0.28 mm; eye
interdistances (mm): AME-AME 0.04,
AME-ALE 0.02, PME-PME 0.10, PME-
PLE 0.08, ALE-PLE 0.04.

Sternum 1.08 mm long, 0.86 mm wide.
Chelieerac> 0.64 mm long with 4 promar-
ginal teeth and 9 retromarginal denticles.

Abdomen 2.99 nun long, 1.98 mm wide.
Epigastric furrow 0.95 mm from tracheal
spiracle, spiracle 0.90 mm from base of
spinnerets.

Legs unmodified. Tibial lengths (mm)
and indices: I 1.39, 15; II 1.31, 16; III 0.75,
28; IV 1.44, 14. Ventral spination: tibiae
I 2-2-2*, II 1-2-0, III 1-1-0, IV 1-1-2;
metatarsi as in male.

Epigynum as in Figure 21, internal geni-
talia as in Figure 22.

Natural history. Mature males have been
taken in November, mature females from
early December through late April. Speci-
mens have been taken by Berlese funnel
sampling of grape bark.

Distribution. Central and southern Cali-
fornia ( Map 1 ) .

Anyphaena dixiana (Chamberlin and
Woodbury), new combination
Map 1; Figures 4, 23, 24, 25

Gaijcnna dixiana Chamberlin and Woodbun',
1929, Proc. Biol. Soc. Washington, 42: 138,
pi. 1, fig. 3 ( 9 ). Female holotype from St.
Ceorge, Utah, in AMNH, examined. Roewer,
1954, Katalog der Araneae, 2: 540 ( G. dixima
[sic]). Bonnet, 1957, Bibliographia Araneorum,
2: 1977.

Anyphaena coloradensis Bryant, 1931, Psyche,
38: 112, pi. 6, figs. 9, 10, pi. 7, figs. .30, 33
{ $, 9 ). Male holotype, female allotNpc from
Boulder, Colorado, in MCZ, examined. Roewer,

1954, Katalog der Araneae, 2: 528. Bonnet,

1955, Bibliographia Araneorum, 2: 343. NEW
SYNONYMY.

Diafinosis. This distinctive species is
closest to AmjpJuiena crebrispina, but may
be quickly recognized by the spur borne
on the dorsal prong of the RTA of males
(Fig. 25) and the greatly reduced epigynal
midpiece of females (Fig. 23).

Male (Cochise Co., Arizona). Coloration
as in Anyphaena celer except that posterior
.spiimerets have dorsal surface sharply di-
vided into dark brown lateral and pale
orange median halves.

Total length 3.85 mm. Carapace 1.67

222 Bulletin Museum of Comparative Zoology, Vol. 146, No. 4

mm long, 1.44 mm wide, cephalic width
0.65 mm, clypeus height 0.07 mm. Eyes:
diameters (mm): AME 0.06, ALE 6.09,
PME 0.09, PLE 0.10; anterior eye row 0.39
mm long, recurved; posterior eye row 0.51
mm long, procurved; MOQ length 0.25
mm, front width 0.16 mm, back width 0.26
mm; eye interdistances (mm): AME-
AME 0.04, AME-ALE 0.03, PME-PME
0.08, PME-PLE 0.08, ALE-PLE 0.05.

Sternum 0.96 mm long, 0.76 mm wide.
Chelicerae 0.53 mm long with 4 promar-
ginal teeth and 6 reti'omarginal denticles.
Endites slightly invaginated at middle.

Abdomen 2.56 mm long, 1.49 mm wide.
Epigastric furrow 0.76 mm from tracheal
spiracle, spiracle 0.76 mm from base of
spinnerets.

Tibial lengths (mm) and indices: I 1.69,
11; II 1.37, 14; III 0.81, 28; IV 1.44, 16.
Ventral spination: tibiae I, II 2-2-2*, III,
IV 1-2-2; metatarsi I, II 2-2-0, III 2-0-2,
IV 2-2-2. Modifications of third leg as in
A. celer.

Palpus as in Figures 4, 25.

Female (Cochise Co., Arizona). Colora-
tion as in male. Total length 4.14 mm.
Carapace 2.03 mm long, 1.57 mm wide,
cephalic width 0.86 mm, clypeus height
0.09 mm. Eyes: diameters (mm): AME
0.05, ALE 0.08, PME 0.09, PLE 0.10; an-
terior eye row 0.43 mm long, recurved;
posterior eye row 0.60 mm long, procurved;
MOQ length 0.26 mm, front width 0.20
mm, back width 0.32 mm; eye interdis-
tances (mm): AME-AME 0.09, AME-
ALE 0.05, PME-PME 0.15, PME-PLE
0.09, ALE-PLE 0.07.

Sternum 1.15 mm long, 0.86 mm wide.
Chelicerae 0.71 mm long with 5 promar-
ginal teeth and 8 retromarginal denticles.

Abdomen 2.50 mm long, 1.69 mm wide.
Epigastric furrow 0.60 mm from tracheal
spiracle, spiracle 0.67 mm from base of
.spinnerets.

Legs unmodified. Tibial lengths (mm)
and indices: I 1.46, 16; II 1.33, 17; III 0.94,
24; IV 1.49, 17. Ventral spination as in
male.

Epigynum as in Figure 23, internal geni-
talia as in Figure 24.

Natural history. Mature males have been
taken from mid- August through mid-May,
mature females from late September
through late April. Specimens have been
taken from 5400 to 9000 feet (1650-2750
m), in yellow pine/ oak and montane for-
ests, in alfalfa, under dead agave and fre-
quently in houses.

Distribution. Northcentral Colorado south
to western Texas, west to southern Cali-
fornia (Map 1 ) .

Anyphaena judicata O. P.-Cambridge
Map 1; Figures 5, 26, 33, 34

Anijphaena iiidicata O. P. -Cambridge, 1896,
Biologia Central! Americana, Aran., 1: 203, pi
26, fig. 4 { S ). Male holotype from Omiltemi,
Guerrero, Mexico, in BMNH, examined. F. O.
P.-Cambridge, 1900, Biologia Centrali Ameri-
cana, Aran., 2: 96, pi. 7, fig. 9, $. Roewer,

1954, Katalog der Araneae, 2: 525. Bonnet,

1955, Bibliographia Araneormn, 2: 345.

Diagnosis. Anyphaena judicata is most
closely related to an unnamed Mexican
species (or group of species) and has no
close relatives among the species occur-
ring north of Mexico. Males may be easily
recognized by the distinctive form of the
dorsal prong of the RTA (Fig. 26). The
female epigynum is closest to that of A.
niarginalis, but the midpiece is proportion-
ately shorter and wider and the sidepieces
are narrower and diminish in width an-
teriorly smoothly, without the sharp de-
crease in width shown by A. marginalis
(Fig. 33).

Male (Cochise Co., Arizona). Coloration
as in Anyphaena celer, except that pos-
terior spinnerets have entire dorsal surface
dark brown.

Total length 3.46 mm. Carapace 1.76
mm long, 1.44 mm wide, cephalic width
0.68 mm, clypeus height 0.09 mm. Eyes:
diameters (mm): AME 0.06, ALE 0.10,
PME 0.09, PLE 0.10; anterior eye row 0.40
mm long, recurved; posterior eye row 0.52
mm long, procurved; MOQ length 0.26
mm, front width 0.17 mm, back width

Spider Family Anyphaenidae • Plotnick 223

0.28 mm; eye interdistanees (mm): AME-
AME 0.05, AME-ALE 0.03, PME-PME
0.11, PME-PLE 0.06, ALE-PLE 0.04.

Sternum 0.95 mm long, ().6S mm wide.
Chelicerae 0.56 mm long with 4 promar-
ginal teeth and 7 retromarginal denticles.

Abdomen 1.80 mm long, 1.15 mm wide.
Epigastric furrow 0.61 mm from tracheal
spiracle, spiracle 0.63 mm from base of
.spinnerets.

Tibial lengths (mm) and indices: I 2.25,
6; II 1.93, 8; III 1.01, 21; IV 1.66, 11. Ven-
tral spination: tibiae I 4-2-2*, II 3-2-2',
III 1-2-0, IV 1-1-2; metatarsi I, II, 2-2-0,
III 2-0-2; IV 1-2-2. Femur III unmodi-
fied. Tibia III ventral spine 1 on retrolat-
eral side missing, ventral spine 2 thickened,
cone-like. Coxae I, II and III ( but not IV )
with a small number of short, thick setae.
Coxae III with a tiibercule.

Palpus as in Figures 5, 26.

Female (Cochise Co., Arizona). Colora-
tion as in male.

Total length 4.72 mm. Carapace 1.76
mm long, 1.37 mm wide, cephalic width
0.81 mm, clypeus height 0.06 mm. Eyes:
diameters (mm): AME 0.07, ALE 0.10,
PME 0.10, PLE 0.10; anterior eye row 0.44
mm long, recurved; posterior eye row 0.60
mm long, procurved; MOQ length 0.29
mm, front width 0.21 mm, back width 0.32
mm; eye interdistanees (mm): AME-
AME 0.07, AME-ALE 0.03, PME-PME
0.13, PME-PLE 0.09, ALE-PLE 0.05.

Sternum 0.97 mm long, 0.77 mm wide.
Chelicerae 0.58 mm long with teeth as in
male.

Abdomen 3.13 mm long, 2.09 mm wide.
Epigastric furrow 1.21 mm from tracheal
spiracle, spiracle 1.31 mm from base of
spinnerets.

Legs unmodified. Tibial lengths (mm)
and indices: I 1.69, 11; II 1.31, 14; III 0.88,
23; IV 1.66, 12. Ventral spination as in
male.

Epigynum as in Figure 33, internal geni-
talia as in Figure 34.

Natural history. Mature males have been
taken from mid-June through mid-August,

mature females from late March to Novem-
ber, most ill July and August. Specimens
have been taken from 5100 to 8000 feet
(1550-2450 m), by sweeping and under
rocks.

Distribution. Arizona south to Guerrero,
Mexico (Map 1).

Anyptiaena marginalis (Banks),
new combination
Map 1; Figures 6, 27, 36, 43

Gayeima marginalis Banks, 1901, Proc. Acad.
Natur. Sci. Philadelphia, 53: 574, pi. 2.3, fig.
22 ( 9 ). Female holotype from Beulali, San
Miguel Co., New Mexico, was probabl>' de-
posited in the MCZ along with the other types
from this paper but was not found by Bryant
when the MCZ t>pes were cataloged; lost,
presumed destroyed. Roewer, 1954, Katalog
der Araneae, 2: 540. Bonnet, 19.57, Biblio-
graphia Araneorum, 2: 1978.

Diagnosis. Amjphaena marginalis is most
closely related to A. hespar, both species
having a simple embolus and elongated
conductor. Males of A. marginalis (Fig.
27), however, do not have the spine-like
dorsal prong of the RTA of A. hespar, and
females of A. marginalis (Fig. 36) do not
have the conspicuous bulge in the epigynal
midpiece which characterizes A. hespar
females.

Male (Graham Co., Arizona). Colora-
tion as in Anyphaena celer.

Total length 3.78 mm. Carapace 1.98
mm long, 1.60 mm wide, cephalic width
0.72 mm, clypeus height 0.09 mm. Eyes:
diameters (mm): AME 0.06, ALE 0.10,
PME 0.08, PLE 0.10; anterior eye row 0.40
mm long, straight; posterior eye row 0.55
mm long, prociuved; xVlOQ length 0.20
mm, front width 0.17 mm, back width 0.28
mm; eye interdistanees (mm): AME-
AME 0.05, AME-ALE 0.02, PME-PME
0.11, PME-PLE 0.08, ALE-PLE 0.04.

Sternum 1.13 mm long, 0.81 mm wide.
Chelicerae 0.54 mm long with 5 promar-
ginal teeth and 6 retromarginal denticles.

Abdomen 2.00 mm long, 1.33 mm wide.
Epigastric furrow 0.52 mm from traclieal

224

Bulletin Museum of Comparative Zoology, Vol. 146, No. 4

spiracle, spiracle 0.59 mm from base of
spinnerets.

Tibial lengths (mm) and indices: I 1.67,
13; II 1.35, 16; III 1.03, 26; IV 1.62, 14.
Ventral spination: tibiae I 4-2-2*, II 2-2-
2*, III, IV 1-2-2; metatarsi I, II 2-2-0, III
2-0-2, IV 2-2-2. Femur III unmodified.
Tibia III ventral spine 1 on retrolateral
side missing. Coxae unmodified.

Palpus as in Figures 6, 27.

Female (Graham Co., Arizona). Colora-
tion as in male of A. celer.

Total length 4.26 mm. Carapace 2.11
mm long, 1.55 mm wide, cephalic width
0.86 mm, clypeus height 0.08 mm. Eyes:
diameters (mm): AME 0.07, ALE 0.10,
PME 0.11, PLE 0.10; anterior eye row 0.44
mm long, straight; posterior eye row 0.64
mm long, procurved; MOQ length 0.30
mm, front width 0.19 mm, back width 0.33
mm; eye interdistances (mm): AME-
AME 0.05, AME-ALE 0.03, PME-PME
0.12, PME-PLE 0.09, ALE-PLE 0.06.

Sternum 1.05 mm long, 0.80 mm wide.
Chelicerae 0.65 mm long with 4 promar-
ginal teeth and 8 retromarginal denticles.

Abdomen 2.52 mm long, 1.53 mm wide.
Epigastric furrow 0.67 mm from tracheal
spiracle, spiracle 0.68 mm from base of
spinnerets.

Legs unmodified. Tibial lengths (mm)
and indices: I 1.44, 18; II 1.21, 21; III 0.99,
25; IV 1.60, 17. Ventral spination as in
male except tibia III 1-1-2.

Epigynum as in Figure 36, internal geni-
talia as in Figure 43.

'Natural history. Mature males have been
taken from late August through late May,
mature females in all months except Janu-
ary and October. Specimens have been
taken from 6000 to 9300 feet (1850-2850
m), in yellow pine/ oak forests and under
rocks. I found this species in great abun-
dance by sorting pine litter at Rustler's
Park in the Chiricahua Mountains of south-
eastern Arizona in August 1972.

Distribution. Arizona, New Mexico and
Colorado (Map 1).

Anyphaena hespar new species
Map 1; Figures 16, 32, 42, 49

Types. Male holotype, female paratype
from Bear Canyon, Santa Catalina Moun-
tains, Pima Co., Arizona, 8 December 1968
(Karl Stephan), deposited in AMNH. Male
and female paratypes from Pima Co., Ari-
zona, deposited in MCZ. The specific
name is an arbitrary combination of letters.

Diagnosis. Anyphaena hespar is most
closely related to A. marginalis. Males of
the former may be distinguished by the
spine-like dorsal prong of their RTA (Fig.
32), females by the conspicuous bulge in
their epigynal midpiece (Fig. 42).

Male (Pima Co., Arizona). Coloration
as in Anyphaena celer.

Total length 3.13 mm. Carapace 1.62
mm long, 1.31 mm wide, cephalic width
0.59 mm, clypeus height 0.08 mm. Eyes:
diameters (mm): AME 0.05, ALE 0.08,
PxME 0.08, PLE 0.08; anterior eye row
0.33 mm long, straight; posterior eye row
0.45 mm long, procurved; MOQ length 0.19
mm, front width 0.14 mm, back width 0.24
mm; eye interdistances (mm): AME-
AME 0.05, AME-ALE 0.03, PME-PME
0.08, PME-PLE 0.07, ALE-PLE 0.04.

Sternum 0.95 mm long, 0.79 mm wide.
Chelicerae 0.39 mm long with 4 promar-
ginal teeth and 8 retromarginal denticles.

Abdomen 1.80 mm long, 1.10 mm wide.
Epigastric furrow 0.56 mm from tracheal
spiracle, spiracle 0.56 mm from base of
spinnerets.

Tibial lengths (mm) and indices: I 1.31,
17; II 1.08, 21; III 0.81, 28; IV 1.39, 18.
Ventral spination: tibiae I 4-2-2*, II 3-
2-2*, III 1-2-2, IV 2-2-2; metatarsi I, II
2-2-0, III 2-0-2, IV 2-2-2. Femur III un-
modified. Tibia III ventral spine 1 on retro-
lateral side missing, spine 2 thickened,
cone-like. Coxae unmodified.

Palpus as in Figures 16, 32.

Female (Pima Co., Arizona). Colora-
tion as in male of A. celer.

Total length 3.06 mm. Carapace 1.55
mm long, 1.26 mm wide, cephalic width
0.67 mm, clypeus height 0.06 mm. Eyes:

Spider Family Anvphaenidak • Plafuick 225

diameters (mm): AME 0.05, ALE O.OS,
PME 0.08, PLE O.OS; anterior eye row 0.33
mm long, .straight; posterior eye row 0.49
mm long, procurved; MOQ length 0.20
mm, front width 0.14 mm, back width 0.26
mm; eye interdistances (mm): AME-
AME 0.05, AME-ALE 0.03, PME-PME
0.10, PME-PLE 0.07, ALE-PLE 0.04.

Sternum 1.04 mm long, 0.70 mm wide.
Chelicerae 0.47 mm long with teeth as in
male.

Abdomen 1.85 mm long, 1.08 mm wide.
Epigastric furrow 0.49 mm from tracheal
spiracle; spiracle 0.41 mm from base of
spinnerets.

Legs unmodified. Tibial lengths (mm)
and indices: I 1.13, 20; II 0.92, 25; III 0.72,
33; IV 1.26, 18. Ventral spination: tibiae
I, II 4-2-2^ III 1-1-2, IV 1-2-2; meta-
tarsi as in male.

Epigynum as in Figure 42, Internal geni-
talia as in Figure 49.

Natural history. Mature males and fe-
males have been taken from late October
through early April. Specimens have been
taken from leaf litter and under rocks.

DistriJmtion. Southeastern Arizona (Map

1)-

Anyphaena rita new species
Map 1; Figures 7, 28, 37, 44

Types. Male holotype, female paratype
from Bear Canyon, Santa Catalina Moun-
tains, Pima Co., Arizona, 8 December 1968
(Karl Stephan), deposited in AMNH. Male
and female paratypes from Pima Co., Ari-
zona, deposited in MCZ. The .specific
name is a noun in apposition derived from
the Santa Rita Mountains, where the
species is abundant.

Diagnosis. Anypliaemi rita is most closely
related to A. cochise, both species having
a conspicuously enlarged region of the
embolus and a slightly recurved tip of the
median apophysis. Males of A. rita (Fig.
28) may be distingui.shed by their smaller
size and by the differences in the dorsal
prong of the RTA. Females of A. cochise
are unknown, l)ut the epigynum of A. rita.

with its extremely broad midpiece, is quite
distinctive (Fig. 37).

Male (Pima Co., Arizona). Colorati(;n
as in Anyphaena celer.

Total length 4.10 mm. Carapace 1.94
mm long, 1.60 mm wide, cephalic width
0.67 mm, clypeus height 0.07 mm. Eyes:
diameters (mm): AME 0.05, ALE 0.08,
PME 0.09, PLE 0.09; anterior eye row 0.36
mm long, recurved; posterior eye row 0.53
mm long, procurved; MOQ length 0.22
mm, front width 0.15 mm, back width 0.27
mm; eye interdistances (mm): AME-
AME 0.05, AME-ALE 0.03, PME-PME
0.09, PME-PLE 0.09, ALE-PLE 0.05.

Sternum 1.13 mm long, 0.77 mm wide.
Chelicerae 0.50 mm long with 4 promar-
ginal teeth and 8 retromarginal denticles.

Abdomen 2.30 mm long, 1.26 mm wide.
Epigastric furrow 0.67 mm from tracheal
spiracle, spiracle 0.65 mm from base of
spinnerets.

Tibial lengths (mm) and indices: I 1.55,
14; II 1.39, 16; III 0.97, 23; IV 1.62, 14.
Ventral spination: tibiae I 4-2-2*, II
3-2-2*, III, IV 2-2-2; metatarsi I, II
2-2-0, III 2-0-2, IV 2-2-2. Femur III un-
modified. Tibia III ventral spines not
thickened. Coxae III and IV with only a
few short thick setae.

Palpus as in Figures 7, 28.

Female (Pima Co., Arizona). Coloration
as in male of A. celer.

Total length 5.04 mm. Carapace 2.05
mm long, 1.53 mm wide, cephalic width
1.03 mm, clypeus height 0.09 mm. Eyes:
diameters (mm): AME 0.05, ALE 6.10,
PME 0.10, PLE 0.11; anterior eye row 0.41
mm long, recurved; posterior eye row 0.58
mm long, procurved; MOQ length 0.32
mm, front width 0.18 mm, back width
0.29 mm; eye interdistances (mm): AME-
AME 0.07,' AME-ALE 0.03, PME-PME
0.09, PME-PLE 0.10, ALE-PLE 0.08.

Sternum 1.13 mm long, 0.81 mm wide.
Chelicerae 0.67 mm long with teeth as in
male.

Abdomen 2.75 mm long, 1.94 mm wide.
Epigastric furrow 1.06 mm from tracheal

226 Bulletin Museum of Comparative Zoologij, Vol. 146, No. 4

spiracle, spiracle 0.95 mm from base of
spinnerets.

Legs unmodified. Tibial lengths (mm)
and indices: I 1.48, 15; II 1.26, 18; III
1.03, 21; IV 1.58, 17. Ventral spination as
in male except tibiae III, IV 1-2-2, meta-
tarsi IV 2-1-2.

Epigynum as in Figure 37, internal
genitalia as in Figure 44.

Natural history. Mature males have been
taken from mid-October through late
March, mature females from early June
tlirough early February. Specimens have
been taken from 4000 to 6800 feet. ( 1200-
2075 m), in oak/ grassland and under rocks.

Distribution. Arizona to Chihuahua,
Mexico (Map 1).

Anyphaena cochise new species
Map 1; Figures 13, 35

Types. Male holotype from Rustlers
Park, 8600 ft. (2625 m), Chiricahua Moun-
tains, Cochise Co., Arizona, 9 September
1950 (W. J. Gertsch), deposited in AMNH.
Male paratype from Cochise Co., Arizona,
deposited in MCZ. The specific name is a
noun in apposition and refers to the type
locality.

Diapiosis. Anypliaena cochise is most
closely related to A. vita, but the dorsal
prong of the RTA is relatively longer in
A. cochise (Fig. 35). Females of this
species are unknown.

Male (Cochise Co., Arizona). Colora-
tion as in Anyphaena celer.

Total length 5.44 mm. Carapace 2.52
mm long, 2.09 mm wide, cephalic width
0.88 mm, clypeus height 0.14 mm. Eyes:
diameters (mm): AME 0.09, ALE 6.13,
PME 0.13, PLE 0.13; anterior eye row
0.53 mm long, straight; posterior eye row
0.75 mm long, procurved; MOQ length
0.30 mm, front width 0.23 mm, back width
0.40 mm; eye interdistances (mm): AME-
AME 0.05, AME-ALE 0.04, PME-PME
0.14, PME-PLE 0.11, ALE-PLE 0.06.

Sternum 1.44 mm long, 1.08 mm wide.

Chelicerae 0.75 mm long with 4 promar-
ginal teeth and 7 retromarginal denticles.

Abdomen 3.38 mm long, 1.94 mm wide.
Epigastric fvuTow 0.92 mm from tracheal
spiracle, spiracle 1.03 mm from base of
spinnerets.

Tibial lengths (mm) and indices: I 2.32,
12; II 2.05, 13; III 1.39, 20; IV 2.14, 14.
Ventral spination: tibiae I 4-2-2*, II 2-2-
2*, III 1-2-2, IV 2-2-2; metatarsi I, II 2-
2-0, III 2-0-2, IV 2-2-2. Femur III un-
modified. Tibia III ventral spine 1 on ret-
rolateral side thickened slightly. All coxae
with a few scattered short thick setae.

Palpus as in Figures 13, 35.

Female. Unknown.

Natural history. Mature males have been
taken in early September at 8600 feet
(2625 m).

Distribution. Known only from the type
locality (Map 1).

Anyphaena autumna new species
IVIap 1; Figures 8, 29, 39, 45

Types. Male holotype, female paratype
from Rustler Camp, Chiricahua Mountains,
Cochise Co., Arizona, 9 September 1950
(W. J. Gertsch), deposited in AMNH.
Male and female paratypes from Cochise
and Graham Co., Arizona, deposited in
MCZ. The specific name refers to the
season of collection.

Diagno.sis. Anyphaena autumna is un-
likely to be confused with any other spe-
cies. The long recurved hook on the RTA
and the peculiar form of the tip of the
median apophysis are mil ike any other
species (Figs. 8, 29). The epigynum is
closest to that of A. gibboides, but the mid-
piece has a characteristic constriction near
its midpoint ( Fig. 39 ) .

Male (Cochise Co., Arizona). Colora-
tion as in Anyphaena celer, though the
paramedian bands on the carapace are
darker and wider than in that species.

Total length 5.51 mm. Carapace 2.50
mm long, 1.98 mm wide, cephalic width
1.03 mm, clypeus height 0.12 mm. Eyes:

Spider Family Anyphaenidae • Platnick 227

diameters (nini): AME 0.09, ALE 0.12,
PME 0.12, PLE 0.13; anterior eye row
0.55 mm long, recnrved; posterior eye row
0.75 mm long, proeun'ed; MOQ length
0.30 mm, front width 0.26 mm, back width
0.38 mm; eye interdistances (mm): AME-
AME 0.08, AME-ALE 0.05, PME-PME
0.15, PME-PLE 0.11, ALE-PLE 0.06.

Sternnm 1.46 mm long, 1.08 mm wide.
Chelicerae 0.79 mm long with 4 promar-
ginal teeth and 9 retromarginal denticles.

Abdomen 3.20 mm long, 2.16 mm wide.
Epigastric furrow 1.04 mm from tracheal
spiracle, spiracle 1.06 mm from base of
spinnerets.

Tibial lengths (mm) and indices: I 2.16,
13; II 1.93, 15; III 1.39, 22; IV 2.16, 14.
Ventral spination: tibiae I 2-2-2, II, III,
IV 1-2-2; metatarsi I, II 2-2-0, III 2-0-2,
I\^ 2-2-2. Third legs unmodified.

Palpus as in Figures 8, 29.

Female (Cochise Co., Arizona). Colora-
tion as in male.

Total length 6.41 mm. Carapace 2.34
mm long, 1.87 mm wide, cephalic width
1.12 mm, clypeus height 0.12 mm. Eyes:
diameters (mm): AME 0.10, ALE 0.13,
PME 0.13, PLE 0.13; anterior eye row
0.59 mm long, recurved; posterior eye row
0.70 mm long, procurved; MOQ length
0.33 mm, front width 0.27 mm, back width
0.42 mm; eye interdistances (mm): AME-
AME 0.06,' AME-ALE 0.03, PME-PME
0.17, PME-PLE 0.12, ALE-PLE 0.07.

Sternum 1.42 mm long, 1.08 mm wide.
Chelicerae 0.99 mm long with 4 promar-
ginal teeth and 8 retromarginal denticles.

Abdomen 3.96 mm long, 2.63 mm wide.
Epigastric furrow 1.33 mm from tracheal
spiracle, spiracle 1.33 mm from base of
spinnerets.

Legs unmodified. Tibial lengths (mm)
and indices: I 1.75, 16; II 1.60, 18; III 1.10,
2.5; IV 1.89, 15. Ventral spination: tibiae
I 4-4-2, II 2-4-2, III 1-1-2, IV 1-2-2;
metatarsi as in male.

Epigynum as in Figure 39, internal geni-
talia as in Figure 45.

Natural history. Mature males and fe-

males have been taken in August and Sep-
tember. Specimens have be(Mi taken at
8200 fec>t (2500 m). I collected a few im-
matiue males (which matured in the labo-
ratory) of this .species in pine litter in the
Chiricahua Mountains, Arizona, where ma-
ture A. mar^inalis were extremely abun-
dant.

Dustribution. Southeastern Arizona (Map

1)-

Anyphaena gibboides new species
Map 1; Figures 14, 30, 40, 46

Types. Male holotype, female paratype
from City Creek Canyon, Salt Lake Co.,
Utah, 22 May 1943 (Wilton Ivie), depos-
ited in AMNH. Male and female para-
t)^es from Lake Co., Oregon, deposited in
MCZ. The specific name is an arbitrary
combination of letters.

Diagnosis. Anyphaena gihJ)oi(Ies is a
distinctive species. Males have a sharply
pointed median apophysis and serrate
RTA which will separate them from the
other known species (Figs. 14, 30). The
epigynum is closest to that of A. autumna,
but lacks the constriction of the midpiece
found in that species ( Fig. 40 ) .

Male (Salt Lake Co., Utah). Coloration
as in Anyphaena celer.

Total length 3.31 mm. Carapace 1.60
mm long, 1.28 mm wide, cephalic width
0.54 mm, clypeus height 0.07 mm. Eyes:
diameters (mm): AME 0.05, ALE 0.08,
PME 0.08, PLE 0.08; anterior eye row 0.34
mm long, straight; posterior eye row 0.48
mm long, procurved; MOQ length 0.23
mm, front width 0.15 mm, back width 0.24
mm; eye interdistances (mm): AME-
AME 0.05, AME-ALE 0.08, PME-PME
0.08, PME-PLE 0.07, ALE-PLE 0.05.

Sternum 0.85 mm long, 0.72 mm wide.
Chelicerae 0.49 mm long with 4 promar-
ginal teeth and 8 retromarginal denticles.

Abdomen 1.94 mm long, 1.24 mm wide.
Epigastric furrow 0.58 mm from tracheal
spiracle, spiracle 0.59 mm from base of
spinnerets.

228 BuUetm Museum of Comparative Zoology, Vol. 146, No. 4

Tibial lengths (mm) and indices: I 1.33,
17; II 1.24, 19; III 0.99, 23; IV 1.47, 16.
Ventral spination: tibiae I 2-2-0, II 1-2-0,
III 2-2-0, IV 2-2-2; metatarsi I, II 2-2-0,
III 2-1-2, IV 2-2-2. Modifications of third
leg as in A. celer save that all coxae have
clumps of short thick setae.

Palpus as in Figures 14, 30.

Female (Salt Lake Co., Utah). Colora-
tion as in male of A. celer.

Total length 3.74 mm. Carapace 1.75
mm long, 1.35 mm wide, cephalic width
0.83 mm, clypeus height 0.09 mm. Eyes:
diameters (mm): AME 0.06, ALE 0.09,
PME O.OS, PLE 0.08; anterior eye row
0.41 mm long, recurved; posterior eye row
0.57 mm long, procurved; MOQ length
0.24 mm, front width 0.18 mm, back width
0.28 mm; eye interdistances (mm): AME-
AME 0.06, AME-ALE 0.03, PME-PME
0.12, PME-PLE 0.09, ALE-PLE 0.06.

Sternum 1.19 mm long, 0.83 mm wide.
Chelicerae 0.62 mm long with 4 promar-
ginal teeth and 6 retromarginal denticles.

Abdomen 2.36 mm long, 1..39 mm wide.
Epigastric furrow 0.72 mm from tracheal
spiracle, spiracle 0.70 mm from base of
spinnerets.

Legs unmodified. Tibial lengths (mm)
and indices: I 1.39, 18; II 1.24, 20; III
0.75, 27; IV 1.39, 18. Ventral spination:
tibiae I 2-2-0, II, III 1-2-0, IV 1-2-2;
metatarsi I, II 2-2-0, III 2-0-2, IV 2-2-2.

Epigynum as in Figure 40, internal geni-
talia as in Figure 46.

Natural history. Mature males and fe-
males have been taken in late May and
June. Habitat data is lacking.

Distribution. Northern Utah west to
southeastern Oregon ( Map 1 ) .

Anyphaena catalina new species
Map 1; Figures 15, 31, 41, 48

Types. Male holotype, female paratype
from Mt. Lemon, Santa Catalina Moun-
tains, Pima Co., Arizona, 13 July 1916 ( F.
E. Lutz), deposited in AMNH. Male and
female paratypes from Pima Co., Arizona,

and Mexico, Mexico, deposited in MCZ.
The specific name is a noun in apposition
and refers to the type locality.

Diagnosis. Anyphaena catalina is most
closely related to A. arbida, though males
of A. catalina may be readily distinguished
by their recurved retrolateral tegular
apophyses (Figs. 15, 31). Females of A.
arbida are unknown; those of A. catalina
may be recognized by the epigynal hood
being roughly equal in size to the epigynal
midpiece ( Fig. 41 ) .

Male (Pima Co., Arizona). Coloration
as in Anyphena celer.

Total" length 3.53 mm. Carapace 1.78
mm long, 1.42 mm wide, cephalic width
0.72 mm, clypeus height 0.09 mm. Eyes:
diameters (mm): AME 0.05, ALE 0.09,
PME 0.08, PLE 0.09; anterior eye row
0.40 mm long, recurved; posterior eye row
0.51 mm long, procurved; MOQ length
0.21 mm, front width 0.17 mm, back width
0.26 mm; eye interdistances (mm): AME-
AME 0.07, AME-ALE 0.04, PME-PME
0.09, PME-PLE 0.08, ALE-PLE 0.04.

Sternum 0.90 mm long, 0.70 mm wide.
Chelicerae 0.56 mm long with 4 promar-
ginal teeth and 6 retromarginal denticles.

Abdomen 1.85 mm long, 0.90 mm wide.
Epigastric furrow 0.61 mm from tracheal
spiracle, spiracle 0.65 mm from base of
spinnerets.

Tibial lengths (mm) and indices: I 2.07,
8; II 1.94, 9; III 1.08, 23; IV 1.80, 10. Ven-
tral spination: tibiae I 4-2-2*, II 2-2-2*,
III, IV 2-2-2; metatarsi I, II 2-2-0, III 2-
0-2, IV 2-2-2. Modifications of third leg
as in A. celer save that femur III lacks short
thick setae and all coxae bear clumps of
them.

Palpus as in Figures 15, 31.

Female (Pima Co., Arizona). Coloration
as in male of A. celer.

Total length 4.57 mm. Carapace 1.84
mm long, 1.42 mm wide, cephalic width
0.94 mm, clypeus height 0.09 mm. Eyes:
diameters (mm): AME 0.07, ALE 0.09,
PME 0.09, PLE 0.09; anterior eye row
0.47 mm long, recui-ved; posterior eye row

Spideu pAisriLY Anyphaemdak • Pintnick 229

0.63 mm long, prociirved; MOQ length 0.26
mm, front width 0.22 mm, back widtli 0.33
mm; eve interdi.stances (mm): AME-
AME 6.0S, AME-ALE 0.04, PME-PME
0.15, PME-PLE 0.11, ALE-PLE 0.07.

Sternum 1.01 mm long, 0.85 mm wide.
Chelicerae 0.68 mm long with 4 promar-
ginal teeth and 8 retromarginal denticle.s.

Abdomen 2.74 nnn long, 1.85 mm wide.
Epigastric furrow 0.86 mm from tracheal
.spiracle, spiracle 0.94 mm from base of
spinnerets.

Legs unmodified. Tibial lengths (mm)
and indices: I 1.51, 14; II 1.33, 17; III 0.94,
23; IV 1.48, 16. Ventral spination: tibiae I,
II 2-2-2*, III 1-2-2, IV 2-2-2; metatarsi
as in male.

Epigynum as in Figure 41, internal geni-
talia as in Figure 48.

Natural Justory. Mature males and fe-
males have been taken in July and August.
Specimens have been taken at 7500 feet
(2300 m) in yellow pine/ oak and douglas
fir/ white fir forests.

Distribution. Southeastern Arizona south
to central Mexico (Map 1).

Anyphaena arbida new species
Map 1; Figures 17, 38

Types. Male holotype from Carr Can-
yon, Huachuca Mountains, Cochise Co.,
Arizona, 26 August 1950 (M. A. Cazier),
deposited in AMNH. Male paratype from
Cochise Co., Arizona, deposited in MCZ.
The specific name is an arbitrary combina-
tion of letters.

Diagnosis. AnypJiaena arhida is most
closely related to A. catalina. Males of the
former (Figs. 17, 38) lack the recurved
retrolateral tegular apophysis of A. cata-
lina; females of A. arbida are unknown.

Male (Cochise Co., Arizona). Colora-
tion as in AnypJiaena celer, except that
posterior spinnerets are as in A. dixiana.

Total length 6.95 mm. Carapace 3.28
mm long, 2.41 mm wide, cephalic width
1.22 mm, clypeus height 0.14 mm. Eyes:
diameter (mm): AME 0.11, ALE 0.13,

PME 0.13, PLE 0.15; anterior eye row
0.64 mm long, recurved; posterior eye row
0.89 mm long, procurved; MOQ lengtii
0.43 mm, front width 0.31 mm, back width
0.44 mm; eye interdistances (mm): AME-
AME 0.09, AME-ALE 0.04, PME-PME
0.18, PME-PLE 0.14, ALE-PLE 0.09.

Sternum 1.62 mm long, 1.33 mm wide.
Chelicerae 1.30 mm long with 4 promar-
ginal teeth and 8 retromarginal denticles.

Abdomen 3.71 mm long, 2.16 mm wide.
Epigastric furrow 1.08 mm from tracheal
spiracle, .spiracle 1.12 mm from base of
spinnerets. Spinnerets surrounded by a
clump of unusually long setae.

Tibial lengths (mm) and indices: I 6.88,
5; II 3.35, 10; III 2.20, 16; IV 3.35, 10.
Ventral spination: tibiae I 4-2-2*, II 3-
2-2*, III, IV 2-2-0; metatarsi I, II 2-2-0,
III, IV 2-2-2. Third legs unmodified.

Palpus as in Figures 17, 38.

Female. Unknown.

Natural history. Mature males have been
collected in August. Habitat data is lack-

ing-

1).

Di.stribution. Cochise Co., Arizona (Map

Anyphaena pectorosa Group

Diagnosis. The pectorosa group is closely
related to the pacifica group, but males
may be distinguished by the spins on their
coxae (Figs. 59-62). Females have the
epigynum on a characteristic sclerotized
plate (Figs. 74, 77, 79) and simple .sper-
mathecae (Figs. 75, 78, 80).

Description. Total length 4.5-6.5 mm.
Carapace longer than wide, narrowed in
front to less than half its maximum width
in males, to slightly more than half its
maximum width in females. Clypeus height
more than 1.5 times the diameter of an an-
terior median eye. Posterior median, pos-
terior lateral and anterior lateral eyes sub-
equal in size, almost twice the diameter of
anterior medians. Procurved posterior eye
row longer than slightly recur\xKl anterior
row. Median ocular (juadrangle almost

230 Bulletin Museum of Comparative Zoolofi.ij, Vol. 146, No. 4

twice as wide in back as in front. Anterior
median eyes separated l:)y sliglitly less than
their diameter, sHghtly closer to anterior
laterals than to each other. Posterior me-
dians separated by slightly more than their
diameter, slightly closer to posterior lat-
erals. Anterior laterals separated by their
radins from posterior laterals. Sternum
longer than wide, with a low hirsute knob
behind its middle in some males. Chelic-
erae with 4 promarginal teeth and 7-9
retromarginal denticles. Abdomen longer
than wide, tiacheal spiracle midway be-
tween epigastric furrow and base of spin-
nerets. Leg formula 1423. Metatarsi I
and 11 with two pairs of ventral spines.
Males with coxae II bearing round knobs,
coxae III and IV bearing spurs. Palpus
with an elongated median apophysis, en-
larged conductor and inconspicuous embo-
lus. Retrolateral tibial apophysis short.
Epigynum on a sclerotized plate, without
a hood. Two simple spermathecae.

Variation. The species in this group
show little intraspecific variation, individ-
ual or geographical, in size, structure or
coloration.

Key to Species

la. Coxae III of males with posterior spur bifid
(Fiffs. 59, 61, 62); sternum of males with a
low hirsute knob behind middle; sclerotized
epigynal plate wider posteriorly than an-
teriorly (Figs. 74, 79) 2

lb. Coxae III of males with posterior spur not
bifid ( Fig. 60 ) ; sternum of males without a
low hirsute knob behind middle; sclerotized
epigynal plate wider anteriorly than pos-
teriorly (Fig. 77) fratema

2a. Distal tip of palpal median apophysis bent
sharply towards cymbium (Figs. 55, 58);
sclerotized epigynal plate with pronounced
posterolateral corners ( Fig. 74 ) 3

2b. Distal tip of palpal median apophysis not
bent sharply towards cymbium ( Fig.
57 ) ; sclerotized epigynal plate without pro-

nounced posterolateral corners (Fig. 79)
alaclma

3a. Distal tip of palpal median apophysis meet-
ing the recessed, dorsal branch of tlie apoph-
ysis ( Fig. 55 ) ; sclerotized epigynal plate
with pronounced posterolateral comers
( Fig. 74 ) pectorosa

3b. Distal tip of palpal median apophysis not
meeting the recessed, dorsal branch of the
apophysis (Fig. 58); females unknown __7flc?:a

Anyphaena pectorosa L. Koch
Map 2; Figures 51, 55, 59, 74, 75

Anyphaena pectorosa L. Koch, 1866, Arachn. Fam.
Drass., 198, pi. 8, figs. 131, 132 { $). Male
holotype from Baltimore, Maryland, in BMNH,
examined. Bryant, 1931, Psyche, 38: 110, pi. 6,
fig. 5, $ . Chickering, 1939, Pap. Michigan
Acad. Sci., 24: 51, figs. 5-8, $,9. Comstock,
1940, Spider Book, rev. ed., p. 577, fig. 636, 9 .
Kaston, 1948, Bull. Connecticut Geol. Natur.
Hist. Surv., 70: 408, figs. 1453, 1477-1480,
$, 9. Roewer, 1954, Katalog der Araneae, 2:
529. Bonnet, 1955, Bibliographia Araneorum,
2: 346.

Aniiphaena calcarata Emerton, 1890, Trans. Con-
necticut Acad. Sci., 8: 187, pi. 6, figs. 3-3d ( $,
9 ). Male holotype, female allotype from West
Haven, Connecticut, in MCZ, examined. Emer-
ton, 1902, Common Spiders, p. 12, figs. 42, 43,
$, 9.

Gaijenna calcarata, Banks, 1910, Bull. U.S. Nat.
Mus., 72: 13.

Gaijenna pectorosa, Comstock, 1912, Spider Book,
p. 563 (in part), fig. 636, 9 (not fig. 637).

Diagnosis. Anyphaena pectorosa is closest
to A. alachua, but may readily be distin-
guished from it by the highly curved me-
dian apophysis of males (Fig. 55) and the
pronounced posterolateral corners of the
sclerotized epigvnal plate of females (Fig.
74).

Male (Fairfax Co., Virginia). Total
length 5.40 mm. Carapace 2.43 mm long,
1.98 mm wide, cephalic width 0.88 mm,
clypeus height 0.11 mm, yellow with thin
dark border and two dark paramedian
longitudinal bands. Eyes: diameters (mm):

Plate 5
Figures 51--54. Left palpi, ventral view. Figures 55-58. Left palpi, retrolateral view. Figures 59-62.
ventral view. 51, 55, 59. Anyptiaena pectorosa L. Koch. 52, 56, 60. Anyphaena fraterna (Banks).
Anyphaena alachua new species. 54, 58, 62. Anyphaena lacka new species.

Male coxae,
53, 57, 61.

Spider I'^aafily Anyphaenidak • Plaliuck 231

232 Bulletin Museum of Comparative Zoology, Vol. 146, No. 4

AME 0.06, ALE 0.11, PME 0.11, PLE 0.12;
anterior e\e row 0.48 mm long, slightly re-
cur^•ed; posterior eye row 0.65 mm long,
procurved; MOQ length 0.28 mm, front
width 0.20 mm, back width 0.35 mm; eye
interdistances (mm): AME-AME 0.07,
AME-ALE 0.04, PME-PME 0.14, PME-
PLE 0.13, ALE-PLE 0.05.

Sternum 1.35 mm long, 1.01 mm wide,
pale yellow with translucent border, dark-
ened extensions to coxae and a low hirsute
knob behind middle. Chelicerae 0.73 mm
long with 4 promarginal teeth and 7 retro-
marginal denticles, pale yellow with boss
outlined in gray. Labium and endites yel-
low, darkest proximally. Endites slightly
invaginated at middle.

Abdomen 3.15 mm long, 1.67 mm wide,
pale white with transverse rows of dark
markings, venter pale. Epigastric furrow
1.01 mm from tracheal spiracle, spiracle
1.06 mm from base of spinnerets.

Legs pale yellow with distal segments
darkest. Tibial lengths (mm) and indices:
I 3.10, 7; II 2.52, 9; III 1.82, 16; IV 2.56, 10.
Ventral spination: tibiae I 2-2-1, II-IV 2-
2-2; metatarsi I, II 2-2-0, III 2-0-2, IV 2-
2-2. Coxae II, III and IV modified as in
Figure 59.

Palpus as in Figures 51, 55.

Female (Fairfax Co., Virginia). Colora-
tion as in male.

Total length 5.44 mm. Carapace 2.41
mm long, 1.91 mm wide, cephalic width
0.97 mm, clypeus height 0.08 mm. Eyes:
diameters (mm): AME 0.07, ALE 0.12,
PME 0.11, PLE 0.12; anterior eye row
0.52 mm long, recurved; posterior eye row
0.71 mm long, procurved; MOQ length
0.33 mm, front width 0.20 mm, back width
0.37 mm; eye interdistances (mm): AME-
AME 0.05, AME-ALE 0.04, PME-PME
0.15, PME-PLE 0.10, ALE-PLE 0.07.

Sternum 1.31 mm long, 1.06 mm wide,
without hirsute knob. Chelicerae 0.72 mm
long with 4 promarginal teeth and 8 retro-
marginal denticles.

Abdomen 3.10 mm long, 1.76 mm wide.
Epigastric furrow 0.70 mm from tracheal

spiracle, spiracle 1.22 mm from base of
spinnerets.

Legs unmodified. Tibial lengths (mm)
and indices: I 2.41, 11; II 2.05, 13; III 1.44,
19; IV 2.20, 12. Ventral spination: tibiae
I, II 2-2-0, III, IV 1-2-1; metatarsi I, II
2-2-0, III, IV 2-2-2.

Epigynum as in Figure 74, internal geni-
talia as in Figure 75.

Natural history. Mature males have been
taken from mid-April through early Sep-
tember, mature females from mid-April
through mid-August. Specimens have been
taken by sweeping foliage, in Malaise and
pitfall ti^ips, and under rocks. Egg cases
taken with females contained 65-95 eggs.

Distribution. New England west to
Michigan, south to western Florida and
eastern Texas ( Map 2 ) .

Anyphaena atachua new species
Map 2; Figures 53, 57, 61, 79, 80

Types. Male holotype, female paratype
from west of Gainesville, Alachua Co.,
Florida, 18 April 1938 (Willis J. Certsch),
deposited in AMNH. Male and female
paratypes from Alachua Co., Florida, de-
posited in MCZ. The specific name is a
noun in apposition and refers to the type
locality.

Diagnosis. Anyphaena alachua is closest
to A. pectorosa but the median apophysis
is not highly curved (Fig. 57) and the
epigynal plate lacks pronounced postero-
lateral corners (Fig. 79).

Male (Alachua Co., Florida). Colora-
tion as in Anyphaena pectorosa.

Total length 4.90 mm. Carapace 2.41
mm long, 2.01 mm wide, cephalic width
0.79 mm, clypeus height 0.13 mm. Eyes:
diameters (mm): AME 0.07, ALE 0.12,
PME 0.12, PLE 0.13; anterior eye row
0.51 mm long, slightly recurved; posterior
eye row 0.70 mm long, procurved; MOQ
length 0.30 mm, front width 0.22 mm, back
width 0.36 mm; eye interdistances (mm):
AME-AME 0.07, AME-ALE 0.04, PME-
PME 0.12, PME-PLE 0.11, ALE-PLE 0.06.

Spider Family Anvi'haemdae • Plalnick 233

Sternum 1.26 mm long, 1.01 mm wide,
with low hirsute knob Ix^hind middle.
Chelicerae 0.76 mm long with 4 promar-
ginal teeth r.nd 9 retromarginal denticle.s.

Abdomen 2.48 mm long, 1.48 mm wide.
Epigastric furrow 0.76 mm from tracheal
spiracle, spiracle 0.8.3 mm from base of
spinnerc>ts.

Tibial lengtlis (nun) and indices: I 2.77,
10; II 2.27,^11; III 1.44, 22; IV 1.94. 14.
\Vntral spination: tibiae I, II 2-2-0, III
1-2-2, IV 2-2-2; metatarsi I, II 2-2-0, III
2-0-2, IV 2-2-2. Coxae II, III and I\'
modified as in Figure 61.
Palpus as in Figures 53, 57.
Female (Alachua Co., Florida). Colora-
tion as in male of A. pectorosa.

Total length 6.17 mm. Carapace 2.45
mm long, 1.80 mm wide, cephalic width
0.94 mm, clypeus height 0.12 mm. Eyes:
diameters (mm): AME 0.08, ALE 0.13,
PME 0.12, PLE 0.13; anterior eye row
0.57 mm long, slightly recurved; posterior
eye row 0.73 mm long, procui-ved; MOQ
IcMigth 0.30 mm, front width 0.22 mm, back
width 0.40 mm; eye interdistances (mm):
AME-AME 0.07, AME-ALE 0.04, PME-
PME 0.15, PME-PLE 0.11, ALE-PLE
0.07.

Sternum 1.35 mm long, 1.08 mm wide,
without hirsute knob. Clielicerae 0.84 mm
long with teeth as in male.

Abdomen 3.53 mm long, 2.02 mm wide.
Epigastric furrow 1.10 mm from tracheal
spiracle, spiracle 1.21 mm from base of
spinnerets.

Legs unmodified. Tibial lengths (mm)
and indices: I 2.30, 13; II 1.91, 14; III 1.31,
22; IV 2.09, 13. Ventral spination as in
male save metatarsi III 2-2-2.

Epigynum as in Figure 79, internal geni-
talia as in Figure 80.

Natural history. Mature males have been
taken in late April and early May, mature
females from late March through mid-May,
by sweeping.

Distribution. Known only from Florida
(Map 2).

Anyphaena lacka new species
Map 2; Figures 54, 58, 62

Type. Male liolotxpe from Lake Corpus
Christi State Park, southwest of Mathis,
San Patricio Co., Texas, 28 Jvme 1962 (J. A.
Beatty), deposited in MCZ. The specific
name is an arbitrary combination of letters.
Dia<i,nosis. Anyphaena lacka is most
closely reflated to A. alachua ])ut has a dis-
tinct point on the tip of the median apoph-
ysis (Fig. 58). Females of A. lacka are
unknown.

Male (San Patricio Co., Texas). Colora-
tion as in Anyphaena pectorosa.

Total length 4.61 mm. Carapace 2.05
mm long, 1.69 mm wide, cephalic width
0.79 mm, clypeus height 0.12 mm. Eyes:
diameters (mm): AME 0.07, ALE 0.12,
PME 0.11, PLE 0.11; anterior eye row
0.47 mm long, slightly recurved; posterior
eye row 0.61 mm long, procurved; MOQ
length 0.26 mm, front width 0.19 mm, back
width 0.32 mm; eye interdistances (mm):
AME-AME 0.05, AME-ALE 0.03, PME-
PME 0.11, PME-PLE 0.08, ALE-PLE 0.04.
Sternum 1.24 mm long, 0.90 mm wide,
with low hirsute knob behind middle.
Chelicerae 0.64 mm long with 4 promar-
ginal teeth and 9 retromarginal denticles.

Abdomen 2.41 mm long, 1.33 mm wide.
Epigastric furrow 0.74 mm from tracheal
spiracle, spiracle 0.90 mm from base of
spinnerets.

Tibial lengths (mm) and indices: I 2.38,
9; II 1.91, 12; III 1.32, 30; IV 1.93, 12.
Ventral spination: tibiae I, II 2-2-0, III,
IV 1-2-2; metatarsi I, II 2-2-0, III 2-0-2,
IV 2-2-2. Coxae II, III, and W modified
as in Figure 62.

Palpus as in Figures 54, 58.
Female. LTnknown.

Natural history and distribution. Known
only from the type specimen.

Anyphaena fraterna (Banks)

iVlap 2; Figures 52, 56, 60, 77, 78

.\]iil})hacna coiispcrsa KeyserlinK. 1887, Verh.
zool. bot. Cos. Wien, 37: 453, pi. 6, fig. 23

234 Bulletin Museum of Comparative Zoology, Vol. 146, No. 4

( $ ). Female holotype from Bee Spring, Ken-
tucky, in MCZ, examined; preoccupied by Any-
phaena conspersa Simon, 1878.

Gaijenna fratema Banks, 1896, Trans. Amer. Ent.
Soc, 23: 63. Male holotype from Sea Cliff,
New York, in MCZ, examined.

Amjphaena fratema, Simon, 1897, Hist. Natur.
Araign., 2: 96. Bryant, 1931, Psyche, 38: 110,
pi. 6, fig. 6, pi. 8, fig. 23, $, 9. Comstock,
1940, Spider Book, rev. ed., p. 577, fig. 637, $ .
Kaston, 1948, Bull. Connecticut Geol. Natur.
Hist. Surv., 70: 408, figs. 1454-1456, $, 9.
Roewer, 1954, Katalog der Araneae, 2: 529.
Bonnet, 1955, Bibliographia Araneonnn, 2: 344.

Sillus consperstis, Petrunkevitch, 1911, Bull. Amer.
Mus. Natur. Hist., 29: 511.

Gayenna pectorosa, Comstock, 1912, Spider Book,
p. 563 (in part), fig. 637, $.

Diagnosis. Anyphaena fratema is a dis-
tinctive species easily recognized by the
long and narrow median apophysis of
males (Fig. 52) and by the female's epigy-
nal plate being wider anteriorly than pos-
teriorly (Fig. 77).

Male (Hall Co., Georgia). Coloration
as in Anyphaena pectorosa.

Total length 4.93 mm. Carapace 2.23
mm long, 1.85 mm wide, cephalic width
0.81 mm, clypeus height 0.10 mm. Eyes:
diameters (mm): AME 0.06, ALE 0.12,
PME 0.11, PLE 0.11; anterior eye row 0.45
mm long, slightly recurved; posterior eye
row 0.64 mm long, procurved; MOQ length
0.30 mm, front width 0.18 mm, back width
0.33 mm; eye interdistances (mm): AME-
AME 0.05, AME-ALE 0.04, FME-PiME
0.12, PME-PLE 0.08, ALE-PLE 0.05.

Sternum 1.28 mm long, 0.99 mm wide,
without hirsute knob. Chelicerae 0.59 mm
long with 4 promarginal teeth and 9 retro-
marginal denticles.

Abdomen 2.83 mm long, 1.60 mm wide.
Epigastric furrow 0.97 mm from tracheal
spiracle, spiracle 0.85 mm from base of
spinnerets.

Legs with scattered dark spots. Tibial
lengths (mm) and indices: I 2.60, 8; II
2.16, 11; III 1.52, 16; IV 2.47, 10. Ventral
.spination: tibiae I, II 2-2-0, III 1-2-2, IV
2-2-2; metatarsi I, II 2-2-0, III, IV 2-2-2.
Coxae II, III and IV modified as in Figure
60.

Palpus as in Figures 52, 56.

Female (Hall Co., Georgia). Coloration
as in male of A. pectorosa.

Total length 5.00 mm. Carapace 2.32
mm long, 1.80 mm wide, cephalic width
0.94 mm, clypeus height 0.10 mm. Eyes:
diameters (mm): AME 0.06, ALE 0.11,
PME 0.11, PLE 0.11; anterior eye row
0.49 mm long, recurved; posterior eye row
0.69 mm long, procui-ved; MOQ length
0.27 mm, front width 0.18 mm, back width
0.36 mm; eye interdistances (mm): AME-
AME 0.05,' AME-ALE 0.04, PME-PME
0.15, PME-PLE 0.09, ALE-PLE 0.05.

Sternum 1.28 mm long, 1.04 mm wide.
Chelicerae 0.75 mm long with teeth as in
male.

Abdomen 2.97 mm long, 1.71 mm wide.
Epigastric furrow 0.85 mm from tracheal
spiracle, spiracle 0.85 mm from base of
spinnerets.

Legs unmodified. Tibial lengths (mm)
and indices: I 2.29, 11; II 1.89, 13; III 1.30,
19; IV 2.16, 13. Ventral spination as in
male.

Epigynum as in Figure 77, internal geni-
talia as in Figure 78.

Natural history. Mature males have been
taken from late March through early July,
mature females from late March through
late August. Specimens have been taken
by sweeping foliage, in Malaise and pitfall
traps, and under logs. I collected this spe-
cies in great abundance by sweeping
honeysuckle (Lonicera sp.) at night in
southern West Virginia during June 1971.

Distribution. Southern New York west
to eastern Kansas, south to western Florida
and eastern Texas (Map 2).

Anyphaena pacifica Group

Diagnosis. The pacifica group is closely
related to the pectorosa group and appears
to displace it in the western United States.
The males have similarly short retrolateral
tibial apophyses (Figs. 69-71), but paci-
fica group males lack the coxal spurs char-
acteristic of the pectorosa group, though

Spider Family ANYPiiAiixiDAt; • Flatnick 235

Anyphoena pectorosa o\ -* >

------v*-

i£

Anyphoena fraterna .«\ -^>

f* ■'

c "^

ijiy. \

VM • ^

r^"^- *'

)• ;' ''*~

• .-••\

•

(• '~~^-r- ;

• 1

V ;'•

_ 1

~'^

";•:• z---

..^

• • -

r^^

Anyphaenc

pacificQ

/ A

-^w

Tn P

^■-^ .__..__

Anyphoena

colifornicQ -

\\v

1 "~

r )

Anyphoena

locko

s , 1

i ,'---,'

; .' t>'-

^. ^ ■'

■• "> /

\ 1

"\ : 1

Anyphoena aperta '

\ 1 -^ ^ .'-

Map 2. Distributions of Anyphaena alachua, A. aperta, A. californica, A. fraterna, A. gertschi, A. lacka, A. pacif-
ica and A. pectorosa.

males of Anyphaena gertschi have rounded
knobs on the coxae. Females lack the
sclerotized epigynal plates found in the
pectorosa group, but have a lightly sclero-
tized atrivun-like area posteromedially
(Figs. 66, 67, 72) and long, sometimes
coiling, ducts (Figs. 68, 73, 76).

Description. Total length 4-6 mm. Cara-
pace longer than wide, narrowed in front
by at least one-third of its maximum width,
often by more than half. Clypeus height
roughly equal to anterior median eye diam-
eter. All eyes subequal in size. Procurved
posterior eye row longer than slightly re-
curved anterior eye row. Median ocular
quadrangle longer than wide in front,
wider in back than long. Anterior median
eyes separated by less than their diameter,
much closer to anterior laterals than to
each other. Posterior medians separated
by more than their diameter, much closer
to posterior laterals. Anterior laterals sepa-
rated by slightly more than their radius
from posterior laterals. Sternum longer

than wide, without a hirsute knob. Chelic-
erae with 3 promarginal teeth and 6-9
retromarginal denticles. Abdomen longer
than wide, tracheal spiracle midway be-
tween epigastric furrow and base of spin-
nerets. Leg formula 1423. Metatarsi I and
II with two pairs of ventral spines. Males
with legs unmodified. (A. pacifica and A.
californica) or with coxae bearing round
knobs and femora II and III bearing
patches of short stiff setae ventrally (A.
gertschi). Palpus with an elongated me-
dian apophysis, enlarged conductor and
inconspicuous embolus. Retrolateral tibial
apophysis short. Epigyiuim not on a scle-
rotized plate, without a hood, with a more
or less pronoimced atrium-like lightly
sclerotized area posteromedially. Internal
genitalia with long ducts that coil in some
species.

Variation. Two species in this group, A.
pacifica and A. californica, show a great
deal of xariation in genitalic structure. In
both species the shape oi the tip of the pal-

236 Bulletin Museum of Comparative Zoology, Vol. 146, No. 4

pal median apophysis and the coihng of
the epigynal ducts are strikingly variable,
and it was initially thought that many
species were involved. Three sources of
evidence, however, have indicated other-
wise. First, many females are found in
which the ducts on one side of the epigy-
num coil differently from those on the
other side. Secondly, when many speci-
mens are taken together at one locality on
a single da}", several variants are often
found. Finally, the retrolateral tibial
apophysis, which usually provides excellent
diagnostic characters in anyphaenids, is
stable within the species as they are de-
fined here. Until such time as biological
evidence on the breeding habits of these
spiders can be obtained, it seems best to
consider both A. pacifica and A. califorjuco
as widespread, variable species.

Key to Species

la. Retrolateral tibial apophysis (RTA) without
a dorsal process (Fig. 69). Median apoph-
ysis with a deep invagination below tip
giving the tip a chelate appearance (Fig.
65). Epigyninii with large wing-shaped
paramedian flaps ( Fig. 72 ) gertschi

lb. Retrolateral tibial apophysis (RTA) with a
dorsal process (Figs. 70, 71). Median
apophysis without a deep invagination be-
low tip (Figs. 63, 64). Epigynum with-
out large wing-shaped paramedian flaps
(Figs. 66, 67) 2

2a. Dorsal process of RTA short, located dis-
tally (Fig. 70). Median apophysis narrow-
ing gradually towards tip (Fig. 63). In-
ternal ducts with many coils (Fig. 68) — _
— pacifica

2b. Dorsal process of RTA long, located prox-
inially (Fig. 71). Median apophysis nar-
rowing abruptly towards tip (Fig. 64). In-
ternal ducts without many coils (Fig. 73)
— californica

Anyphaena pacifica (Banks)
Map 2; Figures 63, 66, 68, 70

Gaijcnna pacifica Banks, 1896, Trans. Amer. Ent.
Soc, 23: 63. Female holot>'pe from Olympia,
Washington, in MCZ, examined.

Anyphaena pacifica, Simon, 1897, Hist. Natur.
Araign., 2: 96. Bryant, 1931, Psyche, 38: 115,
pi. 8, fig. 36, ?. Levi and Levi, 1951, Zoo-
logica (New York), 36: 228, Tig. 25, $. Roe-
wer, 1954, Katalog der Araneae, 2: 529. Bon-
net, 1955, Bibliographia Araneorum, 2: 346.

Anyphaena mundella Chamberlin, 1920, Pomona
Coll. J. Ent. Zool., 12: 12, pi. 5, fig. 3 ( 9 , not
$, = Aysha incursa) . Female holotype from
Claremont, California, in MCZ, examined.
Bryant, 1931, Psyche, 38: 120 (sub Aysha de-
cepta [sic] ). Roewer, 1954, Katalog der Araneae
2: 534 (sub Aysha decepta [sic]). Bonnet,
1955, Bibliographia Araneorum, 2: 836 (sub
Aysha decepta [sic]). NEW SYNONYMY.

Anyphaena intermontana Chamberlin, 1920,
Canad. Ent., 52: 200, fig. 22-6 ( $ ). Female
holotype from Mill Creek, Salt Lake Co., Utali,
in MCZ, examined. Bryant, 1931, Psyche, 38:
114 (sub Anyphaena californica [sic]). Roe-
wer, 1954, Katalog der Araneae, 2: 528 (sub
Anyphaena californica [sic]). Bonnet, 1955,
Bibliographia Araneorum, 2: 343 (sub Any-
phaena californica [sic]). NEW SYNONYMY.

Gayenna saniuana Chamberlin and Gertsch, 1928,
Proc. Biol. Soc. Wash., 41: 185. Male holotype
from Verdure, San Juan Co., Utah, in AMNH,
examined. Roewer, 1954, Katalog der Araneae,
2: 540. NEW SYNONYMY.

Anyphaena saniuana, Bryant, 1931, Psyche, 38:
107. Bonnet, 1955, Bibliographia Araneorum,
2: 347.

Anyphaena pomona Chamberlin and Ivie, 1941,
Bull. Univ. Utah, Biol., 6: 23, pi. 2, fig. 16
( 9 ). Female holotype from Mill Creek, Te-
hama Co., California, in AMNH, examined.
Roewer, 1954, Katalog der Araneae, 2: 529.
NEW SYNONYMY.

Gayenna jollensis Schenkel, 1950, Verb. Naturf.
Ges. Basel, 61: 77, fig. 27 ( 9 ). Female holo-
type from La Jolla, California, in Naturhistor-
isches Museum, Basel, examined. Roewer, 1954,
Katalog der Araneae, 2: 540. NEW SYN-
ONYMY.

Plate 6
Figures 63-65. Left palpi, ventral view. Figures 69-71. Left palpal tibiae, retrolateral view. Figures 66, 67, 72,
74, 77, 79. Epigyna, ventral view. Figures 68, 73, 75, 76, 78, 80. Internal genitalia, dorsal view. 63, 66, 68, 70.
Anyptiaena pacifica {Banks). 64,67,71,73. Anyptiaena californica {Banks). 65,69,72,76. Anyphaena gertschi
new species. 74, 75. Anyphaena pectorosa L. Koch. 77, 78. Anyphaena fraterna (Banks). 79, 80. Anyphaena
alachua new species.

Spider Family Anyphaenidae • Plalnick 237

•^

£S^'

238 Bulletin Museum of Comparative Zoology, Vol. 146, No. 4

Diagnosis. Amjphaena pacifica is closest
to A. californica, but males may be distin-
guished by the short, distal, dorsal process
of the retrolateral tibial apophysis (Fig.
70) and the gradually narrowing tip of the
median apophysis (Fig. 63), while females
have distinctive highly coiled internal
ducts (Fig. 68). Variation in this species
is discussed above.

Male (El Dorado Co., California). Total
length 5.18 mm. Carapace 2.34 mm long,
1.94 mm wide, cephalic width 0.86 mm,
clypeus height 0.12 mm, pale orange with
thin dark border and two dark paramedian
longitudinal bands. Eyes: diameters (mm):
AME 0.09, ALE 0.12, PME 0.10, PLE 0.11;
anterior eye row 0.51 mm long, slightly
procurved; posterior eye row 0.69 mm long,
procurved; MOQ length 0.28 mm, front
width 0.24 mm, back width 0.34 mm; eye
interdistances (mm): AME-AME 0.07,
AME-ALE 0.03, PME-PME 0.14, PME-
PLE 0.10, ALE-PLE 0.05.

Sternum 1.49 mm long, 1.04 mm wide,
pale orange with darker border. Chelicerae
0.67 mm long with 3 promarginal teeth
and 8 retromarginal denticles, dark orange-
brown proximally, pale orange distally,
with boss outlined in gray. Labium and
endites orange, darkest proximally. En-
dites slightly invaginated at middle.

Abdomen 2.81 mm long, 1.69 mm wide,
reddish-brown throughout. Epigastric fur-
row 0.85 mm from tracheal spiracle, spira-
cle 0.92 mm from base of spinnerets.

Legs pale orange, unmodified. Tibial
lengths (mm) and indices: I 2.11, 12; II
1.87, 13; III 1.44, 20; IV 2.07, 15. Ventral
spination: tibiae I, II 2-2-0, III 1-2-2, IV
2-2-2; metatarsi I, II 2-2-0, III, IV 2-2-2.

Palpus as in Figures 63, 70.

Female (Mono Co., California). Color-
ation as in male.

Total length 5.39 mm. Carapace 2.34
mm long, 1.62 mm wide, cephalic width
0.94 mm, clypeus height 0.09 mm. Eyes:
diameters (mm): AME 0.10, ALE 0.12,
PME 0.11, PLE 0.11; anterior eye row 0.51
mm long, slightly recurved; posterior eye

row 0.73 mm long, procurved; MOQ length
0.29 mm, front width 0.25 mm, back width
0.36 mm; eye interdistances (mm): AME-
AME 0.06, AME-ALE 0.03, PME-PME
0.15, PME-PLE 0.10, ALE-PLE 0.07.

Sternum 1.44 mm long, 1.01 mm wide.
Chelicerae 0.71 mm long with teeth as in
male.

Abdomen 3.02 mm long, 1.69 mm wide.
Epigastric furrow 0.81 mm from tracheal
spiracle, spiracle 0.86 mm from base of
spinerets.

Tibial lengths (mm) and indices: I 1.84,
15; II 1.71, 15; III 1.39, 19; IV 2.07, 13.
Ventral spination as in male save tibiae
III 1-1-2 and IV 1-2-2.

Epigynum as in Figure 66, internal geni-
talia as in Figure 68.

Natural history. Mature males have been
taken from late February through late
July, mature females year round. Speci-
mens have been taken in montane forests,
in pitfall traps, under rocks and commonly
in houses.

Distribution. Western North America
from British Columbia south to California,
Ai-izona and New Mexico (Map 2).

Anyphaena californica (Banks)
Map 2; Figures 64, 67, 71, 73

Gaijenna californica Banks, 1904, Proc. California
Acad. Sci., 3: 338, pi. 38, fig. 2(9). Female
holotype from Palo Alto, California, in MCZ,
examined.

Amiphaena mens Chamberlin, 1920, Pomona Coll.
J. Ent. Zoo!., 12: 11, pi. 5, fig. \ {$). Male
holotype from Claremont, California, in MCZ,
examined. Bryant, 1931, Psyche, 38: 113.
Roewer, 1954, Katalog der Araneae, 2: 529.
Bonnet, 1955, Bibliographia Araneorum, 2: 347.
NEW SYNONYMY.

Anyphaena californica, Bryant, 1931, Psyche, 38:
114. Roewer, 1954, Katalog der Araneae, 2:
528. Bonnet, 1955, Bibliographia Araneorum,
2: 343.

Diagnosis. Amjphaena californica is most
closely related to A. pacifica, but males
have a long, proximal, dorsal process on
the retrolateral tibial apophysis (Fig. 71)
and an abruptly narrowed tip of the me-

Spider Family Anyphaenidae • Phifnick

239

diaii apopliysis (Fig. 64), while the inter-
nal dncts of the female are not highly
coiled (Fig. 73). Variation in this species
is discnssed above.

Male (San Diego Co., California). Col-
oration as in AnypJiaena pacifica except
that the abdomen is pale white with trans-
verse rows of dark markings.

Total length 4.68 mm. Carapace 2.21
mm long, 1.78 mm wide, cephalic width
0.68 mm, clypeus height 0.07 mm. Eyes:
diameters (mm): AME 0.07, ALE 6.09,
PME 0.10, PLE 0.11; anterior eye row
0.43 mm long, recm-ved; posterior eye row
0.59 mm long, procurved; MOQ length
0.30 mm, front width 0.20 mm, back width
0.32 mm; eye interdistances (mm): AME-
AME 0.06, AME-ALE 0.04, PME-PME
0.13, PME-PLE 0.11, ALE-PLE 0.07.

Sternum 1.31 mm long, 0.90 mm wide.
Chelicerae 0.60 mm long with 3 promar-
ginal teeth and 8 retromarginal denticles.

Abdomen 2.97 mm long, 1.34 mm wide.
Epigastric furrow 0.79 mm from tracheal
spiracle, spiracle 0.85 mm from base of
spinnerets.

Tibial lengths (mm) and indices: I 3.28,
5; II 3.20, 7; III 2.27, 8; IV 2.93, 7. Ventral
spination: tibiae I 2-2-0, II 2-2-2, III 1-
1-2, IV 1-2-2; metatarsi I, II 2-2-0, III,
I\' 2-2-2.

Palpus as in Figures 64, 71.

Female (Humboldt Co., CaHfornia).
Coloration as in male.

Total length 5.98 mm. Carapace 2.56
mm long, 1.91 mm wide, cephalic width
1.03 mm, clypeus height 0.08 mm. Eyes:
diameters (mm): AME 0.09, ALE 0.11,
PME 0.12, PLE 0.12; anterior eye row 0.52
mm long, recurved; posterior eye row 0.69
mm long, procurved; MOQ length 0.35
mm, front width 0.25 mm, back width 0.37
mm; eye interdistances (mm): AME-
AME 0.07, AME-ALE 0.04, PME-PME
0.13, PME-PLE 0.10, ALE-PLE 0.06.

Sternum 1.44 mm long, 1.08 mm wide.
Chelicerae 0.86 mm long with 3 promar-
ginal teeth and 9 retromarginal denticles.

Abdomen 3.64 mm long, 2.43 mm wide.

Epigastric furrow 1.15 mm from tracheal
spiracle, spiracle 1.33 mm from base of
spinnerets.

Tibial lengths (mm) and indices: I 2.16,
12; II 1.87, 13; III 1.30, 19; IV 2.06, 14.
Ventral spination as in male except tibiae
II 1-2-0.

Epigynum as in Figure 67, internal geni-
talia as in Figure 73.

Natural Jii.story. Mature males have been
taken from early March through mid-July,
mature females from mid-March through
mid-November. Specimens have Ix^en
taken in redwood forests, on citrus trees
and in houses.

Distribution. Oregon and California
(Map 2).

Anyphaena gertschi new species
Map 2; Figures 65, 69, 72, 76

Types. Male holotype, female paratype
from Bluff, San Juan Co., Utah, 11 May
1933 (Wilton Ivie), deposited in AMNH.
Male and female paratypes from Emery
Co., Utah, deposited in MCZ. The specific
name is a patronym in honor of Willis J.
Gertsch, who first recognized the species
as new.

Diagnosis. Anyphaena gertschi is a dis-
tinctive species easily recognized by the
chelate appearance of the tip of the me-
dian apophysis of males (Fig. 65) and
the large wing-shaped paramedian flaps
on the female epigynum (Fig. 72).

Male (Emery Co., Utah). Coloration as
in Anyphaena pacifica except that cara-
pace has paramedian bands only vaguely
indicated and abdomen is pale yellow
throughout.

Total length 4.00 mm. Carapace 1.85
mm long, 1.42 mm wide, cephalic width
0.92 mm, clypeus height 0.14 mm. Eyes:
diameters (mm): AME 0.09, ALE 6.09,
PME 0.09, PLE 0.09; anterior eye row
0.45 mm long, slightly recurv'ed; pos-
terior eye row 0.59 mm long, procurved;
MOQ length 0.26 mm, front width 0.22
mm, back width 0.32 mm; e\e interdis-

240 Bulletin Museum of Comparative Zoology, Vol. 146, No. 4

tances (mm): AME-AME 0.04, AME-
ALE 0.03, PME-PME 0.14, PME-PLE
0.08, ALE-PLE 0.04.

Sternum 1.12 mm long, 0.85 mm wide.
Chelicerae 0.65 mm long with 3 promar-
ginal teeth and 6 retromarginal denticles.

Abdomen 2.11 mm long, 1.31 mm wide.
Epigasti'ic furrow 0.67 mm from tracheal
spiracle, spiracle 0.76 mm from base of
spinnerets.

All coxae with round knobs ventrally.
Femora II and III with patches of short,
thick setae ventrally. Tibial lengths (mm)
and indices: I 2.00, 9; II 1.69, 13; III 1.30,
17; IV 1.87, 12. Ventral spination: tibiae
I 2-2-0, II 1-2-0, III, IV 1-2-2; metatarsi
I, II 2-2-0, III, IV 2-2-2.

Palpus as in Figures 65, 69.

Female (San Diego Co., Cahfornia).
Coloration as in male.

Total length 5.04 mm. Carapace 2.25 mm
long, 1.76 mm wide, cephalic width 0.95
mm, clypeus height 0.12 mm. Eyes: diam-
eters (mm): AME 0.10, ALE 0.13, PME
0.10, PLE 0.13; anterior eye row 0.51 mm
long, straight; posterior eye row 0.68 mm
long, procurved; MOQ length 0.28 mm,
front width 0.26 mm, back width 0.36 mm;
eye interdistances (mm): AME-AME
0.06, AME-ALE 0.03, PME-PME 0.16,
PME-PLE 0.08, ALE-PLE 0.05.

Sternum 1.28 mm long, 0.90 mm wide.
Chelicerae 0.70 mm long with teeth as in
male.

Abdomen 3.10 mm long, 2.02 mm wide.
Epigastric furrow 0.77 mm from tracheal
spiracle, spiracle 1.03 mm from base of
spinnerets.

Legs unmodified. Tibial lengths (mm)
and indices: I 1.62, 14; II 1.49, 15; III 1.17,
21; IV 1.69, 14. Ventral spination as in
male save tibiae III 1-1-0.

Epigynum as in Figure 72, internal geni-
talia as in Figure 76.

Natural history. Mature males have been
taken from late April through late June,
mature females from mid-May through
late September. Nothing is known of the
habits of this species.

Distribution. Southern Utah south to
southern California and Arizona (Map 2).

Anyphaena accentuata Group

Diagnosis. Members of this group can
be immediately differentiated from the
other nearctic Anyphaena by the presence
of only one pair of ventral spines on meta-
tarsi I and II. Only one species occurs in
America north of Mexico.

Description. Total length 4-6 mm. Cara-
pace longer than wide, narrowed in front
to less than half its maximum width in
males, to slightly more than half in females.
Clypeus height roughly equal to anterior
median eye diameter. Median eyes smaller
than laterals. Procurved posterior eye row
longer than recurved anterior row. Me-
dian ocular quadrangle longer than wide
in front, wider in back than long. Anterior
median eyes separated by less than their
diameter, closer to anterior laterals. Pos-
terior medians separated by 1.5 times their
diameter, closer to posterior laterals. An-
terior laterals separated by their radius
from posterior laterals. Sternum longer
than wide, unmodified. Chelicerae with 3
promarginal teeth and 5-7 retrolateral den-
ticles. Abdomen longer than wide, tracheal
spiracle midway between epigastric furrow
and base of spinnerets. Leg formula 1423,
legs unmodified. Metatarsi I and II with
one pair of ventral spines. Palpus with
short median apophysis, short conductor
and conspicuous embolus. Cymbial groove
compressed to retrolateral side of cymbium.
Epigynum with hood. Internal genitalia
with anterior membranous dorsal cover.

Variation. No significant variation was
detected in Anyphaena aperta.

Anyphaena accentuata (Walckenaer)
Figure 134

Aranea accentuata Walckenaer, 1802, Faun. Paris,
2: 226. Type lost, presumed destroyed.

Anyphaena accentuata, Roewer, 1954, Katalog der
Araneae, 2: 522. Bonnet, 1955, Bibliographia
Araneoriini, 2: 338.

Spider Family Axypiiaenidak • Plafnick 241

A drawing of tlic palpus of tliis Enro-
pc^an spider, type species ol tlie genus A/ij/-
pJiaena, is included for pinposes of com-
parison to A. aperta. Confusion exists
between AnypJiaena accentuatii, A. ohscura
(Sundex'all) and A. sabina L. Koch, and
the female is therefore not illustrated and
no description is gi\^en. The male illus-
trated is from England.

Anyphaena aperta (Banks)
Map 2; Figures 135-137

Gaijenna aperta Banks, 1921, Pioc. California
Acad. Sci., 11: 100, fig. ,3 ( 9 ). Female holo-
t\'pe from OKinpia, Washington, in MCZ, ex-
amined.

Ani/pliacna aperta, Bryant, 1931, Psyche, 38: 114,
pi. 8, fig. 35, 9 . Fox, 1938, Iowa State Coll. J.
Sci., 12: 238, pi. 1, fig. 6, $. Roewer, 1954,
Katalog der Araneae 2: 528. Bonnet, 1955,
Bibliographia Araneornm, 2: 342.

Didiinosis. In addition to the diagnostic
character of the species group, Amjphaemi
aperta can readily be distinguished from
all other North American anyphaenids by
the sharply pointed median apophysis of
males (Fig. 135) and the membranous dor-
sal cover of the internal genitalia of females
(Fig. 137). Although the distribution indi-
cates that this might be an inti-oduced
species, no specimens or described species
from the Palearctic or Oriental regions re-
semble Anyphaena aperta.

Male (Yamhill Co., Oregon). Total
length 4.32 mm. Carapace 1.98 mm long,
1.63 mm wide; cephalic width 0.74 mm,
clypeus height O.OS mm, light orange-
brown, darker towards sides, with two
dark paramedian longitudinal bands. Eyes:
diameters (mm): AME 0.07, ALE 0.11,
PME 0.09, PLE 0.11; anterior eye row
0.44 mm long, recurved; posterior eye row
0.62 mm long, procurved; MOQ length
0.26 mm, front width 0.20 mm, back width
0.32 mm; eye interdistances (mm): AME-
AME 0.05, AME-ALE 0.03, PME-PME
0.14, PME-PLE 0.10, ALE-PLE 0.06.

Sternum 1.04 mm long, 0.89 mm wide,
pale orange \\'ith translucent border and

darkened extensions to coxae. Chelicerae
0.55 mm long with 3 promarginal teeth and
5 retromarginal denticles, orange-brown
with boss outlined in gray. Labium and
endites pale orange, darkest proximally.
Endites not invaginated.

Abdomen 2.52 mm long, 1.51 mm wide,
pale white with transx'crse rows of dark
markings, venter pale with a clump of
thick elongate setae posteriorly. Epigastric
furrow 0.86 mm from tracheal spiracle,
spiracle 0.74 mm from base of spinnerets.

Legs pale yellow, unmodified. Tibial
lengths (mm) and indices: I 1.87, 12; II
1.70, 13; III 1.27, 18; IV 1.73, 14. Ventral
spination: tibiae I, II 2-2-2, III 1-2-2,
IV 2-2-2; metatarsi I, II 2-0-0, III 2-0-2,
IV 2-2-2.

Palpus as in Figure 135.

Female (Curry Co., Oregon). Colora-
tion as in male.

Total length 5.83 mm. Carapace 2.65
mm long, 2.05 mm wide, cephalic width
1.17 mm, clypeus height 0.09 mm. Eyes:
diameters (mm): AME 0.10, ALE 0.12,
PME 0.12, PLE 0.13; anterior eye row
0.61 mm long, slightly recurved; posterior
eye row 0.87 mm long, procurved; MOQ
length 0.35 mm, front width 0.30 mm, back
width 0.44 mm; eye interdistances (mm):
AME-AME 0.09, AME-ALE 0.05, PME-
PME 0.19, PME-PLE 0.14, ALE-PLE 0.07.

Sternum 1.46 mm long, 1.04 mm wide.
Chelicerae 0.80 mm long with 3 promar-
ginal teeth and 7 retromarginal denticles.

Abdomen 4.00 mm long, 2.60 mm wide,
without thick setae ventrally. Epigastric
furrow 0.81 mm from tracheal spiracle,
spiracle 1.03 mm from base of spinnerets.

Tibial lengths (mm) and indices: I 1.87,
16; II 1.77, 16; III 1.31, 22; IV 1.87, 17.
Ventral spination as in male except tibiae
I 2-2-0 and IV 1-2-2.

Epigynum as in Figure 136, internal
genitalia as in Figure 137.

Natural history. Mature males have been
taken from late Marcli through early Sep-
tember, mature females from ],\tc March

242 Bulletin Museum of Comparative Zoology, Vol. 146, No. 4

through early November. Specimens have
been taken from redwoods and red cedars.
Distribution. Pacific coast from British
Coknnbia south to southern Cahfornia
(Map 2).

Wulfila O. P.-Cambridge

WuIfiJa O. P.-Cambridge, 1895, Biologia Central!
Americana, Aran., 1: 158. Type species Wulfila
pallidtis O. P.-Cambridge, 1895, designated by
Simon, 1897, Hist. Natur. Araign., 2: 103.

Cragits O. P.-Cambridge, 1896, Biologia Centrali
Americana, Aran., 1: 215. Type species by mono-
typy Cragiis pallidus O. P.-Cambridge, 1896.
NEW SYNONYMY.

Anyphaenella Bryant, 1931, Psyche, 38: 115. Type
species by original designation Clubiona salta-
hunda Hentz, 1847. NEW SYNONYMY.

Diagnosis. Wulfila may be easily recog-
nized by their long, thin, pale white legs.
Leg I in particular is greatly elongated,
with its tibial index usually 5 or less. Pal-
pal structure indicates that this genus is
closely related to Amjphaena. There are
probably more than fifty species in this
genus; most occur in Central America and
the West Indies.

Description. Total length 2.5-4.5 mm.
Carapace longer than wide, narrowed in
front to from one-half to two-thirds its
maximum width. Clypeus height greater
than anterior median eye diameter. Pos-
terior median, posterior lateral and anterior
lateral eyes subequal in size, somewhat
larger than anterior medians. Procurved
posterior eye row longer than straight an-
terior row. Median ocular quadrangle
twice as wide in back as in front. Anterior
median eyes separated by less than their
diameter, by roughly their diameter from
anterior laterals. Posterior medians sepa-
rated by almost twice their diameter, by
their diameter from posterior laterals. An-
terior laterals separated by roughly their
diameter from posterior laterals. Sternum
longer than wide, unmodified. Chelicerae
with 3-6 promarginal teeth, often on ca-
rina, and 5-10 retromarginal denticles.
Abdomen longer than wide, tracheal spira-

cle midway between epigastric furrow and
base of spinnerets. Leg formula 1423, legs
long, thin, pale white. Leg I greatly elon-
gated. Metatarsi I and II with two pairs
of ventral spines. Coxae of males often
with spurs and knobs; leg III spination
often reduced. Palpus with an elongated
median apophysis, enlarged conductor and
conspicuous embolus. Retrolateral tibial
apophysis greatly expanded except in W.
wunda. Epigyna and internal genitalia
small and diverse.

Variation. None of the species in this
genus show any significant individual or
geographic intraspecific variation in struc-
ture, size or coloration.

Key to Species

la. Carapace and abdomen with dark mark-
ings saltabunda

lb. Carapace and abdomen without dark
markings 2

2a. Males 3

2b. Females . 7

3a. At least one pair of coxae modified with
spurs or knobs 4

3b. All coxae unmodified alba

4a. Coxae I and/or II modified with spurs or
knobs 5

4b. Coxae III and/or IV modified with spurs
or knobs - 6

5a. Retrolateral tibial apophysis more than
half the tibial length (Fig. 93) — bryantae

5b. Retrolateral tibial apophysis less than half
the tibial length ( Fig. 95 ) ._ wunda

6a. Retrolateral tibial apophysis greatly ex-
panded at tip (Fig. 86) ..._. tantilh

6b. Retrolateral tibial apophysis not greatly
expanded at tip (Fig. 88) immaculella

7a. Epigynum with long ducts (Figs. 91, 97,
98 ) 8

7b. Epigynum without long ducts (Figs. 90,
96) : 10

8a. Epigynum with a heart-shaped atrium
(Fig. 97) wunda

8b. Epigynum without a heart-shaped atrium 9

9a. Epigynal ducts terminating far anterior of
epigynal openings ( Fig. 91 ) tantilla

9b. Epigynal ducts temiinating near epigynal

openings ( Fig. 98 ) immaculella

10a. Epigynum with anterolateral flaps, with-
out a medial ridge (Fig. 90) alba

10b. Epigynum without anterolateral flaps, with
a medial ridge (Fig. 96) bryantae

Spideh Family Anyphaenidaf, • Platnick 243

i
)

!-•-

\U^

•^

•,

rf ?

•

^ _- •^-^•.^%

•\

\/^

Wulfila saltabundQ^

--—— ^^"^

^^-j

Ifila tantilla

■^

Wulfila immaculella |

U '^

— u-^/- ...

Wulfllo albo

\-- -

1 ' ■

1 — -'i

V .'

^"^ /^

v-4

/■•f

Wulfllo bryantae

Map 3. Distributions of Teudis calcar, Wulfila alba, W. bryantae, W. immaculella, W. saltabunda, W. tantilla and
IV. wunda.

Wulfila pallidas O. P.-Cambridge
Figure 144

Wulfila palUdus O. P.-Cambridge, 1895, Biologia
Central! Americana, Aran., 1: 159, pi. 19, fig.
11 ( 9 ). Female holotype from Teapa, Ta-
basco, Mexico, in BMNH, examined. Bonnet,
1959, Bibliographia Araneorum, 2: 4832.

Wulfila pallida, Simon, 1897, Hist. Natur. Araign.,
2: 94. Roewer, 1954, Katalog der Araneae, 2:
554.

Vulfila pallida, Simon, 1897, Hist. Natur. Araign.,
2: 103.

Thi.s Mexican .species, though belonging
to a distinct species group, closely resem-
bles the North American Wulfila in body
form, leg length and coloration. It is the
type .species of Wulfila.

Wulfila saltabunda (Hentz),
new combination
Map 3; Figures 81, 82, 89, 99

Cluhiona saltabunda Hentz, 1847, J. Boston Soc.
Natur. Hist., 5: 453, pi. 23, fig. 23 ( 9 ). Fe-
male holotype from Alabama in Boston Soc.
Natur. Hist. (Boston Mu.seum of Science), de-
stroyed by beetles.

Amjphaena saltabunda, Fmerton, 1890, Trans.
Connecticut Acad. Sci., 8: 187, figs. 4-4d, $,

$ . Emerton, 1902, Common Spiders, p. 14,
figs. 46, 47, 5 , 9 .

Gayenna saltabunda, Comstock, 1912, Spider
Book, p. 563, figs. 638, 639, $, 9 .

Anyphaenella saltabunda, Bryant, 1931, Psyche,
38: 116, pi. 7, figs. 18, 22, $, 9. Comstock,
1940, Spider Book, rev. ed., p. 576, figs. 638,
639, $, 9. Ka.ston, 1948, Bull. Connecticut
Geol. Natur. Hist. Surv., 70: 406, figs. 1465-
1470, $, 9. Roewer, 1954, Katalog der
Araneae, 2: 530. Bonnet, 1955, Bibliographia
Araneorum, 2: 349.

Dia<i,nosis. Wulfila saltabunda is the
only species in this area which has dark
markings on the carapace and abdomen.
In addition, the shape of the retrolateral
tibial apophysis ( Fig. 82 ) and sperma-
thecae (Fig. 99) serve to distinguish it
from W. alba, its closest relative.

Male (Suffolk Co., New York). Total
length 3.06 mm. Carapace 1.46 mm long,
1.04 mm wide, cephalic width 0.54 mm,
clypeus height 0.07 mm, pale white with
thin dark border and two dark paramedian
longitudinal bands. Eyes: diameters
(mm): AME 0.05, ALE 0.09, PME 0.09,
PLE 0.09; anterior eye row 0.39 mm long,
straight; posterior eye row 0.50 mm long,

244 Bulletin Museum of Comparative Zoology, Vol. 146, No. 4

^■'^.,

■■ 90

Plate 7

Figures 81, 83, 85, 87. Left palpi, ventral view. Figures 82, 84, 86, 88. Left palpi, retrolateral view. Figures
89-91. Epigyna, ventral view. 81. 82, 89. Wulfila saltabunda (Hentz). 83, 84, 90. Wulfila alba (Hentz). 85, 86,
91. Wulfila tantilla Chickering. 87, 88. Wulfila immaculella (Gertsch).

Spider Faaiily Anyphaenidae • Phifnick

245

procurx'ed; MOQ length 0.22 mm, front
width 0.14 mm, back width 0.30 mm; eye
interdistances (mm): AME-AME 0.04,
AME-ALE 0.03, PME-PME 0.12, PiME-
PLE 0.06, ALE-PLE 0.03.

Sternum 0.86 mm long, 0.59 mm wide,
pale white with thick tran.slucent border
witli extensions to coxae and large triangu-
lar dark spots between coxae. Chelicerae
0.40 mm long with 6 promarginal teeth and
7 retromarginal denticles, pale white with
boss outlined in gray and several very long
setae. Laliium and endites pale white,
endites not in\'aginated at middle.

Abdomen 1.60 mm long, 0.97 mm wide,
pale white with transverse rows of dark
spots, venter with thin dark median line
anterior of epigastric furrow and two large
median dark spots between epigastric fur-
row and spinnerets. Epigastric furrow 0.40
mm from tracheal spiracle, spiracle 0.45
nmi from base of spinnerets.

Legs pale white, unmodified, though leg
III spination reduced. Tibial length (mm)
and indices: I 2.70, 4; II 1.42, 9; III 1.08,
13; IV 1.55, 10. Ventral spination: tibiae
I 2-2-0, II 1-1-0, III 0-1-0, IV 1-1-0;
metatarsi I, II 2-2-0, III 0-0-0, IV 1-2-2.

Palpus as in Figures 81, 82.

Female (Suffolk Co., New York). Col-
oration as in male.

Total length 4.18 mm. Carapace 1.78
mm long, 1.28 mm wide, cephalic width
0.70 mm, clypeus height 0.09 mm. Eyes:
diameters (mm): AME 0.06, ALE 0.09,
PME 0.09, PLE 0.10; anterior eye row 0.43
mm long, straight; posterior eye row 0.58
mm long, procurved. MOQ length 0.28
mm, front width 0.15 mm, back width 0.32
mm; eye interdistances (mm): AME-
AME 0.04, AME-ALE 0.05, PME-PME
0.14, PME-PLE 0.08, ALE-PLE 0.04.

Sternum 0.99 mm long, 0.74 mm wide.
Chelicerae 0.56 mm long with 6 promar-
ginal teeth and 10 retromarginal denticles.

Abdomen 2.47 mm long, 2.27 mm wide.
Epigastric furrow 0.85 mm from tracheal
spiracle, .spiracle 0.85 mm from base of
spinnerets.

Tibial lengths (mm) and indices: I 2.56,
5; II 1.31, 12; III 0.90, 20; IV 1.67, 12.
\'entral spination as in male except tibiae
II 2-2-0 and III 1-1-0 and metatarsi III
2-1-0.

Epigynum as in Figure 89, internal geni-
talia as in Figure 99.

Natural Jiistonj. Mature males have been
taken from mid-April through late August,
mature females from late April through
late August. Specimens have been taken
by sweeping and on apple trees.

Distrilnifion. Nova Scotia west to Min-
nesota and Nebraska, south to Florida and
eastern Texas ( Map 3 ) .

Wulfila alba (Hentz),
new combination
Map 3; Figures 83, 84, 90, 100

Cluhiona albeus Hentz, 1847, J. Boston See.

Natur. Hist., 5: 454, pi. 23, fig. 24 {$). Male

holotype from Alabama in Boston Soc. Natur.

Hist. (Boston Museum of Science), destroyed

by beetles.
Anyphaetia alhcns, Marx, 1883, in Howard, A List

of the Invertebrate Fauna of South Carolina, p.

24.
Chimcanthium alhens, Mar.x, 1890, Proc. U.S. Nat.

Mus., 12: 513.
AmiphaeucUa alba, Bryant, 1931, Psyche, 38:

116, pi. 7, figs. 20, 21, S, 9. Roewer, 1954,

Katalog der Araneae, 2: 530. Bonnet, 1955,

Bibliographia Araneorum 2: 349.

Diagnosis. Wulfila alba is closest to W.
sahaJninda but may be distinguished from
it by its lack of dark markings, the .spur-
like retrolateral tibial apophysis (Fig. 84)
and the shape of the spermathecae (Fig.
100).

Male (Orange Co., Florida). Coloration
as in Wulfila saltahunda except that dark
markings are entirely absent.

Total length 3.65 mm. Carapace 1.57
mm long, 1.21 mm wide, cephalic width
0.59 mm, clypeus height 0.07 mm. Eyes:
diameters (mm): AME 0.04, ALE 0.07,
PME 0.07, PLE 0.07; anterior eye row 0.36
mm long, straight; posterior eye row 0.49
mm long, procurved; MOQ length 0.22
mm, front width 0.12 mm, back width

246 Bulletin Museum of Comparative Zoology, Vol. 146, No. 4

0.26 mm; eye interdistances (mm): AME-
AME 0.04, AME-ALE 0.05, PME-PME
0.11, PiME-PLE 0.06, ALE-PLE 0.04.

Sternum 0.95 mm long, 0.70 mm wide.
Chelicerae 0.45 mm long with 6 promar-
ginal teeth and 7 retromarginal denticles.

Abdomen 2.12 mm long, 1.15 mm wide.
Epigastric furrow 0.67 mm from tracheal
spiracle, spiracle 0.79 mm from base of
spinnerets.

Tibial lengths (mm) and indices: I 3.13,
4; II 1.85, 7; III 1.12, 13; IV 2.03, 6. Ven-
tral spination: tibiae I 2-2-0, II 1-2-0, III
0-1-0, IV 1-1-0; metatarsi I, II, III 2-2-0,
IV 1-1-2.

Palpus as in Figures 83, 84.

Female (Indian River Co., Florida).
Coloration as in male.

Total length 4.00 mm. Carapace 1.62
mm long, 1.28 mm wide; cephalic width
0.58 mm, clypeus height 0.06 mm. Eyes:
diameters (mm): AME 0.04, ALE 0.07,
PME 0.08, PLE 0.08; anterior eye row 0.40
mm long, straight; posterior eye row 0.54
mm long, procurved; MOQ length 0.25 mm,
front width 0.14 mm, back width 0.29 mm;
eye interdistances (mm): AME-AME
0.05, AME-ALE 0.06, PME-PME 0.13,
PME-PLE 0.10, ALE-PLE 0.06.

Sternum 0.92 mm long, 0.74 mm wide.
Chelicerae 0.41 mm long with 6 promar-
ginal teeth and 9 retromarginal denticles.

Abdomen 2.66 mm long, 1.51 mm wide.
Epigastric furrow 0.68 mm from tracheal
spiracle, spiracle 0.88 mm from base of
spinnerets.

Tibial lengths (mm) and indices: I 3.13,
4; II 1.91, 8; III 1.05, 13; IV 2.07, 8. Ven-
tral spination as in male except tibiae II
2-2-0 and III 1-2-0 and metatarsi IV 2-
2-2.

Epigynum as in Figure 90, internal geni-
talia as in Figure 100.

Natural history. Mature males have been
taken from late March through early Au-
gust, mature females from early April
through late August. Specimens have been
taken by sweeping, on pines, and in Ma-
laise and pitfall traps.

Distribution. Maryland west to southern
Illinois, south to Florida and eastern Texas
(Map3).

Wulfila tantilla Chickering
Map 3; Figures 85, 86, 91, 101

Cragus palUdus O. P. -Cambridge, 1896, Biologia
Centrali Americana, Aran., 1: 215, pi. 26, fig.
10 { $). Male holotype from Santa Ana, Guate-
mala, in BMNH, examined. Preoccupied by
Wulfila pallidus O. P. -Cambridge, 1895. Roewer,
1954, Katalog der Araneae, 2: 535. Bonnet,
1956, Bibliographia Araneorum, 2: 1246.

Wulfila tantilla Chickering, 1940, Trans. Amer.
Microsc. Soc, 59: 119, figs. 64-66 {$). Male
holotype from El Valle, Panama, in MCZ,
examined. Roewer, 1954, Katalog der Araneae,
2: 555. NEW SYNONYMY.

Wulfila tenella Chickering, 1940, Trans. Amer.
Microsc. Soc, 59: 120, figs. 67, 68 ( $ ). Fe-
male holotype from El Valle, Panama, in MCZ,
examined. Roewer, 1954, Katalog der Araneae,
2: 555. NEW SYNONYMY.

Diagnosis. Wulfila tantilla is very closely
related to W. irnmaculella but may be dis-
tinguished by the greath' expanded tip of
the retrolateral tibial apophysis (Fig. 86)
and by the epigynal ducts terminating far
anterior of the epigynal openings (Fig.
91).

Male (Webb Co., Texas): Coloration as
in Wulfila alba, except that the posterior
declivity of the carapace is darkened.

Total length 3.02 mm. Carapace 1.62
mm long, 1.12 mm wide, cephalic width
0.63 mm, clypeus height 0.09 mm. Eyes:

Plate 8
Figures 92, 94, 107. Left palpi, ventral view. Figures 93, 95, 105. Left palpi, retrolateral view. Figures 96-98,
106. Epigyna, ventral view. Figures 99-104, 108. Internal genitalia, dorsal view. Figure 109. Body, dorsal view.
92, 93, 96, 102. Wulfila bryantae new species. 94, 95, 97, 104. Wulfila wunda new species. 98, 103. Wulfila
irnmaculella (Gertsch). 99. Wulfila saltabunda (Hentz). 100. Wulfila alba (Hentz). 101. Wulfila tantilla Chick-
ering. 105-109. Oxysoma cubana Banl<s. (Figs. 105, 106, 109 by Wilton Ivie, not to scale.)

Spider Family Anyphaenidae • Plalnick 247

248 Bulletin Museum of Comparative Zoology, Vol. 146, No. 4

diameters (mm): AME 0.06, ALE 0.07,
PME 0.09, PLE 0.09; anterior eye row
0.41 mm long, straight; posterior eye row
0.53 mm long, procurved; MOQ length
0.23 mm, front width 0.16 mm, back width
0.30 mm; eye interdistances (mm): AME-
AME 0.05, AME-ALE 0.04, PME-PME
0.13, PME-PLE 0.06, ALE-PLE 0.04.

Sternum 0.74 mm long, 0.63 mm wide.
Chelicerae 0.58 mm long with 5 promar-
ginal teeth on a carina and 8 retromarginal
denticles.

Abdomen 1.57 mm long, 0.97 mm wide.
Epigastric furrow 0.50 mm from tracheal
spiracle, spiracle 0.44 mm from base of
spinnerets.

Coxae III and IV with two small knobs.
Tibial lengths (mm) and indices: I 2.77, 3;
II 1.51, 9; III 0.86, 15; IV 1.51, 9. Ventral
spination: tibiae I, II 2-2-0, III, IV 1-2-0;
metatarsi I, II 2-2-0, III, IV 2-2-2.

Palpus as in Figures 85, 86.

Female (Hidalgo Co., Texas). Colora-
tion as in male of Wulfila alha.

Total length 2.92 mm. Carapace 1.34
mm long, 0.99 mm wide, cephalic width
0.67 mm, clypeus height 0.08 mm. Eyes:
diameters (mm): AME 0.05, ALE 0.06,
PME 0.06, PLE 0.06; anterior eye row 0.37
mm long, straight; posterior eye row 0.50
mm long, procurved; MOQ length 0.20
mm, front width 0.16 mm, back width 0.26
mm; eye interdistances ( mm ) : AME-
AME 0.05, AME-ALE 0.04, PME-PME
0.13, PME-PLE 0.06, ALE-PLE 0.05.

Sternum 0.89 mm long, 0.61 mm wide.
Chelicerae 0.50 mm long with 4 promar-
ginal teeth and 6 retromarginal denticles.

Abdomen 1.62 mm long, 1.15 mm wide.
Epigastric furrow 0.59 mm from tracheal
spiracle, spiracle 0.52 mm from base of
spinnerets.

Legs unmodified. Tibial lengths (mm)
and indices: I 2.36, 5; II 1.21, 11; III 0.77,
18; IV 1.40, 11. Ventral spination as in
male save tibiae IV 1-1-0.

Epigynum as in Figure 91, internal geni-
talia as in Figure 101.

Natural history. Mature males have been

taken from mid-April through mid-Octo-
ber, mature females apparently year-round.
Nothing is known of the habits of this
species.

Distribution. Southern Texas south to
the Canal Zone (Map 3).

Wulfila immaculella (Gertsch),
new combination
Map 3; Figures 87, 88, 98, 103

Amjphac'uella immaculella Gertsch, 1933, Amer.
Mus. Novitates, No. 637: 9, fig. 14 ( ? ). Fe-
male holotype from Sabino Basin, Santa Cata-
lina Movmtains, Arizona, in AMNH, examined.
Roewer, 1954, Katalog der Araneae, 2: 530.
Bonnet, 1955, Bibliographia Araneorum, 2: 349.

Diagnosis. Wulfila immaculella is very
closely related to W. tantilla but may be
distinguished by the unexpanded tip of the
retrolateral tibial apophysis (Fig. 88) and
by the epigynal ducts terminating near the
epigynal openings (Fig. 98).

Male (Sonora, Mexico). Coloration as
in Wulfila alba.

Total length 3.60 mm. Carapace 1.64
mm long, 1.12 mm wide, cephalic width
0.67 mm, clypeus height 0.08 mm. Eyes:
diameters (mm): AME 0.05, ALE 0.08,
PME 0.08, PLE 0.08; anterior eye row 0.40
mm long, straight; posterior eye row 0.52
mm long, procurved; MOQ length 0.22
mm, front width 0.14 mm, back width
0.31 mm; eye interdistances (mm): AME-
AME 0.05, AME-ALE 0.05, PME-PME
0.14, PME-PLE 0.08, ALE-PLE 0.04.

Sternum 0.90 mm long, 0.68 mm wide.
Chelicerae 0.51 mm long with 4 promar-
ginal teeth and 5 retromarginal denticles.

Abdomen 2.07 mm long, 1.00 mm wide.
Epigastric furrow 0.63 mm from tracheal
spiracle, spiracle 0.74 mm from base of
spinnerets.

Coxae III with one, coxae IV with two
small knobs. Tibial lengths (mm) and in-
dices: I 3.42, 3; II 2.05, 5; III 1.30, 11; IV
2.11, 8. Ventral spination: tibiae I 2-2-0,
II 1-2-0, III 0-1-0, IV 1-1-0; metatarsi I,
II 2-2-0, III, IV 2-1-2.

Palpus as in Figures 87, 88.

Spider Family Anyphaenidae • Plat nick 249

Fenmle (Sonora, Mexico). Coloration as
in male of Wiilfila alba.

Total length 3.64 mm. Carapace 1.5S
mm long, 1.08 mm wide, cephalic width
0.61 mm, clypen.s height 0.10 mm. Eyes:
diameters (mm): AME 0.05, ALE 0.07,
PME 0.08, PLE 0.07; anterior eye row
0.40 mm long, straight; posterior eye row
0.53 mm long, procurved; MOQ length
0.22 mm, front width 0.15 mm, back width
0.30 mm; eye interdistances (mm): AME-
AiME 0.05, AME-ALE 0.04, PME-PME
0.12, PME-PLE 0.08, ALE-PLE 0.04.

Sternum 0.94 mm long, 0.71 mm wide.
Chelicerae 0.53 mm long with teeth as in
male.

Abdomen 2.05 mm long, 1.40 mm wide.
Epigastric furrow 0.65 mm from tracheal
spiracle, spiracle 0.74 mm from base of
spinnerets.

Legs unmodified. Tibial lengths (mm)
and indices: I 2.81, 4; II 1.62, 9; III 0.92,
15; IV 1.75, 8. Ventral spination as in male
except tibiae II 2-2-0 and III 1-1-0.

Epigynum as in Figure 98, internal geni-
talia as in Figure 103.

Natural Jiistory. Mature males have been
taken in July, mature females in June and
July. One male was taken on Platanus sp.

Distrihuiion. Southern Arizona and So-
nora (Map 3).

Wulfila bryantae new species
Map 3; Figures 92, 93, 96, 102

Types. Male holotype, female paratype
from 5 miles east of Edinburg, Hidalgo
Co., Texas, 20 April 1937 (S. Mulaik), de-
posited in AMNH. Male and female para-
types from Jim Wells and Cameron Coun-
ties, Texas, deposited in MCZ. The
specific name is a patronym in honor of
Miss Elizabeth Bryant, in recognition of her
pioneering work on North American any-
phaenids.

Diagnosis. Wulfila bryantae is a distinc-
tive species easily recognized by its stubby
median apophysis (Fig. 92) and the medial
ridge on the epigynum (Fig. 96).

Male (Hidalgo Co., Texas). Coloration
as in Wulfila alba.

Total length 3.35 mm. (Carapace 1.44
mm long, 1.08 mm wide, cephalic width
0.81 mm, clypeus height 0.10 mm. Eyes:
diameters (mm): AME 0.06, ALE 0.06,
PME 0.07, PLE 0.07; ant(>rior eye row 0.49
mm long, slightly rcx-urved; posterior eye
row 0.62 mm long, procurved; MOQ length
0.26 mm, front width 0.20 mm, back width
0.30 mm; eye interdistances (mm): AME-
AME 0.07,^ AME-ALE 0.07, PME-PME
0.15, PME-PLE 0.13, ALE-PME 0.06.

Sternum 0.97 mm long, 0.55 mm wide.
Chelicerae 0.73 mm long with 3 promar-
ginal teeth on a carina and 7 retromarginal
denticles.

Abdomen 1.80 mm long, 1.12 mm wide.
Epigastric fvnrow 0.56 mm from tracheal
spiracle, spiracle 0.68 mm from base of
spinnerets.

Coxae I with a small knob, coxae II with
two spurs. Tibial lengths (mm) and in-
dices: I 2.76, 4; II 1.85, 7; III 0.92, 15; IV
1.89, 7. \'entral spination: tibiae I, II 2-
2-0, III 1-2-0, IV 1-1-0; metatarsi I, II 2-
2-0, III, IV 2-1-2.

Palpus as in Figures 92, 93.

Female (Hidalgo Co., Texas). Colora-
tion as in male of Wulfila alba.

Total length 3.78 mm. Carapace 1.44
mm long, 0.99 mm wide, cephalic width
0.74 mm, clypeus height 0.07 mm. Eves:
diameters (mm): AME 0.06, ALE 6.07,
PME 0.06, PLE 0.07; anterior eye row
0.42 mm long, slightly recurved; posterior
eye row 0.59 mm long, procurved; MOQ
length 0.23 mm, front width 0.17 mm, back
width 0.27 mm; eye interdistances (mm):
AME-AME 0.06, AME-ALE 0.05, PME-
PME 0.14, PME-PLE 0.13, ALE-PLE 0.06.

Sternum 0.74 mm long, 0.64 mm wide.
Chelicerae 0.62 mm long with 5 promar-
ginal teeth and 5 retromarginal denticles.

Abdomen 2.59 mm long, 2.16 nun wide.
Epigastric furrow 0.90 mm from tracheal
spiracle, spiracle 0.88 mm from base of
spinnerets.

250 Bulletin Museum of Comparative Zoology, Vol. 146, No. 4

Legs unmodified. Tibial lengths (mm)
and indices: I 2.24, 5; II 1.37, 9; III 0.72,
19; IV 1.35, 10. Ventral spination as in
male except tibiae III, IV 2-2-0 and meta-
tarsi III, IV 2-0-2.

Epigynum as in Figure 96, internal
genitalia as in Figure 102.

Natural history. Mature males have been
taken from late April through early June,
mature females from early April through
early December. Nothing is known of the
habits of this species.

Distribution. Southern Texas and Ta-
maulipas (Map 3).

Wulfila wunda new species
Map 3; Figures 94, 95, 97, 104

Wulfila immaculata, Bryant (not Banks), 1936,
Psyche, 43: 98, fig. 1, $. Male allotype from
Brichell Hammock, Florida Keys, in MCZ,
examined. Not Wulfila immaculata Banks, 1914,
Bull. Amer. Mus. Natur. Hist., 33: 640, pi. 43,
fig. 7, 9 . Female holotype from Vinales, Pinar
del Rio, Cuba, in AMNH, examined.

Types. Male holotype, female paratype
from Tavernier, Monroe Co., Florida, 16
February 1951 (A. M. Nadler), deposited
in AMNH. Male and female paratypes
from Dade Co., Florida, deposited in MCZ.
The specific name is an arbitrary combina-
tion of letters.

Diagnosis. Wulfila wunda is a distinc-
tive species the genitalia of which are
quite different from those of the other
Wulfila in America north of Mexico: the
retrolateral tibial apophysis is very short
( Fig. 95 ) and the epigynum has an atrium
(Fig. 97).

Male (Dade Co., Florida). Coloration
as in Wulfila alba.

Total length 3.42 mm. Carapace 1.55
mm long, 1.08 mm wide, cephalic width
0.68 mm, clypeus height 0.06 mm. Eyes:
diameters (mm): AME 0.05, ALE 0.07,
PME 0.08, PLE 0.08; anterior eye row 0.48
mm long, straight; posterior eye row 0.59
mm long, procurved; MOQ length 0.20
mm, front width 0.14 mm, back width 0.29
mm; eye interdistances (mm): AME-

AME 0.05, AME-ALE 0.09, PME-PME
0.14, PME-PLE 0.12, ALE-PLE 0.04.

Sternum 1.06 mm long, 0.70 mm wide.
Chelicerae 0.85 mm long with 4 promar-
ginal teeth and 6 retromarginal denticles.

Abdomen 1.91 mm long, 1.01 mm wide.
Epigastric furrow 0.70 mm from ti'acheal
spiracle, spiracle 0.76 mm from base of
spinnerets.

Coxae II with a small knob. Tibial
lengths (mm) and indices: I 4.10, 3; II
1.87, 8; III 1.01, 15; IV 2.05, 7. Ventral
spination: tibiae I, II 2-2-0, III 0-1-0, IV
0-2-0; metatarsi I, II 2-2-0, III 0-2-0, IV
2-1-2.

Palpus as in Figures 94, 95.

Female (Dade Co., Florida). Colora-
tion as in male of Wulfila alba.

Total length 3.74 mm. Carapace 1.55
mm long, 1.15 mm wide, cephalic width
0.72 mm, clypeus height 0.07 mm. Eyes:
diameters (mm): AME 0.04, ALE 0.06,
PME 0.07, PLE 0.07; anterior eye row 0.49
mm long, straight; posterior eye row 0.59
mm long, procurved; MOQ length 0.20
mm, front width 0.15 mm, back width
0.30 mm; eye interdistances ( mm ) : AME-
AME 0.06, AME-ALE 0.10, PME-PME
0.16, PME-PLE 0.12, ALE-PME 0.04.

Sternum 0.90 mm long, 0.67 mm wide.
Chelicerae 0.65 mm long with 5 promar-
ginal teeth and 9 retromarginal denticles.

Abdomen 2.16 mm long, 1.15 mm wide.
Epigastric furrow 0.74 mm from tracheal
spiracle, spiracle 0.83 mm from base of
spinnerets.

Legs unmodified. Tibial lengths (mm)
and indices: I 3.13, 4; II 1.44, 10; III 0.76,
20; IV 1.58, 9. Ventral spination as in male
except tibiae III 1-2-0 and IV 0-1-0 and
metatarsi III 1-2-0 and IV 1-2-2.

Epigynum as in Figure 97, internal geni-
talia as in Figure 104.

Natural history. Mature males have been
taken from mid-February through mid-
May, mature females apparently year-
round. Nothing is known of the habits of
this .species.

Spider Family Anyphaenidae • Platnick 251

Distrihuiion. Southern Florida, Culxi,
and Mona Island ( Map 3).

Aysha Keyserling

Aysha Keyserling, 1891, Spinn. Ainer. ( Brasil.
Spiiin.), 3: 83, 129. Type species Aysha pros-
pera Keyserling, 1891, designated by Simon,
1897, Hist. Natm-. Araig., 2: 104.

Diagnosis. Aysha is easily recognized by
the greatly adxanced placement of the
tracheal spiracle, located just behind the
epigastric furrow. The genitalic structure
is quite different from that of Amjphacna
and Wulfila and the genus undoubtedly
represents a different evolutionary line.
There are probably more than thirty spe-
cies in this genus; they occur commonly
in both North and South America.

Description. Total length 4-9 mm. Cara-
pace longer than wide, narrowed in front
to more than half its maximum width.
Clypeus height roughly equal to anterior
median eye diameter. All eyes subequal in
size. Procurved posterior eye row longer
than recurved anterior row. Median ocular
quadrangle longer than wide in front,
wider in back than long. Anterior median
eyes separated by slightly less than their
diameter, slightly closer to anterior laterals.
Posterior medians separated by up to twice
their diameter, closer to posterior laterals.
Anterior laterals separated by their radius
from posterior laterals. Sternum longer
than wide, unmodified. Chelicerae with 3-
4 promarginal teeth and 7-9 retromarginal
denticles. Abdomen longer than wide,
tracheal spiracle much closer to epigastric
furrow than to base of spinnerets. Leg
formula 1423, legs unmodified. Metatarsi I
and II with one pair of \'entral spines. Pal-
pus with greatly enlarged base of embolus,
long curving embolus and short conductor.
Ventral tibial apophysis sometimes present
in addition to retrolateral tibial apophysis.
Epigynum with anterior median opening
and two sidepieces. Internal genitalia with
long, sometimes coiling, ducts.

Variation. Only Aysha gracilis shows

significant variation, and that is in size and
not strnetine or coloration. The size of both
the whole animal and ol the genitalia vary
geographically. The largest .specimens
( males with cymbium length averaging 1.3
mm) occur in Virginia and surrounding
states, with smaller individuals occurring
in the north (New England and Michigan
males with cymbium length averaging 1.1
mm) and in the south (Texas males with
cymbium length averaging 0.9 mm).

Key to Species

la. Males 2

lb. Females -..- 7

2a. Palpus without a ventral tibial apophysis
(VTA) (Figs. Ill, 119) -.._ 3

2b. Palpus with a ventral tibial apophysis
(VTA), sometimes small, transparent, eas-
ily overlooked (Figs. 113, 115, 117, 121) 4

3a. Embolus restricted to distal half of palpal
bulb (Fig. 118) arunda

3b. Embolus not restricted to distal half of pal-
pal bulb (Fig. 110) velox

4a. VTA erect, sclerotized, relatively large
(Figs. 113, 115) 5

4b. VTA recumbent, transparent, relativelv
small (Figs. 117, 121) '. 6

5a. Distal retrolateral tip of tegulum with a
flap covering embolus (Fig. 112) decepta

5b. Distal retrolateral tip of tegulum with a
sharp point underlying embolus (Fig. 114)
— incursa

6a. Base of embolus recurved, with a sharp
spike (Fig. 120) camhridgei

6b. Base of embolus not recurved, forming a
smooth arc (Fig. 116) fitacilis

7a. Internal genitalia with simple uncoiled
ducts (Figs. 124, 127, 141, 143) 9

7b. Internal genitalia coiled or with accessory
ducts (Figs. 125, 142) 8

8a. Internal genitalia highly coiled ( Fig.
125 ) '...... vclox

8b. Internal genitalia not coiled but with loop-
ing accessory ducts (Fig. 142) arunda

9a. Median epig\'nal opening near anterior
rim (Figs. 123, 126, 138) 10

91). Median epigynal opening near middle of

epigynum ( Fig. 140 ) -.. gracilis

10a. Median epig\nal opening much wider than

epigynal sidepieces (Fig. 138) cand)ridgc'i
lOb. Median epigynal opening not wider than

epigynal sidepieces (Figs. 123, 126) 11

11a. Base of epig>nal sidepieces near epigastric
furrow (Fig. 126); internal genitalia with
angular ducts (Fig. 127) incursa

252 Bulletin Museum of Comparative Zoology, Vol. 146, No. 4

. ^ Aysha gracilis

I /

; 1 / HM

~T— 7
I

Ayshc velox

♦^^'^r

cQ^*

" 1 \

)

(

-A

Aysha arunda

.„ V

Map 4. Distributions of Aysha arunda. A. cambridgei, A. decepta, A. gracilis, A. incursa and A. velox.

111). Base of epigynal sidepieces far from
epigastric furrow (Fig. 123); internal
genitalia with rounded ducts (Fig. 124)
decepta

Aysha prospers Keyserling
Figure 145

Ay.sha prospera Keyserling, 1891, Spinnen Amer-
ikas (Brasil. Spinn.), 3: 129, pi. 4, fig. 88
( 9 ). Female holotype from Rio Grande, Brasil,
in BMNH, examined. Roewer, 1954, Katalog
der Araneae, 2: 533. Bonnet, 1955, Biblio-
graphia Araneorum, 2: 838.

This South American species, type spe-
cies of Aysha, is a member of a large, dis-
tinct species group. Somatic characters
clearly ally it with the North American
forms included in the genus.

Aysha gracilis (Hentz)

Map 4; Figures 116, 117, 140, 143

Chihiona gracilis Hentz, 1847, J. Boston Soc.
Natur. Hist., 5: 452, pi. 23, fig. 9(5). Type

specimens from North Carolina and Alabama in
Boston Soc. Natur. Hist. (Boston Museum of
Science), destroyed by beetles.

Anijphaena gracilis, L. Koch, 1836, Arach. Fam.
brassidae, p. 195, pi. 8, fig. 130, 9 . Comstock,
1912, Spider Book, p. 561, fig. 633, $ (not
fig. 632).

Anijphaena rubra Emerton, 1890, Trans. Connecti-
cut Acad. Sci., 8: 186, pi. 6, fig. 1(9). Male
allotype (?) from Franklin Park, Boston, Mas-
sachusetts, in MCZ, examined. Emerton, 1909,
Trans. Connecticut Acad. Sci., 14: 220, pi. 9,
fig. 8-8c, $ .

Aysha gracilis, Bryant, 1931, Psyche, 38: 119, pi.
7, fig. 13, pi. 8, fig. 26, $, 9. Chickering,
1939, Pap. Michigan Acad. Sci., 24: 53, figs.
9-11, $, 9. Comstock, 1940, Spider Book,
rev. ed., p. 575, fig. 633, $ (not fig. 632).
Kaston, 1948, Bull. Connecticut Geol. Natur.
Hist. Surv., 70: 405, figs. 1452, 1459-1464, $,
9 . Roewer, 1954, Katalog der Araneae, 2: 534.
Bonnet, 1955, Bibliographia Araneorum, 2: 837.

Diagnosis. Aysha gracilis is closest to
A. cambridgei but lacks the sharp spike on
the proximal edge of the base of embolus
(Fig. 116) of that .species. Females have

Spider Family Anyphaenidae • Platnick 253

Plate 9
Figures 110, 112, 114, 116. Left palpi, ventral view. Figures 111, 113, 115, 117
110, 111. Aysha velox (Becker). 112, 113. Aysha decepta (Banks)
116, 117. Aysha gracilis (Hentz).

Left palpi, retrolateral view/.
114, 115. Aysha incursa (Chamberlin).

the median epigynal opening near the mid- long, 2.02 mm wide, cephalic width 1.17

die of the epigynum (Fig. 140). Variation mm, clypeus heigiit 0.09 mm, light orange-

in thi.s species is discussed above. brown, darkest anteriorly, with thin dark

Male (Middlesex Co., Massachusetts), border and two dark paramedian longitudi-

Total length 5.73 mm. Carapace 2.56 mm nal bands. Eyes: diameters (mm): AME

254

Bulletin Museum of Comparative Zoology, Vol. 146, No. 4

0.09, ALE 0.11, PME 0.09, PLE 0.11; an-
terior eye row 0.60 mm long, slightly re-
cnrved; posterior eye row 0.80 mm long,
procurved; MOQ length 0.32 mm, front
width 0.26 mm, back width 0.38 mm; eye
interdistances (mm): AME-AME 0.09,
AME-ALE 0.07, PME-PME 0.19, PME-
PLE 0.15, ALE-PLE 0.06.

Sternum 1.44 mm long, 1.08 mm wide,
light orange-brown with translucent border
and darkened extensions to coxae. Chelic-
erae 1.12 mm long with 4 promarginal
teeth and 8 retromarginal denticles, dark
orange-brown proximally, dark brown dis-
tally. Labium and endites light orange-
brown, darkest proximally. Endites sharply
invaginated at middle.

Abdomen 3.20 mm long, 1.73 mm wide,
pale grayish-brown with transverse rows
of dark markings, venter pale. Epigastric
furrow 0.40 mm from tracheal spiracle,
spiracle 1.73 mm from base of spinnerets.

Legs light orange-brown with distal seg-
ments darkest. Tibial lengths (mm) and
indices: I 2.64, 10; II 1.87, 15; III 1.19, 26;
IV 2.09, 15. Vential spination: tibiae I, II
2-2-2, III 1-2-2; IV 2-2-2; metatarsi I, II
2-0-0, III 2-1-2, IV 2-2-2.

Palpus as in Figure 116, 117.

Female (Washington Co., Arkansas).
Coloration as in male.

Total length 8.42 mm. Carapace 2.75
mm long, 2.11 mm wide, cephalic width
1.47 mm, clypeus height 0.10 mm. Eyes:
diameters (mm): AME 0.14, ALE 0.14,
PME 0.13, PLE 0.14; anterior eye row 0.43
mm long, recurved; posterior eye row 1.04
mm long, procurved; MOQ length 0.43
mm, front width 0.36 mm, back width 0.49
mm; eye interdistances (mm): AME-
AME 0.09, AME-ALE 0.08, PME-PME
0.22, PME-PLE 0.18, ALE-PLE 0.06.

Sternum 1.84 mm long, 1.31 mm wide.
Chelicerae 1.57 mm long with teeth as in
male.

Abdomen 5.76 mm long, 3.53 mm wide.
Epigastric furrow 0.68 mm from tracheal
spiracle, spiracle 3.24 mm from base of
spinnerets.

Tibial lengths (mm) and indices: I 2.56,
16; II 1.94, 20; III 1.26, 30; IV 2.30, 17.
Ventral spination: tibiae I 2-2-0, II 1-2-1,
III 1-1-2, IV 1-2-2; metatarsi I, II 2-0-0,
III 2-0-2, IV 2-2-2.

Epigynum as in Figure 140, internal
genitalia as in Figure 143.

Natural history. Mature males and fe-
males have been taken year-round. Speci-
mens have been taken by sweeping, in
pitcher plants, on loblolly pine, in fall web-
worm nests and frequently in houses.

Distribution. New England west to Wis-
consin and Iowa, south to Florida and east-
ern Texas ( Map 4 ) .

Aysha cam bridge! Bryant

Map 4; Figures 120, 121, 138, 141

Aysha cambiidgei Bryant, 1931, Psyche, 38: 119,
pi. 7, fig. 15 { $ ). Male holotype from
Guanajuato, Mexico, in MCZ, examined. Roe-
wer, 1954, Katalog der Araneae, 2: 532. Bon-
net, 1955, Bibliographia Araneorum, 2: 836.

Diagnosis. Aysha cambridgei is closely
related to A. gracilis but has a distinctive
spike on the proximal edge of the base of
the embolus (Fig. 120) and the median
epigynal opening near the anterior rim of
the epigynum ( Fig. 138 ) .

Male (Jeff Davis Co., Texas). Colora-
tion as in Aysha gracilis except that the ab-
domen is pale white with two dark para-
median longitudinal bands.

Total length 5.87 mm. Carapace 2.41
mm long, 1.91 mm wide, cephalic width
0.97 mm, clypeus height 0.11 mm. Eyes:
diameters (mm): AME 0.11, ALE 0.12,
PME 0.11, PLE 0.11; anterior eye row 0.57
mm long, recurved; posterior eye row 0.75
mm long, procurved; MOQ length 0.33
mm, front width 0.28 mm, back width 0.38
mm; eye interdistances (mm): AME-
AME 0.06, AME-ALE 0.05, PME-PME
0.16, PME-PLE 0.11, ALE-PLE 0.05.

Sternum 1.42 mm long, 1.01 mm wide.
Chelicerae 0.98 mm long with 4 promar-
ginal teeth and 7 retromarginal denticles.

Abdomen 3.49 mm long, 1.58 mm wide.
Epigastric furrow 0.68 mm from tracheal

Spideu Family Anyphaenidae • Platnick 255

122

126

127

125

Plate 10
Figures 118, 120. Left palpi, ventral view. Figures 119, 121. Left palpi, retrolateral view. Figures 122, 123,
126. Epigyna, ventral view. Figures 124, 125, 127. Internal genitalia, dorsal view. 118, 119. Aysha arunda new
species. 120, 121. Aysha cambridgei Bryant. 122, 125. Aysha velox (Becker). 123, 124. Aysha decepta
(Banks). 126, 127. Aysha incursa (Chamberlin).

.spiracle, spiracle 1.55 mm from base ot Ventral spination: tibiae I, II 2-2-2, III

spinnerets. 1-2-2, IV 2-2-2; metatarsi I, II 2-0-0, III,

Tibial lengths (mm) and indices: I 3.06, IV 2-2-2.

8; II 1.87, 13; III 1.28, 21; IV 2.16, 12. Palpus as in Figures 120, 121.

256 Bulletin Museum of Comparative Zoology, Vol. 146, No. 4

Female (Henderson Co., Texas). Color-
ation as in male.

Total length 8.50 mm. Carapace 3.35
mm long, 2.52 mm wide, cephalic width
1.69 mm, clypeus height 0.12 mm. Eyes:
diameters (mm): AME 0.14, ALE 0.16,
PME 0.14, PLE 0.14; anterior eye row 0.84
mm long, recm'ved; posterior eye row 1.11
mm long, procurved; MOQ length 0.42
mm, front width 0.37 mm, back width 0.50
mm; eye interdistances (mm): AME-
AME 0.10, AME-ALE 0.07, PME-PME
0.22, PME-PLE 0.20, ALE-PLE 0.05.

Sternum 1.91 mm long, 1.22 mm wide.

'&'

Chelicerae 1.69 mm long with teeth as in
male.

Abdomen 5.04 mm long, 2.88 mm wide.
Epigastric furrow 0.61 mm from tracheal
spiracle, spiracle 3.17 mm from base of
spinnerets.

Tibial lengths (mm) and indices: 12.88,
12; II 2.07, 17; III 1.40, 26; IV 2.57, 15.
Ventral spination as in male except tibiae
I, II 2-2-0 and III 2-2-2.

Epigynum as in Figure 138, internal
genitalia as in Figure 141.

Natural history. Mature males have been
taken from mid-June through early August,
mature females from late May through
early August. Specimens have been taken
on trees and shrubs.

Distribution. South central states from
Alabama to western Texas, south to central
Mexico (Map 4).

Aysha decepta (Banks)

Map 4; Figures 112, 113, 123, 124

Amjphaena decepta Banks, 1899, Proc. Ent. Soc.
Washington, 4: 190. Female holotype from
Brazos Co., Texas, in MCZ, examined.

Aysha mimita F. O. P.-Canibridge, 1900, Biologia
Centrali Americana, Aran., 2: 99, pi. 7, figs.
18-19 { $, ? ). Male holotype, female allotype
from Guatemala, in BMNH, examined. Bryant,
1931, Psyche, 38: 120, pi. 7, fig. 17, $. Roe-
wer, 1954, Katalog der Araneae, 2: 533. Bon-
net, 1955, Bibliographia Araneorum, 2: 838.
NEW SYNONYMY.

Aijsha decepta, Bryant, 1931, Psyche, 38: 120, pi.
7, fig. 16, pi. 8, fig. 27, $, 9. Roewer, 1954,

Katalog der Araneae, 2: 534. Bonnet, 1955,
Bibliographia Araneorum, 2: 836.

Diagnosis. Aysha decepta is very closely
related to A. incursa but has a characteris-
tic flap (on the retrolateral tip of the tegu-
lum) that covers the embolus (Fig. 112),
while the base of the epigynal sidepieces
is a considerable distance from the epigas-
tric furrow ( Fig. 123 ) . Both morphological
and zoogeographical data (Map 4) indi-
cate that these two species are each other's
nearest relatives.

Male (Hidalgo Co., Texas). Coloration
as in Aysha camhridgei.

Total length 4.82 mm. Carapace 2.25
mm long, 1.76 mm wide, cephalic width
1.06 mm, clypeus height 0.10 mm. Eyes:
diameters (mm): AME 0.08, ALE 0.10,
PME 0.11, PLE 0.11; anterior eye row 0.58
mm long, straight; posterior eye row 0.75
mm long, procurved; MOQ length 0.23
mm, front width 0.24 mm, back width 0.39
mm; eye interdistances ( mm ) : AME-
AME 0.08, AME-ALE 0.06, PME-PME
0.18, PME-PLE 0.12, ALE-PLE 0.05.

Sternum 1.37 mm long, 0.85 mm wide.
Chelicerae 0.97 mm long with 4 promar-
ginal teeth and 7 retromarginal denticles.

Abdomen 2.74 mm long, 1.39 mm wide.
Epigastric furrow 0.38 mm from tracheal
spiracle, spiracle 1.28 mm from base of
spinnerets.

Tibial lengths (mm) and indices: I 2.54,
9; II 1.67, 14; III 1.01, 25; IV 1.89, 16.
Venti-al spination: tibiae I 2-2-0, II 1-2-0,
III 1-2-2, IV 2-2-2; metatarsi I, II 2-0-0,
III, IV 2-2-2.

Palpus as in Figures 112, 113.

Female (E. Baton Rouge Parish, Louisi-
ana). Coloration as in male of Aysha cam-
hridgei.

Total length 5.76 mm. Carapace 2.45
mm long, 1.80 mm wide, cephalic width
1.17 mm, clypeus height 0.09 mm. Eyes:
diameters (mm): AME 0.10, ALE 0.13,
PME 0.12, PLE 0.12; anterior eye row 0.60
mm long, recurved; posterior eye row 0.76
mm long, procurved; MOQ length 0.30
mm, front width 0.26 mm, back width

Spider Family Anyphaenidae • Plahiick 257

0.3(S mm; eye interdistances (mm): AME-
AME 0.07, AME-ALE 0.05, PME-PME
0.15, PME-PLE 0.11, ALE-PLE 0.04.

Sternum 1.35 mm long, 0.95 mm widc\
Chelicerae 0.89 mm long with 4 promar-
ginal teeth and 8 retromarginal dentieles.

Abdomen 3.51 mm long, 2.20 mm wide.
Epigastric furrow 0.41 mm from tracheal
spiracle, spiracle 1.78 mm from base of
.spinnerets.

Tibial lengths (mm) and indices: I 1.84,
14; II 1.40, 19; III 0.93, 29; IV 1.82, 17.
Ventral spination as in male except tibiae
III 1-1-2 and metatarsi III 2-1-2.

Epigynum as in Figure 123, internal
genitalia as in Figure 124.

Natural history. Mature males and fe-
males have been taken every month except
January and February. Specimens are
commonly found in great quantities in
wasp nests and occasionally in houses.

Distribution. Northern Florida west to
eastern Texas, south to Costa Rica ( Map
4).

Aysha incursa (Chamberlin)

Map 4; Figures 114, 115, 126, 127

Anypliacna incursa Chamberlin, 1919, Pomona
Coll. J. Ent. Zool., 12: 12, pi. 5, fig. 2(9).
Female holotype from Claremont, California,
in MCZ, examined. Bryant, 1931, Psyche, .38:
120 (sub Aysha dccepta [sic]). Roewer, 1954,
Katalog der Araneae, 2 : 534 ( sub Aysha de-
cepta [sic]). Bonnet, 1955, Bibliographia
Araneorum, 2: 836 (sub Aysha deccpta [sic]).

Anyphaena johnstoni Chamberlin, 1924, Proc.
California Acad. Sci., 12: 662, figs. 105, 106
(5, 9 ). Female holotype, male allotype from
San Pedro Nolasco Island, Gulf of California, in
California Academy of Sciences. Paratype male
from San Marcos Island, Gulf of California, in
MCZ, examined. Bryant, 1931, Psyche, 38: 120
(sub Aysha decepta [sic]). Bonnet, 1955,
Bibliographia Araneorum, 2: 836 (sub Aysha
dccepta jsic] ).

Anyphaena ni^.rifwns Chamberlin and Woodburw
1929, Proc. Biol. Soc. Washington, 42: 137, pi.
1, fig. 4 ( 9 ). Female holotype from St.
George, Utah, in AMNH, e.\amined. NEW
SYNONYMY.

Aysha nigrifrons, Bryant, 1931, Psyche, 38: 121.
Roewer, 1954, Katalog der Araneae, 2: 534.
Bonnet, 1955, Bibliographia Araneorum, 2: 838.

Diapwsis. Aysha incursa is very closely
related to A. decepta but has a distinctive
sharp point on the retrolateral tip of the
tegulum (Fig. 114), while the base of the
epigynal sidepieces is near the epigastric
furr(')w (Fig. 126).

Male (Tulare Co., California). Colora-
tion as in Aysha camhrid^ei.

Total length 6.08 mm. Carapace 3.02
mm long, 2.18 mm wide, cephalic width
1.22 mm, clypeus height 0.12 mm. Eyes:
diameters (mm): AME 0.11, ALE 0.12,
PME 0.11, PLE 0.12; anterior eye row
0.66 mm long, recurved; posterior eye row
0.84 mm long, procurved; MOQ length
0.33 mm, front width 0.31 mm, back width
0.42 mm; eye interdistances (mm): AME-
AME 0.10, AME-ALE 0.07, PME-PME
0.21, PME-PLE 0.17, ALE-PLE 0.05.

Sternum 1.67 mm long, 1.08 mm wide.
Chelicerae 1.22 mm long with 3 promar-
ginal teeth and 8 retromarginal denticles.

Abdomen 3.38 mm long, 1.80 mm wide.
Epigastric furrow 0.50 mm from tracheal
spiracle, spiracle 1.75 mm from base of
spinnerets.

Tibial lengths (mm) and indices: I 3.15,
9; II 2.16, 14; III 1.40, 26; IV 2.34, 16.
Ventral spination: tibiae I 2-2-0, II, III,
IV 2-2-2; metatarsi I, II 2-0-0, III, IV 2-
2-2.

Palpus as in Figures 114, 115.

Female (Santa Barbara Co., California).
Coloration as in male of Aysha camhridgei.

Total length 5.72 mm." Carapace 2.09
mm long, 1.66 mm wide; cephalic width
1.04 mm, clypeus height 0.05 mm. Eyes:
diameters (mm): AME 0.08, ALE 0.09,
PME 0.10, PLE 0.10; anterior eye row
0.50 mm long, recurved; posterior eye row
0.67 mm long, procurved; MOQ length
0.25 mm, front width 0.23 mm, back width
0.33 mm; eye interdistances (mm): AME-
AME 0.07, AME-ALE 0.05, PME-PME
0.14, PME-PLE 0.11, ALE-PLE 0.06.

Sternum 1.30 mm long, 0.85 mm wid(\
Chelicerae 0.70 mm long with 3 promar-
ofinal teeth and 8 retromariiiinal dcMiticles.

258 Bulletin Museum of Comparative Zoology, Vol. 146, No. 4

Abdomen 4.00 mm long, 2.38 mm wide.
Epigastric furrow 0.70 mm from tracheal
spiracle, spiracle 1.91 mm from base of
spinnerets.

Tibial lengths (mm) and indices: I 1.57,
14; II 1.26, 17; III 0.86, 29; IV 1.64, 15.
Ventral spination as in male except tibiae
II, III 1-1-0 and IV 1-1-2 and metatarsi
III 2-0-2.

Epigynum as in Figure 126, internal
genitalia as in Figure 127.

Natural history. Mature males have been
taken from late April through early Sep-
tember, mature females year-round. Speci-
mens have been taken on poplars, in fields,
and in houses.

Distribution. California west to Utah,
south to southern Mexico (Map 4).

Aysha velox (Becker)

Map 4; Figures 110, 111, 122, 125

Anijphaena vcIox Becker, 1879, Ann. Ent. Soc.
Belgique, 22: 83, pi. 2, figs. 5-7 ( 5 ). Female
holotype from Pascagoula, Mississippi, should
be in the Institute Royal des Sciences Naturelles
de Belgique but could not be located there by
Mr. J. Kekenbosch in 1971; lost, presumed de-
stroyed. Banks, 1904, Proc. Acad. Nat. Sci.
Philadelphia, 56: 123, pi. 8, fig. 19, $.

Amjphaena floridana Banks, 1896, Trans. Amer.
Ent. Soc, 23: 63. Female holotype from Lake
Worth, Florida, in MCZ, examined.

Aysha orlandensis Tullgren, 1901, Bih. Svenska
Akad., 27: 19, fig. 4 ( ? ). Female holotype
from Orlando, Florida, in Uppsala Univ. Zool.
Mus., examined. Bryant, 1931, Psyche, 38:
119 (sub Aysha gracilis [sic]). Roewer, 1954,
Katalog der Araneae, 2: 534 (sub Aysha graci-
lis [sic]). Bonnet, 1955, Bibliographia Araneo-
rum, 2: 837 (sub Aysha gracilis [sic]). NEW
SYNONYMY.

Aysha velox, Banks, 1909, Estacion central agrono-
mica de Cuba, Second Report, p. 158. Bryant,
1931, Psyche, 38: 119, pi. 7, fig. 14, pi. 8, fig.
34, $ , 5 . Roewer, 1954 Katalog der Araneae
2: 534. Bonnet, 1955, Bibliographia Araneo-
rum, 2: 839.

Chiracanthium falculum Chamberlin, 1925, Bull.
Mus. Comp. Zool., 67: 220. Male holotype from
Sebastian, Florida, in MCZ, examined.

Diagnosis. Aysha velox is a distinctive
species easily recognized by its short retro-

lateral tibial apophysis and its lack of a
ventral tibial apophysis (Fig. Ill) and the
embolus' not being restricted to the distal
half of the palpal bulb (Fig. 110). The
coiled internal ducts of females (Fig. 125)
are diagnostic.

Male (Alachua Co., Florida). Colora-
tion as in Aysha gracilis except that the
abdomen lacks dark markings.

Total length 7.31 mm. Carapace 3.45
mm long, 2.52 mm wide, cephalic width
1,51 mm, clypeus height 0.13 mm. Eyes:
diameters (mm): AME 0.15, ALE 0.14,
PME 0.13, PLE 0.15; anterior eye row
0.80 mm long, recurved; posterior eye row

1.01 mm long, procurved; MOQ length
0.42 mm, front width 0.38 mm, back width
0.47 mm; eye interdistances (mm): AME-
AME 0.09, AME-ALE 0.09, PME-PME
0.22, PME-PLE 0.18, ALE-PLE 0.05.

Sternum 1.92 mm long, 1.28 mm wide.
Chelicerae 1.58 mm long with 4 promar-
ginal teeth and 8 retromarginal denticles.

Abdomen 4.14 mm long, 1.76 mm wide.
Epigastric furrow 0.31 mm from tracheal
spiracle, spiracle 2.46 mm from base of
spinjierets.

Tibial lengtlis (mm) and indices: I 3.92,
8; II 2.86, 12; III 1.69, 21; IV 2.52, 14.
Ventral spination: tibiae I-IV 2-2-2; meta-
tarsi I, II 2-0-0, III, IV 2-2-2.

Palpus as in Figure 110, 111.

Female (Alachua Co., Florida). Colora-
tion as in male.

Total length 8.42 mm. Carapace 3.96
mm long, 2.88 mm wide; cephalic width
1.87 mm, clypeus height 0.14 mm. Eyes:
diameters (mm): AME 0.15, ALE 0.14,
PME 0.14, PLE 0.14; anterior eye row

1.02 mm long, recurved; posterior eye row
1.31 mm long, procurved; MOQ length
0.48 mm, front width 0.45 mm, back width
0.57 mm; eye interdistances ( mm ) : AME-
AME 0.14, AME-ALE 0.14, PME-PME
0.28, PME-PLE 0.27, ALE-PLE 0.09.

Sternum 2.16 mm long, 1.62 mm wide.
Chelicerae 1.87 mm long with teeth as in
male.

Abdomen 4.50 mm long, 2.41 mm wide.

Spider Family Axyphaexidae • Phi f nick

259

Epigastric furrow 0.36 mm from traclical
spiracle, spiracle 2.48 mm from base of
.spinnerets.

Tibial lengths (mm) and indices: I 3.46,
11; II 2.68,^14; III 1.66, 24; IV 2.79, 14.
Ventral spination as in male.

Epigynum as in Figure 122, internal
genitalia as in Figure 125.

Natural histonj. Mature males and fe-
males have been taken year-round. Speci-
mens have been taken on Casuarina sp.,
Citrus sp., Paurotis sp., Calatuandra sp.,
Pinus sp., Ncluniho sp., and in houses.

Distribution. North Carolina west to Ar-
kansas, south to east Texas and Florida,
Cuba, Haiti, the Dominican Republic and
Bermuda ( Map 4 ) ,

Aysha arunda new species

Map 4; Figures 118, 119, 139, 142

Types. Male holotype, female paratype
from Edinburg, Hidalgo Co., Texas, May
1934 (Mulaik), deposited in AMNH. Male
and female paratypes from Hidalgo Co.,
Texas, deposited in MCZ. The specific
name is an arbitrary combination of letters.

Diagnosis. Aysha arunda is a distinctive
species easily recognized by the restriction
of the embolus to the distal half of the
palpal bulb (Fig. 118) and the triangular
.shape of the epigynum (Fig. 139).

Male (Hidalgo Co., Texas). Coloration
as in Aysha cambridgei.

Total length 6.23 mm. Carapace 3.04
mm long, 2.02 mm wide, cephalic width
1.49 mm, clypeus height 0.10 mm. Eyes:
diameters (mm): AME 0.13, ALE 0.13,
PME 0.14, PLE 0.14; anterior eye row 0.74
mm long, recurved; posterior eye row 0.95
mm long, procurved; MOQ length 0.44
mm, front width 0.34 mm, back width
0.43 mm; eye interdistances (mm): AME-
AME 0.07, AME-ALE 0.06, PME-PME
0.16, PME-PLE 0.19, ALE-PLE 0.07.

Sternum 1.73 mm long, 1.24 mm wide.
Chelicerae 1.62 mm long with 4 promar-
ginal teeth and 8 retromarginal denticles.

Abdomen 3.62 mm long, 1.67 mm wide.

Epigastric Imrow 0.31 mm from tracheal
spiracle, .spiracle 1.75 mm from base of
spinnerets.

Tibial lengths (mm) and indices: I
4.03, 7; II 2.72, 12; III 1.53, 20; IV 2.58, 15.
Ventral spination: tibiae I-IV 2-2-2; meta-
tarsi I, II 2-0-0, III, IV 2-2-2.

Palpus as in Figures 118, 119.

Female (Hidalgo Co., Texas). Colora-
tion as in male of Aysha cambridgei.

Total length 6.59 mm. Carapace 2.99
mm long, 2.23 mm wide, cephalic width
1.33 mm, clypeus height 0.09 mm. Iwes:
diameters (mm): AME 0.11, ALE 6.13,
PME 0.13, PLE 0.13; anterior eye row 0.70
mm long, recurved; posterior eye row 0.91
mm long, procurved; MOQ length 0.40
mm, front width 0.32 mm, back width
0.43 mm; eye interdistances (mm): AME-
AME 0.10, AME-ALE 0.07, PME-PME
0.18, PME-PLE 0.17, ALE-PLE 0.06.

Sternum 1.62 mm long, 1.13 mm wide.
Chelicerae 1.37 mm long with 4 promar-
ginal teeth and 9 retromarginal denticles.

Abdomen 3.76 mm long, 2.12 mm wide.
Epigastric furrow 0.40 mm from tracheal
spiracle, spiracle 2.23 mm from base of
spinnerets.

Tibial lengths (mm) and indices: I 2.65,
12; II 2.00,^6; III 1.22, 27; IV 2.25, 16.
Ventral spination as in male.

Epigynum as in Figure 139, internal
genitalia as in Figure 142.

Natural history. Mature males have been
taken from early May through late Septem-
ber, mature females from early April
through late September. Nothing is known
of the habits of this species.

Distribution. Southern Texas (Map 4).

Oxysoma Nicolet

Oxy.soma NMcolet, 1849, in Gay: Hist. Chili, 10
(3): 511. Type species Oxysonui punctatum
Nicolet, 1849, designated by Simon, 1897, Hist.
Natur. Araign., 2: 100.

Gaycimina Ceitsch, 1935, Anier. Mus. Novitates,
No. 805: 21. Type species by iii()i.()t\p\' Gaijcn-
ninii hritrlwri Certsch, 1935.

Diagnosis. O.xysoma can be (juickly
distintiuished from all other North Ameri-

260 Bulletin Museum of Comparative Zoology, Vol. 146, No. 4

can anyphaenids by the presence of only
two teeth on the cheKceral retromaigin. In
addition, the coloration pattern shown in
Fignre 109 is typical for the genus through-
out its range. Predominantly South Ameri-
can, only one species occurs north of
Mexico. Oxysoma is more closely related
to Aijsha than to Anyphaena or Wulfila.

Description. Total length 5-7 mm. Cara-
pace longer than wide, narrowed in front
to about half its maximum width. Clypeus
height more than twice the anterior median
eye diameter. Posterior median, posterior
lateral and anterior lateral eyes subequal
in size, much larger than anterior medians.
Procurved posterior eye row longer than
recurved anterior row. Median ocular
quadrangle more than twice as wide in
back as in front. Anterior median eyes
separated by their diameter, closer to
anterior laterals than to each other. Pos-
terior medians separated by almost three
times their diameter, closer to posterior
laterals. Anterior laterals separated by
their diameter from posterior laterals.
Sternum longer than wide, unmodified.
Chelicerae with 3 promarginal and 2 re-
tromarginal teeth. Abdomen longer than
wide, tracheal spiracle roughly midway
between epigastric furrow and base of
spinnerets. Leg formula 1423, legs un-
modified. Metatarsi I and II with one
pair of ventral spines. Palpus bulbous, with
elongated conductor and conspicuous em-
bolus. Retrolateral tibial apophysis lacking.
Epigynum on a sclerotized plate. Internal
genitalia with two large spermathecae and
accessory ducts.

Variation. The two males of Oxysoma
cubana known from Arizona are slightly
larger than the eastern specimens. One has
a broken conductor, the other matches the
eastern specimens in genitalic details.

Oxysoma punctatum Nicolet

Oxifsoma punctatum Nicolet, 1849, in Gay: Hist.
Chili, 10(3): 513, pi. 4, fig. 13 (9). Female
holotype from Chile, possibly in Museum Na-
tional d'Histoire Naturelle, Paris, unavailable.

Roewer, 1954, Katalog der Araneae, 2: 544.
Bonnet, 1958, Bibliographia Araneorum, 2:
3269.

Types of this species, type species of
Oxysoma, were unfortunately unavailable
for examination.

Oxysoma cubana Banks
Map 5; Figures 105-109

Oxysoma cubana Banks, 1909, Estacion central
agronomica de Cuba, Second Report, II ( 2 ) :
157, pi. 10, fig. 7 {$). Male holotype from
Havana, Habana, Cuba, in MCZ, examined.
Bryant, 1940, Bull. Mus. Comp. Zool., 86: 435,
pi. 16, figs. 218, 222, pi. 17, fig. 234, $,9.
Kaston, 1948, Bull. Connecticut Geol. Natur.
Hist. Surv., 70: 405.

Gaijennina hritcheri Gertsch, 1935, Amer. Mus.
Novitates, No. 805: 21, figs. 35, 36 ( ? ). Fe-
male holotype from Woods Hole, Massachu-
setts, in AMNH, examined. Kaston, 1948, Bull.
Connecticut Geol. Natur. Hist. Surv., 70: 405.
Roewer, 1954, Katalog der Araneae, 2: 540.
Bonnet, 1957, Bibliographia Araneorum, 2:
1981.

Oxysoma cubanum, Roewer, 1954, Katalog der
Araneae, 2: 543. Bonnet, 1958, Bibliographia
Araneorum, 2: 3268.

Diagnosis. The characters of the genus
distinguish this species from all other
nearctic anyphaenids. The bulbous palp
(Fig. 107) and characteristic epigynum
(Fig. 106), as well as the color pattern
( Fig. 109 ) , are diagnostic. Variation in
this species is discussed above.

Male (Suffolk Co., New York). Total
length 5.22 mm. Carapace 2.68 mm long,
2.14 mm wide, cephalic width 1.08 mm,
clypeus height 0.23 mm, pale yellow with
a median dark band and two submarginal
longitudinal rows of dark spots. Eyes:
diameters (mm): AME 0.05, ALE 0.11,
PME 0.09, PLE 0.08; anterior eye row
0.49 mm long, slightly recurved; posterior
eye row 0.76 mm long, procurved; MOQ
length 0.26 mm, front width 0.19 mm, back
width 0.44 mm; eye interdistances (mm):
AME-AME 0.06, AME-ALE 0.05, PME-
PME 0.26, PME-PLE 0.16, ALE-PLE 0.12.

Sternum 1.42 mm long, 0.95 mm wide,
pale yellow with translucent border. Che-
licerae 0.65 mm long, pale yellow with 3

Spider Family Anyphaenidae • Platnick 261

ft-,--.'

vl'^

\ /
V

i^^^^^^^^

Map 5. Distributions of Oxysoma cubana and Teudis mordax.

promarginal and 2 retromarginal teeth.
Labium and endites pale yellow. Endites
not in^'aginated at middle.

Abdomen 2.97 mm long, 1.39 mm wide,
pale white with a median longitudinal
dark band, venter pale. Epigastric furrow
0.58 mm from tracheal spiracle, spiracle
1.13 mm from base of spinnerets.

Legs pale yellow with scattered dark
markings, unmodified. Tibial lengths (mm)
and indices: I 2.09, 17; II 1.78, 20; III 1.35,
26; IV 2.07, 14. Ventral spination: tibiae
I, II 2-2-2, III 1-2-2, IV 2-2-2; metatarsi
I, II 2-0-0, III 2-0-2, IV 2-2-2.

Palpus as in Figures 105, 107.

Female (Barnstable Co., Massachusetts).
Coloration as in male.

Total length 5.90 mm. Carapace 2.66
mm long, 1.91 mm wide, cephalic width
1.01 mm, clypeus height 0.14 mm. Eyes:
diameters (mm): AME 0.06, ALE 0.12,
PME 0.09, PLE 0.10; anterior eye row 0.46
mm long, recurved; posterior eye row 0.74
mm long, procurved; MOQ length 0.34
mm, front width 0.19 mm, back width
0.41 mm; eye interdistances (mm): AME-
AME 0.05, AME-ALE 0.04, PME-PME
0.25, PME-PLE 0.13, ALE-PLE 0.11.

Sternum 1.42 mm long, 0.86 mm wide.

Chelicerae 0.86 mm long with teeth as in
male.

Abdomen 3.71 mm long, 1.71 mm wide.
Epigastric furrow 1.19 mm from tracheal
spiracle, spiracle 1.13 mm from base of
spinnerets.

Tibial lengths (mm) and indices: I 1.55,
23; II 1.46, 23; III 1.10, 21; IV 1.42, 18.
Ventral spination as in male except tibiae
III 0-2-2 and metatarsi IV 0-0-0.

Epigynum as in Figure 106, internal
genitalia as in Figure 108.

Natural history. Mature males have been
taken from late May through late August,
mature females from late March through
late August. One specimen was taken in a
pitfall trap, but the habits of this wide-
spread but rare species are unknown.

Distrihution. Southeastern Arizona to
Michigan, Massachusetts, Florida, and
Cuba (Map 5).

Teudis O. P. -Cambridge

Teudis O. P.-Canibridge, 1896, Biologia Centrali
Americana, Aran., 1 : 198. Type species Teudis
gentilis O. P.-Cambridge, 1896 (= Teudis
geminus Petrunkevitch, 1911), designated by
F.O. P.-Canibridge, 1900, ihid., 2: 100.

Dia<s,nosis. The limits of this genus are
not known with certaintv. One of the

262 Biilletin Museum of Comparative Zoology, Vol. 146, No. 4

species here placed in Teudis was included
in the genus by its original author, O. P.-
Cambridge. As here construed, the genus
is a large one, including a large number of
neotropic species with diverse genitalia.
The following somatic characters are diag-
nostic: the carapace is only slightly nar-
rowed in front, is reddish brown, darkest
at the sides, with a shiny, glabrous cephalic
area and without dark paramedian longi-
tudinal bands; the legs are short and thick;
the chelicerae are often produced forward.
The affinities of the genus are uncertain,
but it is probably more closely related to
Aijsha and Oxysoma than to Anyphoena
or Wulfila.

Description. Total length 3-5 mm. Cara-
pace longer than wide, narrowed in front
to two-thirds to four-fifths of its maximum
width. Clypeus height roughly equal to
anterior median eye diameter. Posterior
median, posterior lateral and anterior
lateral eyes subequal in size, slightly larger
than anterior medians. Procurved posterior
eye row longer than recurved anterior row.
Median ocular quadrangle longer than
wide in front, wider in back than long.
Anterior median eyes separated by their
diameter, closer to anterior laterals than
to each other. Posterior medians separated
by 1.5 times their diameter, closer to pos-
terior laterals. Anterior laterals separated
by their radius from posterior laterals.
Sternum longer than wide, unmodified.
Chelicerae often produced forward, with
3-4 promarginal teeth and 4-6 retromar-
ginal denticles. Abdomen longer than wide,
tracheal spiracle slightly closer to epigastric
furrow than to base of spinnerets. Leg
formula 1423, legs unmodified. Metatarsi
I and II with one or two pairs of ventral
spines. Palpus with a sharply pointed
median apophysis, short conductor and
conspicuous curving embolus. Retrolateral
tibial apophysis spur-like, retrolateral pa-
tellar apophysis sometimes present. Epigy-
num with conspicuous openings; two
simple spermathecae.

Variation. Teudis rnordax is a polymor-

pliic species. Two forms of males occur,
one in which the chelicerae are similar to
those of females, averaging 1.0 mm in
length and one in which the chelicerae are
greatly elongated, averaging 2.3 mm in
length. This polymorphism occurs in both
areas from which adequate population
samples exist, the southeastern United
States and Panama. The proportion of
males with long chelicerae is about one
in five. The genitalia are identical in
both forms. The paratype male of Gayenna
absohita from Baja California, a synonym,
has normal chelicerae; the holotype male
of Teudis rnordax from Guerrero, Mexico,
has elongate chelicerae. The California
population is unfortunately known only
from females, which are slightly larger
than those from other parts of the range.
The special uses, if any, of the long che-
licerae are unknown.

Key to Species

la. Metatarsi I and II with two pairs of ventral
spines. Chelicerae produced forward. Leg
segments uniform in color. Palpus without a
retrolateral patellar apophysis (Fig. 131).
Epigynum as in Fig. 132 — rnordax

lb. Metatarsi I and II with one pair of ventral
spines. Chelicerae not produced forward. Fem-
ora much darker than other leg segments.
Palpus with a retrolateral patellar apophysis
(Fig. 128). Epigynum as in Fig. 129 . calcar

Teudis gentilis O. P.-Cambridge
Figure 146

Teudis gentilis O. P.-Cambridge, 1896, Biologia
Centrali Americana, Aran., 1: 199, pi. 25, fig.
G { S ). Male holotype from Coban, Guatemala,
in BMNH, examined.

Teudis geminus Petrunkevitch, 1911, Bull. Amer.
Mus. Natur. Hist., 29: 516, nom. nov. for T.
gentilis, possibly preoccupied by Amjphaena
gentilis Keyserling, 1891. Roewer, 1954, Kata-
log der Araneae, 2: 548. Bonnet, 1959, Biblio-
graphia Araneorum, 2: 4366.

This species, type species of Teudis, is
genitalically close to several species from
Panama described by Chickering in the
genus Silhis and is somatically similar to
the species here included in Teudis.

Spider Family Anyphaenidae • Platnick 263

Teudis mordax (O. P.-Cambridge)
Map 5; Figures 131-133

Dclozcugma mordax O. P.-Caiiibridge, 1896, Bio-
logia Centrali Americana, Aran., 1: 182, pi. 22,
fig. 11 ( c5 ). Male holotype from Omiltemi,
Gnerrero, Mexico, in BMNH, examined.

Teudis mordax, O. P.-Cambridge, 1896, Biologia
Centrali Americana, Aran., 1: 198. Roewer,
1954, Katalog der Araneae, 2: 519. Bonnet,
1959, Bibliographia Araneorum, 2: 4368.

Anypliacna fragilis Banks, 1897, Canad. Ent., 29:
194. Female holotype from Jacksonville, Flor-
ida, in MCZ, examined. Bryant, 1931, Psyche,
38: 114, pi. 8, fig. 32, 9. Roewer, 1954, Kata-
log der Araneae, 2: 527. Bonnet, 1955, Biblio-
graphia Araneornm, 2: 344. NEW SYNONYMY.

Gayenna parvtda Banks, 1899, Proc. Ent. Soc.
Washington, 4: 191. Female holotype from
Shreveport, Louisiana, in MCZ, examined.

Gat/eniia ah.soluta Chamberlin, 1924. Proc. Cali-
fornia Acad. Sci., 12: 661, figs. 103, 104 {$,
9 ). Female holotype, male allotype from Con-
cepcion Bay, Baja California, in California
Academy of Sciences. Male and female para-
types from same locality in MCZ, examined.
Roewer, 1954, Katalog der Araneae, 2: 535.
Bonnet, 1957, Bibliographia Araneorum, 2:
1976. NEW SYNONYMY.

Amiphaena laticeps Bryant, 1931, Psyche, 38: 108,
pi. 6, fig. 4, pi. 8, fig. 24 ( 5, 9 ). Male holo-
type, female allotype from Thompson's Mills,
Jackson Co., Georgia, in MCZ, examined. Roe-
wer, 1954, Katalog der Araneae, 2: 529. Bon-
net, 1955, Bibliographia Araneorum, 2: 345.
NEW SYNONYMY.

Silliis coloratus Chickering, 1937, Pap. Michigan
Acad. Sci., 22: 548, pi. 58, fig. 10, pi. 59, figs.
23, 32 ( 9 ). Female holotype from Barro Colo-
rado Island, Panama Canal Zone, in MCZ, ex-
amined. Roewer, 1954, Katalog der Araneae,
2: 545. Bonnet, 1958, Bibliographia Araneorum,
2: 4048. NEW SYNONYMY.

Amiphacna harrowsi Chamberlin and Ivie, 1946,
Bull. Univ. Utah, 36: 9, fig. 12 ( 9 ). Female
holotype from Fort Myers, Florida, in AMNH,
e-xamined. Roewer, 1954, Katalog der Araneae,
2: 524. NEW SYNONYMY.

Teudis fragilis, Barnes, 1953, Amer. Mus. Novi-
tates, No. 1632: 18.

Diagnosis. Teudis mordax may be dis-
tinguished from all other anyphaenids in
America north of Mexico by the chelic-
erae, which project forward. The shape
of the palpal median apophysis (Fig.
131) and the epigynum (Fig. 132) are
also diagnostic. Variation in this species
is discussed above.

Male (Sarasota Co., Florida). Total
length (exclusive of chelicerae) 3.67 mm.
Carapace 1.79 mm long, 1.31 mm wide,
cephalic width 0.99 mm, clypeus height
0.05 mm, light reddish brown, darkest at
sides, cephalic area shiny, glabrous. Eyes:
diameters (mm): AME 0.07, ALE 0.09,
PME 0.08, PLE 0.09; anterior eye row 0.50
mm long, recurved; posterior eye row 0.62
mm long, procurved; MOQ length 0.27
mm, front width 0.22 mm, back width 0.31
mm; eye interdistances (mm): AME-AME
0.07, AME-ALE 0.05, PME-PME 0.14,
PME-PLE 0.13, ALE-PLE 0.04.

Sternum 1.04 mm long, 0.74 mm wide,
pale yellow, darker around borders. Che-
licerae 1.00 mm long with 3 promarginal
teeth and 6 retromarginal denticles, dark
orange-brown. Labium and endites light
orange-brown. Endites slightly invaginated
at middle.

Abdomen 2.00 mm long, 1.13 mm wide,
pale white with transverse rows of dark
spots, venter pale. Epigastric furrow 0.50
mm from tracheal spiracle, spiracle 0.65
mm from base of spinnerets.

Legs light orange-brown, unmodified.
Tibial lengths (mm) and indices: I 1.60,
12; II 1.33, 14; III 0.80, 24; IV 1.26, 18.
Ventral spination: tibiae I 2-2-0, II 1-2-0,
III 1-2-2, IV 2-2-2; metatarsi I, II 2-2-0,
III 2-1-2, IV 2-2-2.

Palpus as in Figure 131.

Female (Sarasota Co., Florida). Colora-
tion as in male.

Total length 3.86 mm. Carapace 1.71
mm long, 1.39 mm wide, cephalic width
1.06 mm, clypeus height 0.09 mm. Eyes:
diameters (mm): AME 0.08, ALE 6.09,
PME 0.09, PLE 0.09; anterior eye row
0.55 mm long, straight; posterior eye row
0.70 mm long, procurved; MOQ length
0.22 mm, front width 0.23 mm, back width
0.34 mm; eve interdistances (mm): AME-
AME 0.06,' AME-ALE 0.07, PME-PME
0.16, PME-PLE 0.14, ALE-PLE 0.05.

Sternum 0.97 mm long, 0.76 mm wide.
Chelicerae 0.82 mm long with teeth as in
male.

264 Bulletin Museum of Comparative Zoology, Vol. 146, No. 4

139

140

138

142

Spider Family Anyphaenidae • Platnich 265

145

146

Figures 144-145. Epigyna, ventral view,
bridge. 145. Aysha prospers KeyserWng.

Plate 12
Figure 146. Left palp, ventral view. 144. Wulfila pallidus O. P.-Cam-
146. Teudis gentilis O. P. -Cambridge.

Abdomen 2.20 mm long, 1.44 mm wide.
Epigastric furrow 0.68 mm from traclieal
spiracle, spiracle 0.85 mm from base of
spinnerets.

Tibial lengths (mm) and indices: I
1.13, 18; II 0.97, 21; III 0.67, 30; IV 1.06,
22. Ventral spination as in male except
tibiae III, IV 1-1-2 and metatarsi IV 2-1-2.

Epigynum as in Figure 132, internal
genitalia as in Figure 133.

Natural history. Mature males and fe-
males have been taken year-round. Speci-
mens have been taken on loblolly pine and
fall webwonn nests.

Distribution. Southern North America,
from Maryland south to Florida, Mexico,
and Panama, west to California and Baja
California (Map 5).

Teudis calcar (Bryant), new combination
Map 3; Figures 128-130

Amjphaena calcar Bryant, 1931, Psyche, 38: 107,
pi. 6, fig. 3 { $ ). Male holotype from Diinedin,
Florida, in MCZ, examined. Roewer, 1954, Kat-
alog der Araneae, 2: 524. Bonnet, 1955, Bibli-
ograpliia Araneorum 2: 342.

Auyphacna scJiwarzi Certsch, 1933, Amer. Mus.
Novitates, No. 637: 10, fig. 12 (9). Female
holotype from Brown.sville, Texas, in AMNH,
examined. Roewer, 1954, Katalog der Araneae,
2: 529. Bonnet, 1955, Bibliographia Araneorum,
2: 347. NEW SYNONYMY.

Diagnosis. Teudis calcar may be dis-
tinguished from all other anyphaenids in
America north of Mexico by the retro-
lateral patellar apophysis of males (Fig.
128) and the epigynum of females (Fig.
129).

Plate 11
Figures 128, 131, 134, 135. Left palpi, ventral view. Figures 129, 132, 136, 138-140. Epigyna, ventral view. Fig-
ures 130, 133, 137, 141-143. Internal genitalia, dorsal view. 128 130. 7eud/s ca/car (Bryant). 131 133. Teudis
mordax (O. P. -Cambridge). 134. Anyphaena accentuata (Walckenaer). 135 137. Anyphaena aperta (Banks).
138, 141. Aysha Cambridge! Bryant. 139, 142. Aysha arunda new species. 140. 143. Aysha gracilis (Hentz).

266 Bulletin Museum of Comparative Zoology, Vol. 146, No. 4

Male (Hidalgo Co., Texas). Coloration
as in Teiidis morclax except that the abdo-
men is uniformly light gray and the femora
are much darker than the other leg seg-
ments.

Total length 3.78 mm. Carapace 1.76
mm long, 1.42 mm wide, cephalic width
0.95 mm, clypeus height 0.10 mm. Eyes:
diameters (mm): AME 0.07, ALE 0.11,
PME 0.11, PLE 0.11; anterior eye row 0.43
mm long, recurved; posterior eye row 0.64
mm long, procurved; MOQ length 0.25
mm, front width 0.22 mm, back width 0.35
mm; eye interdistances (mm): AME-AME
0.07, AME-ALE 0.03, PME-PME 0.14,
PME-PLE 0.11, ALE-PLE 0.03.

Sternum 1.03 mm long, 0.83 mm wide.
Chelicerae 0.75 mm long with 3 promar-
ginal teeth and 4 retromarginal denticles.

Abdomen 2.30 mm long, 1.33 mm wide.
Epigastric furrow 0.49 mm from tracheal
spiracle, spiracle 0.81 mm from base of
spinnerets.

Tibial lengths (mm) and indices: I 1.23,
17; II 1.01, 19; III 0.67, 30; IV 1.12, 21.
Ventral spination: tibiae I, II 1-2-0, III,
IV 1-1-2; metatarsi I, II 2-0-0, III, IV
2-2-2.

Palpus as in Figure 128.

Female (San Patricio Co., Texas). Color-
ation as in male.

Total length 4.97 mm. Carapace 1.94
mm long, 1.55 mm wide, cephalic width
1.00 mm, clypeus height 0.09 mm. Eyes:
diameters (mm): AME 0.07, ALE 0.11,
PME 0.11, PLE 0.11; anterior eye row 0.52
mm long, recurved; posterior eye row 0.72
mm long, procurved; MOQ length 0.27
mm, front width 0.23 mm, back width 0.38
mm; eye interdistances (mm): AME-AME
0.09, AME-ALE 0.04, PME-PME 0.17,
PME-PLE 0.11, ALE-PLE 0.06.

Sternum 1.16 mm long, 0.96 mm wide.
Chelicerae 0.72 mm long with 4 promar-
ginal teeth and 4 retromarginal denticles.

Abdomen 3.13 nmi long, 1.98 mm wide.
Epigastric furrow 0.70 mm from tracheal
spiracle, spiracle 1.16 mm from base of
spinnerets.

Tibial lengths (mm) and indices: I 1.22,
20; II 1.08, 22; III 0.86, 29; IV 1.35, 21.
Ventral spination as in male save metatarsi
III 2-0-2.

Epigynum as in Figure 129, internal
genitalia as in Figure 130.

Natural history. Mature males have been
taken from early April through mid-July,
mature females from late May through
mid-July. Nothing is known of the habits
of this species.

Distribution. Florida and Texas (Map
3).

REFERENCES

Bryant, E. B. 1931. Notes on North American
Anyphaeninae in the Museum of Comparative
Zoology. Psyche, 38: 102-126.

Chickering, a. M. 1937. Anyphaenidae of
Barro Colorado Island, Panama Canal Zone.
Pap. Michigan Acad. Sci., Arts, Letters, 22:
541-561.

. 1940. New Anyphaenidae from Pan-
ama with notes on known species. Trans.
Amer. Micros. Soc, 59: 78-122.

FoRSTER, R. R. 1970. The Spiders of New
Zealand. Part III. Otago Mas. Bull., 3:
1-184.

Hickman, V. V. 1949. Tasmanian littoral spi-
ders with notes on their respiratory systems,
habits and taxonomy. Pap. Proc. Roy. Soc.
Tasmania, 1948: 31-43.

Krombein, K. V. 1967. Trap-nesting wasps
and bees: life histories, nests, and associates.
Smithsonian Press, Washington, 570 pp.

Lamoral, B. H. 1968. On the ecology and
habitat adaptations of two intertidal spiders.
Ann. Natal. Mus., 20: 151-193.

Lehtinen, p. T. 1967. Classification of the
cribellate spiders and some allied families.
Ann. Zool. Fennici, 4: 199-468.

Levi, H. W. 1967. Adaptations of respiratory
systems of spiders. Evolution, 21: 571-583.

Marx, G. 1890. Catalog of the described
Araneae of temperate North America. Proc.
U. S. Nat. Mus., 12: 497-594.

Petrunkevitch, a. 1930. The spiders of Porto
Rico. Part III. Trans. Connecticut Acad.
Arts Sci., 31: 1-191.

Platnick, N. I. 1971. The evolution of court-
.ship behaviour in spiders. Bull. British
Arachn. Soc, 2: 40-47.

Warren, L. O., W. B. Peck and M. Tadic.
1967. Spiders associated with the fall web-
worm, Hyphantria cunea ( Lepidoptera :
Arctiidae). J. Kansas Ent. Soc, 40: 382-
395.

us ISSN 0027-4100

SuLletln OF THE

Museum of

Comparative

Zoology

The Orb-weaver Genus Zygiella
(Araneae: Araneidae)

HERBERT W. LEVI

HARVARD UNIVERSITY

CAMBRIDGE, MASSACHUSETTS, U.S. A

VOLUME 146, NUMBER 5
21 NOVEMBER 1974

PUBLICATIONS ISSUED

OR DISTRIBUTED BY THE

MUSEUM OF COMPARATIVE ZOOLOGY

HARVARD UNIVERSITY

Breviora 1952-

BULLETIN 1863-

Memoirs 1864-1938

JoHNSONiA, Department of Mollusks, 1941-

OccAsioNAL Papers on Mollusks, 1945-

SPECIAL PUBLICATIONS.

1. Whittington, H. B., and E. D. I. Rolfe (eds.), 1963. Phylogeny and
Evolution of Crustacea. 192 pp.

2. Turner, R. D., 1966. A Survey and Illustrated Catalogue of the Teredini-
dae (Mollusca: Bivalvia). 265 pp.

3. Sprinkle, J., 1973. Morphology and Evolution of Blastozoan Echinoderms.
284 pp.

4. Eaton, R. J. E., 1974. A Flora of Concord. 250 pp.

Other Publications.

Bigelow, H. B., and W. C. Schroeder, 1953. Fishes of the Gulf of Maine.
Reprint.

Brues, C. T., A. L. Melander, and F. M. Carpenter, 1954. Classification of
Insects.

Creighton, W. S., 1950. The Ants of North America. Reprint.

Lyman, C. P., and A. R. Dawe (eds.), 1960. Symposium on Natural
Mammalian Hibernation.

Peters' Check-list of Birds of the World, vols. 2-7, 9, 10, 12-15.

Proceedings of the New England Zoological Club 1899-1948. (Complete
sets only.)

Publications of the Boston Society of Natural History.

Price list and catalog of MCZ publications may be obtained from Publications
Office, Museum of Comparative Zoology, Harvard University, Cambridge, Massa-
chusetts, 02138, U.S.A.

THE ORB-WEAVER GENUS ZYGIELLA (ARANEAE: ARANEIDAE)

HERBERT W. LEVI

AusTRACT. The fifteen known species of Zy-
giclla are redescribed and illustrated. Eleven are
Palearctic, one Holarctic, one Nearctic, one Ori-
ental and one Asiatic. Zijgiella atrica and Z. x-
nutata are introduced to North America from
Europe; Z. x-notata has probably been spread by
man, especially to temperate South America.

Zygielhi differs from Araneus in having
the eye region more compact and in liaving
characteristic markings on the dorsoven-
trally flattened, oval abdomen. Also the
epigynal scape, when present, is smooth
and there is a projection or scnlptnring of
the male tegulum. Zijgiella exhibits di-
verse modifications of the paracymbinm;
the paracymbinm is simple and hook-
sliaped in Araneus. In addition, the web
of Zijgiella has an open sector, whereas
that of Araneus is complete.

Despite their diverse genitalia, the spe-
cies of Zijgiella appear closely related.

A few measurements of differently sized
individuals of Japanese Zijgiella sia, pre-
sumably individuals that matured in differ-
ent instars, indicate that in females
growth in leg length is proportional to
carapace length, and distance from lateral
eyes to median eyes increases at a slightly
slower rate than growth of carapace (the
eye diameter presumably grows slightly,
but less than the carapace). Larger males
may have relatively longer legs. There
were hardly any differences in size of the
genitalia.

INTRODUCTION

Spiders of the families Araneidae and
Linyphiidae have far more complicated
genitalia than spiders of other families.
The temptation is to use these excellent
species-separating characters to group the
species into genera. That generic group-
ing has been a problem is well-known.
Simon, perhaps the foremost 19th century
arachnologist, synonymized most araneid
genera in Araneus (1895). I believe this
was an action of despair by a specialist
who minimized the importance of geni-
talia. On the other hand. Archer (1951a,
b) more recently grouped the species into
genera mainly on the basis of the shape of
the median apophysis, one of numerous
sclerites in the male palpus. Neither ex-
treme is satisfactory.

The difficulty of using only genitalia in
separating genera is perhaps best demon-
strated in Mangora ( Levi, in preparation ) .
The high thorax and the feathered tricho-
bothria, a sense organ on the third tibia,
make it easy to separate Mangora species
on first inspection from species belonging
to other genera. The genitalia of the lui-
merous species show great diversity. All
indications are that the body shape and
sense organs reflect close relationship of
the many species and that it is a mono-
phyletic group despite the variability in
the genitalia.

But are there other araneid genera
whose species, while readily recognized as
belonging together, show diversity in geni-

BuU. Mus. Comp. Zool., 146(5) : 267-290, November, 1974 267

268 Bulletin Museum of Comparative Zoology, Vol. 146, No. 5

talia? Zygiella species, most of which are American species and those from other parts

Palearctic, have females with and without of the world, I have to thank many curators

an epigynal scape; males with and without and colleagues for their help in making this

a palpal terminal apophysis. Despite this study possible; as I know from my own

it seems that the species included in Zy- experience as curator, it is very time-con-

giella are closely related. They are easily suming to locate obscure specimens in the

recognized by their body shape as belong- large collections. I wish to thank A.

ing to Zygiella, even though it is not easy Timothea da Costa of the Museu Nacional,

to characterize those features that make Rio de Janeiro; M. Grasshoff, Senckenberg

them distinct. Also the different species Museum, Frankfurt; J. Gruber, Natur-

make a similar web with a vacant sector, historisches Museum, Vienna; M. Hubert

And when the seemingly diverse genitalia of the Museum National d'Histoire Na-

are carefully studied, similarities are found turelle, Paris; C. E. O'Riordan, National

that separate the species from those of Museum of Ireland, Dublin; F. H. Rindge

other genera. and N. Platnick of the American Museum

Modern araneologists subscribe to widely of Natural History; J. O. Hiising and R.

differing concepts of what is a genus, Piechocki of the Martin Luther Univer-

Europeans more than Americans tending sitiit of Halle (Saale) of the German Dem-

to fragment genera in the interest of show- ocratic Republic; J. Proszynski and W.

ing relationships, though the result is just Star^ga of the Institute of Zoology, Polska

the opposite. They forget the usefulness to Akademia Nauk, Warszawa; G. Schmidt;

themselves in having all oak trees in the E. Tortonese of the Museo Civico de

genus Querciis, all pines in Pinus, and Storia Naturale, Genova; G. C. Varley and

forget that araneologists who study taxon- H. Taylor of the Hope Department of

omy produce knowledge used by colleagues Entomology, Oxford; F. Wanless and D.

in fields other than spider taxonomy. Norman of the British Museum (Natural

As the American species of Zygiella have History); and T. Yaginuma. D. McGrath
recently been revised by Gertsch (1964), and D. McGrath, Jr. were helpful in ob-
an attempt was made to revise the group taining specimens. Lorna R. Levi and Ian
world wide. Gertsch's illustrations do not R. Mackay edited the paper and made
show the palpal sclerites sufficiently dis- helpful comments. The research and pub-
tinctly. (This may be a subjective opinion, lication was in part supported by National
just because I did not illustrate them my- Science Foundation research grant GB-
self. ) Some Asian species that have never 36161.
been illustrated before add new dimen-
sions to the interesting problem of genitalic Zygiella F.O. Pickard-Cambridge

differences in related species. I did not ^ . ^ t t^ i loo^ • ti t^ . ui j

, , -11 XT Zygta C. L. Koch, 1834, m Panzer, Deutscmands

describe new species, partly because I do insekten, Heft 123, 17-19. The type species is

not like to describe new forms but also Aranem- calophylla Walckenaer 1802 (? =

because I suspect that additional Zygiella x-notata Clerck) as the only included species,

species already described may be mis- ^ame Zygia preoccupied by Fabricius 1775 for

placed in wrong genera and families. Be- ZySJucTf. O. Pickard-Cambridge, 1902. Ann.

cause I was not describmg new species, Mag. Natur. Hist. (7)9: 15. New genus "to

I did not make an attempt to borrow large replace Zygia with Z. atrica (C. L. Koch)

unsorted collections of Zygiella from other as the type species," as indicated by original

institutions designation. The name is of feminine gender.

Because of the difficulties in visiting Description. Zygiella, unlike Araneus,

European museums and in borrowing type- has the eye region compact, with the eyes

specimens of both previously described closely spaced. The median ocular area

Orb-weaver Genus Zygiella • Levi

269

Plate 1. Female Zygiella atrica (C. L. Koch) from Nahant, Massachusetts, in the laboratory.

is as long as wide in front, always slightly
narrower behind than in front (Figures 1,
12, 57, 65). The anterior eyes are about
equally spaced. But in larger specimens
( Z. sia ) they are separated as much as two
times their diameter from laterals. The
posterior median eyes are only about one
and one-half times the distance of the
median eye interval from the lateral eyes
(to almost 5 times in large specimens of
Z. sia). The carapace is always consider-
ably wider than the eye region in the eye
area, at least in females (Figures 1, 12).
The anterior median eyes are larger to
much larger than the others. Because eye
distance increases almost proportionally
with length (see below), the eyes of only
tlie smallest Araneus species are closely
spaced.
The chelicerae have three to four teeth

on the anterior margin, about as many on
the posterior, and denticles in the groove
between (Figure 58).

The carapace, in contrast to that of
Araneus, is glabrous brown with very few
hairs. The head region is often darker than
the thorax. The abdomen, unlike that of
Araneus, is oval, widest in the middle,
dorsoventrally flattened, as in species of
Nuctenea, but differs from these by being
mainly black and white, not brownish (in
living as well as in preserved animals).
The dorsal folium is almost symmetrical
anterior to posterior, widest in middle; the
cardiac area, however, generally has a
white area while the posterior end of the
folium is darkest (Plate 1; Figures 1, 12,
26, 70, 84, 92, 95). There may be a median
longitudinal line through the white cardiac
spot (Figure 103). The pattern resembles

270 Bulletin Museum of Comparative Zoology, Vol. 146, \o. 5

that of the theridiid genus Enoplog^iatlia,
but not of other araneids. The venter has
a white longitudinal line on each side en-
closing a median black or pigmentless area
between genital furrow and spinnerets
(Figure 27). The epigynum is variable
with a posterior median depression (Fig-
ures 3, 24, 35) or a scape which is not
wrinkled as is the Araneus epigynal scape
(with few exceptions) (Figures 71, 89, 93,
97, 104, 112). The palpus can be readily
distinguished from an Araneus palpus by
the modified diverse paracymbium (Fig-
ures 7, 13, 20, 30, 42, 56, 68, 86, 101, 110,
120). But unlike most Araneidae the pal-
pus has the tegulum often "vertical" in the
palpus, the long axis parallel to the long
axis of the cymbium (Figures 14, 28, 41,
85), and the tegulum bears unique projec-
tions (Figures 42, 56, 60, 68, 75, 85, 118)
or sculpturing (Figure 100). A terminal
apophysis may be present in some species
(Figures 40, 60), but not in Zijgiella atrica,
Z. keijserlingi, Z. x-notata, and Z. minima
(Figures 7, 30).

The male palpal femur lacks the prox-
imal ventral tooth present in Araneus.
The male palpal patella usually has only
one seta in Zijgiella, two in Araneus. How-
ever, Zijgiella sia may have one or two.
The first coxa does not have a hook and
the tibiae of the first two legs of the male
are not modified. The first patella-tibia
of the female is about 1.2-1.5 carapace
lengths, that of the male 1.5-2.0.

Zijgiella differs from Meta, which also
has a modified paracymbium, in having
more complex genitalia (the sculptured
tegulum and complex median apophysis).
I suspect the genera are related.

Most Zijgiella webs have a vacant sector
on the upper half in the direction of the
retreat, but sometimes complete webs are
made.

Zijgiella species have been confused with
species of Enoplognatha and Steatoda,
both Theridiidae. Steatodas abdomen is
usually purplish brown while that of Zij-
giella is black and white. The male Eno-

plogmitho palpus is of very characteristic
shape, having only a minute paracymbial
hook on the lateral edge of the cymbium,
some distance from the tibia. Enoplogna-
tha and Zijgiella females are difficult to
separate on first impression, but female
Zijgiella have three to four teeth on the
posterior margin of the chelicerae and den-
ticles between the two rows of teeth
(Figure 58), while Enoplognatha females
have only one or two teeth on the posterior
cheliceral margin and lack the denticles in
the groove.

Very few specimens were available for
most species examined here. No attempt
was made to borrow the hundreds of
specimens that are usually available for
studies of American species. It would have
been impossible to obtain large series even
if I had tried.

The vast difference in sizes of the few
Japanese Zijgiella sia available indicates a
taxonomic problem easily overlooked (see
under Z. sia below). The larger the araneid
spider, the farther apart the eyes. The
growth in distance between median and
lateral eyes is almost proportional to
growth of carapace width. A careful study
of proportional or allometric growth of
structures used as taxonomic features in
the family may be worthwhile.

The following species described or
placed in Zijgiella do not belong to it or
the types are lost.

alpina, Zilla, Giebel, 1867. Zeitschr. gesanimt.
Naturwissensch., 30: 434. Female holotype
from La Flegere, Chamonix Valley, Switzerland
I? sicj in the Zoology Dept. of the Martin
Luther Universitiit, Halle, Gennan Democratic
Republic (examined). = Theridion sisyphhim
(Clerck). NEW SNYONYMY.

ancora, Epeira, Krynicki, 1837. Bull. Soc. Imp.
Natur. Moscow, 5: 81 from Russia is Steatoda
hipunctata according to Roewer ( 1955, Kata-
log der Araneae, 26: 1477).

aureola, Zilla, Keyserling, 1884. Verhandl. zool.
hot. Ges. Wien, 33: 652, pi. 21, fig. 4, 9 from
the Amazon in the Museum National d'Histoire
Naturelle, Paris, is a species close to Meta. The
type had been marked "Meta aureola Keys." by
Simon.

Orb-weaver Genus Zygiella • Levi 271

caloi)Jii/lhi, Aranea, Walckenaer, 1802, Faune
Parisienne, 2: 200, doubtful uauio. Roewer
(1942) c'ousideis it to be a senior synonyui of
Z. atiirci (C. L. Koch), Bonnet (1959) a
jmiior s\non\in of Z. x-notata. For purposes of
nonienclatural stabilit\ the synon\iny of Bonnet
should be followed.

cniriiiotata. Zilhi, Pokrovskii, 1904. Zap Imp.
Roussk. C;eogr. Obtch., 41: 300, fig. 25, 25a is
not recognizable, but almost certainly is an Eno-
plo^natlia. The author compares it with an-
other Enuploi^uaOia species.

(k'colorata, Zilla, Kevserling, 1893. Spinnen Ameri-
kas, 4: 306, pi. 15, fig. 226, 9. Male holotype
from Brazil (examined) is a Mangora.

gigans, Zilla, Franganillo, 1913. Broteria, 11:
128. Not recognizable.

guttata, Zilla, Kevserling, 1880. \'erhandl. zool.
bot. Ges. Wien, 30: 551, pi. 16, fig. 3, ?.
Female holotype from Peru ( examined ) is a
Lciicauge.

giniaucn.sis, Zilla, Keyserling, 1880. \'erhandl.
zool. bot. Ges. Wien, 30: 554, pi. 16, fig. 5, $.
Male type from Guyana (examined) is of an
unknown genus, not Zygiella.

mclanocephala, Linyphia, Taczanowski, 1874.
Hor. Soc. Ent. Rossicae 10: 70. Types from
Guyana (examined) are Mangora.

uaioazi, Zilla, Dyal, 1935. Bull. Dept. Zool. Panjab
Univ., 1: 186, pi. 11, fig. 6, pi. 16, fig. 124
from India is an Araneus judging by the illus-
trations.

])iinctata, Zilla, Keyserling, 1893. Spinnen Ameri-
kas, 4: 305, pi. 15, fig. 225, $. Female type
from Brazil, lost.

rogcnhoferi, Zilla, Keyserling, 1877. Verhandl.
zool. bot. Ges. Wien 27: 578, pi. 14, fig. 6, 9.
Female holotype from Brazil (examined) is a
Metazygia.

Key to Males of Zygiella Species

(Males of Z. calyptrata, Z. inconveniens and Z.
melanocrania are not known. )
1 Palpal tibia much longer than cymbium

(Figs. 5, 13) - 2

- Palpal tibia of same length or shorter
than c>nibium 3

2(1) Palpal tibia with a distal bulge (Fig.
13); paracymbium pointed at tip (Fig.
13); median apophysis with two long
spines (Fig. 14); Europe keyserUngi

- Palpal tibia with sides parallel ( Fig.
5); paracymbium rounded at tip (Fig.
5 ) ; median apophysis with short spines
(Fig. 6); North America and Europe
__.. atrica

3(1) Tcgulum of palpus without projection
(other than ctmductor) (Figs. 28, 29);
cosmopolitan x-notata

- Tegulum of palpu:^ with a projection
or sculpturing (other than conductor)
(Figs. 42, 56, 68) 4

4(3) Tegulum in "horizontal" position in
CN'iiibium, its long axis transverse to
that of c>nibium (Fig. 119); projection
with teeth "vertical" and surrounding
conductor (Fig. 119); paracymbium a
hook, barely modified (Figs. 118-120);
Japan sia

- Tegulum in more or less "vertical" po-
sition in cymbium, its long a.xis parallel
to cymbium (Fig. 100); projection not
vertical 5

5(4) Tegulum projection in ventral view in
the shape of a human ear (Fig. 100);
paracymbium square (Fig. 101) thorelli

- Tegulum with simple projection; para-
cymbium not a square (Figs. 68, 75) - 6

6(5) In lateral view tegulum projection al-
most as long or longer than tegulum
width (Figs. 68, 110) 7

- In lateral view tegulum projection
shorter than width of tegulum ( Figs.
75, 86) ._ 8

7(6) Tegulum projection pointed (Fig. 68);
paracymbiiun with a dorsally directed
point (Fig. 68); Cdliiornia.. ..carpenteri

- Tegulum projection truncate ( Fig.
109); paracymbium a ventrally directed
lobe (Fig. 110); Europe stroemi

8(6) Palpus with a j,clerite (? terminal
apophysis) more or less parallel to em-
bolus (in ventral view) in distal part
of palpus (Figs. 41, 55); paracymbium
complex with a notch (Figs. 47-49, 59) 9

- Palpus with no sclerite parallel to em-
bolus (in ventral view) (Figs. 19, 74,
85 ) ; paracymbium without notch
(Figs. 20, 75, 86) _-._ 10

9(8) A sclerite (? tenninal apophysis) longer
than embolus in \entral \iew ( Fig.
41); paracymbium with a distal notch
(Figs. 47-49); eastern Asia, North
America - - — - diapar

- Terminal apophysis shorter tlian em-
bolus in ventral view (Fig. 55); para-
cymbium with a ventral notch ( Fig.
59); Europe .montana

10(8) Base of conductor sitting in a depres-
sion surrounded by a rim, or base of
conductor surrounded by wrinkles
(Figs. 75, 85); Eurasia _„ 11

- Base of conductor not surrounded by
wrinkles or a rim (Fig. 19); Canary
Isl. - - .rninitna

11(10) Base of conductor in a depression sur-
rounded by a rim (Fig. 85) kochi

272 BuUctin Museum of Comparative Zoology, Vol. 146, No. 5

— Base of conductor surrounded only by
wrinkles of the tegulum (Figs. 74,
75) caspica

Key to Females of Zygiella Species

1 Posterior rini of epigynum with a semi-
circular lobe (Fig. 71) caspica

— Posterior rim otherwise 2

2(1) Epigynum with a scape (Figs. 89, 93,

97, 104, 112) 3

— Epigynum without a scape (Figs. 3,
10, 16, 22, 34, 77, 82) 7

3(2) Openings ventral underneath heart-
shaped scape (Fig. 112); Japan sia

— Openings posterior 4

4(3) Scape constricted at base (Figs. 89,

93 ) 5

— Scape not constricted at base (Figs.
97, 104) 6

5(4) Scape heart-shaped, slightly longer

than wide (Fig. 89) kochi

— Scape more than twice as long as wide
(Fig. 93); Palestine inconveniens

6(4) Scape a broad lobe with a posterior

median extension (Fig. 97) thorelli

— Scape much longer than wide with
parallel sides (Fig. 104) stroemi

7(2) No depression, openings or sculpturing
visible in ventral view of epigynum, at
most a posterior rim (Figs. 16, 22);
posterior view with two separate open-
ings (Figs. 18, 24) 8

— In ventral view a depression, openings
or sculpturing visible (Figs. 3, 10, 34,
77, 82); no distinct pair of openings in
posterior view 9

8(7) Total length more than 4 mm; epigynum
heavily sclerotized (Figs. 22, 24);
probably cosmopolitan x-notata

— Total length less than 3.5 mm; epigy-
num lightly sclerotized (Figs. 16,
18); Canary Isl. minima

9(7) Semicircular openings bordered on ven-
ter of epigynum (Fig. 82); Burma
nielanocrania

— Openings not so; venter of epigynum
with a median depression or bulge
(Figs. 3, 10, 34, 77) .....10

10(9) A median, posterior, indistinctly bor-
dered, dark depression in ventral view
of epigynum (Fig. 77); fourth coxae
drawn out posteriorly into a spine ( Fig.
80); Malaysia calijptrata

— Posterior depression or bulge distinctly
bordered (Figs. 3, 10, 34); fourth
coxae without a spine; Holarctic region
11

11(10) Median area of epigynum a depression
in ventral view much wider than long
(Fig. 63); California .carpenteri

- Median area at most one and one half
times as wide as long (Figs. 3, 10, 34) 12

12(11) Median area of epigynum a bulging
lobe framed anteriorly only by a lip
(Fig. 10); Europe keyserlingi

- Median area depressed, framed an-
teriorly and laterally (Figs. 3, 34, 52). .13

13(12) Median depressed area extending pos-
teriorly in ventral view (Fig. 3); in
posterior view sides of epigynum lightly
sclerotized and smaller than median
depression (Fig. 4); Europe, North
America atrica

- Median depressed area not projecting
beyond sclerotized area of epigynum in
ventral view (Figs. 34, 38, 52); in pos-
terior view sides of epigynum heavily
sclerotized and sclerotized areas larger
in area than median depression (Figs.
35, 39, 54) 14

14(13) Median area with a constriction as
seen in botli ventral and posterior views
(Figs. 34, 35, 38, 39); Eastern Asia,
North America dispar

- Median area without constriction as
seen in both ventral and posterior views
(Figs. 52, 54); Europe montana

Zygiella atrica (C. L. Koch)
Plate 1; Figures 1-8

EucJuiria atrica C. L. Koch, 1843, Die Arachniden,
12: 103, figs. 1030, 1031, 9, 5. Specimens
came from Germany and France and are pre-
siunably in the museum of the Humboldt Uni-
versitiit, Berlin.

Zilla atrica, - Wiehle, 1931, Tierwelt Deutschlands,
23: 33, figs. 38-40, 9, $.

Zy fiiclla atrica, - Bonnet, 1959, Bibliographia
Araneorum, 2: 4998. Gertsch, 1964, Anier.
Mus. Novitates, No. 2188: 16, figs. 18-20, $,
?.

Diagnosis. This species can be confused
only with Z. keyserlingi. The male differs
from other Zygiella by the long palpal tibia
and, unlike that of Z. keyserlingi, the tibia
has its sides parallel (Figure 5) with setae
equally distributed. The epigynum has a
wide median lobe extending posteriorly in
ventral view; the lobe is depressed in the
middle and the lateral sclerotized areas
are relatively small (Figures 3, 4).

Natural history. This species is common-

Orb-weaver Genus Zygiella • Levi 273

Figures 18. Zygiella atrica (C. L. Koch). 1. Female. 2 4. Epigynum. 2. Posterodorsal view, cleared.
3. Ventral. 4. Posterior. 5-7. Left male palpus. 5. Lateral view. 6. Ventral. 7. Expanded. 8. Eye region
and chelicerae of female.

Figures 9 14. Z. keyserlingi (Ausserer). 9-11. Epigynum. 9. Dorsal, cleared. 10. Ventral. 11. Posterior.
12. Female. 13-14. Male palpus. 13. Lateral. 14. Ventral.

Abbreviations. C, conductor; DH, distal hematodocha; E, embolus; M, median apophysis; P, paracymbium; R,
radix; S, subtegulum; T, tegulum.

Size lines. 0.1 mm, except Figures 1, 8, 12, 1 mm.

est on ocean coasts, but is found in Europe in coastal areas of tlie ocean and of Lake

also in other locations (Wiehle, 1931), on Erie. On the peninsula of Nahant, Massa-

shrubs, junipers, etc. In America the spe- chusetts, it is very common under and be-

cies is certainly introduced and is foiuid tween boulders placed to prevent the road

274 Bulletin Museum of Comparative Zoology, Vol. 146, No. 5

from being washed away by high tides.
As it can be found abundantly among these
boulders year after year, the species can
presumably tolerate the occasional high
waves and salt spray from the ocean. Adult
males and females have been found in this
location in October.

The web (Wiehle, 1931) has more radii
(43-50) than that of Z. x-notata and many
other orb-weavers; most radii are in the
lower half of the web. The free sector
is narrow and the hub has a fine mesh.
There are many frame threads, many close
to the spiral region. The retreat is not as
well built as that of Z. x-notata. Emerton's
picture of the web ( 1902, The Common
Spiders of the United States) shows only
a few radii.

Distribution. Europe; in America from
Nova Scotia to Long Island, New York;
Port Credit, Ontario, and British Columbia
coast. For map see Gertsch ( 1964 ) .

Zygiella keyserlingi (Ausserer)
Figures 9-14

Zilla keyserlingi Ausserer, 1871, Verhandl. zool.
hot. Ges. Wien, 21: 830, pi. 5, fig. 11, ?. Fe-
male holotype from Dalmatia in the Keyserling
collection of the British Museum, Natural His-
tory, not examined. Wiehle, 1931, Tierwelt
Deutschlands, 23: 35, figs. 41, 42, $, $.

Zygiella keyserlingi, - Roewer, 1942, Katalog der
Araneae, 1: 884. Bonnet, 1959, Bibliographia
Araneorum, 2: 5002.

Description. Female from unknown lo-
cality in Europe. Carapace light brown,
cephalic region not much darker. Legs
not banded. Dorsum of abdomen with
characteristic pattern (Figure 12) and
venter with a white line on each side.
Diameter of posterior median eyes 0.9 di-
ameter of anterior medians, anterior laterals
0.8, posterior laterals 0.7 diameter of an-
terior median eyes. Anterior median eyes
one diameter apart, one from laterals. Pos-
terior median eyes one diameter apart, 1.5
from laterals. There are three teeth on the
anterior margin of the chelicerae, three on
the posterior, with denticles between the

margins. Total length 8.0 mm. Carapace
2.7 mm long, 2.3 mm wide. First femur,
3.1 mm; patella and tibia, 4.0 mm; metatar-
sus, 3.1 mm; tarsus, 1.0 mm. Second pa-
tella and tibia, 2.9 mm; third, 1.7 mm;
fourth, 2.6 mm.

Male from unknown locality. Coloration
like that of female. Diameter of posterior
median eyes and of anterior lateral eyes
0.7 diameter of anterior median eyes; that
of posterior lateral eyes 0.6 diameter of an-
terior medians. Anterior median eyes
slightly less than their diameter apart, and
slightly less than their diameter from lat-
erals. Posterior median eyes slightly less
than one diameter apart, 1.5 diameters
from laterals. Total length 6 mm. Cara-
pace 2.9 mm long, 2.0 mm wide. First
femur, 4.1 mm. Second patella and tibia,
3.6 mm; third, 2.0 mm; fourth, 2.9 mm.

Additional female and male specimens
were available from Krivosije, Dalmatia.

Diagnosis. The long palpal tibia is found
also in Z. atrica; however, Z. keyserlingi
has the tibia distally swollen with the
swollen area having more setae than the
basal part ( Fig. 13 ) . The epigynum has
a central bulging area bordered anteriorly
only by a transverse lip (Figures 10, 11).
The species is less pigmented than Z.
atrica.

Distribution. Portugal, Italy, Hungary
to Greece (Bonnet, 1959).

Zygiella minima (Schmidt)
Figures 15-20

Zygiella x-notata minima Schmidt, 1968, Zool.
Beitr., 14: 414, fig. 11, $. Female, male syn-
types in poor physical condition from Esperanza
Forest, Tenerife, Canary Islands, owned by the
author G. Schmidt, but made a\'ailable to me.

Description. Female. Coloration diffi-
cult to determine. Eyes seem about sub-
equal in size. The anterior median eyes
slightly less than their diameter apart,
their radius from laterals. Posterior median
eyes their radius apart, about 0.8 diameter
from laterals. Total length 3 mm. Cara-

Orb-weaver Genus Zygiella • Levi 275

Figures 15-20. Zygiella minima (Schm\6\). 15-18. Epigynum. 15. Ventral, cleared. 16. Ventral. 17. Posterior,
cleared. 18. Posterior. 19-20. Left male palpus. 19. Ventral. 20. Lateral.

Figures 21 31. Z. x-notata (Clerck). 21-25. Epigynum. 21. Ventral, cleared. 22. Ventral. 23. Posterior,
cleared. 24. Posterior. 25. Dorsal, cleared. 26. Female. 27. Female abdomen, ventral. 28-31. Male
palpus. 28. Ventral. 29. Lateral. 30,31. Expanded. 30. Subventral view of bulb. 31. Dorsal view of bulb.

Abbreviations. C, conductor; DH, distal hematodocha; E, embolus; I, stipes; M, median apophysis; P, paracym-
bium; R, radix; T, tegulum; Y, cymbium.

Size lines. 0.1 mm except Figures 26, 27, 1 mm.

276 Bulletin Museum of Comparative Zoology, Vol. 146, No. 5

pace 1.5 mm long. First femur, 2.0 mm;
patella and tibia, 2.4 mm; metatarsus, 1.8
mm; tarsus, 0.7 mm. Second patella and
tibia, 1.7 mm; third, 1.0 mm; fourth, 1.6
mm.

Male. In slightly better physical condi-
tion than female. Eyes subequal in size.
Anterior median eyes their diameter apart,
0.8 diameter from laterals. Posterior me-
dian eyes slightly less than their diameter
apart, their diameter from laterals. Total
length 2.5 mm. Carapace 1.2 mm long,
0.9 mm wide. First femur, 1.6 mm; patella
and tibia, 2.1 mm; metatarsus, 1.6 mm;
tarsus, 0.7 mm. Second patella and tibia,
1.5 mm; third, 0.8 mm; fourth, 1.2 mm.

Diagnosis. Zygiella minima differs from
Z. x-notata in that the female has the epig-
ynum lightly sclerotized and with differ-
ently sized openings in posterior view
(Figures 16, 18); the male has a small
tooth-shaped projection on the face of the
tegulum (Figures 19, 20) absent in Z.
x-notata.

Distribution. Canary Islands.

Zygiella x-notata (Clerck)
Figures 21-31, 57-58

Araneiis x-notatus Clerck, 1758, Aranei Svecici,
46, pi. 2, fig. 5. A Clerck specimen bearing
this name as labeled by Thorell is in the Swed-
ish Museum of Natural History, Stockholm;
not examined.

Zilla bosenhergi Keyserling, 1878, Verhandl. zool.
bot. Ges. Wien, 28; 575, pi. 14, fig. 4, 5, 2 , $.
Female and male syntypes from Uruguay in the
nmseum of the University of Hamburg and
the British Museum (Natural History), exam-
ined. NEW SYNONYMY.

Zilla caUjornica Banks, 1896, J. New York Ent.
Soc, 4: 90. Female holotype from Palo Alto,
California, in the Museum of Comparative Zool-
ogy, examined. Gertsch (in letter, 1957) in-
dicated that Stanford University Museum had
specimens marked types. This spider collection
has since been sent to the Los Angeles County
Museum and was destroyed (C. L. Hogue, per-
sonal communication).

Larinia maulliana Mello-Leitao, 1951, Rev. Chi-
lena Hist. Natur., 51-53: 331, figs. 5, 6, $.
Male holotype from Maullin, Chile, in the
Museu Nacional, Rio de Janeiro, examined.
NEW SYNONYMY.

Zygiella x-notata, - Bonnet, 1959, Bibliographia
Araneorum, 2: 5007. Gertsch, 1964, Amer.
Mus. Novitates, No. 2188, 12, figs. 2, 15-17,
2,5, map.

Diagnosis. The epigynum, unlike that
of Z. minima, is heavily sclerotized. It has
two diagnostic openings seen in posterior
view (Figure 24). The palpus is simple
with the tegulum's long axis parallel to
that of the cymbium (Figures 28-30); the
lack of terminal apophysis (Figures 30,
31) separates males from those of other
species, the lack of a tegulum projection
from males of Z. rninima.

Natural history. Numerous references to
habits and webs can be found in Bonnet
(1959). The web, which has a vacant
sector, has been used in cthological studies.
It is illustrated in Wiehle (1931), figure 37,
and J. Comstock, 1940, The Spider Book,
figure 470. The species is very common
in southern Chile. In the city park of
Osorno, Chile, I found suspended from a
web on a telephone pole a dried, shrivelled,
6-cm long lizard on which a Z. x-notata
had apparently fed (15 March 1965).

Figures 32-50. Zygiella dispar (Kulczynski). 32-39. Epigynum. 32, 33. Posterior view, cleared. 34, 36, 38.
Ventral. 35, 37, 39. Posterior. 32. (Michigan). 33-35. (Mendocino Co., California). 36, 37. British Columbia.
38, 39. (Virginia). 40-50. Left male palpus. 40. Expanded. 41. Ventral. 42. Lateral. 43-46. Embolus and
apophysis. 43. (Alaska). 44. (California). 45. (Manitoba). 46. (Maine). 47-49. Paracymbium. 47. (Alaska).
48. (California). 49. (Manitoba). 50. (Maine).

Figures 51-56. Zygiella montana (C. L. Koch). 51-54. Epigynum. 51. Anterodorsal view, cleared. 52. Ventral.
53. Posterior, cleared. 54. Posterior. 55, 56. Male palpus. 55. Ventral. 56. Lateral.

Abbreviations. A, terminal apophysis; C, conductor; E, embolus; H, hematodocha; I, stipes; M, median apophysis;
T, tegulum.

Scale lirtes. 0.1 mm.

Orb-weaver Genus Zygiella • Levi 277

SSSrSfe'Wiv.i

278 Bulletin Museum of Comparative Zoology, Vol. 146, No. 5

Adult males are found from July until
September on the Pacific coast of North
America.

Distribution. Europe, but probably cos-
mopolitan, carried around the world by
man. It is introduced in America and
found along the Atlantic coast from Maine
to Virginia, the Pacific coast from southern
British Columbia to southern California.
Gertsch ( 1964 ) maps the North American
distribution. It is very common in Chile
and is found in Uruguay and Argentina.

Zygiella dispar (Kulczyhski)
Figures 32-50

Zilla dispar Kulczynski, 1885, Denkschrift. Akad.
Wissenschaften Krakow, 11: 24, pi. 9, fig. 7, 5,
$ . Male type from Kanitchatka, Siberia, in
Polish Academy of Sciences, Warsaw, in poor
physical condition, examined.

Ztjgic'IIa montana, - numerous authors of American
records only.

Zygiella dispar, - Gertsch, 1964, Amer. Mus. Novi-
"tates. No. 2188: 7, figs. 7-10, 9, $.

Zygiella nearctica Gertsch, 1964, Amer. Mus.
Novitates, No. 2188: 4, figs. 3-6, 9, $. Male
holotype from Seba, Alberta, in the American
Museum of Natural History, not examined.
NEW SYNONYMY.

Note. Gertsch (1964) used the name
dispar for the population along the Pacific
coast from Alaska to south-central Cali-
fornia; other specimens he called nearctica.
Gertsch separated Z. nearctica from Z. dis-
par by the following characters: the male
palpus has the apical [= ? subterminal]
apophysis less developed, and has "differ-
ences of the various apophysis"; the female
epigynum has the "fovea" visible from be-
low. The last character is a matter of po-
sition of the epigynum during examination.
Gertsch's figure 7 (dispar) is much more
characteristic of all specimens of the spe-
cies in ventral view than is figure 4 (nearc-
tica), which is the view from slightly pos-
terior. The subterminal apophysis differs
among individuals ( Figures 43-46 ) , as do
the paracymbium (Figures 47-50) and, to
a lesser extent, the median apophysis (not
illustrated). Similarities of the internal

female genitalia also indicate that we have
only one species, not two. California
specimens of the species are the largest;
a male from Alaska was the smallest speci-
men examined.

Gertsch is probably correct in stating
that Z. dispar is distinct from Z. montana
of Europe. Perhaps intermediates will be
found in the vast area between Europe
and Siberia from which no collections have
been examined, but I would not expect
this.

Diagnosis. Females of Z. dispar differ
from those of the related Z. montana in
that the median depression of the epigy-
num has a consti^iction (Figures 35, 39)
in posterior view. The palpus is similar
to that of Z. monta)Ui but differs in that
most sclerotized parts of the palpus have a
different shape and are positioned slightly
differently (Figures 41, 42).

Natural history. The species is found on
trees and rocks (Emerton, 1902, The Com-
mon Spiders, p. 185). Parts of the web
have been illustrated by Emerton (1884,
pi. 40, fig. 2). Emerton collected the
species in the Adirondack Mountains, New
York State, and the White Mountains, New
Hampshire.

Distribution. Kamtchatka, Siberia, and
North America along the Pacific coast,
across Canada, the northern states, south
in the western states, and in the Appala-
chian mountains in the east ( Gertsch, 1964,
fig. 1, a map ) .

Zygiella montana (C. L. Koch)
Figures 51-56, 59-61

Zilla montana G. L. Koch, 1839, Die Arachniden,
6: 146, pi. 536, 537, 9, $. Syntypes probably
from Nassfelder Alpen in Salzburg, Austria, in
the Museum of the Humboldt University, Ber-
lin, not examined. Wiehle, 1931, in Dahl, Tier-
welt Deutschlands, 23: 38, figs. 46-48, 9, $.

Zygiella montana, - Roewer, 1942, Katalog der
Araneae, 1: 886. Bonnet, 1959, Bibliographia
Araneorum, 2: 5003.

Description. Female from Seefeld, Tirol,
Austria. Coloration as in other species.

Orb-weaver Genus Zygiella • Levi 279

Figures 57-58. Zygiella x-notata (Clerck). 57. Eye region and chelicerae. 58. Left chelicera from inside.

Figures 59-61. Z. montana (C. L. Koch). 59. Paracymbium dorsolateral view. 60 61. Left male palpus, ex-
panded. 60. Subventral. 61. Dorsal, cymbium removed.

Figures 62 69. Z. carpenter! Archer. 62-64. Epigynum. 62. Posterior, cleared. 63. Ventral. 64. Posterior.
65. Female. 66. Female abdomen, ventral 67-69. Male palpus. 67. Ventral. 68. Lateral. 69. Paracymbium.

Abbreviations. A, terminal apophysis; C, conductor; E, embolus; M, median apophysis; P, paracymbium; R, radix;
T, tegulum.

Scale lines. 0.1 mm, except Figures 57, 65, 66, 1.0 mm.

280 Bulletin Museum of Comparative Zoology, Vol. 146, No. 5

Secondary eyes 0.8 diameter of anterior
medians. Anterior median eyes 0.7 diam-
eter apart, 0.6 diameter from laterals. Pos-
terior median eyes one diameter apart, 1.2
from laterals. Total length 8.0 mm. Cara-
pace 2.9 mm long, 2.2 mm high. First
femur, 3.0 mm; patella and tibia, 3.7 mm;
metatarsus, 2.9 mm; tarsus, 1.2 mm. Second
patella and tibia, 3.1 mm; third, 1.9 mm;
fourth, 2.6 mm.

Male from Seefeld, Tirol, Austria. Sec-
ondary eyes 0.6 diameter of anterior me-
dians. Anterior median eyes 0.6 diameter
apart, 0.5 diameter from laterals. Posterior
median eyes their diameter apart, 1.6 from
laterals. Total length 6.5 mm. Carapace
3.0 mm long, 2.4 mm wide. First femur,
3.2 mm; patella and tibia, 5.0 mm; metatar-
sus, 4.3 mm; tarsus, 1.0 mm. Second pa-
tella and tibia, 3.9 mm; third, 2.3 mm;
fourth, 3.2 mm.

Diagnods. This European species can
easily be confused with Z. dispar but the
epigynum lacks the constriction of the me-
dian depression in posterior view (Figure
54; the palpus has many sclerites, all
slightly different in shape (Figures 55,
56).

Natural history. According to Wiehle
(1931) this is a mountain species found
in the Alps above 1000 m elevation, most
commonly between 1300-1800 m. The
species is found on buildings, rocks, bark
and branches of trees and shrubs. The web
is similar to that of Z. x-notata with 19-35
radii. The vacant sector is especially wide
and the hub has a rough structure.

Both sexes are mature from June until

September, and may take several years to
mature.

Distribution. European mountains.

Zygiella carpenter! Archer
Figures 62-69

Zygiella carpenteri Archer, 1951, Anier. Mus.
Novitates, No. 1487: 18, fig. 34, 9. Female
holotype from Del Monte Forest, Pacific Grove,
Monterey Co., California, in the American Mu-
seum of Natural History, examined. Gertsch,
1964, Amer. Mus. Novitates, No. 2188: 9, figs.
1, 11-14, 9, 5, map.

Diagnosis. The wide depression of the
epigynum (Figure 63), the long, pointed
projection of the palpal tegulum (Figure
68) and the shape of the paracymbium
(Figures 68, 69) separate the species from
Z. dispar.

Distribution. Sierra mountains of Ore-
gon and Washington. There are also a
few records from near Spokane, Washing-
ton, and the coast of California. There is
a distribution map in Gertsch (1964).

Zygiella caspica (Simon)
Figures 70-75

Zilla caspica Simon, 1889, Verh. zool. bot. Ges.
Wien, 39: 382. Two female, one male syntypes
from Transylvania in the Museum National
d'Histoire Naturelle, Paris, examined.

Zi/fiiella caspica, - Roewer, 1942, Katalog der
Araneae, 1: 883. Bonnet, 1959, Bibliographia
Araneorum, 2: 5002.

Description. Female. Color like that of
other species. Legs not banded, yellowish
brown. Dorsal pattern as is characteristic
in Zygiella (Figure 70), venter with very

Figures 70-75. Zygiella caspica {S\mon). 70. Female. 71,72. Epigynum. 71. Ventral. 72. Posterior. 73-75.
Left male palpus. 73. Mesal. 74. Ventral. 75. Lateral.

Figures 76-80. Z. calyptrata (Workman). 76-78. Epigynum. 76. Dorsal, cleared. 77. Ventral. 78. Posterior.
79. Female. 80. Fourth coxae, ventral.

Figures 81 84. Z. melanocrania (Thorell). 81-83. Epigynum. 81. Ventral, cleared. 82. Ventral. 83. Posterior.
84. Female.

Figures 85-86. Z. kocfii (Thorell), male palpus. 85. Ventral. 86. Lateral.

Scale lines. 0.1 mm except Figures 70, 79, 84, 1.0 mm.

Orb-weaver Genus Zygiella • Levi 281

.:fc-|^.^ vv

282 Bulletin Museum of Comparative Zoology, Vol. 146, No. 5

little black pigment. The posterior median
eyes are slightly smaller than anterior me-
dians, laterals 0.8 diameter of anterior me-
dian eyes. The anterior median eyes are
their radius apart, their radius from lat-
erals. Posterior median eyes their diameter
apart, and slightly more than one diameter
from laterals. Total length 6.5 mm. Cara-
pace 2.4 mm long, 1.9 mm wide. First fe-
mur, 2.8 mm; patella and tibia, 3.6 mm;
metatarsus, 2.6 mm; tarsus, 0.9 mm. Second
patella and tibia, 2.8 mm; third, 1.7 mm;
fourth, 2.3 mm.

Male. Coloration as in female. The eyes
are slightly larger and closer together.
Total length 5.0 mm. Carapace 2.3 mm
long, 1.7 mm wide. First femur, 2.9 mm;
patella and tibia, 4.3 mm; metatarsus, 2.9
mm; tarsus, 1.2 mm. Second patella and
tibia, 2.9 mm; third, 1.7 mm; fourth, 2.3
mm.

Diagnosis. While the short semicircular
scape of the epigynum (Figures 71, 72) is
distinct, the palpus is similar to that of Z.
kochi, but differs in the shape of the tegu-
lum at the base of the conductor and the
terminal apophysis (Figures 73-75).

Distribution. Trans-Carpathian region.

Zygiella calyptrata (Workman)
Figures 76-80

Epeira calyptrata Workman, 1894, Malaysian Spi-
ders, p. 21, plate 21. One female lectotype here
designated and two female paralectotypes from
Singapore in the National Museum of Ireland,
Dublin, examined.

Epeira (Zilla) calyptrata, - Thorell, 1895, Descr.
Catalogue of the Spiders of Burma, p. 188.

Zygiella calyptrata, - Roewer, 1942, Katalog der
Araneae, 1 : 886.

Araneus cahjptratus, - Bonnet, 1955, Bibliogra-
phia Araneorum, 2: 450.

Description. Female lectotype. Carapace
brown; head region very much darker,
glossy. Sternum, legs brown. Dorsum of
abdomen white with black marks (Figure
79). Sides brownish black. Venter gray.
Anterior median eyes much larger than
others. Diameter of posterior median eyes
0.8 diameter of anterior medians; laterals

0.6 diameter of anterior median eyes. An-
terior median eyes their diameter apart,
slightly more than their diameter from lat-
erals. Posterior median eyes slightly less
than their radius apart, 2.5 diameters from
laterals. The fourth coxa has a posterior
distal spine (Figure 80). Total length 4
mm. Carapace 1.7 mm long, 1.3 mm wide.
First femur, 1.5 mm; patella and tibia, 2.0
mm; metatarsus, 1.5 mm; tarsus, 0.6 mm.
Second patella and tibia, 1.7 mm; third,
1.0 mm; fourth, 1.5 mm.

Diagnosis. Unlike females of Z. melano-
crania, those of Z. calyptrata have the me-
dian area of the epigynum dark with an
indistinct border ( Figure 77 ) . It is doubt-
ful that this species belongs to Zygiella and
is related to the other Zygiella species.

Distribution. Malaysia, Burma. (Of
ThorelFs specimens labeled Epeira calyp-
trata in the British Museum, Natural His-
tory, one is this species, the other specimen
is a related species.)

Zygiella melanocrania (Thorell)
Figures 81- 84

Epeira melanocrania Thorell, 1887, Ann. Mus.
Civica Storia Xatur. Genova, (2)5: 209. Fe-
male holotvpe from Teinzo, Burma, in the
Museo Civico di Storia Naturale, Genova, exam-
ined.

Zygiella melanocrania, - Roewer, 1942, Katalog
der Araneae, 1 : 886.

Aranetis melanocranius, - Bonnet, 1955, Bibliog-
raphia Araneorum, 2: 543.

Description. Carapace shiny brown.
Head region dark brown. Chelicerae
brown, darker than head region. Sternum
yellow-brown. Legs brown, first two darker
than last two, with faint indications of
darker rings. Dorsum of abdomen with
characteristic black and white Zygiella pat-
tern (Figure 84). Venter with white pig-
ment spots only. Secondary eyes 0.8 diam-
eter from anterior median eyes. Anterior
median eyes are a diameter apart, slightly
more than one diameter from laterals. Pos-
terior median eyes are their radius apart,
2.5 diameters from laterals. The laterals
are separated by their radius from each

Orb-weaver Genus Zygiella • Levi 283

other. The heiglit of tlie clypcus is about
equal to the radius of the anterior median
eyes. The chehcerae of one specimen have
four teeth on the anterior margin of tlie
fang furrow, but the posterior margin has
four on one cheHcera, three on the other;
there are denticles in the furrow. The ab-
domen is oval and hairy. Total length 5.5
mm. Carapace 2.6 mm long, 2.0 mm wide.
First femur, 2.7 mm; patella and tibia, 3.3
mm; metatarsus, 2.1 mm; tarsus, 0.9 mm.
Second patella and tibia, 2.7 mm; third,
1.6 mm; fourth, 2.2 mm.

Diai:,nosis. The epigynum (Figures 82,
83), with two semicircular openings on the
ventral side, separates this species from all
other known Zygiella.

Distribution. This species is known only
from the type specimen. The specimen il-
lustrated by Dyal ( 1935, Bull. Dept. Zool,
Panjab Univ. 1: 183, pi. 16, fig. 125) is
probably not this species.

Zygiella kochi (Thorell)
Figures 85-91

Zilla kochii Thorell, 1870, Remarks on Synonyms
of European Spiders, p. 33. Synt>'pes from Nice
and Monaco presumably in the Stockholm Nat-
ural History Museum. Rosenberg, 1901, Zool-
ogica, 13: 43, pi. 3. fig. 32, 9, $. Wiehle,
1929, Z. Morphol. Okol. Tiere, 15: 262-308.
Wiehle, 1931, in Dahl, Tierwelt Deutschlands,
23: 41, figs. 52, 53, 9, $.

Zygiella koclii, - Simon, 1929, Arachnides de
France, 6(3): 663, 754, figs. 1021, 1025, 9, $.
Roewer, 1942, Katalog der Araneae, 1: 884.
Ronnet, 1959, Ribliographia Araneoiinn, 2:
5002.

Description. Female from France. Cara-
pace brown, with darker lines going from
eye region to thoracic depression ( Figure
87). Sternum brown. Legs very indis-
tinctly banded. Dorsum of abdomen with
usual pattern ( Figure 87 ) . Venter with a
black spot framed by white on each side.
Secondary eyes 0.7 diameter of anterior
medians. Anterior median eyes 0.7 diam-
eter apart, one diameter from laterals. Pos-
terior median eyes one diameter apart, a
little less than two diameters from laterals.

Total length 7.5 mm. Carapace 3.5 mm
long, 2.5 mm wide. First femur, 3.2 mm;
patella and tibia, 4.3 mm; metatarsus, 3.0
mm; tarsus, 1.3 mm. Second patella and
tibia, 3.2 mm; third, 2.0 mm; fourth, 2.9
mm. The entrance into the seminal recep-
tacles is through pockets and folds rather
than through distinct ducts (Figures 88,
91).

Description of male from unknown lo-
cality. Posterior median eye diameter about
the radius of anterior median eyes; anterior
lateral eyes 0.7 diameter of anterior median
eyes; posterior lateral eye diameter about
the radius of anterior median eyes. An-
terior median eyes their radius apart and
about their radius from laterals. Posterior
median eyes their diameter apart, 1.5 di-
ameters from laterals. There are no modi-
fications on appendages. Total length 7
mm. Carapace 3.1 mm long, 2.3 mm wide.
First femur, 3.2 mm; patella and tibia, 4.7
mm; metatarsus, 3.1 mm; tarsus, 1.3 mm.
Second patella and tibia, 3.6 mm; third,
2.0 mm; fourth, 2.6 mm.

Diagnosis. The heart-shaped scape of
the epigynum (Figure 89) separates this
species readily from others. The scape has
a central depression. The rim on the tegu-
lum surrounding the base of the conductor,
and the shape of the subterminal apophysis
of the palpus separate males from Z. cas-
pica (Figures 85, 86).

Natural histonj. The species is found on
triuiks of trees, cork bark and chestnut in
Corsica; its retreat is in cracks in bark
(Wiehle, 1929, 1931). The web is simihir
to that of Z. x-notata; of fifteen webs four
did not have the vacant sector but had
complete orbs (Wiehle, 1929).

Distribution. Central and southern
Europe, Mediterranean region and North
Africa (Bonnet, 1959).

Zygiella inconveniens (O. P. -Cambridge)
Figures 92-94

Epeira inconveniens (O.P.-Cambridge), 1872,
Proc. Zool. Soc. London, p. 298. Female holo-

284 Bulletin Museum of Comparative Zoology, Vol. 146, No. 5

type and juvenile lectotype from Beirut, Leba-

non.

Zijgiella inconveniens, - Roewer, 1942, Katalog

der Araneae, 1: 883.
Araneiis inconveniens, - Bonnet, 1955, Bibliog-

raphia Araneorum, 2: 522.

Description. Coloration characteristic
for the genus ( Figure 92 ) . The secondary
eyes are about 0.8 diameter of anterior
medians. Anterior median eyes are 0.7
diameter apart, their radius from laterals.
The posterior median eyes are slightly less
than one diameter apart, slightly more than
one from laterals. Total length 5.5 mm.
Carapace 2.5 mm long, 1.9 mm wide. First
femur, 2.3 mm; patella and tibia, 3.2 mm;
metatarsus, 2.4 mm; tarsus, 0.9 mm. Second
patella and tibia, 2.5 mm; third, 1.5 mm;
fourth, 2.2 mm.

Diagnosis. Females differ from Z. kochi
in the longer, narrower scape of the epigy-
num ( Figure 93 ) .

Distribution. Only known from Beirut,
Lebanon.

Zygiella thorelli (Ausserer)
Figures 95-101

Zilla thorelli Ausserer, 1871, Verhandl. zool. bot.
Ges. Wien, 21: 830, pi. 5, fig. 10, ?. Female
from Prater (amusement park), Vienna, Aus-
tria, probably in the Naturhistorisches Museum,
Wien, not examined. Wiehle, 1931, in Dahl,
Tierwelt Deutschlands, 23: 39, figs. 49-51,

?, $.
Zygiella thorelli, - Simon, 1929, Arachnides de
France, 6(3): 663, 664, 755, figs. 1019, 1024,

9 S. Roewer, 1942, Katalog der Araneae, 1:
884. Bonnet, 1959, Bibliographia Araneorum,
2: 5006.

Description. Female from France. Cara-
pace brown, black lines from each pos-
terior lateral eye to thoracic region, fusing
there with a lateral branch; black line
around margin of the thoracic region ( Fig-
ure 95). Chelicerae dark brown. Sternum
dark brown with light brown median longi-
tudinal narrow band. Legs brown with
narrow dark bands. Dorsum with charac-
teristic pattern (Figure 95) containing
black and with white pigment spots. Ven-

ter black between genital furrow and spin-
nerets, with a white line on each side. Pos-
terior median eyes 0.6 diameter of anterior
median eyes. Anterior lateral eyes 0.7 di-
ameter of anterior medians, posterior lat-
eral eyes 0.5 diameter of anterior median
eyes. Anterior median eyes 0.7 diameter
apart, one diameter from laterals. Posterior
median eyes slightly less than their diam-
eter apart, 1.7 from laterals. On the
anterior margin of tlie fang furrow, the
chelicerae have three large teeth; on the
posterior margin, four teeth and, farthest
from fang, a denticle. Total length 10 mm.
Carapace 4.5 mm long, 3.2 mm wide. First
femur, 4.9 mm; patella and tibia, 6.7 mm;
metatarsus, 5.0 mm; tarsus, 1.7 mm. Second
patella and tibia, 5.0 mm; third, 2.9 mm;
fourth, 4.0 mm.

Male from Kochem on the Mosel, Ger-
many. Coloration like that of female. Sec-
ondary eye diameter 0.6 diameter of an-
terior median eyes. Anterior median eyes
slightly less than their radius apart, their
diameter from laterals. Posterior median
eyes 0.7 diameter apart, 1.5 diameters from
laterals. The chelicerae have three teeth
on the anterior margin; three smaller teetli
on the posterior. Total length 7.5 mm.
Carapace 3.9 mm long, 2.8 mm wide. First
femur, 4.8 mm; patella and tibia, 7.0 mm;
metatarsus, 6.5 mm; tarsus, 1.8 mm. Second
patella and tibia, 5.0 mm; third, 2.8 mm;
fourth, 3.5 mm.

Diagnosis. This species has a longer,
narrower carapace than is seen in other
species of Zygiella. Females are distinct
in the shape of the epigynal scape, a lobe
with a distal extension ( Figure 97 ) . Males
are characterized by the sculptured, hu-
man-ear-shaped tegulum (Figure 100). No
other known species is close to Z. thorelli.

Natural history. This central European
species prefers warm locations such as
walls of ruins and cliffs. It has also been
found on wooden buildings. The sexes are
mature in August and September (Wiehle,
1931). The web, a typical Zygiella web,

Ohb-weaver Genus Zygiella • Levi 285

Figures 87 91. Zygiella kochi {ThoreW). 87. Female. 88 91. Epigynum. 88. Dorsal, cleared. 89. Ventral. 90.
Posterior. 91. Posterior, cleared.

Figures 92-94. Z. inconveniens (O. P. -Cambridge). 92. Female. 93, 94. Epigynum. 93. Ventral. 94. Posterior.

Figures 95 101. Z. thorelli {Ausserer). 95. Female. 96 99. Epigynum. 96. Dorsal, cleared. 97. Ventral. 98.
Posterior. 99. Posterior, cleared. 100, 101. Left male palpus. 100. Ventral. 101. Lateral.

Scale lines. 0.1 mm except Figures 87, 92, 95, 1.0 mm.

286 Bitlletin Museum of Comparative Zoology, Vol. 146, No. 5

is pictured in Lendl ( 1891, Potpiiz Tennesz.
kozl., Budapest, 13: 31, figure 8).

Distribution. France, southern Ger-
many, Czechoslovakia, Pohmd to Italy and
Roumania (Bonnet, 1959).

Zygiella stroemi (Thorell)
Figures 102-110

Zilla stroemi Thorell, 1870, Remarks on Synonyms
of European Spiders, p. 235. New name for
Zilla montana, Westring (not C. L. Koch)
from Sweden. Wiehle, 1931, in Dahl, Tierwelt
Deutschlands, 23: 36, figs. 43^5, 9, $.

Zygiella x-notata, - Roewer, 1942, Katalog der
Araneae, 1: 884 (not x-notata Clerck).

Zygiella stroemi, - Locket and Millidge, 1953,
British Spiders, 2: 163, figs. 108b, 109c, 9, $.
Bonnet, 1959, Bibliographia Araneorum, 2:
5005.

Description. Female from Plitvice, Cro-
atia, Jugoslavia. Coloration similar to that
of other species (Figure 103). Diameter
of posterior median eyes 0.8 diameter of
anterior medians. Anterior lateral eyes 0.9
diameter of anterior medians and posterior
lateral eyes 0.8 diameter of anterior me-
dians. Anterior median eyes slightly less
than their radius apart, the same distance
from laterals. Posterior median eyes their
diameter apart, slightly more than their
diameter from laterals. Total length 4.5
mm. Carapace 1.9 mm long, 1.5 mm wide.
First femur, 2.2 mm; patella and tibia, 2.7
mm; metatarsus, 2.0 mm; tarsus, 0.9 mm.
Second patella and tibia, 1.9 mm; third,
1.3 mm; fourth, 1.9 mm.

Male from Plitvice, Croatia, Jugoslavia.
Diameter of secondary eyes 0.7 diameter
of anterior medians. Anterior medians 0.3
diameter apart, the same distance from lat-
erals. Posterior median eyes slightly less
than their diameter apart, slightly more
than their diameter from laterals. Total
length 3.4 mm. Carapace 1.7 mm long, 1.6
mm wide. First femur, 2.2 mm; patella
and tibia, 3.0 mm; metatarsus, 2.6 mm; tar-
sus, 1.0 mm. Second patella and tibia, 2.2
mm; third, 1.1 mm; fourth, 1.5 mm.

Diagnosis. The flat, long scape with al-
most parallel sides (Figure 104) separates

females from all other Zygiella. Males are \
distinguished by the truncate projection
of the tegulum of the palpus ( Figures 108-
110).

Natural historij. The web is on the
trunks of pines (Wiehle, 1931; Locket and
Millidge, 1953); the retreat is under bark,
Wiehle ( 1931 ) reports that the species ma-
tures from May until June, and several
specimens may be found near each other.

Distribution. Most of Europe to Siberia
and Turkestan (Bonnet, 1959).

Zygiella sia (Strand)
Figures 111-120

Aranea (Zilla) sia Strand, 1906, in Bosenberg
and Strand, Abhandl. Senckenberg. Ges., 30
(1-2): 237, pi. 4, fig. 24, $. Adult female,
male, and 5 juvenile syntypes from Japan in
the Senckenberg Museum, Frankfurt, examined.

Zygiella sia, - Roewer, 1942, Katalog der Araneae,
1: 884.

Araneus sia, - Bonnet, 1955, Bibliographia
Araneorum, 2: 598. Yaginuma, 1960. Spiders
of Japan in Colour, Osaka, p. 54, figs. 1, 3,
plate 19, fig. 115, ?, $.

Zilla sia, - Saito, 1959, The Spider Book Illus-
trated in Colours, Tokyo, p. 109, fig. 23, pi. 17,
fig. 129 a, b, pi. 18, fig. 129 d, ?, web.

Description. Female syntype. Carapace
brown, head region darker brown, darker
area coming to a point posteriorly in tho-
racic depression. Some white hairs on sides.
Sternum dark brown. Legs indistinctly to
distinctly banded. Abdomen with the
characteristic pattern. Venter with a white
longitudinal line on each side. Posterior
median eyes 0.6 diameter of anterior me-
dians, anterior lateral eyes 0.6, posterior
laterals 0.5 diameter. Anterior median eyes
0.8 diameter apart, 1.5 diameters from lat-
erals. Posterior median eyes 0.7 diameter
apart, 3.0 diameters from laterals. Lateral
eyes slightly separated. There are three
teeth on the anterior margin of chelicerae.
Total length 7 mm. Carapace 2.7 mm long,
2.2 mm wide. First femur, 2.9 mrn; patella
and tibia, 4.0 mm; metatarsus, 2.7 mm;
tarsus, 0.7 mm. Second patella and tibia,
3.4 mm; third, 1.9 mm; fourth, 2.7 mm.

Orb-weaver Genus Zygiella • Levi 287

Figures 102 110. Zygiella stroemi (ThoreW). 102. Eye region and chelicerae. 103. Female. 104 106. Eplgy-
num. 104. Ventral. 105. Anterodorsal, cleared. 106. Posterior. 107-110. Left male palpus. 107. Expanded.
108. Mesal. 109. Ventral. 110. Lateral.

Abbreviations. A, terminal apophysis; C, conductor; E, embolus; \, stipes; M, median apophysis; R, radix; SA,
subterminal apophysis; T, tegulum.

Scale lines. 0.1 mm except Figures 102, 103, 1.0 mm.

Male syntype from Japan. Coloration as
in female, but abdominal pattern more dis-
tinct. Diameter of secondary eyes about
equal to radius of anterior medians. Pos-
terior lateral eyes slightly smaller than
other secondary eyes. Anterior median
eyes 0.7 diameter apart, one diameter from

laterals. Posterior median eyes 0.6 diam-
eter apart, three diameters from laterals.
There are three teeth anteriorly on chelic-
eral fang margin and three posteriorly.
Total length 6 mm. Carapace 2.6 mm
long, 2.1 mm wide. First femur, 3.0 mm;
patella and tibia, 4.1 mm; metatarsus, 2.9

288 Bulletin Museum of Comparative Zoology, Vol. 146, No. 5

Figures 111-120. Zygiella sia (Strand). 111-115. Epigynum. 111. Ventral, cleared. 112. Ventral. 113. Ven-
tral, scape torn off. 114. Posterior, cleared. 115. Posterior. 116. Female. 117-120. Left male palpus. 117,
118. Expanded. 119. Ventral. 120. Lateral.

Abbreviations. A, terminal apophysis; C, conductor; E, embolus; H, hematodocha; M, median apophysis; P, para-
cymbium; R, radix; T, tegulum; TA, projection of tegulum.

Scale lines. 0.1 mm except Figure 116, 1.0 mm.

Ohb-weaver Genus Zygiella • Levi 289

mm; tarsus, 1.0 mm. Second patella and
tibia, 3.3 mm; third, 1.8 mm; fourth, 2.6
mm.

Note on size. Several other specimens
of this species were examined. They were
much larger. And with the size increase
there was a proportionate increase in the
distance of the laterals from the median
eyes. A female from Shiga Prefecture was
12.5 mm total length. The specimen had
a carapace 4.7 mm long and 4.1 mm wide,
about 1.7 times the size of the female syn-
type. The legs were of proportionate
length, 1.7 times that of the syntype. The
comparative eye sizes stayed the same but
anterior median eyes were about twice
their diameter from laterals (a distance in-
crease of about 1.3 times) and the posterior
medians slightly less than five times from
laterals (a distance increase of 1.6 times
almost proportionate to growth ) . The eyes
thus grew relatively less.

Male specimens from Naga Prefecture
were also larger: total length 7.5 and 10.5
mm; carapace 3.8 and 5.1 mm long, 2.7 and
3.9 mm wide. These measurements are 1.4
times and 2.0 times the corresponding mea-
surements of the syntype; the appendage
articles were, however, 1.7 times and 2.2
times the length of the carapace of the
type. Growth of males' legs thus did not
seem proportional. However, in the two
different-sized males from Naga Prefec-
ture, carapace and leg sizes were in pro-
portion.

The Naga males had the diameter of the
secondary eyes 0.7 diameter of the medians
(the syntypes about 0.5 diameter). The
syntype had the anterior median eyes one
diameter from laterals, the smaller Naga
specimen one and one-half, the larger one
slightly less than two. The posterior me-
dian eyes were three diameters from lat-
erals in the syntype, about four in the
smaller Naga specimen (1.3 times the dis-
tance), about five times in the larger one
( 1.7 times the distance in tlie syntype). The
eye distances increase less than size; there

appears to be only little increase in eye
sizes.

Presumably the specimens had matured
in different instars. But tliesc proportional
differences seem surprising considering the
similarity in proportion and size of the
epigyna and male palpi.

The male specimen whose carapace was
twice as long as the carapace of the syn-
type, also had the palpal tibia 2.5 times
as long as that of the syntype (a propor-
tional increase with leg length), but the
critical palpal cymbium was only 1.4 times
longer than that of the syntype. The larger
specimen thus had relatively a much
longer palpal tibia. No differences were
noted in the position and proportion of the
sclerites held within the cymbium.

Diagnosis. The heart-shaped scape cov-
ering the ventral openings of the epigynum
(Figure 112) separates the female from all
other Zygiella. The scape has a transverse
light mark. The male palpus (Figures 119-
120) is superficially very different from
other species: it has a huge basal hema-
todocha, a minute tegulum bearing a
toothed projection, and the median apoph-
ysis has a projecting hook (Figures 117-
118). As in Z. atrica the palpal tibia is
slightly elongated, but of a different shape.

There is some doubt in placing this
species in Zygiella, because of the wider
spacing of the eyes and the cap on the pal-
pal embolus in the expanded palpus (Fig-
ure 117), not otherwise seen in the genus.
The course of the duct into and through
the tegulum remains uncertain, despite its
having been illustrated in Figures 117, 118.

Distribution. Japan. Fox ( 1938, J. Wash-
ington Acad. Sci. 28: 367) reported speci-
mens from Szechwan Prov. China, but the
specimens of the U.S. National Museum
could not be found.

REFERENCES CITED

Archer, A. 1951a. Studies in the orbweaving
spiders ( Argiopidae), 1. Amer. Mus. Novi-
tates, No. 1487: l-,52.

. 1951b. Studies in the orbweaving spi-

290 Bulletin Museum of Comparative Zoology, Vol. 146, No. 5

ders ( Argiopidae), 2. Amer. Mus. Novitates,

No. 1502: 1-34.
Bonnet, P. 1959. Bibliographia Araneorum,

Vol. 2. Imprimerie Douladoure, Toulouse,

vol. 2, 4231-5058.
Emerton, J. H. 1884. New England spiders

of the family Epeiridae. Trans. Connecticut

Acad. Arts Sci., 6: 295-342.
Gertsch, W. J. 1964. The spider genus Zy-

giella in North America (Araneae, Argiop-

idae). Amer. Mus. Novitates, No. 2188:

1-21.
Roewer, C. F. 1942. Katalog der Araneae.

Verlag Natura, Hamburg, vol. 1.
Simon, E. 1895. Histoire Naturelle des Araig-

nees. Roret, Paris: Libraire Encyclopedique

vol. 1.
Wiehle, H. 1931. Araneidae. In Dahl, F. Die

Tierwelt Deutschlands, G. Fischer Verlag,

Jena 23: 1-136.

Valid names are printed in italics,
illustrations.

alpina, Zilla 270
ancora, Epeira 270
atrica, Eucharia 272
atrica, Zilla 272
atrica, Zijgiella 269*, 272, 273*
aureola, Zilla 270
bosenbergi, Zilla 276
californica, Zilla 276
calophylla, Aranea 271
calyptrata, Epeira 282
calyptrata, Zygiella 281*, 282
calyptratus, Araneus 282
carpenteri, Zygiella 279*, 280
caspica, Zilla 280
caspica, Zygiella 280, 281*
crucinotata, Zilla 271
decolorata, Zilla 271
dispar, Zilla 278
dispar, Zygiella 277*, 278
gigans, Zilla 271
guttata, Zilla 271
guyanensis, ZiUa 271
inconveniens, Araneus 284
inconveniens, Epeira 283
inconveniens, Zygiella 283, 285*
keyserlingi, Zilla 274
keyserlingi, Zygiella 273*, 274
kochii, Zilla 283

INDEX

Page numbers refer to main references, starred page numbers to

kochi, Zygiella 281*, 283, 285*

maulliana, Larinia 276

melanocephala, Linyphia 271

melanocrania, Epeira 282

melanocrania, Zygiella 281*, 282

melanocranius, Araneus 282

minima, Zygiella 274, 275*

montana, Zilla 278

montana, Zygiella 211*, 278, 279"

montana, Zygiella 278

nawazi, Zilla 271

nearctica, Zygiella 278

punctata, Zilla 271

rogenhoferi, Zilla 271

sia, Aranea 286

sia, Araneus 286

sia, Zilla 286

sia, Zygiella 286, 288*

stroemi, Zilla 286

stroemi, Zygiella 286, 287*

thorelli, Zilla 284

thorelli, Zygiella 284, 285*

x-notata, Zygiella 275*, 276, 279*

x-notata, Zygiella 286

x-notatus, Araneus 276

Zygia 268

Zygiella 268

us ISSN 0027-4100

Sulletin OF THE

Museum of

Compardtive

Zoology

The Orb-weaver Genera Ataniella and
Nuctenea (Araneae: Araneidae)

HERBERT W. LEVI

HARVARD UNIVERSITY

CAMBRIDGE, MASSACHUSETTS, U.S. A

VOLUME 146, NUMBER 6
21 NOVEMBER 1974

PUBLICATIONS ISSUED

OR DISTRIBUTED BY THE

MUSEUM OF COMPARATIVE ZOOLOGY

HARVARD UNIVERSITY

Bkeviora 1952-

BULLETIN 1863-

Memoirs 1864-1938

JoHNsoNiA, Department of MoUusks, 1941-

OccAsiONAL Papers on Mollusks, 1945-

SPECIAL PUBLICATIONS.

1. Whittington, H. B., and E. D. I. Rolfe (eds.), 1963. Phylogeny and
Evolution of Crustacea. 192 pp.

2. Turner, R. D., 1966. A Survey and Illustrated Catalogue of the Teredini-
dae (Mollusca: Bivalvia). 265 pp.

3. Sprinkle, J., 1973. Morphology and Evolution of Blastozoan Echinoderms.
284 pp.

4. Eaton, R. J. E., 1974. A Flora of Concord. 250 pp.

Other Publications.

Bigelow, H. B., and W. C. Schroeder, 1953. Fishes of the Gulf of Maine.
Reprint.

Brues, C. T., A. L. Melander, and F. M. Carpenter, 1954. Classification of
Insects.

Creighton, W. S., 1950. The Ants of North America. Reprint.

Lyman, C. P., and A. R. Dawe (eds.), 1960. Symposium on Natural
Mammalian Hibernation.

Peters' Check-list of Birds of the World, vols. 2-7, 9, 10, 12-15.

Proceedings of the New England Zoological Club 1899-1948. (Complete
sets only.)

Publications of the Boston Society of Natural History.

Price list and catalog of MCZ publications may be obtained from Publications
Office, Museum of Comparative Zoology, Hai^vard University, Cambridge, Massa-
chusetts, 02138, U.S.A.

THE ORB-WEAVER GENERA ARANIELLA AND NUCTENEA
(Araneae: Araneidae)

HERBERT W. LEVI

Abstract. The species included in Araniella and
Nuctenea have traditionally been included in Ara-
neiis, but males differ in lacking the embolus cap.
The lack of embolus cap can be related to differ-
ences in mating behavior. Those orb-weavers witii
cap (Amnetis) can mate only once with each
palpus; Nuctenea males lacking a cap can mate
several times.

Four species of Araniella are known, one of
diem Holarctic, the others Palearctic. Some Euro-
pean populations are of interest as there are indi-
cations that the species hybridize.

Of the si.x species known to belong to Nuctenea,
three are Holarctic, and three Palearctic. One of
the Holarctic species may be a recent introduc-
tion to North America; another may be cosmopoli-
tan.

Other species belonging to these two genera
may be hidden among the two thousand si^ecies
placed in Araneus and mostly poorly described.

INTRODUCTION

Araniella and Nuctenea species have
traditionally been placed in Araneus. They
include our commonest orb-weavers. Nev-
ertheless the species are not well known,
and in looking through the collections
available, I found that many specimens
were misidentified.

The species of both genera are mainly
Palearctic with some Holarctic species.

I would like to thank the following for
providing specimens for this study: D.
Bixler; W. J. Gertsch and J. A. L. Cooke of
the American Museum of Natural History,
Cornell and Utah University collections;
J. E. Carico; M. Grasshoff of the Sencken-

berg Museum; C. Dondale and R. Leech,
Canadian National collections, Ottawa; W.
Hackman; G. H. Locket; M. Martelli of
the Zoological Museum of the University
of Florence; W. W. Moss of the Academy
of Natural Sciences, Philadelphia; W. Peck,
Exline-Peck collection; R. X. Schick of
the California Academy of Sciences; J.
Proszynski and W. Star^ga of the Polish
Academy of Sciences; B. Vogel; H. K. Wal-
lace and H. V. Weems, Florida State Col-
lection of Arthropods. Information was
provided by T. Kronestedt of the Natural
History Museum, Stockholm; M. Moritz of
the Zoological Museum of the Humboldt
University, Berlin; and F. Wanless and D.
Newman of the British Musevmi, Natural
History. Lorna R. Levi and Lm Mackay
edited the paper. This investigation and
its publication were supported in part by
National Science Foundation grant number
GB-3616L

One of the striking and puzzling features
of these common orb-weavers is the enor-
mous individual variation in genitalic
structiu-e (Figs. 8-15, 17, 18, 67-76'), while
there is little variation in the size and shape
of the whole animal. This variation is
found in Holarctic Araniella (Usplicata and
A. cucurhitina of Europe as well as in
Holarctic Nuctenea patagiata and Holarctic
N. cornuta. Not infrequently specimens of
these common species are sent to the
museum by collectors who believe them to

Bull. Mas. Comp. Zool., 146(6) : 291-316, November, 1974 291

292 Bulletin Museum of Comparative Zoology, Vol. 146, No. 6

be a new species. In their unusual varia-
tion, Araniella and Nuctenea contrast with
the small Aroneus species (Levi, 1973).
Even though the species are fairly wide-
spread, the differences between small
Araneus species are far less than is found
among individuals of a single collection of
N. cornuto. The larger-sized Araneus
species (A. nordmanni, A. saevus) are in-
termediate in this respect ( Levi, 1971 ) .
Large variation was not found in any of
the widespread, common thcridiids such
as Achaearanea tepidarioriim, although it
does occur in the Tidarren species and
Enoplognatha ovata. Variation in the
genitalic structure of individuals was
found in the zone of overlap between
Araneus gemma and Araneus gemmoides;
all evidence indicates hybridization (Levi,
1971 ) . Araniella species may also hybridize
in Europe. Perhaps the individual varia-
tion of Araniella cucurhitina of Europe is
due to separation of northern and southern
populations during the Pleistocene and the
later hybridization occurred after the reced-
ing of the ice. In none of the species of
Araniella or Nuctenea is the variation
geographic. But I have not studied various
populations in detail. Shortly after Pet-
iimkevitch ( 1925 ) wrote on the remarkable
variation of genitalia of Agelena naevia,
Seyler (1940), following up a hint from
Gertsch (1934), correctly found that what
Petrunkevitch called one species was in
fact several. But I doubt that populations
of Araniella displicata or Nuctenea patagiata
and N. cornuta consist of sibling species.

Of considerable interest is the relation-
ship of Araniella species in Europe. Mr.
Locket made me aware of this. While A.
inconspicua and A. alpica appear distinct
on the Continent, intermediates are found
in Great Britain.

While in many American araneid and
theridiid species specimens from the Gulf
Coast and Florida are noticeably smaller
than those from other parts of North Amer-
ica, Alaskan specimens of N. cornuta are
smaller than those from southern Canada

and the United States. To judge by the
labels, N. cornuta is less dependent on
houses in Alaska and probably competes
with the native Araneus species. In north-
eastern America all three Nuctenea species
are usually found on buildings, but this
is not true throughout their ranges. Nu-
ctenea patagiata may be found under bark
in woods.

It is most unfortunate that at times
names have to be changed as a result of
revisionary studies.

Araniella Chamberlin and Ivie

AmnicUa Chamberlin and Ivie, 1942, Bull. Univ.
Utah, biol. ser., 7(1): 76. Type species Epeira
displicata Hentz, by original designation. The
name is of feminine gender.

Note. Chamberhn and Ivie (1942) do
not give reasons for separating E. displicatus
from Araneus other than that the species
is close to those of Neoscona. I agree with
this opinion.

Diagnosis. There are no good superficial
characters that separate female Araniella
from the small species of Araneus. Females
have a glabrous carapace and an oval
abdomen, widest in the middle, lacking
setae and lacking a folium pattern, but
usually with paired black spots (Plate 1,
Figs. 1, 16). The epigynum has a short,
wide, wrinkled scape (Figs. 8, 25, 34, 40).
The scape is not always clearly set off from
the base of the epigynum. Unlike Araneus
species, however, Araniella has, besides a
single pair of seminal receptacles, a pair
of sclerotized sacs (Figs. 7, 27, 33, 39)
between the external entrance from outside
to the connecting duct and the seminal
receptacles. The entrance of the duct to
the sclerotized sacs on each side is a slit.
The ducts are hard to see in very sclerotized
epigyna (A. cucurhitina).

Unlike other araneids I have examined,
Araniella has three macrosetae (Figs. 31,
42) on the patella of the male palpus.
Species of most genera have only two or
one. However, the palpal femur has a
basal ventral tooth facing a tooth on the

Orb-weaver Auaniella and Nuctenea • LcxA 293

endite as in Araneus. The palpus resembles
that of Neoscona (Berman and Levi, 1971,
fig. 31) in having the sclerites nearly
fused, a small flap-like terminal apophysis
(A in Figs. 20-22), and in laekiug a distal
hematodoeha (Fig. 21). This eontrasts
with the huge terminal apophysis and
distal hematodoeha in Araneus and Nu-
ctenea (Figs. 58, 61). AranicUa speeies
also differ from Neoscona and Araneus in
the hook-shaped, selerotized, median apoph-
ysis ( M ) , dorsally directed toward the
eymbium (Figs. 20, 22), and in the un-
usually complex, large conductor (C). The
embolus (E) lacks a cap. An embolus cap
is always present in virgin males of Araneus
species; the cap breaks off in mating.

Araniella males have a hook on the distal
margin of the first coxa, and the second
femur has a matching depression. The legs
of males are longer than those of females,
and bear many macrosetae.

Most Araniella are green or yellow to
reddish in color when alive, similar to small
species of Araneus. The color, except the
black spots, washes out in alcohol. All
species are about the same size and pro-
portions (Figs. 1, 16).

Natural history. All species build a
small web between leaves; the web may be
liorizontal. The spiders are active at day-
time. The egg sac has a loose woollen ap-
pearance (Plate 1).^

Distribution. The known species are all
Palearctic, except A. displicata which is
Holarctic in distribution.

Misplaced species. Araniella geayi Ca-

' I liad anticipated that Araniella males, which
resemble males of Nuctenea species in lacking a
cap on the embolus, would likewise be able to
mate repeatedly. Therefore, it was with consid-
erable interest that, after submitting my own
manuscript, I received R. Blanke ( 1973, Neue
Ergebnisse zum Sexualverhalten von Araneus cu-
ctirbitinus, Forma et Functio, 6: 279-290).
iilanke did indeed observe differences between the
behavior of A. cucurbitinus and that of Araneus
species: The female approaches the male, the
male taps the tarsi of females, and the male mates
several times.

Plate 1. Araniella displicata (Hentz). Above, female
with egg sac. Below, male. Both laboratory photo-
graphs, from Massachusetts.

poriacco, 1954, Comm. Pontifica Acad. Sci.,
16: 104 from Guyana has as a holot^'pe a
juvenile specimen belonging to the genus
Araneus deposited in the Zoological Mu-
seum of the University of Florence, ex-
amined.

Key to Species of Araniella

1 Male, conductor of palpus with a distal

lobe directed at and close to palpal pa-
tella (Figs. 37, 42); female with epigy-

294 Bulletin Museum of Comparative Zoology, Vol. 146, No. 6

Map 1. North American distribution of Araniella displicata (Hentz).

niim scape having proximal part widest,
much wider than tip (Figs. 34, 40) — _ 2

- Male palpal conductor without distal lobe
toward patella (Figs. 18, 30); female
epigynum with scape having parallel sides
or proximal constriction (Figs. 10,
25 ) - 3

2(1) Median apophysis with fin toward the
embolus ( Fig. 42 ) ; female with base
of epigynum showing on each side of
scape ( Fig. 40 ) ; Europe alpica

- Median apophysis without fin (Fig.
37 ) ; female epigynum base not showing
in ventral view (Fig. 34); Eurasia — _
inconspicua

3(1) Conductor with a set-off piece which
holds terminal apophysis and embolus
(Figs. 22, 30, 31); scape of epigynum
narrower or of equal width to part of base
visible on each side of it ( Figs. 23,
25); Eurasia, perhaps Arctic North Amer-
ica cucu rbitina

- Conductor without set-off piece holding
terminal apophysis and embolus ( Figs.
17, 18); scape of epigynum much wider

than part of base showing on each side
of it (Figs. 8-14); North America and
Eurasia displicata

Araniella displicata (Hentz)
Plate 1 ; Figures 1-21 ; Map 1

Epeira displicata Hentz, 1847, J. Boston Soc.
Natur. Hist., 5: 476, pi. 31, fig. 17. Types
from Alabama, May, Oct., destroyed. Emerton,
1884, Trans. Connecticut Acad. Sci., 6: 313, pi.
34, fig. 4, pi. 36, fig. 20, 342. Kevserling, 1893,
Spinnen Amerikas, 4: 219, pi. 10, fig. 162, 9.
Emerton, 1902, Common Spiders, p. 172, fig.
405. Kaston, 1948, Bull. Connecticut Geol.
Natur. Hist. Surv., 70: 258, fig. 806, ? .

Epeira decipiens Fitch, 1856, Trans. New York
Agric. Soc, 15: 451. Male specimen from
New York, lost.

Epeira sexpunctata Keyserling, 1884, Verhandl.
Zool. Bot. Gesell. Wien, 34: 530, pi. 13, fig. 28,
9 . Female type from North America in the
Museum of Comparative Zoology, examined.
Keyserling, 1892, Spinnen Amerikas, 4: 200,
pi. 9, fig. 148.

Orb-weaver Araniella and Nuctenea • Levi 295

^r^W''%'^

Figures 1-19. Araniella displicata (Hentz). 1-15. Female. 1. Dorsal view. 2. Carapace. 3. Eyes and chelic-
erae. 4. Internal genitalia with epigynum cleared. 5. Epigynum having scape torn off. 6-7. Internal genitalia.
6. Anterolateral. 7. Posterior. 8-15. Epigynum, 8, 10, 12, 14. Ventral view. 9, 11, 13, 15. Posterior view. 8,
9, 10, 11, 14, 15. (California.) 12, 13. (Oregon.) 16 19. Male. 16. Dorsal view. 17, 18. Left male palpus,
mesal view. 19. Palpus, ventral view. 17. (Minnesota.) 18, 19. (Montana.)

Scale lines. 0.1 mm; Figs. 1-3, 16, 1 mm.

296 Bulletin Museum of Comparative Zoology, Vol. 146, No. 6

Epeira alba Keyserling, 1884, Verhandl. Zool. Bot.
Gesell. Wien, 34: 530, pi. 13, fig. 20, 9. Fe-
male t>pe from Kentucky in the Museum of
Comparative Zoology, examined.

Epeira cuctirhitina, - McCook, 1893, American
Spiders, 3: 150, pi. 3, figs. 1-3, pi. 4, fig. 6, ?,
$ . Not A. cuctirhitina (Clerck).

Aranetis croaticus Kulczynski, 1905, Bull. Acad.
Sci. Cracovie, 233, pi. 7, figs. 22, 30, 9 . Female
holotype from Croatia in the Polish Academy
of Sciences, Warsaw, examined. NEW SY-
NONYMY.

Aranea displicata, - Comstock, 1912, Spider Book,
p. 494, figs. 524, 525, ? , web. Wiehle, 1931 in
Dahl, Tierwelt Deutschlands, 23: 109, figs. 167-
170, 9, $. Comstock, 1940, Spider Book, 2nd
ed., p. 508, figs. 524, 525, 9, web. Roewer,
1942, Katalog der Araneae, 1: 798.

Araniella displicata, - Chamberlin and Ivie, 1942,
Bull. Univ. Utah, biol. sen, 32(13): 76.

Araniella displicata octopunctata Chamberlin and
Ivie, 1942, Bull. Univ. Utah, biol. ser., 32(13):
76. Female holotype from Emigration Canyon,
Wasatch Mts., Utah, in the American Museum
of Natural History, paratype, examined.

Araneus displicatns, - Locket and Millidge, 1953,
British Spiders, 2: 149, figs. 96b, 97c, 99c,
lOOc-e, 9, $.

Araneus cucurbitimts displicatns, - Bonnet, 1955,
Bibliographia Araneonnu, 2: 478.

Note. It is of interest that this species
described from Alabama is very rare at
the present time in the Gulf states if present
at all. Only a juvenile from Bankhead
National Forest, Alabama, might be this
species. Others collected by Archer in
Alabama were Armieus gadus Levi that
had been misidentified.

Description. Female from California.
Carapace yellowish with eyes on black
spots. Legs yellow. Dorsum of abdomen
yellowish with three pairs of circular black
spots on the posterior part (Fig. 1).
Carapace smooth, almost without setae.
There is no thoracic depression. Diameter
of posterior median eyes 1.2 diameters of
anterior medians; anterior laterals 0.8 di-
ameters of anterior medians; posterior
laterals subequal in size to anterior medians.
Anterior median eyes L5 diameters apart,
3.2 from laterals. Posterior median eyes
one diameter apart, three from laterals.
The height of the clypeus is about 1.5 di-

ameters of the anterior median eyes. The
chelicerae, which are not very strong, have
four teeth on the anterior margin, three
on the posterior. The legs are relatively
heavy, with many macrosetae. The abdo-
men is suboval, widest in the middle. Total
length 5.3 mm. Carapace 2.5 mm long, 1.9
mm wide. First femur, 1.9 mm; patella and
tibia, 2.4 mm; metatarsus, 1.4 mm; tarsus,
0.6 mm. Second patella and tibia, 2.2 mm;
third, 1.3 mm; fourth, 2.2 mm.

Male from California. Coloration as in
female, except legs tend to be banded or
distal ends of leg articles darker. Carapace
smooth with head region relatively high
and a shallow thoracic depression, having
two lines crossing each other at right angles
(Fig. 16). Eyes are subequal in size. An-
terior median eyes 1.7 diameters apart,
two from laterals. Posterior median eyes
their diameter apart, 2.5 from laterals. The
height of the clypeus is 1.5 diameters of
the anterior median eyes. Total length 4.0
mm. Carapace 1.8 mm long, 1.7 mm wide.
First femur, 1.8 mm; patella and tibia, 2.3
mm; metatarsus, 1.3 mm; tarsus, 0.6 mm.
Second patella and tibia, 2.0 mm; third,
1.2 mm; fourth, 1.9 mm.

Variation. Total length of females 4.8 to
7.2 mm. Carapace 2.0 to 2.7 mm long, 1.7
to 2.0 mm wide. Total length of males
4.0-5.0 mm. Carapace 2.0-2.4 mm long,
1.7-2.2 mm wide. The coloration is much
more variable than the size; it is often
greenish, reddish, brownish or yellowish
on the abdomen. This pigment washes
out, however, and in alcohol the abdomen
is generally white.

Diagnosis. This is probably the only
species of Araniella occurring in North
America, although A. cuctirhitina may
occur in northern Canada and Alaska.
Araniella displicata females can be sepa-
rated by the wider and longer scape, which
often hides the base of the epigynum;
the widest part of the scape is generally
toward its middle (Figs. 8-14). In pos-
terior view the base is shorter, wider ( Figs.
9, 11, 13, 15) than that of A. cucurbitina.

Orb-weaver Aranieu.a and Nuctenfa • Levi 297

Figures 20-21. Araniella displicata (Hentz). 20. Left male palpus, ventral view without cymbium, cleared. 21.
Palpus expanded.

Figures 22-31. A. cucurbitina (Clerck). 22. Male palpus, expanded. 23 26. Epigynum. 23, 25. Ventral. 24,
26. Posterior. 23, 24. (Taunus, Germany.) 25, 26. (Scotland.) 27-29. Epigynum cleared. 27. Ventral. 28.
Lateral. 29. Posterior. 30, 31. Palpus, mesal. 30. (Poland.) 31. (Germany.)

Abbreviations. A, terminal apophysis; C, conductor; E, embolus; M, median apophysis; R, radix; T, tegulijm.

Scale lines. 0.1 mm.

298 Bulletin Museum of Comparative Zoology, Vol. 146, No. 6

Males can readily be separated from A.
cucurbitina and other species by the shape
of the conductor, which holds the tip of
the embolus and terminal apophysis ( Figs.
17-21). That of A. displicata has several
large teeth around its margin while a
special lobe on the conductor of A. cucur-
bitina holds the tip of embolus and terminal
apophysis (Figs. 22, 30, 31). The main
sclerotized part of the conductor of A.
cucurbitina has just one tip.

Habits. Males are mature from late
spring until late summer, females until
fall. The spider is found by sweeping
meadows and low bushes. This species
makes a relatively small orb web, often
among the leaves of bushes or underneath
a single large leaf.

Distribution. Europe, North America,
from the Arctic to North Carolina, probably
Alabama, and Arizona in the south; but
apparently absent from the south-central
states, southern Iowa, Nebraska to the
Gulf (Map 1).

Araniella cucurbitina (Clerck)
Figures 22-31

Araneiis cucurbitimis Clerck, 1757, Aranei Svecici,
p. 44, pi. 2, fig. 4, ? . Cleick's specimens in tlie
Natural History Museum of Stockholm, origi-
nally pinned and labeled by Thorell, not exam-
ined. Locket and Millidge, 1953, British Spi-
ders, 2: 144, figs. 96a, 97b, 98a, 99a, 9, $.
Bonnet, 1955, Bibliographia Araneorum 2: 472
( in part ) .

Epeim proxima Kulczynski, 1885, Pamiet. Akad.
Umiejet. Krakow, li: 19, pi. 9, fig. 11. Male
holotype from Kamchatka in tlie Polish Acad-
emy of Sciences, examined. NEW SYNONYMY.

Araneus cucurbitina opisthographa Kulczynski,
1905, Bull. Acad. Sci. Cracovie, p. 232, pi. 7,
figs. 2, 20, 23, 26, 9, $. Syntypes from nu-
merous localities in Poland in the Polish Acad-
emy of Sciences, Warsaw, examined.

Aranea cucurbitina, - Wiehle, 1931, in Dahl, Tier-
welt Deutschlands, 23: 106, figs. 161, 164, 9,
$. Roewer, 1942, Katalog der Araneae, 1: 785.

Aranea proxima, - Roewer, 1942, Katalog der
Araneae, 1: 790.

Araneus proximus, - Bonnet, 1955, Bibliographia
Araneorum, 2; 571.

Note. I could not find any consistent
differences between specimens labeled
opisthographa and others. Figure 30 was
prepared from a syntype of A. opistho-
grapha.

Diagnosis. The short, narrow scape with
parallel sides which exposes most of the
base of the epigynum in ventral view
(Figs. 23-26) readily separates A. cucur-
bitina from other Araniella females. The
more complex conductor with a distinct,
separate lobe holding the tip of embolus
and terminal apophysis (Figs. 22, 30, 31)
distinguishes males (from other Araniella
species ) .

Natural history. Very common in trees
and bushes in Europe (see Wiehle, 1931).

Distribution. Common and widespread
in Eurasia from Great Britain to Kamchatka
(Wiehle, 1931; Locket and Millidge, 1953).
The species is believed not to occur in
North America. The single record is prob-
ably the result of comparing specimens and
misplacing them, or perhaps the species
may occur in poorly collected Alaska and
northern Canada. The female was from
Fort Smith, Northwest Territory, 20. VI.
1967 (R. Leech), in white poplar.

Araniella inconspicua (Simon)
Figures 32-37

Epeira inconspicua Simon, 1874, Arachnides de
France, 1: 84. Female type in the Musemn
National d'Histoire Naturelle, Paris, not exam-
ined.

Aranea inconspicua, - Wiehle, 1931, in Dahl, Tier-
welt Deutschlands, 23: 112; figs. 174-176, 9,
$. Roewer, 1942, Katalog der Araneae, 1: 787.

Arai^eus inconspicuus, - Locket and Millidge, 1953,
British Spiders, 2: 146; figs. 97a, 98c, 99d,
100b, 9 , $ . Bonnet, 1955, Bibliographia
Araneorum, 2: 521.

Diagnosis. The abdomen usually lacks
the black spots. The female has a triangular
wrinkled scape that completely hides the
base in ventral view (Fig. 34); in pos-
terior view, the median groove is shorter
(Figs. 35, 36) tlian that of A. alpica. The
palpus lacks the distally directed fin of

Orb-weaver Araniella and Nuctenea • Levi 299

.^, .fiftr^i'ji-^'^-

Figures 32-37. Araniella inconspicua (Simon). 32-36. Epigynum. 32. Ventral, cleared. 33. Posterior, cleared.
34. Ventral. 35, 36. Posterior. 32-35. (France.) 36. (England.) 37. Left male palpus, mesal.

Figures 38-42. A. alpica (L. Koch). 38-41. Epigynum. 38. Ventral, cleared. 39. Posterior, cleared. 40. Ven-
tral. 41. Posterior. 42. Male palpus, mesal.

Scale lines. 0.1 mm.

the median apophysis (Fig. 37) present in
A. alpica. While Continental specimens
are readily separated from A. alpica, this
is not true for those of the British Isles.
Perhaps as a result of a recent introduc-
tion of A. alpica they hybridize.

Natural history. Found in trees and
bushes at low elevations (Wiehle, 1931).

Distribution. Europe from Great Brit-
ain, northern Spain to Macedonia ( Wiehle,
1931). There are also references to A.

inconspicua occurring in eastern Asia
(Bonnet, 1955).

Araniella alpica (L. Koch)
Figures 38-42

Epeira alpica L. Koch, 1869, Z. Ferdinandeum
Tirol, (3)14: 173. Specinien.s from Tyrol and
other locaHtios in the Kocli collection of the
Briti.sh Mu.seum (Natural History); presiunahly
types but not examined.

Aranca alf)ica, - Wiehle, 1931, in Dahl, Tier-
welt Deutschlands, 23: 110, figs. 171-173, 9,
$. Roewer, 1942, Katalog der Araneae, 1: 781.

300 Bulletin Museum of Comparative Zoology, Vol, 146, No. 6

Araneus alpiciis, - Locket and Millidge, 1953, Brit-
ish Spiders, 2: 149, figs. 96c, 98d, 99b, 100a,
5 , $ . Bonnet, 1955, Bibliographia Araneomni,
2: 428.

Araniella alpica, - Archer, 1951, Natur. Hist.
Misc., 84: 3, fig. 4, $.

Diagnosis. In ventral view the base of
the epigynum shows as two bulges pos-
terior and lateral to the scape (Fig. 40).
In posterior view the central area is much
longer (Fig. 41) than that of A. iricon-
spicua. The median apophysis has a dis-
tally directed fin ( Fig. 42 ) and the terminal
apophysis is wider than that of other
species. The abdomen has at most four
black spots (see note under A. inconspicua
diagnosis ) .

Natural history. This species is found in
European mountains above 300 m to
krummholz ( 1800 m in the alps ) and is
limited to fir and spruce. Males are ma-
ture until August (Wiehle, 1931).

Distribution. Great Britain ( Locket and
MilHdge, 1953), Scandinavia, Central Eu-
rope (Wiehle, 1931), Balkans (Bonnet,
1955).

Nuctenea Simon

Niictenea Simon, 1864, Histoire Naturelle des
Araignees, p. 261. New subgenus of Epeira
with the type species Epeira umbratica desig-
nated by Bonnet, 1950, Bibhographia Araneo-
rum, 2: 3118. The name is of feminine gender.

Cyphepeira Archer, 1951, Natur. Hist. Misc., Chi-
cago, 84: 4. New subgenus of Epeira with the
type species by original designation Epeira ( Cy-
phepeira) silvicultrix C. L. Koch. The name
is of feminine gender.

Note. The species included here had
been placed by Wiehle ( 1931 ) in groups 4
and 5 of Aranea, by Locket and Millidge
(1953) in groups 3 and 5 of Araneus.
Archer (1951), following F. P.-Cambridge
( 1903 ) , considered the group a distinct
genus, Epeira, with Araneus cornutus
(Clerck) the type of the genus. But this
type designation is an error as Epeira is
an objective synonym of Ara^iea and a sub-
jective synonym of Araneus, having Epeira

diadernata as type designated by Latreille,
1810 (Levi, 1971).

In considering the group included here
as distinct, I am following older authors
and also Archer. In 1959 Yaginuma and
Archer included the species in Cyphepeira,
as did Proszynski and Star^ga (1971).
Wiehle (1927), discussing orb-web build-
ing, included Epeira umbratica, E. sclo-
petaria, E. cornuta and E. patagiata in a
group making an unibraticus-type web.
Gerhardt (1926) separated the group be-
cause of different mating behavior: males
can mate three or four times with each
palpus. Araneus species can mate only
once with each palpus. Also, the female
assumes a different mating position, ap-
proaching and hanging opposite the male,
with cephalothorax lowered and abdomen
raised, and pulls in the male on threads.
The male will court a female that does not
have an orb-web, and in mating the male's
body is not as close to the female's as in
Araneus. Males do not refill their palpi
with sperm immediately after mating, as
males of Araneus have been observed to do
(observations on N. umbratica, N. cornuta,
N. sclopetaria) .

Description. All species are gray to
brown, none brightly colored (Plate 2).
The abdomen is dorsoventrally flattened,
oval in outline, widest in the middle, with
a folium on the dorsum (Figs. 97-109).
The cardiac mark is usually dark. The
venter of the abdomen is black with a pair
of comma-shaped or bracket-shaped white
marks (Figs. 98, 99, 102, 104).

The genitalia are heavily sclerotized.
The opening of the epigynum is hard to
find and the connecting duct difficult to
make out, even in epigyna digested with
10 percent NaOH. The openings of the
N. umbratica epigynum are anterior on
the base (Figs. 45, 46), those of other
species, posterior in a groove (Figs. 71,
72, 81, 82, 85, 86).

The palpus has a simple conductor ( C in
Figs. 58-62), unlike that of Araniella (Figs.

Orb-weaver Araniella and Nuctenea • Levi 301

20, 22) and even simpler than that of opening on (>ach side of the base. The

Araneiis. A complex terminal apophysis palpus of Metazijgia has a terminal apoph-

shiclds the embolus from above (A in Figs, ysis with a large proximal part and is very

58-62) and is connected to the bulb by different from that of Nwctenea. The palpal

distal hematodocha. The distal hema- femur lacks the basal ventral tooth present

todocha may be sclerotized, and reveals in Araneiis and Araniella but has a cor-

its origin by the presence of folds and rc\sponding tooth on the side of the endite.

grooves. Despite sclerotization, parts of Natural history. Nuctenea .species, at

it expand (Figs. 58, 61, 62). The embolus least those found in North America, may

lacks a cap and is a relatively simple be mature the year around; adult males

structure. The tiny structure visible on can be found at all seasons. In contrast,

the opening of the embolus (Figs. 119, 122, in all Araneus species observed in the

123, 125) is found only in mated males and temperate region, mature males can be

is presumably dried up sperm fluid. The found only during a short period of the

median apophysis is on the mesal side and year. Males, as well as females, can mate

may project (except in N. umhratica) . In numerous times; .species belonging to

most species it is biforked (Figs. 61, 62, Araneus, to judge by the work of U.

110-117, 126-129). It is a simple projec- Gerhardt (Levi, in preparation), can mate

tion in N. silvicultrix (Figs. 55, 60) and a only twice, once with each palpus, perhaps

lancet-shaped, appressed sclerite in N. three times if the mating was imsuccessful.

umhratica (M in Fig. 58). Males have a No doubt these differences in habits are a

liook on the distal margin of the first result of the cap, which is found on the

coxa and a corresponding depression on embolus of Araneus, and is absent in Nu-

the second femur; also the second tibia ctenea.

may be swollen and bear macrosetae. The species all build in the evening and

Diagnosis. In the Americas the species are nocturnal. The webs of all the species

of Nuctenea can be confused only with have few radii (fewer than 20) with few

those of Metazijgia. However, the carapace viscous threads, widely separated; those

of Nuctenea is setose (Figs. 105-109), that of adult N. umhratica are separated by 10

of Metazijgia lacks setae. Unlike most mm or more. Thus the web, especially in

species of Araneus, Nuctenea species have wind, gives the impression of being a flimsy

the abdomen dorsoventrally flattened, oval structure. The center is small with rough

in outline, and widest in the middle with a threads and few scaffolding threads. Al-

dorsal folium. The venter is black between though usually made nearly vertical, the

genital furrow and spinnerets, enclosing a web may be horizontal (N. cormita, Plate

white, comma-shaped mark on each side 2). All species make a retreat or sit near

(Figs. 97-109). The cardiac mark on the the web during the day, in the web at

abdomen, if present, is dark, not light as night.

it often is in Araneus. The only Araneus In the northeastern United States the

species that have the abdomen oval are species arc commonly found on houses,

marked differently. More important, Nu- Comstock (1912, 1940) refers to these

ctenea genitalia are much more sclerotized, spiders as the House Araneas, J. H. Emer-

and the embolus lacks a cap. Externally ton calls them House Epeiras. But, as

the epigynum base seems more complex Comstock points out, they are often found

(Figs. 43-46, 50-54, 71-76, 81-92) than in suitable habitats far away from houses,

in Araneus and Metazijgia. The Metazijgia In America the species have a wide

epigynum lacks the scape present in Nuc- distribution, N. cornuta from the Arctic

tenea; in its place there is a ventrally to the tropics. Only N. sclopetaria ap-

extended, laterally flattened lobe with the pears introduced; the other two species,

302 Bulletin Museum of Comparative Zoology, Vol. 146, No. 6

though they are commonly found on build-
ings and in trash (Maps 2-4), as inti'o-
duced species often are, have a continuous
Holarctic distribution.

Misplaced species. Epeira carolinalis
Archer, 1951, Amer. Mus. Novitates, No.
1487: 40, fig. 57, ?, may belong to the
genus Metazygia. The male is not known.

Key to Species of Nuctenea

1 Male with median apophysis of palpus
spHt into two branches (Figs. 110-
117); female with scape of epigynum
originating at anterior of base (Figs.
71-91 ) 3

- Male with median apophysis of palpus
not split into two branches (Figs. 47,
48, 55, 56, 58, 60); female with scape
originating in center of base or lacking
scape (Figs. 43, 51); Eurasia, North
Africa 2

2(1) Terminal apophysis of male palpus a mas-
sive shield (Figs. 48, 49, 58); median
apophysis spindle-shaped (Figs. 47, 58);
epigynum with triangular scape pointing
posteriorly (Fig. 43); Europe and North
Africa umbratica

- Tenninal apophysis of palpus with teeth
(Figs. 55, 60); median apophysis a knob
(Figs. 55, 60); epig>'num with cone-
shaped base drawn out anteriorly ( Figs.
51, 53); northern. Central Europe to
Siberia silvicultrix

3(1) Males 4

- Females 7

4(3) Terminal apophysis a prong with a nar-
row neck (Figs. 61, 110); embolus with
tip hidden by a lamella (Figs. 61, 110,
118, 119), probably cosmopolitan _. cornuta

- Terminal apophysis a sclerotized lobe
without neck (Figs. 112-117); embolus
cylindrical (Figs. 112, 114, 116) 5

5(4) Embolus with a distal, set-off finger (Fig.

116); Central Europe to Iran ixobola

- Embolus cylindrical, pointed at end but
without set-off finger (Figs. 112, 114,
120-125) 6

6(5) Median apophysis massive with the two
prongs about of equal width in ventral
view (Fig. 113); conductor leaning to-
ward median apophysis (Fig. 113);
Holarctic _ patagiata

- Median apophysis more slender with "up-
per" prong narrower than "lower" one
(Fig. 115); conductor bending away from
median apophysis (Fig. 115); Holarctic
sclopetaria

7(3) Epigynal base posteriorly corrugated
(Fig. 84); scape widest near tip witli a
narrow neck (Figs. 78-81); Holarctic
patagiata

- Epigvnal base not corrugated (Figs. 74,
76, 88, 92) 8

8(7) Epigynum covered by a lamella on each
side (Figs. 65-76); base with a ven-
tral, anterolateral fold and swollen pos-
teriorly (Figs. 63, 65, 67, 69, 73, 75);
probably cosmopolitan cornuta

- Epigynal lamellae not visible in ventral
view or hidden behind base ( Figs. 85—
92); base without anterolateral fold and
posterior swelling 9

9(8) Anterior part of scape framed by a lip of
the base (Fig. 87); Holarctic sclopetaria

- Anterior part of scape not framed by lips
from base, base with anterolateral pockets
(Fig. 91); Central Europe to Iran _... ixobola

Nuctenea umbratica (Clerck)

Figures 43-49, 58, 59, 93, 99, 105

Araiieus timbraticus Clerck, 1757, Aranei Svecici,
p. 31, pi. 1, fig. 7, $ . Clerck's specimens from
Sweden in the Museum of Natural History,
Stockholm, labeled by Thorell; not examined.
Locket and Millidge, 1953, British Spiders, 2:
139, fig. 92, ? , $. Bonnet, 1955, Bibliographia
Araneonmi, 2: 621.

Aranea sexpunctata Linnaeus, 1758, Systema Nat-
urae, 10 ed., p. 622. Type specimens believed
lost. Wiehle, 1931, in Dahl, Tienvelt Deutsch-
lands, 23: 93, figs. 138-141, 9, $. Roewer,
1942, Katalog der Araneae, 1: 791.

Epeira (Nuctenea) umbratica, - Simon, 1864,
Histoire Naturelles des Araignees, p. 261.

Epeira umbratica, - Nielsen, 1932, The Biology of
Spiders, Copenhagen, vol. 2, figs. 299-304,
web, egg sac, retreat.

Cliinestcla umbratica, - Archer, 1951, Natur. Hist.
Misc., Chicago, 84. Proszynski and Stargga,
1971, Katalog Fauny Polski, 16: 85.

Description. Female from England. Cara-
pace dark brown. Legs dark brown, with
light bands. Dorsum of abdomen with
usual dark brown pattern. Venter black
with two lateral white marks. Diameter
of posterior median eyes 0.9 diameters of
anterior medians, laterals 0.8 diameters of
anterior medians. Anterior median eyes
their diameter apart, posterior medians
slightly more than their diameter apart.
Height of clypeus equals about the diam-
eter of the anterior median eyes. The abdo-

Orb-weaver Araniella and Nuctenea • Levi 303

Figures 43-49. Nuctenea umbratica (Clerck). 43 46. Epigynum. 43. Ventral. 44. Posterior. 45. Ventral,
cleared. 46. Posterior, cleared. 47-49. Left male palpus. 47. Mesal. 48. Ventral. 49. Apical.

Figures 50-57. N. silvicultrix (C. L. Koch). 50-54. Epigynum. 50. Anterior. 51. Ventral. 52. Posterior. 53.
Lateral 54. Posterior, cleared 55 57. Male palpus. 55. Mesal. 56. Ventral. 57. Tibial macrosetae.

Scale lines. 0.1 mm.

men is much flattened. Total length 12 Male from England. Coloration as in

mm. Carapace 4.7 mm long, 4.1 mm wide, female, with abdominal pattern more dis-

First femur, 3.9 mm; patella and tibia, 5.(S tinct. Diameter of posterior median eyes

mm; metatarsus, 3.6 mm; tarsus, 1.8 mm. 0.8 diameters of anteriors. Laterals very

Second patella and tibia, 5.3 mm; third, slightly smaller than posterior median eyes.

3.0 mm; fourth, 4.6 mm. Anterior median eyes their diameter apart,

304 BuUeiin Museum of Comparative Zoology, Vol. 146, No. 6

Plate 2. Nuctenea cornuta (Clerck). Above, female (Wisconsin). Horizontal web
with flies caught (Minnesota).

posterior medians their diameter apart.
The height of the clypeus is shghtly less
than the diameter of the anterior median
eyes. Total length 8 mm. Carapace 4.0
mm long, 3.6 mm wide. First femur, 5.2
mm; patella and tibia, 7.3 mm; metatarsus,
5.3 mm; tarsus, 2.0 mm. Second patella
and tibia, 5.9 mm; tliird, 3.0 mm; fourth,
4.7 mm.

Diagnosis. Nuctenea umbratica is read-

ily separated from the other species of
Nuctenea by being flatter (Figs. 93, 105),
and by the distinct genitalia (Figs. 43-48).
The openings of the epigynum, unlike
those of other species, are anterior on the
base (Figs. 45, 46) and the median
apophysis is spindle-shaped (Figs. 47, 48,
58).

Natural history. Nuctenea U7nhratica has
its retreat in crevices under bark, between

Orb-weaver Araniella and Nuctenea • Levi 305

Nuctenea cor

Map 2. North American distribution of Nuctenea cornuta (Clerck).

planks, in masonry, in barns, houses and
greenhouses, up to 820 m in tlie Alps, but
in southern Switzerland to 1200 m (Wiehle,
1931). Mature females ean be found at all
seasons, males from June until Oetober.
The eggs are laid in a flattened ball sur-
rounded by loose, woolly silk.

The web is more eccentric than that of

JV. patap,i(ita, with the center closest to the
retreat. The viscous threads may span 70
cm and there are about 20 radii. There is
a line to the retreat. The animal is strictly
nocturnal (Wiehle, 1927, 1931). Mating
has been described by Gerhardt, 1926.

Disirihution. Nuctenea uinbratica is
only found in Europe and North Africa.

306 Bulletin Museum of Comparative Zoology, Vol. 146, No. 6

Nuctenea silvicultrix (C. L. Koch),
new combination
Figures 50-57, 60, 106

Epeira silvicultrix C. L. Koch, 1845, Die Arachni-
den, 11: 131, pi. 932, 933, 9, $. In the Ber-
lin Museum are specimens from the L. Koch
collection, presumably from Nlirnberg, but no
specimens that can readily be interpreted as
types of C. L. Koch. The British Museum
(Natural History) has specimens from Nlirn-
berg belonging to the L. Koch collection and
presumably the types. They were not examined.

Aranea silvicultrix Wiehle, 1931, in Dahl, Tier-
welt Deutschlands, 23: 96, figs. 142-145, $,
$. Roewer, 1942. Katalog der Araneae, 1: 792.

Araneus silvicultor, - Bonnet, 1955, Bibliographia
Araneorum, 2: 598.

Epeira (Cyphepeira) silvicultrix, - Archer, 1951,
Natur. Hist. Misc., Chicago, 84: 4.

Cyphepeira silvicultrix, - Yaginuma and Archer,
1959, Acta Arachnol., 16: 41, fig. 12, $.
Proszynski and Stargga, 1971, Katalog Fanny
Polski, 16: 85.

Description. Female from Schonhaid
[near Neustadt, Bavaria]. Carapace red-
brown, head lighter, with a double, median,
longitudinal darker line. White setae in
the head region. The legs are indistinctly
banded with narrow, light bands. The
dorsum of abdomen has a black folium out-
lined by white. The venter is black with
a white mark on each side. The diameter
of the posterior median eyes is 0.8 diam-
eters of anterior medians, laterals about
0.6 diameters. The anterior median eyes
are about their diameter apart; the posterior
medians slightly less than one diameter.
The ocular quadrangle is slightly longer
than wide, much narrower behind than in
front. The clypeus slants back and its
height is about equal to or sHghtly less
than the diameter of anterior median eyes.
The opening of the epigynum appears to
be on the side of the base and is quite
difficult to see. Total length 7.0 mm.
Carapace 3.1 mm long, 2.8 mm wide. First
femur, 3.0 mm; patella and tibia, 3.9 mm;
metatarsus, 2.7 mm; tarsus, 0.9 mm. Second
patella and tibia, 3.6 mm; third, 2.1 mm;
fourth, 3.3 mm.

Male from Erlangen, Bavaria. Colora-
tion as in female. There is a shallow, round

thoracic depression. Posterior and anterior
median eyes subequal in size, laterals 0.8
diameters of medians. The anterior median
eyes are one diameter apart; posterior ■
medians also one diameter apart. The '
height of the clypeus equals slightly less
than the diameter of the anterior median
eyes. Total length, 5.8 mm. Carapace 3.2
mm long, 2.6 mm wide. First femur, 3.8
mm; patella and tibia, 5.0 mm; metatarsus,
3.7 mm; tarsus, 1.5 mm. Second patella
and tibia, 4.0 mm; third, 2.3 mm; fourth,
3.4 mm.

Diagnosis. The epigynum is triangular,
anteriorly extended (Figs. 50-54); the
terminal apophysis (Figs. 55, 60) and
strong setae on the palpal tibia (Fig. 57)
separate this species from other Nuctenea.

Natural history. In northern Bavaria
the species is found among lichens on
stunted pines growing on infertile, moist
ground; it uses crevices as retreats. Mature
males are collected in April, May and
again in July and August. The web is
similar to that of IV. umbratica (Wiehle,
1931).

Distribution. Norway, Finland, Germany,
Switzerland to Balkans, Ural Mountains
and Turkmen (Bonnet, 1955).

Nuctenea cornuta (Clerck), new
combination, Furrow Spider
Plate 2; Figures 61-76, 94, 97-98,
110-111, 118-119, 126; IVIap 2*

Araneus cornutus Clerck, 1757, Aranei Svecici, p.
39, pi. 1, fig. 11, 9 . Female types from Swe-
den lost. Locket and Millidge, 1953, British
Spiders, 2: 134, figs. 88a, 89b, 90c, ?, $.
Bonnet, 1955, Bibliographia Araneorum, 2: 463.

Aranea foliata Fourcroy, 1785, Entomologia Pari-
siensis, 533. Type specimen from Paris, France,

* Correction, July 1974. K. Thaler recently
found that what has been called Araneus cornutus
in Europe in fact represents two species, a nortliern
one and a southern one (Zool. Anz., in press). The
American specimens are all hke tlie northern
European N. cornuta. I have examined specimens
sent by K. Thaler and agree witli his conclusions.
Figures 63-66 of this paper thus do not belong to
N. cornuta.

Orb-weaver Araniella and Nuctenea • Levi 307

immM'

Figures 58 62. Nuctenea expanded left male palpus. 58, 59. N. umbratica (Clerck). 60. N. silvicultrix (C. L.
Koch). 61, 62. N. cornuta (Clerck).

Figures 63-70. N. cornuta (Clerck), epigynum, variation. 63, 65, 67, 69. Ventral. 64, 66, 68, 70. Posterior. 63
66. (Burgenland, Austria.) 67, 68. (Kamtchatka.) 69, 70. (Poland.)

Abbreviations. A, terminal apophysis; C, conductor; E, embolus; I, stipes; M, median apophysis; R, radix; S, sub-
tegulum; T, tegulum.

Scale lines. 0.1 mm.

308 Bulletin Museum of Comparative Zoology, Vol. 146, No. 6

Nuctenea patagi

Map 3. North American distribution of Nuctenea patagiata (Clerck).

believed lost. Wiehle, 1931, in Dahl, Tierwelt
Deutschlands, 23: 86, figs. 124-127, ?, $.
Roewer, 1942, Katalog der Araneae, 1: 800.

?Epeiia frondosa Walckenaer, 1841, Histoire Nat-
iirelle des Insectes, Apteres 2: 65. The type is
Abbot's manuscript illustration, fig. 326, from
Georgia in the British Museum, Natural History,
copy in the Museum of Comparative Zoology,
examined.

Epeira strix Hentz, 1847, J. Boston Natur. Hist.
Soc, 5: 473, pi. 31, fig. 5, 9. Type specimens
from Pennsylvania and Alabama destroyed.

Epeira vicaria Kulczynski, 1885, Pam Akad.
Umiej. Krakow, 11: 5. Female holotype from
Kamchatka in the Polish Academy of Sciences,
examined.

Aranea frondosa, - Comstock, 1912, Spider Book,
p. 487, figs. 104-106, 128, 186, 194, 509, 513-
516, 9, $, web; 1940, op. cit., rev. ed., p. 501,
figs. 104-106, 128, 186, 194, 509, 513-516, ?,
$ , web.

Epeira cornuta, - Nielsen, 1932, The Biology of
Spiders, Copenhagen, vol. 2, figs. 289, 290,
retreat.

Epeira foliata, - Kaston, 1948, Connecticut Geo!.
Natur. Hist. Surv. Bull., 70: 254, figs. 787, 803,
812, 2043, 9, $, web.

Ciiphepcira cornuta, - Yaginuma and Archer,
1969, Acta Arachnol., 16: 41. Proszynski and
Stargga, 1971, Katalog Fauny Polski, 16: 82.

'Note. Figures 67, 68 were prepared from
the holotype of A. vicaria.

Variation. Total length of females 6.5-
14.0 mm. Carapace 2.4-5.0 mm long, 1.9-
4.5 mm wide; first patella and tibia, 2.6-
5.8 mm long. The total length of males
4.7-8.5 mm. Carapace 2.1-4.2 mm long,
1.8-3.5 mm wide; first patella and tibia
3.0-6.0 mm. The smallest American sped-

Orb-weaveh Araniella and Nuctenea • Levi 309

mens come from Alaska, tlie largest from
the area of New England to Texas.

Dia<^n(ms. The two anterior lateral
lobes of the base of the epigynum are
bent over and face posteriorly (Figs. 63,
65, 67, 69, 73, 75). (Those of N. sclopetaria
are not bent over.) The posterior face of
the epigynum is smooth (Figs. 74, 76),
not grooved as in N. patagiota. The palpal
terminal apophysis has a prong with a tip
wider than its neck (Figs. 61, 110); in
N. sciopetarm and N. patagiata it is wide.
The embolus of N. cornuta has a wide
lamella toward the mesal side (Figs. 61,
110, 118, 119) while in N. sclopetaria and
N. patagiata it is simple and cylindrical.

Natural history. The web is found around
houses, frequently in bushes (Comstock,
1912), often near water (Wiehle, 1931;
Kaston, 1948); it has 15-20 spokes and is
up to 60 cm in diameter (Kaston, 1948).
It is illustrated by figure 516 in Comstock
(1912) and by Kaston (1948). The web
is made at night, when the spider leaves its
retreat. The silken retreat may be in
crevices of walls, on railings, or among
plants. The retreat has been illustrated
by Nielsen ( 1932 ) . Both sexes are mature
all summer, from March in the southern
part of the range; males are more com-
monly mature in spring and late fall
(Kaston, 1948). Egg sacs, according to
Kaston (1948), are 7-10 mm in diameter,
covered with yellowish threads, hidden in
the retreat, and contain 50 to 250 eggs. A
female may make as many as ten egg sacs.

Distribution. Holarctic, perhaps carried
by man worldwide. In North America it
occurs from the Arctic to Central America,
but is most common in the eastern U.S.
and Canada, Newfoundland to Florida
(Map 2).

Nuctenea patagiata (Clerck),
new combination
Figures 77-84, 100-102, 107, 112-113,
120-123, 127; Map 3

?Araneus ocellatus Clerck, 1757, Aranei Svecici, p.
36, pi. 1, fig. 9, 9 . Female holotype from

Sweden, lost. Bonnet, 1955, Bibliographia
Anmeonim, 2: 555.
.\iancu\ i>(iia^.i(itus Clerck, 1757, Aranei Svecici,
p. 38, pi. 1, fig. 10, 9. Locket and Millidge,
1953, British Spiders, 2: 136, figs. cS9c, 90b,

9, $.

Arcnica dumetorum Fourcroy, 1785, Entomologia
l^irisiensis, p. .534. Type from Paris, France,
belie\ed lost. Wiehle, 1931, in Dahl, Tierwelt
Deutschlands, 23: 88, figs. 128, 129, 9, $.

U.pcira itJiaca MeCook, 1893, American Spiders,
3: 152, pi. 4, fig. 3, $. Male lectotype from
Ithaca, New York, in the Academy of Natural
Sciences, Philadelphia, examined. NEW SY-
NONYMY.

Aranca ocellata, - Comstock, 1912, Spider Book,
p. 489, figs. 107, 517-518; 1940, op. cit., rev.
ed., p. 50.3, figs. 107, 517-518, 9, $.

Epeiia patagiata, - Nielsen, 1932, The Biology of
Spiders, Copenhagen, vol. 2, fig. 305, web.

Epeiia (luiiictoiiim, - Kaston, 1948, Connecticut
Ceol. Natur. Hist. Surv. Bull., 70: 255, figs.
788, 804, 813, 9, $.

Cyphcpcim patagiata, - Yaginuma and Archer,
1969, Acta Arachnol., 16: 41. Proszynski and
Stargga, 1971, Katalog Fanny Polski, 16: 84.

Note. According to Bonnet (1955), C.
L. Koch, 1845 (Die Arachniden, 11: 115)
was the first revisor, synonymizing the two
names of Clerck and choosing the name
patagiata.

Variation. Total length of females 5.5-
11.0 mm. Carapace 2.5-4.0 mm long, 2.2-
3.3 mm wide; first patella and tibia, 3.5-5.2
mm. Total length of males 5.8-6.5 mm.
Carapace 2.9-3.8 mm long, 2.3-3.1 mm
wide; first patella and tibia, 4.5-5.5 mm.

Diagnosis. Females are separated from
the other species by the epigynum: its base
is furrowed posteriorly and its scape has a
narrow neck (Figs. 78-84). The male dif-
fers from IV. cornuta by having the terminal
apophysis a flat lobe (as in N. sclopetaria);
it differs from both the other species in
having a deep division in the heavy median
apophysis (Figs. 112, 113, 127) and in
having a finger-shaped embolus ( Figs.
112,120-123).

Natural history. Kaston (1948) indicates
that its habits are similar to those of N.
cornuta. According to Wiehle ( 1931 ) the
web has 20-24 spokes with about 16
viscid threads above, and 23 below center;

310 Bulletin Museum of Comparative Zoology, Vol. 146, No. 6

Map 4. North American distribution of Nuctenea sclopetaria (Clercl^).

the orb is 25 cm across. The retreat is less
silk lined than that of N. cornuta. The web
has been illustrated by Nielsen (1932, fig.
305). I think the species prefers more
arid, shaded areas than N. cornuta. I have
found the retreat under bark in lodgepole
pine at Jackson Hole, Wyoming, in a rather
dry area. According to Wiehle ( 1931 ) and
Kaston (1948) there may be a signal
thread to the reti^eat or the spider may use
a radius to return to it.

Distribution. Some American authors
indicate that this is a more northern species
than N. cornuta. This may not be quite
correct; however, the species is not found
in die southeastern states. It appears to be
common from the Arctic to North Carolina
and Arizona but is much commoner than
N. cornuta in the west, the Rocky Moun-
tains, and the Pacific northwest states
( Map 3 ) ; also Eurasia.

Nuctenea sclopetaria (Clerck),
new combination
Bridge spider, Gray-cross spider
Figures 85-88, 103-104, 108, 114-115,
124, 125, 128; IVIap 4

Araneus sericatus Clerck, 1757, Aranei Svecici, p.
40, pi. 2, fig. 1, 9 . Female type from Sweden

lost. Bonnet, 1955, Bibliographia Araneorum,
2: 594.

Araneus sclopetaritis Clerck, 1757, Aranei Svecici,
p. 43, pi. 2, fig. 3, $ . Type specimen from
Sweden lost. Locket and Millidge, 1953, Brit-
ish Spiders, 2: 136, figs. 88b, 89a, 90a, 9, $.

Aranea undata Olivier, 1789, Encycl. Method. Hist.
Nat. Ins. Paris, 4: 206. New name for sclope-
tariits Clerck, but preoccupied by DeGeer,
1778. Wiehle, 1931, in Dahl, Tierwelt Deutsch-
lands, 23: 90, figs. 130-133.

Aranea ovigera Panzer-, 1804, Syst. NomencL, in
Schiiffer, Icon. Ins. Ratisb., 1: 244, pi. 174,
fig. 3.

Aranea sericata, - Comstock, 1912, Spider Book,
p. 486, figs. 510-512, 9,5; 1940, op. cit., rev.
ed., p. 500, figs. 510-512, 9, $.

Aranea ovigera, - Roewer, 1942, Katalog der
Araneae, 1: 801.

Epeira undata, - Kaston, 1948, Bull. Connecticut
Geol. Natur. Hist. Surv., 70: 256, figs. 789,
805, 814-815, 2044-2046, 2, S, web, egg sac.

Cyphepeira sclopetaria, - Yaginuma and Archer,
1959, Acta Arachnol., 16: 41.

Cyphepeira sericata, - Proszynski and Star§ga,
1971, Katalog Fanny Polski, 16: 84.

Note. According to Bonnet (1955), O.
P.-Cambridge (1874, Trans. Linnean Soc,
30: 330) first synonymized the two simul-
taneously, published names sericatus and
sclopetarius, and chose sclopetarius. How-
ever, this seems to be an error; the names
were first synonymized by Westring ( 1851,

Orb-weaver Araniella and Nuctenea • Levi 311

Figures 71-76. Nuctenea cornuta (Clerck), epigynum. 71, 73, 75. Ventral view. 72, 74, 76. Posterior view.
71-72. Cleared. 71-74. (Panama Canal Zone.) 75,76. (Alberta.)

Figures 77-84. W. parag/afa (Clerck), epigynum. 77-81,83. Ventral. 82,84. Posterior. 81,82. Cleared. 77.
Probably epigynum before last molt. (South Dakota.) 78, 79. (Alberta.) 80. (British Columbia.) 81 84. (On-
tario.)

Figures 85-88. N. sclopetaria (Clerck), epigynum. 85, 87. Ventral. 86, 88. Posterior. 85, 86. Cleared.

Figures 89-92. N. ixobola (Thorell), epigynum. 89, 91. Ventral. 90, 92. Posterior. 89, 90. Cleared.

Scale lines. 0.1 mm.

312 Bulletin Museum of Comparative Zoology, Vol. 146, No. 6

Goteborg Kongl. Vet. Hand!., 2: 34), who
also chose sclopetaria. Thorell (1856, N.
Acta Reg. Soc. Sci. Uppsala, p. 22) also
lists the synonymy under sclopetaria.
Bonnet shows that prior to 1938 the usage
of sclopetaria outweighed sericata, although
sericata has been uniformly used in North
America and also by Bonnet ( 1955 ) . I will
follow European arachnologists and use
sclopetaria as do Locket and Millidge
( 1953, British Spiders, 2). (See Article 24a
of the International Code on Zoological
Nomenclature, 1961.)

Variation. Total length of females 8-14
mm. Carapace 3.9-4.3 mm long, 3.1-4.0
mm wide. First patella and tibia 5.3-7.0
mm. Total length of males 6-7 mm. Car-
apace 3.7-3.2 mm long, 2.9-3.3 mm wide.
First patella and tibia 6.2-7.0 mm.

Diagnosis. This species can usually be
separated from N. cornuta and N. patagiata
by the white hairs around the border of
the carapace and by the fact that the
background of the eye region is lighter
brown than the area behind it and than
the area of the clypeus (Fig. 108). The
female's epigynum has the scape finger-
shaped, as in N. cornuta but not as in N.
patagiata, and the anterolateral margins of
the base lobed and flat ( Fig. 87 ) , not as in
N. cornuta. The openings of the epigynum
are in dark posterior swellings of the base
(Fig. 85). The palpus has a lobe-shaped
terminal apophysis (Figs. 114, 115), some-
what like that of N. patagiata. The median
apophysis is not as deeply divided as that
of N. patagiata (Figs. 115, 128). The
embolus resembles that of N. patagiata
(Figs. 124, 125).

Natural history. This is the least common
of the three Nuctenea species in North
America and is found on houses and other
buildings, often near water. One collec-
tion from West Virginia came from sweep-
ing honeysuckle bushes (Lonicera sp.).
Many webs may be found touching one
another (Kaston, 1948). In Europe the
species is also found on buildings, and
especially on bridges and cliffs above

water (Wiehle, 1931). The orbs have up
to 20 radii, with the viscid spiral separated,
and the web reaches 70 cm in diameter
(Wiehle, 1931). The web is illustrated in
Kaston (1948, figs. 2044, 2046). The spider
rests near the end of one of the frame
threads rather than in a retreat. The egg
sac, according to Kaston, contains 114-337
eggs, and is illustrated in his figure 2045.
Distribution. Eurasia. In America this
species is probably introduced, judging by
its close association with buildings and
its limited distribution, which matches that
of Araneus diadematus Clerck. It is found
from Newfoundland to southern Alaska,
south to North Carolina, and is most abun-
dant in the Great Lakes states ( Map 4 ) .

Nuctenea ixobola (Thorell)
new combination
Figures 89-92, 95, 109, 116-117, 129

Epeim ixobola Thorell, 1873, Remarks on Syn-
onyms of European Spiders, p. 545. Two male
and three female syntypes from Austria in the
Natural History Museum in Stockholm, not
examined.

Aranea ixobola, - Wiehle, 1931, in Dahl, Tier-
welt Deutschlands, 23: 92, figs. 134-137, 9, $.
Roewer, 1942, Katalog der Araneae, 1 : 788.

Araneus ixobolus, - Bonnet, 1955, Bibliographia
Araneorum, 2: 523.

Description. Female from Poland. Car-
apace very flat, no thoracic depression.
Secondary eyes all about 0.7 diameter of
anterior median eyes. Anterior median
eyes slightly more than their diameter
apart, posterior median eyes their diameter
apart. Lateral eyes on tubercles and widely
separated from the median eyes. The
height of the clypeus equals 0.6 diameters
of the anterior median eyes. Total length
13 mm. Carapace 6.5 mm long, 5.3 mm
wide. First femur, 6.3 mm; patella and
tibia, 8.6 mm; metatarsus, 5.7 mm; tarsus,
2.0 mm. Second patella and tibia, 8.0 mm;
third, 4.4 mm; fourth, 6.3 mm.

Male from Poland. Carapace and eye
arrangement as in female. Total length
12 mm. Carapace 5.4 mm long, 4.7 mm

Orb-weaver Araniella and Nuctenea • Levi 313

103

Figures 93 95. Eye region and chelicerae. 93. Nuctenea umbratica. 94. N. cornuta. 95. N. ixobola.

Figure 96. Left anterior femora and carapace of N. patagiata.

Figures 97-104. Abdomen. 97-98. N. cornuta. 99. N. umbratica. 100-102. N. patagiata. 103 104. N. sclo-
petaria. 97, 100, 101, 103. Dorsal. 98, 99, 102, 104. Ventral.

Figures 105-109. Carapace and abdomen, dorsal. 105. N. umbratica. 106. N. silvicultrix. 107. N. patagiata.
108. N. sclopetaria. 109. N. ixobola.

Scale lines. 1.0 mm.

314 Bulletin Museum of Comparative Zoology, Vol. 146, No. 6

wide. First femur, 8.3 mm; patella and
tibia, 11.8 mm; metatarsus, 8.2 mm; tarsus,
3.3 mm. Second patella and tibia, 10.0
mm; third, 5.0 mm; fourth, 7.4 mm.

Diagnosis. Females have a narrow
scape in the epigynum as does Nuctenea
sclopefaria, with which it has been con-
fused. However, the anterior lateral end
of the base differs (Figs. 89, 91) and
posterior lobes of the base do not extend
ventrally. The male differs from that of
N. sclopetaria in having a differently
shaped embolus (Fig. 116).

Natural history. According to Wiehlc
(1931), this species is similar in habits
to N . sclopetaria and replaces it in eastern
Europe. It lives on buildings, fences, and
bridges near water.

Distribution. From Central Europe to
Iran (Roewer, 1942; Bonnet, 1955). Speci-
mens examined came from Leopoldshall
(Anhalt, German Democratic Republic),
Biolowieza, Distr. Hajnowka, and Distr.
Kosice, Poland.

REFERENCES CITED

Berman, J. D., AND H. W. Levi. 1971. The
orb-weaver genus Neoscona in North America
(Araneae: Araneidae). Bull. Mus. Comp.
Zool., 141: 465-500.

Cambridge, F. O. Pickard-. 1903. Araneidea,
In Biologia Centrali-Americana, 2.

CoMSTOCK, J. H. 1912. The Spider Book.
Garden City, New York: Doubleday, Doran
and Co.

. 1940. The Spider Book, rev. ed.

Ithaca: Conistock Publ. Co.

Gerhardt, U. 1926. Weitere Untersuchungen
zur Biologie der Spinnen. Z. Morphol. Okol.
Tiere, 6: 1-77.

Gertsch, W. J. 1934. Further notes on Amer-
ican spiders. Amer. Mus. Novitates, No.
726: 1-26.

Levi, H. W. 1971. The Diademattts group of
the orb-weaver genus Araneus north of Mex-
ico (Araneae: Araneidae). Bull. Mus. Comp.
Zool., 141: 131-179.

. 1973. Small orb-weavers of the genus

Araneus north of Mexico (Araneae: Aranei-
dae). Bull. Mus. Comp. Zool. 145(9):
473-552.

IX PREPARATiox. The presence of the

cap on palpal emboli and mating behavior.
Locket, G. H., and A. F. Millidge. 1953.

British Spiders, 2: 1-449.
Petrunkevitch, a. 1925. External reproductive

organs of the common grass spider Agelena

naevia Walckenaer. T- Morphol., 40: 559-

573.
Proszynski, J., and W. Starega. 1971. Katalog

Fauny Polski, 33: 1-382. '
Seyler, p. J. 1940. The generic and specific

status of four spiders of the genus Agelenop-

sis. Ohio J. Sci., 41: 51-69.
Wiehle, H. 1927. Beitrage zur Kenntnis des

Radnetzbaues der Epeiriden, Tetragr.athiden

und Uloboriden. Z. Morphol. Okol. Tiere,

8: 468-537.
. 1931. Araneidae. In F. Dahl, Die

Tierwelt Deutschlands, 23: 1-136.
Yaginu.ma, T., and a. F. Archer. 1959. Genera

of the Araneine Argiopidae foimd in the

Oriental region and generally placed under

the comprehensive genus Araneus. Acta

Arachnol., 16: 34-41.

INDEX

Valid names are printed in italics. Page numbers refer to main references, starred page numbers to
illustrations.

alba, Epeira 296

alpica, Aranea 299

alpica, Araniella 299*

alpica, Epeira 299

alpicus, Araneus 300

Araniella 292

cornuta, Cyphepeira 308

corniita, Nuctenea 304*, 306, 307*, 311*, 313*, 315*

cornutus, Araneus 306

croaticus, Araneus 296

cucurbitina, Aranea 298

cucurhitina, Araniella 297*, 298

cucurbitina, Epeira 296

cucurbitinus, Araneus 298

Cyphepeira 300

decipiens, Epeira 294

displicata, Aranea 296

(lisplicata, Araniella 293*, 294, 295*, 297*

displicata, Epeira 294

displicatus, Araneus 296

dumetorum, Aranea 309

dumetorum, Epeira 309

OlUJ-\\'KA\'KH ArANIKLLA AND NuCTENEA • Lcvi 315

126ij

Figures 110-117. Left male palpus. 110, 111. Nuctenea cornuta (Clerck). 112, 113. N. patagiata (Clerck). 114,
115. A/, sc/ope/ar/a (Clerck). 116,117. N. ixobola (ThoreW).

Figures 118-125. Embolus of male palpus. 118,119. N. cornuta. 120 123. N. patagiata. 124, 125. N. sclope-
taria. 118,120,121,124. Probably virgin. 119,122,123,125. Probably mated. 118. Mesal-apical. 119 125.
Mesal.

Figures 126-129. Male palpus, apical view. 126. N. cornuta. 127. N. patagiata. 128. N. sclopetaria. 129. N.
ixobola.

Scale lines. 0.1 mm.

316 Bulletin Museum of Coinparative Zoology, Vol. 146, No. 6

foliata, Aranea 306

foliata, Epeira 308

frondosa, Aranea 308

frondosa, Epeira 308

inconspiciia, Aranea 298

inconspiciia, AranieUa 298, 299*

inconspicua, Epeira 298

inconspicuus, Araneus 298

ithaca, Epeira 309

ixobola, Aranea 312

ixobola, Epeira 312

ixobola, Nuctcnea 311*, 312, 313*, 315*

ixobolus, Araneus 312

Niictenea 300

ocellata, Aranea 309

ocellatus, Araneus 309

octopunctata, Araniella displicata 296

opisthographa, Araneus cucurbitina 298

ovigera, Aranea 310

patagiata, Cyphepeira 309

patagiata, Epeira 309

patagiata, Nuctenea 309, 311*, 313*, 315^

patagiatus, Araneus 309

proxima, Aranea 298

proxima, Epeira 298

sclopetaria, Cyphepeira 310

sclopetaria, Nuctenea 310, 311*, 313*, 315*

sclopetarius, Araneus 310

sericata, Cyphepeira 310

sericatus, Araneus 310

sexpunctata, Aranea 302

sexpunctata, Epeira 294

sihicultor, Araneus 306

sihicultrix, Aranea 306

silvicultrix, Cyphepeira 306

sil\icultrix, Epeira 306

silvicultrix, Nuctenea 303*, 306, 307*, 313*

strix, Epeira 308

lunbratica, Chinesteha 302

umbratica, Epeira 302

umbratica, Nuctenea 302, 303*, 307*, 313*

undata, Aranea 310

undata, Epeira 310

vicaria, Epeira 308

us ISSN 0027-4100

SuUetin OF the

Museum of

Comparative

Zoology

The Anatomy of Saurosuchus galilei and the
Relationships of the Rauisuchid Thecodonts

WILLIAM D. SILL

HARVARD UNIVERSITY

CAMBRIDGE, MASSACHUSETTS, U.S. A

VOLUME 146, NUMBER 7
21 NOVEMBER 1974

PUBLICATIONS ISSUED

OR DISTRIBUTED BY THE

MUSEUM OF COMPARATIVE ZOOLOGY

HARVARD UNIVERSITY

Breviora 1952-

BULLETIN 1863-

Memoers 1864-1938

JoHNSONiA, Department of Mollusks, 1941-

OccAsioNAL Papers on Mollusks, 1945-

SPECIAL PUBLICATIONS.

1. Whittington, H. B., and E. D. I. Rolfe (eds.), 1963. Phylogeny and
Evolution of Crustacea. 192 pp.

2. Turner, R. D., 1966. A Survey and Illustrated Catalogue of the Teredini-
dae (Mollusca: Bivalvia), 265 pp.

3. Sprinkle, J., 1973. Morphology and Evolution of Blastozoan Echinoderms.

284 pp.

4. Eaton, R. J. E., 1974. A Flora of Concord. 250 pp.

Other Publications.

Bigelow, H. B., and W. C. Schroeder, 1953. Fishes of the Gulf of Maine.
Reprint.

Brues, C. T., A. L. Melander, and F. M. Carpenter, 1954. Classification of
Insects.

Creighton, W. S., 1950. The Ants of North America. Reprint.

Lyman, C, P., and A. R. Dawe (eds.), 1960. Symposium on Natural
Mammalian Hibernation.

Peters' Check-list of Birds of the World, vols. 2-7, 9, 10, 12-15.

Proceedings of the New England Zoological Club 1899-1948. (Complete
sets only.)

Publications of the Boston Society of Natural History.

Price list and catalog of MCZ publications may be obtained from Publications
Office, Museum of Comparative Zoology, Harvard University, Cambridge, Massa-
chusetts, 02138, U.S.A.

THE ANATOMY OF SAUROSUCHUS GALILEI AND THE
RELATIONSHIPS OF THE RAUISUCHID THECODONTS

WILLIAM D. SILL'

Abstract. Saitrosuclius galilei was a large
quadrupedal carnivorous thecodont from the
Ischigualasto Formation of western Argentina,
which is of approximately Carnian age. Its skull
anatomy indicates that it descended from an
erythrosuchid t\'pe of primitive thecodont. Sauro-
suchus, together with Luperosuchtts, Prestosuchiis,
Ticinos'UcJius, "Mandasiichus" and possibly some
other less well known genera, form a well-defined
lineage that can be trticed throughout most of the
Triassic. Rauisuchus is considered a member of
the same family, and thus the earlier name
Rauisuchidae is retained for the group. Two other
thecodont lineages, the Proterochampsidae and the
Ornithosuchidae, are traced throughout the Tri-
assic. The relationships of the three families
strongly suggest that they are independent deri-
vations of the three Early Triassic primitive
families. Dinosaur origins remain unclear. There
is no good evidence for associating the Raui-
suchidae with early dinosaurs; on the contrary,
there is an unexplained time oxerlap of large
carnivorous dinosaurs and thecodonts that have
nearly identical adaptations.

INTRODUCTION

Saurosiichus g,aUlei is one of the 18 or
more genera of reptiles found in the now
legendar}' Ischigualasto Basin of western
Argentina. Its significance for paleonto-
logic studies lies in the excellent preserva-
tion of the material, particularly of the skull
and tarsus, which makes possible the clari-
fication of the anatomy of the closely
related Brazilian thecodonts, and generally
aids interpretation of the family Raui-
suchidae on a worldwide basis. Together

' Universidad N'acional de San Juan, Dept. Geo-
logia, San Juan, Argentina.

Bull. Mus. Comp.

with Ticinosuchus from the Middle Triassic
of Switzerland, Saurosiichtis provides a key
for tracing a thecodont lineage that was
world-wide in distribution throughout most
of the Triassic Period.

Most of the specimens used for this study
were collected in the Ischigualasto For-
mation by expeditions from the Instituto
Miguel Lillo of Tucuman, Argentina. The
first specimen was collected in 1959, under
the direction of Dr. Osvaldo Reig. Sub-
sequent expeditions, led by Jose Bonaparte,
recovered parts of four additional individu-
als. From these various parts, most of the
skeleton can be reconstructed, although the
forelimb is not represented in any of the
specimens.

Saurosiichus was one of the largest the-
codonts of its time, and no doubt competed
with the emerging dinosaurs for the large
carnivore role. Thecodonts, of course, lost
the competition, and contemporary dino-
saurs, both saurischian and ornithischian,
from the Ischigualasto Formation indicate
that superior locomotion was a factor
related to dinosaurian dominance. At pres-
ent, although Saurosuchus appears to be
the most advanced member of the family
yet described, it is less progressive ana-
tomicalK' than its dinosaiuian contempora-
ries. The lineage of Saurosuclius proxides
e\idence to support the premise that pro-
gressive thecodonts were competitors rather
than progenitors of the dinosaurs.

Abbreviations for the institutions referred
to in this report are as follows:

Zool., 146(7): 317-362, November, 1974 317

318 Bulletin Museum of Comparative Zoology, Vol. 146, No. 7

PVL Institute Miguel Lillo, Tucuman,
Argentina

DGM Division of Mines and Geology,
Brazil

T University of Tubingen, Ger-

many

PIMZ Paleontological Institute, Zurich,
Switzerland

MSJ Museum of Natural Sciences, San
Juan, Argentina

Acknowledgements. This study was made
during a year's stay at the Institute Miguel
Lillo in Tucuman, Argentina. Special
thanks are due to Jose Bonaparte and the
directors of the Institute, whose help and
generosity made the study possible. I am
also greatly indebted to A. W. Crompton of
Harvard University and John Ostrom of
Yale for their technical help and personal
assistance. Many colleagues offered sug-
gestions and gave perspective to the re-
search; among them were A. S. Romer,
Bernard Krebs, Alan Charig, A. Keyser, and
Alick Walker. Drawings were made by
Alexander Gavriloff.

Funds for the Research were provided by
NSF Grant GB-4435X1.

Geologic Setting

The Ischigualasto Basin (Hoyada de
Ischigualasto or Valle de la Luna) fomis
a depression on the western limb of a large
syncline whose axis runs northwest-south-
east. Differential erosion of the soft clay-
stones of the Ischigualasto Formation cre-
ated a prominent depression at the base of
the cliff-foiTning red sandstones of the Los
Colorados Formations (see Fig. 1). The
Triassic sediments extend approximately
one hundred kilometers, and are bounded
on the south by the Valle Fertil mountains
and on the north by the Sierra de Mas
range. Within this area of outcrop there
are numerous minor flexures, principally
anticlines. One such saddle-shaped anti-
cline divides the basin into a northern and
southern portion; this division coincides
with the boundary between the provinces

of San Juan and La Rioja. The southern,
or San Juan, portion is the larger of the two
and has produced most of the fossils known
from the basin. East of the depression, the
opposite limb of the large syncline has ex-
posed the type area of the earlier Chaiiares
Formation.

Interpretation of the time-stratigraphic
relationships of the sedimentary units in the
Ischigualasto basin has varied considerably.
For many years the whole succession was
considered "Rhaetic," or uppermost Tri-
assic. With the discovery of vertebrate
fossils that were more primitive than the
classic Upper Triassic fauna, vertebrate
paleontologists assigned the Ischigualasto
Formation to the Middle Triassic. As new
discoveries are being made a consensus is
forming that the Ischigualasto Formation
is most probably of Camian age, possibly
Late Ladinian, with the underlying Los
Rastros Formation closely equivalent in
time to the Santa Maria Formation of
Brazil. (I have elsewhere summarized the
various interpretations of the South Ameri-
can Triassic: Sill, 1969.)

Although the general geologic relation-
ships between the various formations are
quite straightforward, no attempt has yet
been made to study sedimentary cycles
within the Ischigualasto Fonnation, or to
correlate the occurrence of specific fossils
with different sedimentary regimes.

TAXONOMY AND MORPHOLOGY

Introduction

Taxonomic history of the Rauisuchidae
began with Huene's work on the specimens
he found in the Triassic of Brazil. In a
short paper on thecodont relationships
(Huene, 1936), he proposed the subfamily
Rauisuchinae as a part of the family
Stagonolepidae to include tlie genera
Rauisuchus and Prestosuchus from Brazil.
Later, (Huene, 1942) the group was ele-
vated to familial rank and the genus
Rhadinosuchus, also from the Triassic of
Brazil, was included. At about the same
time (Huene, 1938), he described Stagono-

Saurusuciius and the Rauisuchid Thecodoxts • Sill 319

comes

K ilometers

Figure 1. Generalized geologic map of the southern portion of the Ischigualasto Basin.
Saurosuchus localities.

marks

320 Bulletin Museum of Comparative Zoology, Vol. 146, No. 7

suchus from the Manda Beds of East Africa
and noted its similarity to the Brazilian
forms. However, not until 1956 did Huene
formally place StagonosucJius in the family
Rauisuchidae, at which time he also in-
cluded a number of poorly known theco-
donts that are no longer considered to be
closely related to the family.

Since then, intei"pretations of tlie broader
relationships of the Rauisuchidae have
followed the general pattern of uncertainty
that has been the hallmark of thecodont
taxonomy. Huene ( 1956 ) continued to
maintain the family in close association
with the stagonolepid-aetosaurid groups
and included in the family such diverse
genera as Cerritosaurus and Episcoposau-
riis. Romer (1956) was the first to separate
most of the genera of the Rauisuchidae
from the armored thecodonts, and tenta-
tively placed Raiiisuchus, Prestosuchus,
StagonosucJius, RJmdinosuchus, and Pro-
cerosuchus in the Ornithosuchidae. Hoff-
stetter ( 1955 ) retained the family in the
Stagonolepoidea, but removed Stagono-
suchus to the Stagonolepidae. Reig (1961)
presented a comprehensive review of the
family and showed beyond reasonable
doubt that the family Rauisuchidae should
consist only of the genera Rauisuchus,
Prestosuchus, Stagonosuchiis, and the then
recently discovered Saurosuchus from Is-
chigualasto. He also presented convincing
evidence showing that the family is not
closely related to the Stagonolepidae, and
placed it in the "traditional" thecodont
group which he termed Ornithosuchia (the
equivalent of Pseudosuchia of most au-
thors ) . Hughes ( 1963 ) , on the other hand,
tentatively placed Rauisuchus and Sauro-
suchus in the primitive thecodont group
Proterosuchia as members of the Erythro-
suchidae, a ranking that has not been
accepted by the majority of paleontologists.
Ticinosuchus, on the basis of a complete
skeleton, was added to the family by Krebs
( 1965 ) ; its affinities with the other mem-
bers of the family as described by Reig
are evident, A further genus, Luperosuchus,

from the Chaiiares Formation, was added
to the family by Romer (1971a), and a
closely related form has recently been found
in the Los Colorados Formation (Bona-
parte, personal communication). These
latter discoveries are especially significant,
for they permit the Saurosuchus lineage to
be traced through the major part of the
Triassic in a single basin of deposition.
Romer ( 1966 ) followed Hughes in tenta-
tively associating Rauisuchus and Sauro-
suchus with the Erythrosuchidae, and
adopted the term Prestosuchidae from
Charig's unpublished thesis for the remain-
ing genera Prestosuchus, Procerosuchus,
"Mandasuchus"^ and, tentatively, Stagono-
suchus. However, he later ( 1968 ) replaced
Rauisuchus and Saurosuchus with the
above mentioned forms, but did not sup-
press Prestosuchidae. Meanwhile Presto-
suchidae was carried on by Charig- ( 1967),
who notes that the group is essentially the
same as the Rauisuchidae of Huene (1942)
but with the genus Rauisuchus excluded.

In a more recent work on thecodont
taxonomy Romer ( 1972a ) continued to use
the family name Prestosuchidae on the
grounds that Rauisuchus was too poorly
known. However, he included Rauisuchus
within the family Prestosuchidae (see Dis-
cussion with regard to the affinities of
Rauisuchus) .

Assignment of tlie Rauisuchidae to a
suborder is difficult given the present un-
stable nature of thecodont taxonomy.
Romer ( 1972a ) places the family with the
primitive thecodonts in the Proterosuchia;
other authors, Charig (1967) and Bonaparte
(1971) place it in the usual "catch-all" sub-
order Pseudosuchia. Rauisuchids certainly
appear to have been derived from the
erythrosuchid lineage of the Proterosuchia
(see discussion on thecodont phylogeny).

^ Mandasuchus is technically a nomen nudum,
as it has never been described in print.

- In Charig's paper, origin of the Prestosuchidae
was ascribed to Charig 1967. This paper has not
been published. In an erratum, the family name
was given as Romer 1966.

Saurosuchus and the Rauisuchid Thecodonts • Sill 321

but they are much more spcciaHzed and
progressive than any of its known members.
On the other hand, they do not ha\'e a
great deal in common with the "tv'pical"
ornithosuchid pseudosuchians. As thecodont
relationships become more clearly under-
stood, a new suborder will probably have
to be erected for this and perhaps other
lineages descended from the erx'thro-
suchids, but at present such a step would
be premature.

Discover}' of nearly complete remains of
Ticinosuchtis and Saiirosrichus, represent-
ing what appear to be the earliest and the
latest members of the lineage so far de-
scribed, has provided the means for an
accurate characterization of the family.
Basically, the new evidence tends to con-
firm the definitions of the family given bv
Krebs (1965) and by Reig (1961): Reig's
paper provides an excellent summary of the
taxonomic histor\' of the family and of the
Thecodontia in general.

The family ma}' be defined as follows:
Medium- to large-sized carnivorous qua-
drupedal thecodonts. Skull large, deep, orbit
keyhole-shaped, large elongate antorbital
fenestra, small crescent-shaped accessor}'
antorbital fenestra present in some genera,
teeth flattened, recurved, serrated. Pelvis
triradiate, acetabulum closed, ischium
elongated and rodlike, fused at the midline
along most of its length. Femur long,
slightly sigmoid, without a well-defined
fourth trochanter. Calcaneum and astraga-
lus articulate by a ball and socket joint,
the socket on the calcaneum, the ball on
the astiagalus. Five digits, fifth metatarsal
short and hooked. Many of these features
are generalized characteristics of the primi-
tive thecodonts which have been carried
over in the familv and are retained
throughout their knowii history.

Family distribution. Middle and Late
Triassic; Argentina, Brazil, East Africa,
Switzerland, possibly China. Family Raui-
suchidae Huene 1936 (as a subfamily);
genera Rauisuchus Huene 1936 Brazil, Pres-
tosuchus Huene 1936 Brazil, Stagonosuchus

Huene 193S East Africa, Saurosuchus Reig
1959 Argentina, Ticinosuchus Krebs 1965
Switzerland, Luperosuclius Romer 1971
Argentina, "Mamlasuchus" unpublished
thesis Charig 1956. A number of additional
genera are sometimes included in the
family (see Romer, 1966 and 1972), but
they are not well known. These additional
genera are: Cuijosuchus, HopUtosaurus,
Rhadinosuchus, Pallisteria, Spoiulylosoma,
Procerosuchus, Fenhosuchus.

Saurosuchus Reig 1959

Type species. Saurosuchus galiJei.
Di.striJ)ution. Late Ladinian or Carnian,
Ischigualasto Basin, Western ^\i-gentina.
Diagnosis. As for the species.

Saurosuchus galilei Reig 1959

Type. PVL 2062, nearly complete skull,
posteriormost portion missing.

Hypodigm. The t}'pe and: P\T. 2198,
partial maxilla, left ilium, both ischia, nine
articulated dorsal vertebrae and fragments,
part of the dermal armour, associated ribs
and teeth. PVL 2557, two dorsal vertebrae,
both sacrals, nine caudals, right ilium and
ischium, partial pubis, parts of right femur,
tibia, fibula, complete right tarsus and foot,
associated ribs and chevrons. PVL 2267,
poorly preserved partial ilium, partial
femur, tibia, fibula, well-preserved tarsus,
partial foot. PVTL 2472, poorly preser\'ed
cervical vertebra, tibia, astragalus. MSJ
102, fragment of maxilla and lower jaw.

Horizon. Apparently all levels of the
Ischigualasto Formation, San Juan province,
Argentina. The five specimens of Sauro-
suchus were collected from four localities,
all in the soutliern portion of the outcrop
area. The t\pe, P\'L 2062, consists of a
nearly complete skull and was found in
the upper third of the strata. The more
complete skeletons, P\T. 2198 and PVL
2557, came from the middle part of the
section, and the remaining two indi\'iduals,
PVL 2267 and 2472, wer(> found in the
lower third of the strata, as was MSJ 102
(see map for specific localities).

322 Bulletin Museum of Comparative Zoology, Vol. 146, No. 7

Emended diagnosis. Large carnivorous l^ut the occipital region and braincase are
thecodonts, up to six meters in length. Skull lacking. A fragment of the right maxilla
deep, elongate, finely sculptured, with of PVL 2198 is identical to the correspond-
keyhole-shaped orbit, large antorbital ing region of the type and allows assign-
fenestra elongated anteriorly, small cres- ment of the specimen to the genus with a
cent-shaped accessory antorbital fenestra considerable degree of confidence. The
present between premaxilla and maxilla, lower jaw is known only from a fragment.
Large elongate, nearly vertical external The skull is long, approximately 65 centi-
nares bordered only by premaxilla and meters, triangular in shape, and .stiu-dily
nasal. Teeth robust, recurved, laterally com- constructed. The cranial table is high and
pressed with serrate edges. Four teeth on narrow. Orbits are large, keyhole-shaped
premaxilla, ten on maxilla. Strong orbital openings, of which the upper part is a
arch fomied by the frontal, small supra- well-defined circle high up the side of the
temporal fenestra lying in dorsal plane of skull. A large antorbital fenestra is present,
the skull below the crest of the orbital arch, subtriangular in shape and slightly smaller
Vertebrae amphicoelous, spines broad and than the orbit. It is surrounded by a well-
flat with prominent interspinous notch on defined smooth border set in from the
anterior face. Cervicals apparently elon- sculptured surface of the maxilla. An un-
gated, dorsals strongly compressed laterally, expected feature is the presence of a nar-
rib facets well separated and on different row accessory antorbital fenestra located
levels throughout column. Two sacral between the maxilla and the premaxilla,
vertebrae. Shoulder girdle and forelimb extending from above the tooth-bearing
unknown. Pelvis with closed acetabulum, surface to the posterior tip of the external
pubis almost excluded. Ilium with broad nares.

brevis shelf, ischium long, rodlike, ex- Like the antorbital fenestra, the external

panded at the tip and fused at the midline nares are subtriangular in shape, relatively

along most of its length. Femur slightly large, and situated principally in the verti-

sigmoid, without a large greater trochanter, cal plane of the skull. Notable for their

and with a small fourth trochanter. Fibula small size are the supratemporal fenestrae,

bears a prominent iliofibularis tubercle, which lie in the horizontal plane of the

Tarsus of the "crocodiloid" type, calcaneum skull roof just behind and slightly below

bearing a large tuber and a prominent the heavy orbital arch. Only the anterior

medial socket for articulation of the astra- border of the infratemporal fenestra is

galar ball. Facets for articulation of the preserved, but it indicates a triangular or

tibia and fibula close together. Fourth subrectangular shape approximately the

tarsal large, subtriangular with prominent same size as the orbit,

rounded facet for articulation of fifth meta- The large size of the skull and its sturdy

tarsal. Five digits on foot, first two most construction indicate that Saurosiichus was

robust, third is the longest, fifth is broad, an active predator. Using the head size

flat, and oriented outward. Demial armour index of skull length to length of the pre-

present, two rows of small scutes along sacral vertebral column, a value of either

each side of most of the vertebral column, .27 or .34 is obtained, the latter calculated

leaf-shaped and imbricating. on the assumption that neck vertebrae were

_ , _ . ^, approximatelv the same length as the

General Description / , , .; ,, „ ^.

dorsals, while the smaller ratio assumes

Skull elongated cervicals. Both indices are in

Cranium. Cranial material is represented the range of the large predaceous dino-

almost exclusively by the type, in which saurs; AUosaurus is .28, Tyrannosaunis

most of the dermal elements are preserved, is .41,

Saurosuciius and tiie Rauisuchid Thecodonts • Sill 323

Prcinaxilld. Both promaxillae of tlic type
arc complete and well piescived. The main
body of the bone is a massive rectangle
from which a slender process extends up-
ward and backward around the external
naris to a long o\'erlapping contact with a
similar process of the nasal, and a second
rodlike extension that forms the entire
lower border of the naris and terminates
wx^dged between the nasal and the maxilla.
At its anterior border the premaxilla forms
a straight \'ertieal line from the tip of the
naris to the first tooth position. Below the
narial opening the bone swells to a thick,
slightly undulating ridge that bears four
large teeth. At the most anterior part, just
above the toodi row, lie three foramina.
No sculpturing is present. The rodlike
process that forms the lower border of the
naris is an isolated structure that separates
the accessory antorbital fenestra and the
external naris.

Medially, the premaxillae meet in a long
sturdy symphysis. The ah'eolar margin is
thick and slightly vaulted behind the first
two teeth. Of the four teeth, the third is
the largest. Two deep pits are present in
the ^ aulted area, one beside the second
tooth, the other between the third and the
fourth. A large foramen is present above
the third alveolus. The interalveolar septum
between the third and fourth teeth is ex-
panded on the lingual surface to form a
small interdental plate.

Posteroventrally, a clearly defined suture
is not present between the maxilla and the
premaxilla, but above the thick tooth-bear-
ing portion of the bone the accessory antor-
bital fenestra serves to separate the two
elements.

Maxilla. The maxilla is a large platelike
bone that slopes posteriad and upward
from its suture with the premaxilla to meet
the nasal and lacrimal dorsally and the
jugal ventrally. It is deeply emarginated
by the antorbital fenestra, around which
runs a broad smooth shelf. Outside the
shelf area the maxilla is heavily sculptured
by an irregular network of grooves. It fonus

Tahle 1. Measurements of the skull (in

centimeters) of SAUROSUCHUS GALILEI BASED ON

the type PVL 2062. Note, further preparation

HAS modified some OF THE MEASUREMENTS MADE
HY ReIG (1959) IN HIS Pl^ELlMINAUY ACCOUNT.

Total length of the skull (estimated) 67

Length from tip of snout to anterior border of

the supratemporal lenestra 54

Lengtli from lower anterior corner of infra-
temporal fenestra to tip of snout 47

Diameter of tlie upper portion of the orbit ... 10

Maximum lieight of the orbit _._ 17

Maximum lengtli of the antorbital fenestra - . 19
Maxinnun height of the antorbital fenestra .- 8
Maximum length of the depression smround-

ing the antorbital fenestra 21.5

Maximum height of the depression surround-
ing tlie antorbital fenestra __- 10

Nhiximum lieight of the skull betA\'een top of
the rim of the orbit and bottom of jugal — . 20.5

Length of nasals along tlie midline 32

Length of tlie preniaxil!ar>' tooth row 9

Length of tlie niaxillar\' tooth row 27

Length of the external naris 12

Distance from tip of snout to anterior border

of the antorbital fenestra 21

W'idtli of skull across the supratemporal fenes-

trae ---.. 17

Widtli of skull in front of the orbits 10

Length of teeth alveoli

Premaxilla Left Right

1. 1.5 —

2. 1.5 1.6

3. 2.0 1.8

4. 1.3 1.5

Maxilla

1. 1.8 1.5

2. 2.4 2.3

3. 3.0 3.0

4. 2.6 2.3

5. 2.7 2.4

6. 2.2 2.3

7. 2.3 2.1

8. 1.9 —

9. 1.8 —
10. 1.8 —

Length of maxillar\- teetii, left side, from lateral
edge of the maxilla to die tip of the teeth

Anterior Posterior

Tooth No. curvature curvature

3. 4.6 3.5

5. 5.8 4.7

6. 3.5 2.5

7. 5.0 3.9

8. 3.9 3.1

324 Bulletin Museum of Comparative Zoology, Vol. 146, No. 7

■1

E
o

c
o

(0
D)

3
0}

3
CO
CO

^ o

(D D.

^i Q-

5, <=
D) o

Saurosuciius and ttik Rauisuciiid Thecodonts • Sill 325

the entire ventral, and half of the dorsal union with the frontals, from .5 to 1.5 cm.

borders of the antorbital fenestra, meeting Tlie lateral component is not extensive and

the lacrimal in a broad overlapping snture disappears entirely at the beginning of the

on the smooth shelf portion, and the jugal antorbital fenestra. At their maximum

in a broad zig-zag digitate union. Ten teeth width the joined nasals are approximately

were pr(\sent on the maxilla, of which seven seven centimeters wide, an indication of

were apparently functional at any one time, the narrowness of the anterior portion of

Numerous foramina are present on the the skull. Sculpturing on the nasal is in the

lateral surface just above the tooth row. fonn of irregular longitudinal grooves.

On the medial surface the most promi- Union with the maxilla and lacrimal is in

nent feature of the maxilla is the formation a straight sloping line. The suture with the

of the alveoli by large interdental plates, frontal is an inverted V located at the level

The plates are leaf-shaped extensions of the of the posterior border of the antorbital

alveolar septa and slightly overlap one fenestra.

another at the middle of the tooth body. Prefrontal. The area corresponding to

Above the plates a prominent groove runs the prefrontal is badly fractured, but this

to the dental lamina, which slopes down- element appears to be a small platelike

ward posteriorly to terminate on the ventral bone lying in the horizontal plane above

surface of the maxilla just behind the last the lacrimal. It does not participate in the

tooth. From the groove foramina represent- orbit, but sutures are difficult to distinguish,

ing the replacement teeth open directly Lacrimal. Most of the lacrimal lies in

above each tooth. This morphology ap- the depression surrounding the antorbital

parently represents a standard pattern of fenestra and is therefore completely smooth,

tooth replacement, analyzed by Edmund as is that part of the maxilla that partici-

(1957, 1960), in which the fibrous con- pates in the same depression. The lacrimal

nective bone that surrounds the tooth is is an extensive thin plate, forming most of

partially resorbed during the replacement the smooth shelf around the upper part of

process to form the shield-shaped inter- the antorbital fenestra. Anteriorly it is

dental plates. overlapped by the maxilla. Posteriorly it

Above the tooth row, in the anterior forms a ventral prong that overlies the

portion of the maxilla, a massive buttress dorsal extension of the jugal to fonn the

projects medially to meet the vomer and preorbital bar. The border between lacrimal

form part of the vault of the premaxillary and prefrontal is not discernible, but must

chamber. Dorsally, on the medial surface, lie in the zone behind the smooth depres-

the maxilla terminates in a straight sloping sion of the fenestra. This area is thick and

contact with the nasal and the broad over- heavily sculptured, and from it arises a

lap of the lacrimal. Posteriorly the jugal is prominent lateral ridge that runs down the

laminated to the maxilla just above the surface of the preorbital bar, terminating

tooth row. A large maxillary foramen is at the tip of the ventral prong of the bone,

present just anterior to the jugal suture The lacrimal forms the upper third of the

midway between the tooth row and the posterior border of the antorbital fenestra,

ventral border of the antorbital fenestra, on and virtually all of the anterior border of

the medial surface. the orbit. There is no definite lacrimal

Nasal. Anteriorly, the nasal is a thin bar foramen, but a rounded depression is

above the external naris overlapping the present on the ill-defined internal border

similar element of the premaxilla. From between the lacrimal and the frontal. There

this position it broadens to a dorsal and is no transverse component of the lacrimal,

component thickens considerably near its Jugal. The large skull openings of Satiro-

lateral plate of bon(>. Posteriorly the dorsal suchiis have reduced the jugal to a hori-

326 Bulletin Museum of Comparative Zoology, Vol. 146, No. 7

c
o

in
o

E
*^
u

c
o

(0
D)

CO
CO
CO

■><

3
CO
O

CO
CO

3
CO

O) J3

U- o

Saurosuchus and the Rauisuchid Thecodonts • Sill 327

zontal rod \\'itli two dorsal prongs
projecting npwards to form parts of the
pre- and postorbital bars. It is most ex-
panded anteriorly where it is platelike and
overlapped by the maxilla in a prominent
zig-zag snture. Immediately behind this
union, the dorsal projection of the pre-
orbital bar reaches up medial to the narrow
ridge of the lacrimal. Behind the orbit the
second prong of the jugal extends upward
and backward to form a strong, sloping,
abutted contact with the \'entral expression
of the postorbital. At the ventral border
of the infratemporal fenestra, the jugal is
a relatively narrow uniform bone. It thins
out to a fine edge on its lateral surface,
showing clearly the area where it was
overlapped by the quadratojugal. Sculptur-
ing is present only in the anterior portion
of the bone, where it meets the maxilla,
and even that is light. Only the ventralmost
part of the orbit is formed by the jugal, but
it constitutes nearly all of the anterior
border of the infratemporal fenestia. Di-
rectly below the postorbital bar there is
an outward bulge in the jugal, fomiing a
distinct pocket on the internal surface,
possibly the contact for the ectopterygoid.

Frontal. The frontal is a thick strong
bone dominated by the massive supra-
orbital arch. Medially it curves down from
the arch to the midline where its posterior
portion meets the anterior projection of the
parietal. Anteriorly it joins the nasal and
prefrontal in a zig-zag suture. Sculpturing
is present, principally on the arch, and is
of the pit and groove variety. The thickest
part of the frontal is the area of the mid-
line, which in the type is two centimeters
deep. Internally there does not appear to
be an interorbital septum, but the orbit is
well defined by the medial continuation of
the orbital arch. Anteriorly the arch forms
the previously mentioned pocket at its
junction with the lacrimal. Anterior to the
orbit the frontal thins to slightly over one
centimeter in thickness, and bears a down-
ward-projecting ridge near the midline.
This ridge, presumably the border of the

olfactory tract, is eight millimeters high at
its maximum and tapers off to the level of
the bone at the anterior end of the frontal.
Behind the orbit, at the junction of the
frontal, parietal, postfrontal, and post-
orbital, a prominent circular pocket is
present. This most probably received the
anterior process of the laterosphenoid.

Postfrontal. This is a small semicircular
bone lying on the dorsolateral surface of
the skull between the frontal and the post-
orbital. It does not enter into the supra-
temporal fenestra. On the ventral surface
of the skull it is not possible to distinguish
the borders of the postfrontal.

Postorbital. The postorbital forms nearly
all of the posterior border of the orbit, and
the upper third of the anterior border of
the infratemporal fenestra. Dorsally, just
behind the orbital arch, it bears a promi-
nent, sculptured tuberosity. Ventrally, it
meets the ascending process of the jugal in
a long diagonal contact. The anterior
border of the postorbital bar is emarginated
and beveled at the delimitation of the
circular part of the "keyhole" orbit. On the
cranial table the postorbital fomis most of
the lateral and anterior border of the small
supratemporal fenestra. A well-defined,
smooth margin surrounds this fenestra,
otherwise the upper region of the post-
orbital is sculptured by linear grooves.
Internally, the anteromedial portion of the
postorbital forms the rear part of the socket
for the laterosphenoid articulation. The
posterior part of the postorbital is not pre-
served in the type.

Nothing remains of the cranial table be-
hind the frontal and postorbital bones in
the available material.

Palatal Complex

Palatal remains of Saurosuchus are not
well preserved, but allow reconshuction of
the major features. A primiti\'e character
of the palate is the long triangular inter-
pterygoid vacuity. The internal nares are
somewhat displaced toward the rear and
close to the sides of the maxillae. No traces

328 Bulletin Museum of Comparative Zoology, Vol. 146, No. 7

Figure 4. Palatal view of the skull of Saurosuchus. Ec — ectopterygoid,
Pt^ — pterygoid, PI — palatine, V — vomer. X 1/4.

Saurosuchus and the RAtnsucHiD Thecodonts • Sill 329

of teeth are found on the palatine or on the
pterygoid. Although crushed, the palate
appears to have formed a deep vault rather
than a shelf. The basicranium is not known.

Pterygoid. As usual, the pterygoid is the
largest of the palatal bones, and is divided
into the customary three components:
flange, palatal, and quadrate rami. The
palatine ramus consists of a broad thin
plate of bone that extends forward from
the base of the flange portion and narrows
anteriorly to a V-shaped ridge that meets
the vomer near the midline. The medial
border of the pterygoid is formed by a
rounded ridge and steep shelf of bone that
form the edge of the interpterygoid fe-
nestra. Only at the anteriormost tip do the
pterygoids join at the midline. On the
dorsal surface of the palatal ramus a deep
groove is present just lateral to the wall of
the intei-pter)'goid vacuity. This groove
may continue onto the vomer. The flange
portion of the pter>^goid is massively con-
structed, and bears a thick, rounded
posterior border that curves out to fonii
the "wing." Where the wing meets the
heavy ridge that borders the interpterygoid
fenesti'a a deep pocket is formed. Postero-
medial to this pocket lies a thick remnant
of the basipterygoid articular bar. An-
teriorly the flange thins considerably, be-
coming the same thickness as the palatine.
At the posterior margin the flange is 15 mm
thick, while anteriorly it is only 4 mm. The
angle of inclination of the flange is approxi-
mately 45 degrees.

Ectoptenjgoid. Only the massive portion
of the ectopterygoid that forms the lateral
border of the pterygoid flange is preserved.
This portion forms a strong buttress along
the entire lateral edge of the pterygoid
"wing." There is no identifiable scar on the
maxillae or jugal to indicate the articulation
of the ectopterygoid, although it seems
probable that the bulge just below the
postorbital bar was for reception of the
ectopterygoid strut. The massive nature
of the preserved portion of the bone indi-

cates that the ectopterj^goid served to
strengthen the lateral part of the pterygoid.

Palatine. As preserv^ed, the palatine is a
thin plate, not possessed of unusual char-
acteristics. Anteriorly it forms tlie posterior
half of the internal naris; the suture with
the vomer is well preserv^ed. Laterally it is
applied to the side of the maxilla, opening
posteriorly into the pterygoid fenestra. The
medial border is not well preserved, but
appears to have been of the usual platelike
contact with the pterygoid.

Vomer. The vomer is poorly preser\'ed
and represented only by a distorted and ill-
defined mass of bone anterior to the in-
ternal nares. As near as can be detemiined,
the vomer formed the anterior half of tlie
internal naris, above which it expanded
considerably in the form of a laminar sheet
of bone applied to the medial side of the
massive maxillary buttress. Possibly, a por-
tion of the vomer behind tlie maxilla
formed a secondary buttress behind tlie
laminar part.

Dentition. Most of the 14 sturdy teeth in
the upper jaw are of equal or nearly equal
size. In the premaxilla the teeth are not
preserved, but to judge from the size of
the alveoli, the first and fourth teeth are
slightly smaller than the second and third.
In the maxilla, the last three teeth show a
slight reduction in size compared with the
anterior ones. All of the maxillary alveoli
show the presence of functional teeth, with
the possible exception of the first two,
although at least two and possibly three
growth stages are represented. The third,
fifth, and seventh teeth are the largest, with
the fifth slightly larger than the others. The
fourth, sixth, and eighth are approximately
the same size. The ninth and tenth teeth
are broken off at the alveolar border, but
were similar to the eighth in size. All of
the teeth are of similar shape, heavily
constructed, laterally compressed, sharply
pointed, and recurved. The last three or
four of the maxillary series seem to be more
stiongly recurved than the anterior mem-
bers, but this may be due to defomiation.

330 Bulletin Museum of Comparative Zoology, Vol. 146, No. 7

Roots of the teeth are approximately twice parallels the development found in some

as long as the crown. Near the alveolar prosauropods, and seems to be related to

margin, the teeth are much more com- size.

pressed and elongate than in the main body Cervical vertebrae. Only one cervical is

of the crown, and on the fully erupted teeth known, PVL 2472. It is poorly preserved

a slight depression is present on the lingual and of questionable reference to Sauro-

face of the tooth near the margin formed by suchus. It was found in association with a

the alveolar septum (see Plate 1). The teeth tibia and astragalus, also poorly preserved,

are essentially symmetrical, but the plane that appear to be identical to those of PVL

of symmetiy, taken between the anterior 2267. However, the unusual features of the

and posterior serrations, is slightly rotated vertebrae warrant its inclusion in this study

anteromedially-posterolaterally. Enamel on even though its association with Saurosuchus

the crown is thin and not striated. is not completely reliable. Only the centrum

Serrations are present on the distal three- is preserved; it is an elongated, flattened
quarters of the anterior edge and along all structure generally constricted in the
of the posterior margin. However, this con- middle. The anterior ( ? ) face is sti'ongly
dition can be fully appreciated only on the concave and bears a proti-uding lower mar-
fully erupted teeth; in those teeth that have gin that would seem to indicate a cervical
not reached maturity the serrations con- flexure. The rear (?) surface is only
tinue to the alveolus. Form of the serrations slightly concave. There is no keel. In the
is the same on both edges; they consist of middle portion, the body of the centrum is
simple crosscuts perpendicular to the long not only constricted laterally, but is also
axis of the tooth. Density of the serrations greatly flattened, which transforms the
is 12-14 per 5 mm, and is the same on both whole into a very lightly built structure,
the anterior and posterior edges. There are Prominent pleurocoels are present just be-
no wear facets on the teeth, although the hind the flared articulating surfaces. Thus
larger ones have a somewhat more rounded the lateral border of the centrum is almost
apex. a horizontal plate that curves inward to

the narrow waist (see Fig. 5). The char-

AxiAL Skeleton acteristics of the vertebrae represent an

The exact number of presacral vertebrae ^^^"^^ development of a strong, lightweight

is not known. Two vertebral series are support for the cervical region. As such it

preserved, PVL 2198 and PVL 2557. The ^l,^°^P\^^^/^ "^u rl ^°^^^V .

former consists of nine dorsals, all of which ^^ i' f^'l presence of

bore ribs; the latter series is from the sacral Pleurocoels are not found on any of the

and caudal region and does not duplicate ^^^°^^^ ^^^^f^ vertebrae. This condition

any of the vertebrae of the PVL 2198 series. ^'''^^^' f ^^^^^^^^ ^^ ^^^^^ t^'lV

In PVL 2557 the first two presacrals are ^^"^^ °^ ^^'' ^?^^"g^, ^^^T''i c Measure-

I i 1 .1 . .1 . ments are as follows: length, 18 cm; width

preserved and show that ribs were not ,- , , . ^ ,, .i i r i

. .1 A 4.U *- • u of the anterior surface, 11 cm; width of the

present on these. As the anterior members , . . i . j ,

of the PVL 2198 series do not show char- fo^^tricted waist 4 cm; and approximate

acteristics of cervical vertebrae, it seems ^'^'^^^ ^} ^^^^t, 2 cm.

reasonable to assume that not more than ^^^^«^ veiiebrae. Vertebrae of the dorsal

two vertebrae represent the gap between series are represented by the first two pre-

the presacrals of the two series. Assuming sacrals of PVL 2557 and by nine articulated

the usual presence of seven or eight cervi- members of PVL 2198. The anterior mem-

cals, the vertebral count would fall into the bers of the PVL 2198 series are poorly

23 to 25 characteristic of archosaurs. In preserved. The most striking feature of the

general, structure of the vertebral column Saurosuchus vertebrae is the reduction of

Saurosuchus and the RAuisucmo Thecodonts • Sill 331

Plate 1. A. Type of Skull of Saurosuchus, PVL 2062. X Vs. B. Lingual view of left maxilla, note interdental
plates. X 1. C. Enlarged view of a recently erupted tooth, showing serrations on the posterior edge. X 8.

332 Bulletin Museum of Comparative Zoology, Vol. 146, No. 7

Table 2. Measurements of the vertebral column of Saurosuchus galilei (in centimeters).

PVL 2472 cervical (?)

Maximum lengtli

20.0

Transverse width of posterior face

10.5

Transverse width of anterior face

9.0

Minimum width of constricted waist measured on ventral surface

4.0

Height of anterior face

5.3

Height of centrum body at waist

2.0

Dorsal vertebrae PVL 2198

Posterior

Maximum length of centrimi

7.5

8.5

Height of anterior rim

8.7

Widdi of anterior rim

7.2

Width of posterior rim

6.0

Total height of vertebrae

20.0

22.0

Minimum width of constricted waist

1.2

1.3

Lateral extension of transverse process from the

midline

5.5

4.8

Width of neural spine table

3.0

2.4

Diameter of parapophysis

2.5

Diameter of diapophysis

2.7

Height of neural arch above centrum

12.6

13.0

Lumbar and sacral vertebrae PVL 2557

Presacral 1

Sacral 1

Sacral 2

Length of centnmi

9.0

10.5

10.0

Height posterior rim

12.0

9.5

10.0

Width posterior rim

10.5

10.0

8.5

Height anterior rim

11.0

9.5

Width anterior rim

12.0

10.0

Minimimi width of constricted waist

4.0

3.7

3.1

Width of neural spine table

4.5

3.6

Height of neural arch above centrum

17.6

19.5

Total height of vertebrae

28.0

29.5

Caudal vertebrae PVL 2557

Caudal 3

Caudal 9

Length of centrum

8.0

7.5

Height posterior rim

9.5

7.0

Height anterior rim

9.0

7.0

Widtli anterior rim

9.0

6.0

Width posterior rim

9.0

6.0

Minimum width of constricted waist

3.9

2.3

Width of transverse process from midline

11.5

7.0

Chevrons PVL 2557

No. 1

No. 2

No. 3

No. 4

No. 5

Length

18.0

21.0

21.5

21.2

18.5

Width between articulations

7.5

6.7

6.2

6.0

Width of articular facets

4.0

3.0

2.5

Length from facet to fusion with opposite side

4.5

3.8

4.0

3.5

3.2

the centrum to a thin vertical plate between specialization of the genus. Another notable

the flared rims. This condition is not as feature of the dorsals is the complete sepa-

well developed in other thecodonts and ration between the diapophysis and the

seems to be a unique weight-reduction parapophysis along the entire known series;

Saurosuchus and the Rauisuchid Thecodonts • Sill 333

tlic latter is always found on the neural
arch, never on the centrum. Transverse
processes are larger in the anterior region
than in the posterior, but all are rather
short and stubby. The neural arches sit
high on the centra and bear flat rectangular
spines that are not inclined posteriorly.
Size and shape of the vertebrae appear to
be uniform throughout the series. All centra
are uniformly amphycoelus and do not bear
keels. Morphologic changes along the series
are not prominent and consist principally
of the reduction of the transverse processes
in the lumbar region.

In end view the centra are oval-shaped
with the long axis in the vertical plane. The
rims are flared and rounded, not beveled.
Reduction of the body of the centrum took
place by expansion towards the rims of the
common "hour-glass" constriction. The re-
sults are a steeper angle of the constriction
behind the rims and the formation of a
narrow plate between them (see Fig. 6).
No ridges, rugosities or excavations are
present on the body of the centrimi. Length
of the centrum is 7 to 8 cm, width 6 to 7
cm, height is around 9 cm. The body of
the centrum expands slightly to receive the
neural arch and form the floor of the neural
canal.

The neural arch is a large structure that
sits high up on the centrum. Contact with
the centium is a simple butt union, without
the formation of pedicels. Prezygapophyses
are not well preserved, but form short
processes that sweep forward on either side
of the prominent interspinous notch of the
neural spine, just above the articular facet
for the capitulum. Apparently the prezyg-
apophyses did not overhang the border
of the centrum. The postzygapophyses lie
on the same level as the transverse process
and are formed from lateral expansions that
diverge from the base of the neural spine,
creating a wedge-shaped cleft behind it.
The zygapophyseal facets are relatively
small smooth areas facing downward with
a .slight inclination toward the midline.

Figure 5. Supposed cervical vertebra of Saurosuchus.
Top, ventral, Bottom, dorsal. X Va.

Rib articulations are restricted entirely
to the neural arch. The parapophysis is a
round facet on the anterior vertebrae of the
dorsal series, but becomes laterally ex-
panded into a peduncle on the posterior
vertebrae. On all of the dorsals the
parapophysis lies below and in front of the
transverse process. These processes ai'e
short and robust; those of the shoulder
region are larger than those of the lumbar
series and project posteriorly approximately
30 to 45 degrees. The diapophysis forms
as an expanded foot at the tip of the trans-
verse process. On the anterior dorsals this
expansion is considerably larger than the
parapophyseal facet, while in the lumbar
region it is of the same size. A notable
featui'e of the transverse process is the
presence of strutlike ridges on the under-

334 Bulletin Museum of Comparative Zoology, Vol. 146, No. 7

Figure 6. Three views of a dorsal vertebra of Saurosuctius. Left, right lateral; middle, posterior; right, ventral.

X 1/4.

side and edges. In the shoulder region,
where the transverse process is largest, four
struts are present. One extends to the
prezygapophysis, another to the postzyg-
apophysis, a third down to the parapophysis,
and a fourth down and back to the rim of
the centrum. All of the ridges extend the
entire length of the transverse process, and
form a strong supporting structure. In the
posterior dorsals the strut structure is modi-
fied by a reduction to three struts. The
parapophysis has moved slightly dorsal, al-
most to the level of the prezygapophysis
and the transverse process is smaller. Only
one ridge is present in the anterior portion,
extending from the transverse process to
the parapophysis. The shorter transverse
process and the lateral expanded para-
pophysis change the aspect of the support-
ing strut from that of a ridge to a sheet of
bone (see Fig. 6).

The neural spine on all of the dorsal
vertebrae is a robust rectangular blade,
slightly higher than the centrum. As well

as can be determined, the blade is not in-
clined posteriorly on any of the vertebrae.
On its dorsal surface the spine is expanded
into a spine table, presumably for the at-
tachment of dermal armour. At the anterior
and posterior borders the spine does not
attenuate, but bears prominent grooves for
the interspinalis musculature. On the lead-
ing edge the groove occupies the lower
half of the length and is deeper at the base.
The groove on the posterior margin is shal-
lower but extends the entire length of the
blade.

A distinct lumbar region was present in
Saurosuchus, but it is not possible to de-
termine the number of vertebrae involved.
Specimen PVL 2557 has preserved the t\vo
vertebrae immediately anterior to the
sacrum, and these vary from the other
dorsals principally in their lack of normal
ribs. It is not possible to determine whether
the short downcurved processes are ribs or
transverse processes. They appear to be
transverse processes, arising from the same

Saurosuchus and the Rauisuchid Thecodonts • Sill 335

position on the neural arches as those of
the anterior dorsals. The processes are oval
in cross section and heaxily constructed.
Their origin on the arch is considerabh'
broader than that of the dorsals of PVL
219(S. From the arch they curve slightly
for\\'ard, then strongly downward.

Sacral vertebrae. The sacral \'ertebrae are
known exclusixely from the well-preserved
representatives of PVL 2557. Two sacrals
are present in Saurosuchus. The centra are
slightly more elongate than the other
dorsals, but otherwise are not different. The
sacrals are not fused, but there is a con-
siderable reduction of the rims where the
two meet, with the posterior rim of the
first sacral flared out at the sides and a
corresponding reduction and slight forward
extension of the anterior rim of the second
sacral. This imparts a slightly V-shaped
configuration to the union between the
two vertebrae. This condition is repeated
in the junction between the last sacral and
the first caudal. Such a union must have
essentially immobilized the three vertebrae
in\^olved, providing a partial substitute for
the fusion of the sacrals. The transverse
processes of the first caudal vertebrae are
not preserved, so it is not possible to de-
termine if it participated in supporting the
pelvis. Position and shape of the transverse
processes of the sacrals are essentially of
the type found in primitive archosaurs; the
first is large, oval-shaped, and positioned
near the anterior border of the centrum,
while the second is more crescent-shaped
and arises from the center of the centrum.
Both are impressive structiu-es, greatly en-
larged and heavily constructed.

Neural spines and arches of the sacral
vertebrae are not significantly different
from those of the presacrals that form the
lumbar region. The spines are heavily con-
structed and expanded, but form a well-
matched series with those of the lumbars.
The same is true for the neural arches.

It should be pointed out that the verte-
brae of specimen PVL 2557 do not have the

centra constricted nearly as much as those
of specimen PVL 2198.' Whether this dif-
ference is due to the difference in size be-
tween the two animals (PVL 2557 is
considerably larger than PVL 2198) or to
their different positions in the vertebral
column cannot be ascertained.

Caudal vertebrae. In general the caudals
of Saurosuchus are of shorter length than
the other vertebrae, and ha\'e large rounded
rims. The first three caudals do not bear
chevrons. Diameter of the centia of the
first five caudals is essentially equal to that
of the sacrals. Beginning with the sixth
caudal, there is a gradual reduction in size.
Rims of all of the caudal ^'ertebrae are
broad and rounded compared to the some-
what thinner rims of the other vertebrae.
The area between the rims is not reduced
as in the dorsals; the centra are more
"typical" in their squat rounded shape.
Beginning with the seventh caudal, a slight
groo\'e appears on the ventral surface of
the centrum. At the eighth or ninth the
shape of the centrum changes to the more
elongate and spool-like shape characteristic
of tail vertebrae in general. Large lateral
processes are present on the nine articu-
lated caudals preserved in specimen PVL
2557. The processes of the first four or iive
caudals are large bladelike stnictures that
extend outward and backward from the
level of the dorsal surface of the centra. In
caudals numbers six and seven, the out-
ward extensions of the processes are greatly
reduced, but they retain the blade shape.
In caudals eight and nine, the lateral pro-
cess loses the bladelike expansion and be-
comes a simple short lateral process.

Neural spines of the first four caudals are
large, cover the entire length of the
centrum, and in general are like those of
the dorsals. Beginning with the fifth caudal
there is a relatively sharp reduction in the
anteroposterior length, and in the height
of the spine. The spine becomes more in-
clined caudad and develops a more promi-
nent interligamentum cleft in tlie anterior
border near the base. The sides of the cleft

336 Bulletin Museum of Comparative Zoology, Vol. 146, No. 7

Table 3. Measxjrements of the pelvis of Saurosuchus galilei (in centimeters).

Ilium

Length along dorsal border

Lengtli of anterior spine from the ventral curvature

Length of posterior spine from tlie ventral curvature

Height of dorsal border above acetabulum

Maximum height of ilium

Widtli of dorsal border

Ischium

Total length along curvature

Width of shaft

Height of shaft

Height of terminal expansion

Widtli of temiinal expansion

PVL 2198

PVL 2557

36.0

41.0

4.2

6.5

17.0

23.5

6.5

10.5

23.5

29.0

1.8

2.1

42.5

50.0

3.7

4.1

3.9

4.6

5.3

9.3

5.9

6.1

develop progressively into prominent ridges
that sweep forward to form the prezyga-
pophyses. Position of the zygapophyses
undergoes a slight progressive shift towards
the front; the prezygapophysis begins to
overhang the centrum and is accompanied
by a corresponding anterior displacement
in the postzygapophysis.

Chevrons. Chevron bones first appear on
the fourth caudal. The first four chevrons
are Y-shaped and bear large disclike pedi-
cels for articulation with the vertebrae. The
arms of the "Y" become progressively closer
together until they join at the fifth chevron
(eighth caudal), leaving a small opening
of which only a vestige remains in the sixth
chevron (ninth caudal). Construction of
the chevrons is simple and not unusual; the
expanded pedicel is followed by a long
sturdy shaft fused with its opposite just
below the centrum. A slight ridge is pres-
ent on the posterior surface of the shaft.
The six preserved chevrons are all approxi-
mately the same length (equal to the total
height of the vertebrae) and are strongly
inclined caudad.

Ribs. Few ribs are preserved. Those
available are fragmentary and are covered
with a thick iron-rich matrix. They appear
to be heavily constructed, thick-bodied, and
with a prominent ridge on the upper third

of the anterior edge. The posterior surface
(at least in the proximal section) is flat
with a slight depression down the middle.
Rib articulations appear to be well ossified.
As may be expected, the largest ribs were
the anterior members (size inferred from
the relative size of the articular surfaces on
the vertebrae).

Appendicular Skeleton

Pelvic girdle. The pelvis of Saurosuchus
is well represented except for the distal
portion of the pubis. Elements available
are: left ilium and paired ischia of PVL
2198; right ilium, complete right and par-
tial left ischia of PVL 2557; and a poorly
preserved fragmented ilium of PVL 2267.
The proximal portion of the right pubis of
PVL 2557 is articulated with the ilium of
that specimen.

The usual elements of the pelvis were
present, in typical triradiate forms. There
is no indication of perforation of the acetab-
ulum. A notable feature of the pelvis,
apparently common to the Rauisuchidae,
is the high position of the pubic articu-
lation and the limited participation of this
element in the acetabulum.

Ilium. Two major structural divisions
are present on the ilium; the acetabulum
and the iliac blade. Most of the ilium is

Saurosuchus and the Rauisuchid Thecodonts • Sill 337

•.'.v.'-'i n-'f'fi Hi!Ail!l-A ur

B?HjWv^'«;V^«!>.^^■;.*^K'S^^^..■^■«■T«'■'^''^^''.^.^■^;!. ■'

Figure 7. Ilium of Saurosuchus. x y4.

incorporated into the acetabulum, which is
a large deep depression that faces shghtly
downwards. The dorsal margin of the
acetabulum is formed by a thick lateral
flange positioned just below the anterior
emargination of the iliac blade. Ventrally
the acetabulum wall thins considerably at
its borders with the ischium and pubis.
Anteriorly it expands transversely where it
meets the dorsal border of the pubis, below
which the bone thins, presenting a tear-
drop shape in cross section. A notable
feature of the ilium is its articulation with
the pubis and ischium; the suture of the
pubis occupies nearly all of the anterior
border, starting from a level almost at the
dorsal border of the acetabulum, whereas
the ischium meets the ilium more in the
ventral plane. The ilium is not constricted
above the acetabulum. Rather, the anterior
origin of the iliac blade arises from an
emargination immediately above the thick
dorsal margin of the acetabulum, while the

posterior portion of the blade sweeps up-
ward and backwards from a level slightly
above the midline of the acetabulum. The
anterior tip of the blade is short and tliick;
it does not reach the anterior border of the
acetabulum. The posterior portion of the
blade consists of tliree prominent elements:
1) a rounded dorsal ridge, 2) a horizontal
shelf on the medial side, midway between
the dorsal and ventral borders, and 3) the
very thick rounded ventral border of the
blade. The internal shelf con-esponds to
the structure termed "brevis shelf" by
Romer (1927) and originates just behind
the acetabulum, becoming considerably
heavier and thicker at the terminal end of
the blade. At its posterior tip the iliac
blade is heavily constructed with the brevis
shelf lying perpendicular to the blade.
Rugosities present in the tip region indi-
cate that it was probably continued in carti-
lage. Facets for the sacral ribs lie just
above the level of the acetabulum. Total

338 Bulletin Museum of Comparative Zoology, Vol. 146, No. 7

Figure 8. Two views of \he paired isciiia of Saurosuchus. A, ventral; B, dorsal. X ''A.

length along the dorsal border of PVL 2198
is 36 cm of which 16.5 lie below the acetab-
ular rim. Thus the blade above the
acetabulum is only 6 cm high.

Ischium. The ischium of Saurosuchus
consists of a broad flange followed by a
relatively long shaft that bears a mild
terminal expansion. In general it resembles

somewhat that of the dinosaurs in that it
is rodlike rather than platelike. Proximally
the ischium bears a large expanded head
with a prominent lateral lip. As usual, the
anterior portion of the head is considerably
thinner than the posterior. Anteriorly, be-
low the lip is a deep concavity, where the
bone becomes a thin plate that angles to-

Saurosuchus and the Rauisuchid Thecodonts • Si7/ 339

Table 4. MEAsuREJ\rENTs of the hind limb of Saurosuchus galilei (in centimeters).

PVL 2267

PVL 2557

Femur

Appio.ximatc total lenj:;th
Maximum width of proximal articulation
Distance of 4tli trochanter from head
Thickness of shaft at midpoint
Approximate width of distal articulation

Tibia

Length

Minim mn shaft \\'idth
Widtli distal articulation
Widdi proximal articulation

Fibula

Length approximate
Minimum shaft width

Anteroposterior \\'idth of distal articulation
Transverse width of distal articulation
Distance between distal articulation and
ilio-fibiJaris trochanter

Astragalus

Maximum width across anterior face
Maximum anteroposterior length
Height on anterior face

Calcanium

Maximum anteroposterior lengtli
Maximum height of tuber
Maximum height of anterior face
Maximum transverse width

65.0

7.0

17.0

26.0

5.0

9.0

PVL 2472

PVL 2267

PVL 2557

46.5

45.0

4.0

3.5

6.5

5.5

6.8

12.0

PVL 2267

43.0
2.7
8.5
4.0

21.0

PVL 2472

10.5
8.5

3.8

PVL 2267

12.5

7.5
3.8
8.0

PVL 2557

3.0
8.2
5.3

22.0

PVL 2557

10.6
9.6
5.0

PVL 2557

16.0
11.0

5.5

8.5

wards the midline, terminating in a smooth
rounded border that does not touch its
opposite member. Posterodorsally, the
ischium is a heavy rounded strut arising
from the thick buttress that forms the
posteroventral rim of the acetabulum.
Distal to the expanded head region the two
ischia are solidly fused. The symphysis
forms a slight ridge down the ventral (an-
terior) surface of the paired bones. At
the distal termination the ischium flares out
to a moderately expanded foot, similar to
that of Ticinosuchus, but more rodlike. In
cross section the rod portion is tear-drop
shaped, the thin portion being fused to its
opposite. This fusion created a channel

along the dorsal midline and a correspond-
ing ridge along the ventral surface. The
ischium makes an approximately 45-degree
angle with the iliac blade, and the strut
portion is slightly concave upwards.

Pubis. Only the proximal portion of PVL
2557 is known. It shows that the dorsal
portion of the articulation with the ilium
was a very thick continuation of the an-
terior border of the ilium beneath the
blade. Below this thick rounded border,
the pubis thins rapidly, matching the thick-
ness of the ilium. As noted previously, only
the edge of the pubis actually participates
in the acetabular depression. From the
cross section of the broken portion of the

340 Bulletin Museum of Comparative Zoology, Vol. 146, No. 7

Figure 9. Femur of Saurosuctius, composite drawing.

X 1/3-

pubis, it would appear that the bone
thinned considerably in its anterior portion
below the rounded dorsal margin.

Femur. The femur is known from two
nearly complete specimens. The complete
proximal half of the femur is well preserved
in specimen PVL 2557, and was found
articulated with the corresponding pelvis.
It is well-preserved material but appears to
be slightly compressed. PVL 2267, the
other femur, consists of a complete shaft
but lacks the extreme articular surfaces at
both ends. This specimen was figured by
Reig (1961) and shows a slight intertro-
chanteric depression. The depression is a
deformation of the particular specimen and
not a true anatomical feature.

In its overall aspect, the femur of Sauro-
suchus is of the crocodile type rather than
like that of the dinosaurs. The proximal
portion is a flange with a wedge-shaped
articular head. The shaft is gently sigmoid
and oval-shaped in cross section. Distally
the termination flares out to what must
have been large articular condyles. Owing
to the defoi-mation of PVL 2267, it is not
possible to determine the degree of rotation
of the two extremes, but it appears to have
been slightly greater than that of crocodiles.
The proximal articulation consists of a
rugose tear-drop-shaped surface, the broad
portion of which forms a continuation of
the thick anterior border of the femur. Be-
hind this section the bone thins rapidly to
the posterior edge. In PVL 2557 the broad
portion is 5.5 cm thick, the tapered poste-
rior edge is 2 cm. Curvature of the head
in toward the acetabulum takes the form
of an arc along the anterior border and
reaches a maximum of 4 cm of inward
displacement from the shaft. There is no
fonnation of structures that can be defined
as greater or lesser trochanters. The an-
terior border of the femur is uniformly
thick below the articular head, but in the
upper portion it thins rapidly, forming a
slight depression on the posterior flange
area. Ventrally the bone is smooth, de-
creasing in thickness from the expansion

Saurosuchus and the Rauisuchid Thecodonts • Sill 341

of these; there is no intertrochanteric fossa.
The posterior edge of the fhmge area thick-
ens rapidly, becoming part of the shaft at
the level of the fourth troclianter. Midway
between the articular head and the fourth
trochanter a slight expansion is present on
the posterior edge. The fourth trochanter
is a relatively small rugose bulge arising
in the center of the ventral surface, ap-
proximately one third of the way down
from the proximal articulation, very much
like that of crocodiles. Below the fourth
trochanter, at approximately half the total
length, the proximal flattened expansion
disappears into the oval-shaped shaft.

Distally the shaft expands evenly into
the distal condyles. These are not pre-
served, but a remnant of the intercondylar
fossa on the dorsal surface indicates that
the posterior condyle was the larger of the
two.

Tibia. This bone is known from the com-
plete but poorly-preserved specimens PVL
2472, PVL 2267, and the well-preserved
distal half of PVL 2557. The tibia is a ro-
bust bone approximately twenty percent
shorter than the femur. Proximally, the
head expands to a triangular shape, the
narrow point of which projects anteriorly
and medially to form the cnemial crest.
This crest extends down one third of the
length before merging with the shaft.
Posteriorly, the proximal surface is sepa-
rated by a prominent depression into the
condyles for articulation with the femur.
This area of PVL 2472 is shattered, but
from the area surrounding the depression
it would appeal- that the two condyles were
of nearly the same size. The medial surface
of the proximal expansion formed the
shortest leg of the triangle and bears a
slight depression, probably indicating the
contact for the fibula. Anterolaterally a
broad flat area was present, separating the
cnemial crest from the lateral condyle. The
shaft is long, subround in cross section and
slightly flattened on the anterolateral sur-
face. Distally, the tibia flares out to a
transverse expansion equal in size to the

articulating surface for the femur. How-
ever, it should be noted that in actual artic-
ulation with the astragalus the tibia was
rotated approximately thirty degrees, orient-
ing the cnemial crest directly forward.
Thus the lateral side of the distal termi-
nation rested on the anterolateral portion
of the astragalus and the medial portion on
the posteromedial. The lateral expansion is
broad and oval-shaped, the medial is nar-
row and tapering. Separating the two areas
of expansion is a narrow groovelike depres-
sion on the posterior face that extends up
the shaft to the midpoint (see Plate 2).
The major surface of articulation is con-
cave on the underside. All articulations are
well ossified and have a shiny surface.

Fibula. The fibula is known from the
right distal half of specimen PVL 2557.
The shaft is oval in cross section, the long
axis oriented anteroposteriorly, and is flat
on the medial surface facing the tibia. The
most prominent feature of the shaft is the
large tubercle on the anterolateral face, just
above the midpoint. Presumably, this was
for the insertion of the iliofibularis muscle.
Above the tubercle the shaft curves slightly
outward; below, it is characteristically con-
cave toward the tibia. Distally the tibia
has a flared surface for articulation with
the calcaneum and astragalus. The articu-
lating surface is lower on the lateral side
than on the medial, and bears two grooves
corresponding to the two tarsal elements.
Aiticulation with the calcaneum occurs on
the large lateral groove behind an
anterolateral expansion of the bone. The
astragalar articulation occupied a smaller
diagonal groove on the anteromedial side of
the distal termination.

Tarsus. The tarsus of Saurosuchus was
of the "crocodilian" type, in which the cal-
caneum was functionally a part of the foot
and the astragalus rotated with the crus.
Four elements were present: proximally the
large ti-iangular astragalus and the equally
large tuberous calcaneum, distally a large
lateral tarsal and a much smaller medial
one. Elements preserved are: left and right

342 Bulletin Museum of Comparative Zoology, Vol. 146, No. 7

•*wr"

Plate 2. A. Ilium and Ischium of Saurosuctius, PVL 2198. X Vs. B. Dorsal view of the paired Ischia. X Va.
C. Distal portions of the tibia and fibula of Saurosuctius specimen, PVL 2557. X Va.

Saurosuchus and the RAmsucHiD Thecodonts • Sill 343

Plate 3. A. Exploded and stereo view of the foot and tarsus, PVL 2557. B. Proximal view of the articulated
metatarsals. C. Articulated foot.

344 Bulletin Museum of Comparative Zoology, Vol. 146, No. 7

Plate 4. Stereo views of Tarsal elements of: A. Saurosuchus, B. Crocodylus, C. Undescribed rauisuchid from
Los Colorados Fm. (courtesy of Jose Bonaparte), D. Neoaetosauroides, E. Riojasuchus. (All to same scale.)

Saurosuciius and the RAUisucino Thecodonts • Sill 345

calcaneum of PVL 2262, left astragalus of
PVL 2472, all poorly preserved hut easily
identifiable, and the extremely well-pre-
served complete tarsus and foot of PVL
2557. Adequate description of these com-
plex irregular bones is difficult, and the
reader is referred to the stereo-photographs
(Plates 3 and 4). The tarsus of Sauro-
siichus appears to be virtually identical to
that of Ticinosiichus, as described by Krebs
( 1965), except for minor details. However,
Saiirosuchus was a much larger animal and
the tarsal elements are naturally much
larger and more heavily constructed. All
tarsal elements were well ossified.

Astrafialus. The astragalus is an irregular
triangular block of bone. On its dorso-
medial surface it bears a large, triangular,
saddle-shaped area for articulation with the
tibia. On the lateral side and separated
from this area by a small, steep, forward-
inclined ridge, lies the much smaller facet
for articulation with the fibula. This sur-
face is much more steeply inclined than
that of the tibial articulation, and lies at
approximately seventy degrees to it. An-
teriorly, the surface of the bone bears a
deep excavation, common to most reptiles
that have a crocodiloid tarsus, medial to
which is the bulbous convexity for articu-
lation with the first metatarsal. A notable
feature of the astragalus is its shallow
depth. Thus the anteriormost border of
the tibial facet is practically on the same
level as the first metatarsal articulation.
Posterolaterally the face of the astragalus
is inclined downward from the peak of the
ridge separating the epipodial articulations
to the rounded ball that articulates with the
calcaneum. Just behind the ridge peak a
deep groove is present, which opens up
posteroventrally to a curved depression in
front of the ball joint. This depression fits
over the anteromedial rounded border of
the calcaneum. When thus articulated the
fibular facets of both astragalus and cal-
caneum are brought together and a more
or less double ball and socket joint is
fonued. The posteromedial border of the

astragalus forms a rather featureless thick
rounded border.

Calcaneum. Basically, the calcaneum is
a rectangular block of bone that bears a
posterior upturned tuber and a medial
process that forms the rear border of
an anteromedially directed! hemispherical
socket. Four polished articular surfaces are
present on the bone. The anteriormost
border is formed by the rounded, slightly
ginglymoid articular surface for the fibula.
This area is clearly marked and by its
terminal position indicates that the cal-
caneum must have been strongly rotated
through its transverse axis, elevating the
anterior end and depressing the tuber por-
tion. Medial to the fibular facet is the
rounded convexity that faces anteromedi-
ally at a forty-five degree angle from the
face of the fibular surface, which articu-
lated with the previously described con-
cavity of the astragalus. Again, the area of
movement is well marked by the polished
surface. Immediately behind this area is
the small excavation that forms the anterior
expression of the hemispherical socket that
constitutes the major articulation between
the two proximal tarsal elements. This
excavation is continued medially onto the
anterior face of the medially projecting
process mentioned above, the whole form-
ing a well-developed spherical depression
directed inward at an angle of approxi-
mately forty-five degrees from the anterior
face. The fourth articular surface is a small
rounded depression ventral to the fibular
facet. This was for the reception of the
large fourth tarsal bone. On the dorsal sur-
face of the calcaneum, behind the fibular
facet and lateral to the socket, lies a raised
molding of bone that did not function as an
articular surface and is not marked by
muscle or tendon scars. It appears to have
been an artifact of ossification. Continuing
dorsally, the large tuber calcaneum projects
upward and rearward. The dor.sal and
posterior siuface of the tuber is rugose,
indicating ligament attachment. The lateral
surface is a flat wall, slightly depressed in

346 Bulletin Museum of Comparative Zoology, Vol. 146, No. 7

Table 5. Measurement of the pes of Saurosu-
chus galilei ( in centimeters).

Lateral tarsal

7.0

Transverse width of anterior face
Lengtii of lateral articulation

with fifth metatarsal 5.5

Maximum height 4.0

Medial tarsal

Height
Width

Metatarsals PVL 2557

Length

Minimum shaft width
Width proximal end
Height proximal end
Width distal end
Height distal end

3.3
2.2

III IV

13.7 17.2 17.7 16.5 12.0

2.5 2.6 2.2 2.2 —

4.0 3.5 3.9 3.9 8.0

6.3 7.6 7.2 7.0 —

3.5 4.5 3.8 4.0 4.0

4.7 4.6 3.8 3.5 —

Phalanges

PVL 2557

Ij I, 11^ II„ III^ III^ IVj Vj

Length 5.6 8.5 5.8 4.0 5.8 3.4 4.3 3.7

Height pro.ximal 4.3 3.4 4.1 2.7 3.9 2.9 3.0 3.0
Height distal 2.8 1.2 2.7 2.2 2.5 2.0 2.1 2.2

the center portion. Ventrally the surface is
also flat, but a small pitlike depression is
present at the base of the tuber.

Distal tarsals. Two distal tarsals were
present in the foot of Saurosiichus, ap-
parently corresponding to numbers III and
IV of the primitive reptilian tarsus. The
lateral one is the largest of the two and is
tetrahedral in shape; the ventral surface is
flat, the other three sides form a rounded
pyramid dorsally. The dorsal surface is
slightly divided into a concavity for re-
ception of an expansion on the astragalus,
and a convexity that fits into a shallow pit
on the oalcaneum ventral to the fibular
articulation. Laterally the fourth tarsal
bears a large, saddle-shaped, convex, articu-
lar surface for the fifth metatarsal. Antero-
medially are two convex surfaces, separated
by a prominent groove, for articulation of
the third and fourth metatarsals. At the
extreme medial tip, beside the convexity
for the third metatarsal, lies a small concave

facet for reception of the third tarsal bone
(see Plate 3). This element is a small
rounded bone wedged between the lateral
side of the second metatarsal and the
astragalus.

Pes. The pes of Saurosiichus had five
sturdily constructed digits in the usual
reptilian fashion. Metatarsal V was widely
separated from the others, hooked, and
bore a broad medial expansion. The re-
maining four metatarsals were directed
straight out from the foot, with a prominent
transverse arch in the "instep" region. It is
perhaps notable that the expansion of the
proximal articulation surfaces of the meta-
carpals lies in the vertical rather than the
horizontal plane (see Plate 3B). Virtually
all of the information available comes from
the well-preserved right foot of PVL 2557,
which is complete except for some of the
distal phalanges. Additional elements of
the foot are represented by poorly pre-
served portions of left and right members
of PVL 2267. Apparently the phalangeal
formula was 2, 3, 4, 5, 3 in the usual primi-
tive fashion. However, the fifth toe may
have been reduced to but one or two
phalanges. Metatarsal No. 1 is shorter than
2, 3, or 4, is thick bodied, and bears a pulley-
shaped distal articulation behind which a
prominent diagonal groove traversed the
dorsal surface. Proximally, a concave
facet is present on the medial side of the
articulating surface, the remainder of the
surface being smooth. The lateral margin
of the proximal tip is vertical, its shape
matching the medial border of the second
metatarsal, with which it makes a very close
fit. The first phalanx is relatively large,
almost half the length of the metatarsal,
and bears a proximal concavity with a
ventrally projecting "heel" for articulation
with the rolling surface of the metatarsal.
Distally, the joint with the ungual is a
shallow ginglymus, narrow at the tip and
expanded ventrally. The ungual is a thick-
bodied pointed claw, narrow at the top,
wider on the bottom, and is half the length
of the metatarsal. Largest of the metatarsals

Saurosuchus and the Rauisuchid Thecodonts • Sill 347

is tlie second, although numbers 3 and 4 are ventral. The medial margin is expanded
of similar length. It bears a large narrow at the top to form a bulge, with the afore-
proximal articulation, expanded almost ex- mentioned groove lying just below it. Later-
clusi\'i'l\' in the \ertical plane. On the ally the proximal articulation bears a con-
medial side of the expansion are two facets cavity on the dorsal portion and a small
for the first metatarsal. Laterally, the proxi- convexity ventrally, corresponding to op-
mal articulation forms a straight vertical posite features on the close-fitting fourth
surface with no overlapping contact for the metatarsal. Distally the articular surface
third metatarsal. Midway down the side is of metatarsal number 3 is similar to that of
a prominent pit, corresponding to a simi- number 2 but smaller. The rounded flange
larly sized notch on the medial side of the is more expanded on the medial side than
adjacent metatarsal. Presumably this formed on the lateral, and a groove is present be-
a channel for nerve and blood supply. The hind the flange on the medial side. Only
shaft is thickly built, similar to the con- the first two phalanges are preserved; they
struction of the first metatarsal, and is are virtually identical to those of the second
concave on the lateral margin but straight digit, but somewhat more slender. The
on the medial side. Distally the articulation fomth metatarsal is slightly shorter and
is a large rolling surface with a prominent more heavily constructed than the third,
groove on the \'entral border. Just behind Its proximal articulation is diagonal in the
the articular surface, on the lateral side, an vertical plane like that of the third, but on
indentation is present between the flange the surface itself a prominent excavation is
of the articulation and the body of the present below the side dorsal border for
shaft. Shape and articulation of metatarsals the reception of the bulge of the fourth
show that the axis of the transverse "instep" tarsal. A major feature of the fourth meta-
arch ran between the second and third tarsal is its bowed shape; it is concave on
metatarsals. Two phalanges of the second the lateral side, with the convex medial side
metatarsal are preserved. As might be ex- fitting closely against the side of metatarsal
pected, they are the largest and most number 3. This curvature also serves to
hea\'ily constructed of the digits. The first rotate the plane of the promixal articulation
bears a large concave flange proximally, a approximately twenty degrees from the
short shaft, and a distal articulation similar vertical, toward the lateral side, from the
to that of the metatarsal. However, the plane of the distal articulation. On the
groove is considerably larger than that of lateral surface of the shaft in the proximal
the metatarsal. The second phalanx is sub- region anterior to the articular surface is a
rectangular in shape, and has a smooth prominent triangular depression, apparently
conca\ity proximally and a pulley-shaped for muscles and flesh related to the lateral
articulation distally. Although the ungual plantar pad of the foot. Distally, the
is missing, the size and shape of the distal articular surface consists of a pulley-shaped
articulation indicates that the claw was convexity somewhat different from that of
approximately the same size as that of the the other metatarsals. The groove runs
first ungual. Metatarsal number 3 is ap- diagonally across the articulation from
preciably more slender than the others and ventrolateral to dorsomedial. \^entrally,
is slightly longer than the second or the medial to the groove a prominent heel pro-
fourth. Its proximal expansion is of similar jects downward. Laterally, just behind the
size and shape to that of metatarsal number articular surface lies an expanded process
2, but whereas that of the second is a that continued onto the shaft, making a
straight vertical surface, the third has a pronounced curvature of the lateral border
diagonal proximal surface with the dorsal of the shaft, and giving thc^ distal articu-
portion extended more posteriorly than the lation the aspect of being offset towards

348 Bulletin Museum of Comparative Zoology, Vol. 146, No. 7

Figure 10. Two views of a posterior scute of Sauro-
suchus. Left, dorsal; right, ventral. X V2.

the medial side (see Plate 3). Only one
phalanx of the fourth digit is present; it is
rectangular, heavily constructed, and in
general similar to that of the second digit,
although .somewhat smaller and flatter. The
proximal surface is more clearly divided
into lateral and medial concavities than in
the other digits. Distally the articular sur-
face is considerably flatter, and lacks the
downward extension of the rolling surface
found on the first phalanges of the other
digits. These are indications that the fourth
toe was probably long and relatively
slender. Specimen PVL 2267 has three iso-
lated articulated phalanges that probably
belonged to the fourth digit. These show a
rectangular shape that rapidly diminishes
in length distally with the last of the series,
probably the pre-ungual, little more than a
transverse rectangular chip of bone. How-
ever, the association of these three
phalanges (PVL 2267) is not certain. In
Ticinosuchus all of the phalanges are longi-
tudinally rectangular, as are all of the
proximal ones preserved in PVL 2557.
Metatarsal number five is a massive hook-
shaped element that bears a large hemi-
cylindrical articular surface on its medial
side for the matching concavity of the
fourth tarsal bone. On the anterodorsal
face of the surface is a small facet for the
lateral edge of the fourth metatarsal. Be-
hind the large ball surface, the posterior
border curves laterally and posteriorly to

terminate in a rounded point at the rear
lateral edge. From this point the lateral
margin curves out and forward to the
distal tip. A small expanded process is
present on the lateral edge one third of
the way back from the distal articulation.
The medial surface of the "shaft" curves
smoothly from the anterior tip of the major
proximal articulation to terminate in the
blunt surface of the distal articulation. This
articular surface bears neither flanges nor
grooves, but is a simple, slightly convex
surface. The first phalanx is rectangular
in shape, broader at the proximal end, and
bears an expanded concave articular sur-
face that partially envelopes the convexity
of the metatarsal. Distally, the phalanx
terminates in a simple flat vertical surface
devoid of rounded features. No other
phalanges are known for the fifth digit. The
fifth toe was widely separated from the
other digits.

Dermal Armour

Scutes have been found associated only
with PVL 219(S. These were found partially
articulated with the vertebral column, and
like most of the vertebrae, are poorly pre-
served. Three articulated scutes, much
smaller than the others, were found in as-
sociation with the other bones of the speci-
men, but not in a definable position. As
they are very well preserved, and in general
the degree of preservation becomes better
caudally in PVL 2198, it is assumed that
these scutes were from the posterior dorsal
region. Two paramedian rows of scutes
were present on the dorsal region of Saiiro-
siichus, the total width being 10 cm on
specimen PVL 2198. As preserved, the two
rows do not appear to have been joined by
a strong sutural contact. The dorsal scutes
are slightly asymmetrical and leaf-shaped
in outline, drawn to a point in front and
truncated at the rear. They are imbricated,
the wide rear margin overlapping the point
of the scute just caudal to it. Although the
two rows join at the midline, the medial
border is only slightly thicker than the

Saurosuchus and the Raltisuchid Thecodonts • Sill 349

lateral. The anterior point is slightly asym-
metrical; it is off center toward the medial
border. Dorsally the scutes are gently
arched in cross section, slightly more so on
the lateral side than on the medial. A keel
as such is not present, but there is a slight
longitudinal ridge. Possibly a small inden-
tation was present on the posterior border.
Ventrally there appears to be but a slight
indentation in the posterior portion to re-
ceive the point of the following scute. A
significant change in size takes place along
the length of the series; the posterior scutes
are smaller than the anterior ones, changing
from approximately 5 cm in width to 4.

The three isolated scutes differ consider-
ably from the otliers, but are of the same
pattern and certainly belong to the same
specimen. They are, however, perfectly
symmetrical, with each edge tapered to a
very thin border (see Fig. 10). Anteriorly
the point is longer and more tapered than
in the other scutes, and fits into a wedge-
shaped groove in the preceding scute. The
dorsal surface is prominently ridged in the
center, leading to the point anteriorly and
to an indentation posteriorly. These char-
acteristics suggest that these were members
of a single row of scutes, rather than paired.
A similar condition is reported for Ticino-
suchus by Krebs (1965), and is to be
expected given the other similarities of the
two genera. The greatest difference be-
tween the dorsal and the lumbar scutes is
size; the former are 5 cm wide and ap-
proximately 7 cm long while the latter are
3 cm wide and approximately 4 cm long.
This condition differs from that of Ticino-
suchus in which the scutes of the unpaired
row are larger than the paired. However,
the overall aspect of the armour of Sauro-
suchus is that it is more reduced relative to
the size of the animal than is that of
Ticinosuchus.

DISCUSSION

Origin of the Rauisuchidae

The anatomical characteristics of the figure 11. Pelvis of: A, S/7ans/st;c/,us (after Young);

B, Ticinosuchus (after Krebs); C, Rauisuchus (from a
known members of the family strongly photograph in Huene, 1942); D, Saurosuchus.

350 Bulletin Museum of Comparative Zoology, Vol. 146, No. 7

Figure 12. Left, calcaneum;
Shansisuchus {after Young).

right, astragalus of

suggest direct derivation from the erythro-
suchids, rather than from ornithosuchids or
from a common erythrosuchid-ornitho-
suchid ancestry. Cranial anatomy is but
little modified from the erythrosuchid con-
dition (see Fig. 13). Within the Erythro-
suchidae, the most advanced member (both
anatomically and stratigraphically) appears
to be Shmisisiichus from the Ehrmaying
Series of China (see Young, 1964; Reig,
1970; and Charig and Reig, 1970). This
genus provides a rather good intermediate
between the two families, and indeed was
tentatively included in the "Prestosuchidae"
by Romer ( 1972a ) . However, it still retains
the primitive pelvic girdle and simple tarsal
structure common to the Erythrosuchidae.
As locomotory abilities seem to have been a
principal evolutionary factor within the
Rauisuchidae, it would seem appropriate
to consider the less advanced Shansisuchus
as an erythrosuchid.

Major characteristics of the rauisuchids
that can be traced with a reasonable degree
of confidence through the lineage are:

1) Skull configuration: a keyhole-shaped
orbit, large antorbital fenestra surrounded
by a smooth depression, small supra-
temporal fenestra, high narrow cranial
table, and a posterior prong on the pre-
maxilla. Some of the genera have an ac-
cessory antorbital fenestra between the
premaxilla and the maxilla.

2) Vertebrae: high neural arch, straight
rectangular spine with distal expansion,
deep interspinous clefts.

3) Pelvis: prominent posterior spine,
presence of a brevis shelf, styliform ischium
with an expanded tip, greatly reduced
pubic plate, pubis with slight participation
in the acetabulum.

:-.-.:-.--'s>.-A.
.«;_ ./

Figure 13. Comparison of cranial morphology in A,
Stiansisuctius (after Young); B, Ticinosucfius (modified
from Krebs); C, Luperosuchus (from Romer); D, Sauro-
suctius. Not to same scale.

4) Femur: crocodilelike, without rounded
medial expansion.

5) Tarsus: ball and socket crocodiloid
type, fifth metatarsal hooked.

The several genera that make up the

Saurosuchus and the Rauisuchid Thecodonts • Sill 351

Rauisuchidae can be separated into three
morphologic groupings that reflect both
their stratigraphic position and their prob-
able phylogeny: 1) an early group, repre-
sented by Ticinosticlius from the earliest
Middle Triassic; 2) an extensive intermedi-
ate group represented by Luperosiichus,
StagonosucJnis, "Mandasuchus", Prestosu-
chus, and Raiiisuchus from the later Middle
Triassic; and 3) Saurosuchus and the unde-
scribed form from the Los Colorados, of
earlier and later Late Triassic respectively.
Ticinosuchus, the earliest member of the
family, and the only one known from a com-
plete skeleton, has a sk-ull that has been
higlily fractured and compressed to a
largely two-dimensional state. As recon-
structed by Krebs, the skull is similar, but
not strikingly so, to Saurosuchus and Lu-
perosuchus. However, using the more com-
plete knowledge afforded by the Argentine
specimens, it is possible to reinterpret to
some degree the skull of Ticinosuchus on
the basis of the published photographs. Two
modifications of Krebs' reconstruction ap-
pear feasible: the antorbital fenestra was
probably smaller than shown and was sur-
rounded by a smooth shelf, and the anterior
border of the maxilla was inflected just
above the tooth row, possibly indicating a
small accessory opening similar to that of
Saurosuchus.

Cervical vertebrae represent the only
anatomical character that shows a consider-
able degree of variation among the several
genera of the family. In Ticinosuchus the
cervicals are elongated, but otherwise un-
specialized. A similar condition appears to
be present in "Mandasuchus" but not in
Stagonosuchus, Prestosuchus, or Raui-
suchus. Only one cervical vertebrae is
known from Saurosuchus; it is a highly
specialized elongate structure so different
from other known forms that it is assigned
to the genus with reservation.

In the other comparable features char-
acteristic of the family, there is a remark-
able similarity among the genera definitely
assigned. More subtle differences dis-

tinguish Saurosuchus as the most progres-
sive of the described rauisuchids^; centra
of the vertebrae are constricted, the ischium
is longer and more rodlike, and the femur
is more gracile than the corresponding
features of the other genera.

Within the Ischigualasto Basin three
rauisuchids are found in the sequential
continental sediments. The earliest of these
is Luperosuchus from the Chanares For-
mation (Romer, 1971a). It has already
attained the large size characteristic of most
of the family, but is known only from an
incomplete skull. Changes in the skull from
Luperosuchus to Saurosuchus to the Los
Colorados form were slight; the orbit be-
came more circular in the upper portion
and the smooth shelf around the antorbital
fenestra is larger in the later genera. It
seems reasonable to assume that these three
forms were continuous members of a single
regional lineage. Very possibly Prestosuchus
from Brazil should be included in the
lineage. Prestosuchus is very comparable
to Saurosuchus; apparently the only signifi-
cant difference is that the femur of the
former appears to be more heavily con-
structed and less gracile than that of the
latter. Relationship of the Ischigualasto
Basin forms to other members of the family
is not as close. The vertebrae of Stagono-
suchus are somewhat constricted like those
of Saurosuchus, but the pelvis is more
primitive. "Mandasuchus" is quite similar
to Saurosuchus, and the two may be con-
generic or they may be closely related
forms similar to Prestosuchus and Raui-
suchus. Rauisuchus itself is less like the
other members of the family and its associ-
ation with the group has been questioned
(Charig 1967, Romer, 1972a, Walker, per-
sonal communication). Walker (personal
communication) has suggested that Raui-
suchus may be an ornithosuchid. His sug-
gestion is based principally on some aspects

^ The undescribed rauisuchid from the Los
Colorados Formation is larger than Saurosuchus;
it had a considerably more advanced tarsus, but
a very similar skull (Bonaparte, personal com-
munication ) .

352 Bulletin Museum of Comparative Zoology, Vol. 146, No. 7

of the skull fragments and on the dermal
armour. However, the premaxilla bears the
posterior projecting prong that separates
the external naris from the maxilla, and its
overall shape is similar to that of Sauro-
suchus and Luperosuchus. The ilium
figured by Huene (1942, plate 27) is re-
markably like that of Prestosuchus and
Saurosuchus (see Fig. 7). Other elements
are not as closely comparable, giving rise
to the doubts about the affinities of the
genus. However, the morphology of the
vertebrae and dermal armour are not in-
consistent with that of the other members
assigned to the family, and their resem-
blance to ornithosuchids may be superficial,
as are a number of the resemblances be-
tween the two groups (see discussion of
vertebrae and tarsus). For the present, I
would leave Rauisuchus in the family as-
sociation that is termed "Prestosuchidae"
by some authors, but recognize that it is
less comparable to the larger genera Presto-
suchus, Saurosuchus, and "Manclasuchus"
than these are to each other.

It would seem likely then that the
Brazilian and Argentine genera were part
of a South American radiation, perhaps
from a Luperosuchus-liVe stock. The Afri-
can forms, Manclasuchus and Stagono-
suchus, may represent a separate but
closely related line.

A summary of the evolutionary history of
the Rauisuchidae would then be: origin in
the early Middle Triassic from a progres-
sive group of erythrosuchids, the first
members of the family probably near the
Ticinosuchus level; adaptive radiation in
the Ladinian and Carnian; survival of
specialized members that could compete
with dinosaurs in the uppermost Triassic,
and extinction of the group by the Early
Jurassic (see Fig. 14).

Habits of the Rauisuchidae

On the basis of the known remains, the
rauisuchids can be described as large
quadrupedal animals ranging in total
length from three to six meters. The sharp

serrated dentition leaves no doubt that they
they were carnivores, and the deep narrow
skull would suggest predaceous habits.
During the Middle and Late Triassic they
were probably among the largest of the
terrestrial carnivores. Regarding locomo-
tion, the hind limbs were of the crocodiloid
grade of evolution, and as such the raui-
suchids were reasonably good runners,
although no doubt less agile that the later
dinosaurs and probably less agile than the
contemporary Ornithosuchidae.^ Rise of
the rauisuchids may have been parallel to
the rise of the rhynchosaurs and the gom-
phodont cynodonts during the Ladinian
and Carnian in a predator-prey relation-
ship. It is usually assumed that the large
thecodont predators disappeared during the
Late Triassic owing to the competition
from dinosaurs. However, the presence of
a very large, advanced rauisuchid in sedi-
ments considered to be Late Norian in age
(see Bonaparte, 1972a and Sill, 1969 for
details on the stratigraphic relationships of
the Argentine Triassic), would indicate that
these thecodonts had become adapted to
prey on the early saurischians, many of
which were herbivores. The last known
rauisuchid was a very large animal and had
an advanced digitigrade foot. Nevertheless,
the femur remained at the crocodiloid stage
of development, namely, without the for-
mation of a medial condyle or a shift to the
parasagittal plane of the body. Assuming
that the vertical position of the limbs was
an important adaptation, the rauisuchids
would have been at a disadvantage with
regard to the emerging carnivorous dino-
saurs. Such a relationship presumably
would explain the extinction of the group
as the dinosaurs became dominant.

Thecodont Taxonomy and Phylogeny

Although thecodonts have long been
recognized as the key group in the rise of

^ However, Bakker ( 1972, and in press ) has
shown by experimental data that the physiologic
cost of locomotion is dependent only on speed and
body weight, entirely independently of limb
posture.

Saurosuchus A^a) the Rauisuchid Thecodonts • Sill 353

00
CO

CD
CL

o
O

D
C

Undescribed

Saurosuchus

Stagonosuchus
Mandasuchus

Prestosuchus Rauisuchus

Luperosuchus

Ticinosuchus

Shansisuchus

Figure 14. Suggested phylogeny of the Rauisuchidae.

the archosaur faunas that dominated the
later Mesozoic, they have been a poorly
known and confusing group. As new dis-
coveries have been made in the last few
years there has been a renewed interest in
the order, and at last the prospect emerges
of unraveling the many and varied theco-
dont lineages. Traditionally, thecodonts
have been divided into three groups: 1) the
very primitive forms from the Early Tri-

assic, 2) the highly specialized taxa of the
Late Triassic, phytosaurs and aetosaurs,
and 3) the main stream, Pseudosuchia,
somewhat of an "everything else" suborder.
The n(>w discoveries ha\'e permitted the
clarification of some relationships, and
have added a new lineage, Proterochamp-
sidae, to the order. But the major relation-
ships are still far from settled, and there is
a considerable number of genera that do

354 Bulletin Museum of Comparative Zoology, Vol. 146, No. 7

LATE
TRIASSIC

MIDDLE
TRIASSIC

EARLY
TRIASSIC

LATE
PERMIAN

Figure 15. Suggested phylogeny of some thecodont lineages.

not fit into known families or even sub- by Romer (1972a) and by Bonaparte (1971),

orders. as follows (the sequential order followed

Current thinking on thecodont taxonomy by these authors has been changed to facili-

is perhaps best reflected in recent papers tate comparison):

Saurosuchus and the Rauisuchid Thecodonts • Sill 355

Ronier

Order Thecodontia

Suborder Proterosuchia
Family Proterosuchidae
Family Erythrosuchidae
Family Prestosuchidae ( =Rauisuchidae)
Family Proterochampsidae

Suborder Pseudosuchia^
Family Ornithosuchidae

Family Scleromochlidae

Suborder Aetosauria

Family Aetosauridae ( =Stagonolepidae)

Suborder Parasuchia ( Phytosauria )
Family Phytosauridae

Bonaparte
Order Thecodontia

Sul)order Proterosuchia
Family Proterosuchidae
Family Erythrosuchidae

Suborder Pseudosuchia
Infraorder Ornithosuchia

Family Ornithosuchidae

Family Rauisuchidae

Family Pallisteridae

Family Teleocrateridae ( ? )

Family Scleromochlidae
Infraorder Sphenosuchia

Family Sphenosuchidae

Family Triassolestidae
Infraorder Proterochampsia-

Family Cerritosauridae

Family Proterochampsidae

Suborder Aetosauria

Family Stagonolepidae (= Aetosauridae)

Suborder Parasuchia

Family Phytosauridae

^ The family Sphenosuchidae was placed by Romer in the suborder Protosuchia of the Crocodilia.
Teleocrater and T liassolestes, together with other poorly known genera, are not assigned to families.
"The suborder Archeosuchia was previously erected for the Proterochampsidae (Sill, 1967).

Both of these authors retain tlie usual
categories mentioned previously, but it
is interesting to note the different inteipre-
tations given to the newly defined line-
ages Rauisuchidae and Proterochampsidae.
Romer considers them to be continuations
of the primitive radiation, while Bonaparte
would suggest they are offshoots of the
pseudosuchian stock.

It is perhaps still premature to restructure
thecodont taxonomy, but the new dis-
coveries do make it possible for the first
time to trace some of the lineages through-
out the Triassic.

Primitive thecodonts consist of three
families; the ancestral stem Proterosuchidae
(see Cruickshank, 1972), the large terrestrial
Erythrosuchidae, derived from the Protero-
suchidae, and the progressive Euparkeri-
idae, usually considered the first of the
Pseudosuchia (see Ewer, 1965 and Charig

and Reig, 1970). The proterosuchids were
probably aquatic or semi-aquatic carni-
vores that somewhat resembled crocodiles.
Erythrosuchids show many characters that
relate them to the stem group, but were
fairly large terrestrial carnivores. Euparkeria
was apparently derived from an early line-
age that separated from the Erythrosuchidae
and evolved rapidly towards a more agile
locomotory system. It has usually been
assumed that it was the euparkeriid stock
that produced the later thecodont radiation
(Romer, 1966, and other textbooks). The
new discoveries of fossil thecodonts, in
particular those from South America, make
it possible to connect some evolutionary
lines of all three primiti\'e groups from the
Early to the Late Triassic.

As has been noted previously, the origin
of the Rauisuchidae almost certainly lies in
the Er\throsuchidae. Rauisuchids can be

356 Bulletin Museum of Comparative Zoology, Vol. 146, No. 7

traced through much of the Triassic with stones (Newton, 1894; Walker, 1964). To

closely related forms present in every stage these Romer ( 1972b ) recently added

of the period from the Anisian to the Gracilisuchus from the Chaiiares For-

Norian (see Fig. 14). The Proterochamp- mation. These genera in turn show reason-

sidae represents a newly defined lineage at ably close affinities to Euparkeria, and

present known only from South America, appear to represent a descendant lineage

Earliest members of the family are from from the euparkeriid type of early theco-

the Chaiiares Formation, Chanaresuchus dont.

and Giialosuchus, probably of Early In tracing these families from their
Ladinian or Late Anisian age (Romer origins in the early history of the Theco-
1971b). Later forms occur in the Santa dontia, mention has been made only of
Maria Formation of Brazil, Cerritosaunis those genera that are well enough known
(Price, 1946; Bonaparte, 1971), and in the to show definite relationships; there are, of
Ischigualasto Fonnation of Argentina, Pro- course, still many thecodonts whose system-
terochampsa (Reig, 1959; Sill, 1967). Most atic associations are not clear at present
members of the family show semiaquatic and who are usually assigned to families
tendencies, the Brazilian form Cerritosau- on a rather uncritical basis.
rus less so, and Froterochampsa itself more There remains the two well-known
so. The apparently more aquatic habitus of specialized suborders, the Aetosauria and
Proterochampsa was used by Bonaparte to the phytosaurs. In general these groups are
separate the other genera from it as the limited to the Late Triassic, although an
family Cerritosauridae, but the genera are isolated phytosaur has long been noted,
no doubt closely related and probably and disputed, from the Early Triassic of
should be placed in the same family. Plac- Europe (Jaekel, 1910; Gregory, 1962).
ing the Proterochampsidae as an infraorder Phytosaurs are well known morphologi-
of the Pseudosuchia implies a common cally, except for the tarsus, but no sure
origin after the acquisition of the pseudo- indication exists regarding their relation-
suchian adaptive characteristics. It appears ship to the primitive groups. In general it
more likely that the Proterochampsidae has been assumed that they were pre-
were independent derivatives of the primi- crocodile derivatives of the Pseudosuchia,
tive stem proterosuchians, as suggested by driven into extinction by the appearance of
Romer's classification, but they had ad- the true crocodiles (see Gregory, 1962).
vanced beyond tlie level common to the Howe^'er, phytosaurs were basically primi-
proterosuchids and erythrosuchids. On the tive animals, retaining additional skull
basis of the skull, I previously (Sill, 1967) elements that were lost early in the de-
believed them to be primitive crocodiles, velopment of the other thecodonts. Also,
but the posteranial material of the earlier the pelvic girdle consisted of large platelike
forms described by Romer renders this bones similar to the pattern of the primi-
interpretation unlikely. tive groups (see Camp, 1930; Gregory,
The third lineage to be well documented 1962, 1969). On the basis of the recently
is not new at all, but is the "mainline" fam- described proterochampsids, it seems pos-
ily Ornithosuchidae. As redefined by Bona- sible that phytosaurs may have been de-
parte (1972a) this family would be re- rived from an earlier continuation of the
stricted to the following well-defined aquatic forms of the Proterosuchia.
genera: Venaticosuchus from the Ischi- Aetosaurs are the other closely-knit group
gualasto Formation (Bonaparte, 1972b), of specialized thecodonts. Like phytosaurs
Riojasiichiis from the Los Colorados For- they are known principally from the Upper
mation (Bonaparte, 1969, 1972a), and Triassic, the earliest ones coming from the
Ornithosuchtis itself from the Elgin Sand- Ischigualasto Formation of Argentina (Car-

Saurosuchus and the Rauisuchid Thecodonts • Sill 357

nian).' Those from Iscliigualasto are fully
specialized members of the family, bearing
littl(> indication of primiti\'eness. Aetosaurs
were probably an early specialization for a
rooting, pig-like habit (see Walker, 1961).
Aside from their obvious specializations,
aetosaurs retain many primitive character-
istics common to the Erythrosuchidae and
Euparkeriidae. As noted by Ewer (1965),
Euparkeria was already more advanced in
its locomotory apparatus than the aetosaurs.
Therefore, the origins of the Aetosauria
must have been from a progressive line of
erythrosuchids or an early member of the
Euparkeriidae. If it is true that the
Euparkeria lineage represents an early de-
parture from the Erythrosuchidae, based
largely on limb specialization, then it would
be more likely that the aetosaurs were an
independent derivation from the erythro-
suchid stem, perhaps from the same group
that produced the rauisuchids.

Indirect anatomical evidence supporting
the affinity of Aetosauria with erythro-
suchids is found in the tarsus. It has long
been noted that the astragalus and cal-
caneum of aetosaurs is of the "crocodile-
type" in common with a number of other
thecodonts. The closest comparison of
these elements seems to be with the Raui-
suchidae (see Plate 4).

Another group of thecodonts, which has
long been particularly difficult to interpret
consists of those that share a number of
characteristics of the crocodiles, but are not
true crocodiles. These have been an enigma
since they were first discovered around
the turn of the century. They have been
considered alternately as stages in the
evolution of crocodiles (Huene, 1925),
independent lineages (Haughton, 1924) and
aberrant or primitive members of the
Crocodylia (Sill, 1967; Romer, 1972a).

^ It is possible that an aetosaur was present in
the earher Ehnnaying Series of China. A cal-
caneum figured by Young (1964:81) is very
much hke that of the Ischigualasto aetosaur, and
quite unlike that of ornithosuchians.

Walker ( 1970 ) has recently separated out
the crocodilelike thecodonts and placed
them as a suborder, Paracrocodylia, of
equal rank with the Crocodylia in a new
order Crocodylomorpha. Walker's work,
based largely on re-examination of Spheno-
siichus and HaUopus, indicates the presence
of a possibly unified lineage that shared
many anatomical characteristics of croco-
diles, but were not ancestral to them.
Whether or not a new order should be
erected to place this group in juxtaposition
with the Crocodylia will be decided by
future discoveries. At the moment it does
not seem to be justified. The Crocodylia
are a well-defined group. Walker's Para-
crocodylia is based on the Triassic family
Pedeticosauridae (or Sphenosuchidae), the
genus HaUopus — an apparent Jurassic de-
rivative of the earlier family — and the
Baurusuchidae, which he removes from the
crocodilian suborder Sebecosuchia. Such a
classification does not reflect the same
degree of natural grouping that is found
in the present category Crocodylia. It
would seem more reasonable at present to
consider the Pedeticosauridae as either a
derivation of the thecodont line that gave
rise to the true crocodiles, or as aberrant
crocodiles from the early radiation of the
Crocodylia.

An alternative possibility is that croco-
diles arose from an early branch of the
Ornithosuchidae, possibly a derivative of
the Euparkeria line, or from a continuation
of the Erythrosuchidae, perhaps from the
same stock that produced the Rauisuchidae
(and possibly aetosaurs). Evidence sug-
gesting the possibility of such a relationship
is found in the similarity of the crocodilian
tarsus to that of thecodonts in the above-
mentioned categories. The so-called croco-
dilian tarsal joint, in which the calcaneum
bears a prominent tuber and is functionally
part of the foot while the astragalus is fixed
to the crus, appears to have been better
developed in these lines than in either
Proterosuchus or the Proterochampsidae.
In addition, there appears to be a funda-

358 Bulletin Museum of Comparative Zoology, Vol. 146, No. 7

mental difference between the construction
of the tarsus in ornithosuchids and the
groups presumably derived from erythro-
suchids. In the Ornithosuchidae the major
joint between the proximal tarsal elements
is formed by a ball on the anteromedial
surface of the calcaneum and a correspond-
ing socket on the astragalus. On the other
hand, in rauisuchids, aetosaurs, and croco-
diles, the main socket is on the calcaneum
and the ball is on the astragalus. Both
forms appear to be functionally the same,
but possibly represent parallel evolutionary
paths. Recognition of this condition, first
noticed by Bonaparte ( 1971 ) , tends to
diminish the difficulty noted by Krebs
(1963) and Reig (1970) of explaining ap-
parently unrelated thecodonts that possess
very similar complicated tarsal joints. The
"true" crocodile tarsus then becomes an
impressive argument against derivation of
this group from the Ornithosuchidae-
Euparkeriidae type of pseudosuchian, and
would tend to suggest a closer affinity with
the erythrosuchid lineage, and the pre-
sumed derivatives of that line. Neverthe-
less, not enough is known about the tarsal
joint of the Proterosuchidae, Proterochamp-
sia, or Phytosauridae, to exclude them from
a common ancestry with the Crocodylia.
Tarsal joints of various members of the
Thecodontia are currently under study by
a number of paleontologists, some of whom
feel that the structure may represent a 4cey
to both thecodont and dinosaur phylogeny.
There remains a considerable number of
thecodonts that are not members of any of
the groups mentioned in this paper. Some
of these are almost certainly cladogenetic
derivatives of these groups ( see the generic
list in Romer, 1966, 1972a). The various
phylogenetic possibilities of these forms
have been discussed recently by Reig
(1970) and little more can be said until
additional fossil material is available. In
addition there are a number of "ghost
thecodonts," forms that have been named
and placed in the ordinal hierarchy, but

have never been duly described.^ These
forms, largely from critical Middle Triassic
strata, should provide additional insights
into the thecodont radiation.

Dinosaur origins remain unclear. Both
saurischian and ornithischian representa-
tives are present and clearly recognizable
in the Ischigualasto Formation of Argentina
(Late Ladinian-Early Carnian); saurischi-
ans occur in the earlier Santa Maria For-
mation of Brazil. There is no solid evidence
for linking saurischians with either ornitho-
suchid or rauisuchid thecodonts. However,
Charig ( 1967 ) suggested the possibility of
prosauropods arising from the latter group
(Prestosuchidae in his usage). Reig (1970)
considered it more likely that saurischians
had descended directly from an erythro-
suchid lineage than from a Eiiparkeria type
of thecodont. No clues at all exist regard-
ing the origin of the ornithischian dinosaurs;
the earliest representative {Pisanosaurus
from the Ischigualasto Formation) is a
fully developed member of the group.
It seems to be an inescapable conclusion
that dinosaurs separated from thecodonts
earlier than has usually been assumed, and
that most thecodonts were competitors of
dinosaurs rather than their progenitors.

Thecodont-Dinosaur Transition

It is perhaps paradoxical that the more
we learn about thecodont evolution the less
we know about dinosaur origins. Theco-

^ Mandasuchiis and Teleocrater were described
by Charig in his doctoral thesis of 1956 and the
names then published in an abstract in 1957. The
names were incorporated into the literature by
Huene (1956) and Romer (1966), but no formal
descriptions have ever been published. In a later
paper Charig, Attridge, and Crompton ( 1965 )
referred to the genera, but added a footnote to
the effect that they were nomina nuda. Charig
( 1967 ) mentions both genera, an additional one
from the same area, Pallisteria, and also two fami-
lies, Pallisteriidae and Teleocrateridae. As author
of all three genera and both families, he cites
Charig ( 1967 ) , a paper which has not yet been
published. All of these names, except Pallisteria
and its family, are listed in Romer (1966), but all
appear to be without proper foundation.

Saurosuciius anu the Rauisuciud Thecodonts • Sill 359

LD
00

or
o

THECODONTS

(J)
if)
<

q:
<

Ornithischians

2 3

Aetosourids :^-.r^j

Ornilhosuchids ond
crocodile'like forms

Rauisuchids and
other lorge forms

Phytosours
Proterochompsids

-4=-

■:-.v:-:

DINOSAURS

I Aquolic Carnivores 2 Small Herbivores 3 Large Herbivores 4 Lorge Cornivores t) Small Carnivores

Figure 16. Time-habitat relationships of thecodonts and dinosaurs (see text).

donts evidently were successful, wide-
spread, and diversified during the major
part of Triassic time. Yet dinosaurs, usually
considered as more or less the end result
of thecodont evolution, had their origins
well into the Middle Triassic (see Fig. 16).
Thecodonts and dinosaurs apparently lived
side by side during at least the hist half of
the Triassic. This situation naturally raises
some questions about the selective forces
involved and the nature of the competition
that presumably existed between the two
groups.

The superiority of dinosaurs relative to
thecodonts is usually ascribed to a shift
from a semi-erect to a fullv erect body
stance (Bakker, 1971; Charig, 1972). In

this case the more agile dinosaur loco-
motion supposedly would have driven the
thecodonts into extinction (but see foot-
note, p. 352). However, an early or tran-
sitional stage of dino.saurian limb posture
is not found in any of the known thecodonts,
and in particular there is no evidence of
the shift to the simple hinge t\'pe of foot
characteristic of dinosaurs. Charig (1972)
postulated an as yet unknown thecodont
ancestor in which the calcaneum was re-
duced and rotated with the cms rather
than with the pes. Reig (1970), on the
other hand, would have the dinosaurs origi-
nate directly from a primitive thecodont of
an ervthrosuchid level in the Earlv Triassic,
and evolve essentially independently of the

360 Bulletin Museum of Comparative Zoology, Vol. 146, No. 7

major thecodont radiation of the Middle
and early Late Triassic. However, if this
were the case it would be expected that
dinosaurs rather than thecodonts would
have dominated the Middle Triassic.

The earliest dinosaur remains currently
known come from the Manda and Santa
Maria Formations of approximately Anisian
or Ladinian age (Charig, 1967; Colbert,
1970). These genera, "Nyasasaunis" (un-
described) and Staurikosaiirus are con-
temporaries of rauisuchid thecodonts, found
in the same sediments ( "Mandasuchus" and
Prestosuchus). Staurikosaurus was more-
over a predator of approximately the same
size as Prestosuchus. A similar situation
obtains in the Ischigualasto Formation,
where the carnivorous dinosaur Herrera-
saiirus is found with the same size carnivo-
rous thecodont Saiirosuchiis. The earliest
ornithischian, Pisanosaurus, is found in the
Ischigualasto Formation and, although
poorly preserved, shows that the basic
features of the group had been acquired
by that time (Casamiquela, 1967). The
first theropods occur at approximately the
same time (Charig, 1967), apparently oc-
cupying an ecologic role parallel to that of
the ornithosuchid thecodonts.

Nevertheless, the thecodonts were con-
siderably more abundant and varied in the
sediments of the Middle and lower Late
Triassic. They apparently took over the
carnivore niche previously occupied by the
carnivorous cynodonts, but did not extend
into the herbivore field ( with the exception
of the aetosaurs ) . Dinosaurs produced both
carnivores and herbivores early in their
history. The origins of both categories are
still virtually unknown.

Actual data from the fossil record allow
three well -supported concepts to be stated:
1) dinosaurs were in existence at least
during the last half of the Triassic; 2)
thecodonts were abundant and diverse dur-
ing the Middle and first half of the Late
Triassic, becoming less so during the latter
part of the Late Triassic; 3) although dino-
saurs existed earlier, their major expansion

did not begin until the last half of the Late
Triassic. The reasons for the difference in
expansion phases between the two groups
are not clearly understood, nor can the
apparent ecologic overlap between the
large carnivores be explained on the basis
of current data. However, it seems an in-
escapable conclusion that the more agile
mechanical condition of the dinosaur limbs
was a factor in their eventual replacement
of the thecodonts. It is also possible that
the dinosaurs were undergoing more ex-
tensive physiologic changes, perhaps related
to the changes in locomotion (see Bakker,
1972).

LITERATURE CITED

Bakker, R. T. 1971. Dinosaur physiology and

the origin of mammals. Evolution, 25: 636-

658.
. 1972. Locomotor energetics of lizards

and mammals compared. The Physiologist,

15(3): 278.
-. In press. Lizard locomotor energetics

and the Reptile-Mammal transition.

Bonaparte, J. F. 1969. Dos nuevas "faunas"
de reptiles Triasicos de Argentina. Gondwana
Stratigraphy, lUGS Sumposium, Buenos Aires,
1-15 October 1967, UNESCO, pp. 283-284.

. 1971. Cerritosaiinis binsfeldi Price, tipo

de una nueva familia de Tecodontes (Pseudo-
suchia Proterochampsia ) . An. Acad. Brasil.
Cienc. (1971), 43: 417-421.

. 1972a. Los tetrapodos del sector superior

de la Formacion Los Colorados, La Rioja,
Argentina. Opera Lilloana, 22: 1-183.

1972b. Annotated list of the South

American Triassic tetrapods. Proc. II Gond-
wana Symposium, South Africa, 1970, Pre-
toria, pp. 665-682.

Camp, C. L. 1930. A study of the phytosaurs,
with description of new material from North
America. Mem. Univ. Calif., 10: 1-174.

Casamiquela, R. M. 1967. Un nuevo dinosaurio
ornitisquio Triasico {Pisanosaurus viertii:
Ornithopoda) de la Formacion Ischigualasto,
Argentina. Ameghiniana, 4(2): 47-64.

Charig, A. J. 1957. New Triassic archosaurs from
Tanganyika including Mandasuchus and
Teleocrater. Abstr. Diss. Univ. Cambridge,
1955-56: 28-29.

. 1967. Archosauria. The Fossil Record.

Geological Soc. London, pp. 695-731.

. 1972. The evolution of the archosaur

pelvis and hindlimb: an explanation in func-
tional terms. In Joysey, K. A. and T. S. Kemp

Saurosuchus and the Rauisuchid Thecodonts • Sill 361

(Eds.), Studies in Vertebrate Evolution.
Edinburgh: Oliver and Boyd, pp. 121-155.
— , J. Attiuu(;e, and A. W. Crompton
1965. On the origin of the sauropods and
the classification of the Saurischia. Proc.
Linn. Soc. London, 176: 197-221.
-, AND O. Reig. 1970. The classification of

eines Pseudosuchiers aus der Trias des Monte
San Giorgio (Kanton Tessin, Schweiz).
Paleont. A., 37: 88-95.

1965. Ticinosuchtis ferox nov. gen. nov.

the Proterosuchia. Bio. Jour. Lin. Soc, 2:

125-171.
Colbert, E. H. 1970. A saurischian dinosaur

from the Triassic of Brazil. Am. Mus. Novi-

tates. No. 2405: 1-39.
Cruickshank, a. R. I. 1972. The proterosuchian

thecodonts. In Joysey, K. A. and T. S.

Kemp (eds.), Studies in Vertebrate Evolution.

Edinburgh: Oliver and Boyd, pp. 89-119.
Edmund, A. G. 1957. On the special foramina

in the jaws of many ornithischian dinosaurs.

Cont. Roy. Ontario Mus. Zool. and Palaeo.,

No. 48: i-14.
. 1960. Tooth replacement phenomena in

the lower vertebrates. Roy. Ontario Mus.,

Life Sci. Div., 52: 1-64.
Ewer, R. R. 1965. The anatomy of the thecodont

reptile Etiparkeria capensis Broom. Phil.

Trans. Roy. Soc. London, B, 248: 379-435.
Gregory, J. T. 1962. The genera of phytosaurs.

Amer. Jour. Sci., 260: 652-690.
, AND F. Westphal. 1969. Remarks on

the phytosaur genera of the European Trias.

Jour. Paleont., 43: 1296-1298.
Haughton, S. H. 1924. The fauna and stratig-
raphy of the Stonuberg Series. Ann. So.

Africa Mus., 12: 323-497.
Hoffstetter, R. 1955. Thecodontia. In Traite

de Paleontologie, 5: 665-694.
HuENE, F. V. 1925. Die Bedeutung der Spheno-

suchtis Gruppe fiir Ursprung der Krokodile.

Z. Indukt. Abstamm.-Vererbungsl., 38: 307-

320.
. 1936. The constitution of the Theco-
dontia. Amer. Jour. Sci., Ser. 5, 32: 207-217.
. 1938. Ein grosser Stagonolepide aus der

jiingeren Trias Ostafrikas. Neues Jahrb. Min.

Geol. Pal., Beilage-Bd., 80: 264-278.
. 1942. Die fossilen Reptilien des Siida-

merikanishen Gondwanalandes. Ergebnisse

der Sauriergrabungen in Siidbrasilien, 1928/

1929. Miinchen: C. H. Beck'she. 332 pp.
-. 1956. Paliiontologie und Phylogenie der

niederen Tetrapoden. Jena: G. Fischer. 716
pp.

Hughes, B. 1963. The earliest archosaurian rep-
tiles. S. Afr. Jour. Sci., 59: 221-241.

Jaekel, O. 1910. Ueber einen neuen Belodonten
aus dem Buntsandstein von Bemburg. Sit-
zungsber. Ges. naturf. Freunde, Berlin, 1910:
197-229.

Krebs, B. 1963. Ban und Funktion des Tarsus

sp. Ein neuer Pseudosuchier aus der Trias des
Monte San Giorgio. Schweiz. Palaont. Abh.,
81: 1-140.

Newton, E. T. 1894. Reptiles from the Elgin
sandstone. Description of two new genera.
Phil. Trans. Roy. Soc. London, B, 185:
573-607.

Price, L. I. 1946. Sobre um novo pseudosuquio
deo Triassico superior do Rio Grande do Sul.
Bol. Ser. Geol. Min. Brasil, 120: 7-38.

Reig, O. A. 1959. Primeros datos descriptivos
sobre nuevos reptiles arcosaurios del Triasico
de Ischigualasto (San Juan, Argentina). Rev.
Asoc. Geol. Argentina, 13: 257-270.

. 1961. Acerca de la posicion sistematica

de la familia Rauisuchidae y del genero
Saurosuchus (Reptilia, Thecodontia). Publ.
Mus. Munic. Cien. Nat. Trad. Mar de la
Plata, 1: 73-114.

. 1970. The Proterosuchia and the early

evolution of the archosaurs; an essay about
the origin of a major taxon. Bull. Mus. Comp.
Zool., 139: 229-292.

RoMER, A. S. 1927. The pelvic musculature of
ornithischian dinosaurs. Acta Zoologica, 8:
225-275.

. 1956. Osteology of the Reptiles. Chicago:

Univ. Chicago Press. 772 pp.

. 1966. Vertebrate Paleontology. Chicago:

LTniv. Chicago Press. 468 pp.

. 1968. Notes and comments on vertebrate

paleontology. Chicago: Univ. Chicago Press.
304 pp.

. 1971a. The Chaiiares (Argentina) Tri-
assic reptile fauna. VIII. A fragmentary skull
of a large thecodont, Luperosiichus fractus.
Breviora, Mus. Comp. Zool., No. 373: 1-8.

. 1971b. The Chaiiares (Argentina) Tri-
assic reptile fauna. IX. Two new long-snouted
thecodonts, Chanaresuchus and Gualosuchus.
Breviora, Mus. Comp. Zool., No. 379: 1-22.

. 1972a. The Chaiiares (Argentina) Tri-
assic reptile fauna. XVI. Thecodont classi-
fication. Breviora, Mus. Comp. Zool. No.
395: 1-24.

. 1972b. The Chaiiares (Argentina) Tri-

assic reptile fauna. XIII. An early ornitho-
suchid pseudosuchian, Gracilisuchus stipani-
ciconini, gen. et sp. nov. Brexiora, Mus.
Comp. Zool., No. 389: 1-24.

SfLL, W. D. 1967. Frotcrochanipsa harrionucvoi
and the earh' evolution of the Crocodilia. Bull.
Mus. Comp. Zool., 135: 415-146.

. 1969. The tetrapod-bearing continental

362 Bulletin Museum of Comparative Zoology, Vol. 146, No. 7

Triassic sediments of South America. Am.

Jour. Sci, 267: 805-821
Walker, A. D. 1961. Triassic reptiles from the

Elgin area: StagonoJepis, Dasijgnatluts and

their allies. Phil. Trans. Roy. Soc. London,

B, 244: 103-204
. 1964. Triassic reptiles from the Elgin

area: Ornithosuchus and the origin of carno-

saurs. Phil Trans. Rov. Soc. London, B, 248:

53-134.
. 1970. A revision of the Jurassic reptile

HaUopus victor (Marsh) with remarks on the

classification of crocodiles. Phil. Trans. Roy.

Soc. London, B, 257: 323-372.
Young, C. C. 1964. The pseudosuchians in China.

Palaeont. Sinica, New Ser. C, 19: 106-205.

us ISSN 0027-4100

Sulletln OF THE

Museum of

Comparative

Zoology

The Cranial Foramina of

Protrogomorphous Rodents;

An Anatomical and Phylogenetic Study

JOHN H. WAHLERT

HARVARD UNIVERSITY

CAMBRIDGE, MASSACHUSETTS, U.S.A.

VOLUME 146, NUMBER 8
18 DECEMBER 1974

PUBLICATIONS ISSUED

OR DISTRIBUTED BY THE

MUSEUM OF COMPARATIVE ZOOLOGY

HARVARD UNIVERSITY

Breviora 1952-

BULLETIN 1863-

Memoirs 1864-1938

JoHNSONiA, Department of Mollusks, 1941-

OccASioNAL Papers on Mollusks, 1945-

SPECIAL PUBLICATIONS.

1. Whittington, H. B., and E. D. I. Rolfe (eds.), 1963. Phylogeny and
Evolution o£ Crustacea. 192 pp.

2. Turner, R. D., 1966. A Survey and Illustrated Catalogue of the Teredini-
dae (Mollusca: Bivalvia). 265 pp.

3. Sprinkle, J., 1973. Morphology and Evolution of Blastozoan Echinoderms.
284 pp.

4. Eaton, R. J. E., 1974. A Flora of Concord. 250 pp.

Other Publications.

Bigelow, H. B., and W. C. Schroeder, 1953. Fishes of the Gulf of Maine.
Reprint.

Brues, C. T., A. L. Melander, and F. M. Carpenter, 1954. Classification of
Insects.

Creighton, W. S., 1950. The Ants of North America. Reprint.

Lyman, C. P., and A. R. Dav^^e (eds.), 1960. Symposium on Natural
Mammalian Hibernation.

Peters' Check-list of Birds of the World, vols. 2-7, 9, 10, 12-15.

Proceedings of the New England Zoological Club 1899-1948. (Complete
sets only.)

Publications of the Boston Society of Natural History.

Price list and catalog of MCZ publications may be obtained from Publications
Office, Museum of Comparative Zoology, Harvard University, Cambridge, Massa-
chusetts, 02138, U.S.A.

THE CRANIAL FORAMINA OF PROTROGOMORPHOUS RODENTS;
AN ANATOMICAL AND PHYLOGENETIC STUDY

JOHN H. WAHLERr

Dedicated to

Katherine

Alexander and James

Carol and Daniel

CONTENTS

LIST OF FIGURES 363

ABSTRACT 363

INTRODUCTION 364

CRANIAL FORAMINA OF RODENTS
WITH SPECIAL REFERENCE TO

MARMOTA 366

Systems:

Cranial nerves 367

Arteries . 368

Veins 369

Foramina of the rodent skull 369

PARAMYIDAE

Param ys 374

Leptotomus 379

Reithroparannjs 380

Ischyrotumus 381

Fseudotomus 383

Manitsha 384

SCIURAVIDAE .-... 385

ISCHYROMYIDAE 388

CYLINDRODONTIDAE 393

PROSCIURIDAE 397

APLODONTOIDEA 400

CONCLUSIONS 405

REFERENCES 408

LIST OF FIGURES

1. Marmota monax 370

2. Paramys copei _ _ 375

3. Paramys delicattis 376

4. Auditory and pterygoid regions of

Paramys copei 377

^ American Museum of Natural History, Verte-
brate Paleontology Department, Central Park West
at 79th Street, New York, N. Y. 10024.

5.
6.

9.
10.
11a.
lib.
12.
13.

Reithroparamys delicatissimus 380

Auditory and pterygoid regions of

Ischyrotomus oweni 382

Sciiiraviis nitidus .__. 385

Auditory region of Sciuravus nitidus 387

Ischyromys typus 389

Ardyitomys occidentalis 394

Prosciurus sp. 398

Prosciurus aff. saskatchewaensis 398

Allomys nitens 400

Mijlugauhis laevis 401

Abstract. The cranial foramina and the blood
vessels and nerves passing through them are de-
scribed in detail for the sciurid genus Marmota;
this data serves as the basis for understanding
structures seen in the fossils. The cranial foramina
are described and compared in North American
specimens of the protrogomorphous rodent families
Paramyidae, Sciuravidae, Ischyromyidae, Cylindro-
dontidae, Prosciuridae, Aplodontidae, and Myla-
gaulidae. The least variable foramina are those
that transmit nerves; the most variable, veins.
Presence or absence, relative position, number,
and relative size of foramina are useful characters
in determining relationships. Within the Para-
myidae differences indicate an early radiation of
lineages. Paramyids and sciuravids ha\'e man\'
primitive features in common, but differ in several
details; of especial interest in these families are the
pathways of the internal carotid artery and its
branches. Peculiarities common to the foramina of
ischyromyids and cylindrodontids suggest that the
two groups can be made subfamilies of the family
Ischyromyidae. The Prosciuridae are included like-
wise with the Aplodontidae and Mylagaulidae in
the Aplodontoidea.

Bull. Mus. Comp. Zool., 146(8) : 363-410, December, 1974 363

364 Bulletin Museum of Comparative Zoology, Vol. 146, No. 8

INTRODUCTION

The origin of the Rodentia and their
successful radiation can be attributed to a
unique design for gnawing and chewing.
Perfection of the design has involved modi-
fication of the jaw and skull for more
efficient muscle configuration, and special-
ization of the incisors and cheek teeth in
response to the multitude of specific niches
into which rodents have diversified. The
masticatory system has been subjected to
great selection pressure and has been modi-
fied from the original design in ways that
were limited by genetic potential and by
the efficiency of certain modifications
relative to others. These are the principal
reasons for the parallelism so typical of
rodent phylogeny.

To date, the classification of rodents has
been based primarily on the structure of
the masticatory muscles, the infraorbital
foramen, the lower jaw, and the cheek
teeth. These characters are all part of the
masticatory system, and, when traced
through time, their observed modifications
reveal a complex phylogeny. Gaps in the
sequence, however, cannot always be
filled. Whole families of rodents stand in
uncertain relationships to proposed phylog-
enies. This situation is not surprising; in
a phase of rapid evolution a gap of a few
million years is enough to permit a discrete
group to appear full-blown in the fossil
record. The ancestry of such a group is
often unclear because of parallelism among
the earlier lineages from which it could
have descended.

The cranial foramina, unlike the com-
ponents of the masticatory apparatus, are
not part of a single functional system. There
is no reason to suppose that selection acts
on them as a unit or that selection pressure
from the external environment acts on them
directly. In the main, foramina serve a
passive function; they permit nerves and
blood vessels to pass through the bones of
the skull. It is reasonable to suppose that
foramina may vary in position and soft-

part content so long as tliey satisfy the re-
quirements of the circulatory and nervous
systems. Within these limits selection is
unimportant, and changes fixed in a small
population by random genetic processes
will characterize a new lineage arising from
it. Fusion and division of foramina are pos-
sible examples.

The position or the existence of foramina
may be changed as they are impinged upon
by other structures. Foramina in the orbit
are modified to lead around the roots of
high-crowned cheek teeth. In the temporal
region they may be closed off by enlarged
bullae, and some other combination of
foramina then acquires their function.

A foramen may be taken over by a dif-
ferent functional system. The infraorbital
foramen has been seized upon in the
hystricomorphous and myomorphous ro-
dents for transmission of a part of the
medial masseter muscle. From the moment
of seizure it ceased to behave solely as the
foramen it was and came under the in-
fluence of the selective forces acting upon
the masticatory system. The tough con-
nective tissue around a foramen may
change to accommodate a new stmcture.
In those sciurids which lack an infraorbital
canal, a tough membrane shielding the
transmitted nerves and vessels from the
lateral division of the masseter takes its
place.

Hill (1935 and 1937), Guthrie (1963 and
1969), and Bugge (1970, 1971a, b, and c)
have been the principal contributors to
knowledge of cranial foramina and the
cephalic nervous and vascular systems in
living rodents. They describe differences
that appear to have a systematic basis. But
the very nature of their work, limited pri-
marily to modern examples, precludes dis-
cernment of the primitive and derived
conditions for each aperture. The pattern of
evolution can be seen with certainty only
when the time-dimension of paleontology
is added. Detailed consideration of the
fossils indicates which features in a group
are primitive and eliminates the need to

rely on a so-called, but not in fact, primi-
tive living genus such as Aplodontia.

This paper on the protrogomorphous
North American rodents is the first half of
my Ph.D. dissertation (Wahlert, 1972),
which included the scimomorphs also. De-
scription of the cranial foramina in the
latter families will be presented elsewhere,
and I hope to extend the work to myo-
morphous and hystricomorphous forms. The
Protrogomorpha as defined by Wood (1937
and 1955) contain the families Paramyidae
and Sciuravidae, which are parts of the
initial rodent radiation, and the derived
families Cylindrodontidae, Ischyromyidae,
Prosciuridae\ Mylagaulidae, Aplodontidae,
and Protoptychidae. Protoptychus was
found to be both hystricomorphous and
hystricognathus; a separate paper deals
with its cranial and dental morphology
(Wahlert, 1973).

This approach to the study of rodent
evolution brings with it a special set of
problems. The number of fossil skulls
adequately preserved is very small in com-
parison with the number of fomis known
from teeth. The forms whose skulls can be
examined include representatives of every
family, but they may be from specialized
side branches and not from the main lines
of evolution. Most specimens are incom-
plete. The task of assembling data may be
compared with that of a man in the dark
who attempts to describe an exquisite
topiary arabesque with only the aid of an
unreliable flashlight.

The text is divided into sections, each
dealing with a single taxon; for extinct
lineages this is the family, and for surviving
lineages, the superfamily. Sections are sub-
divided according to the importance of the
included material and the completeness of
the specimens.

Paramyid genera are considered sepa-

^I have followed Wilson (1949c) and assigned
the prosciurids to a taxon of rank eqnal to the
paramyid group. Wood places them in the Para-
mvidae as a subfamily.

Cranial Foramina • Wahlert 365

rately because many rodent lineages may
have originated from within the family.
Differences between genera may be critical
in determining relationships to later forms,
and it is important to recognize that in-
formation about cranial foramina in the
fossils is quite uneven.

The genera within several families and
even within a superfamily are enough alike
that a single section describing each group
is sufficient. The ischyromyids, cylindro-
dontids, prosciurids, and aplodontoids are
treated in this manner.

Discussions at the end of each section
compare features within the groups de-
scribed and compare the most interesting
features of the entire assemblage with those
considered in preceding sections.

The bearing of the evidence provided
by cranial foramina on the phylogeny and
relationships of North American protrogo-
morphous rodents is discussed in the con-
clusion.

A list of the specimens examined is
presented at the beginning of each section
or subsection. Definitions of the strati-
graphic names can be found in Wood (H.
E. Wood et ah, 1941) and Keroher (Keroher
et al, 1966; Keroher, 1970).

Abbreviations are as follows:

AMNH American Museum of Natural
History

CM Carnegie Museum of Natural

History

F:AM Flick Collections, American Mu-

seum of Natural History

FMNH Field Museum of Natural His-
tory

KU University of Kansas Museum

of Natural History

LACM Los Angeles County Museum
(CIT) (California Institute of Tech-

nology Collection)

MCZ Museum of Comparative Zool-

ogy, Harvard University

USNM National Museum of Natural
History

366 Bulletin Museum of Comparative Zoology, Vol. 146, No. 8

SDSM South Dakota School of Mines
and Technology

UCMP University of California Mu-
seum of Paleontology

UNSM University of Nebraska State
Museum

UOMNH University of Oregon Museum
of Natural History

YPM Peabody Museum of Natural

History, Yale University

A letter code, which follows each num-
ber, indicates the completeness of the fossil
specimens:

s - whole skull o - orbit

n - snout t - pterygoid region

p - palate c - cranium

A code such as npo indicates that the snout,
palate, and orbit of the particular speci-
men are preserved and provided informa-
tion for this study; the pterygoid region
and cranium are either gone or are dam-
aged and the detail destroyed.

Measurements of length were taken with
a dial caliper calibrated to 1/20 mm. The
diastemal length is a straight line measure-
ment from the back of the incisor alveolus
to the anteriormost edge of the alveolus of
the first cheek tooth. The sizes of foramina
smaller than 1.0 mm were estimated with
a Dunlap spark-plug gauge.

Most of the figures were drawn with the
aid of proportional dividers; enlargements
of detail and outlines of small specimens
were traced with a camera lucida micro-
scope. I have made no attempt to show
crenulations in the sutures but have taken
care to illustrate the relationship of sutures
to foramina. I have omitted detail from
the teeth because excellent figures of the
dentitions of all species studied are avail-
able in the literature. Solid lines indicate
structures and sutures that I have seen in
at least one specimen of the genus illus-
trated. Dashed lines indicate details that
are less certain but probable in view of
similar features in closely related fonns.
Dotted lines represent guesses. To some

degree all the figures are restorations. I
have attempted to eliminate distortions and
to reconstruct all broken elements; the
figures are not copies of the specimens. The
key to abbreviations in the illustrations is
given in the caption of Figure 1.

Acknowledgements. I would like to
thank Bryan Patterson for his assistance
and criticism throughout the course of this
study, and Albert Wood for his suggestion
of the topic and his assistance in the initial
stages of the work. I am indebted to the
staff members of the institutions, listed
above, for making such a wealth of material
available, and to many of these same
people for their kind hospitality when I
toured museums. I appreciate greatly the
comments of Craig C. Black, Mary R.
Dawson, Robert J. Emry, and T. Mylan
Stout, all of whom read parts of the manu-
script, and of Parish A. Jenkins, Jr., who
read the entire thesis and had many excel-
lent suggestions for improvement of this
manuscript. Barbara Lawrence and Charles
Mack of the Mammal Department, Mu-
seum of Comparative Zoology, provided
me with skulls for sectioning and specimens
for dissection.

Travel was financed by the Departments
of Geological Sciences and of Biology at
Harvard University. Other expenses were
generously sustained by Katherine H.
Wahlert.

Special thanks are due to Carol C. Jones
for incisive criticism of my grammar and
for advice on the figures, to Daniel C.
Fisher and James M. Labaugh, HI, for
help in preparing the final manuscript, and
to Katherine H. Wahlert for devoted
typing.

CRANIAL FORAMINA OF RODENTS
WITH SPECIAL REFERENCE
TO MARMOTA

Hill (1935) was the first to attempt a
complete listing of the foramina in rodent
skulls. From dissections and prepared skulls
he described the position and contents of
each foramen and stated how thev differ

Cranial Foramina • Wahlcrt 367

among scxcral grncra, but he gave no ac-
count of the circulatory and nervous
systems themselves. Unless tliese systems
are understood it is not possible to interpret
and name sex'cral of the foramina. Tandler
(1899. 1901, and 1902), Guthrie (1963 and
1969), and Bugge (1970, 1971a, b, and c)
have examined the cephalic arterial circula-
tion of various rodents, but, apart from
Guthrie, these authors pay little attention
to the foramina involved.

As an introduction to what follows, I
present an account of the foramina and the
circulatory system and cranial nerves in
Marmota monax (Fig. 1). I follow Hyman
(1942) and Greene (1935) for terminology
of the soft parts. Marmota has several ad-
vantages for this purpose. Although fully
sciuromorphous, it retains most of the
cranial foramina met with in the earliest
rodents, and the bones of the skull do not
fuse in the adult.

Systems

Cranial Nerves

The hypoglossal (XII) emerges from one
or more hypoglossal foramina and runs
anteromedially into the base of the tongue.
The foramina are situated just anterior to
the occipital condyle on the ventral side of
the skull.

The vagus (X), accessory (XI), and
glossopharyngeal (IX) emerge from the
jugular foramen. It is between the bulla
and the basioccipital and is lenticular in
shape.

The facial (VII) emerges from the stylo-
mastoid foramen deep between the mastoid
process and the bulla. The main part of
the nerve runs anteriorly and diversifies
over the masseter muscle. The chorda
tympani branch of the facial emerges from
a tiny slot, the canal of Huguier, in the
front surface of the bulla. It runs antero-
medially to join the lingual branch of the
trigeminal (V) nerve.

The mandibular division (3rd) of the
trigeminal nerve (V) emerges from the

large foramen ovale in the pterygoid region.
Initially it runs anterolaterally through a
notch in the lateral pterygoid flange. A
strut of bone may cross the notch to form
a foramcMi which I am calling the foramen
ovale accessorius. The strvit separates the
mandibular nerve and its internal ptery-
goid branch. The auriculotemporal branch
diverges just outside the foramen ovale.
The mandibular nerve continues around
the outside of the external pterygoid muscle
and splits into three parts. The inferior
alveolar branch enters the mandibular
canal of the lower jaw. The mylohyoid and
lingual, which is joined by the chorda
t>anpani, run around the muscle and turn
medially into the soft tissue.

Two portions of the mandibular division,
the masseteric and buccinator nerves, run
dorsolaterally from the foramen ovale
through a canal in the alisphenoid bone;
the canal shares its posterior opening with
the alisphenoid canal, but runs through the
bone above it. These nerves emerge on the
side of the head. Both nerves may pass
through one foramen or each through its
own (masticatory and buccinator foram-
ina); the two cases can occur together on
opposite sides of a single skull. The bucci-
nator nerve runs anteriorly, but a small
branch turns back on leaving the foramen.
The masseteric nerve also has two branches;
the smaller runs dorsally to the temporal
muscle. The main part remains against the
alisphenoid region in a shallow vertical
channel. When this branch reaches the
front of the posterior root of the zygoma,
it turns laterally, passes through the
mandibular notch of the jaw, and descends
to the masseter muscle.

The maxillary division (2nd) of the
trigeminal nerve enters the orbit through
the sphenoidal fissure. On cutting away
the lateral surface of the alisphenoid region,
the alisphenoid canal is exposed. Two large
branches of the maxillary enter the canal
dorsally through two large foramina; the
small zygomatic branch emerges in some
cases from a separate small foramen be-

368 Bulletin Museum of Comparative Zoology, Vol. 146, No. 8

tween them. The posterior foramen (or
two foramina) seems comparable to the
foramen rotundum in other mammals. The
two branches of the maxillary division unite
to form the infraorbital nerve. The vidian
nerve could not be separated or distin-
guished from these. The main trunk of the
infraorbital nerve enters the infraorbital
canal; inside the canal a twig, the anterior
superior alveolar nerve, descends into the
maxillary bone. The trunk continues out
onto the side of the snout. A small medial
branch, the sphenopalatine, comes off the
infraorbital nerve where it enters the orbit;
it re-enters the skull through the spheno-
palatine foramen. As it crosses the orbital
floor it gives off a descending palatine
branch, which enters the dorsal palatine
foramen, runs through the palatine canal,
and emerges on the palate through the
posterior palatine foramen.

In company with the anterior portion of
the maxillary division, the ophthalmic
division ( 1st ) of the trigeminal, and the
abducens (VI), trochlear (IV), and oculo-
motor (III) nerves enter the alisphenoid
canal. The foramen through which they
pass is comparable to the orbital fissure in
other mammals. The nasociliary branch of
the ophthalmic re-enters the skull through
the more anterior of the two ethmoid
foramina. The frontal branch of the
ophthalmic ascends the medial wall of the
orbit and exits onto the top of the skull
through the supraorbital notch.

The optic nerve (II) enters the orbit
through the large, oval optic foramen. It is
situated entirely within the orbitosphenoid.

Arteries

The common carotid artery gives off
three branches when it reaches the back of
the larynx. The superior thyroid artery
diverges medially; the stapedial and oc-
cipital arteries branch off on the lateral
side. The main trunk continues as the
external carotid. The occipital artery turns
posteriorly and crosses ventral to the sta-
pedial. It runs through a channel between

the condyle and the paroccipital process to
the back of the head and neck.

The stapedial artery in company with
the vagus, accessory, and glossopharyngeal
nerves passes through the jugular foramen,
and it enters the stapedial foramen in the
bulla. It exits from the middle ear and
enters the cranial cavity via the stapedial
artery canal in the periotic. A dorsal branch
from it continues out the temporal foramen
to the temporal muscle. The main portion
runs anteriorly and exits via the spheno-
frontal foramen into the orbit; this is the
ophthalmic artery, which supplies the eye
and eye muscles with blood. One branch,
the ethmoidal artery, enters the postero-
dorsal ethmoid foramen. Another, the
superior ophthalmic artery, ascends the
medial wall of the orbit with the frontal
branch of the ophthalmic nerve and goes
through the supraorbital notch onto the top
of the head.

The external carotid artery bends later-
ally and gives rise to auricular, internal
maxillary, and other branches which supply
the lower jaw, jaw muscles, and ear region
with blood. At the bend, the external
maxillary artery diverges and nms an-
teriorly between the masseter and digastric
muscles. It gives off a lingual artery and a
glandular branch. In this region a third
branch proceeds dorsally, gives off a tiny
meningeal twig to the foramen ovale, enters
the alisphenoid canal, and passes as the
internal maxillary into the orbit where it
divides into three branches.

The outermost branch of the internal
maxillary artery, the posterior superior
alveolar, runs anterolaterally to the cheek
region. The middle branch, the infraorbital,
gives off minute branches tliat enter the
nutritive foramina. It passes through the
infraorbital foramen where a miniscule
orbital twig pierces the bone dorsally,
emerges from the malar foramen, and goes
into the tissue anterior to the eye; a ventral
branch, the anterior superior alveolar, in
company with the nerve of the same name,
enters a foramen below. The main trunk

Craxial Foramina • Wahlert 369

continues out onto the snout. The inner-
most branch of the internal maxillar}' arter>'
gives rise to the descending pahitine artery
and continues on into the sphenopalatine
foramen. The descending palatine artery
enters the dorsal palatine foramen, runs
through the palatine bone, and emerges
from the posterior palatine foramen; it
diversifies in the tissue of the palate and
disappears again into the incisive foramen.

Veins

Three distinct ti'unks carry blood from
xarious parts of the head. These are the
anterior and posterior facial veins, which
unite in the neck to form the external
jugular vein, and the internal jugular vein.

The posterior facial recei\'es blood from
the temporal and orbital-palatine regions.
The infraorbital vein begins on the snout
and passes through the infraorbital canal
where it picks up a small tvvig from the
anterior alveolar foramen. In the orbit, as
the internal maxillar\' \'ein, it collects tsvigs
from the nutritive foramina and small
branches from veins passing through the
sphenopalatine foramen and palatine canal.
The descending palatine vein ascends
through the posterior maxillary notch and
joins it at the back of the maxilla. There
are t\vo ethmoid foramina, and the ethmoid
\ein exits through the posterodorsal one. It
joins the ophthalmic, which then unites
with the internal maxillary just before it
enters the sphenoidal fissure. The internal
maxillary occupies most of the space within
the alisphenoid canal, the internal maxillary
artery filling only a small dorsal portion of
the canal. The vein communicates through
the transverse canal in the basisphenoid
with the same vein on the opposite side.
It empties into the pterygoid plexus.

The superficial temporal vein gatliers the
posterior deep temporal, transverse facial,
masseteric, and auricular branches. It is
joined by a large vein from the temporal
foramen and condylar area. This broad
\'essel also continues into the pterygoid

plexus. The inferior alveolar vein enters
the plexus from the mandibular foramen in
the jaw. The plexus anastomoses dorsally
with the internal maxillary vein and ven-
tralh' with the submental vein. A meningeal
branch enters it through a small foramen
bet^veen the bulla and the basisphenoid
bone; this aperture may be a remnant of
the middle lacerate foramen. The pterygoid
plexus changes from a .sack-like structure
into a large vein that proceeds posteriorly
and is called the posterior facial vein.

The anterior facial vein begins on the
snout. It gathers tributaries from the
masseteric and submental regions. It passes
back into the neck where it unites with the
posterior facial vein to form the external
jugular.

The internal jugular vein is quite small.
It collects a branch from the inferior
petrosal sinus in the carotid canal, leaves
the cranium through the jugular foramen
in company with the ners^es and the sta-
pedial artery, turns posteriorly with them,
and passes into the neck.

Foramina of the Rodent Skull

I have followed, as far as possible, the
temiinology used by Hill ( 1935 ) and ha\'e
attempted to find names commonh- used in
the literature for foramina he did not de-
scribe. My main points of departure from
Hill are in the temporal and pterygoid
regions. I have retained the term postgle-
noid foramen but haxe abandoned the
names subsquamosal, postsquamosal, supra-
squamosal, and squamosal in favor of the
general temi temporal foramina. In the
Rodentia the temporal foramina are quite
variable and cannot be categorized. The
new terms post-alar fissure, squamoso-
mastoid foramen, and foramen o\'ale ac-
cessorius are used for apertiu'es that are
different from anything in Hill's list. The
fossils demonstrate that Hill's distinction
betvveen alisphenoid and sphenopterygoid
canals is not universal in the order; only
one canal, the alisphenoid, is present in the
earliest rodents, and the sphenopterygoid

370 Bulletin Museum of Comparative Zoology, Vol. 146, No. 8

appears to be unique to geomyoids among
the groups examined.

The following topographic list of foram-
ina and their contents is based mainly on
the woodchuck (Marmota mormx) unless
otherwise stated. I have indicated also
those foramina present but not figured,
because they are hidden by another struc-
ture. Foramina lacking in Marmota are

described from the rodents in which they
occur. No rodent possesses all of the foram-
ina listed.

1. The unpaired interpremaxillary fora-
men does not occur in the woodchuck.
When present it is situated just behind the
incisors on the median premaxillary suture,
and it transmits a branch of the palatine

op eth suo uml

chu

-t- hiL dpi

1 cm

msc + bu -poTTi

.occ

hy cc etic

Figure 1. Marmota monax (MCZ B9911).

Cranial Foramina • Wahlert 371

artery. As Hill (1935:122) states, it is
relatively large in Aplodontia.

2. The incisive foramina flank the midline
of the diastema. The lateral margin of each
is intersected posteriorly by the premaxil-
lary-maxillaiy suture. Each transmits a
duct from the nasal passage, a branch of
the palatine artery, and a branch from the
palatine vein.

3. The major posterior palatine foramen
is usually .situated in the maxillary-palatine
suture. It transmits the descending palatine
arteries and nerves and a small vein. In
many rodents a posterior pair is present in
the palatine.

4. The posterior maxillary notch is situ-

ated between the end of the maxilla and
the pterygoid extension of the palatine; it
transmits the descending palatine vein. In
many forms tlie notch is enclosed as a
foramen.

5. The infraorbital foramen opens on the
side of the snout in the maxilla. It is the
anterior opening of the infraorl)ital canal
and transmits the infraorbital nerve, artery,
and vein. Protrogomorphous rodents lack
the canal.

6. The anterior alveolar foramen (not
figured) occurs in die floor of the infra-
orbital canal and transmits the anterior
superior alveolar nerve plus a .small artery
and vein.

Key to figures: Foramina and related structures (numbers correspond to those in text):

aa — anterior alveolar (6)

asc — alisphenoid canal (21)

bu — buccinator (24)

bup — posterior aperture, buccinator nerve canal

(24p)

cc — carotid canal (30)

oca — anterior end, carotid canal (30a)

chu — canal of Huguier (40)

dpi — dorsal palatine (16)

euc — Eustacian canal (29)

eth — ethmoid (12)

fac — facial canal (43)

fco — fenestra cochleae (41)

fo — foramen ovale (26)

foa — foramen ovale accessorius (27)

fro — foramen rotundum (20)

fv — fenestra vestibuli (42)

hy — hypoglossal (33)

ifo — infraorbital (5)

in — incisive (2)

iom — depression, origin of inferior oblique eye

muscle

ipm — interpremaxillary (1)

ito — interorbital (13)

ju — jugular (32)

ma — malar (7)

mif — middle lacerate (28)

mn — meningeal

ms — mastoid (38)

msc — masticatory (23)

msp — posterior aperture, masseteric nerve canal

(23p)

nl — nasolachrymal (8)

nu — nutritive (17)

of — orbital fissure (19)

op — optic (14)

paf — post-alar fissure (35)

pgl — postglenoid (34)

pom — posterior maxillary notch or foramen (4)

ppl —

posterior palatine (3)

spf —

sphenofrontal (15)

spl —

sphenopalatine (11)

spn —

sphenoidal fissure (18)

spt —

sphenopterygoid canal (22)

sqm —

squamoso-mastoid (39)

St —

stapedial (31)

stc —

stapedial artery canal (44)

sty —

stylomastoid (37)

suo —

supraorbital notch (10)

t —

temporal (36)

trc —

transverse canal (25)

uml —

unossified area in maxillary-lachrymal
(9)

suture

Bones:

ab —

auditory bulla

as —

alisphenoid

bo —

basioccipital

bs —

basisphenoid

f —

frontal

i —

jugal

1 —

lachrymal

m —

maxillary

mst —

mastoid

n —

nasal

occ —

occipital

OS —

orbitosphenoid

P —

parietal

pet —

petrosal

pi —

palatine

pm —

premaxillary

ps —

presphenoid

Pt —

pterygoid

sq —

squamosal

stippled

areas: cut through bone

solid line: seen in specimen

dashed

line: probable position

dotted li

ine: hypothetical position

372 Bulletin Museum of Comparative Zoology, Vol. 146, No. 8

7. The malar foramen (not figured) is
situated in the orbit where the orbital and
zygomatic portions of the maxilla meet
above the infraorbital foramen. It transmits
the malar artery from the infraorbital canal
to the tissue in front of the eye. It is rarely
present; presumably the artery is usually
external to the bone.

8. The nasolachrymal foramen is situated
in the lachrymal bone and is bounded
anteriorly by the zygomatic portion of the
maxilla. It transmits the lachrymal duct.

9. An unossified area between the
lachrymal bone and the orbital and zygo-
matic portions of the maxilla is not a fora-
men, but the area of origin of the inferior
oblique eye muscle. It has occasionally
been confused with the nasolachrvmal
foramen.

10. The supraorbital notch is an indenta-
tion in the supra-orbital flange of the
frontal bone. It permits passage of the
frontal branch of the ophthalmic nerve and
the superior ophthalmic artery to the top
of the head. No superior ophthalmic vein
was found accompanying them; it may have
been too small to see. The notch occurs in
the Sciuridae, only.

11. The sphenopalatine foramen is situ-
ated at the front end of the orbital process
of the palatine bone above the junction of
the second and third molars. The maxilla
forms the rest of its margin. It transmits
the sphenopalatine nerve, artery, and vein.
The bones participating in the margin of
the foramen differ among rodent groups.

12. The two ethmoid foramina are entirely
within the orbital lamina of the frontal
bone. The anterior one faces ventrally into
a shallow channel and transmits the nasocil-
iary branch of the ophthalmic nerve. The
posterior is larger and more dorsal; it
transmits the ethmoid artery and vein. A
single ethmoid foramen, which carries the
nerve, artery, and vein, is present in the
orbitosphenoid-frontal suture in most ro-
dents.

13. A single or multiple interorbital fora-
men pierces the orbitosphenoid in many
rodents; it is absent in Marmota but present
in some other sciurids. In geomyids it
transmits a sinusoid vein between the orbits
(Hill 1935:124).

14. The optic foramen is large and is en-
tirely within the orbitosphenoid. It trans-
mits the optic nerve.

15. The sphenofrontal foramen is situated
between the orbitosphenoid and alisphe-
noid. It is not quite separate from the
orbital fissure in some specimens. It trans-
mits the ophthalmic artery. The foramen
is absent in many groups.

16. The dorsal palatine foramen leads into
the palatine canal, which runs from the
orbit downward through the palatine bone
and out the posterior palatine foramen. It
transmits the descending palatine artery
and nerve and a small vein.

17. Many nutritive foramina (not figured)
occur in the orbital surface of the maxilla
above the cheek tooth roots. They transmit
minute branches of nerves and arteries, and
are present in all the specimens examined.

18. The sphenoidal fissure has as its outer
wall the alisphenoid bone. The nerves and
vessels transmitted by the orbital fissure
(no. 19), the foramen rotundum (no. 20),
and the alisphenoid canal (no. 21) exit
from it.

19. The orbital fissure (not figured) is
bounded anteriorly by the orbitosphenoid,
and posterolaterally by the alisphenoid. It
transmits the oculomotor, trochlear, and
abducens nerves, and the ophthalmic di-
vision and part of the maxillary division of
the trigeminal nerve. In most rodents the
fissure is united with the foramen rotun-
dum.

20. The foramen rotundum (not figured)
is completely concealed within the ali-
sphenoid canal. It pierces the inner wall
formed by the alisphenoid and transmits
the remainder of the maxillary nerve; the

Cranial Foramina • Wahlert 373

zygomatic branch may have a separate
foramen. The foramen rotundum and
orbital fissure are united in most rodents.

21. The alisphenoid canal passes length-
wi.se through the alisphenoid bone. It trans-
mits the internal maxillary artery and vein.

22. The sphenopterygoid canal is absent
in Marmota. I have found it only in
geomyoids; it leads from the pterygoid
fossa to the sphenoidal fissure. It transmits
the internal maxillary artery, and its walls
are the area of origin of the internal
pterygoid muscle.

23. The masticatory foramen is situated
in the alisphenoid and is often confluent
with the buccinator (no. 24). It transmits
the masseteric branch of the maxillary
nerve.

23p. The posterior aperture of the masse-
teric nerve canal (not figured) can be seen
just anterior to the foramen ovale in some
rodents.

24. The buccinator foramen is antero-
ventral to the masticatory or confluent with
it. It transmits the buccinator division of
the maxillary nerve.

24p. The posterior aperture of the bucci-
nator nerve canal (not figured) can be
seen in a specimen of Paramys; usually this
canal and the masseteric share a common
aperture, as in Marmota.

25. The single transverse canal (not
figured) runs between the alisphenoid
canals through the basisphenoid. It trans-
mits a vein connecting the two internal
maxillary veins.

26. The foramen ovale is situated postero-
laterally in the pterygoid region. It transmits
the mandibular branch of the trigeminal
nerve and a minute meningeal artery.

27. I define as new the foramen ovale
accessorius that is lateral to the foramen
ovale and transmits the mandibular branch
of the trigeminal nerve to the side of the

head. It is present in forms having a sub-
stantial lateral pterygoid flange that reaches
the auditory region.

28. The middle lacerate foramen is be-
tween the pterygoid region and the an-
terior end of the tympanic bulla or periotic
as the case may be. The foramen is absent
in Marmota; a minute aperture in the region
(not figured) transmits a meningeal vein.^

29. The Eustachian canal emerges dorsal
to the anteromedial portion of the tympanic
bulla. It transmits the Eustachian tube.

30. The carotid canal begins at or in front
of the anterior end of the jugular foramen
and runs anteriorly between the basioccip-
ital and the periotic and tympanic. In
many rodents having a canal it transmits
the internal carotid artery. In Marmota,
however, it transmits a vein, the inferior
petrosal sinus, which joins the internal
jugular vein.

30a. In some fossil rodents there is a
foramen leading into the cranium antero-
medial to the periotic. It seems to be the
anterior end of the carotid canal.

31. The stapedial foramen (not figured)
is dorsolateral to the jugular foramen and
shares a common aperture with it. It enters
the middle ear probably in the fused suture
between the tympanic and periotic, and
transmits the stapedial artery.

32. The lenticular jugular foramen is be-
tween the basioccipital and the postero-
medial part of the bulla. It transmits the
vagus, accessory, and glossopharyngeal
nerves, the stapedial artery, and the in-
ternal jugular vein.

^ The function of the foramen is uncertain; no
description exists of its contents in any of the
Recent forms I have examined. In muroids it
transmits the portion of the stapedial artery which
emerges from the anterior part of the middle ear
( C'.uthrie, 1963). In the dog the internal carotid
artery passes through the foramen into the cranium
(Gregory, 1910:430).

374 Bulletin Museum of Comparative Zoology, VoJ. 146, No. 8

33. The hypoglossal foramen in the basi-
occipital is anterior to the condyle and may
be subdivided into two or more parts. It
transmits the hypoglossal nerve.

34. The postglenoid foramen pierces the
squamosal bone ventral to the zygomatic
root and posteromedial to the glenoid fossa.
It is absent in many Marmota skulls. When
present, it transmits a large vein that drains
most of the cranial cavity.

35. The post-alar fissure is absent in
Marmota. I introduce this tenn for an
opening between the alisphenoid wing and
the tympanic bulla; it probably serves a
function similar to that of the postglenoid
foramen. In some forms it separates a part
of the squamosal from the tympanic.

36. The temporal foramen is absent in
Marmota. When present it is within the
squamosal bone or in the squamoso-parietal
suture, usually posterodorsal to the root of
the zygomatic arch. It serves the same
function as the postglenoid foramen and
can take over the entire function of that
opening. In some forms there are two or
more temporal foramina.

37. The stylomastoid foramen is between
the external auditory meatus and the mas-
toid process. It transmits the facial nerve,
and is constant in all rodents that possess
a bulla,

38. The mastoid foramen is on the occip-
ital surface between the occipital bone
and the medial portion of the mastoid
bone. It transmits a small vessel which,
according to Hill (1935:128), is a vein from
the neck muscles to the transverse sinus.

39. The squamoso-mastoid foramen is ab-
sent in Marmota. I introduce the term for
the foramen, which is present in many
rodents, on the occipital surface between
the squamosal and the mastoid. It trans-
mits a vein.

40. The canal of Huguier is a minute slit
in the anterior surface of the bulla. It

transmits the chorda tympani division of
the facial nerve.

The following, which are not, strictly
speaking, cranial foramina, have been
shown in figures of several early rodents.
They are useful as points of reference, and
the canals are, of course, associated with
soft parts intimately related to cranial
foramina.

41. The fenestra cochleae (rotundum) is
a round, membrane-covered aperture lead-
ing into the scala tympani of the cochlea.

42. The fenestra vestibuli (ovale) is an
oval, membrane-covered aperture leading
into the scala vestibuli of the cochlea. The
footplate of the stapes rests on this mem-
brane.

43. The facial canal is in the periotic
dorsolateral to the promontorium and is the
canal in which the facial nerve traverses
the middle ear.

44. The stapedial artery canal is also situ-
ated in the periotic dorsolateral to the
promontorium, and is die canal by which
the stapedial artery exits from the middle
ear. In many of the fossils it appears to be
united, in part, with the facial canal.

PARAMYIDAE

Par amy s

Specimens examined:

Paramys copei (Figs. 2 and 4): Lysite

Member, Wind River Formation: PU

16564 p. Lost Cabin Member, Wind

River Fonnation: AMNH 4755 (type)

npot, 4756 potc.
P. delicatior: Twin Buttes Member

equivalent, Bridger Fonnation: AMNH

55675 po.
P. delicatus (Fig. 3): Blacks Fork

Member, Bridger Formation: AMNH

12506 s, 13090 s; USNM 23556 s; YPM

13381 npo.
P. sp.: Willwood Formation: PU 17421

np.

Cranial Foramina • Wahlert 375

elh

ii> .V

hy ju fi' f ^ ^ bu spf

ifo

I cm

Figure 2. Paramys copei (composite of AMNH 4755 and 4756). See Fig. 1 for key to foramina.

Foramina

The ratio of length of the incisive foram-
ina to diastemal length ranges from .42
to .45. The lateral margins of the foramina
are intersected behind the middle by the
premaxillary-maxillary suture, which runs
posterolaterally away from them.

The posterior palatine foramina are
wholly within the palatine bones. The
larger anterior pair is close behind the
maxillary-palatine suture and medial to
the posterior halves of the first molars. The
smaller posterior pair is more laterally
situated than the anterior and is medial to
the posterior halves of the second molars.
The maxilla ends behind the cheek teeth
in a blunt point. There is a slight posterior
maxillary notch between it and the ptery-
goid extension of the palatine.

In front view the infraorbital foramen is
elliptical; the major axis is inclined so that
the top of the foramen is farther lateral
than the bottom. The axis in P. copei
measures 3.0 mm; in P. delicatus, 4.0 mm.
In lateral view the foramen is approxi-
mately vertical. The anterior alveolar fora-
men, which is in the curve formed by the
orbital wall and floor, is just posterior to
the infraorbital, and is directed antero-
medially. In front of the infraorbital there
is a small foramen, probably for the nasal
branches of the infraorbital arter)' and
nerve. This foramen is more pronounced
and more ventral in P. delicatus dian in
P. copei.

The lachrymal region is preserved onh-
in the type specimen of P. copei, AMNH
4755. The nasolachiymal foramen is dorsal

376 Bulletin Museum of Comparative Zoology, Vol. 146, No. 8

^PJ op eih spl aa

til

1 cm

cc tnlf fo pp7 in

Figure 3. Paramys delicatus (USNM 23556). See Fig. 1 for key to foramina.

to and not far above the infraorbital, and it
is below the lachrymal flange of the zy-
goma. A channel for the lachrymal duct
descends the face of the lachrymal bone to
the foramen. Sutures around the foramen
are not visible.

The sphenopalatine foramen is dorsal to
the junction of the second and third molars.
The maxilla and orbital process of the
palatine make up its borders; the frontal
may reach it dorsally, but this is not clear.
The orbitosphenoid is excluded from the
margin. Wood (1962:15, fig. 3A) figures
the sutures incorrectly and shows the fora-
men surrounded by the maxilla. The
ethmoid foramen is dorsal and posterior to
the third molar. It is within the frontal
bone and is overhung by a lip from it. In
the type specimen of P. copei, the frontal-
orbitosphenoid suture reaches it posteriorly;

in P. delicatus, AMNH 12506, the suture
does not. The optic foramen, which is
within the orbitosphenoid, is nearly 1.0 mm
in diameter. It is dorsal and considerably
posterior to the third molar. In P. copei
it is closer to the sphenoidal fissure than in
P. delicatus.

The dorsal palatine foramen, which is in
the maxillary-palatine suture, is immedi-
ately ventral to the sphenopalatine, and
both are within a single depression. Three
specimens, AMNH 4755, 12506, and 55675,
show this condition clearly. The same
occurs in Thisbemys corrugatus, AMNH
94008 (for which there is no locality data;
this is the only detail known of the foram-
ina of Thisbemys, so I include it here).
Minute nutritive foramina are present, as
in all rodents, in the floor of the orbit above
the roots of the cheek teeth.

Cranial Foramina • Wahlert 377

The sphenoidal fissure at its opening is
separated from the cranial cavity by a
wall of bone; it is situated well behind the
cheek teeth. In P. delicaius a slight ridge
sets off the dorsal portion as a distinct
channel. The alisphenoid canal joins the
sphenoidal fissure laterally.

The prominent sphenofrontal foramen is
in the orbitosphenoid-alisphenoid suture
near its junction with the frontal. In P.
ilclicatus, USNM 23556, a conspicuous
channel leads gradually downward and
forward from the foramen. Wood (1962:
15, fig. 3A) has labeled a puncture in the
bone as the sphenofrontal foramen; actu-
ally, it is indicated in his figure by a dark
area 2 mm behind and 3 mm above the
point he has labeled.

The masticatory and buccinator foramina
are separate, the distance between them
ranging from 1.0 to 3.0 mm. They face
anterodorsally and anteriorly, respectively,
and are a minimum of 3 mm from the
foramen ovale. A minute foramen occurs
between them in P. delicaius but not in P.
copei; this was possibly for a branch that
spht off the masseteric nerve before it
emerged from the masticatory foramen.
The buccinator foramen is farther anterior
with respect to the masticatory in P. deli-
catus. Wood (1962:15, fig. 3) has inter-
preted these foramina differently and, I
believe, incorrectly.

The pterygoid region of P. copei is
bounded medially by a flange and laterally
by a faint ridge that is sufficient to enclose
a foramen ovale accessorius. Medial to the
foramen there is an oval depression. Within
it are four foramina (Fig. 4). The posterior
one leads from the braincase and is clearly
the foramen ovale. The medial foramen
leads into two canals; one, anteriorly di-
rected, is the alisphenoid; the other, medi-
ally directed, is the transverse canal. The
anterior and lateral foramina lead to the
buccinator and masseteric nerve canals
respectively. I am in agreement with Black
(196Sa:291, fig. 18) as regards their inter-
pretation. In P. delicaius the lateral ptery-

cc fac she mlf fo asc*trc

/ C^l

Figure 4. Auditory and pterygoid regions of Paramys
copei (AMNH 4756). Labeled outline drawing (hamular
process hypothetical), and shaded drawing of the
same. See Fig. 1 for key to foramina.

goid ridge is somewhat weaker. A single
large foramen, the foramen ovale, is visible
in ventral view; in one specimen, AMNH
12506, a small foramen, probably the trans-
\'erse canal, opens in its anteromedial wall.
The t\anpanic bones are absent in all
specimens of Paramys, and the middle lac-
erate foramen is completely exposed. In P.
delicaius it seems to be a single, irregular
opening. In P. copei a stmt of bone trans-
forms the medial portion into a separate,
oval-shaped foramen. Wood (1962:42)
interprets this latter as the entire middle
lacerate foramen and states that the larger
lateral part is apparently the foramen ovale.
Since the specimen has a readily identifi-

378 Bulletin Museum of Comparative Zoology, Vol. 146, No. 8

able foramen ovale in the usual position,
Wood's designation cannot be correct.
Black (1968a: 291, fig. 18) has identified
the foramina as I do; the smaller medial
one he identifies as a part of the carotid
canal.

The carotid canal begins at the anterior
end of the jugular foramen, it was probably
open ventrally, and presumably transmitted
a branch of the internal carotid artery. The
canal may emerge from between the basi-
occipital and the petrous portion of the
periotic and enter the cranium through the
medial portion of the middle lacerate fora-
men, but this cannot be determined with
certainty. The jugular foramen, as in all
rodents, is lenticular in shape and is situ-
ated between the periotic and basioccip-
ital. The hypoglossal foramen is single in
P. copei and P. clelicatiis. The posterior
of two hypoglossal foramina shown by
Wood (1962:15, fig. 3E) is a break in the
bone.

The postglenoid foramen, which is be-
hind the glenoid fossa below the zygomatic
root, is in the squamosal bone. Its major
axis is smaller in P. copei than in P. cleli-
catus, about 1.5 and 2.5 to 3.5 mm respec-
tively. Temporal foramina are concentrated
in the squamoso-parietal suture. Antero-
dorsal to the postglenoid is a single fora-
men in the suture. In some specimens it is
accompanied by a second opening either
above in the parietal (AMNH 4756, right
side; AMNH 12506, left side) or below in
the squamosal (USNM 23556, left side).
A smaller foramen, also situated in the
suture, is about halfway between these
foramina and the back of the parietal.

A mastoid foramen is present in the
mastoid-occipital suture. It is well above
the level of the top of the foramen mag-
num. The foramina within the auditory
region are exposed because the tympanic
was not attached and has been lost. The
least distorted periotic is preserved in a
specimen of P. copei, AMNH 4756. The
major features of the auditoiy region are
shown in Figure 4. Wood (1962:43, fig.

14A) and Black (1968a:291, fig. 18) figure
the same portion of this specimen. The
most complete description of a paramyid
periotic is given for P. delicatus bv Wood
(1962:15, fig. 3B and C; page 18)!

A lateral shelf of the periotic begins at
the middle lacerate foramen and continues
posteriorly for two-thirds of the length of
the petrous portion. Behind it the mastoid
portion broadens, curves medially to the
posterior end of the jugular foramen, and
ascends the occipital surface; most of this
region is exposed outside the middle ear.
Lateral to the fossa for the stapedius muscle
there is a protuberance of the mastoid that
is not situated so far posteriorly as the
mastoid process in later rodents.

Medial to the lateral shelf a broad chan-
nel, which narrows posteriorly, runs from
the middle lacerate foramen to the fossa for
the stapedius muscle. In the absence of a
tympanic the stylomastoid foramen is
simply a groove lateral to the fossa on the
medial surface of the mastoid protuberance
and not a foramen as indicated in Wood's
figure. The anterior part of the channel is
presumably the area of origin of the tensor
tympani muscle. In the middle portion are
two posteriorly facing foramina, which are
just internal to the shelf. The anterior one
appears to be the foramen through which
the stapedial artery left the middle ear; the
posterior one, the foramen through which
the facial nerve entered the middle ear.

The medial portion of the auditory
region is occupied by the promontorium. A
faint channel, which marks the course of
the stapedial artery, i-uns from the fenestra
vestibuli to the anterior end of the jugular
foramen. This portion of the channel cor-
responds in position to the indentation for
the stapedial foramen in the bulla of
Reithroparamys (Fig. 5). The fenestra
cochleae is in the posterior surface of the
promontorium.

I do not see, as Wood did (1962:18),
evidence for the position of the auditory
bulla. He states that the ridge paralleling

Cranial Foramina • Wahlert 379

the median margin of the petrosal and over-
hanging the petrosal-basioceipital suture
(in AMNH 12506) seems to have served
for bracing tlie median wall of the bulla.
But the particular specimen he described is
distorted; the petrosal has been tipped and
the basioccipital crushed so that this ridge,
which originally abutted against the basi-
occipital, now stands away from it. The
ridge in its proper position could not have
braced the bulla.

Wood also states that the depression be-
tween the mastoid region and the lateral
shelf of the periotic "... seems to have h(>ld
the meatal tube of the bulla" (p. 18). It is
more likely, however, that the meatus was
lower down, as in Reithroporamys (Fig. 5)
and Sciuravus (Fig. 7) and that the de-
pression contained the dorsal portion of the
tympanic.

Leptotomus

Specimens examined:

Leptotomus hridgeremis: Twin Buttes
Member, Bridger Formation: AMNH
12507 t.

L. costilJoi: Huerfano Formation:
AMNH 55110 s, 55111 (type) s.

L. parvus: Twin Buttes Member, Brid-
ger Fonnation: AMNH 12519 (type)
p, 93030 p.

Foramina

Although two of these specimens are
complete skulls, they are so fractured and
crushed that very little information can be
gotten from them.

The two partial palates of L. parvus
show that the posterior palatine foramina
are wholly within the palatine bont^s, close
behind the maxillary-ioalatine suture. The
large pair is medial to the anterior part of
the second molars. These are connected,
each by a canal through the bone, to their
respective dorsal palatine foramen. The
latter is situated in the maxillary-palatine
suture immediately ventral to the spheno-

palatine foramen and above the anterior-
most part of the third molar.

In lateral view the infraorbital foramen
is vertical; its exact shape and disposition
cannot be determined.

The sphenoidal fissure at its opening is
separated from the cranial cavity by a wall
of bone. It is well behind the last molar.
Details of the region are visible in the
fragiiKMitary specimen of L. hrid<^crensis.
A slight ridge sets off the dorsal portion as
a distinct channel. The alisphenoid joins
the fissure laterally. A foramen in the
medial wall of the alisphenoid canal is
probably the entrance to the transverse
canal; it would be completely hidden in an
unbroken specimen. The exposed channel
through the bone to the sphenofrontal fora-
men is large and runs anteroventrally to a
position that was probably very close to
the top of the sphenoidal fissure.

The pterygoid region is partially pre-
served in AMNH 55110. The foramen ovale
is large, and the lateral pterygoid flange
bridges it ventrally to form a foramen ovale
accessorius.

The carotid canal appears to be like that
of Paramys; it was probably open ventrally
with the lateral lip of the basioccipital
shielding the artery. Whether it carried the
artery, or just the inferior petrosal sinus,
however, cannot be determined.

The postglenoid foramen is in the squa-
mosal under the root of the zygoma. Its
major axis measures about 1.8 mm.

The auditory region is poorly preserved,
but important details can be seen in the
type of L. costilloi. The channel for the
stapedial artery crosses the promontorium
laterally to the fenestra vestibuli as in
Paramys. At a point about a third of the
way along its course another channel about
half as wide diverges anterolaterally.
Within a short distance this channel sub-
divides. The diverging branch runs antero-
medially across the promontorium. This
bifurcating channel, I believe, marks the
course of the promontory artery.

380 Bulletin Museum of Comparative Zoology, Vol. 146, No. 8

1 cm

Figure 5. Reithroparamys delicatissimus (AMNH 12561)
See Fig. 1 for key to foramina.

Reithroparamys

Specimens examined:

Reithroparamys delicatissimus (Fig. 5):
Blacks Fork Member, Bridger Forma-
tion: AMNH 12561 (type) npc.

Foramina

The ratio of length of the incisive foram-
ina to diastemal length is .48. The lateral
margins of the foramina are intersected

very near the back by the premaxillary-
maxillary suture, which runs posterolater-
ally away from them.

A single pair of posterior palatine foram-
ina is present within the palatine bones.
It is situated far laterally, almost on the
maxillary-palatine suture, and is medial to
the anterior ends of the second molars. The
maxilla ends behind the cheek teeth in a
blunt point. There is a slight posterior
maxillary notch between it and the ptery-
goid extension of the palatine.

In front view, the infraorbital foramen
is elliptical. The major axis, which measures
3.5 mm, is inclined so that the top of the
foramen is farther lateral than the bottom.
In side view the foramen is approximately
vertical. The anterior alveolar foramen,
which is in the curve made by the orbital
floor and wall, is just behind the infra-
orbital foramen.

The wall of the orbit and the alisphenoid
bone are missing. Enough of the nasolach-
rymal canal is present to show that the
nasolachrymal foramen was dorsal and
close to the infraorbital.

The pterygoid region (Fig. 5) is mostly
missing. Several details can be made out,
however. The external pterygoid flange,
homologous to the lateral ridge in Paramys,
is substantial and seems to enclose a fora-
men ovale accessorius. Medial to the flange,
the back of the foramen ovale is preserved;
a channel, which is most likely the entrance
to the alisphenoid and transverse canals,
leads anteromedially from the foramen
ovale. The middle lacerate foramen is
completely covered by the auditory bulla.
Just anterior to the bulla and lateral to the
styloid process are two elongate foramina.
A channel extends posteriorly under the
bulla from the medial one. This suggests
that the foramen may have been an
aperture for a branch of the internal carotid
artery, possibly the promontorial. The
Eustachian canal passes over it. The lateral
foramen may have transmitted a meningeal
vessel.

The carotid canal appears to begin at the

Cranial Foramina • Wahlcrt 381

anterior c>nd of tlie jugular foramen and
does not have a distinct entrance. A shc>U
of the periotic is exposed anterohiteral to
the jugular foramen at the point where the
bulla is indented. The stapedial foramen is
in this indentation and between the tym-
panic and periotic. The hypoglossal fora-
men is double on both sides. The larger
foramen opens ventrally and faces antero-
latc^ralh'; its rim continues out toward the
iu<j:ular foramen. The second foramen is
under the rim of the larger and opens and
faces anteromedially.

The postglenoid foramen is within the
squamosal bone. Its major axis is 2.1 mm
long. On the left side of the skull are three
foramina in front of the periotic, which is
sandwiched between the bulla and squa-
mosal. These foramina may be homologous
with the post-alar fissure of some later
sciuromorphous forms. A large temporal
foramen is present in the sciuamoso-parietal
suture above the postglenoid. There seems
to be a much smaller one, behind it, also in
the suture.

The stylomastoid foramen is bounded
by the bulla and mastoid element. There
is a short protuberance of the mastoid
lateral to it. The mastoid foramen is above
the level of the top of the foramen magnum
and is in the mastoid-occipital suture,

Ischyrotomus

Specimens examined:

Ischyrotomus oweni: Blacks Fork

Member, Bridger Formation: USNM

17161 s; 17160 (type) s (specimen

not available to me; information

taken from Wood, 1962).
I. horribilis: Blacks Fork Member,

Bridger Formation: USNM 17159

(type) s.
7. petersoni: Myton Member, Uinta

Formation: AMNH 201S (type) s.

Foramina

The ratio of length of the incisive foram-
ina to diastemal length rang(>s from .18 to
.21. The lateral margins of the foramina

are intersected near the back by the pre-
maxillary-maxillary suture, which runs
posterolaterally away from them.

The pair of larger posterior palatine
foramina is in die maxillary-palatine suture
and is medial to the middle region of the
first molars. The smaller posterior pair,
entirely within the palatine, is in line with
die larger pair and medial to the ant(>rior
halves of the second molars. In I. oweni,
USNM 17161, there are two minute foram-
ina situated more laterally in the palatin(\
The maxilla ends behind the cheek teeth
in a distinct point that is best seen in /.
horribilis. There is a post(>rior maxillary
notch between it and the pterygoid exten-
sion of the palatine.

In front view, the infraorbital foramen
is elliptical, and the major axis is inclined
so that the top of the foramen is farther
lateral than the bottom. The axis in 7.
Jiorribilis is 3.3 mm long; in 7. oweni, 4.4
mm; in 7. petersoni, 4.5 mm. In side view
the foramen is approximately vertical. The
lachrymal region in these specimens is
either missing or damaged. Wood (1962:
189) reports that the nasolachrymal fora-
men is between the lachrymal and maxil-
lary bones in the medial wall of the orbit.

Wood (1962:189-190) states, "The sphe-
nopalatine foramen sometimes lies on the
frontal -maxillary suture and sometimes in
the maxilla as in Faramijs. It is a little
farther to the rear, just behind M'^ instead
of just in front of it." His interpretation of
its position in Paramys is incorrect, as noted
above, and his placement of it in Ischyro-
tomus also seems erroneous. Wood ( 1962:
207, fig. 71), in his figure of USNM 17160,
shows the foramen within the maxilla and
dorsal to the back of the second molar. Its
position cannot be determined in the other
specimens. The ethmoid foramen is above
and between the sphenopalatine and optic
foramina. A slight lip of bone overhangs it.
Sutures in this region are indeterminate.
The optic foramen, which is about 1.0 mm
in diameter, is dorsal and considerably

382 BiiUetin Museum of Comparative Zoology, Vol. 146, No. 8

asc

If fo

/ cm

Figure 6. Auditory and pterygoid regions of Ischyroto-
mus oweni (USNM 17161). See Fig. 1 for key to
foramina.

posterior to the third molar, and it is near
the sphenoidal fissure.

The dorsal palatine foramen was not
seen; the region in which it would occur is
fractured in every specimen. The sphenoi-
dal fissure at its opening is separated from
the cranial cavity by a wall of bone. A low
ridge sets off its dorsal portion as a distinct
channel. Tlie alisphenoid canal unites with
the fissure. The sphenofrontal foramen is
dorsal to the sphenoidal fissure and postero-
dorsal to the optic foramen; a conspicuous
channel leads downward and forward from
it. Sutures in tliis area are indistinct in all
specimens. Available specimens are too
ci-ushed in the alisphenoid region to reveal
whether the masticatory and buccinator
foramina were separate or united.

The pterygoid region of Ischijrotomus
(Fig. 6) is bounded medially by a flange
and laterally by a ridge; it is not developed
into a fossa and is occupied almost entirely
by a depression in which there are two
openings. Posterolaterally the foramen
ovale opens from the cranium; the lateral
ridge is interrupted alongside it, and the
beginning of a foramen ovale accessorius
is suggested by the hook-like termination of
the ridge. Anterior and medial to the
foramen ovale is the second opening; it is

deep within the angle formed where the
lateral ridge and internal pterygoid flange
meet. The alisphenoid canal runs anteriorly
from it, the transverse canal medially. The
dorsal portion of the alisphenoid canal is
slightly damaged, but one small foramen,
which probably transmitted the buccinator
nerve, is clearly visible in its wall. The
middle lacerate foramen is distorted by
crushing in all specimens.

The carotid canal begins at the anterior
end of the jugular foramen. The hypoglos-
sal foramen is either single or double; when
double, the apertures open into a single
pit.

The postglenoid foramen is wholly within
the squamosal bone. The major axis
measures 1.5 mm in 7. horribiUs and 2.7
mm in 7. oweni. Temporal foramina are in
the vicinity of the squamoso-parietal suture.
The largest foramen is dorsal to the post-
glenoid and above the zygomatic root; in
some specimens it is entirely within the
squamosal; in others on the suture. There
is a small foramen anterior to it and another
posterior in the parietal. The occipital sur-
face is damaged, and sutures near the mas-
toid foramen cannot be distinguished.

The major features of the auditory region
(Fig. 6) are essentially as in Paramys, but
there are differences in detail. The mastoid
portion of the periotic has a descending
process lateral and posterior to the fossa
for the stapedius muscle. This mastoid
process is essentially modern in aspect. The
foramina in the petrosal for the stapedial
artery and for the facial nerve bear the
same relationship to each other as in
Paramys. In venti^al view, however, they
are hidden under a single shelf which runs
anterolaterally from a point on a level with
the front of the fenestra vestibuli to a point
overhung by the lateral shelf of the peri-
otic. There is a distinct channel for the
stapedial artery which crosses the promon-
torium. It is broadest at the fenestra vesti-
buli and narrows somewhat near the
anterior end of the jugular foramen. This
portion corresponds in position to the in-

Cranial Foramina • Walilcrt 383

dentation for tlic stapedial foramen in ihc seems to be separated from the cranial

bulla of Reithropommys (Fig. 5) and cavity by a wall of bone. The sphenofrontal

Sciuravus (Fig. 7). foramen is visible on the right side of this

speciracui just dorsal to the sphenoidal fis-

Pseudotomus sure. I do not see a channel leading from it.

The masticatory and buccinator foramina

Specimens examined: ^j.e separatt> and over 4.0 mm anterior to

Pseudotomus hiam: PBlacks Fork Mem- ^}^^, foramen ovale. A broad channel leads

ber, Bridger Formation: AMNII 5025 dorsally from the masticatoiy foramen. The

(type) nptc. buccinator foramen opens anteriorly; it is

Foramina directly under the middle of the mastica-
tory and less than 1.0 mm away from it.

A portion of the external margin of the 7^^ pterygoid portion of Pseiidotomm

right incisive foramen is present. Its curva- j^ similar to that of Ischyrotomus. The

ture suggests that the foramen was rela- foramen ovale has only a hint of a lateral

tively short, as in Ischyrotomus. pterygoid ridge alongside it; there is no

At the back of the maxilla, near the suggestion of a foramen ovale accessorius.

middle of the palate, a slight channel leads j]^q anterior portion of the pterygoid de-

posterodorsally into what was probably the pression leads into the alisphenoid and

larger of tlie posterior palatine foramina. It transverse canals. A posterior projection

was evidently medial to the middle of the fj-Q^i the anterior margin of the middle

first molar and in the maxillaiy-palatine lacerate foramen indicates that the foramen

suture. was partially differentiated into medial and

The infraorbital foramen is broad and lateral portions. The posterior margin is

elliptical. The major axis, which measures j^ot preserved.

3.5 mm, is inclined so that in front view the xhe bullae are missing. The left periotic
top of the foramen is farther lateral than ^nd the anterior end of the right are gone,
the bottom, and in side view the top is ^nd the portion of the basioccipital that
slightly farther anterior than the bottom, nomially abuts the periotic is exposed. The
The anterior alveolar foramen, which is in ventral surface of the basioccipital extends
the curve made by the orbital floor and laterally as a flange that would have over-
wall, is just behind the infraorbital. The lapped the anterior extremity of the peri-
nasolachrymal foramen is dorsal to and not otic. Dorsal to the flange on the lateral
far above the infraorbital and is below the surface of the basioccipital, tliere is a
posterior protuberance of the lachrymal channel that turns up toward the cranium
bone. A wide channel descends the surface j^^t behind the middle lacerate foramen;
of the lachrymal and bends anteriorly into possibly this is the carotid canal, or the
the foramen. The maxilla appears to form course of the inferior petrosal sinus. The
the ventral margin of the foramen. area of the basioccipital contained within

Both orbits are considerably damaged the curve of the channel is sculptured and

and the fragments of bone displaced; was most likely the place where the periotic

sphenopalatine and dorsal palatine foram- attached. The posterior part of the basi-

ina cannot be seen. The anterior part of occipital is missing.

the ethmoid foramen is preserved on the The postglenoid foramen is within the

right side; a lip from the frontal overhangs squamosal bone. Temporal foramina are

it. The optic foramen, about 1.0 mm in present in or near the s(j[uamoso-parietal

diameter, seems to have been dorsal and suture, but their number and exact posi-

considerably posterior to the last molar. tions are indeterminate. The mastoid fora-

The sphenoidal fissure at its opening men is above the level of the top of the

384 Bulletin Museinn of Comparative Zoology, Vol. 146, No. S

foramen magnum in the mastoid-occipital
suture.

Manitsha

Specimens examined :

Manitsha tanka: Chadron Formation^:
AMNH 39081 (type) np.

Foramina

The ratio of length of the incisive foram-
ina to diastemal length is .23. The lateral
margins of the foramina are not intersected
by the premaxillary-maxillary suture, and
it crosses the diastema behind them. The
maxilla ends behind the cheek teeth in a
point appressed to the palatine; there is
no posterior maxillary notch.

The infraorbital foramen seems small
relative to the skull size. In side view it is
approximately vertical. A single hypoglos-
sal foramen is preserved on the left side.

Discussion of the Paramyidae

The paramyid rodents fomi a unified
group with respect to cranial foramina.
There are some differences between genera
and species, among which changes in the
pterygoid region and various patterns of
arterial channels in the auditory region are
the most striking.

The ratio of length of the incisive foram-
ina to diastemal length is high in Paramys
and Reithroparamys, .42 to .48, and low in
Ischyrotomiis, Pseudotomus, and Manitsha,
.18 to .23. The lateral margins of the
foramina are intersected behind the mid-
point, but not at the very back, by the
premaxillary-maxillary suture. In Manitsha
the suture crosses the diastema behind the
foramina and does not run into their mar-
gins.

^ The American Museum catalogue incorrectly
gives the horizon and locality for this specimen as
Lower Brule, North Point of Slim Buttes, and this
misinformation has been perpetuated in the litera-
ture. The correct data, suppHed by M. F. Skinner
(personal communication), are as follows:
Chadron Formation, west side of Reva Pass, Hard-
ing County, South Dakota.

The posterior palatine foramina are
wholly within the palatine bone in
Paranujs, Leptotomus, and Reithroparamys.
In Ischyrotomiis they are in the maxillary-
palatine suture. A posterior maxillary notch
is present in all except Manitsha.

In Paramys, Thisbemys, and Leptotomus
the dorsal palatine foramen is associated
with, but separate from, the sphenopalatine
foramen; its position is uncertain in the
other genera owing to crushing in the
orbital region. The sphenofrontal foramen
is present in Paramys, Leptotomus, Ischyro-
tomtis, and Pseudotomus. Other skulls were
too damaged for it to be found. The
presence of this foramen indicates that the
ophthalmic artery was a branch of either
the stapedial or the internal carotid artery.

Masticatory and buccinator foramina are
separate from each other and not especially
close to the foramen ovale. I do not expect
that Ischyrotomiis will prove to be an ex-
ception when adequate material is found.

Foramina in the vicinity of the foramen
ovale differ among genera and even among
species. The pattern found in Paramys
copei could be that from which later ar-
rangements were derived. In this species
the foramen ovale, masseteric nerve canal,
buccinator nerve canal, and alisphenoid
and transverse canals open into a single
depression. In Ischyrotomus and Pseudo-
tomus entrances to the masseteric and buc-
cinator nerve canals are hidden, and the
depression is differentiated into two parts,
one for the foramen ovale and another for
the alisphenoid and transverse canals. The
transverse canal was hidden in Leptotomus
and a foramen ovale accessorius may have
been present. The only available specimen
of Reithroparamys appears to have been
similar to Ischyrotomus. In Paramys deli-
catus the alisphenoid canal is hidden, and
the foramen ovale is the only conspicuous
opening in the region; the transverse canal
is small but visible in one specimen and
hidden in the other.

The middle lacerate foramen, when
present and undistorted, appears to be

Cranial Foramina • WaJdert 385

^ msc ba spf op ^i^

7 cm

T^-sc ^u spn

OiSC

Figure 7. Sciuravus nitidus (reconstructed from USNM 17683, 18100, and 22477). See Fig. 1 for key to foramina.

divided into hvo parts, the small medial
one possibly for passage of the internal
carotid artery. In Reithroparamijs, only
this small aperture is visible; the t^'mpanic
bulla covers the middle lacerate foramen
if it is present.

In the auditory region of Paramys and
Ischyrotoinus there is a channel for the
stapedial artery which crosses the promon-
torium, and a stapedial foramen is present
in Reithroparamys. In Paramys and
Ischyrotoinus there is no channel indicating
the presence of the promontorial arter\',
whereas in Leptotomus this channel is
clearly marked. The hypoglossal foramen
is single in Paramys, single or double in
Ischyrotomus, and double in Reithro-

paramys. A rudimentary post-alar fissure
is present in Reithroparamys and absent in
the other genera.

SCIURAVIDAE

Specimens examined :

Sciuravus nitidus (Fig. 7): Blacks Fork
Member, Bridger Formation: AM Nil
12531 n, 12551 nptc, 13101 npoc;
USNM 17683 c, 17697 np, 17700 np,
18023 np, 18100 s, 22477 s; CM 9683
np; YPM 13458 p.

Foramina

Accurate measurement of the incisive
foramina is possible in three specimens.
The ratios of their lengths to diastcmal

386 Bulletin Museum of Comparative Zoology, Vol. 146, No. 8

lengths are .41, .45, and .47. The lateral
margins of the foramina are intersected
near the back by the premaxillary-maxillary
suture, which runs posterolaterally away
from them.

The posterior palatine foramina are
within the palatine bones. The large an-
terior pair is close behind the maxillary-
palatine suture and medial to the junction
of the first and second molars in some
specimens, medial to the second molars in
others. The smaller posterior pair is some-
what more lateral in position than the
anterior and is medial to the junction of the
second and third molars. The maxilla ends
behind the cheek teeth in a blunt point.
There is a slight posterior maxillary notch
between it and the pterygoid extension of
the palatine.

In front view, the infraorbital foramen is
elliptical. The major axis is inclined so that
the top of the foramen is farther lateral
than the bottom. The axis ranges in five
specimens from 1.7 to 2.6 mm. In side view
the foramen is approximately vertical. The
anterior alveolar foramen is just behind the
infraorbital foramen in the curve made by
the orbital wall and floor, and plunges
anteromedially.

The structvu-e of the lachrymal region,
although not entirely preserved in any one
specimen, can be determined for the most
part. The nasolachrymal foramen is well
above and slightly posterior to the infra-
orbital. It is directly below the posterior
protuberance of the lachrymal bone and
may be surrounded by that bone, but
sutures are not clear. A short channel
leads into the nasolachrymal foramen and
continues anteroventrally as a canal. The
canal, exposed in AMNH 12531, passes in-
ternal to the infraorbital foramen and turns
medially a short distance in front of it.

The sphenopalatine foramen is dorsal to
the anterior half of the second molar. It
seems to be bounded posteriorly by a long
orbital process of the palatine, and on the
other sides by the maxilla. The frontal may
be barely excluded from its margin; the

orbitosphenoid is completely excluded. The
minute ethmoid foramen is in the frontal
above and about halfway between the
sphenopalatine and optic foramina. The
orbitosphenoid does not seem to reach it. It
is overhung by a slight lip from the frontal.
The optic foramen is not preserved clearly
in any specimen, but seems, at least in
USNM 18100, to have been within the
orbitosphenoid. It is posterodorsal to the
third molar and near the sphenoidal fissure.

The small dorsal palatine foramen is in
the floor of the orbit posterior and slightly
lateral to the sphenopalatine. The suture
between the palatine and maxilla dips into
it. The entire course of the canal descend-
ing from it can be traced in two specimens,
USNM 18100 and YPM 13458 (better seen
in the latter). For a short distance the
canal runs between maxilla and palatine;
then it emerges and continues anteroven-
trally on the internal surface of the palatine
as a channel open into the choanae; finally
it turns anteriorly through the posterior
palatine foramen.

The available specimens are too damaged
to show whether the sphenoidal fissure at
its opening is separated from the cranium.
It is situated well behind the cheek teeth.
The sphenofrontal foramen, seen in one
specimen, USNM 18100, is in the orbito-
sphenoid-alisphenoid suture just below the
point at which the suture meets the frontal.
A short channel leads anteroventrally from
it.

The masticatory and buccinator foramina
are clearly preserved in only one specimen,
USNM 18100. They are close together near
the foramen ovale; channels from them lead
upward and forward respectively.

The pterygoid region is a relatively flat
triangular surface bounded medially by a
flange and laterally by a ridge; it is not
developed into a fossa. There is an elon-
gated depression medial and parallel to the
posterior half of the lateral ridge. The
foramen ovale opens from the cranial cavity
into the posterior part of this depression.
The alisphenoid canal runs forward from

Cranial Foramina • Wahleit 387

the depression, and a lateral canal branches
off from it to lead toward the masticatory
and buccinator foramina. Medial to the
alisphcnoid canal is the opening into the
transverse canal. Details of this region are
clear in two specimens, USNM 18100 and
22477. Dawson (1961:10, plate II) figured
a distorted specimen and did not restore
structures to their original positions; her
sphenopalatine canal is my alisphcnoid
canal.

In specimens lacking the auditory bulla,
the middle lacerate foramen is exposed. Its
exact shape cannot be determined from the
material a\'ailable, but the medial portion
is partiall)' separated from tlie large lateral
part. The medial portion is in the basi-
sphenoid-basioccipital suture and pierces
the side of the cranial floor. A channel
leads posteriorly from the aperture across
the anterior shelf of the periotic onto the
promontorium. At the anterior end of the
jugular foramen there is no space between
the periotic and basioccipital for a carotid
canal, and there is no separate entrance to
a carotid canal elsewhere. The hypoglossal
foramen is single.

The postglenoid foramen is within the
squamosal bone. Temporal foramina can
be seen clearly in USNM 18100. The largest
is in the squamoso-parietal suture about
halfway between the back of the zygomatic
root and the posterior margin of the skull,
A second small foramen is immediately be-
hind it in the suture, and there is a small
foramen in the parietal anterodorsal to
these.

In most specimens the auditory bullae
are missing; the periotic (Fig. 8) is clearly
displayed in three specimens, USNM 18100
and 22477, and AMNH 13101. The major
features are similar to those described for
Paramys, but there are differences in
relative proportions. The fenestra v(>stibuli,
the fenestra cochleae, and the fossa for the
stapedius muscle, indicated by a depres-
sion, are as in Paramys. A channel leads
anteriorly from the stylomastoid foramen
to a single foramen that is slightly anterior

y,^,

sly

f" {ac^slc /'

.5 cm

Figure 8. Auditory region of Sciuravus nitidus (re-
stored from USNM 17683). See Fig. 1 for key to
foramina.

to the fenestra vestibuli; the facial nerve
and the stapedial artery evidently shared
this one opening. Channels showing the
courses of blood vessels are present on the
surface of the promontorium. The main
channel for the internal cartoid artciy, pos-
sibly the promontory branch, begins just
anterior to the jugular foramen; midway
across the promontorium it turns antero-
dorsally and runs to the medial portion of
the middle lacerate foramen. A channel
for the stapedial artery curves postero-
dorsally from the point where the internal
carotid turns anteriorly, and it leads to the
fenestra \'estibuli.

The bulla is preserved on one specimen,

388 Bulletin Museum of Comparative Zoology, Vol. 146, No. 8

USNM 22477. It has been moved out of
place, and no markings on the periotic
indicate precisely where it was situated. I
did not notice a stapedial foramen, but one
must have been present in the margin of
the bulla, as indicated in Figure 7, to per-
mit passage of the internal carotid artery.
It is not possible to see in the specimen
whether the bulla completely covered the
middle lacerate foramen; measurements
suggest that it did not, and I have so shown
it in the figure.

Discussion of the Sciuravidae

The ratio of length of the incisive foram-
ina to diastemal length is high, as in
Paramys and Reithropammys, and is much
higher than tliose of Ischyrotomus and
Manitsha. The posterior palatine foramina
are within the palatine bone, again as in
Paramys and Reithroparamys. A posterior
maxillary notch is present, as in most
paramyids.

The orbital process of the palatine
reaches the back of the sphenopalatine
foramen, whereas the orbitosphenoid does
not. This arrangement occurs in Paramys.
The sphenopalatine and optic foramina are
farther forward relative to the cheek teeth
than those of paramyids.

In Sciuravus the dorsal palatine foramen
clearly is separated from the sphenopala-
tine, a condition perhaps foreshadowed in
paramyids. The sphenofrontal foramen is
in the orbitosphenoid-alisphenoid suture, as
in paramyids. Masticatory and buccinator
foramina are present and considerably
closer to the foramen ovale than they are in
paramyids. The arrangement of foramina
in the pterygoid region is similar only to
that of Ischyrotomus and Pseudotomiis.

Separation of the medial part of the
middle lacerate foramen to receive a
branch of the internal carotid artery occurs
in paramyids. Sciuravus, however, does not
have a carotid canal between the periotic
and basioccipital. Instead, the internal
carotid artery entered the middle ear and
crossed the promontorium in a shallow

channel before entering the cranial cavity.
The carotid circulation in Leptotomus may
be the same, though the carotid canal is
also present. There is only a single aperture
in the petrosal for the stapedial artery and
the facial nerve to exit from the middle ear,
whereas in paramyids a pair of openings is
visible. The postglenoid and temporal
foramina are of about equal size, as in
paramyids.

ISCHYROMYIDAE

Specimens examined:

I have seen and measured so many speci-
mens (approximately 65) that a complete
list would be excessive. The specimens re-
corded below are only those that are nearly
complete or are cited in the text.

Ischyromys douglassi: Chadron Forma-
tion equivalent: CM 1123 pot; 10966
potc.
I. typus (Fig. 9): Orella Member, Brule
Formation: AMNH 694 s; FMNH, P
12747 poc; MCZ 18979 potc; YPM
12521 s; PU 11383 s. Brule Formation:
USNM 15929 npc; 15933 s.
7. sp: Chadron Formation equivalent:
CM 24129 c. Orella Member, Brule
Formation: AMNH 38865 s; CM 9463
npoc. Brule Formation: USNM 16953
s; 175352 tc; 175354 npo; CM 9755 pt.
Titanotheriomys veterior: Chadron For-
mation equivalent: CM 9809 p; 10660
npot. POrella Member, Brule Forma-
tion: MCZ 17202 s.
T. tcyomingensis: Chadron Formation
equivalent: AMNH 14579 (type) npt.
T. sp.: Chadron Formation equivalent:
CM 8924 npt.

Foramina

The ratio of length of the incisive foram-
ina to diastemal length, measured in seven-
teen ischyromyids, ranges from .21 to .30
with a cluster of nine around .24 and a
cluster of eight around .28. The three
specimens of Titanotheriomys in which this
region is preserved fall at the low end of
the range. The lateral margins of the

Cranial Foramina • Wahlcrt 389

spf do

eth

sly / fo*asc

■paf foa

I cm

cc rnn Jo Ire pom ppl in

Figure 9. Ischyromys typus (AMNH 694). See Fig. 1 for l<ey to foramina.

foramina are intersected at the verv back
by the premaxillary-maxillary suture, which
runs posteriorly away from them.

The pair of larger posterior palatine
foramina lies in the maxillary-palatine
suture medial to an area extending from
the middle of the last premolar to the
middle of the first molar. The suture
crosses the palate between the foramina ( in
one specimen, CM 9809, behind them, and
a process from the palatine runs forward
into the back of each foramen). There is
usually a second pair of small foramina
within the palatine medial to the first or
second molars, and in some specimens a
third pair of minute openings is present
behind these. The maxilla ends behind the
cheek teeth in a long point that is appressed

to the pterygoid extension of the palatine;
there is a lenticular aperture between them.

In front view, the infraorbital foramen
forms a pinched ellipse with the acute end
ventral. The major axis has an average
length of 3.6 mm and ranges from 3.0 to
4.6 mm; it is inclined so that in front view
the top of the foramen is farther lateral
than the bottom, and in side view the top
is farther anterior than the bottom. The
anterior alveolar foramen is a short distance
behind the infraorbital foramen in the
curve made by the orbital floor and wall,
and is directed anteromcdially.

The nasolachrymal foramen is dorsal to
the infraorbital foramen and below the
posterior protuberance of the lachrymal
bone. The maxilla appears to participate

390 Bulletin Museum of Comparative Zoology, Vol. 146, No. 8

in its anterior margin. The internal course
of the nasolachrymal canal can be seen
clearly in one specimen of Ischijromijs, CM
1123. A short channel descends the face of
the lachrymal bone, its slight anterior
inclination continuing inside the bone. It
turns sharply forward near the base of the
snout and proceeds anteriorly a short dis-
tance before it bends medially under the
arch of the incisor. In some specimens, be-
tween the lachrymal and the orbital portion
of the maxilla, there seems to be a small
unossified area that is roughly horizontal
and in line with the middle of the infra-
orbital foramen. This may mark the site of
origin of the inferior oblique eye muscle,
as in Marmota.

The sphenopalatine foramen is dorsal to
the first molar. The fused maxilla and pala-
tine surround it almost entirely, and only
a very small arm of the orbitosphenoid
reaches it posteriorly. The ethmoid foramen
is posterodorsal to the sphenopalatine. The
frontal surrounds it on three sides and a
lip from the frontal overhangs it; the orbito-
sphenoid meets it posteriorly. In one speci-
men, USNM 15933, the ethmoid foramen
is double. The anterior opening has the
usual slit-like appearance, and a slender
process from the orbitosphenoid touches it.
The posterior opening is nearly round; a
channel in the orbitosphenoid leads to it
from behind. In this one specimen the
nerve and blood vessels probably entered
the bone through separate foramina, as in
Marmota. The optic foramen, usually 1.0
mm in diameter but slightly larger in a
few specimens, is within the orbitosphenoid.
In most specimens it is above the posterior
part of the second molar or the anterior
part of the third molar; in one, CM 10966,
it is above the posterior portion of the third
molar.

Within the triangle formed by the optic,
ethmoid, and sphenopalatine foramina is a
depression in front of and closest to the
optic foramen. Its deepest and most clearly
defined portion is posterior. A counterpart
of this depression occurs in Marmota in

which it is the place of origin of the rectus
muscles of the eye. Usually there is a small
foramen in the deep part of the depression.
In one fragmentary specimen, USNM
175354, the pit is deepened to a pocket, and
the foramen within was large enough to
clean out. The foramina on the two sides
of the skull proved to communicate across
the midline. There is no exit from this
passage either dorsally or posteriorly
through the orbitosphenoid. In another
specimen, USNM 16953, there is only a
minute foramen within the depression, and
a larger and apparently similar foramen
occurs just anterior and medial to the optic
foramen. The facts that the passage be-
tween the orbits has no other exit and that
the position of the foramen is variable sug-
gest that this was part of the venous system
and that the aperture is a true interorbital
foramen.

The dorsal palatine foramen is hidden.
One broken specimen. CM 9809, shows it
clearly just inside and ventral to the
sphenopalatine foramen. This means that
the descending palatine artery and nerve
entered the sphenopalatine foramen before
passing through the dorsal palatine fora-
men. The sphenoidal fissure at its opening
is separated from the cranial cavity by a
wall of bone. The alisphenoid canal enters
the sphenoidal fissure laterally. A very
slight ridge sets off the dorsal part of the
fissure as a separate channel. A small
sphenofrontal foramen is present in all un-
distorted specimens of Ischijromys; skulls
identified as TitanotJwriomijs are not well
enough preserved for me to determine
whether tlie foramen occurs in them also.
The sphenofrontal foramen is a short dis-
tance posterodorsal to the optic foramen.
In some skulls it is entirely within the
orbitosphenoid and close to the orbito-
sphenoid-alisphenoid suture, whereas in
others it is in the suture. In all a channel
leads gradually downward and forward
from it.

The masticatory and buccinator foramina
are separate from each other and are at a

Cranial Foramina • Walilcrt 391

substantial distance from the foramen ovale
accessorius in specimens of /. (Ioug,lassi
from McCarty's NIountain; in CM 1123 the
distance is 2.0 mm (see also Black, 1968a:
291, fig. 20 for CM 1122). In one specimen
of Ischyroniys from the middle Oligocene,
PU 11383, only a slight bar of bone sepa-
rates these foramina from the foramen ovale
accessorius. In the remaining Orellan speci-
mens of Ischyroniys and TitanotJieriomys
the masticatory and buccinator foramina
are united with the foramen ovale acces-
sorius from which a broad channel runs
anterodorsally across the alisphenoid, re-
vealing the course of the masseteric nerve.
The foramen ovale accessorius is bounded
ventrally by a substantial lateral pterygoid
flange, homologous to the lateral ridge in
paramyids. The pterygoid fossa is deep in
ischyromyids. The foramen ovale is dorso-
medial to the foramen ovale accessorius
and, in ventral view, is mostly obscured by
the lateral pterygoid flange that bridges it.
In some specimens there is a small foramen,
which probably transmitted a meningeal
vessel, medial to the foramen ovale. The
entrance to the alisphenoid canal is at the
anterior end of the foramen ovale and
cannot be seen in ventral view. The middle
lacerate foramen, if present, is hidden by
the large auditory bulla. An aperture in
the middle part of the pterygoid fossa leads
to the transverse canal. In most specimens
it is directed anteriorly; in one, CM 10966,
it heads medially. The transverse canal has
other entrances, too, inside the cranium.
The largest of these is just behind the basi-
sphenoid-presphenoid suture in the side of
the raised cranial floor. A few foramina
open into the canal from the surface of the
raised portion. These details were seen in
one specimen, CM 1123, in which the bone
is translucent; upon shining a light through
the bone, the matrix-filled canals became
apparent.

The posterior opening of the carotid
canal is clearly delimited by the basioccip-
ital and the bulla. It is at the anterior
end of the same slot that contains the

jugular foramen. The bulla has a faint
channel running dorsally into the carotid
canal. The course of the canal can be seen
in one specimen, CM 1123, in which it
runs between the basioccipital and the
bulla and arches up and over the medially
swollen anterior portion of the bulla. It
appears to enter the cranial cavity at the
level where the medial side of the bulla
curves outward. The jugular foramen is
directed dorsally at the back of the slot
between the basioccipital and the bulla.
Black (1968:290, fig. 19) interprets this
as the stapedial foramen. However, there
must be a jugular foramen to transmit the
vagus, accessory, and glossopharyngeal
nerves, and the internal jugular vein; this
opening is the only possibility. There is no
stapedial foramen. The hypoglossal fora-
men is double except in one specimen of
7. typits, USNM 15933, in which it is single.
The two foramina, of nearly equal size,
open into a pit that deepens medially; the
anterior foramen faces posterolaterally, the
posterior, anterolaterally.

The postglenoid foramen is reduced to a
slit-like opening with a major axis measur-
ing 1.5 mm. It is enclosed within the squa-
mosal. A channel can be traced from this
foramen through the bone to the single
temporal foramen (USNM 175352). In
several specimens there seems to be a post-
alar fissure between the alisphenoid bone
and tlie front of the bulla (AMNH 694 and
MCZ 17202 show it clearly). The temporal
foramen, single in most specimens, is
primarily in the squamosal bone; the
squamoso-parietal suture descends into it
on the medial side. Its anterior end is above
the posterior part of the postglenoid fora-
men, and its major axis ranges from 2.5 to
3.0 mm. In one specimen, AMNH 694, a
second minute foramen is present behind it
in the squamosal bone. The squamoso-
mastoid foramen is present on the occipital
surface; one broken specimen, USNM
175352, has a matrix-filled channel running
from it to the temporal foramen; another,

392

Bulletin Museum of Comparative Zoology, Vol. 146, No. 8

CM 24129, preserves the channel without
infilhng.

The minute mastoid foramen is in the
occipital-mastoid suture well above the
level of the top of the foramen magnum.
The stylomastoid foramen is well defined
and very deep because the bulla is large
and the mastoid element inflated.

Discussion of the Ischyromyidae

There is ample skull material of
Ischyromys to permit thorough description
of its morphology. With respect to the
cranial foramina, ischyromyids and para-
myids are so different from each other that
I must agree with those (e.g.. Wood, 1965;
Wilson, 1949c; Schaub, 1958) who con-
sider them to be separate groups of equiva-
lent rank. I did not find any differences in
cranial foramina which would distinguish
Ischyromys from Titanotheriomys.

The ratio of length of the incisive foram-
ina to diastemal length has a range that
overlaps that of Ischyrotomus and Manitsha
but does not overlap those of Pammys,
Reithroparamys, and Scitiravus. The lateral
margins of the incisive foramina are inter-
sected farther posteriorly by the premaxil-
lary-maxillary suture in ischyromyids. The
pair of major posterior palatine foramina
are on the maxillary-palatine suture and
are usually farther forward relative to the
cheek teeth than in paramyids and
Scitiravus. In ischyromyids the posterior
extremity of the maxilla is appressed to the
pterygoid extension of the palatine. There
is a slit between them which may be homol-
ogus to the posterior maxillary notch in
paramyids and Scitiravus.

In side view the ischyromyid infraorbital
foramen is inclined, not vertical as in
paramyids and Sciuravus. The absolute
lengths of the major axes are about the
same as in paramyids although the skull
size is smaller in ischyromyids; if only the
vertical dimension is considered, the length
in ischyromyids is shorter than that in
paramyids.

Relative to the cheek teeth, the major
foramina of the orbit — the optic, ethmoid,
and sphenopalatine — are considerably far-
ther forward than in paramyids and
Sciuravtis. The sphenopalatine is bordered
by a fused maxilla and palatine and is met
posteriorly by the orbitosphenoid; these
features differ in paramyids and Sciuravus.
The dorsal palatine foramen is internal to
the sphenopalatine, whereas in paramyids
the two foramina are within a single de-
pression, and in Sciuravus they are sepa-
rated. In ischyromyids the posterior
palatine foramen, through which tlie de-
scending palatine artery and nerve emerge,
is anterior to its position in paramyids and
Scitiravus. The depression in the orbito-
sphenoid in front of the optic foramen for
the rectus muscles of the eye and the
interorbital foramen accompanying it do
not occur in paramyids or Sciuravus. The
sphenofrontal foramen is small and, in some
specimens, entirely within the orbito-
sphenoid. Where seen in paramyids and
Sciuravus, it is relatively larger and always
in the orbitosphenoid-alisphenoid suture.

Masticatory and buccinator foramina are
united with the foramen ovale accessorius
in most specimens. When both foramina
are present, the channels leading to them
through the alisphenoid bone are very short
in comparison to those of paramyids but
similar to those of Sciuravus. The foramen
ovale accessorius is present in ischyro-
myids and in some paramyids, but lacking
in Sciuravtis. In ischyromyids the external
pterygoid flange is very well developed
and reaches the bulla. The mandibular
division of the trigeminal nerve emerged
from the foramen ovale, bent laterally, and
passed through the foramen ovale acces-
sorius in the extended flange.

The entrance to the transverse canal is
separated from the alisphenoid canal. It is
possible to picture the transition from a
condition like that found in Ischyrotomus
and Scitiravus to this arrangement. As the
external pterygoid flange extended and the
pterygoid fossa developed for muscle at-

Cranial Foramina • Walilert 393

tacliment, the entrance to the ahsphenoid
canal moved as far back as the foramen
ovale. The foramen ovale, through which
the mandibular division of the trigeminal
ner\'e passes, acted as a barrier to further
extension.

The absence of a stapedial foramen indi-
cates that ischyromyids lacked the stapedial
artery, and the ophthalmic artery must have
arisen, therefore, from the internal carotid.
In paramyids and Sciiiravus a channel for
the artery crosses the promontorium. A
double hypoglossal foramen occurs in
ischyromyids, ReitJnoparamys, and Ischijro-
tomus.

Although the postglenoid foramen is
similar in length to those measured in
paramyids, it is reduced in width; this
suggests that the major course of the venous
system that drains the cranial cavity has
shifted away from it. There is one large
temporal foramen rather than a few small
ones as in paramyids. The post-alar fissure
and the squamoso-mastoid foramen do not
occur in paramyids or Sciuravus.

CYLINDRODONTIDAE

Specimens examined:

Ardynomijs occidentalis (Fig. 10): Chad-
ron Formation equivalent: CM 1055
npo, 9991 nptc, 9992 c, 16995 npot,
21701 npot.

Cylindrodon fontis: Chadron Formation
equivalent: CM 17180 np, 17181 np,
17204 pc. White River Series equiv-
alent: AMNH 14584 pc, 14585 np.

C sp.: Chadron Formation equivalents;
F:AM 79100 s, 79102 np, 79104 np,
79105 np, 79109 s; CM 6546 np, 6643
np, 8904 np.

Pseudocylindrodon meditis: Chadron
Formation equivalent: CM 1135 np,
10000 np, 10001 s.

P. neglectus: Chadron Formation equiv-
alent: CM 10100 np.

P. sp.: Chadron Formation equivalent:
CM 1126 n, 6545 np.

Foramina

The ratio of length of the incisive foram-
ina to diastemal length is quite variable in
cylindrodontids. Two specimens of Pseudo-
cylindrodon, CM 10100 and 6545, have
ratios of .37 and .40; the rest range from
.50 to .55. The ratio in Ardynomys ranges
from .28 to .44; in Cylindrodon, from .28
to .38. The lateral margins of the foramina
are intersected near the back in some speci-
mens, and at the back in oth(>rs, by the
premaxillary-maxillary suture. The suture
runs laterally and somewhat posteriorly
away from them.

The pair of posterior palatine foramina
is in the maxillary-palatine suture medial
to the middle part of the first molars. In a
few specimens of Cylindrodon it is slightly
farther posterior, medial to the junction of
the first and second molars. The maxilla
ends in a point behind the cheek teeth, and
in most specimens it is so closely appressed
to the pterygoid extension of the palatine
that there is neither a posterior maxillary
notch nor a foramen. In one specimen of
Pseudocylindrodon, CM 10001, a posterior
maxillary foramen is visible.

In front view the infraorbital foramen is
approximately circular with some flattening
on the dorsomedial side so that the major
axis appears to slant outward. The axis has
a wide size range; in Pseudocylindrodon
from 1.2 to 1.6 mm; in Ardynomys, a
sample of only two, around 2.0 mm; in
Cylindrodon from 0.9 to 1.7 mm. In side
viev^ the foramen is approximately vertical.
In one specimen of Cylindrodon, F:AM
79104, the foramen is double on both sides.
The anterior alveolar foramen is in the
curve made by the orbital wall and floor.
It is within the infraorbital foramen and is
directed ant(>romedially in Pseudocylindro-
do)i and Cylindrodon; in one specimen of
Cylindrodon, CM 17180, it is almost in
front of the infraorbital foramen and heads
medially. In Ardynomys the foramen is
farther posterior and is directed even more
medially.

394 Bulletin Museum of Comparative Zoology, Vol. 146, No. 8

msc

op elh

1cm

fo<^ msc bu

spn

Figure 10. Ardynomys occidentalis (composite of all specimens examined; CM 9991 used as base). See Fig. 1
for key to foramina.

The nasolachrymal foramen is dorsal to
the infraorbital. The maxilla may partici-
pate in its anterior margin, but I cannot
determine this for certain. Some specimens
have a slight channel leading down into the
foramen; others have none. The canal in-
side the bone n.ms anteroventrally past the
infraorbital foramen; then it plunges medi-
ally under the arch of the incisor. For the
first part of its course it is external to the
incisor alveolus, which reaches back into
the orbit in Ardynomys and Cylimlrodon.
In the orbital wall of P.seudocylindrodon

and in some specimens of Ardynomys
(Wood, 1970: fig. 4), there is a slight saddle
posteroventral to the nasolachrymal fora-
men; the saddle most likely marks the area
of origin of the inferior oblique eye muscle,
as in Marmota.

The sphenopalatine foramen is dorsal to
the posterior part of the first molar in
Pseiidocylindrodon and is directed antero-
medially. The maxilla forms its ventral and
anterior margins; sutures are not distinct
enough for other bones reaching it to be
distinguished. The position of the foramen

Cranial Foramina • Wahlert 395

ill Ardynoinys is somewhat variable. In
three specimens, CM 1055, 16995, and
21701, it i.s above the junction of the first
and second molars, while Wood (1970:
fig. 4 of CM 12010) shows it as being
dorsal to the anterior part of the second
molar. It plunges anteroventrally behind
the incisor root capsule. The fused maxilla
and palatine forms the border on three
sides, and a process from the orbitosphenoid
reaches the foramen posteriorly. Specimens
of CyUndrodon are fragmentary and diffi-
cult to interpret; the position of the spheno-
palatine foramen seems to differ from
indi\'idual to individual.

The ethmoid foramen is dorsal to the
back of the sphenopalatine in Pseudo-
cyUndrodon, CM 10001, and CyUndrodon,
CM 17180, and above and about equi-
distant between the sphenopalatine and
optic in Ardynomys. It is overhung by a
lip from the frontal bone. The orbito-
sphenoid reaches it posteriorly.

The optic foramen measures 0.5 mm in
diameter in Pseudocylindrodon, CM 10001;
it is dorsal to the junction of the second and
third molars and very close to the orbito-
sphenoid-maxillary suture. In Ardynomys
its diameter ranges from 0.7 to 0.9 mm, and
its position is dorsal to an area ranging
from the front part of the third molar to a
point slightly posterior to that tooth. Wood
(1970: fig. 4) shows the alisphenoid as
fomiing the back of the optic foramen; this
is not the case in the specimens I have
examined. In one specimen of CyUndrodon
in which the optic foramen is preserved,
CM 17180, it is above the back of the
second molar and about 0.5 mm in diam-
eter. In this same specimen there is a
depression in front of the optic foramen
that is more pronounced on the right side
than on the left. Presumably this marks the
place of origin of the rectus muscles of the
eye.

The dorsal palatine foramen is hidden
in all three genera. There is evidence in
one specimen of Ardynomys, CM 1055, that
it and the sphenopalatine are contained in

a single depression, as in Paramys. The
sphenoidal fissure at its entrance is open
medially into the cranial cavity; the open-
ing extc>nds a short distance in front of the
margin of the fissure. This condition prob-
ably occurs in CyUndrodon also. Burke
(1936:139) indicated two openings in this
region in Ardynomys, CM 1055, not recog-
nizing that the partition between them was
a mineral-filled fracture; the specimen has
a normal, single sphenoidal fissure.

In PseudocyUndrodon the minute spheno-
frontal foramen is level with but quite far
behind the top of the optic foramen, and
it may be on the alisphenoid-orbitosphenoid
suture although this is indistinct in all
specimens. A channel leads anteroventrally
from it. I have found the foramen in one
specimen of Ardynomys, CM 21701, at the
point of intersection of the orbitosphenoid,
parietal, and alisphenoid bones; it is very
small. The foramen seems to be absent in
other specimens of Ardynomys and in
CyUndrodon.

The masticatory and buccinator foramina
are separate, and relatively close to the
foramen ovale, about 1.0 mm from it. A
broad channel runs anterodorsally from the
masticatory foramen; a short one proceeds
forward from the buccinator. Burke (1936:
139; and 1938:260) identified the bucci-
nator as the masticatory and did not detect
the masticatory foramen. The existence of
a foramen ovale accessorius is uncertain.
The lateral pterygoid ridge is expanded
into a flange, and what remains of it de-
fines the anterior end of a foramen or at
least a notch. The flange may have reached
the bulla, in which case there would ha\'e
been a foramen.

The transverse canal, as such, has been
lost, being reduced to a foramen that opens
directly into the cranial cavity from the
middle of the moderately deep pterygoid
fossa. This is seen clearly in a specimen of
Ardynomys, CM 21701, in which the inside
of the cranium is partially cleaned out.
Burke (1936:139; and 1938:259) incorrectly
identified this foramen in the pterygoid

396 Bulletin Museum of Comparative Zoology, Vol. 146, No. 8

fossa as the sphenopterygoid. The ahsphe-
noid canal is small and begins just antero-
ventral to the foramen ovale. It is clearly
defined in Pseudocylindrodon and Cylin-
drodon. In Ardynomys the wall separating
it from the cranial cavity is almost gone; in
CM 1055 little more than a strut of bone
internal to the buccinator foramen is left.
The middle lacerate foramen, if present, is
covered by the tympanic bulla. In Ardyno-
mys and Cylindrodon a foramen of uncer-
tain function pierces the side of the basi-
sphenoid just anterior to the bulla; it is
hidden by the hamular process in Figure 10.

The posterior opening of the carotid
canal is very small and is separate from the
jugular foramen. In Pseudocylindrodon the
aperture seems to be between the bulla and
the basioccipital. In Ardynomys and Cylin-
drodon a short channel leads anterodorsally
across the medial side of the bulla to the
canal. The canal enters and runs anteriorly
through the periotic in Ardynomys; its
course can be seen in one broken specimen,
CM 9991. It also enters the periotic in at
least one specimen of Cylindrodon, CM
17204. There is no stapedial foramen. The
hypoglossal foramen is minute in Pseudo-
cylindrodon and seems to be situated on the
margin of the jugular foramen. The region
is somewhat damaged, and this interpreta-
tion could be incorrect. In the other two
genera it is more medial; in one specimen
of Ardynomys, CM 9991, it is single on the
left side and double on the right; the tu^o
foramina face anteriorly into a single de-
pression.

The postglenoid foramen is within the
squamosal bone. In Pseudocylindrodon,
CM 10001, the major axis measures 0.6 mm;
in two specimens of Ardynomys, 1.1 mm,
The temporal foramen is about 1.0 mm
long in one specimen of Pseudocylindrodon
and 0.8 and 1.3 mm in two specimens of
Ardynomys. In all specimens the region is
too crushed for the possible presence of
other, smaller temporal foramina to be
detected. In Ardynonujs, at least, the
parietal-squamosal suture does not dip

down far enough to reach the foramen.
Burke (1938:258) stated that in Pseudo-
cylindrodon the foramen apparently "marks
the suture between the parietal and squa-
mosal." The suture is not visible, however,
and his surmise is doubtful. He (1936:
136) calls tlie foramen a subsquamosal in
Ardynomys. The squamoso-mastoid fora-
men is conspicuous in all three genera. In
one specimen of Cylindrodon, CM 17204,
the squamosal is broken away, and the
channel from this foramen to the temporal
foramen can clearly be seen in the surface
of the underlying bone.

The mastoid foramen is probably present
but too minute to be found. The stylo-
mastoid foramen is anteroventral to the low
mastoid process just behind the external
auditory opening.

Discussion of the Cylindrodontidae

The three genera of cylindrodontid ro-
dents examined are very similar to one
another in their cranial foramina. They are
quite different from paramyids and Sciu-
ravus and in some features resemble the
ischyromyids.

The ratio of incisive foramen length to
diastemal length has a great range. It
includes the ranges of Paramys, Reithro-
paramys, and Sciuravus, has a slight overlap
at the low end with that of ischyromyids,
and does not overlap the range of the other
paramyids measured. The margins of the
foramina are intersected very far back by
the premaxillary-maxillary suture, as in
ischyromyids. Tlie posterior palatine foram-
ina are on the maxillary-palatine suture, as
they are in Ischyrotomus, Pseudotomus,
and ischyromyids. Relative to the cheek
teeth they are situated as in paramyids and
are slightly anterior to the position in
Sciuravus. The posterior maxillary foramen
is present in Pseudocylindrodon. In Ardy-
nomys and Cylindrodon the end of the
maxilla is appressed to the pterygoid ex-
tension of the palatine as in ischyromyids,
but there is no aperture between them.

Cranial Foramina • Walilert 397

The infraorbital foramen is nearly verti-
cal, as in paramyids and Sciuravus. The
orbital foramina, as in ischyromyids, are
considerably farther forward relative to the
cheek teeth than in paramyids and Sciu-
ravus. The sphcMiopalatine foramen is
bounded by a fused maxilla and palatine,
and is met posteriorly by the orbitosphe-
noid; both characters occur in ischyromyids
and not in paramyids and Sciuravus. The
sphenofrontal foramen i.s minute in Pscudo-
cylindrodon as in ischyromyids, and it is
absent in the other cylindrodontids.

The reduction of bone internal to the
sphenoidal fissure and to the alisphenoid
bone does not occur either in paramyids or
in ischyromyids. The masticatory and
buccinator foramina are separate as in
paramyids, but they are close to the fora-
men ovale as are those seen in Sciuravus
and the earliest skulls of ischyromyids. The
development of a pterygoid fossa with a
foramen, presumably leading to the trans-
verse canal, in the middle of it, the possible
presence of a foramen ovale accessorius,
and the position and relative size of the
entrance to the alisphenoid canal are as in
ischyromyids. The foramen in the basi-
sphenoid just anterior to the front of the
bulla may be homologous to the medial
division of the middle lacerate foramen in
paramyids.

The separation of the entrance to the
carotid canal from that to the jugular
foramen, and the passage of the canal
through the periotic, which occur in
Ardynomys and CyJindrodon, are different
from the conditions in paramyids, Sciu-
ravus, and ischyromyids. Both cylindro-
dontids and ischyromyids lack the stapedial
foramen. The hypoglossal foramen is
usually single, as in Sciuravus and some
paramyids, whereas in ischyromyids the
foramen is always double.

The shift in emphasis of the venous
system away from the postglenoid foramen,
the presence of a single temporal foramen
below the parietal-squamosal suture, and
the opening-up of a squamoso-mastoid

foramen are all changes from the paramyid
condition that occur in ischyromyids.

PROSCIURIDAE

Specimens examined :

Prosciurus relictus: Cedar Creek Mem-
ber, White River Formation: KU 8333

npo, 8345 npo.
P. aff. .saskatchewaensis (Fig. lib):

Orella Member, Brule Formation:

AMNH 1429 c.
P. cf. vetustus: Orella Member, Brule

Formation: FMNH, PM 14674 np.
P. sp. (Fig. 11a): Orella Member, Brule

Formation: SDSM 62365 npo.
?Cedromus sp.: Cedar Creek Member,

White River Formation: KU 8342 ptc.

Foramina

The ratio of length of the incisive
foramina to diastemal length ranges from
.39 to .48. The lateral margins of the
foramina are intersected near the back by
the premaxillary-maxillary suture, which
runs laterally and somewhat posteriorly
away from them.

The pair of major posterior palatine
foramina is in the palatine medial to an
area ranging from the posterior portions of
the second molars to the anterior portions
of the third molars. A second, smaller pair
is medial to the third molars. The maxilla
ends in a blunt point, and a posterior maxil-
lary notch is fomied between it and the
pterygoid extension of the palatine.

In front view, the infraorbital foramen is
nearly circular. The major axis measures
1.3 mm, and it is inclined so that the top
of the foramen is farther lateral than the
bottom. In side view the foramen slopes
forward slightly. The anterior alveolar
foramen, which is in the curve made by
the orbital floor and wall, lies either within
or a short distance behind the infraorbital
foramen and is directed anteromedially.

The nasolachrymal foramen is dorsal to
the infraorbital. A channel descends the
face of the lachrymal bone into the fora-

398 Bulletin Museum of Comparative Zoology, Vol. 146, No. 8

tom.

dpi spl

1 cm

Figure 11a. Prosciurus sp. (SDSM 62365). See Fig. 1
for key to foramina.

men. Immediately posterior and slightly
ventral to the nasolachrymal foramen is a
depression in the bone, whether in the
maxilla or lachrymal cannot be determined.
This was apparently the area of origin of
the inferior oblique eye muscle.

The sphenopalatine foramen is above the
middle or the posterior part of the second
molar. The maxillary-frontal suture inter-
sects it dorsally; the orbital portion of the
palatine reaches it posteriorly, and the
orbitosphenoid is wholly excluded from its
margin. The ethmoid foramen is dorsal to
the posterior part of the second molar. It is
within the frontal bone and overhung by a
lip of that bone. The orbitosphenoid does
not reach it posteriorly. Only the anterior
part of the optic foramen is preserved in

Figure 11b. Prosciurus aff. saskatchewaensis (AMNH
1429; modified from Wood, 1937: plate 13; lateral view
additional). See Fig. 1 for key to foramina.

one specimen, SDSM 62365; it is dorsal to
the posterior part of the third molar. Al-
though damaged, the foramen was about
1.5 mm long. In two specimens, KU 8333
and 8345, a depression is present immedi-
ately anterior to the optic foramen. It con-
tains a minute interorbital foramen and
was probably the area of origin for the
rectus muscles of the eye.

The dorsal palatine foramen in the
orbital floor is above the posterior part of
the third molar and lies within the palatine
near the maxillary suture in KU 8342, and
in the suture in KU 8333. The sphenoidal
fissure is dorsal and wholly posterior to the
last molar. The region in which the spheno-
frontal foramen would be situated is not
preserved in any of the specimens.

My interpretation of the orbital region
is at variance with Galbreath's identifica-
tion of certain foramina in Prosciurus
relictus, KU 8333 (Galbreath, 1953:53, fig.
16). The foramen he labeled the optic is
the interorbital, and the sphenoidal fissure
is the optic foramen. The posterior of two
foramina shown in the orbital floor is the

Cranial Foramina • Wahlert 399

dorsal palatint^; tlie maxillary-palatine
suture is not very clear in the specimen,
but it seems to pass through this foramen.
The anterior aperture is an exposed root of
the second molar.

The alisphenoid region is preserved only
in KU 8342. Galbreath (1953:60, fig. 19)
identified the most conspicuous foramen as
the masticatory. It is situated rather far
Neutrally in comparison with the masti-
catory in other rodent skulls examined, and
it faces anteriorly; both the position and
orientation would indicate that it is the
buccinator foramen. However, a second
aperture leading from the alisphenoid canal
is present anterior to it near the edge of the
alisphenoid; this opening is either the
buccinator foramen or a foramen that does
not occur in any of the other rodent skulls
I have examined. The problem of identify-
ing these foramina with certainty cannot
be solved because the region in which the
mastoid foramen would normally occur is
damaged. The area labeled by Galbreath
as the buccinator foramen is a damaged
foramen ovale accessorius.

The shallow pterygoid fossa is partially
preserved in this same specimen. The
foramen ovale is situated posterolaterally
within the fossa. The alisphenoid canal
begins ventral to the anterior end of the
foramen ovale, the transverse canal medial
to it. The middle lacerate foramen is ab-
sent or covered by the tympanic bulla.

The carotid canal, if present, shares an
opening with the jugular foramen and is
very narrow. The jugular foramen appears
to be broader than in other protrogomorphs,
but this is due to the presence of the sta-
pedial foramen lateral and slightly dorsal
to it in a common depression. The hypo-
glossal foramen is single and medial to the
jugular.

The postglenoid foramen, which is hardly
more than a slit in the squamosal, is in line
with the zygomatic root; its major axis
measures 0.8 mm. There is one temporal
foramen in the squamoso-parietal suture
dorsal and slightly posterior to the post-

glenoid foramen. The squamosal is broken
off on the left side of the specimen, and a
conspicuous channel connecting the tNvo
forannna can be seen on the surface of the
underlying bone. A channel also runs
posteroventrally from the postglenoid fora-
men to the region where the rounded tip
of the squamosal meets the mastoid ele-
ment. This indicates the presence of a
squamoso-mastoid foramen.

The mastoid foramen is very dorsally
situated in the mastoid-occipital suture.
The stylomastoid foramen is in its usual
position between the external auditory
meatus and the mastoid.

Discussion of the Prosciuridae

The ratio of length of the incisive fora-
men to diastemal length has a range that
includes those of Paramys, Reithroparamys,
and Sciuravus. It overlaps the upper end
of the ranges in cylindrodontids and it is
above those in other protrogomorphs. The
major pair of posterior palatine foramina is
situated farther posterior relative to the
cheek teeth than that of any other protrogo-
morphous form. The posterior maxillary
notch is as in paramyids.

The infraorbital foramen is not vertical
as in paramyids and Sciuravus. It slopes
anteriorly though not as much as that in
ischyromyids. The depression for attach-
ment of the inferior oblicjue eye muscle
does not occur in other protrogomorphous
rodents, although the site is indicated in
Isdiyromys; Pseudocylindrodon, and Ardy-
nomys. The sphenopalatine foramen is sur-
rounded by the frontal, maxillaiy, and pala-
tine bones, as in paramyids and Sciuravus,
in which the frontal is barely excluded from
the dorsal margin; the orbitosphenoid does
not reach the foramen in any of these forms.
The ethmoid foramen is within the frontal,
as in Sciuravus and some specimens of Para-
mys. The size of the optic foramen is rela-
tively large for a protrogomorphous rodent.
The depression for attachment of the rectus
muscles of the eye and the interorbital fora-

400 Bulletin Museum of Comparative Zoology, Vol 146, No. 8

men within it are also present in ischyromy-
ids, but not in other protrogomorphs. The
dorsal palatine foramen is separate from
and posterior to the sphenopalatine, as in
Sciuravus.

The pterygoid fossa is shallow and the
foramina within it are situated as in
Ischyrotomus and Sciuravus. The stapedial
foramen indicates the presence of a sta-
pedial artery, which was also present in
paramyids and Sciuravus; ischyromyids and
cylindrodontids lacked it.

The postglenoid foramen is greatly re-
duced, but a temporal foramen and the
squamoso-mastoid foramen are present.
This arrangement is unlike that of para-
myids and sciuravids and similar to that
of ischyromyids and cylindrodontids.

APLODONTOIDEA

Specimens examined:
Aplodontidae:
Allomijs nitens (Fig. 12): John Day

Formation: UCMP 1100 np.
Liodontia furlongi: Barstovian deposits,

Nevada: UCMP 61716 s, 75666 np.
Aplodontia rufa: Recent: UNSM, Z.M.

275: MCZ 799, 1893, 5645, 6369, 6822,

13183, 17810, 18352.
Mylagaulidae:

The taxonomy of mylagaulids is at
present so uncertain that I prefer to list
the specimens examined in stratigraphic
order.
Rosebud Formation:

Promylagaulus riggsi (see McGrew,
1941:6 for figure): FMNH, P 26256
npo.
Marsland Formation and equivalents:

Mesogaulus laevis: UNSM 04953 npot,
04954 p; F:AM 65004 npo.
Sheep Creek Formation:

Troimjkigaulus novellus^: F:AM 65001
np.

Mesogaulus sp.: F:AM 65002 potc,

^ This species has heretofore been placed in the
genus Mijlagaulus, but Stout (personal communi-
cation) now places it in the genus Promylagaulus.

1 cm.

Figure 12. Allomys nitens (UCMP 1100). See Fig. 1 for
key to foramina.

65003 s, 65005 np, 65006 s, 65007 np,

65011 nptc.
Mylagaulus vetus: AMNH 18903 pot,

18904 np, 20507 npt.
Mylagaulus laevis (Fig. 13): AMNH

17576 s.
Ceratogaulus rhinocerus: AMNH 18899

large mylagaulid: F:AM 65016 s, 65017

nptc.
Pawnee Creek Fonnation:

Ceratogaulus rhinocerus: AMNH 9456

(type) npt.
Mylagaulus laevis: AMNH 9043 (type)

np.

Cranial Foramina • Wahlert 401

'ip^ op

7ns

1 cm

Figure 13. Mylagaulus laevis (mostly based on AMNH 17576; basicranium resto ed from various specimens).
See Fig. 1 for key to foramina.

Barstovian deposits, Montana:

Mylagauhis douglassi: UCMP 44694

npo.
Mylagaulus sp.: PU 18186 s.
Deep River beds:

Mylagaulus sp.: AMNH 21307 s.
Valentine Formation:

Mylagaulus sp.: UNSM 04957 n.
Burge Member, Valentine Formation:
Mylagaulus sp.: UNSM 04956 np;
F:AM 65009 potc, 65010 potc, 65012
s, 65013 to.
Clarendonian deposits. South Dakota:
Mylagaulus sp.: UCMP 32323 potc.

Ash Hollow Fonnation:

Mylagaulus sp.: UNSM 04955 n.
Kimball Formation:

Mylagaulus sp.: UNSM 04958 npo.
Pliocene deposits, Nebraska:

Mylagaulus sp.: F:AM 65008 np.,
65014 npo, 65015 np.

Foramina

The interpremaxillary foramen is a con-
spicuous feature in Aploclontia. It is quite
variable in the fossils. Liodontia seems to
have a .small one; the region is not pre-

402 Bulletin Museum of Comparative Zoology, Vol. 146, No. 8

served in AUomijs. In mylagaulids a pit and in other aplodontids, from 2.5 to 4.0 mm,

sometimes a minute foramen are present the low values being from juveniles; in

in its position. Promylagaulus lacks the Promylagaulus, 1.2 mm; and in other

foramen. mylagaulids, from 1.9 to 4.8 mm. In side

The ratio of length of the incisive foram- view the foramen is nearly vertical. The

ina to diastemal length in aplodontoids anterior alveolar foramen in AUomys is

ranges from .28 to .43, and values are quite above the fourth premolar in the curve

evenly distributed. The ratio does not vary formed by the orbital wall and floor. The

with the geologic age of the specimens. No position of the foramen is extremely vari-

ratio can be obtained for AUomys; Promy- able in ApJodontia; it is seen most com-

lagaulus falls at .33. In AUomys, as in all monly in the medial wall of the infraorbital

aplodontids, the lateral margins of the foramen. The foramen was not seen in

foramina are intersected at the back by the most mylagaulid specimens; it is probably

premaxillary-maxillary suture, which runs obscured by the great alveolus of the fourth

laterally and somewhat posteriorly away premolar. In one specimen, UCMP 32323,

from them. In most mylagaulids the ends it lies in the medial wall of the infraorbital

of the foramina are intersected by the foramen.

suture, which runs posteriorly from them; The lachrymal region is very well pre-

in some specimens the maxilla does not ap- served in most specimens. In AUomys the

pear to reach the foramina, which may be lower margin of the nasolachrymal foramen

entirely within the premaxillary bones. is medial to the upper part of the infra-

The major pair of posterior palatine fo- orbital. Sutures in the area are unclear, and

ramina is medial to an area ranging from it is not possible to determine which bones

the middle of the second molars to the an- surround the foramen. In Promylagaulus

terior part of the third molars in all the nasolachrymal is considerably farther

aplodontoids. In AUomys and in juvenile dorsal. In other aplodontids and mylagau-

specimens of Aplodontia the pair is very lids it is a short distance above and slightly

close to the maxillary-palatine suture, but posterior to the infraorbital, and a channel

within the palatine; the suture is not visible leads ventromedially down the face of the

in adult aplodontoids. In one specimen of bone to it. In one juvenile specimen of

Liodontia a second, smaller pair is in line Aplodontia, MCZ 5645, sutures can be

with the first pair and medial to the back seen; the maxilla fonns the anteroventral

of the second molars. Aplodontia commonly edge of the foramen. In one mylagaulid,

has one or two pairs of small foramina in PU 18186, the first part of the canal is

line with and posterior to the major pair, exposed. It slopes to a point just antero-

The maxilla ends behind the cheek teeth in ventral to the infraorbital foramen where

a point, which is fused to the lateral side it turns anteromedially. In many large

of the pterygoid region in all but juvenile specimens of Aplodontia a rounded notch

specimens. The posterior maxillary fora- is present in the posterior projection of the

men, enclosed between the two parts, opens lachrymal; clearly it transmitted a vessel or

above in the floor of the sphenoidal fissure nerve to the top of the head,

and may have transmitted a palatine vein. In order to render the positions of the

The region is not preserved in AUomys and orbital foramina intelligible it is necessary

is unclear in Promylagaulus. to digress and to explain the mode of

In front view, the infraorbital foramen is cheek tooth replacement in mylagaulids.

of variable shape; it may be elliptical with The cheek teeth of mylagaulids are very

the major axis running diagonally or hori- hypsodont, but the fourth premolar greatly

zontally, or it may be nearly round. In surpasses the molars in this respect. This

AUomys the major axis measures 1.8 mm; tooth is shaped like a long wedge, and, as

Cranial Foramina • Wahlert 403

it erupts, the first molar and then possibly gin of the optic foramen is preserved in

the second are eliminated by interdental Allomys and ProniyIap.at(Jtis; it is dorsal to

wear. A cheek tooth dentition may con- the posterior part of the third molar. The

tain F\ M", M-, M\ or P\ M-, M\ or just curvature of both margins suggests that the

P^ M'\ The apparent differences in position optic foramina were about 1.0 mm in

of the orbital foramina relative to the cheek diameter. In Aploduntia the optic foramen

teeth are determined by the degree of is dorsal to an area ranging from the middle

encroachment of the fourth premolar on of the second molar to the front part of the

the molars. In contrast, the cheek teeth of third molar, and it is ncnirly reached by the

Liodontia and Aploclontia are all hypselo- alveolus of the second molar. The diam-

dont; the orbital foramina are high and eter of the foramen ranges from 1.0 to 1.5

above the alveoli, and major differences in mm. Juvenile specimens demonstrate that

position do not occur. it is wholly within the orbitosphenoid. In

The sphenopalatine foramen is above the mylagaulids, other than Promylaiiouhis, the

second molar in AUomy.s; the orbital optic foramen is close behind the spheno-

process of the palatine reaches it posteriorly, palatine and dorsal to the same area as in

and tlie orbitosphenoid is excluded from Aploclontia; it ranges in size from 0.6 to 0.9

its margin. In Aploclontia the foramen is mm.

dorsal to the posterior part of the premolar; An interorbital foramen is present in

the cheek tooth alveoli obscure all sutures Allomys immediately in front of the optic,

in the region. In Promylagaulus the fora- McGrew (1941:7, fig. 2) identifies it as the

men is dorsal to the junction of the first and optic foramen, but it has no connection

second molars, whereas in other mylagau- with the cranial cavity; I believe that his

lids the foramen is above the second molar, sphenoidal fissure is the optic foramen and

posterior to the great alveolus of the last that the sphenoidal fissure itself is missing

premolar. In UCMP 32323, in which the from the specimen he examined. In mature

premolar is almost fully erupted and the specimens of Aploclontia a short process of

second molar worn away, the foramen is bone and a pit anterior to it are present in

above the posterior part of the premolar, front of the optic foramen; this is probably

Orbital sutures are fused in all specimens. the site of origin of the rectus muscles of

The ethmoid foramen is dorsal to the the eye.

posterior part of the second molar in The dorsal palatine foramen is on the

Allomys; sutures in its vicinity cannot be palatine-maxillary suture dorsal to the an-

determined. The foramen is overhung by terior part of the third molar in Allomys.

a slight lip of bone. In Aploclontia the It is hidden in Aplodontia but can be lo-

foramen is posterodorsal to the spheno- cated by pushing a hair through the

palatine and above the junction of the first posterior palatine foramen. The hair comes

and second molars. There are commonly out above the third molar in the crevice

two or three minute foramina close to- between the molar alveolus and the orbito-

gether; the number may differ on the two sphenoid. The foramen cannot be seen in

sides of a skull, and the orientation of in- most mylagaulids; when visible, it is dorsal

dividual apertures in such a group may to the third molar. The specimens of

differ substantially. In juvenile specimens Allomys and Promyhiiiauhis are broken off

the ethmoid foramen is within the frontal at this point. Tht> sphenoidal fissure at its

bone. I could not identify the ethmoid entrance is open medially into the cranial

foramen in Promylagaulus. In other cavity in mylagaulids and Aplodontia.

mylagaulids it is dorsal and slightly an- There is no sphenofrontal foramen,

terior to the sphenopalatine foramen; it The masticatory foramen is separated

opens upward. Only the anteroventral mar- from the buccinator by a distance of over

404 Bulletin Museum of Comparative Zoology, Vol. 146, No. 8

2.0 mm in some specimens of mylagaulids
and of Aplodontia. The buccinator foramen
is much smaller than the masticatory in the
latter. Although the pterygoid fossa is deep,
the lateral flange is short, and a foramen
ovale accessorius is lacking.

In mylagaulids and Aplodontia the ali-
sphenoid canal is very large and situated
anterolaterally in the fossa. The aperture
of the transverse canal is just inside the
medial border of the alisphenoid canal. In
mylagaulids the canal is very broad; one
broken specimen, PU 18186, reveals a pair
of channels running posterodorsally from
the canal into the cranium. This same speci-
men demonstrates that, as in Aplodontia,
the masseteric and buccinator portion of
the mandibular nerve splits off before that
nerve emerges from the foramen ovale. The
foramen ovale is situated posterolaterally in
the pterygoid fossa. There is a slight gap
between the roof of the fossa and the bulla.
In one mylagaulid specimen, UCMP 32323,
the foramen ovale is confluent with this
gap, and in Aplodontia the foramen may
be separate, or confluent, on one or both
sides of the skull.

Aplodontia lacks both carotid and sta-
pedial arteries (Guthrie, 1969; Bugge,
1971b); however, in many specimens a
minute foramen can be seen in the medial
wall of the periotic at the anterior end of
the jugular foramen. This foramen was
noted in one mylagaulid specimen, AMNH
17576; in some others a separate, minute
opening is present between the bulla and
the basioccipital. It is possible that these
foramina are remnants of embryonic ar-
terial passageways. The hypoglossal fora-
men is usually double, but occasionally
single.

The postglenoid foramen is between the
squamosal and the tympanic. Although it
resembles a post-alar fissure, it is too far
dorsal to be that aperture. Temporal
foramina are present near the occipital
crest. A single foramen is the most common
condition, but two foramina either in an
anteroposterior line or side by side are not

unusual. In one specimen of Aplodontia,
MCZ 5645, a channel connecting the
temporal and postglenoid foramina can be
seen inside the cranium.

The mastoid foramen is in the occipital-
mastoid suture slightly above the level of
the top of the foramen magnum. The stylo-
mastoid foramen is in its usual position, al-
though the external auditory meatus is
greatly extended beyond it.

Discussion of the Apiodontoidea

A conspicuous interpremaxillary fora-
men, or a distinct depression in this
position, which is present in many aplo-
dontoids does not occur in other protrogo-
morphous rodents. The ratio of length of
the incisive foramina to diastemal length
has a range that nearly matches those of
Ardynomys and Cylindrodon. It is below
the ranges in Paramys, Reithroparamys,
Sciuravus, Pseudocylindrodon, and pro-
sciurids, and above that in Ischyrotonms;
it overlaps the high end of the range in
ischyromyids. The posterior palatine foram-
ina are farther posterior than in other
protrogomorphs except prosciurids. The
posterior maxillary foramen is similar only
to that of Pseudocylindrodon within this
group.

The infraorbital foramen is vertical, and
of a size range similar to that in most
protrogomorphs. The ethmoid foramen
may be entirely within the frontal through-
out the aplodontoids, although its position
relative to the orbitosphenoid suture can
be determined only in juvenile specimens
of Aplodontia. The foramen is also within
the frontal in prosciurids, Sciuravus, and
some individuals of Paramys. The spheno-
palatine and optic foramina are consider-
ably farther forward relative to the cheek
teeth than in paramyids, but are in positions
similar to those in Sciuravus, Ardynomys,
and prosciurids. The size of the optic fora-
men is approximately as in other protrogo-
morphs. The interorbital foramen resembles
that of ischyromyids and prosciurids. Com-
plete separation of the dorsal palatine

Cranial Foramina • Wohlert 405

foramen from the sphenopalatine is a
resemblance to Sciuravus and proscinrids.
The sphenofrontal foramen is absent, as in
Ardynomys and Cylinchodon among protro-
gomorphs. Reduction of bone internal to
the sphenoidal fissure is a feature that is
also encountered within the suborder. Tlie
presence of separate masticatory and bucci-
nator foramina and the absence of the
foramen o\'ale aecessorius are characters
shared with paramyids and Sciuravus.

The arrangement of foramina in the
pterygoid fossa is very similar to that oc-
curring in Ischyrotomus, Sciuravus, and
proscinrids. It is very different from that
of ischyromyids and cylindrodontids. The
positions of the postglenoid foramen and
the temporal foramina differ from those
seen in other protrogomorphous rodents.

CONCLUSIONS

Cranial foramina vary within definite
limits. Those that transmit nerves are al-
wa\'s present although their number may
increase if a nerve divides inside rather
than outside the bone, or decrease if two
foramina fuse. Foramina transmitting ar-
teries vary by fusion and by loss when an
artery is eliminated; once a new pattern of
arterial circulation is fixed, it is perpetuated
in the lineage concerned. Foramina trans-
mitting veins are the most plastic in num-
ber and presence or absence, within certain
bounds imposed by the requirements of
circulation. With these principles in mind
the characters of the cranial foramina may
be used to test ideas on rodent phylogeny
and taxonomy. To this end, I present, first,
those features that set families and super-
families apart, one from another, and those
that indicate relationships between groups.
Next, I briefly discuss each family and
superfamily in an endeavor to integrate
these data with those presented by various
workers on the basis of other structures.
Following this, I propose a classification of
the groups studied that seems plausible in
light of the available evidence.

The Paramyidae, first appearing in the
late Faleocene, are the earliest known
rodents. Sciuravids, though recovered first
from slightly younger strata, are so similar,
that I shall consider the families together.
The paramyid skulls I have examined rep-
resent three of the four subfamilies desig-
nated by Wood (1962:11)'. They date
from the middle early Eocene (Lysite), and
later, and therefore may differ in some
features from the earliest members of the
family. The skulls of Sciuravus are of early
middle Eocene (Bridger) age and may
differ likewise from older members of the
Sciuravidae.

The dorsal palatine foramen and spheno-
palatine open into a common depression in
the three paramyid genera in which the
region remains intact. I believe this ar-
rangement to be primitive within the
Rodentia. In Sciuravus and in all rodent
families derived from paramyids (except
ischyromyids and cylindrodontids), the
dorsal palatine foramen is in the floor of
the orbit posterior to the sphenopalatine
foramen.

Three patterns of foramina in the pter)'-
goid region occur in paramyids. The in-
clusion of the foramina within a small
depression in Paramys copei could well be
structurally ancestral to the condition seen
in Ischyrotomus and Fseudotomus and in
Paramys delicatus. The pterygoid region
of Sciuravus is like that in Ischyrotomus
and Pseudotomus; from this arrangement
of foramina can be derived those in all
later rodents.

In Paramys, Leptotomus, Ischyrotomus,
and Pseudotomus a canal is present that
begins at the anterior end of the jugular
foramen and runs between the basioccip-
ital and periotic. I have called it the
carotid canal, since it is so termed in living
rodents. However, the mere presence of
this canal is no sure evidence that there was

^ Wood's fifth subfamily, the Prosciurinae, I
exclude froui the Paramyidae for reasons given
below.

406 Bulletin Museum of Comparative Zoology, Vol. 146, No. 8

a medial branch of the internal carotid
artery running through it; the canal also
transmits the inferior petrosal sinus and
would be present if either or both of these
vessels existed. The fact that some later
rodents of apparent paramyid descent have
an internal carotid artery in the canal, e.g.,
ischyromids (almost certainly) and cast-
orids, is evidence that some or all members
of the family had the artery.

The auditory region is preserved in the
genera above, except Pseud otomus. A
channel marking the course of the stapedial
artery runs across the promontorium from
the region of the jugular foramen to the
fenestra vestibuli. If the medial branch of
the internal carotid artery ran through the
carotid canal, then the stapedial branch
diverged from it outside, i.e., medial to, the
middle ear, as it does in later rodent
groups.

Leptotomus has not only a carotid canal
and stapedial artery channel, but also a
groove marking the course of the promon-
torial artery. Did this genus have a three-
branched internal carotid as has been
attributed to early mammals (cf., e.g.,
Szalay, 1972:71) and to primitive insecti-
vores (McKenna, 1966)? If this is the case,
then rodents had a very early origin within
the Mammalia or they are derived from an
insectivore retaining the medial branch of
the internal carotid; the promontorial arter}'
in Leptotomus is, then, a primitive relict.
Perhaps the genus did not have a third,
medial branch of the internal carotid, and
the carotid canal transmitted only the in-
ferior petrosal sinus. If this is the case,
then the carotid circulation was like that in
Scinravus, in which stapedial and promon-
torial branches are present, but a carotid
canal is lacking. This arrangement is like
those in li\'ing and most fossil insectivores
(Tandler, 1899:749; van Kampen, 1905:
422ff.; McDowell, 1958:205), and it is de-
scribed as primitive for primates (Gregory,
1920; Hill, 1953, 1955; McKenna, 1966:7).
If division of the internal carotid within the
middle ear is primitive for rodents, too, and

if the third, medial branch was lost prior to
their origin, then rodents may be related in
some way to these orders. The artery in the
carotid canal in paramyids and later rodents
is, then, the branch that crosses the promon-
torium in Leptotomus and Sciuravus. Ex-
amination of the canal transmitting the
carotid artery in the array of living rodents
reveals a variety of structural detail (Hill,
1935; Wahlert, 1972) which suggests that
the carotid canal arose separately in differ-
ent groups. This is to be expected if the
artery did not run in a canal in the earliest
rodents.

The ischyromyids and cylindrodontids
retain the close association of sphenopala-
tine and dorsal palatine foramina en-
countered in paramyids. In both families
the orbital foramina are farther anterior
relative to the cheek teeth than they are in
any other protrogomorphous group. The
orbitosphenoid reaches the sphenopalatine
in these and in no other groups. The
sphenofrontal foramen is reduced or absent;
the arrangement of foramina in the ptery-
goid region is unique to these two families.
The association of palatine and spheno-
palatine foramina is a primitive feature
also encountered in Paramys. The presence
of a carotid canal in ischyromyids is evi-
dence supporting derivation from the para-
myids. The canal is reduced in cylindrodon-
tids, and has a peculiar course in two of
the genera.

The prosciurids and aplodontoids form a
natural assemblage as regards their foram-
ina, and they have two peculiar features in
common. The ethmoid foramen is well
within the frontal bone rather than in or
near the orbitosphenoid-frontal suture, and
the posterior palatine foramina are, in
general, farther posterior than in other
rodents.

None of the early aplodontids are known
from skulls. The fragment of AUomijs has
an interorbital foramen anteroventral to the
optic foramen, as do some of the specimens
of Prosciurus. The fragment of Promijla-
gaulus indicates only that the genus is a

Cr.\nial Foramina • Wahlert 407

side branch of the MylagauHdae. The
arrangement of foramina in mylagauHds is
so much Hke tliat of Aplodontia that, on
this biisis, tlie two groups could be placed
in one family.

The cranial foramina of prosciurids do
not provide any clear evidence as to
whether tlieir ancestors were sciuravids or
paramyids. Separation of dorsal palatine
and sphenopalatine foramina occurs in
Sciuravus, but this condition in prosciurids
could be derived easily from that in para-
myids. The position of the ethmoid fora-
men just within the frontal seems to occur
in some paramyids and in Sciuravus. The
positions of the orbital foramina, relative
to the cheek teeth, are different from those
of either group. Differences in the ptery-
goid region indicate that Wood's (1962:
243) suggested derivation of prosciurids
from Paramys delicatus is unlikely, but this
by no means excludes the possibility of
descent from some other paramyid.

This is the evidence, based on cranial
foramina, that suggests the unity of certain
assemblages and their affinities to others.
What bearing does it have on the various
hypotheses concerning the relationships of
the groups discussed that have been pro-
posed on other grounds?^

Sciuravids date from the early Eocene
Lysite Member of the Wind River Forma-
tion. At that time sciuravids and paramyids
were so closely related that "... a con-
temporary taxonomist would never have
considered them distinct families" (Wood,
1965:133). The cranial foramina in the
two groups are nearly alike even at the
Ijcginning of the middle Eocene, when the
two are easily separated on the basis of the
dentition. The distinctive characteristics of
paramyids and Sciuravus may well stem
from differential retention of primitive

^Wilson (1949c) presents an excellent review
tracing the history of opinion regarding relation-
ships of all of the groups discussed here. I have
not thought it necessary, therefore, to repeat such
information, but limit myself to contributions
subsequent to his paper.

features in these genera and may not be
consistent throughout their respective fami-
lies. Wood believed the Sciuravidae arose
from, or nvar to, the Microparamyinae, a
group for which no skulls are known. There
is no available evidence that would cast
doubt on this conclusion.

Leptotomus with its primitive carotid
circulation occupies a special place within
the Paramyidae; this is in keeping with
Woods' phylogeny ( 1962:243), which shows
it as part of a distinct lineage since early in
the family's history. Whether or not the
genus should be retained in the Paramyinae
cannot be decided without a better sample
of paramyid skulls. The priniitiveness of
the genus makes it an unlikely candidate
for the ancestor of Isclujromys as proposed
by Wood (1962:248).

Ancestry of the ischyromyids has long
been in doubt because of dental resem-
blances to both paramyids and sciuravids.
Wood (1962:248) has, after earlier hesi-
tation, advocated a paramyid ancestry.
Black ( 1968a ) compared early Oligocene
species of Ischyromijs with paramyids and
sciuravids and found so great a similarity
to the former that he included the genera
of both groups in a single family, the
Ischyromyidae. The evidence from cranial
foramina, although not conclusive, supports
a paramyid ancestry for the family, but
striking differences in the arrangement of
foramina from that in the known paramyids
indicate that the two groups are distinct at
the familial level.

The cylindrodontids have been derived
from sciuravids by Wood ( 1955 and 1959 ) .
The unique features of the cranial foramina
common to cylindrodontids and ischyromy-
ids suggest that the two groups had a
common ancestry, which, on the evidence
of Ischyromys, I suspect to have lain within
the Paramyidae.

Wilson (1949b) and Wood (1962) be-
lieved that the Prosciuridae (Prosciurinae
in their usage) were derived from para-
myids. The cranial foramina, as stated
above, neither oppose nor support this

408 Bulletin Museum of Comparative Zoology, Vol. 146, No. 8

view; they do suggest, however, that the
group is entitled to famiHal rank, and this
is supported by other features. All the
following characters differ from those of
paramyids: stapedial foramen situated
dorsolaterally within jugular foramen; dor-
sal palatine foramen above third molar;
optic foramen, in part, dorsal to third
molar; lateral pterygoid ridge prominent,
possibly enclosing foramen ovale acces-
sorius (in Reithroparamys and Leptotomus
among paramyids); ethmoid foramen situ-
ated well within frontal; cranium distinctly
domed in profile; flattened lyrate area on
skull roof; postorbital process present and
strong; auditory bullae inflated; incisor
enamel uniserial; masseteric fossa extending
ventral to first molar (only in Manitsha
among paramyids). I agree, therefore,
with Wilson (1949c) and with Schaub
(1958) that familial rank is warranted.

The Aplodontidae appear in the latest
Eocene. Resemblances to the prosciurids
were demonstrated in detail by Wilson
(1949a and c). Wood (1962:243, 247)
suggested derivation of the group from a
prosciurine, probably Mytonomys, in the
middle late Eocene. Although Black
( 1968b ) , on the basis of new material,
demonstrated that Mytonomys is not a
prosciurid, the cranial foramina provide
strong support for the descent of aplo-
dontids from prosciurids. Mylagaulids and
aplodontids are so similar that I am in
complete agreement with those who derive
the former from aplodontids in the late
Oligocene. The three groups are very
closely related, and can be considered as
members of a single superfamily.

There remains for consideration the
bearing on formal taxonomy of the evidence
reviewed. Certain changes within the sub-
order Protrogomorpha would seem to be
required, and the following arrangement
of the groups studied emerges:

Protrogomorpha
Ischyromyoidea
Paramyidae

Ischyromyidae

Ischyromyinae

Cylindrodontinae
Sciuravidae

Protoptychidae (inc. sed.)
Aplodontoidea
Prosciuridae
Aplodontidae

Aplodontinae

Mylagaulinae

Subdivision of the Protrogomorpha into
two superfamilies separates the ancestral
and primitive, derived families from the
relatively more advanced aplodontoids.
Until more forms of protoptychids are
known, I prefer to retain the family within
the Protrogomorpha (Wahlert, 1973) and
have placed it tentatively in the superfamily
Ischyromyoidea. In view of the similarities
between ischyromyids and cylindrodontids,
I propose that the rank of each group be
reduced to subfamily and that they be
combined under the older family name,
Ischyromyidae Alston, 1876. This associ-
ation, based on the foramina, merits further
study from other evidence.

The cranial foramina of aplodontids and
mylagaulids are so alike that I have like-
wise reduced the rank of each group to
subfamily and united them under the older
family name, Aplodontidae Trouessart, 1897
( = Haploodontini Brandt, 1855 ) . The pro-
sciurids have been included in a super-
family with them to express the many
characters unique to the two families.

The cranial foramina and relationships
of sciuromorphous, myomorphous, and
hystricomorphous rodents to tliese protro-
gomorphous groups will be the subject of
future publications; work on the first of
these groups is completed (Wahlert, 1972).

REFERENCES

Black, C. C. 1968a. The Oligocene rodent
Ischijromys and discussion of the family
Ischyromyidae. Ann. Carnegie Mus., 39:
273-305.

. 1968b. The Uintan rodent Mytonomys.

Jour. Paleontol., 42: 853-856.

Cranial Foramina • Wa]ileit 409

BuGGE, J. 1970. The contribution of the stapedial
artery to the cephaHc arterial supply in
muroid rodents. Acta anat., 76: 313-336.

. 1971a. The cephalic arterial system in

mole-rats (Spalacidae), bamboo rats (Rhi-
zomyidae), jumping mice and jerboas (Dipo-
doidea) and dormice (Gliroidea) with special
reference to the systematic classification of
rodents. Acta anat., 79: 165-180.

. 1971b. The cephalic arterial system in

sciuromorphs with special reference to the
systematic classification of rodents. Acta anat.,
80: 336-361.

1971c. The cephalic arterial system in

New and Old World hystricomorphs, and in

bathyergoids, with special reference to the

systematic classification of rodents. Acta

anat., 80: 516-536.
Burke, J. J. 1936. Ardynomys and Desmatolagus

in the North American Oligocene. Ann. Car-
negie Mus., 25: 135-154.
. 1938. A new cylindrodont rodent from

the Oligocene of Montana. Ann. Carnegie

Mus., 27: 255-274.
Dawson, M. R. 1961. The skull of Schiravtis

nitidus, a middle Eocene rodent. Postilla, No.

53: 1-13.
Galbreath, E. C. 1953. A contribution to the

Tertiary geology and paleontology of north-
eastern Colorado. Univ. Kansas Paleontol.

Contrib., Art. 4:1-120.
Greene, E. C. 1935. Anatomy of the rat. Trans.

Amer. Phil. Soc. n.s., 27: 1-370.
Gregory, W. K. 1910. The orders of mammals.

Part II. Genetic relations of the mammalian

orders. Bull. Amer. Mus. Nat. Hist., 27:

105-524.
. 1920. On the structure and relations of

Notharctiis, an American Eocene primate.

Mem. Amer. Mus. Nat. Hist, n.s., 3(pt. 2):

49-243, plates 23-59.
Guthrie, D. A. 1963. The carotid circulation in

the Rodentia. Bull. Mus. Comp. Zool., 128:

455-481.
. 1969. The carotid circulation in Aplo-

dontia. Jour. Mammal., 50: 1-7.
Hill, J. E. 1935. The cranial foramina in rodents.

Jour. Mammal., 16: 121-129.
. 1937. Morphology of the pocket gopher

mammalian genus Thomoniys. Univ. Cali-
fornia Publ. Zool., 42: 81-172.
Hill, W. C. O. 1953. Primates. V. I. Strepsirhini.
Edinburgh: University Press. 798pp.

. 1955. Primates. V. II. Haplorhini.

Edinburgh: University Press. 347pp.
Hyman, L. H. 1942. Comparative Vertebrate
Anatomy. Chicago: Chicago University Press.
544pp.

Kampen, p. N. van. 1905. Die Tympanalgegend

des Siiugetierschadels. Morphol. Jahrb., 34:

.321-722.
Kehoher, G. C. 1970. Lexicon of geologic names

of the United States for 1961-1967. U.S.

Geol. Surv. Bull., 1350: 1-848.
ET AL. 1966. Lexicon of geologic names

of the United States for 1936-1960. U.S.

Geol. Surv. Bull., 1200: 1-4341.
McDowell, S. B. 1958. The Greater Antillean

insectivores. Bull. Amer. Mus. Nat. Hist.,

115: 113-214.
McCrew, p. O. 1941. The Aplodontoidea. Geol.

Ser., Field Mus. Nat. Hist., 9: 1-30.
McKenna, M. C. 1966. Paleontology and the

origin of the Primates. Folia primat., 4:

1-25.
ScHAUB, S. 1958. Simplicidentata, p. 659-818.

In J. Piveteau (ed. ), Traite de Paleontologie,

6(2). Paris: Masson et C".
Szalay, F. S. 1972. Cranial morphology of the

early Tertiary Phenacolemur and its bearing

on primate phylogeny. Amer. Jour. Phys.

Anthrop, 36: 56-76.
Tandler, J. 1899. Zur vergleichenden Anatomic

der Kopfarterien bei den Mammalia. Denk-

schr. Akad. Wiss. Math.- Nat. CI., Wien, 67:

677-784.
. 1901. Zur vergleichenden Anatomic der

Kopfarterien bei den Mammalia. Anat. Hefte,

59: 327-368.

1902. Zur Entwicklungsgeschichte der

Kopfarterien bei den Mammalia. Morphol.
Jahrb., 30: 275-373.

Wahlert, J. H. 1972. The cranial foramina of
protrogomorphous and sciuromorphous ro-
dents; an anatomical and phylogenetic study.
Ph.D. Thesis. Harvard Univ. 230pp.

. 1973. Protoptychus, a hystricomorphous

rodent from the late Eocene of North America.
Breviora, No. 419: 1-14.

Wilson, R. W. 1949a. Additional Eocene rodent
material from Southern California. Carnegie
Inst. Washington Publ., 584: 1-25.

. 1949b. On some White Rixer fossil ro-
dents. Carnegie Inst. Washington Publ., 584:
27-50.

1949c. Early Tertiary rodents of North

America. Carnegie Inst. Washington Publ.,

584: 67-164.
Wood, A. E. 1937. Rodentia, p. 155-269. In W.

B. Scott and G. L. Jepsen, The manunalian

fauna of the White River Oligocene. Trans.

Amer. Phil. Soc. n.s., 28 (pt. 2).
. 1955. A revised classification of the

rodents. Jour. Mammal., 36: 165-187.
. 1959. Eocene radiation and phylogeny

of the rodents. Evolution, 13: 354-361.

410 Bulletin Museum of Comparative Zoology, Vol. 146, No. 8

. 1962. The early Tertiary rodents of the

family Paramyidae. Trans. Amer. Phil. Soc.
n.s., 52 (pt. 1): 1-261.

— . 1965. Small rodents from the early
Eocene Lysite Member, Wind River Forma-
tion of Wyoming. Jour. Paleontol., 39: 124-
134.

. 1970. The early Oligocene rodent

Ardynomys (family Cylindrodontidae ) from
Mongolia and Montana. Amer. Mus. Novitates,
No. 2418: 1-18.

Wood, H. E. et al. 1941. Nomenclature and
correlation of the North American continental
Tertiary. Bull. Geol. Soc. Amer., 52: 1-48.

us ISSN 0027-4100

SuLLetln OF TH

seum

Osteology and Classification of the

Neotropical Characoid Fishes of the Families

Hemiodontidae (Including Anodontinae)

and Parodontidae

TYSON R. ROBERTS

HARVARD UNIVERSITY

CAMBRIDGE, MASSACHUSETTS, U.S.A.

VOLUME 146, NUMBER 9
18 DECEMBER 1974

PUBLICATIONS ISSUED

OR DISTRIBUTED BY THE

MUSEUM OF COMPARATIVE ZOOLOGY

HARVARD UNIVERSITY

Breviora 1952-

BULLETIN 1863-

Memoirs 1864-1938

JoHNSONiA, Department of Mollusks, 1941-

OccASioNAL Papers on Mollusks, 1945-

SPECIAL PUBLICATIONS.

1. Whittington, H. B., and E. D. I. Rolfe (eds.), 1963. Phylogeny and
Evolution of Crustacea. 192 pp.

2. Turner, R. D., 1966. A Survey and Illustrated Catalogue of the Teredini-
dae (Mollusca: Bivalvia). 265 pp.

3. Sprinkle, J., 1973. Morphology and Evolution of Blastozoan Echinoderms.
284 pp.

4. Eaton, R. J. E., 1974. A Flora of Concord. 250 pp.

Other Publications.

Bigelow, H. B., and W. C. Schroeder, 1953. Fishes of the Gulf of Maine.
Reprint.

Brues, C. T., A. L. Melander, and F. M. Carpenter, 1954. Classification of
Insects.

Creighton, W. S., 1950. The Ants of North America. Reprint.

Lyman, C. P., and A. R. Dawe (eds.), 1960. Symposium on Natural
Mammalian Hibernation.

Peters' Check-list of Birds of the World, vols. 2^-7, 9, 10, 12-15.

Proceedings of the New England Zoological Club 1899-1948. (Complete
sets only.)

Publications of the Boston Society of Natural History.

Price list and catalog of MCZ publications may be obtained from Publications
Office, Museum of Comparative Zoology, Harvard University, Cambridge, Massa-
chusetts, 02138, U.S.A.

Plate 1. The species of Anodontinae. Upper: Anodus cf. elongatus from the Orinoco (after Steindachner, 1888;
type specimen of Elopomorphus orinocensis). Middle: Anodus elongatus from Iquitos (Academy of Natural Sci-
ences, Philadelphia 122595, 129.3 mm). Lower: Anodus melanopogon from Iquitos (ANSP 122596, 131.0 mm).

OSTEOLOGY AND CLASSIFICATION OF THE NEOTROPICAL
CHARACOID FISHES OF THE FAMILIES HEMIODONTIDAE
(INCLUDING ANODONTINAE) AND PARODONTIDAE

TYSON R. ROBERTS'

CONTENTS

Introduction 412

Osteological observations - - 415

Terminology 415

Material examined — 415

Hemiodontidae - 416

Hemiodontinae - - — -- 416

Cranium — 416

Otoliths 416

Facial bones - 417

Jaws and jaw teeth — 417

Suspensorium 417

Hyoid and branchial arches — 417

Weberian apparatus - 418

Pectoral girdle 418

Pelvic girdle 418

Vertebral counts — -— 418

Caudal skeleton .._ 418

Bivibranchiinae 418

Cranium — — 418

Otoliths 418

Facial bones - 418

Jaws 419

Jaw teeth -- 419

Suspensorium 419

Hyoid arch 420

Gill arches 420

Gill rakers -. 420

Pharyngeal teeth 421

Weberian apparatus 421

Pectoral girdle — - - — 421

Pelvic girdle -- 422

Vertebral counts - 422

Caudal skeleton 422

Anodontinae - - - 422

Cranium _. 422

Otoliths 422

^ Museum of Comparative Zoology, Cambridge,
Massachusetts, U. S. A. 02138.

Facial bones 422

Jaws 423

Suspensorium ... _ 423

Hyoid and branchial arches 423

Gill rakers -.. .- 423

Weberian apparatus 424

Pectoral girdle 424

Pelvic girdle 424

Vertebral counts 424

Caudal skeleton 424

Parodontidae 424

Cranium 424

Otoliths 425

Facial bones 425

Jaws and jaw teeth 425

Suspensorium 426

Hyoid and branchial arches 426

Weberian apparatus 427

Pectoral girdle 427

Pelvic girdle .- - 427

Vertebral counts 427

Caudal skeleton 427

Discussion 427

Relationship between Hemiodontidae and

Parodontidae 427

Transfer of Anodontinae from Curimatidae

to Hemiodontidae 429

Remarks on lower taxa 432

Genera of Hemiodontinae 432

Genera of Bivibranchiinae — 432

Genera and species of Anodontinae 433

Genera of Parodontidae 433

Conclusion 434

Literature cited 434

Figures 437

Abstract. The family Parodontidae is highly
specialized and readily distinguished from Hemio-
dontidae and all other characoid families. The
three genera it contains differ relati\'ely little from
each other and clearly cannot be separated at a

Bull. Mus. Comp. Zool, 146(9) : 411-472, December, 1974 411

412 Bulletin Museum of Comparative Zoology, Vol. 146, No. 9

level higher than that of genus. In contrast, the
se\'en genera of Hemiodontidae fall into four sub-
families, each characterized by specialized trophic
structures unlike those of any other characoids.
The osteology of Parodontidae and of three hemio-
dontid subfamilies is described and figured, based
on study of all genera except Atomaster and
Pterohcmiodus. Although similarities in the jaws
and suspensoria are suggestive of shared special-
izations, the evidence that the two families are
closely related is inconclusive. The genera and
species of Hemiodontinae are still poorly defined
and need systematic study. The trophic structures
of Bivihranchia, considered as a whole, are more
specialized than those of any other characoid.
Although Argonectes is much more primitive than
Bivihranchia, its specializations clearly indicate
that it belongs in Bivibranchiinae. Bivibranchiinae
are the only characoids with protrusible jaws, and
the mechanism of protrusion is unique. The sub-
family Anodontinae, which is transferred into
Hemiodontidae from Curimatidae, consists of a
single genus with probably only two valid species,
Anodxis elongatus and A. melanopogon.

The genera and subfamilies of the two families
can be classified as follows

Parodontidae

Parodon, Apareiodon, Saccodon (Parodontops

a synonym of Saccodon)
Hemiodontidae

Hemiodontinae

Hemiodus (Hemiodopsis a synonym),
Pterohcmiodus (of doubtful validity)

Bivibranchiinae

Argonectes, Bivihranchia, Atomaster (prob-
ably a synonym of Bivihranchia)

Micromischodontinae
Micromischodiis

Anodontinae

Anodus { Eigenmannina a synonym)

INTRODUCTION

The poorly known characoid families
Hemiodontidae and Parodontidae include
some of the most interesting fishes in the
fresh waters of South America. Hemio-
dontidae range throughout the lowlands of
the Guianas and the basins of the Orinoco
and Amazon, and southwards to include the
Plata basin. Mostly six inches to a foot long
when adult, they form small schools in
lakes and large rivers. All are swift swim-
mers with streamlined bodies. Seven genera
and about 27 species are currently recog-
nized, belonging to four subfamilies:

Hemiodontinae, Bivibranchiinae, Anodon-
tinae, and Micromischodontinae. The best
known genera are Hemiodus Miiller 1842,
Bivihranchia Eigenmann 1912, and Anodus
Spix 1829 (in Agassiz and Spix, 1829).
Bivihranchia and its poorly known relatives
Argonectes Bohlke and Myers 1956 and
Atomaster Eigenmann and Myers 1927 are
the only characoids with truly protrusible
upper jaws. The greatly reduced premaxil-
laries are freed from the cranium, and the
maxillaries and anterior end of the suspen-
sorium are very loosely bound to it. When
the lower jaw is depressed, the entire upper
jaw and the palatine-ectopterygoid portion
of the suspensorium move slightly forward
and strongly downward and away from the
cranium. The mouth is apparently able to
close when the upper jaw is protruded. The
highly specialized mechanism of protrusion
is probably unlike that in any other teleosts.
Anodus, hitherto assigned to Curimatidae,
is one of the few Amazonian fishes with
pharyngeal stioictures apparently special-
ized for filter feeding on plankton (Roberts,
1972: 138-40). It has edentulous jaws and
far more gill rakers than any other chara-
coid. The gill arches are exceptionally
elongate, and the number of gillrakers
undergoes a considerable increase with
growth of individuals.

Parodontidae occur in the eastern half
of Panama, on the Pacific and Caribbean
coasts of Colombia and the Pacific coast of
Ecuador, in the Orinoco and Amazon
basins, in the Guianas, and southwards to
include the Plata basin. There are about
25 species, mostly four to six inches long
when adult. All are mountain stream fishes,
typically found in swift streams at altitudes
from 100 to over 1000 meters. Several
species occur in headwaters on the pe-
riphery of the Amazon basin, but no speci-
mens have been reported from the Middle
or Lower Amazon. Their expanded and
flattened pectoral fins are presumably
adapted to help maintain position in swift
current. Nuptial tubercles occur in several
species (Wiley and Collette, 1970). The

Hemiodontidae and Parodontidae • Roberts 413

pedicellate teeth and the apparently very
mobile (but nonprotrusible ) premaxillaries
are highly specialized for browsing on
algae. Three genera are currently recog-
nized: Tarodon Valenciennes 1849 {in
Cuvier and Valenciennes, 1849), Saccodon
Kner and Steindachner 1863, and Aparcio-
don Eigenmann 1916. Saccodon, restricted
to coastal basins in Panama, Colombia, and
Ecuador, is of exceptional interest to evo-
lutionary biologists because populations of
the two principal species exhibit extraordi-
nary poK'morphism with respect to trophic
structures (Roberts, 1974). Greater di-
versity of premaxillary tooth morphology
occurs in a single population of Saccodon
dariensis than in all species of Parodon and
Apareiodon combined.

The pedicellate, multicuspid dentition in
the nonprotrusible upper jaw of Hemiodus
is very similar to that in many Parodonti-
dae, and parodontids with relatively termi-
nal mouths such as Saccodon terniinalis
(cf. Roberts, 1974) and some species of
Apareiodon bear a striking resemblance to
Hemiodus. Most students of characoids,
including Regan (1911), Eigenmann (1912),
Fowler (1950), Gery (1959), and Greenwood
et al. (1966), either stated that Hemiodonti-
dae and Parodontidae are closely related or
else at least provisionally accepted this
hypothesis by placing them either in the
same family or subfamily or next to each
other in a classification.

The scant literature on the osteology of
Hemiodontidae and Parodontidae can be
rapidly reviewed. Regan (1911) gave an
extremely brief account of the jaws and
suspensorium of Hemiodontidae and Paro-
dontidae. Travassos (1951; 1952) gave
extensively illustrated accounts of the
opercular series, jaws and dentition of
Parodontidae. The highly specialized re-
placement tooth trenches characteristic of
the family can be seen in many of his
figures. Gery (1959) briefly compared the
superficial skull bones of Parodontidae and
Hemiodontidae. Gery ( 1963a ) described
the jaws and suspensorium of Bivibranchia;

unfortunately, there are several errors in
this account. Lastly there is my account of
the osteology of Micromischodus (Roberts,
1971), a new genus from the rio Negro and
Middle Amazon. These six references con-
stitute all that has been published on the
osteology of the two families.

The present account deals with the
osteology of Parodontidae and of the
hemiodontid subfamilies Hemiodontinae,
Bivibranchiinae and Anodontinae. The
osteology of the one remaining hemiodontid
subfamily, the monotypic Micromischo-
dontinae, has already been treated (Rob-
erts, 1971), but may be reviewed briefly
here. Micromischodus is the only hemio-
dontid in which the lower jaws bear teeth
throughout life. In all others the lower
jaw is either toothless throughout life or
becomes toothless early in life. The special-
ized unicuspid pedicellate dentition on the
jaws and pharyngeal toothplates of Micro-
mischodus is unique. My original account
should have mentioned that Micromischo-
dus, like other hemiodontids, has three
openings into each posttcmporal fossa.
Excepting the teeth and elongated slender
pharyngeal toothplates, the osteology of
Micromiscliodus is very similar to that of
Hemiodus.

Following the general osteological ac-
counts, diagnoses are presented for all of
the higher taxa: Hemiodontidae, Hemio-
dontinae, Bivibranchiinae, Anodontinae,
and Parodontidae. The possible relationship
of Hemiodontidae and Parodontidae to
each other is reviewed in the Discussion
section. The question of relationships of
Parodontidae and Hemiodontidae to other
characoids is left in abeyance, pending
osteological and morphological investi-
gations of other groups including Chilo-
dontidae and Anostomidae. I have already
pointed out that the relationship of the
"semifunctional" preformed replacement
teeth to the functional unicuspid pedicel-
late teeth on the pharyngeal toothplates of
Micromischodus may parallel an e\olution-
ary stage on the way to the multicuspid

414 Bulletin Museum of Comparative Zoology, Vol. 146, No. 9

pharyngeal teeth characteristic of Chilo-
dontidae and Anostomidae (Roberts, 1971).
The most important concUision about classi-
fication of characoids reached in this paper
is that Anodontinae belong to Hemiodonti-
dae rather than to Curimatidae.

The biology of Hemiodontidae and
Parodontidae is largely unknown. Knoppel
(1972) gave an account of the stomach
contents of several species of Parodontidae
and Hemiodontidae (Hemiodontinae and
Bivibranchiinae ) . His findings may be
summarized as follows: Parodontidae and
Hemiodontinae ingested mainly sand, detri-
tus, algae (diatoms mentioned for Parodon,
filamentous algae for Hemiodus), and
higher plants. Of the ten species of
Hemiodontinae examined, only Hemiodus
immacidatus had fed on chironomid larvae.
The stomach contents of this species also
included ephemeropteran larvae but no
plant material of any kind. In three species
of Bivibranchia Knoppel found sand, detri-
tus, and chironomid larvae. One of the
species also contained algae (kind not
specified) and a copepod (Harpacticidae?).
Stomach contents of Micromischodus (Rob-
erts, 1971: 8) include an assortment of
bottom material or detritus, some of which
may be droppings of other fishes, and
many small insects, especially larval
Diptera and a corixid. There is no pub-
lished information on the food items of
Argonectes and Anodus. The behavior and
reproductive biology of Parodontidae and
Hemiodontidae have not been studied.

This paper provides further documenta-
tion for the hypothesis that diversification
of feeding structures has played a major
role in the evolution of higher taxa in the
suborder Characoidei (Roberts, 1967; 1971;
1973). An attempt to compare the amount
of trophic specialization among various
characoids seems worthwhile. Regardless
of whether the conical tooth morphology
and the arrangement of the teeth in
Salminus are correctly interpreted as primi-
tive (Roberts, 1969), the feeding structures
of this genus are clearly generalized for

Table 1. Amount of trophic specialization
among various characoid taxa.

aj u

•a

a
o

o
a
>.
U

a
o
■a
o

•a
.S
S

Jaws

11112 2 2

Jaw teeth

2 12 2 2 2 2

Replacement teeth

10 112 11

Lips

0 0 0 0 0 0 0

Palate

0 0 2 0 0 0 0

Suspensorium

0 0 0 0 112

Pharyngeal arches

0 2 0 10 11

Gill rakers

12 0 0 111

Pharyngeal teeth

0 0 0 2 0 0 0

Pharyngeal epi-

thelium

1-2

0 0 0 0 0 0 0

Total

4-5

5 6 6 7 8 8 9

characoids. Using Salminus as a guideline,
I have judged the character states of ten
structures involved in the mechanics of
feeding as they occur in a variety of
characoid taxa. The structures are scored
either "0" (generalized condition), "1"
( some specialization ) , or "2" ( high special-
ization). The individual scores for the ten
structures are then totalled to give the
amount of trophic specialization in each
taxon. Subjectivit}' is probably inevitable
in any methods that could be devised to
determine the amount of specialization
between organisms as unalike as the pro-
verbial apples and oranges, and it has not
been eliminated here. The method has
been arbitrarily designed, and evaluation of
the character states is still relatively sub-
jective. On the other hand, the steps have
been broken down, so that they can be
followed by other workers, subjected to re-
evaluation and further analysis, extended to
additional taxa, and perhaps improved
upon. The data for a number of taxa is
presented in Table 1. Note that the range
of theoretically possible total scores is from
0 to 20. Salminus, generalized in all re-
spects, would automatically score 0. Of

Hemiodontidae and Parodontidae • Roberts 415

the groups tabulated, Bivibranchia is most
specialized, with a score of 16. I do not
know of any other characoid group that
would score this high. The group with the
next most specialized feeding structures is
probably the Prochilodontidae, which
scores 13 ( see Roberts, 1973, for an account
of the feeding structures of this iliophy-
tophagous family). Argonectes, which shares
several trophic specializations with Bivi-
branchia, scores only 9. The piscivorous
family Cynodontidae also scores 9. Most
piscivorous groups, including the Serrasal-
minae (piranhas), score low, mainly be-
cause they tend to have generalized
pharyngeal trophic structures. The ilio-
phagous Curimatidae are generalized in
most features and score very low. Of groups
not tabulated, Anostomidae and Chilodonti-
dae have perhaps the most specialized
feeding structures and would probably
score around 8 or 10.

Acknowledgments. It is a pleasure to
express my thanks for specimens received
for this study from the following individu-
als and their respective institutions: Loren
P. Woods of the Field Museum of Natural
History; James E. Bohlke of the Philadel-
phia Academy of Natural Sciences; William
Fink and Stanley H. Weitzman of the
Smithsonian Institution; Jacques Gery;
Warren C. Freihofer of the California
Academy of Sciences; and Heraldo A.
Britski of the Museu de Zoologia of the
Universidade de Sao Paulo. I also thank
Stanley Weitzman for helpful comments on
the manuscript.

OSTEOLOGICAL OBSERVATIONS

Terminology

The osteological terminology used in this
account is based primarily on that in Weitz-
man (1962), with three main exceptions,
namely that for gill arches, precaudal and
caudal vertebrae, and caudal skeleton. The
gill arch terminology followed is that
recommended by Nelson ( 1969 ) . It is
essential to define gill arch elements pre-

cisely and to standardize their terminology
if the information they represent is to be
useful in phyletic analysis. Nelson empha-
sized the distinction between endoskeletal
elements and dermal toothplatcs, an im-
portant clarification. An adequate nomen-
clature for the bony elements collectively
known as basihyals and basibranchials has
yet to be developed. In characoids these
apparently include dermal as well as endo-
skeletal elements, and yet the dermal ele-
ments in most cases are not toothplatcs.
In many characoids the basihyal is a com-
pound element, consisting of an anterior
(dermal?) portion and a posterior endo-
skeletal portion. Concerning the vertebrae,
caudal vertebrae are herein defined as the
first vertebra bearing a complete hemal
canal and all of the vertebrae succeeding
it, counting the compound ural centrum
as one. The precaudal count includes the
four vertebrae bearing the Weberian ap-
paratus. The terminology for the caudal
skeleton is that proposed by Monod (1969).

Material Examined

This account is based on alizarin prep-
arations of the following specimens of
Hemiodontidae and Parodontidae (MCZ
= Museum of Comparative Zoology, Har-
vard; MZUSP = Museu de Zoologia da Uni-
versidade de Sao Paulo; CAS = California
Academy of Sciences; USNM = Smith-
sonian Institution; FMNH = Field Museum
of Natural History, Chicago ) :

Anodus elongattis, MCZ 20671. One speci-
men, 180 mm. Lago Alexo, Amazonas,
Brazil. Thayer expedition.

Anodus melanopogon, MZUSP 5959. Three
specimens, 54.0 to 55.0 mm. Mouth of the
rio Purus, Amazonas, Brazil. Expedigao
Permanente da Amazonia, 1-5 April 1967.

Apareiodon affinis, FMNH 71228. Three
specimens, 26.5 to 46.8 mm. Rio Uruguay
at Soriana, south of Dolores. C. C. San-
born, 25-31 January 1927.

416 Bulletin Museum of Comparative Zoology, Vol. 146, No. 9

Apareiodon itapicuruensis, FMNH 56991.
Two specimens, 19.8 and 21.0 mm. Rio
Itapicuru at Queimadas, Bahia, Brazil. J.
D. Haseman, 2 March 1908.

Argonectes longiceps, MCZ 20635. One
specimen, 198 mm. Rio Xingu, Amazonas,
Brazil. Thayer Expedition.

Bivihranchia protractila, CAS(SU) 48847.
One specimen, 39.9 mm. Rio Cassiquiare
at mouth of Rio Curamoni, Venezuela. C.
Ternetz, 7 March 1925; CAS(SU) 48608.
One specimen, 115.2 mm. Rio Negro at
Camanaos rapids, Brazil. C. Ternetz, 23
January 1925.

Hemiodtis quadrimaculatus, MCZ 29926.
One specimen, 35.0 mm. Lower Potaro
River at Tumatumari, Guyana. C. H.
Eigenmann, 1908.

Hemiodtis semitaeniatus, MCZ 49072. Two
specimens, 54.4 and 55.8 mm. Jaure ranch,
North Savannah, Rupununi, Guyana. C.
Hopkins, 18 May 1971.

Parodon caliensis, MCZ 47682. One speci-
men, 63.5 mm. Probable locality data: small
streams of Sierra Azul, easternmost range
of Andes [Peru?]. O. Barton, December
1945.

Farodon guijanemis, MCZ 48969. One para-
type, 38.5 mm. Upper Mana River at Saut-
Fini, French Guiana. Lelong, 25 November
1957.

Saccodon dariensis (Meek and Hildebrand),
USNM 208503. Two specimens of dental
morph I, 83.4 and 89.4 mm, and two speci-
mens of dental moiph IV, 97.2 and 100.7
mm. Rio Sabalo, Bayano basin, one mile
above Naragandi, Panama. Battelle Me-
morial Institute NW Lab., 21 March 1967.

Saccodon ivagneri Kner and Steindachner,
MCZ 48745. Four specimens of dental
morph I, 45.5 to 57.2 mm; and MCZ 48746.
Five specimens of dental morph IV, 52.0
to 103.3 mm. Arroyo Bambine (tributary
of Rio Cristal), Guayas basin, at Montalvo.
T. R. Roberts, R. Gilbert, F. Silva M., 6
November 1971.

The only genera not represented in this
list are the hemiodontids Pterohemiodus
Fowler 1940 (closely related to, and perhaps
indistinguishable from, Hemiodus), and
Atomaster (a close relative and possible
congener of Bivihranchia). Alizarin prepara-
tions of the following Curimatidae have
been examined: Acuticurimata macrops,
Cruxentina nasa, Curimata cyprinoides,
Ctirimatorhis oceUatus, Gasterotomiis latior,
Potaniorhina pristigaster, Semitapicis pJani-
rostris, and Suprasinelichthys laticeps.

Hemiodontidae

Hemiodontinae

Cranium {Figures 1-5). Frontoparietal
fontanel complete; frontal portion narrow
anteriorly, widening in front of epiphyseal
bar; parietal portion uniformly wide. Eth-
moid narrow, its head with a dorsomedian
projection and lateral projections (one on
either side) articulating with premaxillaries.
Supraoccipital crest flattened dorsally; tip
of supraoccipital spine rounded in dorsal
view, extending to end of cranium. Post-
temporal fossae well developed, with three
openings. Lateral opening, largest in size,
bordered entirely by epiotic and pterotic;
dorsomedial opening, bordered by epiotic,
parietal and supraoccipital; ventroposterior
opening, lying in posteromedian portion of
epiotic. Lateral and dorsomedial openings
separated from each other only by a narrow
bony bridge formed by epiotic. Dilator
groove well developed, frontal participat-
ing in its formation.

Vomer with a concave lamellar com-
ponent ventrally and two separately formed
lamellar components dorsolaterally. Lateral
ethmoid large, with anterior processes
articulating with vomer. Rhinosphenoid
present. Interorbital septum large, rhino-
sphenoid and orbitosphenoid widely sepa-
rated from parasphenoid. Posterior portion
of parasphenoid cleft for most of its length.
Subtemporal fossa well developed.

Otoliths. Otoliths of generalized chara-
coid morphology, similar in size and shape

Hemiodontidae and Parodontidae • Roberts 417

to those in Bnjcon ( cf. fig. 7 in Weitzman,
1962).

Facial bones {Fi<^ure 6). Facial bones
present no unusual features. Circumorbital
series complete, with a supraorbital, antor-
bital, and six moderately large infraorbital
bones bearing infraorbital canal of cephalic
laterosensory system. Infraorbital canal
with posteriorly directed branches on infra-
orbital bones 3 and 4. Infraorbital bones
flat, ventral portions of infraorbitals 1-4 not
strongly curving beneath head. Infraorbital
1 not greatly enlarged. Size and shape of
infraorbitals as in Micromischodus: infra-
orbital 3 largest, infraorbital 2 relatively
slender. Infraorbital six contacting or
closely approaching supraorbital.

Opercular bones of generalized characoid
morphology, apparently without conditions
of phyletic significance at familial or
generic levels. Lateral surface of opercle
smooth. Dorsoanterior corner of opercle
strongly notched.

Jaws and jaw teeth {Figures 6-10). Pre-
maxillary, with a short ascending process,
loosely joined to maxillary; maxillary larger
than premaxillary, its proximal end with a
median knoblike projection. In H. semi-
taeniatus descending limb of premaxillary
lies medial to anterior edge of maxillary;
in H. quadrimacidatus descending limb of
premaxillaiy much shorter, scarcely or not
at all extending medially to maxillary.
Upper jaw with a single row of 10-15 or
more loosely articulated (movably at-
tached) multicuspid teeth. Tooth crown
rounded, cusps in a straight row, number
of cusps increasing with age. In H. semi-
taeniattis a maximum of two replacement
teeth for each functional tooth; in H.
quadrimaculatiis up to four replacement
teeth per functional tooth. Inner surface
of premaxillaries without bony partitions
separating rows of replacement teeth. In
H. semitaeniatus about eight teeth articu-
lated with premaxillary and seven with
maxillary; in H. quadrimacidatus about
four or five teeth with premaxillary and
five with maxillary. In H. quadrimacuJatus

premaxillary curving around replacement
tooth rows more strongly than in H. semi-
taeniatus.

Lower jaw toothless, elongate. Portion
of dentaries meeting at symphysis slender;
middle portion of dentary moderately
elevated. A single fenestra near anterior
edge of dentary. Coronomeckelian bone
low set, lying well below dorsal margin of
articular bone.

Suspemorium (Figures 6-8, 11). Palatine
straight, immovably attached to ectoptery-
goid, its anterior end moderately expanded.
Ectopterygoid loosely (movably) attached
to quadrate. Mesopterygoid loosely at-
tached to quadrate and metapterygoid;
metapterygoid loosely attached to hyo-
mandibular. Anterior margin of metaptery-
goid separated by a gap from dorsoposterior
margin of quadrate. Metapterygoid-quad-
rate fenestra relatively small. Anteroventral
limb of metapterygoid (entering border
of metapterygoid-quadrate fenestra) de-
veloped more strongly and anterior margin
of hyomandibular more oblique in H.
quadrimaculatus than in H. semitaeniatus
(compare Figs. 7 and 8).

Hijoid and branchial arches {Figures
12-14). Hyoid arch without unusual
features, of generalized characoid morphol-
ogy. Four or five branchiostegal rays ( four
in H. semitaeniatus, five in H. quadri-
maculatus). Proximal portion of next to
last branchiostegal ray not enlarged (cf.
Argonectes and Micromischodus) . Basihyal
slender, consisting of two ossifications, an
anterior (dermal) bone and a posterior
( endoskeletal ) bone. Urohyal fairly long,
its lateral wings moderately depressed and
its dorsomedian lamina moderately high
and extending slightly posterior to lateral
wings.

Branchial arches including cerato-
branchial 5 of relatively generalized
characoid morphology, endoskeletal ele-
ments excepting infrapharyngobranchials
and basibranchials similar in size and shape
to those in Brycon. Three basibranchials;
basibranchial 1 minute, basibranchials 2

418 Bulletin Museum of Comparative Zoologij, Vol 146, No. 9

and 3 very long. Three infrapharyngo-
branchials infrapharyngobranchials 2 and 3
greatly enlarged.

Gill rakers variable in form, often with
large ctenii ( not true teeth ) ( Fig. 14 ) , and
relatively numerous: 45 gill rakers on first
gill arch of a 35.0-mm H. quadrimacuJatus,
58 in a 55.8-mm H. semitaeniatus (Fig. 13),
70 in a 235-mm H. 7nicroIepis and 75 in a
113.5-mm H. notatus.

Pharyngeal dentition consisting of nu-
merous small conical teeth, either non-
pedicellate or with very short pedicels,
confined to three or four pairs of upper
pharyngeal toothplates and a single pair
of lower pharyngeal toothplates. Upper
pharyngeal toothplates on ventral surface
of infrapharyngobranchials 2 and 3 and
loosely associated with median ends of
epibranchials 3 and 4.

Weberian apparatus (Figure 15). Webe-
rian apparatus, very similar to that of
Micromischodus and Brycon meeki, ap-
parently without features of phyletic sig-
nificance at generic or familial levels.

Pectoral girdle (Figure 16). Pectoral
girdle of generalized characoid morphol-
ogy, apparently lacking in features of
phyletic significance at generic or even
family levels. Three postcleithra, post-
cleithrum 3 slender for most of its length
but with a lamellar portion near its proxi-
mal end (as in Micromischodus, Brycon
dentex and many other characoids).

Pelvic girdle (Figure 17). Pelvic bone
simple anteriorly; ischiac process relatively
large. Four radials associated with each
pelvic bone; innermost radial comma-
shaped. Pelvic fin with 11 rays. Pelvic
girdle evidently without features useful for
phyletic analysis at generic or family levels.

Vertebral counts. Total vertebral and
precaudal plus caudal vertebral counts in
two specimens of H. semitaeniatus 40
(26 + 14) and 41 (26 + 15).

Caudal skeleton (Figure 18). Caudal
skeleton with two or three epurals and two
uroneurals. Hypurals 3-6 separate. Hypural
2 fused to hypural 1 and not to complex

ural centrum. No bony projections com-
parable to parhypurapophyses or hypur-
apophyses. Principal caudal rays 10 + 9.

BlVIBRANCHIINAE

Cranium (Figures 19-24). Cranium of
Argonectes similar in most respects to that
in Hemiodontinae and Micromischodonti-
nae. Anterior end of ethmoid of distinctive
shape: narrow, slightly up-turned, with a
dorsomedian groove terminating as a notch
in tip of ethmoid (Fig. 19). Shapes of
frontoparietal fontanel and of three open-
ings into each posttemporal fossa as in
Hetniodus and Micromischodus. Pterotic
spine pronounced, epiotic spine moderate.
Dilator fossa well developed. Rhinosphe-
noid moderately large and relatively simple
in shape (Fig. 20). Posterior portion of
parasphenoid cleft for virtually its entire
length. Subtemporal fossa well developed.
Hyomandibular fossa formed by contiguous
surfaces of sphenotic and pterotic (Fig. 21).

Cranium of Bivibranchia broadly similar
to that of Argonectes but with some notable
departures: anterior end of ethmoid de-
pressed, without a dorsomedian groove but
with anterolateral projections; vomer rela-
tively broad; rhinosphenoid greatly en-
larged and of distinctive shape (Fig. 23);
pterotic and epiotic spines enlarged; an-
terior portion of parasphenoid immediately
in front of attachment of gill arches with a
strong, ventrally projecting keel (absent in
Argonectes). Dilator fossa less prominent
than in Argonectes. Condition of fronto-
parietal fontanel, openings into post-
temporal fossae, and hyomandibular fossa
similar to Argonectes. In Argonectes and
Bivibranchia ventromedian opening into
posttemporal fossa partially closed by thin
irregular bridges of bone from epiotic.

Otoliths. Otoliths of Bivibranchiinae not
examined.

Facial bones (Figures 25-29). In Argo-
nectes facial bones departing only in minor
details of size and shape from those of
Hemiodus: infraorbitals 1 and 4 relatively
small. Infraorbital 6 contacting supraorbital.

Hemiodontidae and Parodontidae • Robciis 419

In Bivibranchia facial bones differ in size
and shape from those in Argonectes and
Hemiodus: infraorbital 2 enlarged; infra-
orbital 1 relati\'ely elongate; infraorbital 6
widely separated from supraorbital. Dorsal
margin of opercle straight, without in-
dentation characteristic of other hemio-
dontids.

Jaws (Figures 25, 27-29). Lower jaw
similar in shape to that of Hemiodontinae;
Argonectes and Bivibranchia differ from
Hemiodus in having a posteriorly directed
knob on symphyseal process of dentary and
coronomeckelian bone elevated in position.
In Bivibranchia coronomeckelian more ele-
vated than in Argonectes, forming part of
dorsal margin of lower jaw (Fig. 28);
articular bone more elongate in Bivi-
branchia, its anterior end loosely fitted into
dentary (not directly applied to internal
surface of dentary), its posterior end with
a knoblike projection that fits into a recess
in median surface of quadrate when lower
jaw is depressed.

Upper jaw highly specialized. Premaxil-
lary minute. Maxillary slender and highly
curved, more so in Bivibranchia than in
Argonectes.

Jaw teeth [Figures 27-29). Jaw teeth
highly specialized, diagnostic of subfamily.
A single row of eight or nine minute, tri-
cuspid, pedicellate teeth loosely attached
to premaxillary and concave proximal por-
tion of maxillary. Three or four preformed
replacement teeth for each functional tooth
in Argonectes, one or two in Bivibranchia.
Lower jaw edentulous.

Suspensorium (Figures 27-28). In
Argonectes, palatine, mesopterygoid and
hyomandibular relatively generalized in
morphology, not greatly different from
these bones in other characoids. Mesoptery-
goid contacting and movably articulated
with ectopterygoid and metapterygoid, not
contacting palatine or quadrate. Ectoptery-
goid, quadrate and metapterygoid highly
specialized: proximal end of ectopterygoid
with a slender rounded head loosely fitting
into a deep recess or socket in anterior

margin of quadrate just above quadrato-
mandibular joint. Metapterygoid movably
articulated with mesopterygoid and hyo-
mandibular, as in Hemiodus. Metaptery-
goid-quadratc joint more extensive (and
less movable?) than in Hemiodus, meta-
pterygoid quadrate foramen relatively
small. Ventroposterior edge of metaptery-
goid contacting dorsal margin of symplectic
distal end.

In Bivibranchia every bone in suspen-
sorium highly specialized in shape and in
relationships to other bones. Palatine bone
S-shaped, its anterior end slender, forming
a concavity into which elongated distal end
of maxillary rides. Proximal end of ecto-
pterygoid slender, forming the finger of a
highly specialized, loose-fitting "finger and
ring" joint with quadrate bone. Anterior
margin of quadrate dorsal to quadrato-
mandibular joint with a completely open
ring of bone inside of which rides proximal
end of ectopterygoid. Median surface of
quadrate ventral to quadrato-mandibular
joint with a well-developed concavit\' into
which free-ending posterior projection from
articular bone slides when lower jaw is
depressed. Ectopterygoid of a highly modi-
fied and characteristic shape, much more
specialized than in Argonectes. Ectoptery-
goid temiinating anteriorly in a slender
process to which ectopterygoid-vomerine
ligament attaches. Ectopterygoid with an
extensive movable articulation with quad-
rate, contact with metapterygoid reduced
compared to Argonectes, no direct contact
with ectopterygoid. Metapterygoid-quad-
rate articulation even more extensive than
in Argonectes, rectilinear. Metapterygoid-
quadrate foramen reduced in size, as in
Argonectes, but metapterygoid and distal
end of symplectic separated by a distinct
gap. Metapterygoid-hyomandibular articu-
lation specialized; hyomandibular with a
strutlike process immovably joined to
dorsoposterior margin of metapterygoid.
Metapterygoid strengthened by a well-
developed longitudinal keel on its median

420 Bulletin Museum of Comparative Zoology, Vol. 146, No. 9

surface extending anterior to and immedi-
ately below strut from hyomandibular.

Htjoid arch {Figures 30, 34-36). In
Argonectes basihyal slender and moderately
elongate, composed of separate anterior
and posterior ossifications firmly sutured
together; anterior ossification about twice
as long as posterior one. Dorsal and ventral
hypohyals closely joined. Hyoid bar per-
haps slightly shorter than in Hemiodus
but otherwise not strongly modified. Five
branchiostegal rays, four on ceratohyal and
one on epihyal. Proximal portion of
branchiostegal ray 4 expanded so that it
reaches dorsal edge of ceratohyal, as in
Micromischodus. Urohyal moderately long,
its lateral wings relatively narrow and only
slightly depressed (ventral surface of
urohyal moderately concave). Dorsomedian
lamina moderately high, extending to
posterior end of urohyal. Head of urohyal
(attaching to ventral surface of ventral
hypohyals) moderately wide, "neck" im-
mediately behind head narrowed.

In Bivibranchia basihyal slender and
moderately elongate, composed of separate
anterior and posterior ossifications; anterior
ossification about one-third as long as
posterior one, and firmly attached to an-
terior end of posterior ossification, but not
sutured to it as in Argonectes. Dorsal and
ventral hypohyals closely joined. Hyoid bar
slightly shorter than in Hemiodus, ex-
panded dorsally and ventrally at junction
of ceratohyal and epihyal; edge of dorsal
expansion sharp; middle of hyoid bar
thickened. Five branchiostegal rays, four
on ceratohyal and one on epihyal. Branchio-
stegal ray 4 with its proximal portion only
slightly enlarged. Urohyal specialized, its
lateral wings wider and more depressed
than in Argonectes, dorsomedian lamina
fomiing a strong low-lying ridge behind
head of urohyal but increasingly weak on
posterior half of urohyal until it is almost
absent. Head and neck of urohyal wider
than in Argonectes.

Note. In Argonectes branchiostegal
membranes relatively free from isthmus.

gill openings extending forward almost to
point of attachment of branchiostegal rays
to hyoid bar. In Bivibranchia branchi-
ostegal membranes broadly joined to
isthmus below middle of eye, gill openings
extending forward only to posterior ends of
branchiostegal rays 1 and 2.

Gill arches {Figures 31-37). In Argo-
nectes gill arches relatively generalized for
characoids, similar to those in Hemiodus,
without any obvious specializations. Ap-
parently three basibranchials; basibranchial

1 minute. Small accessory ossifications (der-
mal bones?) between basibranchials 1 and

2 and basibranchials 2 and 3. Three hypo-
branchials. Four infrapharyngobranchials
(a separate ossification dorsal to posterior-
most upper pharyngeal toothplates in-
terpreted as infrapharyngobranchial 4).
Epibranchials 1-3 and ceratobranchials 1^
morphologically generalized, their ad-
pharyngeal surfaces simply rounded (as in
Hemiodus). Epibranchial 4 with a moder-
ately large abpharyngeal laminar extension
( a widespread feature in characoids ) .

In Bivibranchia gill arches highly
specialized. Three basibranchials; basi-
branchial 1 much larger than in Argonectes
and most other hemiodontids, almost as
long as basibranchials 2 and 3. Three hypo-
branchials. Hypobranchial 1 specialized:
elongate and enlarged, peculiarly bowed
inwards, its anterior end distinctly ex-
panded. Hypobranchial 2 less modified
than hypobranchial 1, but also relatively
elongate. Three epibranchials. Epibranchi-
als 1-3 and ceratobranchials 1—4 highly
specialized, their adphaiyngeal surfaces
with a greatly expanded, thin bony lamina.
These laminae completely separate gill
rakers on leading edge of each gill arch
from those on trailing edge, and are
covered by highly specialized epithelium.

Gill rakers {Figures 34, 36). In Argonectes
gill rakers present on leading and trailing
edges of gill arches 1-4 and on leading
edge of gill arch 5 (ceratobranchial 5). A
177-mm specimen has 14 + 1 + 8 rakers on
arch 1, 8 + 1 + 9 on arch 2, 8+1 + 9 on

Hemiodontidae and Parodontidae • Roberts 421

arch 3, 7+1 + 6 on arch 4, and 8 on
arch 5.

In Bivibranchia gill rakers present on
leading and trailing edges of most of gill
arches 1^ and on leading edge of arch 5.
Gill rakers absent on basibranchials, infra-
pharyngobranchials, hypobranchials 1 and
2, and anterior half of ceratobranchial 1.
Gill rakers slender and moderately elongate,
only tripodlikc basal portion and sometimes
a small part of shaft absorbing alizarin. A
39.9-mm specimen has 12 + 1 + 11 rakers
on arch 1, 12 + 1 + 15 on arch 2, 13+1
+ 17 on arch 3, 10 + 1 + 17 on arch 4, and
23 on arch 5. A 115.2-mm specimen has
12+1 + 10 on arch 1, 14+1 + 17 on arch
2, 10 + 1 + 19 on arch 3, 12 + 20 on arch
4, and 23 on arch 5. (Note: in smallest
rakers basal portion ossifies in two places;
cf. rakers on epibranchials 3 and 4 in Fig.
36).

Remarks. In Bivibranchia the adpharyn-
geal epithelium is thrown into a uniform
series of prominent, finely papillose ridges.
One such ridge extends from the base of
each gill raker on the leading edge of the
gill arches to the base of the gill raker at
the same position on the trailing edge. Thus
the number of ridges corresponds to that
of rakers. The ridges of successive arches
interdigitate when the branchial apparatus
is contracted. Eigenmann (1912: 258) re-
ferred to these epithelial structures as
"broad laminae with papillated ridges."
The ridges are well shown in the photo-
graphs of the gill arches accompanying
Eigenmann's original description (op. cit.,
pi. 33, figs. 2-4). In these figures the gill
rakers themselves are poorly or not at all
discernible except on the lower portion of
the first gill arch (fig. 2a).

The functional anatomy of the hyoid and
branchial arches should be examined in live
Bivibranchia. To judge from alizarin prep-
arations, the hyoid arch may move in-
dependently of the branchial arches, using
basibranchial 1 as a pivot. When the hyoid
arch is forward, basibranchial 1 lies hori-
zontally. When the hyoid arch is pushed

backward, basibranchial 1 assumes a verti-
cal orientation with its posterior end
directed ventrally (Figs. 34, 35). This
presumably would effect the pharyngeal
pumping mechanism.

Pharyniical teeth (Figures 31-34, 36-37).
In Ariionectes pharyngeal toothplates and
pharyngeal dentition generalized. Upper
pharyngeal dentition occurs on four pairs
of separate toothplates that are intimately
associated with ventral surface of infra-
phaiyngobranchials 2, 3 and 4 and epi-
branchial 4. Lower pharyngeal dentition
restricted to a pair of toothplates of fifth
ceratobranchials. Dentition on all tooth-
plates consisting of dense patches of
moderate -sized simple conical teeth.

In Bivibranchia pharyngeal dentition
highly specialized. Upper pharyngeal denti-
tion restricted to a single pair of toothplates
loosely associated with medially directed
processes of infrapharyngobranchial 3 and
epibranchials 3 and 4 (infrapharyngobran-
chial 4 absent). Upper toothplate of each
side bearing two nearly coextensive, close-
set rows of eight to ten tricuspid, pedicel-
late teeth (rows more regular in 115.2-mm
specimen than in 39.9-mm specimen).
Lower pharyngeal dentition restricted to a
single pair of toothplates intimately associ-
ated with dorsal surface of ceratobranchial
5. Lower pharyngeal dentition consisting
of three or four irregular rows of pedicel-
late teeth. In 39.9-mm specimen only
anteriomiost tooth row consisting of tri-
cuspid teeth, remaining teeth unicuspid.
In 115.2-mm specimen first two or three
tooth rows almost entirely tricuspid, last
row unicuspid. Tricuspid pharyngeal teeth
moiphologically similar to jaw teeth. Teeth
on uppcT pharyngeal toothplates of equal
size throughout tooth rows; teeth on lower
toothplates substantially larger towards
middle of pharynx.

Weberian apparatits. Weberian apparatus
similar to that in Uemiodus, apparently
without modifications useful for phyletic
analysis within Hemiodontidae.

Pectoral girdle. Pectoral girdle of

422 Bulletin Museum of Comparative Zoology, Vol. 146, No. 9

generalized characoid morphology, similar
to that of Hemiodus, except for postcleithra
of Bivihranchia. In Argonectes three post-
cleithra, similar in shape and size to those
of Hemiodus. In Bivihranchia only two
postcleithra, corresponding to postcleithra
2 and 3 in other characoids. Shape of
postcleithrum 3 as in Hemiodus, strutlike
distally, with a proximal lamina. Shape of
postcleithrum 2 specialized, with a strutlike
extension dorsally.

Pelvic girdle. Shape of pelvic girdle
similar to that in Hemiodus. Base of ischiac
process directed medially, perpendicular to
pelvic bone. Argonectes with 12 pelvic
rays, Bivihranchia with ten.

Vertebral counts. Total vertebral and
precaudal plus caudal vertebral counts:
Argonectes, 43 (24 + 19); Bivihranchia,
40(22+18) (in two specimens).

Caudal skeleton ( Figures 38-39 ) . Caudal
skeleton of typical hemiodontid morphol-
ogy.

Anodontinae

Cranium (Figures 40-42). Cranium rela-
tively narrow and shallow. Frontoparietal
fontanel complete; frontal portion about
half as wide as parietal portion and uni-
form in width for most of its length, widen-
ing slightly immediately in front of
epiphyseal bar. Frontal portion prolonged
anteriorly by opening between diverging
posterior limbs of ethmoid bone (Fig. 40).
Ethmoid moderately wide, more so than in
Hemiodontinae but much less than in
Curimatidae, its anterior end with processes
articulating with premaxillaries and maxil-
laries (as in Hemiodus). Supraoccipital
crest flattened dorsally but narrower than
in Hemiodus and failing to reach end of
cranium. Exoccipitals and basioccipital
extend considerably further posteriorly
than supraoccipital crest. Posttemporal
fossae with three openings as in Hemio-
dontidae, Curimatidae and many other
characoids. Dilator groove well developed,
opening broadly onto dorsal surface of
cranial roof. Posterior margin of epiotic

strongly angulated but without posteriorly
directed spine. Pterotic with very elongate,
slender, posteriorly directed process.

Vomer with a relatively flat ventral sur-
face (no ventrally directed median ridge),
and a pair of relatively small, separately
formed lamellae on its dorsal surface, these
dorsolateral elements considerably smaller
than comparable elements present in
Hemiodontinae (comparable elements ab-
sent in Curimatidae?) Ventral portion of
vomer with strongly demarcated anterior
and posterior portions; anterior portion
circular in ventral view, posterior portion
circular anteriorly but drawn out to a point
posteriorly. Parasphenoid relatively straight,
gently sloping upwards anteriorly, with a
poorly developed, ventrally directed median
lamella. Posterior portion of parasphenoid
cleft for about one-eighth of its length only
(cleft for all or almost all of its length in
other Hemiodontidae). Lateral ethmoids
similar in shape to those in Hemiodontinae,
although somewhat smaller, and with an-
terior processes articulating with vomer
more elongate. Rhinosphenoid present.
Interorbital septum smaller than in Hemio-
dontinae. Subtemporal fossa shallow but
distinct. Intercalar bone large, with a short,
posteriorly directed process bearing liga-
ment of attachment with lower limb of
posttemporal bone. Auditory fenestra large.
Lagenar capsules moderate in size (not
greatly enlarged as in Curimatidae).

Otoliths (Figure 43). All three otoliths
in Anodontinae relatively elongate com-
pared to those in Hemiodus and Brijcon.
Lapillus enlarged, even larger than asteris-
cus.

Facial hones (Figures 44-45). Circum-
orbital series complete. Infraorbital 4 with
a posteriorly directed canal from infra-
orbital branch of cephalic laterosensory
system. Infraorbitals flat and relatively
small, especially infraorbital 6, leaving
much of cheek (nearly all of preopercle)
exposed. Nasal bone an elongate open canal
or trough, without a lamellar portion,
Antorbital small, contacting supraorbital

Hemiodontidae and Parodontidae • Roberts 423

but separated by a gap from infraorbital 1.
Opercle and intcropercle relatively large;
gill cover prolonged posteriorly by sub-
opercle (similarly developed opercular
bones occur in curimatids such as Gastero-
tomus and SuprasinelichtJujs). Dorsal
border of opercle notched, as in Hemio-
dontinae (in Curimatidae dorsal margin of
opercle invariably rounded). Lateral sur-
face of opercle without a flange external to
hyomandibulo-opercular joint but with an
oblique flange just behind and parallel to
its anterior margin (Fig. 44).

Jaius (Figures 44-45). Jaws toothless in
adults and in smallest specimens known
thus far (down to 46.4 mm). Gape large,
extent of mouth-opening greater than in
any other Hemiodontidae or in any Curi-
matidae. Premaxillary small, maxillary and
dentary enlarged. A large oval fenestra
near anterior margin of dentary. Articular
bone moderately elongate. A small, hori-
zontally oriented slit where dorsal margin
of articular passes internally to dentaiy.
Coronomeckelian bone small, its position
relatively ventral (not elevated as in Bivi-
branchiinae and Parodontidae).

Suspemorium (Figure 45). Palatine
small, immovably attached to ectoptery-
goid. Ectopterygoid slender, failing to con-
tact quadrate posteriorly. Mesopterygoid
greatly expanded, immovably (?) attached
to ectopter>^goid and metapterygoid but
movably attached to quadrate. Symplectic
and posterior limb of quadrate elongate.
Metapterygoid-quadrate fenestra large.
Metapterygoid without anteroventral limb
forming posteroventral border of meta-
pterygoid-quadrate fenestra. Main body
of hyomandibular vertically oriented, its
anteroventral border greatly indented or
concave; posterior end of metapterygoid
lying in this indentation but failing to con-
tact hyomandibular.

Hyoid ami l)ranchial arches (Figures
48-50). Basihyal slender and elongate,
seemingly a single ossification. Hyoid bar,
especially ceratohyal, relatively elongate.
Interhyal small. Dorsal and ventral hypo-

hyals separated from each other slightly,
not so closely fitted together as in other
hemiodontids. Ceratohyal with dorsal and
ventral processes extending from its an-
terior end and applied to lateral surface of
dorsal and ventral hypohyals, respectively
(Fig. 48). Five branchiostegal rays; four
on ceratohyal and one on epihyal. Proximal
end of branchiostegal ray 4 not greatly ex-
panded. Branchiostegal rays elongate and
moderately slender. Urohyal extremely
elongate. Lateral wings of urohyal narrow
and strongly depressed (ventral surface of
urohyal deeply indented). Dorsomedian
lamina of urohyal well developed, its crest
extending posteriorly bevond lateral wings
(Fig. 48).

Branchial arches with three basibran-
chials and three hypobranchials, basi-
branchials 1-3 and hypobranchials 1-3
successively more elongate. Basibranchial
1 very small, basibranchial 2 moderately
elongate, and basibranchial 3 extremely
elongate (t\vice as long as basibranchial
2). Hypobranchials 1 and 2 flat or nearly
so; hypobranchial relatively elongate, its
dorsal surface moderately excavated (much
less so than in Hemiodus). Ceratobran-
chials 1-4 successively shorter; cerato-
branchial 1 very elongate; ceratobranchial
4 distinctly shorter and ceratobranchial 5
narrower than ceratobranchials 1-3. Four
epibranchials and three infrapharyngobran-
chials (a very thin, slivery and elongate
ossification where infrapharyngobranchial
4 would be located is interpreted as an
upper pharyngeal toothplate; it bears
minute teeth and is similar to lower pharyn-
geal toothplate in appearance). Epibran-
chials 1-3 and infrapharyngobranchials 1-3
with well-developed apophyses. Infra-
pharyngobranchials 1-3 successively larger,
infrapharyngobranchial 3 twice as long as
infrapharyngobranchial 2. Pharyngeal den-
tition consisting of minute conical teeth
rather widely spaced on elongate and ex-
tremely narrow toothplates.

Gill rakers ( Figures 48-51 ) . Numerous
gill rakers on both leading and trailing

424 Bulletin Museum of Comparative Zoology, Vol. 146, No. 9

edges of first four gill arches and on lead-
ing edge of fifth gill arch ( ceratobranchial
5). Gill rakers of arch 5 interdigitating
with those on trailing edge of arch 4 and
extending distally into a sort of pocket or
epibranchial organ, although arrangement
of distal gill rakers basically similar to that
in other Hemiodontidae and in Parodonti-
dae ( cf . Fig. 13 of Hemiodus and Fig. 74 of
Saccodon). Rakers very elongate and
slender, directed foreward at an acute
angle; long axis of rakers on leading and
trailing edges of first three gill arches
closely parallel. One margin of each raker
smooth, with a narrow membranous flap
for its entire length; opposite margin amied
with a double row of minute ctenii. Rakers
on leading edges of arches with ctenii di-
rected medially, opposing laterally-directed
ctenii on rakers of trailing edges. Ctenii
increasing in number with age. In a 55.0-
mm A. melanopogon most rakers about 3
mm long, with a double row of 40 + 40
ctenii. A 212-mm individual has most rakers
about 11 mm long, with just over 130 + 130
ctenii. A 200-mm A. elongatus has most
rakers about 10 or 11 mm long, with about
130 + 130 or more ctenii. In large speci-
mens ctenii approximately 0.2-0.3 mm long
and 0.1 mm or less apart (Fig. 51). Rakers
of successive arches not diminished in size,
and with ctenii equally numerous. Rakers
on trailing edge of arches as long as those
on leading edge.

Number of rakers increasing with age,
first arch with 90-100 rakers in specimens
of 50-55 mm and almost 200 in large
specimens. A 55.0-mm A. melanopogon
has 35+1 + 62 rakers on its first arch, a
212-mm specimen 73 + 1 + 115; a 200-mm
A. elongatus has 80 + 1 + 110. Number of
rakers on upper limb of arch greater in
arches 2-4 than in arch 1. A. melanopogon
of 55.0-mm with 35+1 + 62 rakers on arch
1 has 48 + 1 + 65 on arch 2, 58 + 1 + 51 on
arch 3, and 51 + 1 + 63 on arch 4. Rakers
at extremes of arches may extend beyond
bones of arches, most notably on lower
limb of arch 4, where a long series of rakers

extends anteriorly from anteromedial end
of ceratobranchial (Fig. 50) (no compar-
able series of rakers in Hemiodus).

Weberian apparatus (Figure 52). Web-
erian apparatus differing from that of
Hemiodus only in details of shape and
proportion of some parts, most notably
neural complex, which is very elongate and
low-lying.

Pectoral girdle (Figure 53). Pectoral girdle
complete, each half with extrascapular,
posttemporal, supracleithrum, cleithi-um,
three postcleithra, mesocoracoid, coracoid,
scapular, four proximal radials, and an
irregular series of small distal radials (not
figured). Postcleithra relatively small but
shaped as in Hemiodus. Coracoid small,
failing to contact anterior end of lower
limb of cleithrum. Posttemporal, supra-
cleithiiim and upper limb of cleithrum
unusually slender and elongate, as in
curimatids such as Gasterotomus and
Suprasinelichthys, which have greatly ex-
panded gill covers and gill openings super-
ficially similar to those of Anodus.

Pelvic girdle (Figure 54). Pelvic girdle
similar to that in Hemiodus. Pelvic bone
slightly narrower and ischiac process more
elongate than in Hemiodus. Pelvic fin with
11 rays.

Vertebral counts. Anodontinae exhibit
total vertebral and precaudal plus caudal
vertebral counts as follows: A. elongatus,
45 (32 + 13); A. melanopogon, 44 (29 + 15)
and 45 (29 + 16).

Caudal skeleton (Figure 55). Caudal
skeleton of generalized hemiodontid mor-
phology. Three slender epurals. Hypurals
1 and 2 fused to each other and both sepa-
rate from ural centrum. Principal caudal
rays 10 + 9.

Parodontidae

Cranium (Figures 56-59). Cranial roof
including supraorbital portion of frontal
bone convex and smooth. Frontal and
parietal bones of opposite sides broadly
overlapping at midline of cranium for their
entire length (frontoparietal fontanel ab-

Hemiodontidae and Parodontidae • Roberts 425

sent). Dilator fossa roofed by frontal bone,
not extending onto dorsal surface of
cranium. Ventroposterior extension of
supraorbital process of frontal bone with
a large foramen open into dilator fossa.
\^entral rim of foramen formed by a thin
semicircle of bone. Supraoccipital crest
absent; posterior margin of supraoccipital
either with a bifid dorsal process (P.
guijoncmis, A. affinis), bilateral, horizontal
flanges (P. caliensis), or evenly rounded
(Saccodon). Posttemporal fossa with three
openings (as in Hemiodontidae, Curimati-
dae, etc.). Head of ethmoid with a thin
ventromedian lamella and lateral projec-
tions for articulation of movable premaxil-
laries. Posterior portion of ethmoid with a
strong \ entromedian process broadly joined
to vomer. Ethmoid and vomer rigidly joined
to each other; posterior laminae of ethmoid
loosely attached to ventroanterior surface
of frontals, apparently moving freely be-
neath them. Articulation of vomer with
parasphenoid and lateral ethmoids appar-
ently movable (permitting head of ethmoid
to move in a vertical plane?). Rhinosphe-
noid absent. Orbitosphenoid enlarged,
ventroanterior process of orbitosphenoid
sutured to posteromedially directed pro-
cesses from lateral ethmoids, orbitosphenoid
failing to contact parasphenoid. Anterior
portion of parasphenoid straight, not curv-
ing upwards as in Hemiodontidae and
many other characoids. Hyomandibular
fossa developed on lateral margin of prootic
bone and anterior portion of pterotic (in
most characins, including Hemiodontidae,
the main articular surfaces of the hyo-
mandibular fossa are on the sphenotic and
pterotic). Subtemporal fossa well de-
veloped, lying entirely on pterotic bone.
In Saccodon exoccipitals with a narrow but
deep bony ridge extending laterally from
lagenar capsules to intercalar bone. Lage-
nar capsules reduced in size. Intercalars
relatively large. Epiotic with moderately
elongate posterior process. Posterior process
of pterotic short and rounded.

Otoliths {Figure 60). Otoliths of relatively

generalized characoid morphology, not
greatly different in size and shape from
those in Brycon.

Facial bones (Figures 61-65, 70). Nasal
bone slender and moderately elongate, al-
most entirely tubular. Circumorbital series
complete, with a supraorbital, antorbital
and six relatively large infraorbitals. Antor-
bital a small, triangular element, its apex
separated by a slight gap from supraorbital,
its base resting on dorsal margin of infra-
orbital 1. Infraorbital 1 covering posterior
portion of premaxillary. Infraorbital 2
covering most of dentar)^ Infraorbitals 2-4
almost entirely hiding preopercle from
view. Ventral portion of infraorbitals 2 and
3 curving strongly underneath head.

Opercular series complete, with a pre-
opercle, interopercle, subopercle and oper-
cle. No supraopercle or suprapreopercle.
Lateral surface of opercle smooth, dorsal
margin of opercle broadly rounded. Poste-
rior portion of interopercle greatly deep-
ened. Subopercle relatively shallow for its
entire length, with a well-developed ascend-
ing process internal to "junction" of pre-
opercle, opercle and interopercle. Anterior
end of preopercle immovably joined to
quadrate; anterolateral surface of preoper-
cle completely covered by bony extension
of quadrate. Opercle and subopercle
oriented parallel to main axis of head. An-
terior ends of preopercle and interopercle
inclined inwards towards quadrato-man-
dibular junction.

Jaws and jaw teeth (Figures 63-71).
Premaxillaries greatly expanded, with from
2 -j- 2 to 5 + 5 functional teeth in a single
row. Replacement tooth trenches for each
functional tooth position separated from
each other by bony ridges on inner face of
premaxillaries with from four to 30 pre-
formed replacement teeth. Replacement
teeth largely enclosed by lateral and
median walls of premaxillary and in varying
degrees by a ventrally directed extension
arising from dorsoposterior margin of pre-
maxillary. In Apareiodon affinis this ex-
tension is short and replacement teeth lie

426 Bulletin Museum of Comparative Zoology, Vol. 146, No. 9

relatively exposed posteriorly. It extends
further ventrally in Parodon caliensis. In
dental morph I of S. tvagneri, entire dorso-
posterior margin of premaxillary extends
ventrally to cover almost half of extent of
replacement tooth rows. In dental morph
IV of S. loagneri dorsoposterior margin of
premaxillary extends ventrally to cover
two-thirds of extent of replacement tooth
rows. Flattened median walls of premaxil-
laries of opposite sides firmly united to
each other by tough connective tissue.
Premaxillaries with dorsomedian process
articulating to ethmoid. Dorsomedian pro-
cesses greatly expanded in A. affinis and P.
caliemis, moderately expanded in P. giiyan-
ensis, and relatively small in S. wagneri. A
longitudinal depression or groove in dorsal
surface of premaxillary, lateral to dorso-
median process, into which fit lateral pro-
jections from head of ethmoid.

Maxillary relatively small and slender,
strongly curved, edentulous or with one to
three small teeth. One preformed replace-
ment tooth for each functional tooth lies in
a pocket in maxillary. Ascending process of
maxillary loosely attached to premaxillary.
Some specimens of S. ivagneri with a small
supramaxillary at tip of ascending pre-
maxillary process.

Lower jaw extremely similar in form to
that in Hemiodontidae, albeit of relatively
stouter construction. Dentaries with an
oval fontanel near anterior margin, slender
symphysial processes joined to each other
by tough connective tissue, and elevated
lateral walls just inside rictus of jaws.
Dentary either toothless (Apareiodon,
Saccodon, young Parodon) or with one to
three large stout teeth planted on inner sur-
face of lateral expansion just inside rictus
of jaws. Articular bone and posterior por-
tion of dentary slender. Coronomeckelian
bone large, its position near dorsal edge of
articular much higher than in most other
characoids.

Suspensorium {Figures 63-65, 70, 72).
Palatine large, its anterior end expanded,
its posterior end immovably attached to

ectopterygoid and lying in a long ti-ough
or groove in dorsoanterior surface of ecto-
pterygoid. Ectopterygoid moderately ex-
panded anteriorly, slender posteriorly, its
posterior end forming a narrow, movable
joint with dorsal edge of quadrate. Meso-
pterygoid movably articulated with median
surfaces of ectopterygoid and metaptery-
goid. Posterior portion of metapterygoid
loosely bound to hyomandibular. Meta-
pterygoid-quadrate foramen moderately
large. Anterior end of hyomandibular in-
clined inwards towards quadrato-man-
dibular joint. Quadrate firmly united to
preopercle, a bony extension from quadrate
overlying lateral surface of anterior end of
preopercle. Quadrato-mandibular condyle
very large. Lateral surface of quadrate
posterior to condyle greatly thickened.

Hijoid and branchial arches {Figures 73,
74). Basihyal moderately elongate, narrow
posteriorly and slightly expanded an-
teriorly. Hyoid bar of generalized chara-
coid morphology. Dorsal and ventral
hypohyals separate (not fused together).
Interhyal relatively small. Apareiodon with
three branchiostegal rays, Saccodon and
Parodon with four. In Saccodon branchios-
tegal rays 1 and 2 ending basally in slender
flanges articulating on inner surface of
ceratohyal. Branchiostegal rays 3 and 4
with relatively stout basal flanges articulat-
ing with external surface of ceratohyal and
epihyal, respectively. Urohyal short and
stout, its horizontal laminar portions rel-
atively broad, its dorsal median lamina
temiinating in a high crest.

Three basibranchials. Basibranchial 1
short, basibranchials 2 and 3 elongate.
Three hypobranchials. Hypobranchial 1
flat, hypobranchials 2 and 3 with strongly
concave dorsal surfaces. Epibranchials and
ceratobranchials relatively shorter than in
Hemiodontidae. Epibranchials oriented al-
most at a right angle to main axis of
body when viewed from above; cerato-
branchials at an angle of about 45
degrees. Three infrapharyngobranchials;

Hemiodontidae and Parodontidae • Roberts 427

infrapharyngobranchials 2 and 3 moderately
enlarged.

Gill rakers moderately elongate or short
and platelike, without ctenii. Number of
gill rakers moderate, increasing slightly
with age. Saccodon iragncri of 50 to 55
mm have 32-35 gill rakers on first gill arch.
A 103.3-mm specimen of dental morph IV
with gill rakers on leading and trailing
edges of gill arches as follows: arch 1,
20+1 + 22 on leading edge and 30 + 27 on
trailing edge; arch 2, 15 + 1 + 39 and 23 +
1 + 36; arch 3, 32+1 + 42 and 23 + 43;
arch 4, 23 + 46 + (?) and 0 + 35; and arch
5, 0 + 28 ( on leading edge only of cerato-
branchial 5). A 63.5-mm Parodon caliemis
has 16 + 1 + 23 rakers on leading edge of
arch 1, a 46.8-mm Apareiodon affinis, 12 +
1 + 22. In A. affinis gill rakers on leading
edge of arches 2 and 3 enlarged, much
larger than those on arch 1; in other paro-
dontids examined, largest rakers occur on
leading edge of arch 1 (Fig. 74). Row of
gill rakers on leading edge of ceratobran-
chial 5 terminating distally in a semicircle
of rakers of diminutive size, joining a
complementary semicircle of similar rakers
from trailing edge of ceratobranchial 4
(Fig. 74). A comparable condition occurs
in Hemiodontidae ( Hemiodontinae and
Anodontinae) and perhaps in other chara-
coids.

Pharyngeal dentition consisting of nu-
merous, small conical teeth, apparently
non-pedicellate, well developed only on last
pair of upper pharyngeal toothplates ( those
loosely associated with medial ends of
epibranchials 3 and 4) and on lower
phaiyngeal toothplates associated with dor-
sal surface of ceratobranchial 5. Toothplates
on ventral surface of infrapharyngobran-
chials 2 and 3 either absent or small and
with few teeth.

Weberian apparatus {Figure 75). Webe-
rian apparatus similar to that in numerous
characoids, apparently without useful fea-
tures for phyletic studies.

Pectoral girdle (Figure 76). In A. affinis
pectoral girdle, including postcleithra 1-3,

relatively similar to morphologically gener-
alized pectoral girdle of Hemiodus (no
close relationships implied). In Saccodon
and Parodon cleithrum enlarged, its an-
terior and posterior limbs curving strongly
inwards; medially directed lamina from an-
terior and ascending limbs of cleithrum
very strong; postcleithrum 2 greatly ex-
panded, postcleithrum 3 strutlike, narrow
for its entire length. In Saccodon, first two
pectoral fin rays simple; in Parodon and
Apareiodon only first pectoral ray simple.

Pelvic girdle (Figure 77). Anterior por-
tion of pelvic bone simple (not bifid or
trifid), broader in Saccodon than in
Apareiodon or Parodon. Ischiac process
relatively short and simple in Saccodon. In
P. caliensis and A. affinis ischiac process
elongate and thin posteriorly, especially in
P. caliensis, and bearing a short, lateral
projection near its base. Eight or nine
pelvic fin rays.

Vertebral counts. Parodontidae exhibit
total vertebral counts and precaudal plus
caudal vertebral counts as follows: P.
caliensis 35 (14 + 21); A. affinis 41 (20 + 21,
21 + 20); dental morph I of S. wagneri 38
or 39 (18 + 20, 20+18, 19 + 20); and
dental morph IV of S. wagneri 38 ( 18 + 20
in two specimens, 19 + 19, 20 + 18).

Caudal skeleton (Figure 78). Caudal
skeleton morphologically generalized for
characoids, similar to caudal skeleton of
most Characidae, Curimatidae, etc. Epurals
two. Uroneurals two. Hypurals six; hypural
2 slender, separate from hypural 1, fused
to complex ural centrum ( as in most chara-
coids ) . Principal caudal rays 10 + 9.

DISCUSSION

Relationship Between Hemiodontidae
and Parodontidae

The superficially similar appearance of
Hemiodontidae and Parodontidae together
with the unique disposition of the jaw teefli
common to many species in botli families,
has led most students of characoid classifi-
cation to regard them as closely related

428 Bulletin Museum of Comparative Zoology, Vol. 146, No. 9

Table 2. Distingltishing characteristics of Parodontidae and Hemiodontidae

Parodontidae

Hemiodontidae

Premaxillary

Maxillary

Frontoparietal
fontanel

Rhinosphenoid

Adipose eyelid

Branchiostegal rays

Hypural 2

Vertebrae

Highly mobile.
Greatly enlarged.

From four to over 30 preformed re-
placement teeth for each functional
tooth lying in trenches with bony
partitions.

Lies entirely lateral to premaxillary,
without contacting ethmoid.

Absent.

Three or four.

Fused to ural centrum and separate
from hypural 1.

35 to 41.

Immobile or only slightly mobile.

Minute to considerably enlarged but never
as much as in Parodontidae.

A maximum of four to six preformed replace-
ment teeth for each functional tooth; no
trenches or bony partitions separating them.

Tends to articulate directly with ethmoid.

Present.

Four or five.

Separate from ural centrum and fused to
hypural 1.

40 to 45.

(Regan, 1911; Eigenmann, 1912; Fowler,
1950; Gery, 1959; Greenwood et al, 1966).
Only Regan (1911) and Gery (1959) pre-
sented osteological evidence bearing on
this problem, and their evidence is brief
and not very informative. Regan pointed
out some similarities in the suspensorium
(similarities shared with the unrelated
fomis Nannostomtis and Characidium),
stated that the hyomandibular is single-
headed (I find it bicipital in both families),
and incorrectly indicated that the maxil-
laries in both Hemiodontidae and Paro-
dontidae articulate with the ethmoid (in
some Hemiodontidae the maxillaries articu-
late with the ethmoid, but in Parodontidae
they lie entirely lateral to the expanded
premaxillaries ) . Gery ( 1959 ) compared the
superficial cranial features of Parodon
g,mjanensis with those of Heniiodiis quadri-
maculatus and found them similar in a
number of respects but the similarities in-
volved do not include any specialized
features.

The similarity of the jaws ( especially the
lower jaws) and dentition of many Paro-
dontidae and Hemiodontidae is undeniably

striking. The shape of the upper jawbones
in H. quadrimaculatus, together with the
decrease in number of functional teeth and
increase in replacement teeth in this spe-
cies (Fig. 10), does suggest a step towards
the highly specialized premaxillaries and
maxillaries of Parodontidae. On the other
hand, the premaxillaries of H. quadrimacu-
latus show no indication of bony septae
separating the replacement teeth for each
functional tooth position (such septae are
diagnostic of Parodontidae). The similarity'
in shape of the lower jaws in the two
families can be seen by comparing Figures
7 and 8 with 65 and 70. The specialized
suspensorium is also very similar. The
slightly overlapping condition of the quad-
rate with the anterior end of the preopercle
in Hemiodontinae suggests a step towards
the highly specialized condition in Paro-
dontidae (compare Figs. 11 and 72). It
seems, however, that phyletically suggestive
osteological similarities do not extend be-
yond the jaws and suspensoria. The
cranium of Parodontidae is highly special-
ized and differs in many respects. The
hyoid and branchial arches, Weberian ap-

Hemiodontidae and Pakodontidae • Roberts 429

paratus, pectoral and pelvic girdles and
vertebral columns in both families are of
relatively generalized characoid morphol-
ogy and do not seem to offer useful char-
acters in the present context. The caudal
skeleton of Parodontidae is more general-
ized than that of Hemiodontidae in having
hypurals 1 and 2 separate and hypural 2
fused to the ural centrum. Whether Hemio-
dontidae and Parodontidae are indeed close
relatives will have to be resolved by further
study. In any event, there are considerable
differences between the two families
(Table 2).

Transfer of Anodontinae from
Curimatidae to Hemiodontidae

Anodontinae has been placed in Curi-
matidae by most students of characoid
classification (Eigenmann and Eigenmann,
1889; Regan, 1911; Fernandez-Yepez, 1948;
Greenwood et al., 1966) because its mem-
bers lack jaw teeth and superficially re-
semble curimatids such as Gasterotomtis.
Evidence that Anodontinae should be trans-
ferred from Curimatidae to Hemiodontidae
can be marshalled as follows:

1. Curimatidae are characterized by a
strong flange on the lateral surface of the
opercle, just above the hyomandibulo-oper-
cular joint ( Fig. 46 ) . This feature is readily
observable, without dissection, in alcoholic
specimens, and is present in all species of
Curimatidae seen by me. I do not know
of any other characoids with a comparable
opercular flange. The flange is absent in
Anodus.

2. In the lower jaw of Curimatidae the
articular is relatively short; both the den-
tary and articular are elevated; where the
dentary and articular join there is a large
vertical gap or fenestra, and there is no
oval fenestra in the dentary (Figs. 46-47).
In Hemiodontidae the articular is relatively
elongate; the dentary is elevated in varying
degrees, but the articular is never elevated;
where the dentary and articular join there

are cither no gaps or only horizontally
oriented gaps (cf. BivihrancJiia), and there
is an oval fenestra in the dentary. In all of
these features the lower jaw of Anodus
agrees with that of Hemiodontidae. Anodus
has an exceptionally large oval fenestra in
the dentary.

3. In Curimatidae the gill rakers are absent
or poorly developed: with little or no ossi-
fication, short and "fleshy" and relatively
few in number. In most hemiodontids the
gill rakers ate well developed: well ossified,
elongate, often bearing ctenii, and rel-
atively numerous. The gill rakers of
Anodus are exceptionally well developed:
well ossified, very elongate, provided with
minute ctenii, and more numerous than in
any other characoids ( increasing in number
with age).

4. Curimatidae usually (always?) have
four branchiostegal rays. Hemiodontidae
often have five branchiostegal rays (four
or five in Hemiodus, five in Micromischo-
dus, Argonectes, Bivibranchia) . Anodus has
five branchiostegal rays. In Curimatidae
the branchial membranes are united to the
isthmus. In Anodontinae and in other
Hemiodontidae excepting Bivibranchia the
branchial membranes are free from the
isthmus and the gill openings extend rel-
atively far forward ventrally.

5. Curimatidae are characterized by a joint
between the palatine and lateral ethmoid
which is absent in Hemiodontidae and
Anodontinae. In Curimatidae the dorsal
surface of the posterior end of the palatine
is expanded into a facet (Fig. 47) which
either attaches directly to a facet on the
ventral surface of the lateral ethmoid just
internal to its lateral wing or else is joined
to such a facet by dense connective tissue
(sometimes cartilaginous). There is no
indication of these facets on the palatine
and lateral ethmoid in Hemiodontidae or
Anodontinae.

6. Curimatidae are characterized by a
greatly enlarged lagenar capsule. In Ano-

430 Bulletin Museum of Comparative Zoology, Vol. 146, No. 9

dontinae, as in Hemiodontidae, the lagenar
capsule is relatively small.

7. In Curimatidae body shape is extremely
variable, but rarely, if ever, is it truly sub-
cylindrical and fusiform as in all Hemio-
dontidae and in Anodontinae. Many
curimatids are deep bodied or laterally
compressed. The ventral abdominal surface
is frequently squared off, keeled, or cari-
nate. The body contour is frequently
modified in front of or along the base of
the dorsal and anal fins.

8. Of more than 100 species, only one
curimatid — Curimatorbis ocellatus Eigen-
mann and Eigenmann 1889 — has an oval
spot on the middle of the body just behind
the dorsal fin. Such a spot is more char-
acteristic of Hemiodontidae, being present
in several species of Hemiodus, in Argo-
nectes, and in two undescribed species of
Bivihranchia from Surinam (Gery, personal
communication). A comparable spot is
always present in A. elongatus and some-
times present (although faint) in A.
melanopogon (Plate I).

9. Excepting very small species and young
stages, Curimatidae have an intestine with
numerous coils. In Hemiodontidae and in
Anodxis the intestine is moderately coiled.

10. Curimatidae have low vertebral counts,
from 30 to 36, about evenly divided be-
tween precaudal and caudal vertebrae.
Anodontinae have 44 or 45 vertebrae, pre-
caudal vertebrae twice as numerous as
caudal. Hemiodontidae other than Ano-
dontinae have 40 to 43 vertebrae, and
precaudals about twice as many as caudals.

11. In Curimatidae hypural 2 is fused to
the complex ural centiiim and entirely
separate from hypural 1. In Anodus, as in
Hemiodontidae, hypural 2 is fused to
hypural 1 and has lost its connection with
the ural centrum.

In comparing the skull of Anodus with
those of other characoids, one is struck by
how much space is occupied by the gill

arches and expanded gill covers and how
little by the elongate, narrow and relatively
depressed cranium. In curimatids such as
Gasterotomus and Suprasinelichthijs in
which the gill covers are also enlarged and
the external appearance of the head strongly
resembles that of Anodus, the cranium
is relatively heavy, broad and deep. The
general appearance of the cranium of
A^iodus is more like that of Hemiodus than
of any curimatids I have examined.

Anodus agrees with both Hemiodontidae
and Curimatidae in having three openings
into each posttemporal fossa and a rhino-
sphenoid bone. These characters are of
minor phyletic significance, since three
openings into the posttemporal fossa occur
in many noncharacid characoids, African
as well as South American, and the rhino-
sphenoid is present in many South Ameri-
can groups. The only character shared
exclusively by Curimatidae and Ano-
dontinae is absence of jaw teeth, a loss
character, and therefore of dubious sig-
nificance.

The characteristics of Anodontinae are
approached in some ways by Hemiodus
microlepis. The superficial resemblance
between this species and A. elongatus is as
striking as that between A. elongatus and
curimatids such as Curimatorbis ocellatus
or Gasterotomus latior. H. microlepis has
about 105-115 scales in a lateral series and
an oval spot on the body just posterior to
the dorsal fin. The gill rakers, although not
elongate as in A. elongatus, are relatively
numerous (probably increasing in number
with growth) and denticulate. A 235-mm
specimen has 30 + 40 gill rakers on the first
gill arch. Each comblike raker bears about
10 large ctenii (Fig. 14). Fowler (1906:
319) reported 22? + 38? ciliate rakers in a
specimen nine inches long. Gill rakers of
the last arch are relatively large and inter-
digitate with trailing gill rakers of the
fourth arch, as in other species of Hemiodus
and in Anodus. The teeth, although typical
in shape for Hemiodus, are relatively small
and numerous (about 20-22 on each side

Hemiodontidae and Parodontidae • Roberts 431

<
a

S
u

b
O

C/3

a
jj

u
K

O
en

P
en

2
h
<

U
o

H
en

n
<

iMpi-:H

l-o

itions com-
in Hemio-

3
3

S o oi

3 .

rt 4-.

D m

-g

. >'0 t^D >^ ^

cX g P ^ a'd

.« S ^ oa Ji c P

^a -i 1 s s g 1 a

P-i S "i
'^ .3 >

P rt^i

« ;-H W5

z s §

3 m
O en

S 2 c
O aT3

3^
0 :3

-^ §

(D
a-*
ci en

13 t,
0 <u

s s

IJH

a o o3

1 1

a. 2

o P

" g

O
C
<

Si 23

XI 3

0) "a

'^ .s >
o

3 'T ^

13
3 lU

O en

4->

-^ o

4-J G en
rt 3 .

0 3 aJ 0

u §^
13 . 3

O rt .^^

23 S

CD "-^ '^

3 « >>_2

■H

h !-■ '^

§ gl'^

3 -a >N

Z G O

C S 3

0 ai3

i! 2 a rt

W 3 ^ .S

>'T3 t: -£2 rj

g c 5 <D -s
-< i 3 O §

■ — 1 ^ r ■ n-i a

1 ^^1 Si

a-Sj 3

S 5 -S 3

"^ TO ^ •«
. >.== "

<D -73 UH 3

S S ° 2

en C;3 13 MH

2 3 ^ 2i

•^ J— 1

5-. • 1 ' (D

.2 g 2'?^
fc 1 2 3'§

^ 1^ •^ b 3
4) > >~> in 3

-- 3 > a -b 13

> c;

.rt en

"■a

2 - cs 0 X 0
- 2^ § .S -o

3 ^

- a
u a

0< 5 4J s^

2 >>c 2

>^^ ^ . -i:^ a

2 -0 3

0) -

^•2

.5P ?3 ^ --g

bOi3 b g iJl3
ffi OJ ^ 3 2 rt

13 >

3
0
fa

;-H rt <U OQ Oi

X ^^H

S "

1 —^ -u 1

3 a -fi 13

:2.2^-s^

irT id >^

9; 3 0)

^ OU

3 (U >S

= 1 :..

.3 TTJ <U "

2'c

U 3 OJ -3

C
o
-a

c t^ o g

° S ft ^•
^ ° ^ rt -5

. O
en o

2 a .

■gl53 0 ^
13 .2, tu jJ '0

ll.fi

.sISs

3 0 G

rj Tf >> O tn (U

<u CO T= -t; •'-'

di a • — ' a v

o a^

C 1^ S-

o bc S

0 s s a &

ii
3
0

^r-<u_ ^ d, u ^
hJ o C O (D fi

2i S

•g

f-j

•4-*

1— >

432 Bulletin Museum of Comparative Zoology, Vol. 146, No. 9

of the upper jaw). The tapered posterior
chamber of the swimbladder extends to or
shghtly beyond the base of last anal fin
ray. I have examined two specimens (MCZ
20652; 229 and 235 mm; Tefe, Amazonas,
Brazil. Thayer Expedition ) . The osteology
of this species has not been investigated.

The distinguishing characteristics of the
Anodontinae and of the other three sub-
families of Hemiodontidae are presented in
Table 3.

Remarks on Lower Taxa

Genera of Hemiodontinae. Three genera
of Hemiodontinae are currently recognized:
Hemiodus Miiller 1842, Hemiodopsis Fow-
ler 1906, and Pterohemiodus Fowler 1940.
Anisitsia Eigenmann and Kennedy 1903 is
a synonym of Hemiodus; Hemiodus in-
cludes only those species in which the
scales below the lateral line are larger than
those above it (Gery, 1961; 1963b; 1964).
The three genera are split very finely.
There is a complete gradation between
species in which scales above and below
the lateral line are the same size and
species in which scales below the lateral
line are larger. Larger scales below the
lateral line occur also in Micromischodus
(Roberts, 1971: 7, fig. 1) and in many
species of Curimatidae. The scales below
the lateral line are slightly larger than
those above it, even in some species of
Hemiodus in which they are stated to be
of equal size, e.g., H. microlepis. In itself
this character is insufficient grounds for
recognition of a separate genus, and no
other distinction between Hemiodus and
Hemiodopsis (cf. Gery, 1964) has been
offered. Employment of single characters
leads to proliferation of taxa, many of
which will be polyphyletic. If a genus is
based on such a simple character as scale
size there is no way of determining whether
it is monophyletic or polyphyletic. For
these reasons I do not recognize Hemio-
dopsis. Similar grounds can be given for
rejecting Pterohemiodus, erected for a
species with a filamentous extension of the

dorsal fin. Gery ( 1961 : 338 ) gave a more
detailed diagnosis of Pterohemiodus, and
there is perhaps more reason for retaining
this genus than for Hemiodopsis. The
vertical disposition of the teeth, indicated
by Gery as diagnostic of Pterohemiodus,
is a characteristic feature of Hemiodus
quadrimaculatus as well as most Parodonti-
dae. Systematic revision of the species of
Hemiodus is sorely needed. Systcmatists
should note that in this genus the numbers
of jaw teeth, of cusps on the jaw teeth, and
of gill rakers tend to increase with age.

Genera of Bivibranchiinae. Three genera
are currently recognized in Bivibran-
chiinae: Bivibranchia Eigenmann 1912,
Atomaster Eigenmann and Myers 1927, and
Argonectes Bohlke and Myers 1956. I have
not examined Atomaster. Eigenmann and
Myers distinguished it from Bivibranchia
because of its small ctenoid scales. Myers
(personal communication) is inclined to
think that Atomaster is too much like
Bivibranchia to be recognized as a separate
genus.

Bohlke and Myers (1956) regarded
Argonectes as a closely allied, specialized
derivative of Hemiodus, having some of the
same specializations in the jaws and denti-
tion as Bivibranchia. I find that the two
genera share several unique specializations
in the jaws, jaw teeth and suspensoria, and
thus they constitute a well-defined sub-
family (see definition above). Their pro-
tractile jaws and suspensoria are highly
specialized in much the same way, and yet
in many respects, even in the structure of
the jaws, Argonectes is clearly more primi-
tive than Bivibranchia. The posterior
portion of the maxillary is much more
elongate and curved in Bivibranchia, per-
haps an indication of greater protractility.
Argonectes lacks the specialized valves
found in the roof of the mouth in Bivibran-
chia (and in Atomaster) (Eigenmann, 1912;
Gery, 1963a; Eigenmann and Myers, 1927).
The shapes of the palatine and mesoptery-
goid, highly specialized in Bivibranchia,
are relatively generalized or only slightly

Hemiodontidae and Parodoxtidae • Roberts 433

modified in Argonectes. The ectopten-
goid-quadrate joint, highl\- specialized in
Argonectes, is even more specialized in
Bivihranchia (compare Figs. 27 and 28).
The gill arches, pharyngeal epithelium and
pharyngeal dentition, strikingly specialized
in BwU)ranchia, are generalized in Argo-
nectes, although Argonectes has even fewer
gill rakers than Bivihranchia. The infra-
orbital bones and postcleithra, generalized
in Argonectes, are moderately specialized
in Bivihranchia. Argonectes has a complete
adipose eyelid, Bivihranchia the more usual
condition in which there is an opening
in the adipose eyelid over the pupil.
Argonectes apparenth' grows larger than
Bivihranchia. I ha\e not seen any speci-
mens of Bivihranchia larger than 115 mm
(the largest reported by Eigenmann. 1912),
whereas Argonectes attains 198 mm (MCZ
20635), and I ha^•e seen specimens from
several localities larger than 115 mm.

Genera and species of Anodontinae
{Plate I: Frontispiece). Two genera of
Anodontinae are currently recognized:
Anodus Spix 1829 and Eigenmannina
Fowler 1906. {EJopomorphus Gill 1878 is
a synonym of Anodus: Eigenmann and
Eigenmann, 1889. ) The generic t}pe of
Anodus is A. eJongatus Spix 1829, by
monot}p\', a well kno\\'n and wideh' dis-
tributed Amazonian species. Elopomorphus
jordanii Gill 1878 and Anodus st eat ops
Cope 1878 have been placed (correctly, I
think) in the synonym)- of A. eJongatus
(Eigenmann and Eigenmann, 1889; Fowler,
1906; Fernandez-Yepez, 1948; Fowler,
1950). "EJopomorphus" orinocensis Stein-
dachner 1888 (pp. 66-67, pi. 2), from the
Orinoco (no other locality' given) is prob-
ably another synonym of A. elongatus. This
is the onlv non-Amazonian record of
Anodontinae, and has been overlooked
bv subsequent authors ( Feniandez-Yepez,
1948; Mago Leccia, 1970). ("Anodus"
latior, recorded from Venezuela b)' Mago
Leccia [1970], is the generic type of
Gasterotomus Eigenmann 1910 and should
be known as G. latior. It is a curimatid. )

Only one other species of Anodus has
been described, A. melanopogon Cope
1878, and it is clearly distinct. Widely
distributed in the Amazon, it has been
taken together with A. elongatus .i the Rio
Maranon at Iquitos and in the rio Xegro
near lago Alexo. It differs from A. elongatus
in having a more slender build and smaller
scales (ca. 125-130 scales in a laterrl series
vs. 100-105) (Plate I). Specimens of the
b.vo species of comparable 1 -ngth seem to
have about the same number of gill rakers,
but this should be studied in more ex-
tensive series than are presently available.
Eigenmannina Fowler 1906 takes A.
melanopogon as its generic t\pe, but A.
melanopogon is so similar to A. elomiatus
that generic separation is miwarranted. I
have compared Cope's type specimens of
A. melanopogon with a series of freshly
preserx'ed \oung specimens of the same size
( MZUSP 5959 ) . Although the t>-pes are in
\'er)- poor condition, m\- examination indi-
cated that tlie>' are the same species as the
fresh material. The "strongly concave upper
profile of the head," used as a generically
diagnostic character by Fowler (1906: 306-
307, fig. 10 ) , is clearly an artifact of preser-
\'ation. Both lots ha\'e the extensi\e black
coloration on the underside of the jaws,
high scale counts and narrow caudal
peduncle, and protruding lower jaw that
are characteristic of the species. In half-
grown and adult specimens of A. melano-
pogon the lower jaw projects onl\- slightly
or not at all, as in larger specimens of A.
elongatus. The largest specimens of A.
melanopogon examined. 209 and 212 mm,
are \er)- similar to A. elongatus of the same
size.

Genera of Parodontidae. Three genera
of Parodontidae are currently recognized:
Parodon Valenciennes 1849 (in Cuvier and
\'alenciennes, 1849), Saccodon Kner and
Steindachner 1863, and Apareiodon Eigen-
mann 1916. Parodontops Schultz and Miles
1943 has been placed in the s\-non\-my of
Saccodon as a result of my studies on
trophic polymorphism in that genus (Rob-

434 Bulletin Museum of Comparative Zoology, Vol. 146, No. 9

erts, 1974). Parodon, Saccodon and
Apareiodon are so poorly defined in the
literature (cf. Schultz and Miles, 1943;
Gery, 1959) that their distinctness may be
questioned. Parodon differs from the other
two genera in having teeth in the lower
jaw; according to Gery ( 1959 ) , the cusps
on the premaxillary teeth are more deeply
incised. Schultz and Miles stated that the
lower jaw is sometimes toothless in young
specimens of Parodon and questioned the
validity of Apareiodon. Saccodon differs
from the other two genera in having two
unbranched rays in each pectoral fin in-
stead of only one (Roberts, 1974). Since
my observations have been extended to
only two species of Parodon and two of
Apareiodo7i (see Material Examined), I
lack sufficient information to provide
meaningful definitions of Parodontidae at
the generic level.

CONCLUSION

Modification of trophic structures ob-
viously has played a major role in the
adaptive radiation of Hemiodontidae and
Parodontidae. The four subfamilies of
Hemiodontidae are definable primarily in
terms of trophic adaptations. Diversifica-
tion of jaw teeth, complete loss of jaw
teeth, radical innovations in functional
anatomy of the jaws, and modification of
oral and pharyngeal epithelia, of gill rakers,
of pharyngeal teeth, and even of endoskele-
tal elements in the gill arches have
occurred. Within Hemiodontidae, Hemio-
dontinae and Micromischodontinae, al-
though specialized in certain respects, are
relatively generalized. In Bivibranchiinae
Argonectes is clearly more primitive than
Bivihranchia in many important respects.
I see no reason to doubt that Bivihranchia
had ancestors extremely similar to the liv-
ing Arg07iectes. Bivibranchiinae differ
radically from all other characoids in their
highly modified suspensoria and protrusible
upper jaws. If Argonectes or a form very
similar to it gave rise to Bivihranchia, then
the development of protrusibility set the

stage for further modification of the
suspensorium and for structural innovations
in the oral and pharyngeal epithelia, endo-
skeletal gill arch elements and pharyngeal
dentition found in Bivihranchia. In no
other characoids are the trophic structures
more highly modified than in Bivihranchia.
Hemiodontinae are generalized enough
relative to Bivibranchiinae to have been
near the ancestral line to Argonectes. Apart
from lacking jaw teeth, Anodontinae have
little in common with Curimatidae. Unlike
Curimatidae (which feed on the bottom
and ingest quantities of mud) and other
Hemiodontidae (which also feed mainly
on the bottom but more selectively than
Curimatidae ) , Anodontinae presumably
strain small organisms from midwater.
Examination of the guts of preserved
Anodtis reveals only that they do not ingest
any mud. Determination of their food items
will probably require microscopic exami-
nation of fresh stomach contents. Hemio-
dontidae and Parodontidae offer excellent
opportunities for multidisciplinary investi-
gations of the role of feeding habits and
trophic adaptations in the evolution of
higher taxonomic categories.

LITERATURE CITED

Agassiz, L., and J. B. DE Spix. 1829. Selecta
genera et species piscium brasiliensium. i-xv
-h I-II + 138 pp, 100 pis. (irregularly num-
bered ) .

BoHLKE, J. E., AND G. S. Myers. 1956. Studies
on fishes of the family Characidae. — No. 11.
A new genus and species of hemiodontins
from the Rio Orinoco in Venezuela. Acad.
Nat. Sci. Philadelphia, Notulae Naturae, No.
286: 1-6.

Cope, E. D. 1878. Synopsis of the fishes of the
Peruvian Amazon obtained by Professor Orton
during his Expeditions of 1873 and 1877.
Proc. Amer. Phil. Soc, 17: 673-701.

CxrviER, G., AND A. Valenciennes. 1849. Histoire
Naturelle des Poissons. Paris. Vol. 22, xx +
532 pp.

EiGENMANN, C. H. 1910. Catalogue of the fresh-
water fishes of tropical and south temperate
America. Reports Princeton Expeds. to Pata-
gonia 1896-1899, zool., 3(2): 375-511.

. 1912. The fresh-water fishes of British

Hemiodontidae and Parodontidae • Roberts 435

Guiana. Mem. Carnegie Mils., 5, xxii + 578

pp.
. 1916. On Aparciodon, a new genu.s of

characid fishes. Ann. Carnegie Mus., 10:

71-76, pis. 11-12.
— , AXD R. S. EIGE^^rA^■^■. 1889. A revision

of the edentulous genera of Curimatinae.

Ann. New York Acad. Sci., 4: 409-440.
— , AND C. H. Kennedy. 1903. On a col-
lection of fishes from Paragua>', \\ith a

synopsis of the American genera of cichlids.

Proc. Acad. Nat. Sci. Philadelphia, 59: 497-

537.
-, AND G. S. Myers. 1927. A new genus of

Brazilian characin fishes allied to Bivibran-
chia. Proc. Nat'l. Acad. Sci. (Washington),
13(8): 565-566.

Fernandez-Yepez, a. 1948. Los curimatidos
(peces fluviales de Sur America). Catalogo
descripti\'o con nuevas adiciones genericas y
especificas. Ministerio de Agricultura y Cria
(Caracas). 79 pp. + inde.x.

Fowler, H. W. 1906. Further knowledge of some
heterognathous fishes. Part 1. Proc. Acad.
Nat. Sci. Philadelphia, 58(2): 293-351.

. 1940. A collection of fishes obtained by

Mr. William C. Morrow in the Ucayali ri\'er
basin, Peru. Proc. Acad. Nat. Sci. Philadel-
phia, 91: 219-289.

. 1950. Os pei.xes de aqua doce do Brasil

(part 2). Arq. Zool. Est. Sao Paulo, 6:
205-104.

Gery, J. 1959. Contributions a I'etude des pois-
sons characoides. Paiodon giitjancnsis n. sp.
de Guyane Francaise, avec quelques consider-
ations sur le groupe (Hemiodontinae). Bull.
Mus. natl. Hist, nat., ser. 2, 31(6): 481-490.

. 1961. Pteroheiniodtis luclingi sp. nov.,

un curieux poisson characoi'de a nageoire
dorsale filamenteuse, avec une cle des genres
d'Hemiodontinae ( Ostariophysi-Erythrinidae) .
Bonner zoologische Beitriige, 314(12): 332-
342.

. 1963a. L'appareil protracteur buccal de

Bivibranchia (Characoidei) avec une note
sur Phractolaemus (Chanoidei) (Pisces). Vie
et Milieu, 13(4): 729-740.

. 1963b. Sur la nomenclature et la syste-

matique du genre Hemiodus Miiller (Pisces,
Characoidei). Bull. Mus. nat'l. Hist, nat., ser.
2, 35(6): 598-605.

1964. New species of hemiodin chara-

coid fishes fomiing the Hemiodopsis quadri-

maculatus group. Tropical Fish Hobbyist,

November 1964: 11-15, 69-74.
Gill, T. 1878. On a remarkable generic type of

characins. Ann. Mag. Nat. Hist., ser. 5, 2:

112.
Greenwood, P. H., D. E. Rosen, S. H. Weitz-

MAN, AND G. S. Myers. 1966. Phyletic
studies of teleostean fishes, with a provisional
classification of li\ing forms. Bull. Amer.
Mus. nat. Hist., 131(4): 339-456, pis. 21-23.

Kner, R., and F. Stelndachner. 1863. Eine
Uebersicht der ichthyologischen Ausbeute des
Hcrrn Professors Dr. Mor. Wagner in Central-
Amerika. Sitzungsber. Akad. Wiss. Miinchen,
1863, vol. 2, pp. 220-30 (attributed in text
to Kner and Steindachner).

Knoppel, H. a. 1972. Zur Nahrung tropischer
Siisswasserfische aus Siidamerika- Einige aus-
gewiihlte Arten der Anostomidae, Curimati-
dae, Hemiodidae imd Characidae (Pisces,
Characoidei). Amazoniana (Kiel), 3(2):
231-246.

Mago Leccia, F. 1970. Lista de los peces de
Venezuela, inchuendo un estudio preliminar
sobre la ictio-geografia del pais. Ministerio
de Argricultura y Cria (Caracas). 283 pp.

Monod, T. 1969. Le complexe urophore des
teleosteens. Mem. Inst. Fond. Afrique Noire
(Dakar), No. 81: 1-705.

MiJLLER, J. 1842. Beobachtungen iiber die
Schwimmblase der Fische, mit Bezug auf
einige neue Fischgattungen. Ber. Akad. Wiss.
Berlin, 1842: 177-186, 202-210.

Nelson, G. J. 1969. Gill arches and the phylo-
geny of fishes, with notes on the classification
of vertebrates. Bull. Amer. Mus. nat. Hist.,
131(4): 475-552, pis. 79-92.

Regan, C. T. 1911. The classification of the
teleostean fishes of the order Ostariophysi. —
L Cyprinoidea. Ann. Mag. Nat. Hist., ser. 8,
8: 13-32, pi. 2.

Roberts, T. R. 1967. Tooth formation and re-
placement in characoid fishes. Stanford Ichth.
Bull., 8(4): 231-47.

. 1969. Osteology and relationships of

characoid fishes, particularly the genera
Hepsetus, SaJminus, Hoplias, Ctcnohicius,
and AcestwrhijncJuis. Proc. California Acad.
Sci., 35(15): 391-500.

. 1971. Micromischodtis sugiUatus, a new

hemiodontid characin fish from Brazil, and its
relationship to the Chilodontidae. Breviora,
No. 367; 1-25.

. 1972. Ecology of fishes in the Amazon

and Congo basins. Bull. Mus. Comp. Zool.,
143(2): 117-47.

. 1973. Osteology and relationships of

the Prochilodontidae, a South American family
of characoid fishes. Bull. Mus. Comp. Zool.,
145(4): 213-35.

. 1974. Dental polymoiphism and sys-

tematics in Saccodon, a neotropical genus
of freshwater fishes (Parodontidae, Chara-
coidei). J. Zool. (London), 173(3): 303-
321.

436 Bulletin Museum of Comparafive Zoologij, Vol. 146, No. 9

ScHULTZ, L. P., AND C. MiLES. 1943. Descoptions
of a ne\\' genus and a new species of Paro-
dontinae, characinid fishes from South
America. J. Washington Acad. Sci., 33(8):
251-255.

Steindachxer, F. 1888. Ichthyologische Beit-
riige (XIV). Sitzungsber. math.-nat. CI.
Akad. Wiss. (Wien), 96(2): 56-68, pis. 1-4.

Travassos, H. 1951. Contribuicao ao estudo da
subordem Characoidei Berg, 1940 - VII.
Estudo da regiao opercular de tres especies
de Parodontinae Eigenmann, 1910 (Actino-
pterygii - Cypriniformes ) . Bol. Mus. Nac.
(Rio de Janeiro), zool., No. 103: 1-17, 7 pis.

. 1952. Contribuicao ao estudo da sub-

ordem Characoidei Berg, 1940-VIII. Estudo
de alguns ossos da regiao oromandibular de
tres especies de Parodontinae Eigenmann,
1910. ( Actinopterygii - Cypriniformes). Bol.
Mus. Nac. (Rio de Janeiro), zool.. No. 108:
1-17.

Weitzman, S. H. 1962. The osteology of Bnjcon
meeki, a generalized characid fish, with an
osteological definition of the family. Stanford
Ichthyological Bulletin, 8(1): 1-77.

Wiley, M. L., and B. B. Collette. 1970. Breed-
ing tubercles and contact organs in fishes:
their occurrence, structure, and significance.
Bull. Amer. Mus. nat. Hist., 143: 143-216.

antorbital

Hemiodontidae and Parodontidae • Roberts 437

FIGURES

supraorbital epiphyseal bar

ethmoid
vomer

lateral ethmoid

1 mm

frontal
fronto-parietal fontanel

sphenotic pterotic

epiotic

supraoccipital

exoccipital

posttemporal fossa

parietal

Figure 1. Hemiodus semitaeniatus, 55.8 mm. Cranium with nasal bone, antorbital and supraorbital of the right
side in place (dorsal view).

frontal

posttemporal fossa
pterotic

ethmoid

vomer

lateral ethmoid

parietal

rhinosphenoid

orbitosphenoid

pterosphenoid

/ prootic

parasphenoid
I 1 mm I intercalar

Figure 2. Hemiodus semitaeniatus, 55.8 mm. Cranium (lateral view).

supraoccipital

basioccipital
lagenar capsule

438 Bulletin Museum of Comparative Zoology, Vol. 146, No. 9

orbitosphenoid
rhinosphenoid

Figure 3. Hemiodus quadrimaculatus, 35.0 mm. An-
terior portion of cranium (lateral view); inset, ethmoid
bone (dorsal view).

vomer

ethmoid

lateral ethmoid

pterosphenoid
orbitosphenoid
frontal

sphenotic

pterotic

1 mm

parasphenoid
infrapharyngobranchial 1

hyomandibular fossa

auditory foramen
exoccipital

lagenar
capsule

basioccipital

intercalar

prootic

Figure 4. Hemiodus semitaeniatus, 55.8 mm. Cranium with infrapharyngobranchial 1 of the right side in place

(ventral view).

Hemiodontidae and Parodontidae • Roberts 439

supraoccipital

parietal

epiotic

posttemporal fossa

pterotic

foramen magnum
cavum sinus imparis

. . .1 , -^^ I / parasphenoid

basioccipital ^^ ^-^ *^ "^

Figure 5. Hemiodus semitaeniatus, 55.8 mm. Cranium (occipital view).

antorbital
premaxillary

maxillary

supraorbital

nasal

dentary

articular

angular

quadrate

I 1 "^m I

preopercle

interopercle

opercle

subopercle

Figure 6. Hemiodus semitaeniatus, 55.8 mm. Jaws, facial bones, and suspensorium (lateral view).

440 Bulletin Museum of Comparative Zoology, Vol. 146, No. 9

premaxillary

hyomandibular

dentary

coronomeckelian bone

symplectic

opercle

preopercle
subopercle

1 mm

inter opercle

Figure 7. Hemiodus semitaeniatus, 55.8 mm. Jaws, suspensorium and opercular bones (medial view).

palatine

subopercle

symplectic

1 mm

interopercle

Figure 8. Hemiodus quadrimaculatus, 35.0 mm. Lower jaw, suspensorium and opercular bones (medial view).

Hemiodontidae and Parodontidae • Roberts 441

premaxillary

maxillary

Figure 9. Hemiodus semitaeniatus, 55.8 mm. Upper
jaw showing functional and replacement teeth (medial
view).

Figure 11. Hemiodus. Relationship of quadrate to
preopercle (lateral view of bones from the right side).
Upper: H. quadrimaculatus, 35.0 mm. Lower: H. semi-
taeniatus, 55.8 mm.

premaxillary

basihyal

hypohyals

ceratohyal

urohyal

maxillary •

Figure 10. Hemiodus quadrimaculatus, 35.0
mm. Upper jaw showing functional and re-
placement teeth (medial view).

epihyal
interhyal

Figure 12. Hemiodus semitaeniatus, 55.8 mm. Upper: hyoid
bar and branchiostegal rays (lateral view). Lower: hyoid arch
and urohyal (dorsal view).

442 Bulletin Museum of Comparative Zoology, Vol. 146, No. 9

hypobranchials

ceratobranchials

epibranchials

basibranchials

1 mm

infrapharyngobranchials

upper tooth plates

ceratobranchial 5

lower toothplate

Figure 13. Hemiodus semltaeniatus, 55.8 mm. Branchial arches (dorsal view, with lateral elements of the right
side removed).

Hemiodontidae and Parodontidae • Roberts 443

1 mm

Figure 14. Hemiodus microlepis, 235 mm. Series of gill rakers from leading edge of ceratobranchiai 1; rakers of
trailing edge are nearly identical mirror images (dorsal view).

neural complex

claustrum
scaphium

intercalarium

transverse process of third vertebra
neural arch of third vertebra

neural arch and spine
of fourth vertebra

lateral process of

second vertebra

1 mm

basapophysis of fifth vertebra
' pleural rib

tripus Qg suspensorium

Figure 15. Hemiodus semitaeniatus, 55.8 mm. Weberian apparatus (lateral view).

444 Bulletin Museum of Comparative Zoology, Vol. 146, No. 9

posttemporal

extrascapular

mesocoracoid

coracoid

supracleithrum

postcleithra 1-3

I 1 mm I

proximal
radials 1-4

pelvic rays 1-11
pelvic splint .

pelvic bones

I 1 mm I

ischiac process

radials 1-4

Figure 17. Hemiodus semitaeniatus, 55.8 mm. Pelvic
girdle (ventral view).

Figure 16. Hemiodus semitaeniatus, 55.8 mm. Right
half of pectoral girdle (medial view).

urostyle
modified neviral process

neural spines

epurals 1-2

complex ural centrum

hemal spines

j 1 mm I

parhypural

uroneurals 1-2

Figure 18. Hemiodus semitaeniatus, 54.4 mm. Caudal skeleton (lateral view)

Hemiodontidae and Parodontidae • Roberts 445

pt erotic

ethmoid

fronto-parietal fontanel

sphenotic parietal

Figure 19. Argcnectes longiceps, 198 mm. Cranium (dorsal view).

epiotic

supraoccipital

exoccipital
basioccipital

posttomporal fossa

frontal

sphenotic

pterotic

parietal

ethmoid

lateral ethmoid

rhinosphenoid

orbitosphenoid

pterosphenoid

parasphenoid

hyomandibular fossa

subtemporal fossa

Figure 20. Argonectes longiceps, 198 mm. Cranium (lateral view).

supraoccipital

posttemporal fossa
epiotic spine

epiotic

exoccipital

pterotic spine

basioccipital
lagenar capsule

int ere alar

446 Bulletin Museum of Comparative Zoology, Vol. 146, No. 9

auditory foramen pterotic

lateral ethmoid

I 5 mm ]

frontal

orbitosphenoid

pterosphenoid

parasphenoid

pterotic spine

epiotic

exoccipital

basioccipital

intercalar
subtemporal fossa

sphenotic

Figure 21. Argonectes longiceps, 198 mm. Cranium (ventral view).

hyomandibular fossa

parietal

vomer

ethmoid
lateral etiimoid

epiphyseal bar
frontal

fronto-parietal fontanel

5 mm

sphenotic

pterotic

epiotic

exoccipital

basioccipital
supraoccipital

posttemporal fossa

Figure 22. Blvibranchia protractila, 115.2 mm. Cranium (dorsal view).

Hemiodontidae and Parodontidae • Robeiis 447

frontal

ethmoid

sphenotic Pterotic parietal

vomer
lateral ethmoid

rhinosphenoid

orbitosphenoid

pterosphenoid

parasphenoid
I 5 mm I

epiotic

prootic intercalar
infrapharyngobranchial 1

supraoccipital

posttemporal
fossa

exoccipital

basioccipital

Figure 23. Bivibranchia protractila, 115.2 mm. Cranium (lateral view).

lateral ethmoid

auditory foramen
sphenotic

ethmoid

vomer

orbitosphenoid

parasphenoid

5 mm

pterosphenoid

hyomandibular fossa

pterotic

intercalar

exoccipital

basioccipital

epiotic
subtemporal fossa

prootic

Figure 24. Bivibranchia protractila, 115.2 mm. Cranium (ventral view).

448 Bulletin Museum of Comparative Zoology, Vol 146, No. 9

supraorbital

antorbital

premaxillary

\ 1

maxillary

dentary articular

1 mm

nasal

angular

quadrate
interopercle

opercle

subopercle

preopercle

Figure 25. Bivibranchia protractila, 39.9 mm. Lateral bones of skull (lateral view) (upper jaw of specimen
slightly damaged).

supraorbital

nasal

I 5 mm I

Figure 26. Argonectes longiceps, 198 mm. Nasal, antorbital and circumorbital bones (lateral view).

Hemiodontidae and Parodontidae • Roberts 449

■>

m
a>

c
o

o
a>

■o

c
CO

3
(0

w>
(0

E
E

Q.
p

c
o
p>

o>
iJZ

450 Bulletin Museum of Comparative Zoology, Vol. 146, No. 9

Hemiodontidae and Parodontidae • Roberts 451

c
_>_

d)
c
o

"to
>.

J=
o

■a

E
E

CX3

Q.
O)
o

c
o

£0
Q)

c
o

CO
<D .

LL >

452 Bulletin Museum of Comparative Zoology, Vol. 146, No. 9

ii -D

u *;

£. —

il o

E-2

00 *;

O) o

1- o

n —

Q. CO

O 0}

O O)

■?^

C i-

O (0

-- x:

i> .

•*^ i-

O (1)

£ 5

S o

o —

" 15

2^^

3 c c

o) CO o

Hemiodontidae and Parodontidae • Roberts 453

•i?.2

ichial 5, e
ted upper a
ique poster

TO .5 J3

m. Ceratobr
3 and assoc
e (slightly o

■I

, 39.9 nni
branchial
right sid

^LO TO O CD

. := cn-c

nchial

tract
aryn
of t

cu o -5 CO

t. >- D. m

^ CL TO i!i

O >- TO

OJ TO C "q.

\chi

Of i

othi

c O

rt s

¥ rt

bra
ion

ll t

b£'^

— . -^ TO

c a

^ 'X o

^-o

"sal

^ o

- >-

a"^

t^ TO

<^ TO -5

t— 1

— Q.

2-g^^

=j c m 5

D) TO g dj
LL XI ° >

/n "■

<A ^

X 1 o

>^ if >»

;f x:

55*^'

o>
at
CO

■2

*3
O
TO

\\^

.^ \\ /

SS^

^^w\

•a

x--*^ \

is --;-

•§

<d.2

w >

O TO
D) O

ir -o

454 Bulletin Museum of Comparative Zoology, Vol. 146, No. 9

epurals 1-3 uroneurals 1-2

urostyle
modified neural process

neural spines

complex
ural centrum

hemal spines

parhypural

5 mm

Figure 38. Argonectes longiceps, 198 mm. Caudal skeleton (lateral view).

epurals 1-3

urostyle

modified neural process
neural spines

complex ural centrum

hemal spines

1 mm

parhypural

uroneurals 1-2

Figure 39. Bivibranchia protractile, 39.9 mm. Caudal skeleton (lateral view).

Hemiodontidae and Parodontidae • Roberts 455

fronto-parietal fontanel

frontal

vomer

ethmoid

lateral ethmoid

epiphyseal bar

1 mm

parietal

Figure 40. Anodus melanopogon, 55.0 mm. Cranium (dorsal view).

lower limb of
posttemporal bone

exoccipital

basioccipital

supraoccipital
epiotic

sphenotic pterotic posttemporal fossa

sphenotic pterotic parietal

ethmoid

hyomandibular fossa
frontal

vomer

lateral ethmoid

rhinosphenoid
orbitosphenoid

parasphenoid

pterosphenoid

1 mm

supraoccipital

posttemporal fossa

epiotic

prootic
Figure 41. Anodus melanopogon, 55.0 mm. Cranium (lateral view).

intercalar
auditory foramen

exoccipital

basioccipital

456 Bulletin Museum of Comparative Zoology, Vol. 146, No. 9

VrocUe parasphenold

Uleral ethmoid

orblloaphenold

hyomandlbular Toasa

Figure 42. Anodus melanopogon, 55.0 mm. Cranium (ventral view).

lower Umb oi
^^:Z>^^ po«Hempor«l booc

1 mm

Figure 43. Anodus melanopogon, 55.5 mm. Otolithis of left side. A. Lapillus (ventral view). B. Asteriscus (lateral
view). C. Sagitta (lateral view).

Hemiodontidae and Parodontidae • Roberts 457

mesopterygoid

hyomandibular

ectopterygoid
nasal

supraorbital

antorbital
palatine

premaxillary

dentary

maxillary

articular

angular

1 mm

quadrate

symplectic

preopercle

interopercle

Figure 44. Anodus melanopogon, 55.0 mm. Lateral bones of skull (lateral view).

opercle

subopercle

opercle

subopercle

hyomandibular

mesopterygoid

ectopterygoid

palatine
premaxillary

maxillary

dentary

articular
coronomeckelian bone

quadrate

angular

symplectic

Figure 45. Anodus melanopogon, 55.0 mm. Jaws, suspensorium and opercular bones (medial view).

preopercle

interopercle

458 Bulletin Museum of Comparative Zoology, Vol. 146, No. 9

supraorbital

nasal

antorbital

premaxillary \ 1^^

i%.

iHcixiiiary

''~Sv/^^^^^>s.

cr^f^"^"*"^

'^m^v^-^ss^

dentary

^^^

/'"'^^^^^^^

■^^ ■r"^-^

' / — ^
articular /

S-^?-^-^-^

angular

quadratf

1 5 mm 1

opercle

preopercle

Figure 46. Gasterotomus latior, 76.7 mm. Lateral bones of skull (lateral view).

subopercle

interopercle

palatine

premaxillary

maxillary

ectopterygoid

mesopterygoid
metapterygoid

hyomandibular
opercle

dentary
coronomeckelian bone

articular

angular

r quadrate

I 5 mm I ^ „ , ^.

I 1 symplectic interopercle

Figure 47. Gasterotomus latior, 76.7 mm. Jaws, suspensorium and opercular bones (medial view).

Hemiodontidae and Parodontidae • Roberts 459

Infraphu-rniobranrhlal I

cplbrftnchUI ]

bnncUoalpgal ray* I-S

Figure 48. Anodus melanopogon, 55.0 mm. Hyoid arch, urohyal bone and first branchial arch (lateral view).

infrapharyngobranchials 1-3.

upper pharyngeal toothplate

epibranchials 1-4

I 1 mm I

lower pharyngeal toothplate

ceratobranchials 4 and 5

Figure 49. Anodus melanopogon, 55.0 mm. Upper half of branchial
arches (dorsal view).

460 Bulletin Museum of Comparative Zoology, Vol. 146, No. 9

basibranchials

hypobranchials 1-3

I 1 mm I

ceratobranchials 1-5

Figure 50. Anodus melanopogon, 55.0 mm. Lower half of branchial arches (dorsal view)

xu..uAxianja.aximuJaiUUJjLu.i.^^ '" " '''■"'''''''''■'''''•'''■>■'•<' '^^^^^^^'J'UUjmu^uuUU^

I 1 mm .

Figure 51. Anodus elongatus, 180 mm. Gill raker from leading edge of ceratobranchial 1 showing ctenii.

Hemiodontidae and Parodontidae • Roberts 461

CO
Q.
Q.
CO

a
a>

E
E
o
id

IT)

c
o

a
o

-5

•o
o

c
"?

w
to

462 BuUetin Museum of Comparative Zoology, Vol 146, No. 9

U-)

Q.
O

-2

C
CD

■=3: -o

■^

'5)

L o

supraorbital

ethmoid

Hemiodontidae and Parodontidae • Roberts 463

sphenotic

parietal

lateral ethmoid frontal

\ 1 mm I

epiotic

supraoccipital

exoccipital

posttemporal fossa

pterotic
Figure 56. Saccodon wagneri (dental morph IV), 55.4 mm. Cranium with right supraorbital in place (dorsal view).

ethmoid

frontal

sphenotic

pterotic

parietal

lateral ethmoid

orbitosphenoid

pterosphenoid

supraoccipital

epiotic

posttemporal
fossa

exoccipital

basioccipital

1 mm

parasphenoid prootic auditory foramen

Figure 57. Saccodon wagneri (dental morph IV), 55.4 mm. Cranium (lateral view).

464 Bulletin Museum of Comparative Zoology, Vol. 146, No. 9

hyomandibular fossa

ethmoid

lateral ethmoid

auditory foramen

parasphenoid
frontal

1 mm

orbitosphenoid

pterosphenoid

intercalar

exoccipital

basioccipital

sphenotic pterotic

Figure 58. Saccodon wagneri (dental morph IV), 55.4 mm. Cranium (ventral view).

epiotic
subtemporal fossa

prootic

parietal

epiotic

intercalar

exoccipital

supraoccipital

posttemporal fossa

pterotic

foramen magnum
cavum sinus imparls

basioccipital

I 1 mm I

Figure 59. Saccodon wagneri (dental morph IV), 55.4 mm. Cranium (occipital view).

1 mm

Figure 60. Saccodon wagneri (dental morph I), 52.0
mm. Otoliths. A. Lapillus (ventral view). B. Sagitta
(lateral view). C. Asteriscus (lateral view).

Hemiodoxtidae and Pahodontidae • Roberts 465

extrascapular

posttemporal

supraorbital
lateral ethmoid

nasa
ethmoid

antorbital

premaxillary

maxillary

opercle

subopercle

1 mm

branchiostegal rays 1-4
Figure 61. Saccodon wagneri (dental morph I), 57.2 mm. Lateral bones of skull (lateral view)

Figure 62. Parodontidae. Nasal, antorbital and circumorbital bones. Left, Saccodon wagneri (dental morph 1),
57.2 mm. Right, Parodon guyanensis, 38.5 mm.

466 Bulletin Museum of Comparative Zoology, Vol. 146, No. 9

Hemiodontidae and Parodontidae • Roberts 467

supramaxillary
premaxillary

ectopterygoid

mesopterygoid
metapterygoid

hyomandibular

opercle

quadrate

preopercle

interopercle ' subopercle

Figure 64. Saccodon wagneri (dental morph IV), 55.4 mm. Jaws, suspensorium and opercular bones (lateral
view) (teeth removed).

opercle

premaxillary

mesopterygoid
ectopterygoid
palatine

maxillary

dentary
coronomeckelian bone

articular

subopercle

preopercle
interopercle

Figure 65. Saccodon wagneri (dental morph IV), 55.4 mm. Jaws, suspensorium and opercular bones (medial
view) (teeth removed).

468 Bulletin Museum of Comparative Zoology, Vol. 146, No. 9

—:. c

5ro

.5^ E

_ o

(0 -^

1—

<D CO

■*-• i-

CO (1)

""" CO

^ ~

Q) £

<r> o

\ «\

■^ o

i 5 ^

^. CO flj

1.^^

T3 —

C OJ

CO o

U> TO

5 ®

5 Q-

-3 .

^ It)

E "

^-1

^.E

in 3

- (U,-:.

> ^ S

— CO

._ k_

■|E S

—

O TO C

t O)

C ><

CO > -^

c ^ -a

■o c CO

— '— o

gner

jsual

(ind

5 c £

S 3 (U

c cS

O <X)

"g.i g

o O o

CO CO c

. <n ^

1^ !E «_
«Oh o

2f . CO

3 <D CB

i- 1 S3

E i

■<r

^-^

C11

•*-•

»^

«J

S E

-S -^

O CO

o -r-

o w 5

o g 0)

to CO >

GO — _

. 53 S

CD Q. C

CD Q. O

3 r-^ CO

D) t CO

E E £

u. ^

E
E

C3)

C31

<.i

Cl)

(1)

Hemiodontidae and Parodontidae • Roberts 469

opercle

I palatine

premaxillary

maxillary

dentary
coronomeckelian bone articular angular

1 mm [

quadrate

symplectic

subopercle
preopercle

interopercle

Figure 70. Parodon guyanensis, 38.5 mm. Jaws, suspensorium and opercular bones (medial view).

premaxillary

Figure 71. Parodon caliensis, 63.5 mm. Upper jaw and
portion of lower jaw showing dentition (lateral view).
Inset on right, functional tooth from premaxillary (me-
dial view).

symplectic

quadrate

Figure 72. Saccodon wagneri (dental morph iV), 55.4
mm. Relation of quadrate to preopercle (lateral view).

470 Bulletin Museum of Comparative Zoology, Vol 146, No. 9

dorsal hypohyal ceratohyal epihyal

ventral hypohyal

branchiostegal rays 1-4

Figure 73. Saccodon wagneri (dental morph IV), 52.0 mm. Hyoid arch with basihyal removed (lateral view) and
urohyal bone (lateral and dorsal views).

basihyal

ceratohyal
basibranchials 1-3

epihyal

hypobranchials 1-3

interhyal

infrapharyngobranchials 1-3

epibranchials

upper pharyngeal toothplates

ceratobranchials

lower pharyngeal toothplate

ceratobranchial 5

Figure 74. Saccodon wagneri (dental morph IV), 55.4 mm. Hyoid and branchial arches (dorsal view).

p
I

Hemiodontidae and Parodontidae • Roberts 471

neural complex

claustrum
scaphium

inter calarium

lateral process

of centrum 2

I 1 mm I

transverse process of third vertebra
neural arch of third vertebra

neural arch and spine of fourth vertebra

neural spines

tripus

pleural ribs

neural prezygapophysis

basapophyses

Figure 75. Saccodon wagneri (dental morph IV), 55.4 mm. Weberian apparatus (lateral view).

posttemporal

postcleithra

supracleithrum

cleithrum

first pore-bearing scale of lateral line

postcleithra

1 mm I

coracoid

scapula

Figure 76. Saccodon wagneri (dental morph IV), 55.4 mm. Right half of pectoral girdle (lateral and medial views).

472 Bulletin Museum of Comparative Zoology, Vol. 146, No. 9

pelvic rays

pelvic splint

pelvic bone

ischiac process
pelvic radials 1-4

I 1 mm I

Figure 77. Saccodon wagneri (dental morph IV), 55.4 mm. Left half of pelvic girdle (ventral view).

epurals 1-2

uroneurals 1-2
urostyle

modified neural process
neural spines

complex xu-al centrum

hemal spines

1 mm

parhj^ural

Figure 78. Saccodon wagneri (dental morph IV), 55.4 mm. Caudal skeleton (lateral view).

us ISSN 0027-4100

Isulletin of the

Museum of

Comparative

Zoology

The Taphonomy and Paleoecology

of Plio-Pleistocene Vertebrate

Assemblages East of Lake Rudolf, Kenya

ANNA K. BEHRENSMEYER

HARVARD UNIVERSITY

CAMBRIDGE, MASSACHUSETTS, U.S.A.

END OF VOLUME

VOLUME 146, NUMBER 10
21 FEBRUARY 1975

PUBLICATIONS ISSUED

OR DISTRIBUTED BY THE

MUSEUM OF COMPARATIVE ZOOLOGY

HARVARD UNIVERSITY

Breviora 1952-

BULLETIN 1863-

Memoirs 1864-1938

JoHNSONiA, Department of Mollusks, 1941-

OccAsiONAL Papers on Mollusks, 1945-

SPECIAL PUBLICATIONS.

1. Whittington, H. B., and E. D. I. Rolfe (eds.), 1963. Phylogeny and
Evolution of Crustacea. 192 pp.

2. Turner, R. D., 1966. A Survey and Illustrated Catalogue of the Teredini-
dae ( Mollusca: Bivalvia). 265 pp.

3. Sprinkle, J., 1973. Morphology and Evolution of Blastozoan Echinoderms.
284 pp.

4. Eaton, R. J. E., 1974. A Flora of Concord. 236 pp.

Other Publications.

Bigelow, H. B., and W. C. Schroeder, 1953. Fishes of the Gulf of Maine.
Reprint.

Brues, C. T., A. L. Melander, and F. M. Carpenter, 1954. Classification of
Insects.

Creighton, W. S., 1950. The Ants of North America. Reprint.

Lyman, C. P., and A. R. Dawe (eds.), 1960. Symposium on Natural
Mammalian Hibernation.

Peters' Check-list of Birds of the World, vols. 2-7, 9, 10, 12-15.

Proceedings of the New England Zoological Club 1899-1948. (Complete
sets only.)

Publications of the Boston Society of Natural History.

Price list and catalog of MCZ publications may be obtained from Publications
Office, Museum of Comparative Zoology, Harvard University, Cambridge, Massa-
chusetts, 02138, U.S.A.

THE TAPHONOMY AND PALEOECOLOGY OF PLIO-PLEISTOCENE
VERTEBRATE ASSEMBLAGES EAST OF LAKE RUDOLF, KENYA

ANNA K. BEHRENSMEYER'

CONTENTS

Abstract 474

Introduction 474

The Taphonomy of Macro-Vertebrate

Assemblages — 476

Factors Relating to Mode of Death 476

Factors Relating to Weathering and

Decomposition - 479

Transport and Burial 480

Diagenetic Factors 482

Conclusions 482

Characteristics of Recent Bones as Sedi-
mentary Particles - 483

Properties of Bones as Sedimentary

Particles 485

Measurements of Bone Size and Density 485
Relative Dispersal Potential of Bones - _ 488
The Hydraulic Equivalence of Bones and

Quartz Grains 490

Experiments in Bone-settling

Velocities 492

The Hydraulic Equivalents of

Fossil Bones 493

Current Velocities and Bone

Transport 495

Additional Factors Affecting Bone

Transport in Natural Situations 497

Bottom Morphology and Current

Profile 498

Fluid Densitv 498

Burial Potential 498

Conclusions: Bones as Sedimentary

Particles 499

Sedimentary Environments of the Koobi Fora

Formation, East Rudolf 500

Geologic Setting 500

Regional Stratigraphy 503

RecLMit Limnology 505

Stratigraphy of tlie Koobi Fora Formation 507
Sedimentary Environments of the Fossil

Vertebrate Localities 509

Designation of Sampling Localities 509
Method of Geologic Analysis 509

^ Department of Paleontology, University of
California, Berkeley, California 94720

Sedimentary Environments of the

Sample Localities 514

Discussion and Conclusions 525

Sorting in Bone Assemblages of the Koobi

Fora Formation 526

Sampling of Bone Assemblages 526

Choosing the Sample Areas 527

Sample Size 528

Method of Representing Fossil

Abundance 528

Characteristics of the Bone

Assemblages 529

Identification of Bones 530

Significance of the Frequency Data 531
Comparisons of Overall Bone

Concentrations — 531

Relative Abundance of Skeletal Parts 532
Correlations Based on Bone

Abundance 532

Factor Analysis of the Bone

Assemblages 534

Comparisons with Voorhies Groups 536

Single Skeleton Comparisons 536

Discussion of Evidence for Transport

Sorting 537

Additional Aspects of the Bone

Assemblages 538

The Reptilian Assemblages 540

Conclusions Concerning the Bone

Assemblages 540

Faunal Assemblages of the Koobi Fora

Formation 541

Method of Identification 541

Abundance of Vertebrate Classes 542

Reptiles 543

Mammals — - 545

Discussion of the Most Abundant

Mammalian Groups 547

Abundances of Selected Mammalian

Groups - 549

Suids - , 549

Equids _ - 551

Bovids - 551

Hippos - 552

Conclusions Regarding the Faunal

Assemblages 553

Bull. Mus. Comp. Zool., 146(10): 473-578, February, 1975 473

474 Bulletin Museum of Comparative Zoology, Vol. 146, No. 10

Paleoecology of the Vertebrate Assemblages

of the Koobi Fora Formation 554

Ecological Comparisons of the Samples - 557

Aquatic and Terrestrial Faunas 557

Open and Closed Habitat Mam-
malian Faunas 557

Comparisons of Koobi Fora Forma-
tion Faunas and Recent Terrestrial

Faunas 558

Hominid Paleoecology 561

East Rudolf in Relation to Other Studies

in Vertebrate Paleoecology 562

Summary 564

General Conclusions 564

Taphonomy 564

Paleoecology 565

Conclusions for the Vertebrate Assem-
blages of the Koobi Fora Formation,

East Rudolf 565

Acknowledgments 566

References 567

Appendix 1. Bone Measurements: Density,

Volume and Weight 570

Appendix 2. Calculation of Hydraulic

Equivalence 573

Plates 574

Abstract. The object of this study is to show
that paleoecologic information can be derived from
fossil vertebrate assemblages fragmented prior to
burial if a taphonomic history can be established
for these assemblages. Such paleoecologic informa-
tion can lead to knowledge of the character and
evolution of vertebrate communities through time.
Within the Koobi Fora Formation of East Rudolf,
Kenya, vertebrate bones are preserved in fluvial
and lacustrine depositional environments through
a time span between about 3.0 and 1.2 million
years Before Present. A total of seven bone
assemblages were collected from lake margin, chan-
nel and floodplain deposits. Detailed geological
studies were done in the seven laterally extensive
sedimentary environments sampled for bones. The
bone samples were taken from surface lag con-
centrations utilizing widely spaced 10 X 10 meter
stjuares. The seven assemblages were analyzed
for numbers of different skeletal parts and verte-
brate taxa. Theoretical and experimental evidence
for the characteristics of bones as sedimentary
particles formed the background for the analysis
of the East Rudolf assemblages. The flume experi-
ments of M. Voorhies (1969), plus measurements
of whole-bone densities and settling velocities,
supported the hypothesis that bones form distinc-
tive dispersal groups when sorted by various
taphonomic processes, especially fluvial transport.
Therefore, the dispersal groups represented in
fossil assemblages can indicate the taphonomic
histories of these assemblages. Mammalian skeletal
parts from the seven sample assemblages were
analyzed for the percentages of different dispersal
groups. The numbers of different parts were also

compared with the ratios of parts in a single,
average skeleton. The fossil assemblages from
fluvial deposits showed distinctive alteration from
the ratios of parts in undisturbed skeletons. This
resulted from the absence of the lighter bones
which form the most easily transported dispersal
group. The delta margin assemblages showed little
alteration of bone percentages from those in an
undisturbed skeleton, indicating little selective
sorting prior to burial in this environment. The
bones, hence the faunas, did not appear to be sub-
stantially mixed or transported from the general
ecological province inhabited by the living animals.
This information allows paleoecological inferences
to be drawn from faunal distributions in the
different environments. Faunal assemblages were
analyzed for the patterns of occurrence of reptilian
and mammalian groups. Aquatic and nonaquatic
vertebrates (excluding fish) are preserved in ap-
proximately equal abundance in the delta margin
deposits. Nonaquatic forms are significantly more
abundant in the fluvial deposits. Several members
of the mammalian faunas, particularly hippos,
suids and bovids, show differential abundances in
the two environments. Certain members of the
Bovidae and Suidae have patterns of occurrence
indicating preferences for delta margin or fluvial
habitats. The paleoecologic information from the
more abundant vertebrate groups helps to establish
an ecological framework for the hominid fossils
from the Koobi Fora Formation.

INTRODUCTION

The primary object of this study is to
determine the paleoecology of vertebrate
faunas that occur in the East Rudolf de-
posits of northern Kenya. Taphonomic
analysis will provide the background for
the paleoecologic interpretations. The citi-
cial hnk between a fossil assemblage and
the original ecosystem from which it was
derived lies in the taphonomy of the as-
semblage, the history of its passage from
the biosphere into the lithosphere. When
bones of different animals are found to-
gether in a particular sedimentary deposit
it is essential to know whether these bones
were buried together because they were
transported together (perhaps from dif-
ferent points of origin), or whether their
close association indicates that the animals
lived and died in the same habitat. The
taphonomic history of a bone assemblage
can provide this information.

East Rxjdolf Paleoecology • Bchrensmcijer 475

Mctliods for establishing the* taphonomic
liistories of fossil vertebrate assemblages in
East Rudolf are explored and d(>veloped in
this stud)' in order to permit the fossil
faunas to be related to former living verte-
brate communities. These methods apply
specificalK- to the East Rudolf bone as-
semblages, which represent thanatoeocmoses
(death assemblages) of large vertebrates
that were disarticulated, fragmented and
transported prior to burial. However, many
of the conclusions concerning the inter-
action of tliese bones with processes of
transport and weathering will have broad
implications for paleoecologic interpreta-
tions of vertebrate assemblages from other
regions and time periods.

This study consists of two major parts,
the first providing theoretical and experi-
mental models for the second, which
analyzes particular bone assemblages from
the East Rudolf deposits. The first part
discusses factors contributing to bone dis-
persal and destruction in recent East Afri-
can habitats and then examines in detail the
properties of bones as sedimentary particles,
including their dispersal potentials when
subjected to fluid stress and their hydraulic
equivalence to quartz particles. The second
part describes the geologic context of East
Rudolf bone assemblages sampled from
seven different localities, analyzes their
taphonomic histories, and then interprets
the paleoecology of the faunas represented
in each of the seven samples.

Many aspects of the East Rudolf region
proved extremely advantageous as a back-
ground for the study of the taphonomy and
paleoecology of a series of fossil vertebrate
assemblages. The East Rudolf Research
Expedition, led by R. E. Leakey of the
National Museums of Kenya, has been
active in the area since 1967 and has
brought together a large team of scientists
representing a wide range of disciplines.
The collection of fossil vertebrates from the
region as a whole has established the
composition of the local Plio-Pleistocene
faunas and has provided evidence for faunal

succession between about 4.5 and 1.3 my.
B.P. The regional geology has been worked
out through the combined efforts of several
teams of geologists, and the overall stratig-
raphy and dating are reasonably well
established. Archaeological investigations
are providing evidence relating the cultural
artifacts of primitive man to the broader
context of the Plio-Pleistocene faunas and
environments. In addition, the Recent
faunas and environments of East Rudolf
are comparable in many ways to those of
the Plio-Pleistocene and serve as readily
available analogues for the interpretation
of the taphonomy and of the paleoecology
of former time periods. A study of the
recent taphonomy of lake margin areas is in
progress (A. Hill, Bedford College, London)
and this should give further valuable evi-
dence for comparison with the fossil as-
semblages. On a broader scale, geological
and paleontologie information currently
available for sedimentary deposits through-
out East Africa, plus the wealth of data on
the recent ecosystems, have greatly en-
hanced and broadened the scope of this
study.

As an added attraction to all of the other
advantages that characterize East Rudolf,
the area is one of the richest known locali-
ties for fossil man. At present, over 120 speci-
mens have been recovered, and these repre-
sent at least two taxa of contemporaneous,
Plio-Pleistocene hominids. This study pro-
vides a background for the paleoecologic
context of fossil man at East Rudolf in
terms of faunal associations, environments,
and possible habitat separation between the
two forms.

The initial decision to undertake a study
of the East Rudolf bone assemblages was
in part inspired by prc>\'ious, intriguing re-
search in vertebrate paleoecology. The out-
standing works that have helped to shape
many of the viewpoints to be presented
later include: Olson (1952, 195S), Shot-
well (1955, 1963), Clark, Beerbower and
Keitzke (1967), Voorhies (1969), and
Dodson (1971, 1974). Information has been

476 Bulletin Museum of Comparative Zoologij, Vol. 146, No. 10

drawn from various other studies relevant
to the interpretation of assemblages con-
sisting of bones of the larger vertebrates.
These include the investigations of recent
carcass decay and dispersal by Weigelt
(1927) and Schafer (1972). The overall
theoretical background for taphonomy is
derived primarily from Efremov ( 1940,
1953), the founding father of this line of
scientific investigation.

Research on the East Rudolf bone as-
semblages and their relationships to dif-
ferent sedimentary environments was begun
in the summer of 1971. Prior to this, I had
spent two field seasons working on the
stratigraphy and sedimentary environments
of East Rudolf, as well as five weeks at
Lothagam Hill on the southwest side of
Lake Rudolf. Field work on the East
Rudolf assemblages encompassed two field
seasons of three months each in 1971 and
1972. Surface bones associated with dif-
ferent lithofacies were collected according
to a consistent procedure that permitted
later statistical comparisons between as-
semblages. All collecting and bone identi-
fications were done by me or under my
close supervision.

THE TAPHONOMY OF MACRO-
VERTEBRATE ASSEMBLAGES

Many processes can influence the pro-
gression of bones from the living animal to
the final place of burial and fossilization.
All of these must be considered in order to
derive paleoecological information from a
fossil assemblage. Efremov (1940:85) ap-
plied the term taphonomy to this special
area of geological and biological problems,
and specified it as "the study of the tran-
sition (in all its details) of animal remains
from the biosphere into the lithosphere."
Processes included in taphonomy have been
discussed in the general context of verte-
brate assemblages by various authors (e.g.,
Efremov, 1940; 1953; Clark et al, 1967;
Miiller, 1957). As an introduction to the
study of the East Rudolf fossil assemblages,
it will be useful to consider these processes

in some detail where they are relevant to
the East African situation.

Factors Relating to Mode of Death

Causes of death initially determine which
bones and which animals will have a poten-
tial for fossilization. Causes of death may
include predation, disease, physical ac-
cident, poison, starvation and intraspecific
strife (Clark et al, 1967: 115). Predators
have the greatest initial influence on the
widespread dispersal of bones. Animals
that die of other causes are likely to be
similarly dispersed by scavengers, at least
in East Africa, unless the carcass is some-
how protected.

Lion, leopard, cheetah, hyaena, jackal,
man and crocodile are the most important
predator/scavenger members of the com-
munities of large vertebrates in East Africa
today. All of these both hunt and scavenge
(Kruuk, 1972; Van Lawick-Goodall, 1971;
Schaller, 1972). Vultures also are important
contributors to bone dispersal. The effects
of these predators and scavengers on a
carcass are referred to as "carnivore
activity" in the following discussion. Dis-
association of a skeleton can be amazingly
rapid and thorough. Kruuk (1972:126)
reports that a young wildebeest killed by
hyaenas had its parts completely dispersed
from the site in 13 minutes. Both hyaenas
and lions will kill in shallow water and may
have their meal there rather than dragging
the carcass onto land. However, lions often
will drag v/hole carcasses or parts away
from the site of death. Hyaenas may carry
favorite parts long distances (away from
other hyaenas). Kruuk (1972:119) ob-
served hyaenas caching parts of carcasses
in 30-50 cm of standing water, with vari-
able success in retrieving the cache later on.
When a shallow pool dried up, many bones
were exposed, apparently as a result of this
behavior.

The preferences of carnivores for con-
suming particular parts of a carcass have an
important bearing on what would remain to
be fossilized and which bones would be

East Rudolf Paleoecolocy • Behrensmeijer 477

PROCESSES IN VERTEBRATE TAPHONOMY

Accident resulting
in immediate burial

Reworking

Bioturbation

Leaching

Compaction and

movement of sediment
Mineralization

iFOSSIL ^ \
' ASSEMBLAGE |

/
/

I I
/ /
/ /
/ /

(Adapted from Clark et al., 1967)

Figure 1.

likely to remain near the site of death.
Kruuk (1972:126) reports: "If any part of
a corpse is left by hyaenas in both Ngoron-
goro and Serengeti, the skull is most likely
to remain uneaten, followed by the verte-
brae, ribs, pelvis and ends of long leg
bones." Lions will also consume everything
but "horns, teeth and a few bone splinters
and scraps," of small prey such as gazelle
(Schaller, 1972:269). Carnivore activity by
man would probably result in less total
destruction of bones, but limb shafts might
be split for marrow and skulls broken open
for the brain (White, 1955, Brain, 1967b).
It is obvious that a crucial factor in the
number of bones which survive to become
a thanatocoenose (death assemblage) will
be the proportion of carnivores to the num-
ber of food animals at any one time. In
Kruuk's study areas, the number of predators
was high, and the number of carcasses left
intact was low. Observations of carcasses
in the East Rudolf region reveal a fairly
large number of intact or partial skeletons,
reflecting a relatively low number of car-
nivores. The reciprocal relationship be-
tween numbers of carnivores and numbers
of complete bones or carcasses has been

generally noted in East Africa (e.g., Estes,
1967:39).

Not only numbers, but relative sizes of
carnivores as well as their degree of adap-
tation for bone mastication will have im-
portant effects on a thanatocoenose. The
strength of bones with nutritional value
must exceed the crushing force of the jaws
of hyaenas and lions in order to remain
intact, unless there is a surplus of food
(e.g., mass deaths), which would make
bones less attractive food items.

Crocodile predation or scavenging will
tend to bring bones into close association
with depositional environments Init may
totally destroy bones that are consumed.
The proportion of large animals in a
crocodile's diet increases with its size, and
prey can range up to the size of an adult
black rhino (Selous, 1908:201). Crocodiles
generally kill by dragging animals into the
water and drowning th(>iu. They are also
active scavengers, ranging as much as half
a mile from water, and may even compete
with lions for a kill (Cott, 1961:302-303).
Carcasses are torn apart or may be cached
(underwater) until decomposition makes
dismembering easier (Guggisberg, 1972:

478 Bulletin Museum of Comparative Zoology, Vol. 146, No. 10

94). Bones and teeth are demineralized
during digestion, and remaining organic
material (mainly collagen) disintegrates
rapidly after defecation (D. Fisher, per-
sonal communication). Thus, only bones
that are too large or cumbersome to ingest
(e.g., skulls), or that are neglected by the
crocodile, would survive. However, these
would have optimal chances for burial.

Drowning, disease, starvation and other
causes leading to mass deaths can have
different effects on a thanatocoenose. Any
mass death situation is likely to create a
surfeit of food for the local carnivores, and
many skeletons may be left more or less
intact. Schaller (1972:215) notes that lions
do not seem to scavenge from such deaths.
It takes only a short time for carcasses to
mummify in dry conditions and become
unattractive to most scavengers. Once
mummified, a carcass could probably sur-
vive a certain degree of transport without
becoming disarticulated.

Mass death by drowning seems to be
fairly common among the East African un-
gulate species. The social behavior of herd-
oriented species (e.g., zebra, wildebeest)
can result in mass panics in which the
fright of a single animal may cause a group
stampede. Such events often occur at
waterholes, and many animals may be
trampled and drowned (R. Estes, personal
communication). Schaller (1972:215) re-
ports the death of 62 wildebeest in a water-
hole at one time, and 83 in the same place
a few years later. Apparently many herd-
oriented animals also drown during river
crossings. Abel (1912:12) mentions panic
as one of the causes of massed vertebrate
remains, but the idea has not generally
been used in the interpretation of massed
fossil assemblages.

In a discussion of the causes of mass
deaths leading to fossilization, Kurten
(1953:72) favors seasonally occurring floods
or, in a more general sense, linked causes
of death and deposition. Voorhies (1969:
52) supports a catastrophic event of this
kind in the case of the Pliocene Verdigre

Quarry in Nebraska, and suggests large-
scale drought or winter storm followed by
flash-flooding as the cause of the deposit.
Drought may concentrate animals that are
not normally herd oriented. However, in
some cases it would be worth considering
social catastrophes as well as climatic
events in interpreting massed assemblages
of fossil vertebrates.

The drowning of single animals, as in
river crossings, initially provides good con-
ditions for the burial of a whole skeleton or
articulated parts. Scavenging by crocodiles,
and possibly also by fish and turtles, can
destroy such carcasses, and if they float
long enough to decay, various parts may
gradually drop off and be widely dispersed.
However, drowning can lead to the preser-
vation of whole or partial skeletons. As a
consequence, wildebeest, and other non-
aquatic animals, might be among the best
preserved fossils in channel and point-bar
deposits.

The above observations arc relevant to
paleoecological interpretations based on the
preservation of vertebrates in channel and
point-bar deposits. Dodson (1971:69) has
suggested that the excellent, complete
preservation of hadrosaur skeletons in the
channel deposits of the Oldman Formation
(western Canada, Cretaceous) indicates
that a good portion of the hadrosaurs' time
was spent living in the channels. By
analogy, it might be possible to conclude
that wildebeest occupy aquatic habitats,
based on where the most complete skeletons
would be preserved. Tlie fact is that wilde-
beest are terrestrial in their habits, but
occasionally die in channels. In many cases,
the exception (e.g., drowning) may pro-
duce the fossil, while "normal" habits will
leave little or no record. The hadrosaurs
may have been partly aquatic, but their
place and state of preservation in the chan-
nel deposits should be used only in support
of other evidence for aquatic habits.

In general, it appears that two major
kinds of thanatocoenoses can be derived
from modern large- vertebrate communities;

East Rudolf Paleoecolocy • Bchrensmcycr 479

an incomplete, broken and dispersed one
resnlting from carni^•ore aeti\ity, and a
relatively complete one (in terms of whole
bones and associated skeletons) resulting
from mass deaths. TIk^ first can form fossil
assemblages which sample faunas over
periods of months or years, while the
second ma\' lead to assemblages which
sample tlie standing crop of vertebrates at
specific points in time. Intermediate kinds
of thanatocoenoses will depend on the
numbers of carnivores and the effects of
scavenging on the available carcasses.

In East Africa today, many thanatocoe-
noses are composed of broken and in-
complete skeletons, owing to the large
numbers of carnivores. The parts that are
left to continue toward fossilization chiefly
include skulls, horn cores, vertebrae, ribs,
ends of limb bones and teeth, of the larger
onhnals (primarily ungulates). These are
subjected to weathering and transport and
are treated as sedimentary particles by the
various geologic processes. More complete
carcasses occur in situations where they are
not subject to scavenging. From the recent
evidence, it appears that a thanatocoenose
composed of fragmented skeletons is likely
to occur in an area broadly representative
of the habitats of the living animals, al-
though later the bones may be dispersed
from the area by fluvial processes. A
thanatocoenose composed of massed, com-
plete skeletons is more likely to represent
localized conditions of death and/or trans-
port. Thus, a fossil assemblage composed
of bones from a fragmented thanatocoenose
that has not been tramportecl should pre-
serve the best evidence for the paleoecology
of the fauna.

The foregoing discussion applies to the
larger vertebrates, and it is assumed that
the bones of animals of sheep size and
smaller will be much less likely to survive
carnivore activity. However, small verte-
brates such as turtles and fish probably
form thanatocoenoses comparable to those
of the large mammals and reptiles, reflect-
ing both carnivore activity and occasional

mass deaths. Processes leading to thanato-
coenoses composed of small mammals are
not well understood and are worthy of
further investigation. However, since the
East Rudolf deposits have so far yielded a
negligible number of small mammals, such
an investigation will not be undertaken in
this study.

Factors Relating to Weathering
and Decomposition

The biological and chemical properties
of the place where an animal dies will have
an important effect on bone destruction.
Humid surface environments will facilitate
the decay of organic material and will
cause dissolution of bone minerals. Dry
en\'ironments dehydrate the organic com-
ponent of fresh bones, resulting in cracking
and splitting. For the most part, since teeth
have less residual organic matter, they will
survive surface weathering better than
other parts, although large teeth tend to
split when dehydrated.

The rate of decay of muscles and liga-
ments is of interest in determining how
long parts will remain articulated. In the
absence of vertebrate carnivores, insect
activity is an important process in deflesh-
ing a skeleton and is enhanced by warm,
humid, subaerial conditions. In such con-
ditions (summer in South Carolina, U.S.A.),
Payne (1965:597) has observed complete
remo\'al of flesh from the carcass of a baby
pig in eight days when insects were present,
but flesh remained after 100 days when
insects were absent. Voorhies observed
sheep carcasses in the drier conditions of
Nebraska and reports complete disarticu-
lation after 90 days of normal insect
activity. Carcasses of mammals that are
submerged in water may disarticulate in 1
to 3 months (Dodson, 1974:79; Schafer,
1972:21). The evidence suggests that the
most rapid rates of disarticulation owing to
insects and the activity of micro-organisms
may be achieved on land. However, under
certain conditions, bones can remain artieu-

480 Bullciin Mu.scuni uf Comparative 7A)ulu^y, Vul. 146, No. 10

latcd for weeks or months in eitlier sub-
aerial or aquatie en\'ironnients.

Ycry little is known about the textural
characteristics of bones weathered under
different conditions. However, there is
some data on the length of tim(> bones can
survive .surface weathering. \'oorhies (1969:
31) reports that bones left for a year in the
Nebraska climate were soft and cracked,
showing noticeable signs of disintegration.
Bone-weathering experiments in East Africa
show nearly complete destruction in 7 to
8 years (Isaac, 1967:40). Obserxations on
bones in xarious Eiist African game parks
indicate they can last several years, but are
usually in good condition for only a few
months (A. Hill, personal communication).
In the semi-arid climate of Lake Rudolf
(annual rainfall ~ 250 mm [10 inches]),
bones acquire distinctive characters indica-
tive of surface weathering, including flak-
ing, splitting and splintering (Plate 1) during
the first few months of exposure. Bone is
e\idently a very .short-lived material in
surface environments, and must be left in
actively aggrading depositional situations
in order to survive intact. If bones are to
maintain a fresh and unweathered appear-
ance, they must be buried soon after the
death of an animal.

The characteristic appearance of natu-
rally weathered bone surfaces is often dis-
tinguishable after they have been fossilized.
This can provide in\'aluable evidence for
the taphonomic history of a bone or bone
assemblage. Typical surface textures are
shown in Plate 1. Certain types of fractures
also appear to result only from breakage in
fresh bones. These include spiral, fibrous
and sawtooth fractures as shown in Plate 2.
Some work has been dont^ on the character-
istic weathering and fracture patterns of
bones (e.g., Sadek-Kooros, 1966; Brain,
1967a; Reif , 1971 ) . Preliminary experiments
using tumbling mills show that bones can
be abraded and rounded during transport
without extensiv(> fragmentation. Pro-
jections on the bones are usually broken off
and surfaces become smooth and rounded

(C. Jepsen, personal communication). More
experimental work is needed in order to
establish the causes of the observed fracture
patterns and surface textures of recent
bones.

The major factor determining bone ap-
pearance and survival potential under sur-
face conditions appears to be the content of
organic matter. Crystals of hydroxyapatite
which form bone are supported in an ex-
tensi\"e system of organic material. Bones
that ha\'e this organic material removed
artificially can be reduced to powder with
very little force ( a vertebra can be crushed
in one's hand). Bones that have been de-
mineralized, leaving only the organic
material (principally collagen), maintain
their form and can be bent and twisted like
rubber (F. A. Jenkins, Jr., personal com-
munication). Weathered bones that have
obviously lost much of their organic content
can be easily crushed and broken and are
particularly friable when wet, with relative
strength proportional to the compactness
of the bone structure. These bones would
be easily abraded and destroyed in trans-
port situations, while bones with greater
residual organic content would survive
longer under similar conditions.

Bone can be regarded as a very labile
kind of sedimentary particle that is char-
acteristically altered by the geologic pro-
cesses to which it is subjected. Therefore,
bones may re\'eal a great deal about the
influence of weathering and transport on
their pre-burial history.

Transport and Burial

Bones can be transported from the place
of death either within a floating carcass or
as isolated objects that behave as discrete
sedimentary particles. Flotation could lead
to considerable transport of bones away
from the habitat of the living animal with
little damage to bone surfaces. There are
no experimental data as yet to show how
far isolated bones can be transported. How-
ever, a certain amount of information can
be derived from a theoretical consideration

East Rudolf PALi;fjp:fX)Lor,Y • Bnhrensrncycr 481

of the propr-rtif'S of bonos as sedimentary
particles. Tliese will be given detailed
examination in the following chapter, since
an understanding of bone transport is
crucial for paleoecological interpretations
from assemblages of isolated bones such as
those found in the deposits of East Rudolf.

Whole or nearly whole carcasses can be
floated intact to places of deposition as
long as gases remain trapped iaside
(Schiifer, 1972). In East Africa today,
transport of floating carcasses for long
distances is probably rare owing to the
prevalence of crocodiles. However, in at
least one case, long-distance carcass trans-
port has been observed. A skeleton of a
topi (Damalisctis) was found on the shores
of Xorth Island in the center of Lake
Rudolf, 24 km. (15 miles j from land across
waters in which crocodiles are abundant
(I. Findlater, personal communication).

Transport of articulated parts that do not
float will also occur. Ligament softens when
immersed in water but may still hold bones
togetlier, particularly body parts such as
feet and limb joints (Dodson, 1974:79).
The principal factors limiting long-distance
transport of articulated parts would be the
rate of ligament decay and the combined
size and shape of the object.

The place where an animal dies will have
a great effect on the dispersal potential of
its bones. For example, a skeleton lying in
dense bush on a floodplain or levee has a
lower chance for dispersal than one on open
flats, simply because of the obstruction
caused by vegetation. The trapping effect
of vegetation, particularly along levees,
would greatly reduce the probability of
transporting bones from floodplains to
channels or vice-versa. Levees might ef-
fectively trap bones during flood stages
when the potential for burial is high, and
thus would preserve a mixed or allochtho-
nous fauna. Bones of animals that actually
died on the levees would be preserved there
only if they were buried before being de-
stroyed by surface weathering. On the
deltas, recent examples show a trapping

and binding effect of grasses on bones
which would prevent their movement
unless the vegetation was destroyed (Plate

3).

Transport of parts of carcasses and iso-
lated bones by predators and scavengers is
also a factor in bone dispersal. This may be
significant for individual carcasses, but it
is probably not effective in moving an
entire thanatocoenose away from the
general area of the biocoenose, at least
for large animals with fairly broad habitats.
Specific cases may be important tapho-
nomically, such as the dispersal of fish
and crocodile remains away from aquatic
environments by carnivore activity.

Re-excavation of buried bone is poten-
tially important in floodplain situations with
laterally eroding and aggrading channels
and in lacustrine transgressions involving
erosion of former shoreline deposits. Experi-
mental evidence^ shows that bones continue
to lose organic matter (presumably used by
soil bacteria j after burial and become very
soft and friable after a few years of burial
in v/et sediment. Re-excavation would
rapidly destroy all but the most durable
parts, leaving teeth and compact bone
fragments. Beaches or channels that erode
into previous deposits would tend to con-
centrate teeth and the most durable bone
fragments and redeposit them. In some
cases, bones might become well enough
mineralized during burial to survive re-
excavation intact, particularly if they were
protected by carbonate concentrations
formed in floodplain or levee soils.

^ Experiments are in progress on the shore of
Lake Rudolf, where controlled samples of recent
Vjone have been buried below the water table.
These were examined in 1970 and 1972 for changes
in color, weight, and surface texture. All liga-
ments joining originally articulated parts dis-
appeared after one year of burial. .After two years
of burial, bones had lost 10-20% of their dry
weight (loss of organic material) and had acquired
a characteristic brown patina on their external
surfaces. Otherwise there was minimal change in
appearance, but all bones were soft and friable.

482 Bulletin Museum of Comparative Zoology, Vol. 146, No. 10

Diagenetic Factors

Bones can be destroyed after burial by a
variety of processes. Bioturbation, in which
sediment is mixed by the action of roots
and burrowers, is proljably an important
factor in some cases of bone destruction.
Once bone is softened underground by the
loss of organic matter and by ground water,
it could easily be penetrated and disrupted
by the agents of bioturbation. This may
account for the observed low frequency of
bones in paleosol horizons in the East
Rudolf deposits. In floodplain deposits,
bones would be best preserved when the
increments of sediment added during a
flood were thicker than the average depth
(e.g., 10-50 cm) of the root and burrow
penetration that would affect the new land
surface. Accumulations of this magnitude
have been observed after a single major
flood of a small river in central Colorado
(McKee et al, 1967:835). Rapidly aggrad-
ing floodplains lacking extensive plant
cover would provide an ideal depositional
situation for preserving floodplain thanato-
coenoses with minimal subsurface destmc-
tion.

Compaction of sediment can crush and
distort buried bones, particularly if tliey
are wet and friable because of subsurface
conditions. Distortion could occur if the
bones are somehow decalcified and rubbery
or if organic material has been lost, leaving
a fragile structure of hydroxy apatite crystal-
lites. The effects of compaction will be
greater in clay-rich sediments that lose a
significant volume of water when com-
pacted. Clay units also may undergo a
considerable degree of expansion and con-
traction, creating minor slickensides and
joint systems. As noted by Dodson (1971:
55) this can cause breakage of enclosed
bones, although the broken pieces may
themselves be well preserved (B. Patterson,
personal communication). Sandy units are
less subject to compaction and fracturing,
and enclosed bones will be less disturbed'
Thus, for purely physical reasons relating
to sediment type, bones buried in clays and

silty clays will have much less chance of
undisturbed preservation than bones buried
in coarser sediments.

An abundance of CaCO^ in a deposit, or
the seasonal movements of ground water
charged with a Ca^+ and COo, may either
help to preserve a buried bone or to destroy
it. Bones can serve as centers of CaCOs
nodule formation and are often permineral-
ized with CaCO.s. In some cases, however,
bones can be "exploded" by the outward
growth of a carbonate nodule. East Rudolf
fossils provide examples of this as well as
bones which have breaks that are "healed"
by CaCO.t deposits. The processes of car-
bonate deposition in association with bones
are poorly known. In some cases the original
apatite is not altered by fossilization, as
shown by unaltered carbon contents of
recently fossilized bone apatite (Haynes,
1968). There is some evidence that the
amount of organic matter in a buried bone
will influence its fossilization, with fresher
bones tending to encourage nodule for-
mation (Konizeski, 1957:141). Environ-
ments with locally high concentrations of
calcium, such as those associated with high-
calcium vulcanism in the East African Rift
System (Bishop, 1968:38), seem to promote
thorough permineralization of bones and
later resistance to surface destruction of the
fossils.

Carbonate concentrations imply fairly
alkaline soil conditions (Millar et al, 1966:
143), which will be more likely to preserve
bone than acid conditions. Rates for the
dissolution of bones in association with acid
soils are not known, but over time, even
slight acidity (undersaturation of Ca++)
would contribute to their destruction.

Conclusions

The points of taphonomic interest for the
interpretation of East Rudolf (and other)
fossil assemblages include the following:

1) Mode of death will be of primary
importance in determining which
skeletal parts will be available for

East Rudolf Paleoecolocy • Bchrcnsmcycr 483

fossilization. M^ien carnivores exert
tlieir maximum effect on carcasses of
relati\'ely large animals (> 150 kg),
skulls will be the most common resi-
due, followed by vertebrae, ribs, pel-
ves, and ends of long bones. Smaller
animals may be d{\stroyed entirely ex-
cept for teeth and parts such as horn
cores. Burial of articulated skeletons
will occur rarely, and then only when
scavenging is held to a minimum. The
proportion of complete bones that
survive scavenging will be a function
of the density of carnivores in relation
to prey at any particular place and
time; the greater the carnivore den-
sity, the fewer the complete bones.

2) Terrestrial animals may have the best
chances for complete preservation in
environments that are not their
normal habitat (e.g., drowned wilde-
beest buried in channel deposits).
The parts of a thanatocoenose that are
not subject to immediate transport
will be the most useful in reconstruct-
ing a biocoenose. Once bones are dis-
articulated and exposed to weather-
ing, their chances for dispersal away
from the immediate area, without
significant alteration of the thanato-
coenose, are greatly reduced. Dis-
persal potentials are then dependent
on the size, density and weight of
each bone and its rate of destruction
in situ or in transport.

3) Bones quickly show the effects of
surface weathering after initial ex-
posure (usually a matter of weeks or
months), and they will not last more
than 3 to 10 years under most surface
conditions in any sort of intact state.
Fossil bones with intact fresh-appear-
ing surfaces were probably buried
relatively soon after the death of the
animal,

4) Teeth should outlast bones in most
taphonomic situations, and small teeth
will probably have a higher survival
potential than large. Compact bone

weathers and abrades more slowly
than porous. Bones that have lost a
good proportion of their residual or-
ganic content (collagen) will be more
quickly abraded and destroyed during
transport than fresh bones.

5) Well-preserved fossil bones record
the fact that the deposit they are in
has not been extensively reworked.
Reworking should result in the frag-
mentation or total destruction of
bones unless they are thoroughly
mineralized. However, tc^eth may
survive and be concentrated from
older deposits.

6) Buried bones can be disrupted or de-
stroyed by bioturbation, acid ground
water, carbonate nodule formation,
and movement of enclosing sediment
with a high clay content. Alkaline
conditions with available CaCO^ are
an optimal chemical environment for
bone preservation.

CHARACTERISTICS OF RECENT
BONES AS SEDIMENTARY
PARTICLES

Vertebrate remains can be transported
along with other material moving from
place to place on the earth's surface. Bones
from different sources can be mixed, and
some can be carried long distances while
others lag behind. In order to determine
the ecological provenance of the bones, it is
first necessary to understand their behavior
in transport situations.

A few experiments have been done on
bone transport under controlled conditions
in laboratory flumes. Voorhies' (1969)
study on the transport of disarticulated
sheep and coyote skeletons provides data on
bones of moderately large vertebrates. Dod-
son (1974) conducted fhmie experiments on
the dispersal of movisc bones. His data
show that mouse bones can be easily trans-
ported by relatively low velocity currents
(6-35 cm/sec), and he concludes that
dispersal will b(> so great as to render the
bones essentially useless for paleoccologic

484 Bulletin Museum of Comparative Zoology, Vol. 146, No. 10

Table 1. Voorhles dispersal groups: boxes of sheep and
coyote which are transported together in a flume with cur-
RENT VELOCITIES UP TO 152 CM/sEC.

GROUP I

GROUP II

GROUP III

Imnediately transported Transported later than
by flotation or by Group I, usually by
saltation. traction.

Resisted transport,
lagging far behind
other groups.

RIBS

VERTEBRA

SACRUM

STERNUM

(scapula)

(phalanx)

(ulna)

FEMUR

TIBIA

HUMERUS

METAPODIAL

PELVIS

RADIUS

(scapula)

(ramus)

(phalanx)

(ulna)

SKULL

MANDIBLE

(ramus)

Parentheses indicate
occurrence in more
than one group.

interpretations (Dodson, 1973:82). How-
ever, Voorhies' data show that most sheep
and coyote bones require greater current
velocities to move, and that distinct groups
of bones with different dispersal potentials
form at velocities comparable to those
found in natural streams (~ 20-150 cm/sec.
(Leopold et al, 1964:166)).

The dispersal groups of bones (referred
to as "Voorhies Groups"), are shown in
Table 1. These are formed of skeletal parts
that tend to be transported together as the
current velocity in a flume is increased to
a maximum of 152 cm/sec. Fifteen sepa-
rate trials were run and the results averaged
to give the three dispersal groups (Voorhies,
1969:66).

Voorhies' work shows that different bones
require different minimum fluid shear
stresses for transport. This provides a valu-
able basis for interpreting fossil assemblages

(From Voorhies, 1969)

of bones of sheep and coyote size. An
assemblage composed primarily of either
Group I, Group II or Group III would
indicate that sorting processes related to
transport had been operating on the original
thanatocoenose. An assemblage composed
of Group III would retain more paleo-
ecologic information than one consisting of
Group I or II, since this group requires the
greatest stress for transport.

The flume data are restricted to sheep
and coyote bones and provide little direct
information on how bones of other animals,
of other sizes, will sort under current action.
However, Voorhies' data can be used to
indicate the relative importance of basic
characters of bones in forming the dispersal
groups. These characters are size, density
and shape, the important parameters of any
sedimentary particle. In order to show how
these parameters affect bone transport, it

East Rudolf Paleoecoi.ocy • Bchicnsmcycr 485

is necessary to understand liow tliey affect
the transport of sedimentary particles in
general. The principles and formulae that
describe fluid-particle interactions arc re-
viewed in a number of recent texts, and
references used for this study include J.
Allen (1970), Pettijohn ct aJ. (1972), and
Shapiro (1961).

Properties of Bones as
Sedimentary Particles

Data on the sizes, densities and shapes
of bones are necessary for a theoretical
consideration of bones as sedimentary
particles. Since there is little or no in-
formation of this kind available, it was
necessary to carry out a series of measure-
ments to determine the general range of
densities and sizes (volumes and weights)
of bones. These provide the basis for sub-
sequent discussion of bone transport po-
tentials.

Measurements of Bone Size and Density

In order to relate the characters of
measured bones to the East Rudolf fossil
assemblages, bones of recent representa-
tives of fossil taxa were used. These in-
cluded museum skeletons of hippopotamus,
zebra, a large and a small antelope
(Rcdiinca, Damaliscus), and a pig {Hijlo-
chocrus), as well as parts of two crocodiles
and various fish species. In addition, the
skeleton of a sheep was used for comparison
of size and density characters with Voorhies'
flume data. The bones included variable
amounts of residual organic material, but
in general were thoroughly degreased.

Volumes and weights were measured for
c>ach bone. Volumes were measured by a
simple water-displacement method. Bones
were soaked for 5 minutes, or until bub-
bling stopped, and then measured for
volume which included the absorbed water.
Wet-weight, also including this water, was
measured, in order to derive a wet density
for each bone. Wet density is the param-
eter of interest if bones are transported
while wet and take up water quickly upon

immersion. Rates of water uptake will be
discussed further below.

Densities, weights and volumes of most
of the skeletal parts of the animals listed
above are given in Appendix 1. In general,
densities range from less than 1.0 to about
2.3, volumes from 1.0 to 3000 cc, and
weights from 1.5 to 4900 grams (g). Bone
densities are comparable for all mammals
except hippos, which are generally slightly
higher. Crocodile and fish bones have
generally higher densities than mammal
bones. Within each skeleton, the range of
densities is very broad, from foot bones and
vertebrae that float to teeth, which are the
heaviest parts for their size.

How representative are these measure-
ments of bones that are actually parts of
natural thanatocoenoses? Several possible
sources of error can be examined and their
overall importance analyzed:

1) Differential uptake of water. It was
apparent during the measurements of
volumes that bones varied in their
rates of immediate water absorption.
Some floated for several hours with
essentially no water gain, while others
were immediately permeated. In
nearly all cases the major weight gain
from water uptake occurred in the
first 1-5 minutes of immersion. Figure
2 shows the rates of water absorption
for various bones. It is significant
that the naturally weathered bovid
femur gained all of its water in the
first moments of immersion, while the
museum femur continued to gain
weight after 70 hours of soaking. This
indicates air pockets in the un-
weathered bones which are probably
blocked by organic matter. All pores
are open in the weathered bone and
water quickly permeates it.

Natiually occurring, unweathered
bones also will have trapped air
pockets that can lower their densities
during initial ti'ansport. However, the
rates of surface weathering are rapid
enough so that most exposed bone

486 Bulletin Museum of Comparative Zoology, Vol 146, No. 10

▲

-M

+-> en

3 C

C -f-

(1)

•r- -i^

E ns

LO to

J- H-

50 60 70 80 90

Hours
(Log Scale)

Figure 2. Water uptake rates of different mammal bones. Weathered bones and small or compact bones gain
essentially all of their water content within five minutes of soaking. Large, unweathered bones with residual
organic matter gain weight more slowly as pockets of trapped air are gradually displaced.

would become permeable in a matter
of a few months. Therefore, the
museum assemblage differs from
natural ones in its greater number of
blocked pore spaces, which will tend
to lower the measured densities. For
the larger and more porous parts, the
wet weights and densities recorded in
Appendix 1 are thus minimum esti-
mates of these measures for naturally
weathered bones.
The original organic content of the

hones. Aside from contributing to air
entrapment in a bone, residual or-
ganic matter probably does little to
affect its overall weight and density.
The density of cartilage is 1.1 and
tendon 1.3'(Currey, 1970:30). Other
tissues have densities close to 1.0 since
they are composed primarily of water.
A bone filled with water should have
about the same weight as a bone
filled with tissue, and after initial
submergence, the two would have

East Rudolf Paleoecology • Bchrcnsmcyer 487

ALCELAPHINE HUMERI

1.6

1.4

1.2

1.0

^*N

.''"■■^

\a\

\ \

*^ ^

\ • 1

\OjA

^-^

Vc\

V 1

>.f

200

300

Volume in cc.

400

A = A1 eel aphus
D = Dama 1 i scus
c = Connochetes

Figure 3. Graphs showing density variation in a sample of 12 alcelaphine humeri. Dotted lines encircle the left
and right bones from the same animal. Density differences between right and left of a pair are due to variation in
the amount of air initially trapped in the bones. Density differences between pairs may be due in part to original
variations in the amount of bone per cubic centimeter in the individual animals.

approximately the same density, other
factors being equal.
3) Variations in bones from different
individuals. The same skeletal part
may vary in density in individuals of
the same group of animals because of
body size, age, diet, etc. How great
is this variation? A series of humeri
from a single bovid group were
measured in order to answer this
question. Figure 3 shows that the
humeri vary from 1.27 to 1.52 in
density. Density variations owing to
differential water uptake are indi-
cated by the differences between
right and left sides. Compared with
the range in volume, there is relatively
little variation in the densities of
humeri from the different genera. It

appears that differential uptake of
water is more important in affecting
the densities of fresh bones than vari-
ations between individuals of dif-
ferent sizes or genera. Experimental
work on the densit)' variations of
weathered bones should be done, but
it seems probable that indi\'idual
variation in similar skeletal parts is
not an important factor affecting bone
dispersal potentials.
4) Experimental error. Repeated trials
showed this to be much less than
other causes of density and weight
variation. The error was generally
less than 5 percent for wet weight and
\'olume.

The differential uptake of water in fresh

488 Bulletin Museum of Comparative Zoology, Vol. 146, No. 10

bones appears to be the major factor
that may cause the measurements given in
Appendix 1 to be different (i.e., lower)
than the actual properties of naturally
occurring, waterlogged bones. Measure-
ments of weights and densities of weathered
bones and thoroughly waterlogged fresh
bones (Fig. 2) indicate that the densities
given in Appendix 1 are within about 15
percent of the true densities for large,
porous bones and are much closer for
smaller, more compact ones.

Published data on bone density indicates
a range similar to that given in Appendix
1. Samples of porous (cancellous) and
compact (cortex) human bone have densi-
ties of 1.78 and 1.88, respectivelv (Johnson,
1965:550). Currey (1970:30)" lists the
specific gravity of "bone" as 2.0, "ear bone"
( = petrous part of temporal ) as 2.4 and
tooth enamel as 2.6. The density of the
bone mineral, hydroxyapatite, is 3.1 to 3.2
(Berry and Mason, 1959:454). (Published
determinations of bone density usually re-
late to a cubic centimeter of average boney
material, not the bulk density of whole
bones, which is less owing to the number
of naturally occurring open spaces and the
presence of trapped air pockets.)

From the above discussion, it can be
concluded that measurements of bone size
and density as given in Appendix 1 are
generally representative of the variation
that would be present in a natural bone
assemblage, and they can be used to pre-
dict hydraulic behavior if the limitations
of the data are kept in mind. For the fol-
lowing discussion of the relative dispersal
potentials of bones, absolute measures of
bone densit)^ etc., are less important than
consistent differences in the relative proper-
ties of different skeletal parts.

Relative Dispersal Potential of Bones

How does the dispersal potential (prob-
ability of transport) of bones relate to
characters of size, density and shape? This
can be shown by plotting density versus
wet weight for all the skeletal parts of the

sheep, as given in Appendix 1, and compar-
ing the disti-ibution of bones with the
Voorhies Groups. Figure 4 shows that
Group I consists of the smallest, lowest
density bones. Groups II and III are com-
posed of bones that are denser than those
of Group I but not necessarily larger. Den-
sity appears to be more important than size
in determining whether bones will disperse
with Group I.

The scapula and mandibular ramus (with
teeth) fall within Group II on Figure 4
although they do not always belong there
according to the flume experiments (Table
1). The scapula can belong to Group 1,
yet its size and density do not show this.
The shape factor must be operating to in-
crease the dispersal potential of the scapula,
and this is reasonable considering its high
surface area to volume ratio. Its form, with
the spine projecting at right angles to the
scapular blade, also would contribute to
bottom instability. The ramus, on the other
hand, is more of a lag element ( Group III )
than its size and density indicate. Its sur-
face area to volume ratio is fairly high, and
this should operate to make it more trans-
portable. However, as noted by Voorhies
(1969:67), rami have a convex-up (bucchal
side up) stable bottom position and are
relatively flat. Once they attain a stable
bottom position, the size and density com-
bine with a relatively small cross-sectional
area (as seen by the current) to decrease
transport potential.

It is clear that densit)' and size characters
of bones can explain their dispersal
potential as shown by Voorhies Groups,
with shape becoming an important factor
only for particular bones. Although natural
transport conditions will vary greatly from
those of the flume, the sorting of groups
of bones with different transport potentials
would seem an inevitable consequence of
their differences in size, density and shape.
Dispersal groups of bones of sheep size and
above should form at current velocities
found in natural flow conditions, unless
these flows are competent enough to carry

LlJ
O

East Rudolf Paleoecology • Bchrcnsmcyer 489

2.20

2.00

1.80

1.40

1.20

1.60 ►•»\*

VOORHIES
GROUP III

r\ -^*

*v>

<i°\

.c\ ^\o^ GROUP II

VOORHIES

.v^

vC^'v.

I ••^^ •^e.-^\^

^.^---^-^

— .1.00

.80

<}^

■< Zone of Overlap

• ^^^^

VOORHIES

GROUP I

FLOTATION

100

150

200

250

300

350

WET WEIGHT IN GRAMS

Figure 4. Plot of density versus wet w^eight for sheep bones showing that these two variables can be related to
the Voorhies Dispersal Groups. Group I is the most easily transported according to Voorhies' flume experiments,
and Group I elements have low densities and/or weights. The scapula and ramus, which do not plot within their
dispersal group, indicate that shape has an important effect in their potential for transport under fluid stress.
Abbreviations: po = podial, pat = patella, v = vertebra, ph = phalanx, stern = sternum.

all bones together (e.g., high density mud-
flows ) .

Using density and weight data for the
other animals given in Appendix 1, it is
possible to predict, in general, which bones
of these animals would sort out with
Voorhies Groups I, II and III, under the
same experimental conditions. Those ele-
ments which fall in the Group I zone and
in the mixed Groups I/II zone are listed
for each animal in Table 2. If these were
placed in the flume with the sheep bones,
their size and density should cause them to
sort out with the easily dispersed group,
and they would be transported away from
the other skeletal parts. For the different

animals, the same skeletal parts are con-
sistently present in Group I, but the num-
ber of Group I elements decreases with
increased animal size.

Since Voorhies Group I is the most easily
affected by transport, its presence or ab-
sence in fossil assemblages can provide
specific information on the sedimentary
history of these assemblages. For example,
if one of the East Rudolf iissemblages were
composed only of bovid vertebrae and ribs,
equid vertebrae, terminal phalanges of
hippos, etc., then this would almost surely
represent a transported, allochthonous bone
concentration. Various models of this kind
can be constructed for later comparison

490 Bulletin Museum of Comparative Zoology, Vol. 146, No. 10

Table 2. Bones of different sized recent animals which should disperse as Voorhies Group I or
Group I/II under fluid stress. The dispersal potential of most of these bones is determined by

THEIR LOW WEIGHT AND/OR DENSITY. AlL OTHER BONES SHOULD LAG BEHIND. BONES IN GrOUP I/II ARE

SHOWN IN LOWER CASE.

OVIS

REDUNCINE

SUID

ALCELAPHINE

EQUUS

HIPPO

CROCODYLUS

FISH

Phalanges

Ulna

Podial

Phalanges(t)

Calcaneum

Phalanges(t)

Scutes

Pectoral

Sacrum

Sternum

Vertebrae
Axis
Cervical

Podial

Sacrum

Vertebrae
Atlas
Axis

Phalanges

Patella

Sesamoids

rib

Rib

Ulna-P
Vert. Cent.
Sesamoids

Podials

Phalanges

Vertebrae
Cervical
Lumbar

Sesamoids

phalanges(t)

phalanges

sesamoids

scutes

small
teeth

Spine
(Bagrus)

Skull
(Lates)

Thoracic
Lumbar

Patella

Cervical
Thoracic

Patella

astragalus
podial

podial
phalanges

Patella
Vert. Cent.

pectoral

pectoral
+ spine

Humerus-P
Metacarp-D

Ulna-P
Vert. Cent.

phalanges

vertebrae
cervical
lumbar

Sesamoids
vert.-thor.

skull parts
(Clarius)

Vert. Cent.

Sesamoids

patella

rib

Rib

Phalanges

rib

calcaneum

Sesamoids

calcaneum

astragalus

phalanges

podial

vert.-lumb.

vert. -atlas

rib

Vert.

= Vertebra

metatars-P

(t)
P

= Terminal
= Proximal

metacarp-P

= Distal

phalanges

with the fossil assemblages. These are
summarized in Figm'e 5. Three basic kinds
of assemblages can occur, with transitional
phases: "undisturbed" (Groups I, II, III),
"winnowed" and "lag" (Groups II, III), or
"transported" (Group I), in decreasing
order of paleoecological importance.

The proportions of different Voorhies
Groups in fossil assemblages should provide
evidence for the proximity of fossils to the
original thanatocoenose and the habitats of
the living animals. This is a different ap-
proach from that of Shotwell (1955, 1963),
who attempted to distinguish proximal and
distal (more transported) paleocommuni-
ties on the basis of numbers of different
skeletal parts representing each fossil
taxon. The basis for Shotwell's method was
the idea that the proximity of an animal's
habitat to the site of deposition would in-
crease the number of different skeletal parts
that were likely to be preserved together.
As Voorhies (1969:53) has pointed out, the

number of different parts is less important
than their characteristics of sorting. Figure
5 shows that a single large or dense Group
III bone, such as a skull, could have more
paleoecologic significance than a large
number of Group I bones, such as verte-
brae, ribs, podials, etc. For both small
animals and large, bones vary enough in
relative dispersal potential so that the pro-
portions of different dispersal groups, rather
than the absolute numbers of different
bones, will provide the most useful tapho-
nomic information.

The Hydraulic Equivalence of Bones
and Quartz Grains

The size of a quartz grain that is
hydraulically equivalent to any given bone
can be calculated by using either direct
measurements of the bone's settling velocity
or calculations based on the properties of
size, density and shape of the bone. The

East Rudolf Palkokcolocy • Bchniisnwycr 491

o
a:

CO
Ql
LU
Q-
CO

I—I

1—

(IL

1-^

BEYOND
LIMITS OF
DISPERSAL

VOORHIES
GROUP I

VOORHIES
GROUPS Ij ll

VOORHIES
GROUP I

VOORHIES
GROUPS I, II

VOORHIES
GROUP II

VOORHIES
GROUP b(ll)

VOORHIES
GROUPS I, II

VOORHIES
GROUP II

VOORHIES
GROUPS 11,111

VOORHIES

GROUPS

IjlljlU
(undisturbed)

VOORHIES
GROUPS 11,111

(winnowed)

VOORHIES
GROUP III

(lag)

(complete
removal )

INCREASING CURRENT VELOCITY^

SITE WHERE
TRANSPORT
PROCESSES
BEGIN TO
AFFECT BONES

Figure 5. The formation of hypothetical dispersal groups of bones according to current velocity and proximity
to the place where bones begin transport (usually the s!te of death). The bones included in each Voorhies
Group are given in Table 2. For bones that have been disarticulated prior to transport, the three groupings on
the bottom part of the chart represent the skeletal associations most "proximal" to the place of disarticulation.
For fossil vertebrates, such associations can provide the most paleoecologic information on the habitats of the
living animals, when examined in the context of the sedimentary environment where the bones were preserved.

equations for calculating hydraulic equiva- data on bone-settling velocities. A series of

lence are given in Appendix 2. Since the such experiments was conducted which

shape factor is very difficult to quantify, provides information on the relative im-

calculations of quartz equivalence are of portance of the shap(> factor in affecting

limited value without actual experimental settling velocities.

492 Bulletin Museum of Comparative Zoology, Vol. 146, No. 10

Table 3. Experimental data on bone settling velocities and

hydraulic equivalents. measured settling velocities shown

below are averaged from 10 separate trials.

Bone

Vertebral
Centrum #1

Vertebral

Centrum #2

Phalanx
(Suid)

Patella
(Ovis)

Calcaneum
(Bovid)

Scapula
(Ovis)

Axis
(Bovid)

Metatarsal
(Bovid)

Tibia-Dist.
(Bovid)

Astragalus
(Bovid)

Premolar
(Hippo)

Molar
(Equus)

Molar
(Ovis)

Rib
(Bovid)

Rib Part
(Ovis)

Scute
(Croc.)

Bone

Measured Diameter Predicted (v )-(v )

Volume Density Settling of Quartz Settling
Velocity Equivalent Velocity

1.1,.

yiM/cc

(V3)
cm/ sec

(for a
sphere)

cm/sec

25.2

1.10

16.7

2.6

15.4

-1.3

29.0

1.14

21.9

4.4

18.7

-3.2

13.4

1.14

20.0

3.7

16.4

-3.6

4.3

1.33

17.9

3.0

20.9

3.0

13.2

1.39

22.7

4.8

27.3

4.6

39.0

1.41

33.6

10.5

33.6

0.0

88.0

1.42

34.5

10.1

38.9

4.4

48.0

1.51

29.8

8.2

30.0

0.2

10.3

1.56

27.1

6.8

31.4

4.3

7.0

1.76

31.8

9.4

34.3

2.5

16.7

2.03

38.5

13.8

46.2

7.7

40.0

2.12

53.8

26.9

55.7

1.9

2.6

2.12

30.5

8.6

35.4

4.9

25.0

1.43

18.3

3.1

31.9

13.6

14.2

1.46

20.8

4.0

30.2

9.4

4.5

1.69

18.4

3.1

24.5

6.1

Experiments in Bone-Settling Velocities

The settling velocities of 16 vertebrate
bones of varying size, density, and shape
were measured by using a stop watch and
a large plexiglass tank ( 30 cm X 60 cm X 60
cm). The length of fall was 50 cm, and
each bone was dropped ten times from the
same initial orientation just below the sur-
face of the water. All bones were thoroughly
soaked before the settling experiments. The
average settling velocities for each bone are
given in Table 3.

The diameters of quartz spheres ( d(, ) that

would settle at the same rates as the bones
can be calculated by using the equation:

1 2-(vJ^

" 4/3-g-(p„-l)
dq = .000928 -vs-

dq = quartz diameter
Vs = bone settling velocity
g = 980 cm/sec-
pq = quartz density, = 2.65

The quartz equivalents given in Table 3
show that the bones are comparable to
coarse sand- to pebble-grain sizes. The

East Rudolf Paleoecolocy • Bchrcnsmeyer 493

DIAMETERS OF QUARTZ GRAINS WITH EQUIVALENT SETTLING VELOCITIES

OVIS MOLAR

d = 14.8 imi

ASTRAGALUS
DAMALISCUS

d„= 2.6 mm

VERTEBRAL CENTRUM
HYLOCHOERUS

d = 3.1 ran

DERMAL SCUTE
CROCODYLUS

EQUUS MOLAR

Figure 6. The hydraulic equivalents of different recent bones, as determined by settling velocity experiments.
The equivalent quartz grain sizes were calculated using the method shown on Page 492 and the data given in
Table 3. Density variation in the bones is the primary factor causing the variability of the hydraulically equiva-
lent grain sizes. (Bones and grains are drawn to correct relative sizes.)

variation in bone-quartz equivalence is
shown graphically in Figure 6. It is obvious
that the lighter bones, such as the vertebral
centrum, and bones with high surface area
to volume ratios, such as the crocodile
scute, will be more easily transported than
the heavier and more spherical bones. The
quartz equivalents agree well with the evi-
dence for differential dispersal potentials
of these bones from Voorhies' ( 1969 ) flume
study. Combined evidence from the settling
velocity and flume experiments provide the
general background necessary for predict-
ing the behavior of bones in transport situ-
ations. More work is needed, however,
since in specific cases, the hydraulic equiv-
alence and flume data do not agree. The
sheep scapula, for instance, has a relatively
large quartz equivalent (10.5 mm), which

is inconsistent with its high potential for
dispersal in Voorhies Groups I/II.

The Hydraulic Equivalents of
Fossil Bones

It would be useful to be able to predict,
in general, the original quartz equivalent of
any given fossil bone. Such data could then
be compared with matrix grain sizes associ-
ated with the fossils. The quartz equivalent
of any object can be calculated if its density
and volume are known, and if shape can be
disregarded or corrected for. The basic
equation is:

d„ = (p„-i) -d./Les

Nominal diameter of the

d„

bone = \yi.91 X Volume
Bone densitv

494 Bulletin Museum of Comparative Zoology, Vol. 146, No. 10

For bones, the crucial question is, "How
important is shape in affecting the size of
the quartz equivalent?" This can be clari-
fied by comparing the actual settling
velocities of bones with predicted rates
based only on volume and density. Table
3 shows that most predicted settling
velocities are faster than the measured
rates, by an average of about 12 percent.
For the lightest bones, the predictions
indicate slower rates than are actually ob-
served, with an average error of about 14
percent. The ribs and crocodile scute show
a much greater difference, with predicted
settling velocities exceeding the observed
velocities by 33 to 74 percent.

For most bones, it will be possible to
estimate quartz equivalents by using
volume and density, and to estimate a
possible range of error owing to the shape
factor. For bones that have predicted set-
tling velocities within ± 15 percent of the

Table 4. Average densities of porous and com-
pact BONES ANT) TEETH.

Porous

Compact

Teeth

No. in sample

Mean

1.11

1.65

1.96

Range

1.01-1.29

1.36-2.00

1.70-2.24

actual settling velocities, the quartz equiv-
alents can be estimated within about ± 25
percent. Most bone shapes will fall within
this range. Although the range of possible
equivalent quartz sizes is broad, it should
be possible to equate bones with general
grain-size groups such as coarse sand,
pebbles, etc. Bones with high surface area
to volume ratios, such as ribs and crocodile
scutes, will have a much broader range of
possible quartz equivalents, and cannot
be satisfactorily approximated using the
method described above.

10 100

Bone Volume in cc.

10.000

djj = 3/1. 91 (Bone Volume)

Figure 7. A log-log graph relating bone volumes to hydraulically equivalent spherical quartz grains for three
average bone densities: 1.11, 1.65 and 1.96. Given the volume of any recent (or fossil) bone and a measure (or
estimate) of its density, a range of hydraulic equivalents can be read off the ordinate. This estimate should be
within ±25% of the actual quartz equivalent (d,,) for most bones, with the exception of high surface area to vol-
ume bones such as ribs. See Appendix 2 for method of calculating hydraulic equivalence (d,, = nominal
diameter of bone).

East Huudlf Pale()E(;()i.(k;y • Bchrcnsinctjcr 495

Quartz cquixalcnce gix'cs the size of
quartz grains tliat would settle at the same
rate as a gi\'en bone. This is not necessariK'
the size of quartz grains that would be
transported with the bone. However, set-
tling velocities are related to transport po-
tential, as shown by the Drag and Critical
Stress formulas (Pettijohn et al, 1972:335),
and hydraulic equivalence provides a
general idea of how bone and sediment
sizes should be related if both are trans-
ported by similar processes.

For fossil bones, an important factor in
calculating hydraulic e(|ui\'alence lies in
correctly estimating the original densitv'.
Mammalian skeletal parts are composed of
three basic structural components of dif-
fering densities; porous bone, compact bone
and enamel/dentine. Most whole bones in-
clude both porous and compact parts, and
overall densities range between the two. It
may not be possible to tell the original
proportions of porous and compact bone in
a fossil, but the original densitv should lie
between the average values for each
structin-al type. This will put upper and
lower limits on predicted values for quartz
equivalents.

In order to obtain representative average
densities, the data given in Appendix 2
were averaged for teeth and a selection of
the most porous and the most compact
bones. The porous bones include patellae,
vertel)ral centra and terminal phalanges,
while the compact bones include metatar-
sals, distal tibiae, scapulae and ribs. These
provide generalized but realistic average
densities for porous and compact bones
(Table 4). Using the average densities, it
is possible to construct a graph relating
bone volume to equivalent quartz grain
sizes (Fig. 7).

Figure 7 can be used to relate fossil
bones to hydraulically equivalent quartz
grains in a general way. A rather low le\'el
of resolution is all that can be expected
considering both the wide range of (juartz
diamc>ters that are possible owing to the
shape factor ( ± 25' f ) and th(> problems

encountered in estimating the densities of
fossil bones. IIoweNcr, this is enough to
pro\'ide useful information. Thus, a 100
cc tooth (e.g., a large hippo molar), is
theoreticalh' eciuivalent to a quartz pebble
between about 34 and 56 mm (nominal
diameter). The tooth is within the large
to \'ery large pebble size range. A bovid
astragalus of 7.0 cc, considered as a com-
pact bone, has an estimated range of ((uartz
equivalents between 7.3 and 12.3 mm (9.8
mm ± 25'/' )• The actual (juartz txjuivalent
of an astragalus of this size was measured
at 9.4 mm (Table 3). Thus, Figure 7 can
be used to estimate ranges of quartz equiv-
alents possible for fossil bones (excluding
ribs, scapulae, etc.), and in most cases the
actual ({uartz efjuivalents will probably be
close to the median of this range.

Current Velocities and Bone Transport

Since bones of different sizes and densi-
ties can be related in a general way to
hydraulically equivalent quartz grains, it
is theoretically possible to predict what
current velocities are needed to mo\'e bone
particles. This can be done bv using the
graph of J. Allen (1965: 109) which relates
current velocities to (juartz grain sizes in
terms of transport and deposition (Fig. 8).
The scale for quartz grain size is simply
converted to scales for bone grain size at
each of the three average density \'alues.
Thus, it should take a flow velocity of about
80 cm/sec. to move a bovid metapodial of
100 cc (nominal diameter = 5.7 cm). To
move large mammal bones ( 1000 cc, nomi-
nal diameter = 12.4 cm ) should require
flows of over 150 cm /sec.

From a theoretical standpoint, bones of
the size range for most East African mam-
mals should be transportable in flow
velocities of between 10 and 150 cm /sec.
Bones > 1000 cc of animals such as hippo,
rhino and elephant will disperse nuich less
readily, and only at flow velocities of > 150
cm sec. Predictions for the transport
\'elocities of any bone or bone assemblage

496 Bulletin Museum of Comparative Zoology, Vol. 146, No. 10

400-

RANSP
SUSPE

OR!
NDE

■ C
:d

F
M

\TERI

BED-LOAD

TRANSPORT

1.-

^oc}5''

100

1 >^

r X

*\r>>

^ ^-t:^«

i^^

1 40

t :

^^V

.^^c

^0^-

/■

-—^

-f

A"^

^''

o
o

•~ 10-

/ N(

i TRAf

)R-

■

.^'^ C I

-^ '■JO

DEP
"SUSP

OSITION OF

ENDED

MATER

[AL

10(

Scales for Equivalent Nominal Diameters
10 1.0 10

100

.10

1.0

100

.10

1.0

100

00 mm (Quartz)

TOOTH
/^ =1 .96

COMPACT BONE
/o =1.65

POROUS BONE
/o =1.11

Nominal Diameter (mm)

Figure 8. Theoretical transport velocities for bones, using the graph of J. Allen (1965: 109), which plots quartz
grain sizes against current velocities. The three lower scales show nominal diameters of bones of three average
densities. These scales give the calculated hydraulic equivalents of the quartz grain sizes shown directly above
them on the abscissa of the graph. Bones and quartz grains are considered as ideal spheres. The actual shapes
of bones will cause their quartz equivalents to range between about ±25% of the quartz equivalent for a sphere
(with a broader range for ribs and other high surface area to volume bones).

are possible if volumes are known. The However, until such experimentation can

theoretical framework for bone transport be carried out, the theory provides a general

should be tested experimentally to de- framework for understanding bones as

termine how closely predictions fit facts.- sedimentary particles.

- Allen's ( 1965 ) graph refers to the current
velocity necessary to move a particular grain on a
bed of similar-sized grains. Therefore, the analogy
to bones must be restricted to those bones which
are associated with a bed of clasts of similar nomi-
nal diameter ( or, more precisely, on a bed of
bones of similar shape and size ) . Preliminary
flume experiments conducted in 1974 (subsequent
to the completion of the above manuscript) indi-

cate that bones on a bed of smaller grain size will
move at lower current velocities than those pre-
dicted from Figure 8. Experiments in a natiu^al
stream show that large bones (e.g., a cow tibia)
on a sand and gravel bottom may not move even
at mean flow velocities of 150 cm/sec. However,
in both flmne and stream experiments, the Voor-
hies Groups for bone sorting remain valid.

East Rudolf I'ALiiOECOLOGY • Bchrcn.smcijcr 497

Current Profile

Main Flow

Turbulent
Boundary Layer

Laminar
Sublayer

No Laminar
Sublayer

Figure 9. The effects of current profile and bottom morphology on bone transport: A, a small tooth remains at
rest in the low-velocity laminar sublayer while an astragalus is moved by the turbulent boundary layer; B, on a
coarser bottom with no laminar sublayer, both tooth and astragalus are transported together; C, a tooth is
dropped on the downstream side of migrating ripples, buried, and re-excavated, losing its roots during the ero-
sional period; D, a metapodial (seen end-on) creates turbulence and a scour pit, rolls into the scour pit and is
buried.

Additional Factors Affecting Bone and shape. These primarily eoncern tlie
Transport in Natural Situations character of the environment where trans-
Many factors can affect bone transport port occnrs (e.g., channels) and the nature
besides the basic parameters of size, density of the fluid flow.

498 Bulletin Museum of Comparative Zoology, Vol. 146, No. 10

Bottom Morphology and
Current Profile

Stream velocities in natural situations are
on the order of 20-150 cm/sec., with flood
velocities reaching over 400 cm/sec. (Leo-
pold et ah, 1964:167). However, these are
mean velocities for the flow, measured at
about 40 percent of the total depth above
the bottom of the channel (for a channel
10 m deep, mean velocity is at about 4 m
above the channel bottom). Velocities at
the bottom of a channel are usually much
less, and vary widely according to bottom
morphology. The rate of decrease of cur-
rent velocity near the base of a flow is
plotted as a "current profile" (Fig. 9).

If the channel bottom is smooth, a thin,
low velocity sublayer will separate turbu-
lent flow in the main part of the channel
from the sediment interface (Pettijohn et
al, 1972:333). Particles that are smaller
than the depth of the sublayer may be left
behind as lag while larger particles that
penetrate into the main flow are trans-
ported. In this way, very small or flat
bones and teeth could be sorted from
larger bones, creating lag and transport
groups that might not fit predictions based
on relative settling velocities. It is possible
that this had some effect on the lag be-
havior of jaws in Voorhies' (1969:66)
experiments, since his flume had a fine-
grained, smooth-surfaced bed.

If a channel bottom is rough, the turbu-
lent boundary layer extends to the surface
of the bed, and the velocity of the flow
increases upward less abruptly than for
smooth beds (Pettijohn et al, 1972:334).
Ripples and dunes or coarse sediment can
cause this effect. A relatively large bone
transported as part of the bed load over a
smooth bottom would tend to be retarded
if it encountered a rough bottom by the
decrease in the velocity gradient. In areas
of active dunes or ripples, the bone should
have a good chance of being buried by the
advancing bed forms (Fig. 9). Bones
would also tend to be retarded or trapped
in areas of coarser sediment such as gravel

bars. Not only does the bottom velocity
decrease over gravel, but the critical
boundary stress for a bone among large
particles greatly increases. This is related
to the "kinematic wave" effect of Langbein
and Leopold (1968), in which large par-
ticles tend to concentrate other large par-
ticles during sediment transport.

Fluid Density

Fluid density is generally considered to
be close to 1.0. However, if it is increased
by a large suspended load, then bones will
be transported more easily. A high density
boundary layer or a sediment-charged flow
resulting from a flood can significantly
increase the ability of a flow to transport
bones. Sorting of bones that are less than
or equal to the density of the flow would
not occur since all would "float," regardless
of size. This is one of the only ways to
produce a completely unsorted but trans-
ported bone assemblage, where current
velocity and fluid density are large enough
that bones of all sizes and densities become
part of the suspended load. This would be
a very unusual natural transport situation.

Burial Potential

The more easily a bone can be buried,
the less likely it is to be transported any
significant distance. Bottom conditions will
have an important effect. Loosely packed
sand or soft mud can effectively anchor a
bone, particularly if it has projecting parts.
A scapula would be unstable if oriented
spine-down on a hard bottom, but would be
quite stable if the spine were buried in
bottom sediment. This would instantly con-
vert a Voorhies Group I/II bone to Group
III "lag."

The potential for deep burial of large
particles in active channels is low, even
under heavy flow conditions (Leopold et
al, 1966:213). Low density elements such
as bones would have little potential for
deep burial during sediment movement un-

East Hudolk Paleoecoloc.v • Bchrcnsuwijcr 499

less covered b\' ad\'ancing dune or ripple
fronts. Flow .sc^paration and turbnlenee on
the leeward side of a ripple or dnne is likely
to trap larger particles at the base of the
slip face where they can be buried ( Fig. 9 ) .
As the ripple moves on, the bone may be
re-e\ca\'ated and carried further. Thus,
progression of a bone down a channel with
active bed forms would be a series of stops
and starts, with a good deal of abrasion oc-
curring during each re-excavation.

Larger bones that are beyond the carry-
ing capacity of the flow may also move
slowly along the Ijottom due to localized
effects on the flow. A large particle on a
sandv bottom creates turbulence and eddies
on its leeward side which will tend to
remove the sand, creating a scour pit ( Leo-
pold et ah, 1966). The bone could be
tipped into its scour pit and buried, or a
new scour pit could form, thus moving it
slowly downstream (Fig. 9). A good deal of
abrasion would occur during this process.

The most likely places for final burial are
in the actively aggrading parts of a channel
such as point bars and sand or gravel bars,
or in the fill phase of scour-and-fill. Bones
will move along a channel, suffering pro-
gressive abrasion, until they encounter such
a situation.

Conclusions: Bones as Sedimentary
Particles

A number of points can be made which
are relevant to the interpretation of fossil
assemblages. These are as follows:

1 ) Densities of bones soaked in water for
five minutes vary from less than LO
to about 2.00, and teeth range from
about L7 to 2.24. Variation in the
densities of bones available for trans-
port is high owing to differences in
densities of different skeletal parts
and to the presence of trapped air
pockets in bones with remaining or-
ganic matter.

2) For mammal bones of sheep size and
above, current velocities typical of

sedimentary systems can form distinct
dispersal groupings of elements (\' oor-
hies Groups). The theoretical dispersal
potential of bones appears to depend
primarily on density and size, with
shape b(>coming more important for
those with high surface area to
volume ratios (e.g., ribs).

3 ) C]lose proximity of a bone assemblage
to the original habitats of the living
animals can be indicated by the
presence of all dispersal groups in
association (but disarticulated), or in
some cases by the presence of lag ele-
ments. The total number of different
skeletal parts in a disarticulated as-
semblage is not a valid measure of
proximity unless these represent the
full range of dispersal potentials ( e.g.,
patellae to skulls).

4) Settling velocity experiments indicate
that shape factors will increase or de-
crease bone settling rates by about L5
percent from rates predicted on the
basis of density and size alone. The
nominal diameters of quartz grains
that are hydraulically equivalent to
fossil bones can be approximated
within a range of ± 25 percent
(owing to the shape factor), using
estimated original densities. Such ap-
proximations cannot be used for bones
with high surface area to volume
ratios (ribs, scapulae, etc.). Most
bones of mammals smaller than
hippopotamus are equivalent in set-
tling velocity to quartz particles of
sand to pebble size. An association of
fine-grained sediment and relatively
large bones suggests (but does not
prove) that different processes ma\'
have led to their deposition (e.g.,
bones dropped into fine sediment by
a floating carcass rather than trans-
ported along with the sediment).

5) Considering bone-quartz e{iui\'alent
grain sizes and the standard current
velocities recjuired to transport (juartz
particles, it can be shown that bones

500 Bulletin Museum of Comparative Zoology, Vol. 146, No. 10

under 1000 cc should be transported seven sedimentary units that were sampled

by currents of from 10 to 150 cm/sec. for fossil content. The following two

However, in order to achieve the sections characterize the fossil assemblages

higher velocities in this range, mean in terms of the sorting of skeletal parts and

velocities would normally have to ex- in terms of kinds of animals represented in

ceed 200 cm/sec. (flood conditions), each of the seven units.
Voorhies Group I would move at

normal flow velocities, but Groups II Geologic Setting

and III in general would require flood r^, t- t^ i ir i i i

conditions for significant transport. ^he East Rudo f region lies on the north-

6) Bottom conditions can have signifi- 7^^^^™ f^^ of the Lake Rudolf Basm in

cant effects on bone dispersal, with *e northern part of the East African Rift

irregular, coarse-grained or loosely System ( Fig. 10 ). This part of the Rift has

compacted beds tending to retard ^een tectonically and volcanically active

bones and decrease their transport since at least 25 million years (my.) Before

potential. This may alter the com- Present (B P. , when general downwarping

position of the dispersal groups. began Wa sh and Dodson, 1969; Baker

and Wohlenberg, 1971). There are no well-
The overall conclusion relevant to the defined boundary faults forming a single
following analysis of the East Rudolf fossil "rift valley" in this region. Instead, Lake
assemblages is that bones, as sedimentary Rudolf lies along the axis of a broad de-
particles, are particularly sensitive to pression containing numerous horst and
sorting according to density and size factors, graben structures of various ages. The
In the following pages, it should be kept majority of the faults are north-south and
in mind that the interpretations of the East are steeply dipping to vertical. The main
Rudolf material depend on the assumption segment of the Rift System passes through
that Voorhies Groups will tend to form in the southern end of Lake Rudolf and east
natural systems because of the hydraulic of the major part of the basin, through Lake
effects of bone size, density and shape. Stephanie and north to the Red Sea.
Various factors discussed above may alter Several cycles of faulting and vulcanism
the compositions of the bone dispersal have affected the Lake Rudolf Basin during
groups in natural systems, and the interpre- the last 3 my., and the region remains
tations of the East Rudolf assemblages tectonically active today. The structural
should be accepted with some caution until instability has given rise to numerous local
this assumption can be tested with experi- sediment traps that have been subsequently
ments in natural systems. uplifted and exposed at the surface. This

activity is superimposed on the broader

SEDIIVIENTARY ENVIRONIVIENTS OF downwarping of the basin as a whole,

THE KOOBI FORA FORMATION, which has acted as a large-scale sediment

EAST RUDOLF trap since the Miocene. The lake itself is

shallow (~ 100 m) and the depth of sedi-

The objective of this part of the study is ment along the basin axis is unknown,

to characterize particular East Rudolf sedi- Given the large drainage area (now a

mentary environments in terms of their closed drainage system of 146,000 km-

geology and fossil assemblages. The result- [Butzer, 1971a:l]), the amount of sediment

ing geologic, taphonomic and biologic data accumulated since the Miocene is probably

will then be analyzed for paleoecologic considerable.

information. This section presents a geo- The overall sedimentary situation, which

logic background for the East Rudolf provides for relatively rapid deposition, is

deposits followed by detailed discussions of ideal for fossil preservation. Throughout

East Rudolf Paleoecolocy • Bchrcnsmcijcr 501

::i37'^::::::

■ ■''^. V

Mursi\'

;'-!Usno

^^:.^-*\.^ 'ky'^^^i Fm.

,^--^( OMO DEPOSITS

5(Shungura
Fm.

;::::::::::x:::::::j^:^;:;:!:!:vx^ ' S^ I ]^ ^

wm^mMMMWmm gu i f /:( ^. )

m^mmmmmmmm '-^ f/ '

|:ETHibpj[A||:
■:;:;x;:::i::;?;:x:::::::::::::;

•: LAKE

;:x:!:i:;:::i:;:i:i:;:x:x:::x::;x:!;:;::x;:;x::i:;:;:;:;:x:;:x::(

::x;:;x;:;:;:;:;:;:::;:;x::jx:x:<x:x:x:x;x;x;:!:;:;x;:;:;\
•:;x;:;x;:;x;:;:;x;:;:;>i;:;:;:;:it;:;: : : :::x;x:x;x:x;x::;^^

iiSTEPHANIE^Sg

iifelil

::xj::x::;:ix::i:;:;:;:;:;:x:x::^
-,^<f Fault

Lodwar

Lake Rudolf
Basin

>Xx3 outside
:-:-:''.v:-3 Rudolf Basin

Plio-Pleisto-:x::
cene Deposits

60 kmi

-♦•::-:-:-x

^:xj&^:xf::::::x::::|$^

EAST RUDOLFliii;:^^^^^^^^^
DEPOSITSiii^^^^

•::*x:i:if:x:xx:x:x;:;:x:::::;

•:::Jx!x;i:;x-ix;:;x;x4%

XxX:X:?X:: LuyXyXXvX

ivli^ixij::::::*-^;:::::::!:::::::::::::-:
•xl.x.:.+X::: oc ::::X:X:X:X:X:

■.•.•.•.•••odv

m^^

Figure 10. Map of the Lake Rudolf Basin. Thie area presently draining into tfie lake is shown in white.

502 Bulletin Museum uf Comparative Zoology, Vol. 146, No. 10

o
o

O
O

3
O

' I

»—

CD
O

Cr-

El
»i

0
ML

o
o

A AA

o E .v.-.v-x-:>v>---.-

-Jo ■

c
o

•;;;v;

<E-

I--

I/I

m <

_,..^_.l

▲▲▲▲ A

— j^in

-Hi"

:^5

E t

cat

c E-_:

0 «~^ *•'

o t •

-0 tiii

^ Z

▲ ▲

3N3DOiSI31d

3N3DOnd

1 f

o o

^ i

S o

3
w

— o
o .

c —

0-J

3N3DOIW
, I

r»

a> c

•^^ ^~

o p

</] CO

o —

° 5

2 0, TO

CO W -

<" 2, a>

(D ra li.

o E to

O CD <1>

CO
. TJ

c c

O 3

- O •

C/) ^^ C

^^■°

_ LLJ <"

« 0,

§ oi §;

W "5 <U

2 o

° §.c

- R ^

1. CJ w.

" ° s

- S *

■> CO

^§§

>,-^ CO
XI c

>. o
■o TO '55

CD 05 <J

O t O
O ^ O

CO cc

Q.— x:

CO CO *-

CU CD "O

S> (U

C: CO (0

0) en

c <D

"^ T3

en 2

— </i

c ^ •

■- O O)

_ "O c
CO ^ -c
U QC (D

■^ >

>^6

0) B

.? c o c
u_ CO en CO

East Rudolf Paleoecolocv • Bchrcnsmcyer 503

tlu' Rift System, vertebrate fossils liavc been
collected from various time horizons, and
eventually should pro\'ide a continuous
picture of faunal succession in this part of
East Africa from Miocene to Rcx'cnt tim(>s.

Regional Stratigraphy

Three major sedimentary deposits in the
Lake Rudolf Basin are known in detail: the
Omo sequence, Lothagam Hill, and East
Rudolf. Extensive deposits south of Lotha-
gam, including Kanapoi and Ekora, have
provided important faunal collections and
are currently under geologic investigation.
These areas record periods of sedimenta-
tion for the last 5 to 8 my. in the south-
western, northern and northeastern parts
of the basin (Fig. 10). Correlations be-
tween areas are based on evolutionary stages
of the faunas and Potassium-Argon (K Ar)
dating, with paleomagnetic reversal chro-
nology currently showing promise as a third
method. The stratigraphic relationships of
the three areas are given in Figure IL Gen-
eralized lithofacies shown in Figure 11 also
demonstrate the variability of sedimentary
conditions from region to region on any
given time plane. For the most part, the de-
posits represent local conditions relating to
fluctuations in base level which may or may
not have been basin-wide in scale. Average
rates of sediment accumulation, calculated
for continuous sequences between two
dated horizons, vary from 50 cm 1000 years
for the Shungura Formation (Omo) to 10
cm/1000 years for parts of the Koobi Fora
Formation (East Rudolf) ( Behrensmeyer,
1974).

The Omo Basin includes several sedimen-
tary formations of varying ages and litholo-
gies. Of these, the Shungura Formation
(Fm.) represents the thickest continuous
sequence, on the basis of radiometric age
control. K/Ar dating on a succession of
volcanic tuffs has provided a time scale
for over 500 m of section. The Shungura
Fm. has also produced a large assemblage
of vertebrate fossils which can be ac-

curately placed in the stratigraphic se-
({uence. These factors combined have made
it a standard reference for faunal and time
correlations between 1.7 and 3.8 my. B.P.
in the Lake Rudolf Basin. The sediments
represent fluvial deposition in a large-scale
river system (the ancestral Omo), with a
change to littoral and lacustrine deposition
above "Tuff G," at about 1.9 my. B.P.
(Butzer, 1971b; de Heinzelin ct al, 1971).

The Lothagam Group includes a thick
sequence of volcanics and sediments ex-
posed in a tilted fault block on the south-
west side of Lake Rudolf. The area of
exposure is only a few square kilometers,
much less than for the Omo or East Rudolf
sequences. Dates from a basalt flow and
an intrusive sill, plus evolutionary stages
of the fossil faunas, put the Lothagam sedi-
ments between about 8 and 3.5 my. B.P.
Deposits of Lothagam-1 and -3 bear litho-
logic similarities to the Shungura Fm. and
are fluvio-deltaic in origin whereas Lotha-
gam-2 is clearly lacustrine ( Patterson et al.,
1971).

East Rudolf covers some 5500 square
kilometers (900 sc[uare miles) and its sedi-
ments represent an overall accumulation of
over 300 m (Fig. 12). The oldest unit, the
Kubi Algi Fm., occurs primarily in the
southern part of the region. It is sparsely
fossiliferous and has not been studic^l in
detail, thus it is not discussed in this study.
The Koobi Fora Fm., which forms the bulk
of the fossiliferous deposits, is spread over
a wide area and is extremely variable in
composition, so that correlation is difficult
even along continuous exposures. A K/Ar
date of 2.6±.26 my. B.P. (Fitch and Miller,
1970) and paleomagnetic chronology (Brock
and Isaac, 1974) indicate a time span of 3.0
to 1.4 my. for this unit.

The East Rudolf deposits are bounded on
the east by Miocene volcanics, which they
lap onto unconformabh'. The Kokoi Ridge,
which divides the Ileret and Kool)i Fora
areas, is formed of recently uplifted Plio-
cene basalts with interbedded lacustrine
sediments (Bowen and Vondra, 1973:391).

504 Bulletin Museum of Comparative Zoology, Vol. 146, No. 10

'^^ Ethiopia;*'

^ , KENYA'*

Figure 12. Generalized geologic map and stratigraphic section of East Rudolf, based on Bowen and Vondra
(1973) plus unpublished work by Bowen. Numbered areas were used for controlled sampling of the vertebrate
fossils which provided the material for this study.

East Rudolf Paleoecology • Bchremmeyer 505

The palooslope, as indicated b\- numerous
determinations of palc>oeurrents, inclined
toward the soutlnvest, as it does today. The
Stephanie Basin prohahh- drained across
this area periodically during the Plio-
Pleistocene (B. E. Bowen and C. F. Vondra,
personal communication), but it is doubtful
that a major ri\er s\'stem such as the Omo
Ri\er existed in the East Rudolf region.
Rather, the geomorphic and sedimentary
situation can be reconstructed as set of
coalescing piedmont fans and deltas derived
from mod(>rate relief ( about 600 m or
2000') volcanic and basement terrain to-
ward the east. The distance from the major
part of the Plio-Pleistocene lake margin
deposits to the areas of highest relief (if
similar to today ) was on the order of 40-50
km, and the regional gradient would thus
ha\'e been 600 m /50 km or about 12 m/km.
The geologic evidence indicates that this
part of the Rudolf Basin contained a lake
which acted as base level during the period
of Plio-Pleistocene deposition. At present,
it is not known whether this lake was con-
nected to the Nile drainage during the
deposition of the Koobi Fora Fm. The fossil
and modern aquatic faunas (vertebrates
and invertebrates ) show close Nilotic
affinities, and it is fairlv certain that at
least periochc connections have existed
(Butzer et ah, 1972). The abundant and
diverse molluscan fauna of the Koobi Fora
Fm. indicates fresh water conditions. It
seems likely that both closed and open
drainage situations, and saline and fresh-
water conditions, existed periodically dur-
ing the Plio-Pleistocene in the Lake Rudolf
Basin.

Recent Limnology

A summary of the recent characteristics
of Lake Rudolf is useful for comparison
with Plio-Pleistocene conditions. Today
Lake Rudolf has a surface area of 7500 km-
and a catchment area of 146,000 km-
( Butzer, 1971a: 1). Most of the influx of
water comes from the Ethiopian Plateau
via the Omo River during its seasonal

floods in July (Butzer, 1971 a:37). The
lake basin itself is semiarid, with about
380 mm (15") of rainfall annually (19.36-
1970), as measured along its shores (But-
zer, 1971a). Maximum daily temperatures
range from 34.0-36.0°C, and this plus the
strong southeast trade winds encourages
evaporation from the lake surface. The
seasonal fluctuation in water level is over
one meter. Longer-range fluctuations have
caused the lake to drop 20 m between 1896
and 1940, and to rise 5 m in the last ten
years. During the period between 9500-
7500 B.P., Lake Rudolf was approximately
80 m above its present level (Butzer, 1971a:
15).

The present alkalinity of the lake is
.0194-0.210, and the pH about 9.5 (Beadle,
1932: 187). Ca is low in the lake water
owing to the high pH, and Beadle ( 1932:
186) noted evidence for active precipitation
of CaCOs. Alkalinity is due to high K and
Na content. The overall conditions of the
lake are presently outside the tolerance of
most of the molluscan forms that are typical
of the Plio-Pleistocene and Holocene depo-
sits, and only three living species have been
recovered ( T. Hopson, personal communica-
tion). Fish and plankton are abundant, how-
ever. Recent, apparently living stromatolite
formations have been dredged up by a
fisheries research vessel near South Island
(T. Hopson, personal communication).

It is evident simply on faunal grounds
that the lake has changed greatly since tlu>
Plio-Pleistocene, and also in the past 10,000
years. The water level has fluctuattxl
widely, and along with it, alkalinity and
pH have altered. All evidence indicates
that the lake has been fresher, and the
regional climate probably wetter (at least
periodically), than at present.

The recent sedimentary environments as-
sociated with the Omo Delta have been
examined by Butzer ( 1971a). These include
"flood basin" (alluvial flats associated with
the Omo river channel), "delta fringe"
(distributary and inter-distributary flats)
and the "prodeltaic zone" (suba((uatic

506 Bulletin Museum of Comparative Zoology, Vol. 146, No. 10

200

Meters

FLUVIAL SYSTEM

Ch CHANNEL

Pb POINT BAR

Lv LEVEE

Fp FLOODPLAIN

DELTA

DELTA MARGIN
Mf MUDFLATS
LAGOON
BEACH RIDGE
BARRIER BAR
DISTRIBUTARY
PRODELTA

Br
Bb
Ds
Pd

DELTAIC PLAIN
Id INTERDISTRIBUTARY

DISTRIBUTARY

Figure 13. Diagrammatic representation of the fluvial and deltaic sedimentary environments that are recorded
in the Plio-Pleistocene deposits. The diagrams are based on characters of recent fluvial and deltaic environ-
ments near the shores of Lake Rudolf.

East Rudolf P ai.kokcoi.ocy • lichrcnsmcyer 507

scdimcmtary extension of tlic d(>lta). Char-
acteristics of tliese environments, inclndint:;
information on .sediment t\pe, \-e<:;etation,
and sedimentary structm'es, are inxalnable
in interpreting tlie Plio-Pleistocene sedi-
ments aronnd the Lake Rudolf Basin.
Observations (by tlie autlior) of reecMit
t>nvironmc>nts along the eastern shore of the
lake, outside the Omo delta system, have
also proved useful in the interpretation of
the older sedimentary environments.

In order to discuss the paleoenvironments
of the East Rudolf sediments, it is necessary
to provide a terminolog)' that describes the
environments present in the combined lake-
margin and fluvial system of the Lake
Rudolf Basin. Figure 13 and Plate 4 give
a verbal and pictorial description of the
recent environments to be used for inter-
pretation of the Plio-Pleistocene deposits.

Stratigraphy of the Koobi
Fora Formation

The Koobi Fora Fm. as a whole has been
interpreted as a "prograding deltaic com-
plex," with general upward coarsening
indicating outward growth and thickening
of the deltaic deposits through time (Bowen
and Vondra, 1973:392). The paleogeog-
raphy of the deltaic complex at particular
time horizons has yet to be understood, but
a primary lobe probably originated from
the Lake Stephanie region. Sedimentary
sequences in the Ileret and Koobi Fora
areas are different during similar periods of
time, and the Kokoi volcanics may have
divided the two regions into separate
depositional basins during early phases of
uplift.

Stratigraphic nomenclature and sedi-
mentary relationships have been established
by Bowen and Vondra (1973). Although
direct stratigraphic correlation between the
discontinuous outcrops typical of East
Rudolf is difficult, there are several marker
horizons of reworked volcanic ash which
are recognizable over much of the area
covered by the Koobi Fora Fm. Three of

th(>se tuffs, the "KBS," th{> "Tulu Bor" and
the "Suregei," are indicated in the strati-
graphic sections given in Figure 14. They
lie within about 150 m of predominantly la-
custrine and prodeltaic deposits. The Ileret
Member (Mb.), above the KBS Tuff, in-
cludes d(>ltaic and prodeltaic sediments that
pass upward into subaerial deposits indicat-
ing floodplain or deltaic plain conditions.
These are followed by an erosional uncon-
formity overlain 1)\' the Guomde Fm., a
primarily lacustrine unit of undeterim'ned
age. In the Koobi Fora region, the deltaic
and lacustrine deposits which include the
KBS Tuff are followed by an erosional
unconformity that may be of regional
significance. Fluvial deposits follow in the
northeastern part of Koobi Fora, and these
pass laterally into deltaic deposits toward
the southwest, which is down the regional
paleoslope. The Guomde Fm. is absent
in the Koobi Fora region. The Holocene
Galana Boi beds cap the sedimentary sec-
tions both at Ileret and Koobi Fora.

Deposits north and south of the Kokoi
Ridge show increased structural deforma-
tion near the present lake shore. In the
Koobi Fora region, extensive faulting oc-
curs west of a north-south hinge-line
approximately 5 km from the Koobi Fora
peninsula. The Upper Member (Mb.) of
the Koobi Fora Fm. also thickens toward
the west at this point. There is evidence
that minor tectonic events occurred in this
area during the time of sedimentation of
the Upper Mb. These events are represented
by truncated normal faults, and at least
three episodes have been recognized (G. D.
Johnson, personal commvniication). It is
likely that faulting and increased subsidence
in this region have affected rates of sedi-
mentation and, along with this, bone
preserN'ation.

With the general stratigraphic and tec-
tonic framework of East Rudolf as a back-
ground, the following sections will deal
specifically with the characteristics of tlie
sedimentary units sampled for vertebrate
fossils.

508 Bulletin Museum of Comparative Zoology, Vol. 146, No. 10

•a:
▼

c
o
-o
o

S-
13

(Ul

Meters
r20

LU
LU

Od
LU

Tuff

ILERET

KOOBI FORA

■■■■•

-''^^Chari Tuff

^"middle tuff"
8+6-0104

103-0267

■

Koobi
Fora Tuff

^ "lower tuff"

niiiimirm

Uff

130-0201

103-0256
102-0201

105-1311

^2.6 mr

105-0208

Tulujor

Tuff_

5(yre

96 7

7"aff

iiiiiiiiiiiii

Od
LU
PQ

O-

51
Ul

CZ)

PQ
CD
CZ5

U-

rs
s-

oi
o'

i-o|

•r— I

tor

o
o

East Rudolf Paleoecology • Behrcnsmcyer 509

Sedimentary Environments of the
Fossil Vertebrate Localities

Designation of Sampling Localities

Vertebrate fossils are particularly abun-
dant in the upper part of the Koobi Fora
Fni. from just below the KBS Tuff to the top
of the unit. Extensive areas of surface fossil
concentrations closely associated with par-
ticular sedimentary environments provided
ideal situations for bone sampling. Seven
sampling localities were chosen to provide
data on bone-sediment associations.

The discontinuous nature of East Rudolf
exposures has led to a numerical system of
outcrop "area" designation. This has been
used informally by the various East Rudolf
research groups. In the interest of con-
sistency, the fossil sampling localities used
for paleoecologic analysis will be coded
according to this system. Each locality is
thus designated by a two-part number, as
follows: 103-0256; "103" indicating that
the locality is in Area 103 and "0256" indi-
cating a stratigraphic section ("02") and a
bed or horizon in the section ("56").

For the purposes of this study, a locality
is defined as an area of outcrop where there
is a clear association between a particular
lithofacies and an assemblage of vertebrate
fossils. The location of each general area
where sampling took place is indicated in
Figure 12. Stratigraphic positions of the
seven sampling localities are shown in
Figures 15 and 16. Choice of sampling
localities depended primarily on the nature
of the fossil assemblage, and the criteria
used will be explained in the chapter deal-
ing with the fossils. An effort was made,
however, to choose localities where bone
was associated with a variety of lithofacies,
so that assemblages from different sedi-
mentary environments, such as channel,

lake margin and floodplain, could be com-
pared.

Method of Geologic Analysis

Most of the information used for the
interpretation of the fossiliferous sedi-
mentary environments was obtained in the
field. Each of the sample localities was
documented by using closely spaced strati-
graphic sections. Lithologic samples were
collected from each section and examined
later in the laboratory for specific infor-
mation on sediment size and texture. In
addition, lateral facies changes and overall
stratigraphic context were examined and
mapped in each localit^^

The important criteria for distinguishing
lithofacies in the Koobi Fora Fm. have been
worked out over the course of several field
seasons. The sample localities were spe-
cifically examined in terms of these char-
acters, which are as follows:

1 ) Grain size and sorting

2) Thickness and lateral continuity of
consistent lithologies

3) Presence and nati.ire of clay clasts
and/or reworked CaCo;; nodules

4) Limonitic nodules and mottling

5) Primary CaCO^ nodules and/or beds

6) Presence or absence of root casts

7 ) Evidence of bioturbation, particularly
burrows

8) Cross-stratification (large- and small-
scale )

9) Lateral persistence of well-defined,

Figure 14. The stratigraphy of East Rudolf, after Bowen and Vondra (1973). Faunal Zones as determined by
Maglio (1972) are given for the Koobi Fora Fm. The lleret Mb. is only in part the time equivalent of the Upper
Mb. (see Fig. 12). The Kubi Algi Fm. continues downward from the top of the Suregei Tuff but is not included
in this figure. The two-part numbers (e.g., 130-0201) designate the fossil sampling localities used for this study
and show their relative stratigraphic positions.

510 Bulletin Museum of Comparative Zoology, Vol. 146, No. 10

KOOBI FORA/

LAKE
RUDOLF

Map showing locations of Localities in Areas 101, 102, 103

Scale for all Locality Maps
0 .5 1.0

km.

/y/, Holocene and Recent Deposits

Koobi Fora Formation, Upper Mb.

Paleocurrent Direction
Fault

s 102-0201
□ 103-0256

■ 103-0267

nr I I I ■ 1 ■

Figure 15. Maps showing the distribution of bone sampling squares (10 X 10 m) in the area near the Koobi
Fora Peninsula. An index map shows the relative positions of the three more detailed maps. All maps showing
square distribution are drawn to the same scale. The squares themselves are shown slightly larger than true
scale.

East Rudolf Paleoecolocy • Bchrcnsmeyer 511

Placement of Sample Squares in Area 103

LAKE
RUDOLF

H 102-0201

D 103-0256

'I'l'i'i'i

103-0267

Figure 15 Continued.

even horizontal bedding and small-seale
laminations

10) Desiccation structures (mudcracks)

11) Slickensides, prismatic cracking, evi-
dence for paleosol development

12) Invertebrate fossil content

Generalized stratigraphic sections for
each of the sample locaHties are given in
Figure 17. Each locality' is described in
Table 5 according to the sedimentary char-
acters outlined above, and tlie following

interpretations of sedimentary environments
are based on this e\'idence. Vertebrate e\'i-
dence is not included for the specific pur-
pose of keeping this separate from other
characters used in interpreting the environ-
ments of deposition. This permits the sedi-
mentary evidence to be related to the
\ertebrate assemblages without danger of
circular reasoning. Surface textures, hy-
draulic equivalents and other aspects of
the bone assemblages are discussed below,
after conclusions are drawn concerning
each enxironment of deposition.

512 Bulletin Museum of Comparative Zoology, Vol. 146, No. 10

LAKE
RUDOLF

Map showing locations of Localities in Areas 6, 8, 130, 105

LOCALITY 8+6-0104

LOCALITY 130-0201

\ c "-^

Holocene and
Recent Deposits

.5 1.0

km.

Figure 16. Maps showing the distribution of bone sampling squares (10 X 10 m) in the lleret region and east
of the Kol<oi Ridge. An index map shows the relative positions of the four more detailed maps. Locality maps
are drawn to the same scale as those in Figure 15, and squares are slightly larger than true scale.

East Rudolf Paleoecolocy • Bchnnsmcycr 513

Locations of sample squares in Area 105

IkOOBI FORA FMfe^
cLOWER_M_E_MBER[i=i:

Figure 16 Continued.

514 Bulletin Muscwn uf Comparative Zoology, Vol. 146, No. 10

LOCALITY

8+6-

.0104

CaCO,

LOCALITY
102-
0201

Meters „ „„ r
g or G

1.0 s or S

z or Z

Gravel . ^,

Sand c or C = Clay
Silt ^ O"" T'Tuff

LOCALITIES
103-0267&
103-0256

. Q '*' Root Casts
<='• CaC03 Nodules

"Y~T-Mudcracks
^ Armored Mudballs
* Invertebrates

LOCALITIES
105-1311&
105-0208

LOCALITY
130-
0201

Figure 17. Detailed stratigraphic sections for each of the fossil sampling localities. The sections represent the
total stratigraphic interval sampled for fossil bones, and combine the sedimentary data from many sections
measured in each locality.

Sedimentary Environments of the
Sample Localities

Locality 103-0256: Deltaic Flats.

Most of the fossil material is derived
from a thin and very extensive sand which
overlies mudcracked silty clays. In the few
places where the sand was not deposited,
the horizon can be recognized by the mud-
cracked surface. There are no obvious
lateral changes in the grain size of the sand
unit, which is dominantly coarse to fine
over the entire area. This sand is overlain
by the tuffaceous silts and sands that form
the base of the Koobi Fora Tuff, a 12-15 m
thick unit that also covers some 2-3 km-.
The tuff is extensively cross-stratified in its
lower part but generally is horizontally
bedded with persistent horizons of mollusk
shell fragments.

The sediments underlying the mud-
cracked surface are much less unifonn
laterally and have interbedded silty clays,
sandy silts and occasional lenses of coarse,
clean sands. There is a general tendency
toward fining from northeast to southwest,
roughly following the regional paleoslope.
Root casts and CaCO^ nodules <3 cm in
diameter are typical of the silty clay beds.
The mudcracks on the upper surface of the
unit are up to 15 cm in depth and are fimi
evidence for subaerial exposure. They are
filled with the overlaying fine to coarse
grained sand.

The evidence indicates a deltaic mudflats
environment of deposition for the silty clays
and silts underlying the mudcracked sur-
face. The interbedded sand lenses represent
distributary channels. Bioturbation, partly
due to root growth, has obscured evidence

East Rudolf Paleoecolocy • Bchrcnsmexjer 515

Q
U
1-1

u
o

t/3

•41

c«

b
O

(ZJ

n
<

C CO

'»

•H 01

•H

•«

•H

in c

j: 01

u-i

g,^

-0

••-(

•<T

•«r

0 T)

_^^

kl C
4J <0

°^°°

p=»

01 X

D^ gv

kl 3

■O -1

10 0

V -H

C C

rH k4

•-<

v£

^ 10

•H -H

1 4->

0 X(

10 *J

-• .

•

s -o

l4 «

rH 0>

tJi 0

10 rH

iQ 3

u u

•-<

■0

(fl

0) m

cr j: I

<0
in

lO 3 ki

w-t

0) ^

c c

r-l 0 0

O
1

•H -H
10 4-1

1 ki u-i in

c
c

<0 73

U 10

rH u a

-0

£ C

CT g

(0 01 (0 in

-1

E -" c

•

in <o 10

U rH

ifl a

t 10

rH £

C H C

01 1 0^

10 -o

10 "0

•H 0) 0

•

> rH 3

<u c

•V c

■a

O H

O 10

•o

0» -^ 0

U 10

0) (0

o> a 4J

•rl *J

■O « kH

3 0)

XI in

It 10

E '<•'

■4J rH

•H

>. in

C 5

rH l«H

rH 01

C 0

■H 0

fl ^

S "2

—1

V< "0 rH

0 rH

•-*

> 0

Q 3

•

0 01 10

- 01

■H M

0 a o

- Xi

+ ja

o> J3 cr

a 0 01

■a

^~'^''~

■"""

01 £

« u

<e 0

4J rH «

r^ U

f*.

3 « 0)

1 4-1

X) c m

•-<

<H C

•o

»»

r^ 01

k4 10 "O

10 ^

•P £ c
m 0 10

flj

•

g s

tri

■H

1 1 x:

> 01

■0

0( rH cr

01 >

■o

H ^ 3

T3 '^

0 s 0

« c

UH E tH

5 c 8

•H

a 4J

■rl

0 -H

■

U Xi

rH -r) tT>

in
o

•p

10 > 01
c a rH

ID 0)

S S>3

•H

10 -P 10

<a i

rH 0) 0

Q< in in

tJ> 0

•-t c

■H

"0 -H W

o cr

«l

n 0)

5.S

•H

1 £ -a

■f

rH O' C

•K

rH 3 <0

CO 0

•rt

•

%ii'^

♦

M W

(*>

Q
U

feg

2 ><

S H

b ^

< M

(/)

ri8

Z M

S O

o
•i

(0 -1

M <J

w to

Q 10

M O

5»

K
O

> h

•o

•0

516 Bulletin Museum of Comparative Zoology, Vol. 146, No. 10

for successive land surfaces, although in
some areas paleosols appear to be present
(G. D. Johnson, personal communication).
A laterally continuous beach sand trans-
gressed over the deltaic flats, but ap-
parently nowhere formed preserved beach
bars or ridges. Predominantly lacustrine
conditions followed, with influxes of tuffa-
ceous material from distributary mouths
and spreads of shell debris over level sub-
aqueous surfaces.

The lack of extensive erosion on the
mudflats with the coming of lacustrine
conditions indicates the character of the
transgression. Although the mudclasts and
CaCOa nodules incorporated in the sand
are no doubt derived from the mudflat (as
are the fossil bones), the mudcracks on the
surface have not been eroded away. This
can only mean a very low gradient shore-
line, low wave energy, and probably a
relatively rapid transgression. Otherwise,
it is difficult to explain why the increased
energy level which carried the sand would
not have formed beach ridges and eroded
beach fronts, destroying the upper surface
of the mudflats.

The Cretaceous Wealden Lake environ-
ment in the Anglo-Paris Basin provides
some close analogues for the transgressive
deposits of the Koobi Fora Fm., and par-
ticularly those of 103-0256. P. Allen (1959)
reports graded sheets of pebbly sand that
were spread extensively over deltaic de-
posits as the Wealden Lake rose. One of
these, the "Top Ashdown Pebble Bed," is
a graded unit with pebbly sands fining
upward to sands and silts. It is only 10-20
cm thick, and truncates all underlying
structures and sediments. The base is
erosional, and the components of the bed
are derived from underlying deposits (P,
Allen, 1959:292). This is directly compa-
rable in most characteristics to the 103-0256
transgressive sand, but differs in that
103-02.56 does not appear to be derived
from the underlying beds, except for the
mudclasts and carbonate nodules. The base
is less erosional than in the Wealden trans-

gressive sheets. The sand in 103-0256 was
evidently redistributed from former beach
and distributary mouth deposits and carried
shoreward by the advancing lake.

The fossil bones derived from the trans-
gressive sand and the mudflats deposits are
concentrated on the slope below a strike
ridge created by the westward tilted,
resistant sand. They are highly mineralized,
although the pore spaces of many of the
fossils are not filled with cement of any
kind, a unique characteristic of this as-
semblage. There is evidence for mixing of
bones with varying degrees of predeposi-
torial weathering. Some retain fresh, un-
cracked and unflaked surfaces while others
are weathered and have cracked or worn
surfaces preserved rmder their sandstone
matrix cover.

The quartz equivalents for the fossils,
estimated according to their densities when
fresh (Table 4) range from 1.0->20 mm.
This is a very different size range than that
of the quartz sand which forms the matrix
of the fossils (<. 1-1.0 mm). The distributary
sands associated with tlie mudflats contain
grains up to 5 mm in diameter, yet this size
range is absent from the transgressive sand
and apparently was not present on the
deltaic flats. If the bones were derived
from the distributaries, it is reasonable to
expect them to be associated with sand
larger than 1.0 mm. It is possible to con-
clude that most of the bones were probably
not brought into the area by fluvial pro-
cesses, but were derived from a death
assemblage that lay upon the deltaic flats.
The presence of many fresh, unabraded
bone surfaces further supports a locally de-
rived fossil deposit. The bones were prob-
ably redistributed by the transgression, but
final burial was evidently rapid and
abrasion minimal.

Locality 130-0201: Delta Margin

Vertebrate fossils were sampled from a
relatively large stratigraphic thickness (7.0
m) of tilted and faulted sediments. A
variety of lithologies occur, and overall

East Rudolf Palkoecology • BcUnnsmcycr 517

<2;rain sizes range from <. 1-6.0 mm. The
dominant litliologies are evenly stratified
sandy silts, silt}^ elays and medium-grained
Flagstone sands. The sands are generally
elean and rieh in biotite. Coarser, more
poorl\- sortcxl sediment oeeurs in laterally
restrieted lenses.

This locality lies within the marginal
deltaic facies of the Lower Mb. of the
Koobi Fora Fm. The units sampled for
fossil vertebrates appear to be on an
actively aggrading margin of the deltaic
complex. The sedimentary characteristics
listed in Table 5 agree in many respects
with Butzer's (1971a:79) description of the
modern Omo interdistributary basins (la-
goonal mudflats and marsh), including the
presence of limonitic mottling in the silts
and clays. The more evenly bedded and
extensive silts and sands may belong to the
prodeltaic zone as well. The lack of evi-
dence for surface exposure and root-bio-
turbation suggests generally subaqueous
conditions, with water depths greater than
the maximum tolerated by aquatic vegeta-
tion (about 1-2 m).

The poorly sorted gravelly sands are re-
sti'icted to lenses that represent channels.
Pebbles up to 6 mm in diameter occur in
these lenses as floating grains in a coarse
sand matrix. Mudclasts are also present.
The combined evidence suggests at least
periodic currents over 100 cm/sec, and
possibly flood deposition of the kind lead-
ing to the very poor sorting and large
floating grains (Pettijohn, 1957:254-255).
(Vertebrate bone fragments and teeth are
often extremely abundant in these gravelly
sands, and include a high proportion of
nonaquatic forms, in contrast to the aquatic
assemblages derived from laterally associ-
ated lithologies.)

Cross-stratification is often well-developed
in the medium- to fine-grained sandstones.
These include small-scale structures com-
parable to "Kappa" and "Nu" cross-strati-
fication that indicate linguoid ripples ( Fig.
18). The well-sorted medium to coarse
sands show planar foresets, and in some

cases the cross-stratification suggests beach
or barrier bar deposition comparable to that
reported for recent barrier environments
(Davies et «/., 1971). Current directions
for the various forms of cross-stratification
are highly variable^. The bed forms and
grain sizes indicate water movement in the
lower flow regime.

The deltaic margin interpretation of 130-
0201 agrees well with the lacustrine and
deltaic models of Visher (1965) for sedi-
ment types and bedding characteristics.
The transgressive sand-pebble sheets of the
Wealden Lake and 103-0256 are absent or
poorly developed. Instead, the delta of
130-0201 appears to have been continu-
ously aggrading into a subsiding basin,
with occasional periods when sediment
accumulation overtook subsidence and shal-
low water features (root casts, sand and
gravel lenses) developed.

Hydraulic equivalents for the bones
range up to 50 mm, which is much larger
than the maximum size of other associated
particles. However, when large aquatic
animals are eliminated (e.g., hippopotamus
and crocodile), the mammalian remains
have an estimated maximum hydraulic
equivalence of 20 mm and most are less
than 10 mm. This is closer to the matiix
grain size in the channel lenses. The bones
that are close to being hydraulically equiva-
lent to their matrix grains also show more
evidence of abrasion and weathering. These
may have been carried to the delta margin
during periods of high discharge (i.e.,
floods), and therefore may be derived from
a variety of upstream source areas.

Locality 105-0208: Delta Margin and
Lagoon.

The sediments are predominantly silty
sands, poorly sorted and ripple-laminated
with abundant mica. These form a recog-
nizable 2-3 m thick unit over much of 105,
bounded above and below by finer units
of silty clays. Abundant \'ertebrate bone
occurs in association with the silty sands
but is rare in the silty clays. The silty sands

518 Bulletin Museum of Comparative Zoology, Vol. 146, No. 10

Cross-Stratification typical of the sediments sampled
for vertebrate fossils. (Localities in parentheses)

Beta-cross-stratification
(102-0201, 103-0267)

Mu-cross-stratification
(102-0201)

Kappa-Cross-Stratification Nu-cross-stratification

(130-0201, 105-0208) (130-0201, 105-0208, 103-0267)

Onikron-cross-stratifi cation
(102-0201)

Pi -cross-stratification
(102-0201)

(From J. Allen, 1963)

Figure 18.

East Rudolf Paleoecology • Bchrcnsmcycr 519

arc characterized by liorizontal, cvcmly
bedded cosets of small-scale "Kappa" and
"Nu" cross-stratification (Fig. 18). The
individual beds are between 2 and 5 cm
in thickness. Contorted bedding and un-
even lenses of silt are present, but the
bedding shows little evidence of dis-
turbance from bioturbation. Bedding struc-
tures indicate aggradation from the advance
of successive ripple fronts, such as might
be expected in a prodeltaic, lagoonal
en\'ironment.

Coarser sands are interbedded in dis-
continuous sheets and lenses. In one lens,
the overall characteristics suggest a barrier
or beach bar. This sand overlies the mud-
cracked surface of a thin lens of silty clay,
and incorporates clay clasts in its lower 10
cm. The sand body is elongate and extends
for over 100 m before pincliing out. Large
tubular structures resembling root casts are
abundant. Upward the bed becomes better
sorted and has well-developed low angle
planar cross-bedding that closely resembles
the cross-stratification reported for barrier
and beach environments ( e.g., Davies et al.,
1971).

The more tabular, thinner sand bodies
interbedded in the siltv sands are com-
monly cross-stratified, with single sets of
planar and concave-upward laminae. Shal-
low troughs are also common. Ripple
formation at variable current velocities and
depths is indicated, as in 130-0201. The
sands pinch out into discontinuous nodular
layers. Root casts are common, in associ-
ation with the sheet sands, but mudcracks
at the lower bedding contact are rare. The
sands occasionally preserve a variety of
fresh-water mollusks that are unbroken and
locally autochthonous, including BeUamya,
Cleopatra, Melanokles, Pila and Pseudo-
hovaria. The invertebrate fauna indicates
"prodeltaic or even marshy" conditions (D.
Van Damme, personal communication).

There are no poorly sorted, coarse-grained
channel lenses within this part of the Area
105 section, in contrast to 130-0201. The
overall sedimentary characteristics suggest

a lagoonal (>nvironment, with sand and silt
pr(nided from nearby distributary mouths.
This compares well with delta margin con-
ditions in actively aggrading sectors of the
Omo Delta (Rutzer, 1971a:75). Beach
ridges and barrier bars formed at the lake-
ward side of the lagoonal complex and
occasionally transgressed shoreward over
lagoonal sediments. The water in the
lagoon probably varied in depth with shal-
lower phiises represented by coarser sand
lenses with root casts indicating the spread
of shoreline vegetation. The area may have
been periodically (perhaps seasonally) sub-
aerial, although most characteristics indi-
cate overall shallow subaqueous conditions.

Localities 10.5-0208 and 130-0201 are
closely comparable in stratigraphic position
within the Koobi Fora Fm. Both lie near
the top of the Lower Member; 105-0208 is
about 8 m below the KBS Tuff, and 130-
0201 between 10 and 15 m below the tuff.
130-0201 is probably the older of the two.
The localities are about 15 km apart, and
represent related depositional situations on
the margins of the prograding delta system.

Rones of aquatic and nonaquatic animals
are abundant and well preserved, and often
consist of associated skeletal parts. Bone
surfaces are generally fresh, with only
occasional evidence of predepositional
weathering and abrasion. Most of the bones
of nonaquatic animals are fragmented, with
spiral and saw-tooth fractures indicating
predepositional breakage.

The largest bones are of hippo, and these
reach hydraulic equivalents of 15-30 mm,
well outside the sediment range for the
coarser sands. Most of the other bone frag-
ments and teeth are between 1.0 and 20 mm
in quartz equivalent sizes. This overlaps
the size range for other sediment in 105-
0208, but most of the bones, and particularly
the teeth, exceed 2 mm in e(iuivalent size.
Larger sediment grains occur in laterally
associated facies to the east and northeast,
but are absent in 105-0208. The combined
cN'idence points to a local source of bones
from the lagoon and shoreline environ-

520 Bulletin Museum of Comparative Zoology, Vol. 146, No. 10

ments, with perhaps a small component
from the distributaries, including some
floating carcasses. Bones were probably
redistributed and buried during the mi-
gration of beach sands, a process similar in
some respects to that proposed for 103-
0256.

Localitij 105-1311: Channel Complex.

The fossil-bearing unit is up to 4.5 m
thick and overlies an erosional surface with
up to 12 m of relief. Coarse gravels, includ-
ing CaCO.s nodules reworked from the
underlying beds, arc concentrated primarily
near the base of the unit. The lithology is
consistently a coarse sand with lenses of
gravel. As shown in Figure 16, the main
body of the sand is linear, with current
directions indicating a west to southwest
bend. To the south, the sands and gravels
intertongue with silts and silty clays repre-
senting levee and floodplain deposition
lateral to the channel.

Well-developed large scale cross-stratifi-
cation is present throughout the sand unit.
Troughs are the most common form and are
between 20-50 cm in diameter. Gravel
lenses are present at the bases of many of
them. These compare well in morphology
and size with "Pi" cross-stratification (Fig.
18) and with the cross-strata occurring in
ephemeral streams in Central Australia
(Williams, 1971). Such stratification is
formed by downstream-migrating ripples of
varying size in the lower flow regime
(Allen, 1968:110; Williams, 1971:37).

All the above characteristics indicate that
the 105-1311 sand body is a fluvial channel.
The upward fining of the sands is typical of
point bar formations (J. Allen, 1965:140),
and it is likely that much of the sand was
deposited in the point bar formed by the
lateral migration of the channel bend. Root
casts are abundant in the sands and laterally
related silty clays, but are less common in
the coarse gravels near the base of the unit.
This indicates vegetation lateral to the
active channels, and it is likely that a gal-
lery forest existed along the channel. The

geologic evidence does not reveal the extent
of this forest, or whether the water flow
was permanent or ephemeral.

The fossil bones in 105-1311 form dis-
tinct groups according to surface texture.
In one group, bones are highly rounded
and polished and are less than 5 cm in
diameter. They can accurately be described
as "bone pebbles." These compare well
with the second-cycle bones of Rief (1971),
which were mineralized prior to final trans-
port and burial. Although it is possible for
bones to be thoroughly mineralized in
relatively short periods of time (e.g., 5000
years for bones from Lake Rudolf Holocene
deposits), the degree of rounding of the
bone pebbles in 105-1311 indicates long-
tenn abrasion. It is more likely that they
were derived from earlier fossiliferous sedi-
ments associated with Miocene volcanics
to the east than from floodplain deposits
associated with the 105-1311 channel. The
second group includes a wide range of sizes
of relatively well-preserved teeth and
bones. These are generally fragmental, and
show signs of abrasion in their rounded
edges, broken processes and exposed tra-
beculae. A third group consists of only a
few specimens, including whole skulls col-
lected outside the sampling areas, which
show little or no weathering or abrasion.
The latter two assemblages are composed
of bones that had not been mineralized
prior to transport, and that had undergone
variable degrees of surface weathering and
abrasion prior to burial. These can be
referred to as "first cycle" bones.

The largest bone fragments are hydrauli-
cally equivalent to quartz particles up to
40 mm, and most of the teeth fall in the
5-25 mm range. This is well within the
particle size range of the associated gravels,
which range up to 60 mm in maximum
diameter. A very different taphonomic situ-
ation exists in 105-1311 compared with the
three localities discussed previously, which
have bones that exceed the associated
quartz particles in hydraulically equivalent
grain sizes. It is clear that the bone as-

East Hitdolf Paleoecology • Behrensmeyer 521

scmblagc in 105-1311 is much more likely
to reflect the processes that ha\'e affected
the associated sediment, i.e., abrasion and
sorting through hydraulic transport. The
same forces tliat moved sediment through
the channel could also have moved the
bones. A large proportion of these are
probably deri\'ed from upstream sources,
with a more local component derived from
the imdercutting and reworking of previous
floodplain deposits by the laterally migrat-
ing channel. Both of these assemblages
should consist of isolated teeth and the
more durable parts such as ends of limb
bones, all showing some degree of abrasion.
The third component, consisting of the
best-preserved material, would come from
bones left in the immediate vicinity of the
channel and rapidly buried. The bone
assemblage of 105-1311 is thus a mixture
of autochthonous and allochthonous ma-
terial, and most of the bones show the
effects of being in a fluvial system. Since
the general environment of deposition is
fluvial, the bones should belong to animals
found in tiie floodplain or channel habi-
tats, as opposed to the deltaic or lacustrine
habitats.

Locality 102-0201: Channel.

The sequence is tilted some 15-20° west
both in Area 102 and its continuation in
Area 103. Current directions indicated in
the sand are dominantly NNE to SSW, so
that the strike of the beds is roughly
parallel to the current. The deposits of
102-0201 overlie a scoured surface on silty
clays with paleosol development, and they
are followed by widespread sheet sands
with stromatolites and shell debris. The
stromatolites indicate shallow-water la-
custrine conditions (S. Awramik, personal
communication). 102-0201 represents a
brief period of channel cutting between
two longer lacustrine and deltaic deposi-
tional phases.

The dominant lithology is a coarse sand
with gravel near the base, fining upward
to medium and fine-grained sand and

finally to silt. There is no obvious trend
to\\^ard downstream fining in the 3 km
segment examined. Large-scale lenses of
gravel up to 1.5 m thick are common in the
lower 3 m of the unit. The upper 2 m have
only occasional small gravel lenses and
dispersed pebbles. Mudclasts and carbonate
nodule clasts, which are abundant near
the base of the unit, appear to be derived
from the underlying silty clays. Otherwise
the gravel is composed of mixed quartz,
feldspar, volcanic material such as welded
tuff and pieces of silicified wood, all well-
rounded. There are a few polished bone
pebbles and occasional large polished bone
fragments indicating a source of previously
mineralized material.

Cross-stratification includes planar fore-
sets 10-25 cm in height, and a variety of
trough cosets. Many cosets of the planar
cross-beds are comparable to "Beta"-type
stratification (Fig. 18). In some cases the
cross-strata are more upwardly concave
than planar, comparing well with "Mu"
and "Omikron" stratification (Fig. 18),
Allen (1963:110) attributes the formation
of the latter types of cross-strata to mi-
grating asymmetrical ripples. "Beta" cross-
strata result from the downstream migration
of single, straight edged ripple trains over
a planar eroded surface (J. Allen, 1963:
102 ) . It seems that both of these conditions
of ripple bedding, plus intermediates, took
part in the formation of 102-0201. The
troughs are generally large-scale ( 10-50 cm
across) and compare with Allen's "Pi" or
"Nu" types of cross-stratification (Fig. 18).
These are attributed to the migration
of large-scale asymmetrical ripples with
curved crests and projecting lobes or
tongues (J. Allen, 1963:110)." All of the
above structures can be formed by flowing
water in the lower flow regime.

The evidence is conclusively in favor of a
channel origin for the 102-0201 sand. The
gravel concentrations near the base repre-
sent channel bars and channel lag deposits.
In one case, where the coarse material in-
cludes an unusual amount of bone, bedding

522 Bulletin Museum of Comparative Zoology, Vol. 146, No. 10

structures and local upward fining suggest
point bar fomration. Root casts are more
common in the finer sands of the upper
part of the unit than in the gravels. Verte-
brate fossils show a sharp upward decrease
in abundance and were clearly concen-
trated along with coarse sediment near the
base of the channel.

The stratigraphic context of 102-0201
indicates much closer proximity to the lake
than for 105-1311. In fact, 102-0201 can be
regarded as a channel or complex of chan-
nels incised into a temporarily inactive
delta. Evidently, base level was lowered
due to either tectonic or climatic processes.
The period of cut and fill separating two
deltaic-lacustrine units may reflect one of
the local tectonic events which affected this
part of the Koobi Fora Fm. during its de-
position (G. D. Johnson, personal com-
munication ) .

The channel-cutting and gravels of 102-
0201 may be the downstream counterpart
of 105-1311. Both lie near the base of the
Upper Member of the Koobi Fora Forma-
tion, and are in the MetridiocJioerus faunal
zone (Fig. 14). The composition of the
gravels is similar, and some evidence for an
extensive erosion surface analogous to that
in Area 105 has been found in the vicinity
of Area 103. If 102-0201 is the deltaic-
distributary counterpart of 105-1311, then
it is probably slightly earlier in time. After
an erosional phase, the areas closer to base
level (i.e., 102) would begin to aggrade
earlier than more upland areas such as 105.

The fossil vertebrate material is variable
in surface texture and overall preservation.
Bones of aquatic and semiaquatic forms
show minimal abrasion and are often com-
plete. Other vertebrates are represented
by teeth, limb parts, etc., usually broken
and weathered. This indicates probable
transport and a subaerial source (i.e., chan-
nel banks) for the bones of nonaquatic
animals. A few relatively complete parts,
such as a complete rhinoceros jaw, indicate
closer sources and less transport.

The bone fragments are occasionally

over 50 mm (e.g., the rhinoceros jaw) in
hydraulically equivalent quartz sizes. How-
ever, most are equivalent to grains less than
20 mm, and thus are similar to the size
range of the gravels. As in 10.5-1311, most
of the bone in 102-0201 has probably been
subjected to winnowing and abrasion dur-
ing transport. The close association be-
tween bones and gravels in 102-0201 im-
plies similar concentrating processes. This
may be an example of Langbein and Leo-
pold's "kinematic wave" effect (1968),
where large particles tend to concentrate
other large particles and form gravel bars.
The sediment particles, including bones,
are a mixture of allochthonous and autoch-
thonous material. The more complete
skeletal parts, the mudclasts and the
armored mudballs, are examples of locally
derived material from the channel banks or
channel bed. The gravels, including the
polished bone fragments, have been trans-
ported from upstream sources. The largest
proportion of bones and teeth may have
either local or distant sources, and prob-
ably represent animals which inhabited
channel and floodplain environments as
well as the temporarily dry and emergent
deltaic plain.

Locality 103-0267: Distributary complex.

The fossil-bearing horizon is exposed in
widely separated areas covering over four
square kilometers. The dominant lithology
is a poorly sorted gravelly sand. The sands
are of variable thickness and occasionally
cut several meters into the underlying beds.
Coarse sediment fines upward and inter-
tongues with silts and silty clays near the
top of the unit.

The 103-0267 sands and gravels overlie
the Koobi Fora Tuff, which is predomi-
nantly lacustrine in origin and is capped by
a widespread, oolitic carbonate sand with
stromatolites. The deposits of 103-0267 are
followed by lacustrine silts and shell beds.
Thus, the channeling and sand deposition
represent a brief period of subaerial ex-
posure and erosion similar to that of 102-

East Rudolf P.vleoecology • Behrensmeyer 523

0201. About 50 m of continuous section
separates the two units. G. D. Johnson
(personal communication) has suggested
that a tectonic event mav be responsible for
10.3-0267 as well as 102-0201.

The upper part of 10.3-0267 is occasion-
ally characterized by a discontinuous hori-
zon of CaCO.T concentration with abundant
root casts. The root casts are truncated by
the overlying sediment. CaCOs layers are
formed of linked, irregular nodules which
become more massi\e upward. This la\er
appears to bear a primar>- relationship to
the associated sediments; i.e., it fomied at
the time when the top of 103-0267 was a
land surface. The carbonate la\er is thus
tentativeh- identified as a caliche. It is
comparable in sti-ucture and form to
caliches of the .\merican Soutliwest
(Reeves, 1970; Aristarain, 1962; Bretz and
Horberg. 1949). Although the processes
leading to caliche fomiation are not well
known, seasonal upward and downward
percolation of ground water is usually indi-
cated b\' such carbonate concentrations in
soil horizons (Reeves, 1970: 353).

Cross-stratification is more widely vari-
able in scale than in the 10.5-1311 or 102-
0201 channels. The largest sets are up to
20 m across and are broadly concave up-
ward. They compare with "Pi" cross-strati-
fication (Fig. 18) and "festoon" bedding of
the mega-ripple zone (Msher, 1965:47). A
variet)- of smaller scale cross-stratifications
are also present, including "Beta" and "Xu"
t\'pes (Fig. 18). Troughs are well developed
in sandy gravels near the base of the unit,
while the festoon bedding occurs near the
middle in coarse sands with gravel lenses.

Characteristics of 10.3-0267 suggest a
distributar\' complex, with some redistri-
bution of sediment by shoreline processes.
Current directions are highl\- variable, from
NW to S. The deposits represent laterally
extensive channel cut and fill with sub-
sequent aggradation over emergent deltaic
flats. The large-scale cross-strata indicatc>
distributary channels with flow depths of
several meters. This contrasts with the

channels in 10.5-1311 and 102-0201, which
lack cross-stratification of comparable scale
and probabK- carried shallower flows.

The bones of 10.'3-0267 are concentrated
in the lower 3 m, and are usually associated
with pebbles of about 1 cm in diameter.
Large-scale gravel and bone concentrations
such as in 102-0201 are absent and bones
are more or less evenh' dispersed o\er tlie
area co\ered b\- the deposit. There is a mLx-
tiire of bone surface textures indicating
\arious kinds of weathering and abrasion
before burial. Parts of aquatic animals are
generally the best preserved. Second cycle
"bone pebbles" are present, as in the channel
deposits of 102-0201 and 10.5-1311.

Grain size equivalents for the bones
range up to 30 mm in diameter, but most
fall between .5-15 mm. Since grain sizes in
the gravels are up to 30 mm, the bones are
within the o\erall sediment size range.
Many have been transported, and the as-
semblage includes both autochthonous and
allochtlionous bones, as in the 102-0201 and
105-1311 channels.

Locality 8+6-0104: Floodplain.

This unit is composed of lithofacies
unique to the upper part of the Koobi Fora
Fm., occurring only in the Ileret Mb. and
in the Upper Mb. in Areas 130 and 131
(Fig. 12). The dominant lithology is a
lidit-colored tuffaceous silt. The environ-
ment of deposition evidently extended over
a wide area, and the silts are exposed in
Areas 6 and S, which are some 2.5 km apart.
The unit is stratigraphicalh' marked by the
"middle tuff complex," which inchides
locally discontinuous lenses of reworked
\olcanic ash and pumice.

The silts are remarkabl\- consistent in
textine and appearance. They are inter-
bedded at regular inter\'als with zones of
silt>- clays. These show \ertical prismatic
structure and cla>- concentrations suggest-
ing paleosol development. Zones of CaCO:<
nodules occin- within the silts and at con-
tacts of silt>- clays on silts or sand>- silts. The
nodule horizons are often laterallv continu-

524 Bulletin Museum of Comparative Zoology, Vol. 146, No. 10

ous and formed of elongate or flattened,
irregular carbonate concentrations. Inter-
nally these are composed of fine sand float-
ing in structureless micrite. They vary in
size from 2-15 cm maximum diameter.
Smaller nodules of CaCOa are dispersed
throughout the clays and silts. The sedi-
ments themselves have very little dispersed
carbonate, and do not react to the HCL
test. Fragments of nodules are incorporated
in sand lenses representing small channels
interbedded in the silts. In some cases the
nodule horizons are truncated by later beds.
The combined evidence leads to the con-
clusion that the carbonate concentrations
formed during the deposition of 8+6-0104.

The nodule horizons can best be ex-
plained as incipient caliches or carbonate
concentrations formed in the "B" soil zones
of successive subaerial deposits. Lobova
(1967:290-299) describes the formation of
similar carbonate concentrations in desert
soils of the USSR. He suggests they are
formed by biogenic carbonate concentrated
in water percolating downward from the
surface which later evaporates, leaving
CaCOs precipitates. These nodule horizons
commonly form at depths of 20-60 cm be-
low the surface. The presence of such
horizons in primary association with the
sediments of 8+6-0104 indicates seasonal
fluctuations of water content in soils with
a local (biogenic?) source of CO;r and a
source of Ca^^ (clays?). The absence of
extensive, thicker caliches is perhaps due to
the steady aggradation of the floodplain,
with continued burial of former land sur-
faces.

Root casts are abundant throughout the
unit. They are usually less than 1 cm in
diameter and are formed of CaCOs similar
to that found in the nodule horizons.
CaCOa-filled root casts are also found in
the desert soils of the USSR, and are used
as evidence of biogenic formation of
carbonate concentrations (Lobova, 1967:
290). Some of the root casts are truncated
by the channel scour-and-fill structures.
The silts are riddled with tubes which may

be burrows rather than root holes. These
are usually 1.5 mm in diameter and have
distinctive clay rims.

One well-developed channel can be
traced NNW across the exposures in Area
8. It is approximately 40 m across and is
filled with medium- to coarse-grained sand
plus pumice cobbles up to 10 cm in
diameter. Some of the silt beds (and per-
haps the Area 8 lens of "middle tuff)
represent levee and overbank deposits from
this channel. Other channels occur within
the silts and clays. Most are small scale,
with variable current directions. The chan-
nel sands are often well sorted and ce-
mented with CaCOs, and in some cases the
cemented sands weather out as rounded,
resistant blocks and nodules.

It would be difficult to assign 8+6-0104
to any environment other than a floodplain.
In general, the characteristics fit Allen's
(1965) concept of vertically accreting flood-
basin deposits. The whole complex of small
channels and silt deposits may represent a
zone intermediate between deltaic fan and
floodplain, similar to that 1 to 2 km east
of the margin of the present-day Tulu Bor
Delta at Ileret. In Area 6, the "middle tuff"
fonns a widespread, mudcracked surface
indicating deposition in a pond or lagoon,
with later desiccation. It is possible that
more deltaic conditions existed farther to
the west of Area 8 in Area 6.

The bones of 8+6-0104 are generally
very well preserved and often covered with
a CaCOs crust. Some show surface weather-
ing and cracking, and there are abundant
isolated teeth. Associated skeletal parts of
terrestrial mammals are also fairly common.
This evidence suggests variable degrees of
surface weathering and rates of burial.

The bones in 8+6-0104 are associated
with much smaller grain sizes than in the
channels. Hydraulic equivalents of most of
the bones fall well above the 1 mm maxi-
mum grain size of the silts and sandy silts
in which they occur. If individual bones
had been carried in the channels and spread
over the floodplain during floods, then they

East Rudolf Paleoecology • Bcliiciisnwycr 525

should be found in association with grains If the composition of tlic bone assem-
closcr to their hydrauHc e(iuivalents, i.e., blages is hnkcxl to sedimentary processes,
coarser sand and gravel. Sediment of this then the channel assemblages should be
size is available in channels lateral to the more like other channel assemblages than
silt deposits. Since it is not found with the like floodplain or deltaic assemblages. Th(^
bones, and since these show a gcMieral lack characteristics that should be similar within
of abrasion, most of the bones are probably similar deposits include the degree of bone-
autochthonous to the floodplain environ- sorting and the degree of weathering and
ment. The presence of associated skeletal abrasion. In the extreme case, the de-
parts may indicate carcasses buried in situ positional processes could sort and partially
or floated in during the floods. Most of the destroy a given thanatocoenose so as to
bones probably were buried by the periodic obscure all of the original ecological infor-
influxes of floodstage silts. The trapping mation in the assemblage,
effect of floodplain vegetation may have Thus, the first step in recovering eco-
been influential in anchoring the bones logical information from East Rudolf as-
until they could be buried. Some of the semblages is to isolate those cases where
lighter elements may have been dispersed the effects of depositional processes are
by these floods, but most of the thanato- minimal. The evidence presented so far
coenose remained in place as a lag deposit against extensive alteration of a thanato-
to be covered, or destroyed by later coenose by sedimentary processes includes:

wea leimt,. ^^ Bones with fresh, unabraded surfaces

2) Complete bones, sk-ulls with teeth and

Discussion and Conclusions ^^^.^^^^ structures intact

The seven localities can be grouped into 3) Associated skeletal parts (indicating

three broad categories on the basis of simi- l^ck of reworking)

larities in lithofacies: On these criteria, the assemblages of

1) Delta: 103-0256, 130-0201, 105-0208, Localities 103-0256, 105-0208 and 8+6-0104
/■iQn_9Qgy\ have been least affected by depositional

2) Channel: 102-0201, 105-1311, (103- processes, and retain a maxiinum amount of
0'?67') paleoecologic information. The other loeali-

3) Floodplain: 8+6-0104 ties have assemblages with mixed histories,

and ecological information may be more

These groupings are similar in lithology, difficult to isolate.

bedding structures, and lateral facies Fossil assemblages from the channel
relationships. The deltaic localities are environments (including 103-0267) are
more diverse in these characteristics than similar in that they all bear evidence for
the channels, with 103-0256 representing a bone abrasion and include mixed autoeh-
transgressive beach, 130-0201 distributaries thonous and allochthonous material. The
and a delta margin, and 105-0208 a beach lacustrine-deltaic environments are less
and lagoon complex. 103-0267 can also be similar among themsehes. with Locality
regarded as deltaic, since it represents a 130-0201 combining the characteristics of
distributary complex rather than a single, transported and nontransported assem-
well-defined channel. However, its lithol- blages, while the others appear primarily
ogy and sedimentary structures are more untransported. In general, however, it ap-
like those of the channels. Henc(% it is pears that some aspects of the bone as-
intermediate between the deltaic and chan- semblages are similar in similar lithologies,
nel groupings, and is included parentheti- and thus reflect the processes operating in
cally in both. the different sedimentary environments.

526 Bulletin Museum of Comparative Zoology, Vol. 146, No. 10

SORTING IN BONE ASSEMBLAGES
OF THE KOOBI FORA FORMATION

The primary object of this section is to
estabhsh the relative numbers of different
skeletal parts in the seven bone assemblages
and to discuss their taphonomic implica-
tions. Different skeletal parts have very
different potentials for dispersal, as dis-
cussed in the section on bones as sedi-
mentary particles. Depositional processes
operating on bones should affect the ratios
of skeletal parts, particularly those that
have widely different densities, such as
teeth and vertebrae or phalanges. Assem-
blages that have a concentration of elements
with similar dispersal potential indicate
sorting of the original components of the
thanatocoenose. Assemblages with a mix-
ture of heavy and light, large and small
bones indicate either less alteration of the
thanatocoenose before burial, or a mixture
of bones with different taphonomic
histories.

Most skeletons are incomplete when their
parts become sedimentary particles, due
primarily to destruction by carnivores. The
initial assemblage, after carnivore activity
(such as in East Africa today), consists of
teeth, skulls, horn cores, vertebrae and limb
ends, with more parts surviving for large
animals than small. This results in an
assemblage of bones with a wide range of
sizes and densities, which will be subject
to sorting in transport situations. The bones
and teeth also have different survival po-
tentials in most situations, with the former
being more readily destroyed by tapho-
nomic processes than the latter.

Sampling of Bone Assemblages

Most of the fossil vertebrates of the East
Rudolf deposits occur in surface lag con-
centrations due to the removal of surround-
ing sediment. In general, movement of
fossils away from their source rocks is
minimal, and they remain in clear associ-
ation with particular sediments. Conditions
of preservation and recent erosion are such

that even delicate fossils usually remain
reasonably intact, once exposed, and com-
pact objects such as teeth may last for long
periods of time ( 100+ years? ) on the sur-
face. This provides a large amount of
accessible material for collection.

Bones from the seven localities described
in the previous section were collected using
the following procedure: Grid squares of
10 X 10 m were laid out over the chosen
area of outcrop. The first square in each
locality was positioned using an arbitrary
spot on an aerial photograph or simply by
selecting a local landmark (e.g., a tree or
conspicuous outcrop), without specific
reference to the degree of surface bone
concentration. Subsequent squares were
measured off from the first, with a mini-
mum of 20 m between squares. On hori-
zontal sti'ata, the squares were laid out on
an orthogonal 30 X 30 m grid. On dipping
strata, the squares were positioned along
the strike of the units being sampled. The
grid system was adjusted, where necessary,
to avoid patches of recent sediment and
vegetation. The selection of squares was
not adjusted to sample particularly attrac-
tive patches of bone fragments, in order to
prevent subjective biasing of the bone
samples. Collecting was done by system-
atically traversing a square first east-west,
then north-south (for a square oriented
NSEW). All the surface bone larger than
5 cm (mamixum length) was collected in
addition to those smaller bones that could
be identified to class (Fish, Mammal, Rep-
tile, Bird). During the first field season
all samples were removed for identification
and study. During the second season, after
workers were familiarized with the verte-
brate taxa and skeletal parts, it was pos-
sible to do most identification in the field.
This greatly simplified the logistics of the
sampling, and enabled workers to leave the
field with a card for each square recording
taxa and skeletal elements plus geological
data. This was a welcome alternative to
carrying out 50-60 lbs. of fossil bone frag-
ments after each day of collecting.

East Rudolf Paleoecology • Behrensmeyer 527

Table 6.

SAMPLING
LOCALITY

130-0201
105-0208
103-0267
103-0256
102-0201
105-1311
8+6-0104

Stratigraphic data and sample size of the seven fossil sampling localities. Sample
squares are 10 x 10 meters, representin(; 100 m^ each.

# OF
SAMPLE
SQUARES

21
20
20
27
34
25
66

STRATIGRAPHIC

INTERVAL SAMPLED

(IN METERS)

7.0
2.5
3.0
.75

5.0

3.0

4.5

BASIC
LITHOLOGY

Sand, silt,
and clay

Sand and
gravel

Sand

Sand and
gravel

Silt

GENERAL
DEPOSITIONAL
ENVIRONMENT

Delta margin

Delta margin
and lagoon

Distributary-
beach complex

Transgression over
deltaic mudflats

Channel

Floodplain

KOOBI FORA FM. STRATIGRAPHIC
FAUNAE UNIT UNIT

Mesochoerus

Metridiochoerus

Loxodonta

Lower Mb. ,
Koobi Fora Fm.

Upper Mb. ,
Koobi Fora Fm.

Upper Mb. ,
Koobi Fora "^m.

Ileret Mb.,

Koobi Fora Fm.

Maps of each locality showing the
positioning of the sample squares are given
in Figures 15 and 16, and the number of
squares collected in each locality is given
in Table 6. The major problems encountered
in the sampling were: 1) choosing localities
that showed a clear relationship between
the surface bones and the sedimentary
units, 2) obtaining comparable samples
from each locality that adequately repre-
sented the bone assemblages.

Choosing the Sample Areas

The primary goal was to collect an
assemblage of bones that represented the
material buried in a well-defined sedi-
mentary deposit. In selecting the sampling
localities, the following guidelines were
established:

1) A locality was chosen on beds, or a
series of beds, representing deposition
in one of three broad environmental
categories: channel, floodplain or
delta.

2) The topographic situation was such
that contamination of the fossil con-
centrations with material from other
horizons was minimal. Efforts were
made, for example, to sample beds on
drainage divides rather than in val-
leys.

3 ) Vegetation and recent sediment in the
area were minimal.

4) Previous collecting in the area was
minimal, or collection sites were
marked and the removed fossils re-
corded.

5) The locality was extensive enough so
that a representative sample of the
fossil assemblage could be collected.

Fortunately, the East Rudolf region pro-
vided many areas that satisfactorily met all
these requirements. Since stratigraphic
series of environmentally related beds
rather than single beds were used, the
chances of contamination from different
series of beds representing different de-
positional environments was greatly re-
duced. In the course of sampling, the actual
bone-producing beds were often indicated
by matrix adhering to fossils, and some of
the samples could be assigned to particular
horizons. Such evidence further supported
the association of bones with the enxiron-
mental units of interest.

The advantage of sampling different
lithologies that are genetically related (e.g.,
sands, silts, and clays, all deposited in
deltaic conditions) is that this will give a
more general picture of the faunal and
skeletal elements preser\'ed in a rather
broadly defined environment. This con-

528 Bulletin Museum of Comparative Zoology, Vol. 146, No. 10

trasts with sampling a particular bed ( as in
some quarry deposits ) , which is more likely
to be the result of very local or special
conditions. The sampling method described
above allows coverage of extensive areas
(square kilometers) of outcrops represent-
ing single, broadly defined sedimentary
environments. This permits sampling on a
scale more comparable to the habitat sizes
of many East African vertebrates (on the
order of square kilometers to thousands of
square kilometers). Sampling by widely
spaced squares should establish faunal and
bone abundances that represent broad-scale
differences between sedimentary environ-
ments and the habitats associated with
them. Moreover, sampling through several
meters of sedimentary strata representing
extended periods of time should reveal
more general pictures of bone and sediment
associations than assemblages representing

single events.

Sample Size

Surface bones are so abundant in the
sampling localities that even a few 10 X 10
meter squares provided large numbers of
fragments, and over 9,000 were collected in
the total sample from 213 squares. More
than 7,000 {787c) of these were identifi-
able as to skeletal part or vertebrate group
or both. Very few of the sample squares
lacked fossil material, even though they
were laid out without regard to fossil dis-
tribution.

The abundance of fossil material was suf-
ficient to provide an average of 34 identifi-
able pieces per square and to give a good
representation of the most common parts
and animals. Field collecting was aimed
at obtaining the largest possible compar-
ative samples from all the localities. Since
the surface concentration of bone varied
from locality to locality, the number of
squares collected in each varied as well.
Thus, it was necessary to collect over 60
squares for 8+6-0104, which had a low
surface concentration, but only 20 for 105-

0201. At least 20 squares (=2000 m-) were
collected in each locality.

Method of Representing
Fossil Abundance

It is possible to represent the relative
abundance of different bones in more than
one way. For instance, within each locality
the total number of fragments identifiable
as vertebrae can be compared with the total
number of tooth fragments. Percentage
representations of these totals can be com-
pared between localities. Alternatively, the
total number of squares with vertebrae can
be compared with the total number of
squares with teeth, etc. For reasons de-
scribed below, the second method of repre-
senting relative abundance is used in all the
following analyses of the fossil assemblages.

Difficulties in using total numbers of
parts for comparative purposes include the
following:

1) One tooth, for instance, can weather
on the surface into dozens of frag-
ments which are still identifiable as
teeth, but a vertebra may only pro-
duce a few fragments that can defi-
nitely be identified as vertebrae. In
both cases the numbers of broken
fragments, if totaled, would count for
more than the whole elements, and
give erroneous data on the relative
numbers of these elements. This prob-
lem is particularly pertinent to a frag-
mented surface sample, and is almost
impossible to correct for by attempt-
ing to calculate the "minimum num-
bers" of fragments per bone in the
manner of Shotwell (1955).

2) A single skeleton, if disassociated
prior to burial or during recent
erosion, may be counted as several
individuals of the same animal group,
while the whole skeleton would be
counted as one individual. This can
lead to errors in representing the
actual abundance of different ani-
mals. Shotw ell's method of using mini-

East Rudolf Paleoecology • Bchrensmeyer 529

muni numbers of indi\'iduals^ lielped
to resolve this probl{>m for his quarry
samples (1955). In the East Rudolf
surface assemblage*;, with many verte-
brate groups represi'uted by a wide
range of identifiable bone fragments,
the minimum numbers method was
not feasible.

The more satisfactory method of repre-
senting bone abundance for the East Rudolf
localities is to use the number of squares
with a particular skeletal part. This is done
as follows: if one vertebra, or several, or
dozens of pieces of the same one, occur in
a sample square, this is counted as 1 oc-
currence. If one tooth of the same taxon
occurs in each of 5 squares, this is counted
as 5 occurrences. The number of occur-
rences of each bone can be converted into
a "square frequency" by dividing by the
total number of squares in each locality.
Thus, 5 occurrences out of a sample of 20
squares gives a frequency of .25 or 25%,

This method has a number of advantages
\\'hich make it a valid measure of bone
abundance in the broadly defined sedi-
mentary units of interest for this study:

1) It gives a measure of the dispersed
abundance of the different bones in
space and time, which should be a
result of the overall conditions of each
sedimentary environment.

2) The problems encountered in using
fragment totals are essentially elimi-
nated, since using occurrences in
squares will greatly reduce the effects
of differential identifiability and frag-
mentation of the surface bones. Also,
since the squares are widely spaced,
the probability of sampling parts of
the same bone or even of the same
animal more than once is very low.

A comparison of the two measures of
abundance, by fragment number and by

^ The relative abundance of different taxa is
represented by the number of the most common
similar skeletal part (e.g., left femora) of each
taxon (Shotwell, 1955:331).

squares, illustrates the advantages of the
latter method. In Figure 19, the frequency
of vertebrae in each locality is given ac-
cording to total numbers of fragments
identifiable as vertebra, and bv the fre-
ciuency in terms of scjuares with vertebrae.
Numbers of vertebral fragments that are
high relative to the square frequencies, as
in 8+6-0104, imply localized concentra-
tions. In fact, for 8+6-0104 the large num-
ber of vertebrae results from two associated
partial skeletons of bovids. In contrast, a
high square frequency and a low fragment
number shows a widely dispersed sample of
isolated vertebrae, as in 103-0256 and 103-
0267, where only one or two vertebrae
occur per square. The representation of the
dispersed abundance is more useful in
comparing bone assemblages that result
from interrelated processes in channel,
floodplain or deltaic environments. The
"square frequency" of bones ( = the number
of squares with a particular bone or taxon
divided by the total number of squares per
locality) will thus be used in the following
sections.

Characteristics of the Bone
Assemblages

During the collecting of the bone sample,
and prior to numerical analysis, it was ap-
parent that some parts, such as teeth, were
more abundant in some localities than
others. However, most of the differences
in bone proportions among the localities
became apparent only after relative abun-
dances were tabulated in the laboratory.

The bone sample contains abundant
skeletal fragments from mammals, reptiles
and fish, and a few from birds. Analysis of
bone frequencies is restricted mainly to the
mammals, which form the largest and most
diverse component of the sample. Fre-
quencies of the skeletal parts are given in
Table 7. Discussion of the method of
identification, which can influence tlie ap-
parent abundance of parts, will precede
analysis of the data.

530 BuUetin Museum of Comparative Zoology, Vol. 146, No. 10

120

100

I/)

I 80

i-
<u

>•

»*-
o

rl.OO

.80

,60

.40

c
<u

(U
ro

.20 $

.00

LOCALITIES ►

130-
0201

105-
0208

103-
0267

103-
0256

102-
0201

105-
1311

Delta

Channel —

8 + 6-
0104

Flood-
Plain

Figure 19. A comparison of two ways to represent the abundance of vertebrae in tine fossil samples. The
shaded bars on the histogram represent the total number of vertebrae and vertebral fragments from each local-
ity. The white bars represent the square frequency of vertebrae, the proportion of squares in each locality
which contained at least one vertebral part. Localities with a large number of fragments in proportion to the
square frequency generally have associated vertebral columns (e.g., 105-0208, 8+6-0104). Localities with high
square frequencies but low numbers of fragments indicate wide dispersal of vertebrae.

Identification of Bones

A large proportion of the collection was
identifiable as to skeletal part. For reptiles
and mammals, 24 categories include all the
identifiable bones. Mammals have 19 cate-
gories which can be assigned to a specific
class with certainty (Table 7). Rib and
diaphysis fragments, plus some phalangeal,
vertebral and pelvic fragments, cannot al-
ways be assigned to class. These are not
included in the frequencies for either mam-
mals or reptiles.

The number of identifiable fragments of
different bones is variable and can lead to

a bias for greater apparent abundance of
the bones witli more identifiable parts.
However, since all the bone fragments were
identified in a consistent manner by one
person (A.K.B.), there should be little or
no effect on the locality to locality com-
parisons. There will be some effect on the
absolute abundance of certain elements
within each locality. Teeth, for example,
are always more identifiable than other
parts and thus will appear to have higher
frequencies than is actually the case. The
use of "square frequency" helps to minimize
this effect, since only squares with at least
one partial tooth (excluding fragments of

East Rudolf Paleoecology • Behrensmeyer 531

dentine or enamel) were eounted. Differ-
ences in the numbers of identifiable frag-
ments of other bones, particularly for
mammals, probably have an insignificant
effect on their relative "scjuare frequencies"
in this study. Most bon(\s were represented
in each sqnare by at least one relatively
eomplt>te part (e.g., ends of limb bones,
vertebral centra, whole phalanges, etc.).

Significance of the Frequency Data

The data given in Table 7 show that
most of the skeletal parts are represented
in each locality. Some have consistently
high frequencies, such as teeth; some low,
such as patellae, and some are variable. The
lower frequencies indicate occurrence in
only a few squares out of the total for each
locality. Both high and low frequencies are
of interest in comparing the samples.

To assess the statistical significance of
the frequencies, one must ask, "How repre-
sentative of the actual bone assemblage in
each locality are the 'square frequencies'?"
In some respects the problem is comparable
to establishing binomial sampling limits for
accurately detecting character frequencies
in any given population (Simpson et ah,
1960:199). In such cases, tables are avail-
able for relating actual frequencies to ob-
served frequencies using various sample
sizes. For example, a character with 40%
frequency in the actual population could
vary from 12-35 occurrences in a sample of
60, with a probability of only .001 that
fewer than 12 or greater than 35 occur-
rences would be observed.

The binomial sampling limits for square
frequencies can be calculated using the
Harvard Tables (1955). For a sample of
34 squares, a frequency of .32 (11 squares)
could represent a possible range of actual
frequencies between .17 and .48, with
a probability of only p = .05 that the actual
frequencies in the bone assemblage would
fall outside of this range. The sampling
error indicated by simple binomial prob-
ability is potentially rather large. However,
it can be assumed that the square fre-

quencies are more closely representative of
the true bone fre(|uencies because: 1), each
sample square consists of a 10 X 10 m area,
which greatly increases the probability of
finding a particular bone if it is present in
the assemblage and 2), many square's in-
clude more than one bone of a particular
kind, and the actual frequency is higher in
these cases than representation by square
frequency would indicate. Therefore, the
square frequencies will l)e treatcxl as repr(^-
sentative frequencies for the following data
analysis. The bone abundances, as repre-
sented by these frequencies, should be
comparable from locality to locality. The
statistical significance of specific differ-
ences or similarities between localities was
tested using Chi-Square analysis.

Comparisons of Overall
Bone Concentrations

The relative concentration of identifiable
bones varies greatly in the sample squares
of the seven localities. Overall bone abun-
dance can be conveniently expressed by
dividing the cumulative total of bone oc-
currences in squares by the number of
squares in each locality. These figures are
given for identifiable mammal and reptile
parts in Table 7. Locality 8+6-0104 has
the lowest concentration and 105-0208 the
highest. The three channel assemblages are
no more concentrated than the lacustrine-
deltaic ones for mammals, but are slightly
less prolific in terms of reptiles. There does
not appear to be any consistent correlation
between sediment grain sizes and identifi-
able bone abundance in the deposits
sampled.

The localities with more bones per square
do not appear to have more of any par-
ticular elements. Rather, they show an
increase in the frequencies of all skeletal
parts. This implies better conditions for
preserving bones of all kinds, regardless of
size and density, and argues against ac-
cumulation due to sel(>cti\'e proees.ses of
sorting (which would tend to concentrate
bones of similar sizes or densities or both ) .

532 Bulletin Museum of Co7nparative Zoology, Vol. 146, No. 10

Table 7. The square frequencies of reptile and mam-
mal SKELETAL PARTS IN THE SEVEN SAMPLE LOCALITIES. FRE-
QUENCIES ARE CALCULATED AS THE NUMBER OF SQUARES WITH
A PARTICULAR ELEMENT DWIDED BY THE TOTAL NUMBER OF
SQUARES IX EACH LOCALITY. ThE FREQUENCIES OF ASSOCIATED
PARTIAL SKELETONS AND JUVENILE BONES ARE CALCULATED IN
THE SAME MANNER. MaMMAL AND REPTILE BONES ARE COM-
bined in the second listing to include those which could
not be definitely assigned to one or the other class.
This shows the relatively high proportion of rib and
diaphysis fragments in the total bone sample.

DELTA

— CHANNEL—

FLOOD-

REPTILE

PLAIN

130-

0201

105-
0208

103-
0267

103-
0256

102-
0201

105-
1311

8+6-
0104

Tooth

.86

.85

.50

.33

.26

.68

.12

Skull/jaw

.10

.05

.30

.07

.15

.08

.00

Vertebra

.19

.05

.15

.19

.18

.00

.02

Limb

.14

.05

.15

.09

.00

Scute

.24

.40

.50

.30

.24

.40

.02

Phalanx

.10

.15

.00

.04

.00

.04

.00

Carapace/
plastron

.24

1.00

.70

.81

.24

.32

.12

# occurrences

1.9

2.5

2.2

1.9

1.1

per square
(average)

MAMMAL AND
REPTILE

Tooth

.95

1.00

.80

.59

.76

1.00

.67

Rib

.76

.90

.85

.63

.53

.64

.32

Pelvis

.14

.25

.04

.06

.16

.05

Diaphysis

.57

1.00

.95

.89

.71

.92

.55

Phalanx

.52

.55

.35

.30

.15

.32

.12

Vertebra

.57

.85

.55

.59

.29

.48

.15

Relative Abundance of Skeletal Parts

The frequency data in Table 7 can be
analyzed : 1 ) , in terms of the most common
bones in each locality and 2), in terms of
the correlations between localities caused
by similar proportions of different mam-
malian bones. Teeth are the most common
mammalian element in all localities except
103-0256. Otherwise, the patterns of fre-
quency are variable, with some indication
that vertebrae and phalanges concentrate
in the deltaic environments. In order to
clarify possible correlations between lo-

calities, two numerical analyses were used:
a multiple regression analysis, which gives
correlation coefficients for locality to lo-
cality comparisons, and a Q-Mode Factor
Analysis, which shows groupings of the
localities in terms of skeletal parts.

Correlations Based on Bone Abundance

Figure 20 shows a correlation matrix re-
sulting from multiple regression treatment
of skeletal part frequencies in the squares.
The correlation is "Pearson's product
moment correlation" which assumes con-

East Rudolf Paleoecology • Behrensmeyer 533

Table

nr 1 ■

7 (CONT.)

T A

p U A

FLOOD-

DLL in

tHAniNtL '

MAMMAL

PLAIN

130-
0201

105-

0208

103-
0267

103-
0256

102-
0201

105-
1311

8+6-
0104

Tooth

.67

.85

.70

.56

.62

1.00

.52

Jaw part

.24

.10

.04

.21

.08

.09

Maxilla

.05

.00

.02

Cranial part

.05

.25

.15

.07

.12

.08

Horn core

.19

.25

.35

.15

.18

.36

.03

Vertebra

.48

.75

.50

.59

.15

.36

.15

Sacrum

.00

.05

.04

.00

Scapula

.14

.45

.15

.12

.20

.09

Pelvis

.10

.20

.04

.03

.12

.05

Humerus

.19

.50

.30

.11

.06

.20

.14

Radius/ulna

.14

.40

.20

.11

.15

.20

.15

Femur

.14

.40

.05

.07

.18

.16

.08

Tibia

.10

.30

.25

.07

.06

.28

.14

Patella

.05

.00

.04

.00

Metapodial

.10

.40

.22

.18

.32

.14

Astragalus

.10

.20

.25

.11

.03

.08

.14

Calcaneum

.10

.15

.00

.12

.08

Podial

.10

.45

.10

.19

.24

.20

.18

Phalanx

.48

.65

.35

.26

.15

.28

.12

Total # squares

# occurrences

3.4

6.4

1.2

2.9

2.4

4.1

2.2

per square
(average)

Associated

parts
Juveniles

.04
.14

.20
.30

.05
.05

.04
.04

.00
.06

.00
.12

.06
.00

% hippo bones

18^;

16%

21%

12%

Total # squares for all localities: 213
Average occurrences per square: 690/213 =3.2

tinuous data and normal bivariate distri-
butions. Both conditions are satisfactorily
met by the squares data. Correlations are
based on the five most common elements:
teeth, vertebrae, phalanges, scapulae and
radii/ulnae.

An obvious feature of all the correlations
is that they are high (> .5). This shows a
basic similarit)' in the ratios of the five
skeletal elements in all the sample assem-
blages, although these elements vary
greatly in size and densit)'. Therefore, the

differences in the sedimentary environ-
ments were not enough to alter the basic
similarity of the thanatocoenoses sampled
in each deposit. This similarity is probably
produced by those bones most likely to
survive carnivore activity and become sedi-
mentary particles.

Many of the correlations shown in Figure
20 are significantly different, in spite of the
overall similarity. The highest and lowest
coefficients differ significantly, with a
probability of <.05 according to the "z test"

534 Bulletin Museum of Comparative Zoology, Vol. 146, No. 10

DELTA

CHANNEL

FLOOD-

PLAIN

130-
0201

1 .000

1.000

1 .000

105-
0208

.958

103-
0256

.851

.961

103-
0267

.950

.951

.915

102-
0201

.715

.629

.566

.816

105-
1311

.807

.751

.697

.897

.985

8+6-
0104

.716

.641

.595

.834

.995

.983

1.000

LOCAL- ►
ITIES

130-
0201

105-
0208

103-
0256

103-
0267

102-
0201

105-
1311

8+6-
0104

DELTA

■^ '^ CHANNEL^

FLOOD-
PLAIN

Figure 20. Correlation coefficients (Pearson's product moment correlation) between sampling localities accord-
ing to the proportions of the five most common skeletal parts: teeth, vertebrae, phalanges, radii/ulnae and
scapulae. Highest correlations show strong similarities between channel and floodplain environments in terms
of the proportions of different skeletal parts.

for significance (Simpson et ah, 1960:246).
Other coefficients are indicative of trends
even when their differences are not within
the acceptable limits of significance
(pC05).

The coefficients show that tlie channel
assemblages, 105-1311 and 102-0201, are
closely correlated with each other and with
the floodplain, 8+6-0104. The deltaic as-
semblages have relatively low correlations
with the floodplain, variable degrees of
correlation with the channels, and high
correlations among themselves. Thus the
proportions of the five different bones are
similar in similar sedimentary environ-
ments, showing the effects of processes
operating within these environments. Some
of the close interenvironmental correlations,
such as between the floodplain and channel
assemblages, and between 130-0201 and
103-0267, suggest processes that are com-
mon to more than one sedimentary situ-

ation. These can be further clarified by
examining which bones are influential in
causing the interlocality correlations.

Factor Analysis of the
Bone Assemblages

Factor analysis was used to indicate
which skeletal parts cause similarities or
differences among the seven bone assem-
blages. The Q-Mode Factor Analysis,
"CABFAC," was run on the frecjuency data
from all of the mammalian skeletal parts. A
solution of three varimax factors (axes
placed within the data array) explains 97%
of the total variance in the assemblages.
The projection of the data for each locality
on these axes is plotted on the triangle
diagram shown in Figure 21. The diagram
shows graphically how the three factors
group (cluster) the bone assemblages.

The three factors consist of 1) vertebrae

East Rudolf Paleoecology • Behrensmeyer 535

METAPODIALS.
TIBIAE. ETC.

(Voorhies Group II )

DELTA

CHANNEL AND
FLOODPLAIN

VERTEBRAE
AND PHALANGES

(Voorhies Group l)

TEETH

(Voorhies Group III)

Figure 21. Triangle diagram showing the results of a three-factor analysis of the frequency data for all mam-
malian bones. The factors correlate with Voorhies' dispersal groups, showing a relatively high proportion of
Group I (most easily dispersed) in the deltaic assemblages and Group III (lag) in the channel and floodplain
assemblages.

and phalanges, 2) teeth, 3) hmb parts such
as tibiae, metapodials, and astragaH. The
triangle diagram shows a clear separation
of assemblages on the basis of Factors 1 and
2. The three deltaic localities have a high
proportion of vertebrae and phalanges,
while the channels and the floodplain have
high proportions of teeth. 103-0267 falls
between the two groupings, and is some-
what anomalous in its lack of similarity to
the channel assemblages. Localities 103-

0267 and 8+6-0104 both show that there
is no strict correlation between tooth con-
centrations and coarse-grained sediment.
103-0267 is a coarse-grained deposit lack-
ing a high proportion of teeth; 8+6-0104
is fine-grained, but is characterized by a
high tooth concentration.

It is clear that the high correlation
coefficients between assemblages from
similar sedimentary environments are due
to tlie proportions of teeth, vertebrae and

536 Bulletin Museum of Comparative Zoology, Vol. 146, No. 10

Table 8. The relative frequencies of skeletal parts in a single skeleton, the average of bovid,

SUID, EQUID and HIPPO SKELETAL PROPORTIONS. UNDERLINED PARTS ARE THOSE WHICH ARE MOST COMTvION
OR MOST CONSISTENTLY PRESENT IN THE FOSSIL ASSEMBLAGES.

No. in

No. of each part/total

No. in

No. of

each part/total

average

no. of parts in

average

no. of

parts in average

skeleton

skeleton (156)

skeleton

skeleton (156)

Teeth

.26

Radii 'Ulnae

.01

Jaw

.01

Femora

.01

Maxilla

.01

Tibiae

.01

Cranium

.01

Patellae

.01

Horn Cores

.01

Metapodia

.03

Vertebrae

.18

Astragali

.01

Sacrum

.01

Calcanea

.01

Scapulae

.01

Podials

.14

Pelvis

.01

Phalanges*

.27

Humeri

.01

TOTAL

156

* Including metapodials of suid and hippopotamus.

phalanges. These two groups of skeletal
parts have very different properties of
density and destructibility. The deltaic
environments preserve more of the easily
transported and destructible elements, the
vertebrae and phalanges. The channel and
floodplain environments presei"ve more of
the denser and durable parts, primarily
teeth. The experimental data on bone
transport discussed in the section on bones
as sedimentary particles can be used to
interpret these differences in the fossil
assemblages.

Comparisons with Voorhies Groups

The Voorhies Groups consist of bones
with very different dispersal potentials. For
animals from suid- to equid-size, phalanges
and vertebrae are included in Group I, limb
parts in Group II to III and teeth in Group
II-III. Group I is most easily transported,
Group III least easily transported and
Group II intermediate, in currents up to
150 cm/sec, given Voorhies' ( 1969 ) experi-
mental conditions.

The three factors shown in Figure 21 are
closely comparable to the three Voorhies
Groups. Group I is more typical of the
deltaic assemblages and Group II of the
channel and floodplain assemblages. This

provides evidence that transport sorting
may be an important process in creating
differences between the bone assemblages,
i.e., the lag group is left behind in the
channels while the transportable group is
carried out to the deltaic and lacustrine
deposits. The loss of Group I in the flood-
plain may result from winnowing of the
lighter elements during floods, if the cur-
rent velocities on the floodplain exceed
10-20(?) cm/sec.

Single Skeleton Comparisons

Comparisons of the bone frequency data
with the percentages of different bones in
a single, whole skeleton show how the as-
semblages have been altered from their
original states. If all bones had been pre-
served together, then the correlations be-
tween the proportions of different parts in
the sample assemblages and a single skele-
ton should be high.

Average proportions of parts in a single
skeleton were calculated, combining the
most common mammal groups in the fossil
assemblages. These consist of bovids,
hippos, suids and equids. Frequencies of
the different parts are given in Table 8.
Figure 22 shows the comparison of fossil
and single skeleton bone frequencies for

East Rudolf Paleoecolocy • Bchrcnameyer 537

SQUARE FREQUENCY

no ^ a>
o o o

.'"^

..'V

.-^^""^

■f'/^

■

'/ /

>4 4

•■

-TOOTH

— VERTEBRA-

— SCAPULA

•-RADIUS/ULNA

-PHALANX'

SQUARE FREQUENCY

ro
o

4:^

C7>
O

cx>

\^ ^ *

\ ^ /

\ /'

/ \ /

y \,-^^ y

-^ ^^^ -^

X ^<

^^^^ /

^ rf^

1 ■/

1 / ■
1 / •.

L i» i?

^^ V ■

^^^
^V-

o Vll

» 4

FLOODPLAIN

CHANNEL

_.. < _..

CL (S 3

» T U3

0> — '

C 00

-■• o ^
CL -h (B

cr (B

3- O <-!•

-•• < o
•a -1. 3
•a Q.
o •

DELTAIC LOCALITIES

o I oj I

tn I o '

Figure 22. The square frequencies of the five most common mammalian skeletal elements in each locality
compared with the proportions of the same elements in a single, average skeleton. The localities separated by
factor analysis (Fig. 21) are distinct in their degree of alteration from single skeleton proportions.

the 5 most common or most consistently
occurring parts. The assemblages fall into
two obvious groups: 103-0256, 130-0201,
10.5-0208 and 103-0267 are closely cor-
related with the single skeleton, and 105-
1311, 102-0201 and 8+6-0104 are not.

It appears that the lacustrine-deltaic
environments, plus the 103-0267 channel-
beach complex, preserve skeletal parts with
a minimum of change from the original
proportions. This implies the absence of
processes that would sort the bones accord-
ing to size, density or destructibility. In
contrast, the channels, 102-0201 and 105-
1311, and the floodplain, 8+6-0104, pre-
serve altered assemblages with a high pro-
portion of the heavier and more durable
parts and a much lower proportion of the
lighter and more destructible elements.

Discussion of Evidence for
Transport Sorting

The combination of evidence from the
comparisons of bone assemblages with
Voorhies Groups and single skeletons leads
to important conclusions regarding the
histories of the bones in each locality. In
the deltaic deposits, bones from all Voorhies
Groups are present in proportions similar
to those of an average single skeleton.
Therefore, the major component of Group
I in these deposits is probably not trans-
ported from elsewhere (i.e., the channels).
If it were, then it has combined with lag
assemblages to closely approximate the
proportions in one skeleton. A better inter-
pretation for the deltaic assemblages is that
they have not been sorted. The relatively

538 Bulletin Museum of Comparative Zoology, Vol. 146, No. 10

fresh, unabraded surface textures of many
of the bones, plus their lack of hydraulic
equivalence with matrix grain sizes, further
supports this interpretation. The high pro-
portion of Voorhies Group I in the deltaic
assemblages is a product of nonselective
taphonomic processes rather than selective
ones.

In the 105-1311 and 102-0201 channels,
the concentration of teeth is the result of
sorting by fluvial processes. This sorting
combines the lag concentration of teeth
because of their greater density and because
of their greater durability in transport
situations. In addition, teeth are probably
concentrated from floodplain deposits as
the channel migrates laterally, eroding its
banks. Other parts derived from re-
excavated skeletons w^ould not be likely to
survive erosion unless already mineralized.

The 103-0267 distributary-beach complex
combines the sedimentary characteristics of
the other channels with a bone assemblage
similar to the deltaic ones. The assemblage
shows lack of selective sorting, and ap-
parently a large lag component of teeth was
not a product of the fluvial processes
operating in 103-0267. Why this should be
so is as yet unexplained.

The floodplain assemblage shows selec-
tive preservation of teeth in a fine-grained
sedimentary context. As mentioned on p.
536, this indicates the removal of lighter
elements from an untransported thana-
tocoenose. Such removal could result from
winnowing out of the light parts or from
surface weathering and preferential de-
struction of vertebrae and phalanges
relative to teeth. The relative importance
of these two processes can be determined
by future experimental work on the critical
entrainment velocities of vertebrae and
phalanges, and by observation of thana-
tocoenoses on modern floodplains. The low
frequency of horn cores in 8+6-0104 may
provide a clue indicating selective destruc-
tion by weathering, since in modern situ-
ations horn cores are often destroyed by
womis that feed on their organic constit-

uents (H. B. S. Cooke, personal communi-
cation) (Plate 3).

The extent to which sedimentary pro-
cesses have altered the bone assemblages is
clear from the examination of bone fre-
quencies. The deltaic assemblages are least
altered, and probably represent autochtho-
nous accumulations in sedimentary environ-
ments where the potential for rapid burial,
without re-excavation, is high. The flood-
plain assemblage is also autochthonous but
has been altered by taphonomical processes
so that it resembles the channel assem-
blages. These show the most extensive
alteration of bone ratios due to sedimentary
processes. 103-0267 is intermediate in the
degree to which taphonomic processes have
affected the bone assemblage.

The most useful localities for paleo-
ecologic information are thus established
as the deltaic and floodplain environments.
The channels will also prove useful, since
the factors contributing to their bone as-
semblages are known. They will include a
mixture of animals from the vicinity of a
fluvial system, in contrast to deltaic de-
posits, which preserve animals that fre-
quented lake margin habitats.

Additional Aspects of the
Bone Assemblages

More infomiation regarding taphonomic
history can be drawn from bone character-
istics unrelated to relative abundance.
These include the occurrences of associated
skeletal parts and the ratios of proximal
and distal ends of limb bones.

Associated parts of skeletons are rare in
the East Rudolf deposits in general. The
frequencies of these in the sample localities
are included in Table 7. The channels 105-
1311 and 102-0201 have none, while the
floodplain and deltaic localities, including
103-0267, have at least one. Most of these
consist of associated vertebrae, with more
complete partial skeletons occurring in 105-
0208 and 8+6-0104.

The associated skeletal parts may result

East Rudolf Paleoecology • Behrensmeyer 538

Table 9. Totals of proximal (P) and distal (D) limb ends

IN THE fossil ASSEMBLAGES FROM EACH SAMPLE LOCALITY.

-lA

— CHA

NNEL-

FLOOD-
PLAIN

130-
0201

105-
0208

103-
0267

103-
0256

102-
0201

105-
1311

8+6-
0104

TOTAL

Humerus

Radius/
ulna

1
1

3
2

2
1

10
4

Femur

Tibia

Metapodial

Total

106

from carcasses buried at the site of death
or from carcasses transported by flotation.
There are very few criteria that could be
used to distinguish between these two
possible taphonomic histories. However, the
associated parts do indicate a minimum of
reworking of the bone assemblages after
initial burial. This agrees well with other
evidence for lack of reworking of the delta
margin and floodplain assemblages. The
absence of associated parts in the channels
is consistent with the abraded surface tex-
tures of the bones as an indication for ex-
tensive reworking of sedimentary particles
in the channel environments.

Most of the limb bones in the samples
are represented by one end or the other.
It is of interest to determine whether some
ends are more common than others, as an
indication of preferential sorting or de-
struction prior to burial.

The numbers of proximal and distal ends
of the major limb bones are listed in Table
9. For all localities and all limbs combined,
the totals of 95 proximal and 106 distal are
very close to a 1:1 ratio. This might be
interpreted as indicating that no more
proximal than distal ends are preserved, or

vice versa. When the totals for each limb
in all localities are examined, however, the
relative frequencies of proximal and distal
ends prove to be quite variable. There are
nearly twice as many distal as proximal
ends of humeri and femora, and many more
proximal ends of radii/ulnae than distal.
From the density measures of proximal
and distal recent bones given in Appendix
2, it is apparent that the denser end is
more commonly preserved in humeri and
radii/ulnae, while the lighter end is more
common in femora. A model for differential
preservation because of transport sorting or
more rapid weathering of low density ends
does not fit this evidence.

The differences in frequency of the
proximal and distal ends can best be ex-
plained by carnivore activity. In animals
killed by carnivores or scavenged after
death, the limbs are usually pulled off
the carcass at the proximal articulation
( humerus/scapula and femur/pelvis joints)
(Muller, 1957:256-258). Proximal ends of
the humerus and femur would be subjected
to stress and later exposed for gnawing. In
contrast, the elbow and knee joints are
more likely to remain held together by

540 BuUetiu Museum of Comparative Zoology, Vol. 146, No. 10

ligaments and survive the l)one-cnisliing
activity directed at more nourisliing parts
such as marrow-filled diaphyses. This ex-
plains the relatively high proportion of
distal humeri and femora. The dispropor-
tionate number of proximal "radii/ulnae"
actually consist primarily of olecranon
processes from the ulnae. These are liga-
ment-covered, lack a marrow cavity, and
thus would survive better than the distal
ends of radii.

The pattern of proximal and distal limb
element frequencies can be regarded as
good evidence for carnivore activity in fos-
sil assemblages in general. Such evidence
has also been used by Voorhies (1969:20)
to indicate carnivore activity prior to the
final burial of the Pliocene Verdigre Quarry
bone assemblage. For the East Rudolf
localities, the evidence for carnivore ac-
tivity can be detected in spite of differences
in the taphonomic histories of the bone
assemblages in the different depositional
environments.

The Reptilian Assemblages

Reptilian parts form less consistent fre-
quency patterns than those of mammals.
The most common elements, as shown in
Table 7 are crocodilian teeth and chelonian
shell parts. The relative numbers of crocodil-
ian parts are very similar from locality to
locality. There is no indication of any in-
crease in similarity between assemblages
from similar sedimentary environments.
This is probably a result of the universal
availability of crocodile bones in the
aquatic (generally depositional) environ-
ments where crocodiles live. The low fre-
quency of crocodiles in the floodplain
environment, which crocodiles do not usu-
ally frequent, emphasizes this point. The
chelonian shell parts are variable in oc-
currence, and are slightly more abundant
in the deltaic deposits including 103-0267.
This suggests some correlation between the
more aquatic sedimentary environments
and the chelonian occurrences.

Conclusions Concerning the
Bone Assemblages

The evidence given in the preceding two
major sections of this study brings out a
number of taphonomically important fac-
tors that can be combined to support a
definite history for any given vertebrate
fossil assemblage comparable to those oc-
curring in the Koobi Fora Fm.:

1 ) The correlation of bone assemblages
with dispersal groups from Voorhies'
flume study

2) The correlation of bone assemblages
with the proportions of a single skele-
ton

3) The comparison of hydraulic equiva-
lents of bones with grain sizes in the
associated sediments

4) The completeness of bones, and sur-
face characteristics that indicate pres-
ence or absence of weathering or
abrasion prior to burial

5) Presence or absence of articulated or
associated skeletal parts

6) Ratios of proximal and distal ends of
limb bones that deviate from 1:1

All of these factors provide a basis for
interpreting the East Rudolf data. The
bone frequencies for each locality thus can
be used to determine the taphonomic
histories of the vertebrate assemblages. The
following points can be made:

1) The different sedimentary environ-
ments of East Rudolf show a general
similarity in the compositions of their
mammalian bone assemblages. The
same bones are present in all environ-
ments, and none of the assemblages
consist exclusively of one of the three
Voorhies Dispersal Groups.

2) Evidence for a certain degree of
sorting and redistribution of bones is
present in the different sedimentary
environments. Significant differences
in relative numbers of different bones
are shown by the concentrations of
teeth in the channels (105-1311 and

East Rudolf Paleoecology • Behrcnsmcyer 541

102-0201) plus the floodplain (8+6-
0104), and by the concentrations of
\'ertebrae and phalanges in the deltaic
localities ( 130-020l/ 105-0208, 103-
0256, 103-0267 ) . These can be corre-
lated with the sorting effects of tapho-
nomic processes in the channels and
on the floodplain, and the absence of
sorting on the delta margins.
3) Consideration of the bone frecjuencies
in the light of Voorhies Groups,
hydraulic equivalence and single-
skeleton comparisons shows that au-
tochthonous and allochthonous assem-
blages of fragmental vertebrate bones
can be distinguished. The deltaic and
floodplain localities consist of basic-
ally autochthonous vertebrate fossils,
while the channels contain a mixture
of allochthonous and autochthonous
assemblages.

FAUNAL ASSEMBLAGES OF THE
KOOBI FORA FORMATION

Taphonomic analysis has shown that all
of the sample fossil assemblages can be
considered autochthonous in the broadly
defined deltaic and fluvial environments.
It is now possible to examine the faunal
compositions of the seven bone assemblages
and to relate these to the different sedi-
mentaiy environments. Comparisons can
be made from environment to environment
which should indicate ti'ue paleoecologic
differences or similarities in the faunas.
In the following discussion, several aspects
of the paleoecology of the Koobi Fora Fm.
and its vertebrates will be given particular
attention. These include the differences
in numbers of aquatic and nonacjuatic
x'ertebrates, the relatixe frequency of dif-
ferent terrestrial mammals in the different
environments, and the patterns of occur-
rence of mammalian groups that havc^ close
counterparts in modern ecosystems.

The fauna from the square sample as a
whole includes 14 out of the 20 major
vertebrate groups listed by Maglio (1972:
380-381) for the Koobi Fora Fm. The

sample assemblages also include most of
the genera of bovids, suids, equids and
hippos. The carnivores listed by Maglio
(1972:380-381) are the most poorly repre-
sented groups in the samples used for this
study.

Method of Identification

The fossil collections consist of material
that can be identified at a number of dif-
ferent taxonomic levels. Major groupings
of vertebrates used for faimal comparisons
among the sample localities were desig-
nated so that each member of a group has
approximately equal numbers of identifi-
able parts. In practice, for example, this
amounted to teeth, skull parts, limb ends
and foot parts for mammals. The mammals
listed below could be identified e(}ually
well using any of these parts. Consideration
of this factor was necessary to prevent
undue biasing of the square frequency for
a form with substantially more or less
identifiable parts. The fossil assemblages
can be divided into faunal groups, cor-
responding roughly to several taxonomic
Categories, as follows:

1) Class: Mammal, Reptile, Bird, Fish.
Identifications were based primarily on
the morphology of the bone fragments.
Bone micro-structure was useful as a
distinguishing character for very small
fragments. In some cases parts of pelves,
scapulae, ribs and diaphyses could not
be certainly assigned either to mammal
or reptile, and such parts are not in-
cluded in any of the totals.

2) Groups of Reptiles and Mammals.

Mammals Reptiles

Elephant Suid Crocodyliis

Deinothere Eciuid Etitliccodou

Hippopotamus Primate Trionychid

Rhinoceros Carnixore Peloni<Hhisid

Giraffe Rodent C.cocliclonc

B()\ id \'aranid

The common denominator in these
groups is that they are approximately
equally identifiable, within each list. For

542 Bulletin Museum of Comparative Zoology, Vol. 146, No. 10

example, using the figures for relative
numbers it will be possible to determine
whether there are more hippos than
elephants in the sample assemblages, but
not whether there are more hippos than
trionychids. Some of the reptiles produce
many more abundant and readily identi-
fiable bones than do the mammals (e.g.,
crocodile scutes), and may appear rela-
tively more abundant in the samples.
Thus, relative frequencies are only com-
parable icithin the mammals and reptiles,
and some caution should be observed in
comparing the different reptilian groups
as well. Crocodylus and Eiithecodon are
comparable since they were identified
only on the basis of teeth, and the three
chelonian groups should be comparable
since only shell parts were used. Prob-
lems in separating small Geochelone
from Pelomedusids may lower the abun-
dances of these two groups relative to
the Trionychids.

3) Groups of Suids, Equids. Bovids and
Hippos.

Suids
Mesochoents
Notochoents
Metridiochoerus

Bovids*
Alcelaphinae
Large

= Mepalotragus
Small
= Damaliscus-like

Reduncinae
Tragelaphini
Bo villi (Pelovis)

Equids

Eqiius

Hipparion

( = "Stylohipparion" )

Hippos
Hippopotamus sp. (large)
H. sp. nov. (small)

(S. C. Savage,

personal communication )

* Nomenclature after
Ansell (1971).

These groups occur in significant abun-
dance in the sample assemblages and can
be compared between localities. Other
groups are represented by only one or
two occurrences and cannot be used for
comparative purposes. These include the
bovid groups Antilopini, Hippotraginae,
Cephalophinae and Neotragi, and a suid
that is probably Potomachoerus.

Identification of the suids was done
only on the basis of teeth. The equids
were distinguished using teeth and meta-
podials, and the hippos using relative size
of the skeletal parts identifiable as hippo.
Bovids were identified primarily on the
basis of teeth, but also from horn cores,
ends of limb bones, podials and
phalanges. The bone samples were
equally identifiable within each of these
groups. However, when the groups are
compared, the suid and equid genera
will have lower apparent abundances
than the bovid tribes and subfamilies
since they are represented by many
fewer identifiable parts.

Although various other groups of verte-
brates are identifiable to genera and spe-
cies, they are not discussed in this study.
These include primarily the fish and
chelonians. Future work may reveal
interesting patterns of abundance for
members of these groups in the different
sedimentary environments.

The relative numbers of all vertebrate
groups will be expressed as "square fre-
quencies," the percentage of squares in a
locality that contain a given animal. This
is the same measure as that used for skele-
tal parts (p. 529), and represents the "dis-
persed" or overall abundance of an animal
in a sample locality.

Abundance of Vertebrate Classes

The class frequencies (Table 10) show
that mammals, reptiles and fish are well
represented in all seven localities. Birds
occur in low frequency in two deltaic lo-
calities and in the floodplain. Mammals
are at least as abundant as reptiles and fish
in nearly all localities. Since most of the
mammals are terrestrial, while the fish and
most of the reptiles are aquatic, the relative
numbers of vertebrate classes in the chan-
nels and deltas is not directly related to
which animals were actually living within
the depositional environments. Otherwise,
fish and reptiles should be more common

East Rudolf Paleoecology • Behrensmeyer 543

Table 10. Class auundance in terms of square freqxjency.

Delta

Channel

Floodplain

130-0201

105-0208

103-0267

103-0256

102-0201

10.5-1311

8+6-0104

Mammal

.90

1.00

.82

.83

Reptile

.90

1.00

.95

.89

.62

.18

Bird

.00

.15

.00

.07

.00

.0.3

Fish

1.00

.95

1.00

.79

.30

tlian mammals. The large proportion of
mammalian fossils probably means that
there was a greater large vertebrate bio-
mass in the terrestrial habitats close to the
depositional environments. In the flood-
plain (8+6-0104), the relative numbers of
the classes are more closely correlated with
nonaquatic habitat preferences, and fish
and reptiles are much less abundant than
the predominantly terrestrial mammals.

Overall, the deltaic environments have a
larger proportion of fish and reptiles than
the channels. This is consistent with the
more permanent and extensive lacustrine-
deltaic aquatic environments, particularly
if the nondeltaic channels were seasonally
dry.

Calculated correlation coefficients based
on the class frec]uencies are high among
all localities (.97-.99) except for 8+6-0104.
Here, the larger number of terrestrial
animals over aquatic reflects a basic paleo-
ecologic distinction consistent with the
geologic interpretation of the floodplain
environment for 8+6-0104. In this case,
and probably for many similar fossil-bear-
ing environments throughout the geologic
record, the fauna which is preserved is pri-
marily terrestrial. The deltaic and channel
environments, in contrast, consistently pre-
serve mixtures of aquatic and nonaquatic
animals.

Abundance of Vertebrate Groups

Much paleoecologic information is avail-
able from the relative fre(|uencies of the
groups of mammals and reptiles listed
on p. 541. The East Rudolf samples are
best suited for such analysis since they

provide abundant, easily identified fossil
material representing these groups. The
relative frequencies of mammals and rep-
tiles are given in Table 11 and discussed
below.

Reptiles

Three groups of reptiles are represented:
crocodilians, chelonians and squamata. The
last is rare, and the sample consists of a few
vertebrae comparable in size to those of
Varanus (Monitor Lizard). Crocodilians
and chelonians occur in all areas, with
chelonians more variable in relative abun-
dance.

The crocodilians can be separated into
two groups (Genera): Crocodijhis (repre-
sented by at least two species) and the
long-snouted Euthecodon (represented by
at least one species ) . The two groups occur
in similar frequencies except in 130-0201
and 102-0201, where Crocodylus is more
abundant, and 103-0256, where Eutheco-
don is more common. It is unlikely that
sorting during transport had any significant
effect on the frequencies, since the hydrau-
lic properties of the teetli are similar.
Therefore, it is valid to conclude that, in
general, the t\vo crocodilians occupied over-
lapping ranges, i.e., both were present in
the deltaic and channel environments. The
living, long-snouted relatives of Euthecodon
(Tomistoma) are found in cjuiet, open
water (A. Greer, personal communication).
Specific habitat prc^ferences of Euthecodon
are not clear from its patterns of abimdance
in the sample localities at East Rudolf.
However, the conditions in 103-0256 after
the transgression of the lake seem to have

544 Bulletin Museum of Comparative Zoology, Vol. 146, No. 10

Table 11. The square frequencies of the mammal and

REPTILE FAMILIES, CALCULATED AS THE NUMBER OF SQUARES

WITH IDENTIFIABLE PARTS OF EACH FAMILY DIVIDED BY THE

TOTAL NXTNIBER OF SQUARES FOR EACH AREA.

DELTA

-CHANNEL-

FLOOD-

MAMMALS

PLAIN

130-

105-

103-

102-

105-

8+6-

0201

0208

0267

0256

0201

1311

0104

Elephant

.05

.25

.10

.04

.09

.36

.05

Dinothere

.00

.12

.08

.00

Hippopotamus

.62

.85

.70

.37

.41

.92

.21

Rhinoceros

.00

.10

.00

.06

.12

.02

Giraffe

.00

.15

.05

.00

.03

.28

.06

Bovid

.62

.75

.70

.63

.50

.96

.58

Suid

.29

.60

.20

.19

.32

.60

.27

Equid

.14

.30

.15

.11

.12

.56

.12

Primate

.05

.20

.00

.04

.06

.12

.03

Carnivore

.00

.05

.00

.04

.03

Rodent

.05

.00

.04

.00

REPTILES

Crocodilus

.86

.60

.45

.04

.21

.52

.09

Euthecodon

.48

.65

.55

.33

.12

.52

.03

Trionychid

.19

.70

.45

.30

.18

.25

.03

Pelomedusid

.05

.75

.20

.48

.00

.60

.08

Geochelone

.00

.05

.25

.33

.03

.04

.00

Varanid

.00

.05

.00

.04

.00

.04

.02

been exceptionally favorable to Euthecodon
and not so to Crocochjhis. Many articulated
parts, including complete skulls, have been
found in 103-0256 in addition to the
samples from the squares.

The chelonian sample consists of three
family groups: Trionychids, (soft-shelled
aquatic turtles), Pelomedusids, (semi-
aquatic to aquatic turtles), and Geoche-
lone, (land tortoise) (Loveridge, 1941;
Loveridge and Williams, 1957). Both
aquatic forms are common in all localities
except 130-0201, 102-0201 and 8+6-0104.
Trionychids are most consistently abundant.
The occurrences of Geochelone and the
Pelomedusids do not conform to any con-
sistent pattern that can be related to habitat
preferences of their recent counterparts.
The shell fragments have fairly low hy-

draulic equivalents ( < 3 mm except for
large Geochelone) and are readily trans-
portable. This may affect the proportions
of large and small turtles preserved in an
assemblage, and would tend to concentrate
large Geochelone in lag deposits. Sorting
should not affect the relative frequencies
of the other forms, if they are of compar-
able size ranges.

The pattern of chelonian occurrences in
the fossil samples is not readily explained
either by sorting or by inferred habitat
preferences. Other factors are involved,
and one of these probably concerns eco-
logical preferences of the fossil turtles that
cannot be adequately inferred from the
modern ones without identification of the
fossil material to species. This is further
complicated by the probability that factors

East Rudolf Paleoecology • Bchrensmcyer 545

influencing turtle abundance (e.g., water
turbidity, vegetation) may not be evident
from the sedimentary record.

The low frequencies of chelonians in the
floodplain assemblage are consistent with
all other evidence for its predominantly
nonaquatic environment of deposition. In
other localities, at lea.st some crocodilians
and chelonians inhabited the environments
of deposition. This is indicated l)y the
variety of skeletal parts of crocodilians
present in all deltaic and channel assem-
blages (parts with a wide range of hy-
draulic equivalents). It is also inchcated
by the presence of associated, unreworked
parts of both crocodilians and chelonians
in many of the localities.

Mammals

The numbers of different mammalian
groups represented varies among the locali-
ties (Table 11). The 105-1311 channel
assemblage is most diverse, with 11 groups
represented. 105-0208, which has the
largest number of identifiable bones (2389),
has a relatively low faunal diversity (8
groups). In general, the deltaic environ-
ments have lower numbers of different
terrestrial mammals. The short time span
postulated for the deposition of 103-0256
(the transgressive sand) may help to ac-
count for its low faunal diversity. As would
be expected, the more terrestrial deposi-
tional environments preserve more kinds of
terrestrial animals. As many mammal bones
are preserved in the more aquatic environ-
ments as in the nonaquatic ones, but they
represent fewer terrestrial groups. The
relative diversity of the fossil mammal
assemblages apparently gives a true repre-
sentation of the greater di\'ersity of forms
in the more terrestrial habitats.

The most common groups in all localities
are bovids, hippos, suids and equids. Bovids
are the dominant forms in all assemblages
except 130-0201, 105-0208, and 103-0267,
where hippos are slightly more numerous.
The frequencies given in graphic form in
Figure 23, show similar patterns for most

of the assemblages. For the autochthonous
assemblages of the deltaic and floodplain
en\'ironments, these probably reflect the
actual frequencies of animals in the death
assemblages over the time periods sampled.
The channel assemblages also should be
broadly representative of the original rel-
ative abundance of the mammalian groups
in the habitats sampled by fluvial processes.
However, there is a definite bias against
the smaller animals due primarily to their
greater destiiictibility and transportability
in all environments.

Rodents are rare throughout the East
Rudolf vertebrate-bearing deposits. Since
they are abundant in recent terrestrial habi-
tats in the area, there can be little doubt
that they are under-represented because of
taphonomic processes that did not lead to
preservation with the larger vertebrates.
These probably involve carnivore destruc-
tion, rapid surface weathering of bones
( due to high surface area to volume ratios )
and high dispersal potentials (Dodson,
1974). The difficulties in collecting very
small bones from the sample squares is
another potentially important factor, al-
though care was taken to minimize this
during the sample collection.

Carnivores and primates have low fre-
quencies in the samples, and this is partly
a result of the same size factors affecting
the rodent sample. However, some of the
primates and carnivores reach sizes com-
parable to those of the smaller bovids and
suids, and factors relating to size do not
entirely explain the low frequencies. For
carnivores, a low abundance compared with
herbivores is consistent with their low
relative biomass in the Eltonian Pyramid,
in which a few carnivores are ecologically
balanced with large numbers of herbivores.
The scarcity of carnivore fossils probably
reflects this ecological character. Primates
are comparable in abundance to giraffes
and rhinos in most localities. The samples
consist primarily of baboons (Sitnopithe-
cus), plus two occurrences of hominids (cf.
Atistralopithecus, in the 105-1311 channel).

546 Bulletin Museum of Comparative Zoology, Vol. 146, No. 10

l.OOi

DELTA CHANNEL

-^ 130-0201 • . 102-0201

105-0208

- 105-1311

FLOODPLAIN

8+6-0104

a-
(/I

n —

Figure 23. Comparisons of the square frequencies of mammal and reptile groups in the seven sample localities.

East Rudolf Paleoecology • Behrensmeyer 547

The relatively low frequency of primates is
probably due to a combination of destruc-
tibility, high dispersal potential and lower
original abundance or more localized
groups than the better represented ungu-
late families.

The extinct Deinotheres are represented
only in the 10.5-1311 and 102-0201 chan-
nels, together with rhinos. Rhinos also
occur in the floodplain assemblage, in the
Loxodonta Faunal Zone (Fig. 14), which
may represent a time (~ 1.3 my.?) when
deinotheres had become locally extinct.
Neither is found in the deltaic environ-
ments sampled. Elephants, whose parts
ha\'e similar low dispersal potentials, are
found in all localities. The patterns of oc-
currence are best explained by habitat
preferences, with the deinotheres and
rhinos preferring dry, upland savanna and
bush habitats while the elephants were
more ubiquitous.

Discussion of the Most Abundant
Mammalian Groups

The relative frequencies of the four most
common groups, hippos, bovids, suids and
equids, can be combined with data on the
skeletal representation of each family for
more detailed analysis of taphonomic and
paleoecologic factors.

Comparisons of the frequencies given in
Table 11 show that there is a close simi-
larity^ between localities in the proportions
of the four families, except in the cases of
103-0256 and 8+6-0104. The mammals of
these localities include high frequencies of
bovids and low frequencies of hippos. For
the floodplain environment, this correlates
nicely with the low representation of other
aquatic forms. However, for the trans-
gressive deposit in 103-0256, with its abun-
dant aquatic fauna, the low proportion of
hippos is anomalous.

All four common groups are represented
by teeth and other skeletal elements of
different sizes and densities. These would

have different dispersal potential, accord-
ing to the Voorhies Groups (Table 2) and
size-density considerations. If any of the
mammal groups were transported into the
fo.ssil assemblages as disarticulated skeletal
parts, then they should be represented by a
higher proportion of phalanges, podials,
etc. Autochthonous animals should have
mixed representation with teeth, limb parts,
phalanges, etc.

For bovids and hippos, elements of
widely variable original size and density
occur together in all localities. These in-
clude teeth, phalanges and ends of limb
bones, which fall into all three of Voorhies'
dispersal groups. It is evident that these
assemblages do not reflect extensive sorting
by taphonomic processes. These processes
may have been operating, particularly in
the channels, but the bone input was
enough to maintain an unsorted character
in the assemblages. The best explanation
for the skeletal representation is that the
bone input was generally local, and the
bovids and hippos are autochthonous with
respect to the deltaic and fluvial environ-
ments.

Suids and equids are less abundant as
fossils and are also represented by a lower
diversity of skeletal parts. Both teeth and
phalanges are present in the suid samples
from all areas except 105-1311. Equids are
represented by teeth, with only a few
examples of other parts. The suid samples
are more likely to be autochthonous since
they combine elements of widely differing
transport potential. The equid sample con-
sists mainly of lag elements, main- of which
show evidence of transport abrasion. These
are probably allochthonous in the channels,
but are at least partly autochthonous in the
floodplain and deltaic assemblages. For
both suids and equids, the relativelv low
frefiuenci{\s and poor skeletal representa-
tion imply lower original abundance than
bovids in the floodplain and deltaic habi-
tats.

The comparison of tooth frequency with
the frequencies of all other elements is a

548 Bulletin Museum of Comparative Zoology, Vol. 146, No. 10

130-OJOl 105-010* I03-OJ56 103-OlWT 103-0301 103- I3i

8-6-010+

DELTA

CHANNEL

FLOOD
PLAIN

130-oaoi

IOS-010« lOB-OaSfc l03-01t7 101-0301 )05-lBll

«.t-0lO4-

.80

>•

o .60

on
u.

^ .40
a

.20

East Rudolf P.\leoecology • Behrensmcyer 549

useful measure of the eff(X'ts of taphonomic
processes (weathering and transport) on the
fossil assemblages. Figure 24 shows the
frequencies of the four families in terms of
all identifiable elements and in terms of
teeth only. Where the lines divc>rge, a large
proportion of the sample consists of parts
other than teeth. Representation of hippos
and bovids is similar in all localities except
103-02.56 and <S+6-0104. The low fre-
quency of hippos in these t\vo assemblages
is due to a lack of elements other than
teeth. Since the bones from both localities
are basically autochthonous, this empha-
sizes that there must have been relatively
few hippos leaving bones in the original
environments. The plots for suids and
equids demonstrate the relatively greater
numbers of teeth in these samples. The
autochthonous or allochthonous context of
these animals in the bone assemblages can-
not be inferred from their skeletal repre-
sentation, but is indicated by their associ-
ation with autochthonous or allochthonous
assemblages, as determined by the tapho-
nomic histories of each assemblage as a
whole.

The assemblage of 103-0267, which com-
bines characteristics of channel and deltaic
environments, includes a high proportion
of different hippo and bovid skeletal parts,
in conti-ast to the other channel assem-
blages. This indicates a closer similarity to
the taphonomic histories of assemblages in
130-0201 and 105-0208, than to those of the
other localities. Considering all evidence
for 103-0267, it appears to have been much
more permanently aquatic than the other
channels. This would fit an interpretation
of distributary channels close to the lake in
association with beach environments.

It is of interest that the known aquatic
or nonaquatic habits of bovids and hippos

cannot be inferred from their overall fre-
(juencies or skeletal representation in the
deltaic or chamiel assemblages. Without
knowledge of recent ecology, it would only
be valid to say that both hippos and bovids
are autochthonous in the deltaic and fluvial
environments. Bevond this, anatomical
studies would be necessary for interpreting
ecological differences. The lesson in this is
important for fossil asscMiiblages with no
modern counterparts: hippo.s and bovids
are an example of two distinct groups with
comparable fossil representation that does
not necessarily reflect their ecological dif-
ferences, even though one is more closely
tied to aquatic habitats than the other.

Abundances of Selected
Mammalian Groups

Sums

Three genera of suids are common in
the fossil assemblages and occur in variable
frequencies from localit)' to locality. Table
12 gives the frequency data for these
genera, based only on teeth. Notochoerus
and MetricUochoerus are combined since
molar fragments of the two are difficult
to distinguish. They are similar in hav-
ing high-crowned, elongate third molars
adapted to eating the relatively abrasive
grasses, and both are thought to belong to
the same group as Phacochoerus, the
modern warthog (Cooke and Maglio, 1972:
312), which is generally an open-habitat
form (Dorst and Dandelot, 1970:174).
Mesochoerus is easily distinguished from
the other suids by its low-crowned third
molar. Its dentition is adapted to softer
vegetation than in MetricUochoerus and
Notochoerus. Mesochoerus is beliexed to
be close to the ancestral stock of the recent

Figure 24. Comparisons of skeletal representation of the four most common mammal groups in each sample
locality. Solid circles and dashed (lower) lines indicate the frequencies of all squares with teeth. Open circles
and solid (upper) lines indicate the total square frequency, counting all teeth and bones. The space between the
lines is large if a group is represented by parts other than teeth, Bovids and hippos show comparable skeletal
representation except for the delta flats (103-0256) and the floodplain (8+6-0104).

550 Bulletin Museum of Comparative Zoology, Vol. 146, No. 10

Table 12. Square freqtjencies of Stno genera.

Delta

Channel

Floodplain

130-0201

105-0208

103-0267

103-0256

102-0201

105-1311

8+6-01014

Notochoenis/
Metridiochoerus

Mesochoerus

.00
.14

.20
.30

.05
.00

.04
.04

.24
.03

.24
.04

.01
.08

Hijlochoerus (Giant Forest Hog), a dense blages, and Mesochoerus is more typical of

bush or forest animal (Cooke and Maglio, the deltaic assemblages. Chi-square tests

1972:311). (Simpson et al, 1960) show that this dif-

Figure 25 shows that the Notochoerus ference is significant with p ^ .05. Paleo-

group {= Notochoerus + Metridiochoenis) ecologic separation of the two groups is

is far more common in the channel assem- strongly indicated, and in a way that is con-

>.20

.10

Mesochoerus

y \ Metridiochoerus/

Notochoerus

J-

130-
0201

105-
0208

102-
0201

105-
1311

DELTA

CHANNEL

8+6
0104

♦
FLOODPLAIN

Figure 25. A comparison of tlie square frequencies of the two suid groups, Mesochoerus and Notochoerus/
Metridiochoerus. Mesochoerus is considered here to be a more closed habitat (bush) form due to its relation-
ship to the modern Hylochoerus (Giant Forest Hog), and to its low-crowned molars, which appear to be adapted
for relatively soft vegetation. Notochoerus/ Metridiochoerus suids are more closely related to the modern Phac-
ochoerus (Warthog) and have high-crowned molars adapted for abrasive vegetation. These suids may have
been more open habitat (grassland) forms. Localities 103-0256 and 103-0267 are omitted due to the low fre-
quencies of suids identifiable to genus (Table 12).

East Rudolf Paleoecology • Behrensmeyer 551

Table 13. Generalized ecological characteristics of Recent bovid groups that are commox in
THE East Rudolf fossil assemblages. ( Modified from R. Estes, ms. in press. )

Water projcimity

IIal)itat

Near

Far

Food h.tbits

Social habits

Tiagelaphini

Dense Bush

(X)

Browsers

small groups

Reduncinae

Woodlands, Floodplains

Grazers

small groups

Alcelaphinae

Open grasslands

(X)

Grazers

large herds

sistent with predictions based on tooth
morpliology and recent analogues. The
channel assemblages might be expected to
sample the more open-country habitats,
particularly if the gallery forests fring-
ing the channels are not very extensive.
Deltaic environments, if comparable to
the most vegetated areas of the recent
Omo Delta, would have more forested
habitats. The paleoecologic evidence associ-
ates Mesochoerus with deltaic, potentially
more densely vegetated environments and
Notochoerus/Metrkliochoerns with fluvial,
mixed- to open-habitat environments.

The third molars of the two suid groups
are different in size, and there is a possi-
bilitv that the smaller Mesochoerus teeth
have been sorted out of the channel de-
posits. Mesochoerus third molars are
between about 40 and 60 mm long and 20
mm in height, while Notochoenis/MetrkUo-
choerus third molars are from about 50 to
75 mm in length and 40 to 60 mm in height.
However, the hydraulic equivalents of the
teeth in both channels fall within the 10-25
mm range, which is near the median for the
total range of other teeth in the deposits as
well as the associated sediment. One cannot
logically assume that sorting would sepa-
rate the pig teeth but nothing else with
similar size differences. Therefore, sorting
can be eliminated, and ecological factors
provide the best explanation for the sepa-
ration of the two suids in the fossil assem-
blages.

EQuros

Although the equid sample is poor and
consists mainly of teeth (Figure 24) there

is some suggestion of habitat separation of
Eqtms and Hipparion in the samples.
Eqitus is most abundant in 105-1311 and
102-0201, and Hipparion in 130-0201 and
105-0208. There is a time separation be-
tween these two groups of samples, but
Eqtms is known to occur elsewhere in East
Rudolf at the same level as Hipparion. The
correlation of Eqtms with the channel envi-
ronments and Hipparion with lake margins
is comparable to the pattern of occurrences
of N otochoertis /Metridiochoerus and Meso-
choerus. Hipparion is preserved in associ-
ation with the environment most likely to
have been densely vegetated, and Eqims
is found in the deposits more likely to have
sampled open country, savanna forms.

BoviDS

Three bovid groups are abundant enough
in the sample assemblages for detailed
analysis. These include the Alcelaphinae
(hartebeest, etc.), Tragelaphini (kudus,
elands, etc.) and Reduncinae (bush buck,
waterbuck, etc. ) . Recent members of these
groups are well known in terms of habitat
preference (Bigalkc, 1972; Estes, in press;
Dorst and Dandelot, 1970). Ecological
characters are listed in Table 13. Fre-
quencies of bovid tribes in the fossil as-
semblages are given in Table 14.

All localities combined, alcelaphines and
reduncines are nearly equal in abundance,
while Tragelaphines are less common. The
high frecjuencics of the smaller alcelaphines
in 103-0256 and of both large and small in
105-1311 are significantly larger than the
frequencies of the other groups in these
localities, with p ^ .05 (Chi-square tests).

552 Bulletin Museum of Comparative Zoology, Vol. 146, No. 10

Table 14. Frequencies of fossil bovids and the small hippopotamus in the East Rudolf

assemblages.

Delta

Channel

Floodplain

130-0201

105-02

103-0267

103-0256

10;

J-0201

105

-1311

8+6-0104

Tragelaphini

.24

.10

.20

.00

.12

.16

.02

Reduncinae

.24

.25

.15

.11

.15

.12

.15

Alcelaphinae

( Damaliscus-size )

.14

.15

.05

.33

.06

.40

.12

Alcelaphinae
( Megalotragus )

.00

.15

.04

.00

.48

.03

Hippopotamus

.05

.20

.10

.26

.09

.00

.06

sp. nov.

Reduncines and tragelaphines are of similar
frequency except in 103-0256 and 8+6-
0104, which have a high proportion of
reduncines and few tragelaphines. These
differences are probably related to eco-
logical factors since it is difficult to
imagine any other processes which could
preferentially sort the tribes. (They are of
approximately equal body size.) All are
represented by multiple skeletal parts
where they are abundant. The patterns of
occurrence are not as well defined as for
the suids, but bush forms ( reduncines ) are
generally more common in the deltaic
environments while alcelaphines are as-
sociated with the potentially more open
environments sampled in the 105-1311
channel and in the 103-0256 mudflats.

The alcelaphine Megalotragus is a large,
extinct form known only from teeth in the
sample assemblages. It has an unusually
high frequency in the 105-1311 channel.
This is significantly different from its oc-
currences in the other localities and can-
not be explained except by ecological fac-
tors. Megalotragus is associated with the
grassland suids and equid, and may well
have been an open-country form itself. Its
absence in 102-0201 is somewhat puzzling,
if it is typically preserved in fluvial deposits.
However, another large bovid (probably
Pelorovis) is represented in 102-0201 with
a frequency of .18. The data may indicate
a rather finely resolved habitat separation

between the two forms which can only be
clarified by additional sampling.

Hippos

The habitat of the extinct small hippo,
H. sp. nov., can be generally inferred from
its frequency in the bone assemblages
(Table 15). It occurs in all localities ex-
cept 105-1311 and is most abundant in
103-0256. Nearly all the hippo remains
in 103-0256 belong to this form, and a
variety of skeletal parts exist in the sample
squares together with teeth. It is generally
more abundant in the deltaic environments,
including 103-0267, and is associated with
both bush and open country animals. It is
definitely autochthonous in the deltaic
mudflats environment of 103-0256. From
this evidence, it can be concluded that H.
sp. nov. was probably a lake margin fonn,
preferring deltaic flats with mixed bush and
grassland environments. It may have been
less aquatic than the larger hippos, but this
can only be validly inferred from morpho-
logical data, not from the taphonomic evi-
dence now available.

A large extinct hippo, peculiar to the
East Rudolf Plio-Pleistocene, cannot be
definitely identified in the fossil assem-
blages from the squares. Ecological data
on this hippo would be interesting since
somehow three or more forms of Hippo-
potamus managed to coexist at East Rudolf.

East Rudolf Paleoecology • Behrensmeyer 553

SEMI-AQUATIC TO AQUATIC

(hippo, Crocodylus Euthecodon)

TERRESTRIAL

(bovid, suid, equid)

PRIMARILY AQUATIC

(Euthecodon. trionichid,
pelomedusid)

Figure 26. Triangle diagram showing the results of a three factor analysis of the frequency data from all groups
of mammals and reptiles. The deltaic localities are spread between two "aquatic" factors, and the channel and
floodplain localities are distributed closer to the "terrestrial" factor.

At least one of these is not present in the
entire Omo seqvience, suggesting significant
ecological differences between the two
regions, at least as far as the hippos were
concerned. We might assume that these
differences were expressed in the utilization
of broadlv different environments. If so,
further careful sampling of the sedimentary
e\'idence should reveal more about the
ecology of the different hippos.

Conclusions Regarding the
Faunal Assemblages

Much of the information provided b\' the
faunal frequencies is summarized in the

triangular diagram in Figure 26. This
shows the results of a CAi3FAC Q-Mod(>
Factor analysis for three varimax axes
(which explains 97'/' of the variance). The
data consist of the square frequencies for
all the animal groups given in Table 11.
The three factors can be direct!)' r(>lated to
the acjuatic or nonacjuatic affinities of the
various animals. Factor 1 includes the
terrestrial forms and the other two factors
include aquatic and semiaciuatic forms with
affinities for channel or deltaic-lacustrine
habitats. These separate the sample^ locali-
ties into three groups depending on their
components of ac^uatic animals.

Thus, the evidence at various taxonomic

554 Bulletin Museum of Comparative Zoology, Vol. 146, No. 10

levels indicates faunal differences between
the localities which agree with environ-
mental interpretations based on geological
data. These are real and meaningful
ecological differences between environ-
ments; differences expressed in the square
frequencies of the faunas and supported by
the geologic and taphonomic characters of
the sediments and their bone components.
The important points brought out by the
faunal data include:

1) The relative numbers of the different
vertebrate classes in the fossil assem-
blages is more dependent on their
original numbers and proximity of habi-
tat to a sedimentary environment, than
on their aquatic or nonaquatic habits.

2 ) Relative numbers of animals can indi-
cate whether they were aquatic or
nonaquatic in nonaquatic sedimentary
environments, but not in aquatic ones.

3) The more terrestrial sedimentary en-
vironments preserve a greater diversity
of terrestrial animals. However, the more
aquatic environments may preserve an
equivalent or greater number of bones
from terrestrial animals, representing
fewer groups.

4) For terrestrial families larger than a
baboon or a small antelope, the frequen-
cies expressed in the sample assemblages
should be roughly proportional to their
original numbers.

5 ) The more abundant mammals (bovids
and hippos) are generally represented by
more kinds of skeletal parts of different
sorting potential, indicating autochtho-
nous accumulations of bones. This is
consistent with the generally autochtho-
nous nature of the assemblages on the
deltaic and floodplain environments.

6) The different sedimentary environ-
ments clearly preserve different ratios of
some animals with different habitat
preferences. Terrestrial animals which
prefer grassland habitats are found in
greater abundance in fluvial deposits,
while bush or forest mammals occur in
greater abundance in the deltaic deposits.
This indicates deltas with denser vege-
tation than the gallery forests which
fringed the channels while the sample
bones were accumulating.

PALEOECOLOGY OF THE
VERTEBRATE ASSEMBLAGES
OF THE KOOBI FORA
FORMATION

Much independent but cross-supporting
evidence provides a basis for interpreting
the paleoecology of the East Rudolf fossil
assemblages. These lines of evidence in-
clude:

1) Geologic evidence. Characteristics of
the overall sedimentary environments
and the processes operating within them

(Fig. 27).

2) Taphonomic evidence, a) The extent
and type of sorting in the bone assem-
blages, interpreted with the aid of
theoretical considerations from experi-
mental evidence for bone disposal,
b) Relationships of the hydraulic equiva-
lences of the bones and of the associated
matrix sediment, c) Characteristics of
the bone fragments, interpreted accord-
ing to observations on weathering, frac-
turing and abrasion of modern bones.

3) Faunal evidence. Interpretations based
on the faunal composition of the fossil
assemblages, using the ecology of modern
analogues to Plio-Pleistocene animals.

Figure 27. Block diagrams showing reconstructions of East Rudolf sedimentary environments. Circles show
the interpretation of the general sedimentary environment of each fossil sampling locality, indicated by locality
numbers. The representation is schematic; the localities do not occur on the same time planes or closely adja-
cent to each other as might be construed from the diagrams. 102-0201 is more closely associated with an
emergent delta than can be shown on the diagram. 103-0267 includes a more extensive complex of distributary
mouth and beach environments than is indicated by the encircled area.

East Rudolf Paleoecology • Behrensmeyer 555

8+6-0101

FLUVIAL SYSTEM

103-0256

130-0201

105-0208

DELTA

556 Bulletin Museum of Comparative Zoology, Vol. 146, No. 10

1.00

.80

o
o

or
S.

.60

.40

.20"

.00

Increase in
Aquatic component

^
§

I Oa Aaaac

130-0201
103-0267

103-0256

Delta Margin

Distributary and
Beach Complex

Deltaic Mudflats

8+6-0104 Floodplain

Figure 28. A comparison of the square frequencies of four vertebrate groups with aquatic, semiaquatic and ter-
restrial habits in four different depositional environments. These environments range from primarily aquatic to
primarily terrestrial in terms of their geologic characteristics. The frequencies of the aquatic and semiaquatic
animals increase as the depositional environments become more aquatic (floodplain to delta margin). However,
the frequencies of terrestrial forms remain essentially constant in all environments.

East Rudolf Paleoecology • Bchrensmeyer 557

Other lilies ot exidence can be important
in paleoecologie interpretations but are not
at present available for the East Rudolf
assemblages. These include botanical and
gcochemical data, which can reveal im-
portant factors about vegetation, climate,
and salinity of the lake. Continuing re-
search should eventually provide such data.

Ecological Comparisons
of the Samples

The overall similarities and differences
among localities show that sedimentary en-
\'ironments can be characterized according
to distinct taphocoenoses and biocoenoses.
At East Rudolf, three broadly defined
sedimentary environments are represented:
delta, channel and floodplain. The faunas
are all basically autochthonous in each of
these environments and reveal meaningful
ecological differences among them.

Aquatic and Terrestrial Faunas

The more aquatic sedimentary environ-
ments as detennined from geologic evi-
dence have an increased rejjresentation of
a({uatic animals but show no decrease in
the absolute number of terrestrial animals.
This is demonstrated in Figure 28 by the
increase in the frequencies of crocodilians
and hippos relative to bovids and equids.
I The absolute frequency of bovids and
eciuids does not change significantly from
environment to environment, even though
these range from floodplain to delta margin.
The pattern of aquatic and terrestrial
occurrences can be represented for faunas
from each locality as shown in Figure 29.
The ratio of terrestrial animals increases as
environments become more terrestrial, at
the expense of the acjuatic forms. The ratio
of the semiaquatic hippo, which spends
approximately half its time in and half out
of the water (Dorst and Dandelot, 1970:
172), changes little from aquatic to non-
aquatic environments. These patterns are
the result of geologic and taphonomic
processes which should have similar effects

on fossil assemblages other than these East
Rudolf examples. The crucial variables
appear to be: 1) the total volume of bones
available from acjuatic and nonaquatic
animals and 2) the pro\imit\- of an animal's
habitat to an actively aggrading sedimen-
tary environment. The habits of a fossil
vertebrate cannot be inferred from its abun-
dance in an a(iuatic sedimentary environ-
ment unless this can be compared with more
terrestrial environments from about the
same time.

Open and Closed Habitat
Mammalian Faunas

The mammalian assemblages provide
evidence for two terrestrial faunas with
preferences for open (grassland) or closed
(bush) habitats. In order to establish these
ecological differences, the habitats of the
fossil mammals must be inferred from
morphologic evidence plus analogy to re-
lated living forms. For the paleoecologie
interpretation of the East Rudolf fossil
assemblages, Mesochoerus, reduncines and
tragelaphines are used to represent the
closed habitat fauna, and DamaJiscus-size
alcelaphines, Notochoems/MetridiocJwerus
suids and Equus represent the open habitat
fauna. The evidence for relating these
mammals to the different ecologic situ-
ations has been discussed previously. The
ecological separation of such groups accord-
ing to habitat preference for grassland or
bush environments is a common feature of
recent East African ecosvstems (Lamprey,
1963; Harris, 1970; Estes,' 1973).

The relative percentages of closed and
open habitat forms in the seven fossil
localities are shown in Figure 30. All
localities include both, but the deltaic
environments in general include more
closed habitat forms and the channels more
open habitat forms. The deltaic mudflats
(103-0256) have an open habitat fauna,
in agreement with geologic e^'idenee for an
extensive, unforested delta margin environ-
ment. The patterns of faunal occurrence
indicate that the deltaic, chaimel and flood-

558 Bulletin Museum of Comparative Zoology, Vol. 146, No. 10

o 80

TERRESTRIAL

VERTEBRATES

Environments Increasingly Terrestrial ^

Figure 29. The ratios of aquatic, semiaquatic and terrestrial vertebrate groups represented in the East Rudolf
fossil assemblages. The sample localities are arranged from the more aquatic to least aquatic depositional en-
^'ironments on the basis of geologic interpretations. Aquatic animals include crocodilians and chelonians except
Geochelone; semiaquatic includes only hippopotamus, and all other groups are considered to be terrestrial.
Abundance is calculated as the % of the cumulative square frequencies for each locality.

plain sedimentary environments sample
both closed and open habitats, but that
closed habitats were more abundant on the
deltas. A comparison between 130-0201,
representing a deltaic fauna, and 105-1311,
representing a nondeltaic fauna, shows the
most distinct ecological difference among
any of the localities (Fig. 31).

Comparisons of Koobi Fora Formation
Faunas and Recent Terrestrial Faunas

Bovids, suids and equids are the most
abundant large mammals in the fossil as-
semblages and also in most of the recent
undisturbed East African ecosystems (e.g.,
Foster, 1967; Sheppe and Osborn, 1971).
The ratios of these mammals in the fossil

East Rudolf Paleoecology • Bchrcnsmcijcr 559

TOO

u
o

i-
<u

»-

LU
Q-
O

CQ
<

Q
LU
CO
O

—1

130-
0201

105-
0208

103-
0267

103-
0256

102-
0201

DELTA

J \.

CHANNE

105-
1311

8+6-

0104

FLOOD'

PLAIN

OPEN HABITAT FAUNA

CLOSED HABITAT FAUNA

Figure 30. Histogram showing the relative percentages of closed and open habitat mammals in each of the
sample localities. Closed habitat forms include Mesochoerus, reduncines and tragelaphines; open habitat forms
include Damaliscus-s'\ze alcelaphines, Notochoerus/ Metridiochoerus and Equus. Percentages were calculated
on the basis of the total square frequencies for the closed and open habitat groups.

faunas is compared witli tlieir relative
numljers in recent faunas in Figure 32.
Bovids are most common in both cases.
However, equids are less common than
suid.s in all seven of the fossil sampling
localities, and the faunal proportions are
most similar to the recent Kafue Park fauna
in Zambia. Taphonomic causes do not

adequately explain the greater fre(|uency
of suids in the fossil assemblages. There-
fore, e(juids may have lower representation
than suids bc^cause they were more eco-
logically .separated from the sedimentar\'
environments, or because they were gener-
alK less abundant in the East Rudolf region
during the time period represented by the

560 Bulletin Mmcum of Comparative Zoology, Vol. 146, No. 10

200

CLOSED HABITAT

OPEN HABITAT

S 30

cC^^"

\ \

^ 20
o

<4-

\ V

:io

o
+J

^■S^

■^^

\
\

c
u

, 1

-^ W=^

Meters

Figures 31. A comparison of the relative abundances of the six mammal groups chosen to represent open and
closed habitat faunas in the two localities that show the clearest separation of these two faunas, 105-1311 and
130-0201. Percentages were calculated on the basis of the cumulative totals of square frequencies of the six
animals in each locality. The reconstructions represent the general sedimentary contexts of each fossil assem-
blage.

fo.ssil deposits. A greater number of fossil
suids agrees well with the greater diversity
of this group in the Plio-Pleistocene, with
at least 5 species (Maglio, 1972) present
in the East Rudolf area.

At present East Rudolf supports a mam-
malian fauna of at least six bovid species,
two species of zebra and a single species
of suid, the warthog. Giraffe, hippo,
baboon, man and a variety of carnivores
and rodents are also present. Rhinoceros
has only recently become extinct in the
area, and elephants were recorded there
near the end of the last century (R. E.
Leakey, personal communication). Most

of the fossil mammalian groups are repre-
sented in the recent ecosystem, with the
exception of the deinotheres, now extinct.
However, in terms of species and genera.
East Rudolf today is much less diverse than
in the Plio-Pleistocene. The fossil faunas
are more similar in terms of numbers of
species represented (Maglio, 1972) to the
recent faunas of wetter areas such as
Nairobi National Park or the Kafue Flood-
plain in Zambia. Environmental change in
the East Rodolf region during at least the
past 1.5 my. has been great, and apparently
is continuing to affect the vertebrate com-
munity.

TOO

c
u

<? 40

EAST
RUDOLF

Plio-
Pleistocene

(All

localities)

East Rudolf Paleoecology • Behrensnieyer 561

EAST KAFUE

RUDOLF PARK,

Recent ZAMBIA

(Stewart, (Dowset, 1966)
1963 and
suid estimate
by A.K.B.)

NAIROBI

NATIONAL

PARK

1961

NAIROBI
NATIONAL

PARK

1966

(Foster, 1967)

QUEEN

ELIZABETH

NATIONAL

PARK

(Field and
Laws, 1970)

SUID

EQUID

BOVID

Figure 32. Connparisons of the relative percentages of bovids, suids and equids in the seven Pljo-Pleistocene
localities used for this study, and in Recent ecosystems. All relative numbers for the recent examples are based
on numbers of individuals (game counts). The abundances shown for East Rudolf (Plio-Pleistocene) were aver-
aged from the square frequencies for bovids, suids and equids in all of the seven localities, but are consistent
with relative abundances in each separate locality.

Hominid Paleoecology

The hominid fossils that occur in the
Koobi Foia Fm. should reflect the same
taphonomic processes as the remains of
other vertebrates. Therefore, they can be
examined in the context of the rest of the
fauna for possible paleoecological factors.
The abundance of hominids in the East
Rudolf collections is due to the intensive
program of hominid collecting and does
not reflect their relative abundance in the
total fauna. In reality, they are fairly rare,
as indicated by an occurrence in only 2
out of 213 sample squares in the seven
localities used for this study. This is com-
parable to the relative abundance of rodent
fossils in the squares. It would be necessary
to sample hundreds or thousands of squares
to provide enough hoiuinid material for
valid quantitative comparisons of fre-
quencies in different sedimentary environ-

ments. This is not feasible, but it will be
possible to relate the hominid fossils to the
different sedimentary environments and to
the closed or open habitat faunas using the
sedimentary and faunal context of each
specimen. Based on the patterns of occur-
rence of other mammals, it seems possible
that the two lineages of hominids recog-
nized from East Rudolf (Leakey, 1973) were
associated with the two different habitats.
The East Rudolf hominid collection as of
April, 1973, consisted of 50 specimens as-
signed to Australopithecus, 34 assigned to
Homo, and 10 unassigned (M. Leakey,
personal communication). Both taxa are
represented by a wide variety of skeletal
parts. The greater number of Australopi-
thecus specimens is due primarily to a
larger number of partial mandibles of this
form. This may reflect taphonomic pro-
cesses (e.g., carnivore activity, surface

562 Bulletin Museum of Comparative Zoology, Vol. 146, No. 10

weathering) that operated to selectively de-
stroy the less robust Homo mandibles.
There is no clear pattern as yet indicating
occurrences of the two forms in different
sedimentary environments.^ Both occur in
deltaic and floodplain deposits. Australopi-
thecus is possibly the only hominid oc-
curring in the 10.5-1311 channel, where it is
relatively common (at least 7 separate speci-
mens). This is interesting in that it cor-
relates with other faunal peculiarities of
105-1311, which has a high proportion of
open habitat mammals plus forms (deino-
theres, rhinos) that are not present in the
deltaic assemblages. However, since a
variety of habitats were sampled in 105-
1311, it would be premature to draw any
conclusions on the habitat preferences of
Australopithecus.

The sample assemblage from 105-0208
occurs in delta margin deposits several
meters below the KBS Tuff (Fig. 17),
which is the horizon bearing the oldest
known hominid "campsites" (Isaac et ah,
1972). The relative abundance of the dif-
ferent vertebrate groups ( Fig. 23 ) is prob-
ably broadly representative of the fauna
that was extant on the delta margin at the
time of the hominid habitation sites. At
this time, however, the delta margin had
changed in position, probably receding
farther to the southwest. The deposits
directly associated with the KBS Tuff
( which is primarily a channel fill ) are fine-
grained silty clays, which indicate extensive
interdistributary marshes and mudflats that
were probably seasonally dry. Such en-
vironments do not seem to be conducive
to fossil preservation on the recent Omo
Delta (Butzer, 1971:103), and fossils are
indeed rare in the silty clays associated
with the habitation sites. Thus, evidence
for the faunal context of the tool-manu-

* Additional research conducted in 1973 indi-
cates that the Homo lineage sample is much more
restricted to lake margin deposits that the Austra-
lopithecus sample, which is abundant in both lake
margin and fluvial deposits ( Behrensmeyer, In
press ) .

facturing hominids must come indirectly
from the older 105-020(S fauna. This pro-
vides at least regional, if not local, evi-
dence for the vertebrate fauna most closely
associated with the hominids.

East Rudolf in Relation to
Other Studies in
Vertebrate Paleoecology

Although vertebrate paleoecology has
long been an area of recognized research
value, there are relatively few compre-
hensive studies in print. Those that are
available provide useful comparisons for
paleoecologic interpretations of the Koobi
Fora Fm. and show how this study relates
to broader research on the evolution of
vertebrate communities and ecosystems
through time.

The major studies that have defined ter-
restrial paleo-communities include Olson
(1952, 1958), Shotwell (1955, 1963) and
Clark et al. (1967). Olson's interpretation
of the Permian Vale and Choza Fauna of
Texas shows a correlation between environ-
mental change (increasing aridity), as de-
termined from geologic evidence, and
significant changes in the vertebrate fauna.
A study of the Oligocene Chadron fauna of
South Dakota by Clark et al. (1967:69-73)
reveals two distinct ecological assemblages
of mammals, referred to as "savanna" and
"aquatic-wet forest" faunas. Environmental
changes to cooler, more arid conditions led
to restriction and finally to the elimination
of the wet-forest fauna during the time
span represented by the Chadron Fm.
Shotwell's studies of Pliocene faunas of the
Juntura Basin of eastern Oregon included
pioneer work in quantitative methods for
reconstructing ecological associations of
animals from quarry samples. Using these
methods, the fossil mammals of the Drewsey
Fm. are assigned to four paleo-communi-
ties: woodland, savanna, open grassland
and pond-bank. Change through time shows
reduction in the woodland and savanna
faunas with the development of the open

East Rudolf Paleoecology • Bchrcnsmcycr 563

grasslands fauna (Shotwcll, 1963:19). All
of these studies rely on geologic and faunal
evidence plus a variety of taphonomic
assumptions. The paleoecologic interpre-
tations, particularly those of Shotwell,
could be further supported or perhaps
altered by more detailed taphonomic
analysis.

Olson ( 1952 ) developed the concept of a
"chronofauna" to describe the nature of the
vertebrate fauna of the Texas Permian. A
chronofauna is defined as "a geographicall}-
restricted, natural assemblage of interacting
animal populations that has maintained its
basic structure over a geologically signifi-
cant period of time" (1952:181). Accord-
ing to Olson, the membens of a chronofauna
may change by such processes as expansion
into unoccupied niches or the substitution
of one species for another in any given
niche, but the \'ertebrate communit}' is in
adaptive equilibrium with the en\'ironment
and will maintain its stiiicture until en-
vironmental change occurs. Clark et al.
( 1967 ) refer to their aquatic-wet forest and
savanna faunas as chronofaunas, and Shot-
well's communities might also be con-
sidered as chronofaunas. The resolution of
evidence for the Cenozoic faunas leads to
more refined ecological interpretations and
to the consideration of shorter time spans
than is possible for the Permian.

In these three studies of vertebrate
paleo-communities, the chronofaunas show
change due to increasingly arid conditions.
This leads to the expansion of grassland
mammals in the Cenozoic faunas and a
decrease in the diversity of the Pennian
reptile and amphibian fauna. The absence
of evidence for significant morphological
change in the vertebrate species during
the time span represented by the chrono-
faunas has been noted by Olson (1952:193)
and Clark et al (1967:73). During periods
of environmental change, extinction, mi-
gration and niche modification apparently
were more common patterns of faimal re-
sponse than rapid morphologic modifica-
tion.

The fauna of the Koobi Fora Fm. records
interaction between vertebrate communi-
ties and en\'ir()nmental conditions between
about 3.0 and 1.5 my. B.P. A comparison
between the Plio-Plcistocene fauna and the
Recent one shows a restriction of species
di\'ersity which is more comparable to the
tenninal stages of the Permian chrono-
fauna than to the shifts in the Cenozoic
paleo-communities of South Dakota or
Oregon. The general significance of the
faunal change at East Rudolf may reflect
either the short- or long-term effects of
increasing aridity, but it is clearly an ex-
ample of how en\'ironmental change may
affect vertebrate communities.

The evidence now available suggests at
least one significant difference between the
evolution of East African faunas and the
evolution of the Texas Permian, the Juntura
Basin Mio-Pliocene and the Chadron Oligo-
cene faunas. Some of the East African
species, particularly the elephants and
suids, show rapid evolutionary change dur-
ing their Plio-Pleistocene histories (Cooke
and Maglio, 1972). Such change has not
been recognized in the other faunas, and
it is unlikely that this contrast is an
artifact of taxonomy or of inadequate
sampling. Documentation and comparison
of the histories of vertebrate faunas at dif-
ferent time levels and in different regions
should do much more to clarify the patterns
of evolutionary response to environmental
change.

Further reconstruction of ^XTtebrate com-
munities in the Lake Rudolf Basin will
provide a more detailed historv' of faunal
change over the past 5-6 my. The open
(grassland) and closed (bush) habitat
faunas of Ea.st Rudolf probabh' represent
distinct ecological communities that can be
documcMitc'd elsewhere in East Africa and
followed through time. The Omo secjuence
offers an ideal opportunity for comparing
vertebrate faunas of a major river-flood-
plain complex with the lake margin faunas
of East RudoU. Using tlu> combination of
geologic, taphonomic and faunal evidence,

564 Bulletin Museum of Comparative Zoology, Vol. 146, No. 10

it will be possible to compare fossil verte-
brate communities throughout East Africa
and to reconstruct changes in chronofaunas
through much of the latter part of the
Cenozoic.

SUMMARY

This study has developed methods for
deriving paleoecologic information from
fossil assemblages of fragmented vertebrate
bones subjected to various geologic pro-
cesses before burial. These methods have
been applied to paleoecologic interpreta-
tion of the Plio-Pleistocene bone deposits
of East Rudolf, Kenya. The conclusions
relate to vertebrate assemblages in general
as well as to the assemblages of East Rudolf
and the Lake Rudolf Basin in particular.

General Conclusions

Taphonomy

1) The amount of fragmented bone
buried in any given sedimentary environ-
ment will depend on the rate of sedimenta-
tion and the amount of bone originally put
into that environment. The important fac-
tors which control bone input are: a) verte-
brate abundance, b) carnivore activity, c)
the proximity of bones to depositional en-
vironments, d ) the rates of surface weather-
ing of bones, and e) the dispersal potential
of bones. The composition of the resulting
fossil assemblage will also in part depend
on diagenetic factors.

2) Carnivore activity will have a major
effect on the composition of a thanato-
coenose. Intense mammalian carnivore ac-
tivity results in fewer bones of small
animals and increased fragmentation of
bones of large animals. The evolution of
bone-crushing dentitions in mammals has
changed the character of Cenozoic tapho-
coenoses compared with those of the
Mesozoic, when reptilian carnivores lacked
the capacity for bone mastication.

3) Bones are disarticulated and acquire
characters of surface weathering in months

to years if exposed on a land surface.
Hydrodynamic transport will tend to leave
features of rounding and abrasion on bones.
Therefore, well-preserved bones with frag-
ile parts intact and surfaces unflaked or
uncracked record conditions of rapid burial
without subsequent re-excavation.

4) Bones vary greatly in density, size
and shape and are sensitive to hydro-
dynamic sorting. Disarticulated thanato-
coenoses include bones with a wide range
of dispersal potentials. This will result in
the formation of dispersal groups if the
bones are subjected to normal or flood-
stage current velocities ( 10-150+ cm/sec. ) .
The dispersal groups will move at different
rates from the point of origin. If bones with
a wide range of dispersal potentials are
found in sedimentary^ association, this indi-
cates that the assemblage is not a product
of selective transport sorting of the original
thanatocoenose.

5) Mammal bones immersed in water
for 5 minutes have densities from < 1.0 to
2.0, and teeth have densities between 1.7
and 2.3. Reptile and fish bones are be-
tween 1.3 to 2.3 density. Bones are gen-
erally hydraulically equivalent to quartz
particles of smaller nominal diameter. Cur-
rents should transport bones together with
quartz particles that are roughly equivalent
hydraulically. Therefore, sedimentary as-
sociations of quartz grains and bones of a
much larger hydraulic equivalence (e.g., a
hippopotamus skull in a siltstone) may indi-
cate other modes of origin for the bone-
sediment association. These include in situ
death, flotation of carcasses, or predator/
scavenger transport of bone.

6) Theoretical considerations indicate
that velocities of 80 to 200+ cm/sec. must
be achieved near the bottom of a flow in
order to move bones of moderate density
(~ 1.5) and size (100+ cc). Therefore, most
disarticulated, water-logged parts of large
vertebrates are unlikely to move far from
their point of origin except in special trans-
port situations such as floods in channels.

East Rudolf Paleoecology • Behrensmcycr 565

Paleoecology

i) Ecological characteristics ol fos-
sil vertebrates can be dc^fined using a
combination of geologic and taphononiic
evidence, independent of ecological inter-
pretations based on \'ertebrate morphology
or the adaptation of li\'ing analogues. Such
evidence can link habitat preferences with
preservation in particular sedimentary en-
vironments. This correlation can be inferred
solely from the geologic context and the
taphonomy of a gi\'en bone assemblage.
Such e\idence can then be combined with
morphological and recent-counterpart data
to support paleoecologic interpretations.

2 ) Fragmented bone assemblages can be
used with confidence for paleoecologic
interpretations if they: a) consist of bones
with a wide range of dispersal potentials,
b) are not hydraulically equivalent to
associated sediment and c) retain fresh,
unweathered or unabraded surfaces. As-
semblages with these attributes can be
interpreted as generally autochthonous to
their environment of deposition. Most of
the animals represented in such an
assemblage were preserved in the general
context of their original habitats.

3) Aquatic environments of deposition
can preserve variable amounts of bone from
aquatic and terrestrial vertebrates, de-
pending on the relative bone input from
each ecological group. Bone assemblages
of terrestrial and aquatic animals in aquatic
deposits (e.g., channel, delta margin) may
differ only in the better preservation of the
latter, not in their greater abundance.

4) Terrestrial environments of deposition
(e.g., floodplains) preserve a high pro-
portion of terrestrial vertebrates along with
a few aquatic ones. Semiaquatic vertebrates
tend to occur in both terrestrial and acjuatic
deposits, with better representation in
aquatic environments.

5) The bone input from groups of large
terrestrial vertebrates into fragmented,
autochthonous taphocoenoses should gen-
erally reflect their relative numbers in the
original ecosystem. The fossil abundances

can be used to approximate relative num-
bers of different \ertebrate groups in a
gixen en\'ironment. This provides a basis
for reconstructing paleo-communities and
comparing them through time.

6) Vertebrate communities at different
time horizons or in different regions may
differ in their response to broad-scale
environmental change in ways that can be
detected in paleoecologic studies. These
responses include rapid morphological evo-
lution, shifts in the relative numbers of
animals suited to particular habitats, and
a general decline in species diversity ac-
companied by the extinction of forms in all
the available habitats.

Conclusions for the Vertebrate
Assemblages of tfie Koobi Fora
Formation, East Rudolf

1) Fossil-bearing deposits reveal sedi-
mentation and bone preservation in at least
three major depositional environmc^nts:
delta margin, channel and floodplain.

2) The three depositional environments
show a basic similarity in their representa-
tion of different skeletal parts, with teeth
the most abundant component. However,
the relative numbers of certain skeletal
elements differ in ways that reflect the dif-
ferent processes operating in the three
environments. Teeth are relatively more
abundant in the channel deposits and in
the floodplain, while vertebrae and
phalanges are more abundant in the delta
margin deposits. This can be related to
the concentration of heavy, durable parts
in the channels through sorting and re-
working of bones, and to the absence of
such processes in the delta margin environ-
ments. The taphononiic characters of the
floodplain assemblage indicate preferential
removal of the lighter elements without
transport or reworking of thc> associated
heavier l:)()nes.

3) The sum of taphonomic and geologic
evidence shows that the delta margin and
floodplain bone assemblages are autoch-
thonous with respect to the overall sedi-

566 Bulletin Museum of Comparative Zoologtj, Vol. 146, No. 10

mentary environment. Channels contain a
mixture of allochthonous and autochtho-
nous bones and show the most evidence for
taphonomic alteration of the original than-
atocoenose.

4) The East Rudolf faunas include
aquatic, semiaquatic and terrestrial verte-
brates that vary in abundance according to
sedimentary environment. Analogies with
recent East African ecosystems indicate
that the relative fossil abundance of ter-
restrial mammalian families probably re-
flects their abundance in the original
ecosystem. Bovids, suids and equids are
the most common groups in the fossil as-
semblages and in most recent undisturbed
East African faunas.

5) Two terrestrial faunas can be defined
for the East Rudolf assemblages, based on
ecological analogies between recent and fos-
sil mammals. The open habitat fauna in-
cludes alcelaphines, Metridiochoerus/Noto-
choerus suids and Eqiius. The closed
habitat fauna is characterized by Meso-
choerus, reduncines and tragelaphines.
There is overlap of these faunas in all of
the sample assemblages. However, delta
margin deposits generally preserve a greater
proportion of closed habitat forms, and the
channels preserve more open habitat forms.

7) The paleoecologic results for East
Rudolf show that it is possible to define
ecological groups of terrestrial vertebrates
from surface samples of fragmental bone
assemblages. Similar sampling of fossil
assemblages at different time horizons can
provide a basis for establishing East African
chronofaunas and for reconstructing their
interaction with environmental changes
through time,

ACKNOWLEDGMENTS

The interdisciplinary nature of this work
led to a great deal of productive and en-
joyable interaction with researchers in a
broad range of disciplines, including Geol-
ogy, Paleontology, Zoology, Anthropology,
and Ecology. An interdisciplinary approach

inevitably generates a large number of
well-deserved acknowledgments. I attempt
here to express my gratitude to those people
who have made particular contributions to
this study, but I wish to preface this with
a simple and very sincere note of thanks to
all who provided help and encouragement
during the course of the project.

I would like to give special thanks to
Bryan Patterson, Richard E. Leakey, Glynn
L. Isaac and Vincent J. MagHo. Without
their unfailing encouragement, help and
ideas, this work would not have been pos-
sible. Gratitude is due to Parish A.
Jenkins, Jr., Raymond Siever, Robert T.
Bakker, Daniel C. Fisher, Peter Dodson
and H. B. S. Cooke for many useful com-
ments and suggestions during the prepa-
ration of the manuscript. Discussions with
Andrew Hill, Richard D. Estes, Stephen J.
Gould, Jack Sepkoski, F. B. Van Houten,
G. Jepsen, John Fleagle, Allen Greer, T.
Hopson and Stanley Awramik resulted in
many additional ideas and references. Dirk
van Damme (Geologish Institute, Ghent,
Belgium) also deserves thanks for identifi-
cation of the invertebrate fossils. I have
greatly appreciated the exchange of geo-
logical information with co-workers on the
East Rudolf Expedition, including Bruce
Bowen and Carl Vondra of Iowa State Uni-
versity, Gary Johnson of Dartmouth College
and Ian Findlater of Birkbeck College,
University of London. Field work ( consist-
ing of long hours of sample collecting in
10 X 10 meter squares) was accompHshed
through the assistance and stoicism of
Susan Abell, Penny Bowen, John Bart-
helme, John W. Harris, John M. Harris,
John Onyango-Abuje, Jonathan Karoma,
John Kimingitch, Fred Lucas, Dinah
Grader, Diane Gifford, Dan Stiles, Paul
Abell, Kelly Stewart and Andrew Hill. John
Barthelme and John Harris deserve special
thanks for their help in laboratory analysis
and identification of the fossil material.
Barbara Lawrence and Charles Mack of the
Department of Mammalogy, Museum of
Comparative Zoology ( Harvard ) were very

East Rudolf Paleoecology • Bchrvnsmexjer 567

h(>lptul in providing recent skeletal material
used for this stud>-. Photographic reproduc-
tion of the figures and plates was done by
John Lupo (Biological Laboratories, Har-
\'ard) and additional assistance was pro-
vided by Al Coleman (MCZ Laboratories,
Harvard). Typing was the joint effort of
Karen Mason, Maureen Sepkoski and Agnes
Martin. Their work is gratefully acknowl-
edged with special thanks to Agnes Martin
for taping the final draft. Finally, among
the many friends who have provided mis-
cellaneous assistance in time of need, I
would particularly like to thank Vickie
Rowntree, Catherine Badgley, A. Gordon
Brown and Elizabeth Whitehouse.

This i;tudy was done as a Ph.D. Disser-
tation in the Department of Geological
Sciences, Harvard University, and was
completed in June of 1973. The work re-
ceived financial support through grants by
the National Science Foundation (Grant
No. 28607a) and the National Geographic
Society to the East Rudolf Research Project.

REFERENCES

AuEL, O. 1912. Grinidzuge der Palaeobiologie
der Wirbeltiere. Stuttgart- E. Schweizer-
bart'sche Verlagsbuchhandliing Niigele und
Dr. Sproesser. 470 pp.

Allen, J. R. L. 1963. The classification of
cross-stratified units, with notes on their
origin. Sedimentol., 2: 93-114.

. 1965. A review of the origin and

characteristics of recent alluvial sediments.
Sedimentol., 5: 89-191.

. 1970. Physical Processes of Sedimenta-

tion. London: George Allen and Unwin, Ltd.
248 pp.

Allen, P. 1959. The Wealden environment:
Anglo-Paris Basin. Roy. Soc. London, Phil
Trans. (B), 242(692): 283-346.

Ansell, W. H. F. 1971. Part 15: Order Artio-
dactyla. In: Meester, J. and 11. \V. Setzer,
(eds.). The Mammals of Africa. An Identifi-
cation Manual. Washington, D.C: Smith-
sonian Institution Press.

Aristarain, L. F. 1962. Caliche deposits of
New Mexico. Ph.D. Dissertation. Harvard
Universit>-, Department of Geological Sci-
ences.

Baker, B. H., and J. Wohlenberg. 1971. Struc-

ture and evolution of the Kenya Rift Valley.
Nature, 229: .538-542.

Beadle, 1^. C. 1932. The waters of some East
.\frican lakes in relation to their fauna anci
flora. J. Linn. Soc. (Zool.), 38: 135-136.

BEimKNSMEYEH, A. K. 1974. Late Cenozoic
sedimentation in the Lake Rudolf Basin,
Kenya. Annals Geol. Soc. Egypt, IV: 287-306.

. In press. The habitat of Plio-Pleistocene

hominids in East Africa; taphonomic and mi-
cro.stratigraphic evidence. In C. Joly (ed. ),
African Hominidae of the Plio-Pleistocene:
Evidence, Prolilems and Strategies. New
York: Duckworth, Inc.

Berggren, W. a., and J. Van Couvering. 1973.
Late Neogene chronostratigraphy, biostratig-
raphy, biochronology and paleoclimatology.
Woods Hole Oceonogr. Inst. Tech. Rep.,
WHOI-73-40. 334 pp.

Berry, L. G., and B. Mason. 1959. Mineralogy.
San Francisco: W. H. Freeman and Co. 630
pp.

Bigalke, F. C. 1972. The contemporary mam-
malian fauna of Africa. In Keast, A., B. Glass,
F. C. Erk, (eds.). Evolution, Mammals and
Southern Continents. Albany, New York:
State University of New York Press. 141-194.

Bishop, W. W. 1968. The evolution of fossil
environments in East Africa. Trans. Leicester
Lit. and Phil Soc, 62: 22-44.

BowEN, B. E., and C. F. Vondra. 1973. Strati-
graphical relationships of the Plio-Pleistocene
deposits. East Rudolf, Kenya. Nature, 242:
391-393.

Brain, C. K. 1967a. Bone weatliering and the
problem of bone pseudo-tools. S. Afr. J. Sci.,
63(3): 97-99.

. 1967b. Hottentot food remains and

their bearing on the interpretation of fossil
bone assemblages. Sci. Pap. Namib Des. Res.
Sta., No. 32.

Bretz, J. H., AND L. HoRBERG. 1949. Caliche
in southeastern New Mexico., J. Geol. 57:
491-511.

BrIGGS, L. I., D. S. McCULLOCH, AND F. MOSER.

1962. The hydraulic shape of sand particles.
J. Sediment. Pet., 32(4): 645-656.

Brock, A., and G. Isa.a.c. 1974. Paleomagnetic
stratigraphy and chronology of hominid bear-
ing sediments east of Lake Rudolf, Ken\a.
Nature, 247: 344-348.

BuTZER, K. W. 1971a. Recent History of an
Ethiopian Delta. Res. Pap. No. 136, Dept.
Geogr. Chicago: The Uni\ersit\- of Chicago
Press. 184 pp.

. 1971b. The Lower Onio Basin- geol-
ogy, fauna and hominids of Plio-Pleistocene
formations. Naturwiss., 58: 7—16.

BuTZER, K. W., G. L. Isaac, J. L. Ric:iiardson,

568 Bulletin Museum of Comparative Zoology, Vol. 146, No. 10

AXD C. Washboukn-Kamau. 1972. Radio-
carbon dating of East African lake levels.
Science, 175(4027): 1069-1076.

Clark, J., J. R. Beerbow^r, and K. K. Kietzke.
1967. Oligocene sedimentation, stratigraphy
and paleoecology and paleoclimatology in the
Big Badlands of South Dakota. Fieldiana,
Geol., 5. 158 pp.

Cooke, H. B. S., and V. J. Maglio, 1972. Plio-
Pleistocene stratigraphy in East Africa in
relation to proboscidean and suid evolution.
In Bishop, W., and J. Miller, (eds.). Cali-
bration of Hominoid Evolution. New York:
Scottish Academic Press. 303-329.

CoTT, H. B. 1961. Scientific results of an in-
quiry into the ecology and economic status of
the Nile Crocodile (Crocodylus niloticiis) in
Uganda and Northern Rhodesia. Trans. Zool.
Soc. Lond., 29(4): 211-356.

CuRKEY, J. 1970. Animal Skeletons. London:
Edward Arnold, Publishers.

Davies, D. K., F. G. Ethridge, and R. R. Berg.
1971. Recognition of barrier environments.
Bull. Am. Assoc. Pet. Geol., 55(4): 550-565.

Dodson, P. 1971. Sedimentology and taphonomy
of the Oldnian Formation (Campanian),
Dinosaur Provincial Park, Alberta (Canada).
Paleogeogr., Paleoclimatol., Paleoecol. 10:
21-74.

. 1974. The significance of small bones

in paleoecological interpretation. Contrib. to
Geol., Univ. of Wyoming. Spec. Pap. No. 2
Laramie, Wyoming: Univ. Wyoming Press.

DoRST, J., AND P. Dandelot. 1970. Larger
Mammals of Africa. London: Collins Press.
287 pp.

DowsETT, R. J. 1966. Wet season game popu-
lations and biomass in the Ngoma area of
the Kafue National Park. The Puku (Occas.
Pap. Dept. Game and Fish, Zambia), No. 4:
135-145.

Efremov, J. A. 1940. Taphonomy: A new
branch of Paleontology. Panam. Geol., 74:
81-93.

. 1953. Taphonomie et annales geologi-

ques. Ann. du Cent. d'Etud. et de Doc.
Paleontol., No. 4. 164 pp.

Estes, R. D. 1967. Predators and scavengers.
Nat. Hist., 75(2, 3): 20-29, 38-47.

. (1973), in press. Social organization of

the African Bovidae. I.U.C.N., Symposium
on Ungulate Behavior and Management.

Field, C. R., and R. M. Laws. 1970. The dis-
tribution of the larger herbivores in the Queen
Elizabith National Park, Uganda. J. Appl.
Ecol., 7(2): 273-294.

Fitch, F. J., and J. A. Miller. 1970. Radio-
isotopic age determinations of Lake Rudolf
artefact site. Nature, 226(5242): 226-228.

Foster, J. B. 1967. Nairobi National Park
Game Census, 1966. East Afr. Wildl. J., 5:
112-120.

Guggisberg, C. a. W. 1972. Crocodiles. New-
ton Abbot, England: David & Charles, Pub-
lishers. 195 pp.

Harris, L. D. 1970. Some structural and
functional attributes of a semi-arid East
African ecosystem. Ph.D. Dissertation, Michi-
gan State University.

Harvard Tables of the Cutmulative Binomial
Probability Distribution. 1955. Cam-
bridge, Massachusetts: Harvard University
Press. 503 pp.

Haynes, V. 1968. Radiocarbon: analysis of in-
organic carbon of fossil bone and enamel.
Science, 161: 687-688.

de Heinzelin, J., F. H. Brown and F. C.
Howell. 1971. Pliocene/Pleistocene for-
mations in the Lower Omo Basin, southern
Ethiopia. Quaternaria, 13: 247-268.

Isaac, G. L. 1967. To\\'ards the interpretation
of occupation debris: some experiments and
observations. The Kroeber Anthropol. Soc.
Pap. 37: 31-57.

Isaac, G. L., R. E. F. Leakey and A. K. Behrens-
MEYER. 1971. Archeological traces of early
hominid activities, east of Lake Rudolf, Kenya,
Science, 173: 1129-1134.

Johnson, L. C. 1965. Morphological analysis in
Pathology. In Frost, H. M. (ed. ). Bone
Biodynamics. (Henry Ford Hospital Int.
Symp. ). Boston: Little, Brown and Co.
543-654.

KOxNiZESKi, R. L. 1957. Paleoecology of the
middle Pliocene Deer Lodge Local Fauna,
western Montana. Bull. Geol. Soc. Am. 68:
131-150.

Kruuk, Hans. 1972. The Spotted Hyaena.
Chicago: The University of Chicago Press.
335 pp.

KuRTEN, B. 1953. On the variation and popu-
lation dynamics of fossil and recent mammal
populations. Acta Zool. Fenn., 76: 5-118.

Lamprey, H. F. 1963. Ecological separation of
the large mammal species in the Tarangire
Game Reserve, Tanganivika. East Afr. Wildl.
J., 1: 63-92.

Langbein, W. B., and L. B. Leopold, 1968.
River channel bars and dunes — theory of
kinematic waves. U. S. Geol. Surv. Prof. Pap.
No. 422-L: 1-20.

Leakey, R. E. F. 1973. Fiuther evidence of
Lower Pleistocene hominids from East Rudolf,
North Kenya, 1972. Nature, 242: 170-173.

Leopold, L. B., M. G. Wolman and J. P. Miller.
1964. Fluvial Processes in Geomorphology.
San Francisco: W. H. Freeman and Co.
522 pp.

East Rudolf Paleoecology • Belirensmeijer 569

Leopold, L. B., W. VV. E^t^IETT and R. M.
Myrick. 1966. Channel and hillslope pro-
cesses in a semiaiid area. New Mexico. V. S.
Geol. Surv. Prof. Pap. No. 352-G: 193-253.

LOBOVA, E. V. 1967. Soils of the Desert Zone
of the USSR Jenisak'iii: Israel Program for
Scientific Translations. 405 pp.

LovERiDGE, A. 1941. Revision of the African
terrapin of the family Pelomedusidae. Bull.
Mus. Comp. Zool., 88(6): 462-524.

LovERiDGE, A., AND E. E. WiLLiAMs. 1957. Re-
vision of the African tortoises and turtles of
the suborder Cryptodira. Bull. Mus. Comp.
Zool, 115(6): 163-557.

Maglio, V. J. 1972. Vertebrate faunas and
chronology of hominid-bearing sediments east
of Lake Rudolf, Kenya. Nature, 239: 379-
385.

McKee, E. D., E. J- Crosby and H. L. Berryhill,
Jr. 1967. Flood deposits. Bijou Creek,
Colorado, June, 1965. 1. Sediment. Pet., 37
(3): 829-851.

Millar, C. E., L. M. Turk and H. D. Foth.
1966. Fundamentals of Soil Science. New
York: John Wiley & Sons. 491 pp.

Muller, a. H. 1957. Lehrbuch der Palao-
zoologie. Vol. 1: Allgemeine Grundlagen.
Jena: Gustav Fischer Verlag. 322 pp.

Olson, E. C. 1952. The evolution of a Permian
vertebrate chrono-fauna. Evolution, 6(2):
181-196.

. 1958. Fauna of the Vale and Chosa:

14. Fieldiana, Geol., 10(32): 397-448.

Patterson, B., A. K. Behrensme\'er, and W. D.
Sill. 1970. Geology and fauna of a new
Pliocene localitv in north-western Kenya. Na-
ture, 212(5062): 918-921.

Payne, J. E. 1965. Summer carrion study of
the baby pig, Siis scrofa. Ecology, 46(5):
592-602.

Pettijohn, F. J. 1957. Sedimentary Rocks.
New York: Harper & Row, Publishers. 718
pp.

Pettijohn, F. J., P. E. Potter and R. Siever.
1972. Sand and Sandstone. New York:
Springer-Verlag. 618 pp.

Reeves, C. C. 1970. Origin, classification and
geologic history of caliche on the southern
high plains, Texas and eastern New Mexico.
J. Geol., 78: 352-362.

Reif, Wolf-Ernst. 1971. Zur Genese des Mus-
chelkalk-Keuper-Grenzbonebeds in Siidwest-
deutschland. Neues lahrb. Geol. Palaontol.
Abh., 139(3)- 369-404.

Rittenhouse, G. 1943. Transportation and de-
position of heavy minerals. Bull. Geol. Soc.
Am., 54(12): 1725-1780.
Sadek-Kooros, H. 1966. Jaguar Cave: an early

man site in the Beaverhead Mountains of
Idaho. Ph.D. Dis.sertation. Harvard Uni-
\ersit\-. Department of Anthropology.

Sanders, H. L. 1968. Marine benthic diversity:
a comparative study. Am. Nat., 102(925):
243-282.

Sciiafer, W. 1972. Ecologj' and Palaeoecology
of Marine Environments. Craig, G. (ed.), I.
Oertel (transl.). Chicago: The University of
Chicago Press. 568 pp.

ScHALLER, G. B. 1972. The Serengeti Lion.
Chicago: The University of Chicago Press.
480 pp.

Selous, F. C. 1908. African Nature Notes and
Reminiscences. London: MacMillan and Co.
356 pp.

Shapiro, A. II. 1961. Shape and Flow: The
Fluid Dynamics of Drag. New York: Double-
day & Co., Inc., Anchor Books. 186 pp.

Sheppe, W., and T. Osborne. 1971. Patterns
of use of a floodplain by Zambian mammals.
Ecol. Monogi-., 41(3): 181-205.

Shotwell, J. A. 1955. An approach to the
paleoecology of mammals. Ecology, 36(2):
327-337.

. 1963. The Juntura Basin: studies in

earth history and paleoecology. Trans. Am.
Phil. Soc, (N. S.), 53(1): 3-77.

Simpson, G. G., A. Roe, and R. C. Lewontin.
1960. Quantitative Zoology. New York:
Harcourt, Brace & World, Inc. 440 pp.

Stewart, D. R. M. 1963. Wildlife census. Lake
Rudolf. East Afr. Wildl. J., 1: 121.

Van Lawick-Goodall, H. and J. 1971. Inno-
cent Killers. Boston: Houghton Mifflin Co.
222 pp.

Visher, G. S. 1965. Use of vertical profile in
environmental reconstruction. Bull. Am. Assoc.
Pet. Geol., 49(1): 41-61.

VooRHiES, M. 1969. Taphonomy and population
dynamics of an early Pliocene \'ertebrate
fauna, Knox County, Nebraska. Contrib. to
Geol. Univ. of Wyoming. Spec. Pap. No. 2.
Laramie, Wyoming: Uni\'. Wyoming Press.
69 pp.

Walsh, J. and R. G. Dodson. 1969. Geology of
Northern Turkana. Rep. No. 82, Geol. Surv.
Kenya. 42 pp.

Weigelt, J. 1927. Rezente Wirbeltierleichen
und ihre paliiobiologische Bedcutung. Leip-
zig: Verlag \on Max Weg. 208 pp.

White, T. 1955. Obser\'ations on the butcher-
ing techniques of some aboriginal peoples,
numbers 7, 8, and 9. Am. Antiq., 21(2):
170-178.
Williams, G. E. 1971. Flood deposits of the
sand-bed ephemeral streams of Central Aus-
tralia. Sedimentology, 17: 1—40.

570 Bulletin Museum of Comparative Zoology, Vol. 146, No. 10

APPENDIX 1

Measurkments of densities, volumes and wet weights of modern bones. Densities are calcu-
lated FOR THE BONES AFTER THEIR PORE SPACES WERE FILLED WITH WATER.

DENSITIES

Skeletal part

Ovis

( sheep )

MCZ

1939

Redunca

( reedbuck )

MCZ

14917

Hylochoenis

( forest hog)

MCZ

27851

Damaliscus
(topi)
MCZ
15724

Equus

( zebra )

MCZ

5003

Hippo-
potamus
MCZ
5020

HUMERUS

1.53

1.40

1.55

1.51

1.77

1.74

RADIUS 1
ULNA j

1.64

1.58
1.16

1.66

1.72
1.41

1.45

1.69

FEMUR

1.45

1.36

1.41

1.37

1.36

1.79

TIBIA

1.46

1.23

1.54

1.71

1.45

1.55

METATARSAL

1.44

1.33

1.07

1.62

1.68

1.50

METACARPAL

1.35

1.45

1.15

1.51

1.52

1.34

ASTRAGALUS

1.68

1.81

1.28

1.66

1.16

1.46

CALCANEUM

1.62

1.37

1.22

1.50

1.00

1.52

PODIAL #1

1.43

1.13

1.30

1.46

1.23

1.31

—

1.25

1.20

1.48

1.01

1.46

PHALANX #1

1.45

1.34

—

1.40

1.00

1.37

1.60

—

1.36''»

1.34

1.02

1.29">

1.06

—

1.02" >

1.01'"

1.05<"

1.07' '>

TEETH M

2.19

—

2.23

2.08

2.00

—

2.24

1.97

—

1.53

—

1.83

—

1.53

1.88

1.74

RIB #1

1.11

1.54

1.41

1.36

1.22

1.63

—

1.43

1.20

1.08

1.84

VERTEBRA ATLAS

1.24

.78

1.56

1.43

1.28

1.64

AXIS

1.07

.94

1.41

1.33

1.24

1.87

CERVICAL

1.04

1.13

—

1.11

.98

1.82

THORACIC

1.06

1.05

1.21

1.30

1.11

1.26

LUMBAR

.89

1.23

1.13

.99

1.36

SACRUM

1.11

.92

1.18

1.07

PATELLA

1.07

1.01

1.30

.64

1.24

PELVIS

1.19

1.17

—

STERNUM

.97

—

SKULL

1.42

1.39

JAW (1/2)

1.43

1.74

1.58

SCAPULA

1.65

1.88

1.53

VERT. CENT. #1

.98

.75

1.60

1.00

1.06

1.29

1.09

—

1.40

1.00

ULNA, PROX.

—

.90

1.21

SESAMOID #1

—

1.46

—

— .

1.18

HUM. PROX.

1.26

1.34

1.42

1..32

1.63

1.55

DIST.

1.75

1.96

1.69

1.81

1.83

1.96

R/U. PROX.

1.64

1.47'"^

1.65

1.96"^'

1.29

1.74

DIST.

1.59

1.72'«'

1.67

1.52""

1.50

1.63

FEM. PROX.

1.47

1.44

1.50

1.58

1.33

1.83

DIST.

1.42

1.30

1.29

1.21

1.45

1.64

TIB. PROX.

1.32

1.20

1.27

1.43

1.27

1.30

DIST.

1.64

1.28

1.96

2.30

1.77

1.91

MT. PROX.

1.31

1.48

1.49

DIST.

1.56

—

1.55

1.36

MC. PROX.

1.38

1.33

1.40

DIST.

1.25

—

1.37

1.40

SCAPULA (GLENOID)

1.30

1.48

1.58

1.29

1.32

R = Radius

only.

t = Terminal Phalanx

East Rudolf Paleoecology • Behrensmeyer 571

VOLUMES ( Cubic Centimeters )

Skeletal part

Ovis

( sheep )

MCZ

1939

Rediinca

( recdbvick )

MCZ

14917

Hijlochoenis

( forest hot; )

MCZ

27851

Damaliacus

(topi)

Mc;z

15724

Equits

(zebra)

MCZ

5003

Hippo-
potamus
MCZ
5020

HUMERUS

53.5

67.0

404

225

310

2542

RADIUS )
ULNA j

39.6

39.0
11.0

232

148
39.0

303

1700

FEMUR

65.0

116

383

296

635

3000

TIBIA

56.0

128

186

246

411

1852

METATARSAL

21.0

46.0

23.0

117

140

144

METACARPAL

20.5

40.0

25.5

116

176

174

ASTRAGALUS

4.1

7.2

27.5

20.4

296

CALCANEUM

5.5

14.1

44.5

36.0

352

PODIAL #1

2.8

6.4

12.2

16.2

10.4

150

—

2.0

8.8

5.6

14.8

PHALANX #1

2.9

4.5

12.0

17.9

1.0

4.8

7.4'"

8.1

—

17.8'"

1.7

8.5'"

7.9'"

20.0'"

10.7'"

TEETH M

1.7

—

3.6

25.4

73.0

—

17.6

42.2

—

290

—

0.8

—

130

RIB #1

10.0

5.2

55.0

14.0

26.5

229

—

7.7

31.8

33.0

25.0

—

VERTEBRA ATLAS

25.0

24.4

75.0

63.0

139

866

AXIS

30.5

19.8

56.0

67.0

155

500

CERVICAL

24.0

18.4

—

62.0

170

450

THORACIC

14.0

8.4

57.0

22.8

64.0

480

LUMBAR

15.6

16.6

45.0

40.0

49.0

480

SACRUM

30.0

33.0

165

125

—

PATELLA

2.9

5.2

28.0

20.0

45.0

244

PELVIS

107

64.0

—

STERNUM

3.2

—

SKULL

209

124

—

JAW (1/2)

39.5

23.0

—

119

—

SCAPULA

26.0

20.5

—

110

—

VERT. CENT. #1

5.1

6.0

25.0

20.0

31.0

233

4.7

25.0

30.0

28.0

—

ULNA, PROX.

7.8

20.0

SESAMOID #1

—

7.0

—

10.6

HUM. PROX.

32.5

35.0

220

129

168

1420

DIST.

23.4

24.0

184

94.0

150

1122

R/U. PROX.

19.5

21.0"^^

117

65.0""

170

822

DIST.

20.1

18.0""'

115

84.0""

147

878

FEM. PROX.

32.0

55.0

179

128

325

1466

DIST.

33.0

61.0

204

168

299

1634

TIB. PROX.

31.0

66.0

113

154

235

1100

DIST.

25.0

62.0

73.0

96.0

168

750

MT. PROX.

11.5

—

61.0

68.0

—

DIST.

9.7

58.0

68.0

MC. PROX.

10.0

—

60.0

82.0

DIST.

12.0

58.0

90.0

SCAPULA (GLENOID)

16.6

13.0

97.0

65.0

100

572 Bulletin Museum of Comparative Zoology, Vol. 146, No. 10

WET WEIGHTS (Grams)

Skeletal part

Ovis

( sheep )

MCZ

1939

Redunca

( reedbuck )

MCZ

14917

Hylochoenis

( forest hog )

MCZ

27851

Damaliscus
( topi )
MCZ
15724

Equus

( zebra )

MCZ

5003

Hippo-
potamus
MCZ
5020

HUMERUS

82.1

93.5

626

340

550

4413

RADIUS )
ULNA {

64.9

61.6
12.8

385

255
55

440

2869

FEMUR

94.5

158

539

407

866

5364

TIBIA

82.1

158

286

420

597

2873

METATARSAL

30.3

70.8

24.7

190

244

216

METACARPAL

27.7

62.3

29.3

171

176

234

ASTRAGALUS

6.9

13.2

35.1

33.9

432

CALCANEUM

8.9

19.9

54.3

54.0

536

PODIAL #1

4.0

7.2

15.8

23.7

12.8

196

—

2.5

10.6

8.3

15.0

137

PHALANX #1

4.2

6.3

—

21.9

107

1.6

7.1

12.2'"

11.4

23" >

1.8

—

10.2"'

7.5">

21">

11.5'"

TEETH M

4.4

—

6.7

68.7

146

—

11.4

*'35.0*

—

64.5

—

530

—

1.5

226

RIB#1

11.1

8.0

77.8

18.7

34.2

374

—

11.0

38.2

35.8

46.0

VERTEBRA ATLAS

30.9

19.0

117

90.0

178

1418

AXIS

32.6

18.7

78.7

89.0

193

934

CERVICAL

24.9

20.7

69.0

166

818

THORACIC

14.8

8.8

69.2

29.7

71.3

606

LUMBAR

13.9

20.4

55.3

45.0

48.9

652

SACRUM

33.4

30.5

195

134

PATELLA

3.1

5.9

28.4

25.9

29.0

276

PELVIS

127

75.0

...

STERNUM

3.1

SKULL

296

171

JAW (1/2)

56.5

40.0

SCAPULA

43.0

38.6

168

VERT. CENT. #1

5.0

4.5

25.0

33.0

300

5.1

—

28.0

ULNA, PROX.

—

6.4

30.0

SESAMOID #1

—

10.2

—

12.5

HUM. PROX.

41.0

47.0

313

170

275

2206

DIST.

41.0

47.0

313

170

275

2206

R/U. PROX.

32.0

31.0'^^

192

127(R)

220

1434

DIST.

32.0

31.0"^^

192

127<«)

220

1434

FEM. PROX.

47.0

79.0

269

203

433

2682

DIST.

47.0

79.0

269

203

433

2682

TIB. PROX.

41.0

79.0

143

220

298

1436

DIST.

41.0

79.0

143

220

298

1436

MT. PROX.

15.1

90.0

88.0

DIST.

15.1

90.0

88.0

MC. PROX.

13.8

85.0

122.0

DIST.

13.8

85.0

122.0

SCAPULA (GLENOID)

21.5

19.3

153

84.0

132.0

* M = Molar.

East Rudolf Paleoecology • Behrensmeyer 573

APPENDIX 2 The value of r„ represents an idealized

r^^i^..i^i- X Lj J r r- I quartz equivalent for the bone which dis-

Calculation of Hydraulic Equivalence j .i rr .. r i

regards the effects of shape.

Processes of sediment transport are gen- It is difficult to predict the effects of
erally explained in terms of quartz grains shape on bone-quartz equivalents. In some
with a standard density of 2.65. Some work cases bone shape may decrease settling
has been done on the hydraulic equivalence velocity by increasing the frictional drag
("equivalent settling velocity") of quartz on the bone, and this will reduce the size
and particles with greater densities to show of the equivalent quartz grain. On the
how small, dense grains sort out with larger, other hand, a bone shape (e.g., a stream-
lighter ones ( Rittenhouse, 1943; Briggs, lined one) that reduces drag may increase
1962). However, there is a lack of infor- the size of the quartz equivalent. The
mation on the hydraulic equivalence of orientation of a bone may have great effects
quartz with particles of lower density such on settling velocit)^ and quartz equivalence,
as bones. Thus, a metapodial dropped parallel to its

Hydraulic equivalence can be considered lonig ^^^s may fall faster than a sphere of

in terms of any two particles that have the equivalent volume, but the same bone

same settling velocity. Given a particular dropped with its long axis horizontal could

bone, it is possible to determine what size settle at a rate slower than that of a sphere,

of quartz grain will settle at the same rate The same bone can alter from equivalence

as the bone. For spheres, hydraulic equiva- to small or large quartz grains by slight

lence to quartz can be easily calculated changes in orientation. In actual transport

using the Impact law. If the settling veloc- situations, some bones tend to orient with

ity for quartz (vq) is to equal the settling long axes parallel to current direction

velocity of a bone (vb), then: (Voorhies, 1969:66-67), and these will

rq =

1 '^07 / 1 ^ — 1 QA7 have maximum hydraulic equivalents for

160 / • ( pq - 1 ) • r„ - 1307 • ( pb - 1 ) • Tb tj^pjy volume. Bones also tend to orient per-

"^Tfi^^"- l^^~n " pendicular to the current, and these will
^q ~ y^~ '' ^^ have smaller effective quartz equivalents.
I Pb - J- ; • Th -pi^g bones that orient transverse to the
1-65 current should be more mobile in transport
— one; situations.
7y = bone density ^^^^^^ ^^ ^ ^reat need for experimental
r.i = radius of quartz grain '^°^^ '!^^^^^^ '""'^ ^^^°^ *,^^^ relationship he-
rb = V. the nominal diameter ^^^^" ,^°"^ '^"^^j^S velocities and quartz
of 1 ffiven bone equivalents and the actual current veloci-
ties necessary for bone entrainment and
If pb = l.5 and rb = 1.0 cm, then r,, = .30 cm. transport.

574 Bulletin Museum of Comparative Zoology, Vol. 146, No. 10

Plate 1. Surface textures of weathered and unweathered bones.

A: a) Unweathered recent bovid radius, showing a smooth, "fresh" surface texture; b) Naturally
weathered bovid femur from the recent East Rudolf thanatocoenose, showing slight roughening
and cracking of the bone surface; c) Distal end of an equid femur from the recent East Rudolf
thanatocoenose, showing extreme flaking and roughening of the bone surface. {Scale in 1 cm and
2 cm intervals)

6: a) Fossil astragalus from the 102-0201 channel sand, showing pre-burial abrasion; b) Fossil astrag-
alus from the 103-0267 distributary channel and beach complex, showing considerable pre-burial
abrasion; c) Recently weathered astragalus from the modern East Rudolf thanatocoenose, showing
the typical cracked weathering pattern on its articular surface (Note: pattern lacking in a) and b));
d) Recent, unweathered astragalus; e) Unweathered and unabraded fossil astragalus from Lo-
cality 8+6-0104 (floodplain). (Scale in 1 cm and 2 cm intervals)

C: a) Distal end of a fossil humerus that was probably weathered prior to burial, showing the typical
cracking pattern on its articular surface; b) Distal end of a recently weathered humerus, showing a
similar cracking pattern; c) Recent, unweathered humerus; d) Fossil humerus from Area 8, East
Rudolf, showing no sign of pre-burial weathering or abrasion. (Scale in 1 cm and 2 cm intervals)

Plate 2. Fracture patterns in recent and fossil bones.

A: a) "Sawtooth" fracture (right side of bovid pelvis); b) "Step" fracture (bovid metapodial); c) "Splin-
tered" fracture (sheep rib); d) "Spiral" fracture (distal end of bovid humerus; e) Weathered bovid
humerus (distal end) with a spiral fracture incurred prior to weathering. (Scale in 1 cm intervals)

6: Spiral fracture on the metatarsal of a recently killed giraffe, presumably caused by a hyaena.
(Scale in 10 cm intervals)

C: a) and b) Typical fracture patterns of bones after mineralization; c) Recent humerus (distal) show-
ing spiral fracture; d) Fossil fragment of a diaphysis, showing a spiral fracture probably incurred
prior to burial and mineralization. (Scale in 1 cm and 2 cm intervals)

Plate 3. The trapping effect of surface vegetation on bones in the recent thanatocoenose on the delta of Laga
Tulu Bor, lleret, East Rudolf.

A: Bovid femur bound by shoreline grass and partially buried. (Scale in 10 cm intervals)
B: Bovid skull and vertebrae, showing loose entrapment by grass. The horn cores are bound firmly to
to the ground by warm tubes (just to right of camera lens cover).

Plate 4. Recent sedimentary environments south of lleret. East Rudolf.

A: A beach bar on the shore of the delta of Laga Tulu Bor, with the open lake to the right and a
closed lagoon or back beach pond to the left. Pebbles and bone debris litter the beach. Beach
bars such as this move shoreward seasonally with the annual rise in lake level (about 1 m fluctua-
tion per year). Depositional environments such as this were probably active in the formation of
Plio-Pleistocene deposits such as those of Localities 130-0201, 105-0208, 103-0267 and 103-0256.

B: Laga Tulu Bor after a brief but heavy rainstorm, with a flow depth of about 1.5 m. The channel is
normally dry for most of the year. A break in the gallery forest that fringes the channel is visible
in the upper right of the photograph. This opens onto the grass-covered floodplain. Some charac-
teristics of this depositional environment are probably comparable to Localities 102-0201 and 105-
1311.

C: The upper part of the deltaic plain of Laga Tulu Bor, showing flooding of a low area (atrophied
channel) after a heavy rain. This area lies in the transition zone between floodplain and deltaic
plain. The sediment is primarily silt. This depositional environment is probably comparable to that
of Locality 8+6-0104 (floodplain).

East Rudolf Paleoecology • Behrensmeyer 575

Plate 1

576 Bulletin Museum of Comparative Zoology, Vol. 146, No. 10

Plate 2

East Rudolf Paleoecology • Behrensmeyer 577

Plate 3

578 Bulletin Museum of Comparative Zoology, Vol. 146, No. 10

Plate 4

827r 070

Harvard

lllllll II II III I HI II

MCZ

LIbrarv

llllllHI

II III I II I II

2044 066 304 205

I i {Ml

Internet Archive Audio

Featured

Top

Images

Featured

Top

Software

Featured

Top

Books

Featured

Top

Video

Featured

Top

Mobile Apps

Browser Extensions

Archive-It Subscription

Save Page Now

Full text of "Bulletin of the Museum of Comparative Zoology at Harvard College"

See other formats