sitar tats

eseoepesese)

Bae eee tia tee ae So Boe acannon

oe
7 a et ol

a
i cy

i

ADVANCES IN HERPETOLOGY
AND
EVOLUTIONARY BIOLOGY

Ernest E. Williams

ADVANCES IN HERPE'TOLOGY
AND
EVOLUTIONARY BIOLOGY

Essays in Honor of
Ernest E. Williams

EDITED BY
ANDERS G.J. RHODIN
KENNETH MIYATA

1983
Museum of Comparative Zoology
Cambridge, Massachusetts

Museum of Comparative Zoology
Harvard University
Cambridge, Massachusetts 02138 U.S.A.

Copyright © 1983 by the Presidents and Fellows of Harvard College

ISBN 0-910999-00-7
Library of Congress Card Catalog Number 73722

All rights reserved
Printed in the United States

PREFACE

In July of 1980 Emest E. Williams
officially retired as Curator of Herpetol-
ogy at the Museum of Comparative Zool-
ogy at Harvard University, a position he
held for a quarter of a century. Emest is
particularly fond of parties, especially if
champagne and various delicacies are
served, and so we marked his retirement
with a celebration in the Romer Hall of
Fossil Vertebrates on a warm summer
afternoon. A host of his close associates
and former graduate students converged
on Cambridge to have a good time (some
may have regretted their efforts the next
day) and to visit a man whom we feel is
unique. On that day, we had the pleasure
of springing a major surprise on Ernest:
this volume in his honor.

This Festschrift was initially con-
ceived as an informal homage to a man
whom so many respect. We contacted as
many of his former students and col-
leagues as possible and solicited papers
for possible publication in the volume.
The response was overwhelming, and it
quickly became obvious that the Fest-
schrift would be a tome of major propor-
tions. The Museum of Comparative
Zoology and its director, Dr. A. W.
Crompton, then generously provided the
Support necessary for an undertaking of
this magnitude, and at the party we were
able to present Emest with a prospectus
of the forthcoming volume. It was one of
the few times we have seen him show
Surprise.

Though the majority of his own work
has been in herpetology, focusing pri-
marily on museum and field investiga-
tions, Ernest’s interests encompass much
of modem evolutionary biology. This
volume is a testament to the extraordi-

nary breadth of his interests in the evolu-
tionary biology of both vertebrates and
invertebrates. More than that, it is a testa-
ment to the enormous influence he has
had on students and colleagues in foster-
ing this wide range of interests. Emest’s
ability to encourage and stimulate re-
search is remarkable, and we can think of
no one else who has influenced such a
diversity of perspectives, approaches,
and personalities. His openness to new
ideas, his willingness to tolerate unusual
approaches and behavior, and his ability
to stimulate and guide without making it
obvious are all unusual, if not unique,
traits in a man of his stature. We doubt
that a single one of his associates has ever
regretted working with him, and more
than one graduate student has character-
ized him as being the “best of all possible
major professors.”

Although he is now retired as Curator
of Herpetology, Williams remains a
Professor of Biology and his research
continues unabated. This past year he re-
ceived his second Guggenheim Fellow-
ship, and he has been busy collaborating
with colleagues across two continents.
Undoubtedly, his influence will continue
to grow in the years to come.

Reflecting Emest’s diverse interests,
this volume in his honor encompasses a
wide spectrum of research investiga-
tions. There are 50 papers by 69 authors,
split into seven broad topics: Syste-
matics, Comparative Morphology, Zoo-
geography, Ecology, Behavior, Evo-
lution, and Conservation. The papers
deal with an enormous array of animal
taxa, from protozoans and arthropods to
birds, bats and primates. However, the
bulk of the volume deals with reptiles

vi Advances in Herpetology and Evolutionary Biology

and amphibians, as do most of Emest’s
own research investigations. There are
49 genera and 144 species of amphibians
discussed, including one new genus and
three new species. There are 338 genera
and 643 species of reptiles represented,
including one new genus, four new
species and one new subspecies. Within
the lizard genus Anolis, central among
Ernest’s interests, there are 97 taxa men-
tioned in the text.

We offer this Festschrift volume in
tribute to Ernest E. Williams and join
with the authors represented here in
expressing our thanks for his friendship,
his support of the investigative process in
biology, and his stimulation of zoological
inquiry among those colleagues fortunate
enough to be within his sphere of influ-
ence.

ACKNOWLEDGMENTS. We wish to
thank all those individuals who helped
with the enormous task of the review
process: Robin M. Andrews, Stephen C.
Ayala, John L. Behler, John W. Bickham,
Robert Bleiweiss, Peter Brazaitis,
Donald Brinkman, Donald G. Broadley,
Lisa Brooks, Walter C. Brown, Charles C.
Carpenter, Robert L. Carroll, Susan M.
Case, William E. Cooper, James R.
Dixon, C. Kenneth Dodd, Jr., William E.
Duellman, Eugene S. Gaffney, Alfred L.
Gardner, Alice G. C. Grandison, Neil
Greenberg, Allen E. Greer, Max K.
Hecht, Paul E. Hertz, Robert D. Holt,
Raymond B. Huey, Robert F. Inger,
Farish A. Jenkins, Jr., Thomas A.
Jenssen, Kenneth V. Kardong, A. Ross
Kiester, F. Wayne King, John Kirsch,
George Lauder, Raymond F. Laurent,
John D. Lynch, Gregory C. Mayer,
Timothy C. Moermond, Scott M. Moody,
Charles W. Myers, Thomas S. Parsons,
Raymond A. Paynter, Jr., Peter C. H.
Pritchard, A. Stanley Rand, Jonathan
Roughgarden, Rodolfo Ruibal, Anthony

P. Russell, Joseph J. Schall, Thomas W.
Schoener, Albert Schwartz, Kathleen K.
Smith, William Stubblefield, Richard
Thomas, Garth Underwood, Romulus
Whitaker, Don E. Wilson, Roger Conant
Wood, and George R. Zug. We also wish
to thank several colleagues of Williams
who extended their encouragement in
this commemorative effort, but were not
among the contributors or reviewers:
Archie Carr, David Crews, Arthur C.
Echternacht, Richard Etheridge, Kurt
Fristrup, Jose Maria Gallardo, William
Garstka, Terry E. Graham, W. Ronald
Heyer, Paul F. A. Maderson, Ernst Mayr,
George Meszoly, Juan A. Rivero, George
Gaylord Simpson, Hobart M. Smith,
Barbara Wu, Rainer Zangerl, and Richard
G. Zweifel.

In addition, we wish to express our
gratitude to those people who helped in
the overall production of this volume: A.
W. Crompton, who provided welcome
institutional and =moral support;
Raymond A. Paynter, Jr., who shepherd-
ed the volume through its early editorial
organizational stages; Elizabeth Camp-
bell Elliott, who did a superb job as tech-
nical editor for the volume; Jose Rosado,
James Knight, and Catherine McGeary of
the Herpetology Section, who assisted
with certain technical aspects; Pere Al-
berch, Williams’s successor as Curator of
Herpetology, for his continuing support;
Laszlo Meszoly, who designed the book
cover; and Susan DeSanctis Rhodin, who
provided most of the initial impetus to
mobilize this massive effort, then pre-
pared the feast which marked the retire-
ment party and presentation of the pros-
pectus, and finally provided the patient
support and encouragement needed to
bring the volume to its fruition.

Anders G. J. Rhodin
Kenneth Miyata

INTRODUCTION

Emest Edward Williams was born in
1914 in Easton, Pennsylvania, the only
child of middle-aged parents. He attend-
ed Lafayette College, where one of his
teachers was B. W. Kunkel, a capable
vertebrate anatomist and an important
early influence. After obtaining his B.S.
at the age of nineteen, financial difficul-
ties forced Williams to partially interrupt
his studies for several years. Fortunately,
having fought in World War II, he was
able to fully resume in 1946 thanks to a
“G.I. Bill” studentship. He enrolled at
Columbia University for a Ph.D. under
W. K. Gregory, the father image of verte-
brate morphology in those days. How-
ever, Gregory retired soon after, and
Williams proceeded on his own.

Williams’s thesis subject, originating
from his association with Kunkel during
the Lafayette days, was variation and
selection in the cervical central articula-
tions of living turtles. The work was
done mostly at the American Museum of
Natural History, where Williams met a
number of his future colleagues and
collaborators, including Samuel B.
McDowell, Max K. Hecht, Karl Koopman
and Rodolfo Ruibal.

The thesis was presented in 1949 and
published in 1950. It is the work of a fine
anatomist with unusually broad views
and interests. Unfortunately, a myopic
committee mutilated it, excising a final
chapter on the relationships between.
selective pressures and variability that,
in reality, had against itself being ahead
of the times. The thesis contains, as an
appendix, a new classification of turtles
to the subfamily level; this type of
“appendix” was an American Museum

vil

trademark that was to become part of
Williams’s style.

Williams’s first employment was as an
instructor in A. S. Romer’s course in
Comparative Anatomy at Harvard (Romer
was also a former Gregory student). This
is how his association with the Museum
of Comparative Zoology (and, incidental-
ly, with me) began.

Kunkel had gotten Williams interested
in turtles, and this was a field well suited
to his temperament—inquisitive, shy,
and independent. It was not crowded,
materials being abundant and easily made
available. It afforded access to a broad
range of subjects, from embryology and
comparative anatomy to paleontology,
and leading into systematics above the
species level and zoogeography. Be-
tween 1950 and 1961 Williams published
23 papers on chelonological matters; one
was coauthored with McDowell and
another with T. S. Parsons, both graduate
students at the time. In 1954 he received
a Guggenheim fellowship to study fossil
turtles in European institutions, es-
pecially the British Museum. The results
then obtained stand to this day, as wit-
nessed, for example, by Gaffney’s con-
tribution to this volume, one of its eight
papers on turtles.

The anatomy of the chelonians, espe-
cially the osteology of the head and neck,
led Williams into similar investigations
of other reptilian groups, a field in which
he published some highly original papers
(e.g., on Gadow’s theory of arcualia and
on the occipito-vertebral joint) and to
which he has sporadically returned. The
articles in this volume by Allen Greer, a
former student, and by T. Frazzetta, a col-

Vili

laborator since 1970, are a reflection of
this phase.

Some of the turtle work on fossils from
Cuba and elsewhere started Williams’s
involvement with the West Indies. A
1950 trip to Jamaica strengthened his
interest in the area and marked the be-
ginning of his work on lizards, in collabo-
ration initially with Garth Underwood,
then teaching in Jamaica, and Rodolfo
Ruibal, a Cuban who was living in the
U.S. but keeping close ties with the
island. Williams’s interest in the Antilles
in the beginning centered on paleontol-

ogy: besides the turtle work there are.

early papers on fossil and sub-fossil mam-
mals with Koopman and with Reynolds
and Koopman. Another of the West
Indian papers, on a fossil tortoise from
Mona, contains the currently accepted
classification of the Testudinidae which
originated the famous footnote by
Loveridge in the coauthored revision of
the African cryptodires. In this footnote
Loveridge contrasted himself, a “work-
ing herpetologist,” with his coauthor,
who represented a “current trend in
herpetology.” This was a_ perceptive
statement, as the next phase of Williams’s
career consisted essentially of contribu-
ting to dilate the borders of lacertilian
systematics to coincide with those of
biology in general. In the process, the
standards of “working herpetology” were
considerably raised and strengthened.

The instructorship at Harvard was for a
period of three years, after which
Williams was promoted in 1952 to Assis-
tant Professor, having added to his former
duties those of coordinating Romer’s
Evolution course in General Education.
Romer really did not care much for the
course, and Williams’s appointment was
more than anything a device to keep him
at Harvard until Loveridge’s retirement
from the herpetology curatorship. It was
an obvious step. Williams had been
working on the collection, both on recent
turtles and on anatomical matters, and
was beginning to become interested in
lizards. Also, his curatorial instincts were
awakening.

Advances in Herpetology and Evolutionary Biology

Loveridge had arrived at the MCZ in
the early twenties, essentially to renovate
the herpetological collections. This he
rapidly did, and he also promoted a pro-
gram of exchanges that could be called
intensive for its time, emphasizing as
complete a representation as possible
of all faunas, but definitely not long se-
ries. An uncommon lover of tidiness,
Loveridge had meticulously planned his
life in order to have completed the
corpus of a Herpetology of East Africa by
the time he was due to retire. (That he
was about one year late he stoutly credit-
ed to the coauthored cryptodiran mono-
graph.) This meant full concentration on
African matters, with only a few brief
digressions into other faunas. As a result,
Williams inherited in 1957 a collection
that was systematically very complete,
but not especially rich in adequate series
or skeletal materials. The department
library, successively kept and added to
by Agassiz, Garman, and Barbour, and
further enriched by the addition of
Loveridge’s library (purchased by
Williams and donated to Harvard) consti-
tuted an extremely valuable resource, but
it was severely restricted to systematics
at the alpha level.

After taking over the curatorship,
Williams immediately began a program
of strengthening the general collections,
of building and steadily adding to a rep-
resentative skeletal collection. Concomi-
tantly, the library grew both in size and
scope, embracing much of ecology and
evolutionary biology. In conjunction
with this, he began taking on graduate
students.

Having become a curator contributed
somewhat to a cooling of Williams’s
enthusiasm for turtle taxonomy. For
instance, he and I had planned a revision
of the South American chelids; the avail-
able materials were borrowed, innumer-
able measurements taken, and graphs
prepared. The project was dropped, how-
ever, due to the absence of significant
series and the logistic problems in handl-
ing any quantity of these very awkward
animals. More recently, this research

has been resurrected by Rhodin and
Mittermeier (this volume).

Among the reputedly strongest parts of
the MCZ herpetological collections was
that of West Indian anoles, mostly as-
sembled by Thomas Barbour. To it
Williams turned his attention. His initial
aim was properly broad: to study the
genus “in depth and at several levels, not
only in terms of the descriptive taxonomy
but also in terms of ecology and be-
havior.”

The first four years of anoline work
consisted of intensive collecting in the
Greater and Lesser Antilles, clearing up
the alpha systematics as far as possible,
and getting acquainted with the types of
problems that mere work on museum col-
lections cannot reveal. Research was also
started on the continent of South Ameri-
ca; as I see it, this rapidly became impor-
tant not so much in itself, but chiefly in
focusing attention on the problem of
generic borders in the Iguanidae, a sub-
ject which to this day looms large in the
work of Williams and his associates
(Estes, Hall, Moody, in this volume). By
this time (1964) the groupings within the
anoles also began to be envisaged, stem-
ming from Etheridge’s initially unpreten-
tious concept of alpha and beta fracture
plane types in tail vertebrae.

Another initial preoccupation was with
digital pads, so characteristic of anoles.
This line of work has progressed with
moderate impetus, but fine results
(Peterson, this volume).

The next phase of work (say to 1970)
brought about a sharpening of the focus.
The main problems perceived were
those related to competition: coadapta-
tion, exclusion, fragmentation. A very
important result of this
Williams’s classic paper on the ecology of
colonization, published in 1969 in the
Quarterly Review of Biology. By this
time Schoener had begun his detailed
studies on resource sharing (Moermond,
this volume). Cytogenetics and_bio-
chemistry were adopted as everyday
tools, in the spirit of the times with a bit
of excessive optimism (‘the ultimate

phase was

INTRODUCTION - Vanzolini ix

tool’), that still lasts in some members of
the group (Gorman et al., this volume).
Thermoregulation, a field that seemed at
first extremely promising, as I see it,
failed to fulfill its promise.

By 1972 Williams had performed the
first test of a model of an intra-island
radiation. Central to the analysis was the
concept of “ecomorphs’—massive
opportunism in diverse lineages leading
to convergence and parallelism to an ex-
treme degree. The concept, to which
Williams had been led by his joint work
with Stan Rand, was not intrinsically
new, but had previously been applied
only to groups differing at least at the
family level and inhabiting broadly sep-
arated geographic areas. Its application to
members of a single genus within a sin-
gle zoogeographic subregion required an
enormous fund of information, and I
believe West Indian anoles constitute the
only viable case so far.

The following (and current) steps of
the concerted attack on West Indian
anoles address still more basic biological
problems: species borders and species
interactions, radiation, and species pack-
ing. “Most species packing in West
Indian Anolis has been achieved by
speciation rather than by accumulation of
already differentiated colonizing species,
and incompletely speciated populations
still exist on West Indian islands. A back-
ground for studies of speciation in West
Indian anoles is provided by the con-
siderable information at hand on the
taxonomy, ecology, behavior, biogeo-
graphy, and morphological and karyo-
typic evolution of West Indian anoles.”

In 1970 Williams was appointed Pro-
fessor of Biology, thus automatically
receiving a Harvard M.A. and becoming,
in Harvard parlance, a “child of the
house,” no longer a foreigner. In 1972 he
was made Agassiz Professor, a position of
real significance, as witnessed by a list of
fellow incumbents, including, among
others, A. S. Romer, P. J. Darlington, Jr.,
Bryan Patterson, and Frank Carpenter.
Since his assumption of the curatorship,
the Harvard system had also allowed him

x Advances in Herpetology and Evolutionary Biology

to have graduate students. In keeping
with Harvard administrative tradition, it
is a moot point who was Williams’s first
student, Carl Gans or Stan Rand. How-
ever that may be, there has been a steady
flow ever since, and the type of relation-
ship between professor and students has
always been to me a matter of admiration.
The main characteristic of the group has
been diversity in concentration, i.e., to
bring all sorts of approaches to bear on a
theoretical problem embodied in a sys-
tematic unit. Within this very broad
mandate there has been full freedom to
pursue individual lines of thought. In
fact, I can think of no other environment
where a man as singular as Robert
Trivers could have matured his signal
contributions to evolutionary theory.

I have stated above that in the fifties
zoology, as practiced in museums, and
especially in the U.S., underwent a
marked change in emphasis and even in
scope, embracing theoretical and experi-
mental aspects that had previously been
outside the range of the profession. A
notable forerunner was Charles M.
Bogert, of the American Museum of
Natural History, who, besides impec-
cable field and taxonomic research, did
basic work in such new areas as thermo-
regulation in reptiles and acoustic com-

munication in amphibians. The trend
developed rapidly; notwithstanding all
the favorable aspects, in one important
respect the consequences have been less
fortunate. Emphasis on the experimental
approach, demanding a different type of
training, entailing stiff academic compe-
tition and an element of fashion, has
tended to lower the prestige and the
quality of curatorship. It has become in
Many cases increasingly difficult to base
museum policy on the good of the collec-
tions—even their intrinsic worth has
been questioned. Williams has been a
steadfast and articulate advocate of the
collection as the heart and kemel of the
museum department, and its most endur-
ing legacy. For Williams, the basic task of
a curator is still to enlarge, structure, and
make available a collection. His career
has been better known through a consis-
tent broadening and deepening of scien-
tific preoccupations; that this has gone
hand in hand with a definite philosophy
and practice of selfless curatorship
rounds off a complete and entire scientif-
ic personality. The esteem and respect
evinced in this volume bear ample wit-
ness to this personality.

P. E. Vanzolini

LIST OF CONTRIBUTORS

ANGEL C. ALCALA

Ros

Professor of Biology and Vice President for
Academic Affairs, Silliman University,
Dumaguete City, Philippines. M.A., Ph.D.
(herpetology and ecology), Stanford University.
Association with E. E. Williams indirect
through W. C. Brown. Research interests in-
clude herpetology and ichthyology and conser-
vation of Philippine land vertebrates and
marine mammals.

IN M. ANDREWS

Associate Professor of Biology, Virginia Poly-
technic Institute and State University, Blacks-
burg, Virginia 24061; Research Associate,
Smithsonian Tropical Research Institute, Bal-
boa, Panama. Ph.D. (zoology), University of
Kansas. Post-doctoral fellowship under E. E.
Williams and A. S. Rand. Research interests in-
clude life history evolution of reptiles, physio-
logical ecology of reptiles, especially with re-
gards to eggs, energetics and thermoregulation.

WALTER AUFFENBERG

Curator in Herpetology, Florida State Museum,
University of Florida, Gainesville, Florida
32611. M.S. (herpetology), Ph.D. (paleontology
of reptiles), University of Florida. Former spe-
cial student and teaching assistant under E. E.
Williams at Harvard. Research interests in-
clude behavioral ecology of varanid lizards and
paleontology and behavior of tortoises.

ANGUS D’A. BELLAIRS

Emeritus Professor of Vertebrate Morphology,
St. Mary’s Hospital Medical School, University
of London, Paddington, London, W.2., United
Kingdom. D.Sc., London; M.R.C.S. (Member of
the Royal College of Surgeons), London. Pro-
fessional associate of E. E. Williams for many
years. Research interests include the develop-
ment, structure and evolution of the vertebrate
skeleton and biology of the Reptilia.

JupITH A. BLAKE

Curatorial Assistant, Division of Herpetology,

Museum of Comparative Zoology, Harvard.

University, Cambridge, Massachusetts 02138;
Biologist, Ruecker Wildlife Refuge, 137
Seapowet Ave., Tiverton, Rhode Island 02878.
Ph.D. (evolutionary biology), Harvard Univer-
sity under E. E. Williams. Research interests
include evolutionary biology, genetics of
speciation and conservation biology.

DONALD G. BROADLEY

Curator of Herpetology, National Museum,

Bulawayo, Zimbabwe. M.Sc., Ph.D. (herpetol-
ogy), University of Natal. Professional associate
of E. E. Williams for many years. Research in-
terests include taxonomy, ecology and zooge-
ography of the reptiles of Africa.

WALTER C. BROWN

Emeritus Professor of Biology, Menlo College,
Menlo Park, California 94025; Fellow and
Research Associate in Herpetology, California
Academy of Science, San Francisco, California
94118. Ph.D. (biology), Stanford University.
Professional associate of E. E. Williams for
many years. Research interests include system-
atics, evolution and ecology of amphibians and
reptiles, with special interest in the Philippines
and the Pacific.

DONALD G. BUTH

Assistant Professor of Biology, Department of
Biology, University of California, Los Angeles,
California 90024. Ph.D. (ecology, ethology and
evolution), University of Illinois under T. H.
Frazzetta (former post-doctoral fellow under E.
E. Williams at Harvard). Association with E. E.
Williams through both T. H. Frazzetta and G.
C. Gorman. Research interests include bio-
chemical systematics of reptiles and fishes.

ROGER CONANT

Adjunct Professor, Department of Biology,
University of New Mexico, Albuquerque, New
Mexico 87131; Research Associate, Academy of
Natural Sciences, Philadelphia, and American
Museum of Natural History, New York; Past-
President and former Secretary, American
Society of Ichthyologists and Herpetologists;
Director and Curator of Reptiles Emeritus,
Philadelphia Zoological Garden. Honorary
Sec.D., University of Colorado. Professional as-
sociate of E. E. Williams for many years. Re-
search interests include herpetology of the
Chihuahuan desert and biology of Agkistrodon.

ISABEL D. CONSTABLE

Undergraduate Student, Brown University,
Providence, Rhode Island 02912. Association
with E. E. Williams indirect through R. A.
Mittermeier; also, father associated with A.
Loveridge as undergraduate student in herpe-
tology at the MCZ. Research interests include
primates, tropical forests and wildlife conserva-
tion.

JAMES R. DIxON

Professor and Chief Curator, Texas Coopera-
tive Wildlife Collections, Department of Wild-

xil

Advances in Herpetology and Evolutionary Biology

life and Fisheries Sciences, Texas A & M Uni-
versity, College Station, Texas 77843; Past-
President, Society for the Study of Amphibians
and Reptiles; Board of Governors, American
Society of Ichthyologists and Herpetologists;
Chairman, Board of Scientists, Chihuahuan
Desert Research Institute. Ph.D. (zoology),
Texas A & M University. Professional associate
of E. E. Williams for many years. Research in-
terests include herpetology of the Americas.

LUDMELA DJATSCHENKO

Dentist, 282 West Acres Dr., Guelph, Ontario,
N1H 7P1, Canada. D.D.S., University of Toron-
to. Association with E. E. Williams indirect
through T. S. Parsons.

WILLIAM E. DUELLMAN

Curator, Division of Herpetology, Museum of
Natural History, and Professor, Department of
Systematics and Ecology, University of Kansas,
Lawrence, Kansas 66045; Past-President,
Herpetologists’ League. Ph.D. (zoology), Uni-
versity of Michigan. Professional associate of E.
E.. Williams for many years. Research interests
include evolutionary biology of amphibians
with emphasis on systematics, biogeography,
and reproductive biology of Neotropical
anurans.

PAUL ELIAS

Field Editor for Europe, Advanced Book Pro-
gram, Addison-Wesley Publishing Co., Read-
ing, Massachusetts 01867. M.A. (zoology),
University of Califormia, Berkeley; A.M. (bi-
ology), Harvard University under P. Alberch
(E. E. Williams’s successor as Curator of
Herpetology). Research interests include the
constraints imposed upon evolution by devel-
opment and their ramifications in systematics.

NELLY CARRILLO DE ESPINOZA

Chief, Section of Herpetology, Museo de His-
toria Natural “Javier Prado,” Universidad
Nacional Mayor de San Marcos, Avenida
Arenales 1256, Lima, Peru. Doctor (herpetol-
ogy), Universidad Nacional Mayor de San
Marcos. Professional associate of E. E. Wil-
liams for many years. Research interests in-
clude Neotropical snakes and lizards.

RICHARD ESTES

Professor of Zoology, San Diego State Univer-
sity, San Diego, California 92182, and Research
Associate, Museum of Paleontology, University
of California, Berkeley, California 94720;
Former Research Associate, Museum of Com-
parative Zoology, Harvard University. Ph.D.
(paleontology), University of California,
Berkeley. Association with E. E. Williams
through research at the MCZ. Research inter-
ests include morphology and systematics of
lower vertebrates, paleoherpetology, faunal
evolution and biogeography.

KARL F. FORSGAARD

Attomey, Houghton Cluck Coughlin & Riley,
1111 Third Ave., Suite 1500, Seattle, Washing-

ton 98101. Association with E. E. Williams as
undergraduate student at Harvard and former
curatorial assistant in herpetology at the MCZ.
Research interests include high-altitude and
polar zoology, colubrid and iguanid ecology
and taxonomy, and wildlife protection law.

THOMAS H. FRAZZETTA

Professor of Ecology, Ethology and Evolution,
University of Illinois, Urbana, Illinois 61801;
Associate in Herpetology, Museum of Com-
parative Zoology, Harvard University, Cam-
bridge, Massachusetts 02138. Ph.D. (vertebrate
zoology), University of Washington. Post-doc-
toral fellowship under E. E. Williams at
Harvard. Research interests include evolution-
ary biology, biomechanics of vertebrate jaws,
especially reptiles and galeoid sharks, popula-
tion biology.

EUGENE S. GAFFNEY

Curator, Department of Vertebrate Paleontol-
ogy, American Museum of Natural History,
79th St. and Central Park West, New York, New
York 10024. Ph.D. (geology), Columbia Univer-
sity. Professional associate of E. E. Williams for
many years. Research interests include turtle
systematics.

CarL Gans

Professor of Zoology, The University of Michi-
gan, Ann Arbor, Michigan 48109; Research
Associate, Carnegie Museum, Pittsburgh, and
American Museum of Natural History, New
York; former Guggenheim Fellow; Past-
President, American Society of Ichthyologists
and _ Herpetologists. M.S. (engineering),
Columbia University; Ph.D. (biology), Harvard
University under E. E. Williams (his first
graduate student). Research interests include
functional morphology, reptilian adaptations.

GEORGE C. GORMAN

Professor of Biology, University of California,
Los Angeles, California 90024; former Re-
search Associate, Museum of Comparative
Zoology, Harvard University. Ph.D. (herpetol-
ogy), Harvard University under E. E. Williams.
Research interests include evolution of Anolis,
population genetics, biochemical evolution,
reproductive biology of reptiles.

ALLEN E. GREER

Curator of Reptiles and Amphibians, The
Australian Museum, 6-8 College St., Sydney,
New South Wales 2000, Australia. Ph.D. (biol-
ogy), Harvard University under E. E. Williams.
Post-doctoral fellowship under E. E. Williams
at Harvard. Research interests include biology
and evolution of reptiles, especially squamates.

STELLA GUERRERO

Graduate Student, Department of Entomology
and Parasitology, University of California,
Berkeley, California 94720; former Research
Assistant, Smithsonian Tropical Research Insti-
tute, Balboa, Panama. Association with E. E.
Williams indirect through A. S. Rand and R. M.

Andrews. Research interests include ecology
and effects of parasites on wild populations,
especially lizards.

WILLIAM P. HALL
Post-Doctoral Fellow, Queen’s College, Uni-
versity of Melbourne, Parkville, Victoria 3052,
Australia; former Visiting Assistant Professor of
EPO Biology, University of Colorado, Boulder;
former Visiting Lecturer, University of Mary-
land, College Park. Ph.D. (biology), Harvard
University under E. E. Williams. Research
interests include functional relationships be-
tween fixation of chromosomal mutations and
speciation in iguanid lizard radiations, and
roles of hybrid zones in maintaining genetic
isolation in lizards and frogs.

Max K. HECHT
Professor of Biology, Queens College, City
University of New York, Flushing, New York
11367. Ph.D. (vertebrate zoology), Cornell
University. Association with E. E. Williams
through field work in the West Indies. Re-
search interests include vertebrate systematics.

PAUL E. HERTZ
Assistant Professor of Biological Sciences,
Barnard College, Columbia University, 606
West 120th St., New York, New York 10027.
Ph.D. (biology), Harvard University under
E. E. Williams. Research interests include rep-
tile ecology, the evolution of thermal biology,
and mechanisms of adaptation.

ROBERT A. HICKS
P.O. Box 420, Ipswich, Massachusetts 01938;
former Employee, Massachusetts Audubon
Society, National Marine Fisheries Service,
and Marine Resources Conservation Founda-
tion. B.A., Harvard University, field research
assistant with E. E. Williams and R. L. Trivers
in the West Indies.

ROBERT D. HOLT
Assistant Professor, Department of Systematics
and Ecology, and Assistant Curator, Museum of
Natural History, University of Kansas,
Lawrence, Kansas 66045. Ph.D. (biology),
Harvard University under E. E. Williams. Re-
search interests include theoretical ecology,
island biology, predator-prey interactions and
tropical ecology of birds and lizards.

RAYMOND B. HUEY
Associate Professor of Zoology, Department of
Zoology NJ-15, University of Washington,
Seattle, Washington 98195. M.A. (zoology),
University of Texas; Ph.D. (biology), Harvard
University under E. E. Williams. Research
interests include physiological ecology, defen-
sive behavior and interspecific competition.

ROBERT F. INGER
Curator, Amphibians and Reptiles, Field
Museum of Natural History, Chicago, Illinois
60605. Ph.D. (zoology), University of Chicago.
Professional associate of E. E. Williams for
many years. Research interests include evolu-

LIsT OF CONTRIBUTORS xill

tionary ecology and systematics of Southeast
Asian amphibians and reptiles and structure
and organization of communities of reptiles and
amphibians in tropical forests.

THOMAS A. JENSSEN

Associate Professor of Zoology, Department of
Biology, Virginia Polytechnic Institute and
State University, Blacksburg, Virginia 24061.
Ph.D. (zoology), University of Oklahoma. Post-
doctoral fellowship under E. E. Williams at
Harvard. Research interests include behavioral
ecology of amphibians and reptiles, especially
display behavior of Anolis.

A. Ross KIESTER

Assistant Professor of Biology, Department of
Biology, Tulane University, New Orleans,
Louisiana 70118. Ph.D. (biology), Harvard
University under E. E. Williams. Junior
Fellow, Society of Fellows, Harvard. Research
interests include evolutionary and population
biology.

KARL F. KOOPMAN

Curator, Department of Mammalogy, American
Museum of Natural History, 79th St. at Central
Park West, New York, New York 10024. Ph.D.
(zoology), Columbia University. Fellow grad-
uate student and teaching assistant with E. E.
Williams at Columbia, professional associate
for many years. Research interests include
systematics and zoogeography of Chiroptera.

RAYMOND F. LAURENT

Professor and Principal Investigator, Consejo
Nacional de Investigaciones Cientificas y
Tecnicas, Fundacion Lillo, Miguel Lillo 205,
4000 Tucuman, Argentina; former Associate
Professor, Museum National d’Histoire Na-
turelle, Paris; former Research Zoologist with
E. E. Williams at Harvard. Ph.D. (zoology),
University of Bruxelles. Research interests
include African and South American herpetol-
ogy, especially systematics, evolution and
zoogeography.

JAMES D. LAZELL, Jr.

President, The Conservation Agency, 8
Swinbume St., Conanicut Island, Rhode Island
02835; Associate in Herpetology, Museum of
Comparative Zoology, Harvard University,
Cambridge, Massachusetts 02138. M.A. (bi-
ology), Harvard University under E. E. Wil-
liams; M.S. (zoology), University of Illinois;
Ph.D. (biology), University of Rhode Island.
Research interests include field and evolution-
ary biology of reptiles, amphibians, and
mammals.

Joun D. LYNCH

Professor of Life Sciences, School of Life Sei-
ences, University of Nebraska, Lincoln,
Nebraska 68588. Ph.D. (zoology), University of
Kansas. Professional associate of E. E. Williams
for many years. Research interests include
systematics and biogeography of neotropical
frogs, especially Eleutherodactylus.

Advances in Herpetology and Evolutionary Biology

Grecory C. MAYER

Graduate Student, Museum of Comparative
Zoology, Harvard University, Cambridge, Mas-
sachusetts 02138. Ph.D. candidate (biology),
Harvard University under E. E. Williams.
Research interests include the genetics, ecol-
ogy and geography of speciation.

SAMUEL B. MCDOWELL

Professor of Zoology, Rutgers University,
Newark, New Jersey 07102. Ph.D. (zoology),
Columbia University under G. G. Simpson.
Fellow graduate student with E. E. Williams,
illustrated his Ph.D. thesis, co-authored an
early paper on turtle morphology, continued
professional association conceming relation-
ships of West Indian insectivores. Research
interests include New Guinea reptiles, evolu-
tion of the vertebrate skull, origin of snakes,
notacanthiform fishes, lipotyphlan mammals.

FEDERICO MEDEM

Professor, Universidad Nacional de Colombia,
Bogota, and Director, Estacidn de Biologia
Tropical “Roberto Franco,” Apartado aereo 22—
61, Villavicencio (Meta), Colombia; Member,
IUCN/SSC Steering Committee and Crocodile
Specialist Group; former Guggenheim Fellow.
Dr.rer.nat. (zoology, paleontology, geology),
Humboldt University, Berlin. Professional as-
sociate of E. E. Williams for many years. Re-
search interests include ecology and zoo-
geography of turtles and crocodiles.

RUSSELL A. MITTERMEIER

Chairman, IUCN/SSC Primate Specialist
Group and Director, World Wildlife Fund—
U.S. Primate Program, World Wildlife Fund—
U.S., 1601 Connecticut Ave., NW, Washington,
D.C. 20009; Associate in Herpetology, Mu-
seum of Comparative Zoology, Harvard Uni-
versity, Cambridge, Massachusetts; Vice-
Chairman, IUCN/SSC Freshwater Chelonian
Specialist Group. Ph.D. (biological anthropol-
ogy), Harvard University under I. DeVore.
Research interests include primatology, herpe-
tology and conservation.

KENNETH MIYATA

Post-Doctoral Fellow, Division of Amphibians
and Reptiles, Smithsonian Institution, National
Museum of Natural History, Washington, D.C.
20560; Research Associate in Vertebrate Zool-
ogy, Smithsonian Institution; Associate in
Herpetology, Museum of Comparative Zool-
ogy, Harvard University. Ph.D. (biology),
Harvard University under E. E. Williams.
Research interests include biogeography and
ecology of neotropical herpetofauna.

Timotuy C. MOERMOND

Associate Professor, Department of Zoology,
University of Wisconsin, Madison, Wisconsin
53706. Ph.D. (evolutionary biology), Harvard
University under E. E. Williams. Research
interests include feeding behavior of animals,
optimal foraging strategies and community

structure, especially in Anolis lizards and birds,
co-evolution of frugivorous birds and bird-dis-
persed plants.

Scott M. Moopy

Assistant Professor of Biomedical Science,
Department of Zoology and College of Osteo-
pathic Medicine, Ohio University, Athens,
Ohio 45701. M.S., Ph.D. (evolutionary and
ecological biology), University of Michigan.
Post-doctoral fellowship, Deutscher Akademi-
sches Austauschdienst. Association with E. E.
Williams as undergraduate student at Harvard
and former curatorial assistant in herpetology at
the MCZ. Research interests include compara-
tive and functional morphology and phylogene-
tics, zoogeography and paleontology of lizards,
especially Agamidae.

Jose ALBERTO OTTENWALDER

Curator, Department of Mammalogy, Museo
Nacional de Historia Natural, Plaza de la Cul-
tura, and Department of Zoology, Investigation
and Conservation, Parque Zoologico Nacional,
Santo Domingo, Dominican Republic; Mem-
ber, IUCN/SSC Crocodile, Insectivore and
Elephant Shrew Specialist Groups; Museum
Specialist, Carnegie Museum of Natural
History; Field Associate, Florida State Mu-
seum. Licenciado (mammalogy), University of
Santo Domingo. Professional associate of E. E.
Williams. Research interests include zooge-
ography, conservation and insular biology of
West Indies vertebrates.

STEPHEN W. PACALA

Assistant Professor, Ecology Section, Biological
Sciences Group, University of Connecticut,
Storrs, Connecticut 06268. Ph.D. (ecology),
Stanford University under J. Roughgarden.
Association with E. E. Williams indirect
through J. Roughgarden. Research interests
include development and testing of population
dynamics and niche theories.

FRED PARKER

717 Ross River Rd., Kirwan, Townsville,
Queensland 4814, Australia; former District
Officer, Papua New Guinea; former Assistant
Secretary for Wildlife, Department of Natural
Resources, Papua New Guinea; Member,
IUCN/SSC Freshwater Chelonian Specialist
Group and Snake Specialist Group. Association
with E. E. Williams through many years of col-
lecting New Guinean and Australian herpeto-
fauna for the Museum of Comparative Zoology.
Research interests include herpetology of
Papua New Guinea and the Solomon Islands.

THOMAS S. PARSONS

Professor of Zoology, University of Toronto,
Toronto, Ontario M5S 1M1, Canada; former
Instructor in Biology, Harvard University.
M.A., Ph.D. (comparative anatomy), Harvard
University under A. S. Romer. E. E. Williams
was on Ph.D. thesis committee, and an early
collaborator on turtle morphology. Research

interests include comparative anatomy and
paleontology of vertebrates, especially reptiles.

JANE A. PETERSON
Assistant Professor of Biology, University of
California, Los Angeles, California 90024.
Ph.D. (anatomy), University of Chicago under
D. B. Wake. Post-doctoral fellowship at
Harvard University under E. E. Williams and
A. W. Crompton. Research interests include
adaptation in locomotor behavior and morphol-
ogy, particularly of lower tetrapods, biomech-
anics, evolution of complex functional units,
scaling in the musculoskeletal system.

MakkK J. PLOTKIN
Associate in Ethnobotanical Conservation,
Harvard Botanical Museum, Oxford St., Cam-
bridge, Massachusetts 02138, and Ethnobotany
Project Director, World Wildlife Fund-U:S.,
1601 Connecticut Ave., N.W., Washington,
D.C. 20009. M.F.S. (forestry), Yale University.
Association with E. E. Williams as former cura-
torial assistant in herpetology at the MCZ.
Research interests include neotropical conser-
vation, ethnobotany and herpetology.

A. STANLEY RAND
Senior Scientist, Smithsonian Tropical Re-
search Institute, Balboa, Panama. Ph.D. (zo-
ology), Harvard University under E. E. Wil-
liams. Research interests include behavior and
ecology of tropical amphibians and reptiles.

ANDERS G. J. RHODIN
Instructor and Chief Resident, Section of Or-
thopaedic Surgery, Yale University School of
Medicine, New Haven, Connecticut 06510
(present address: Orthopaedic Associates, P.C.,
Nichols Rd., Fitchburg, Massachusetts 01420),
and Associate in Herpetology, Museum of
Comparative Zoology, Harvard University,
Cambridge, Massachusetts 02138; Member,
IUCN/SSC Freshwater Chelonian Specialist
Group. M.D., University of Michigan. Associa-
tion with E. E. Williams also as former cura-
torial assistant in herpetology at the MCZ.
Research interests include systematics,
phylogeny, biology and zoogeography of
turtles, especially neotropical and Australo-
New Guinean chelids, comparative chondro-
osseous development of vertebrates, particu-
larly marine turtles.

Marc G. M. vAN ROOSMALEN
Ecologist and Primatologist, Research Institute
for Nature Management, Broekhuizerlaan 2,

P.O. Box 46, 3956ZR Leersum, Netherlands;-

Member, IUCN/SSC Primate Specialist Group.
Ph.D. (ecology), Agricultural University of
Wageningen, Netherlands. Association with E.
E. Williams indirect through R. A. Mittermeier.
Research interests include ecology and con-
servation of neotropical primates and tropical
ecosystems.
FRANKLIN D. Ross
Curatorial Assistant, Department of Reptiles

LIST OF CONTRIBUTORS XV

and Amphibians, Museum of Comparative
Zoology, Harvard University, Cambridge,
Massachusetts 02138. Research interests in-
clude ecology and systematics of reptiles and
amphibians, in particular the paleontology and
zoogeography of crocodilians.

JAMES PERRAN Ross
Associate in Herpetology, Museum of Com-
parative Zoology, Harvard University, Cam-
bridge, Massachusetts 02138; Member, IUCN/
SSC Marine Turtle Specialist Group. Ph.D.
(zoology), University of Florida. Research
interests include physiological ecology and
distribution of sea turtles, and application of
ecology to conservation of endangered species.

JONATHAN ROUGHGARDEN
Professor of Biological Sciences, Stanford
University, Stanford, California 94305. Ph.D.
(biology), Harvard University under E. E. Wil-
liams. Research interests include theoretical
and community ecology.

JoHN D. RUMMEL
Graduate Student, Department of Biological
Sciences, Stanford University, Stanford, Cali-
fornia 94305. Ph.D. candidate (ecology), Stan-
ford University under J. Roughgarden. Assoei-
ation with E. E. Williams indirect through J.
Roughgarden. Research interests include ex-
perimental ecology, niche theory and the effect
of population processes on ecosystem charac-
teristics.

AMY SCHOENER
Research Assistant Professor, Department of
Oceanography, University of Washington,
Seattle, Washington 98195. Ph.D. (biology),
Harvard University. Participated with E. E.
Williams and T. W. Schoener on lizard studies
in the West Indies. Research interests include
island biogeography and marine ecology.

THomMas W. SCHOENER
Professor of Zoology and Environmental
Studies, Zoologist in the Agricultural Experi-
ment Station, University of California, Davis,
California 95616; former Professor, Depart-
ment of Zoology, University of Washington,
Seattle, Washington; former Associate Profes-
sor, Department of Biology, Harvard Univer-
sity. Ph.D. (biology), Harvard University under
E. E. Williams and E. O. Wilson. Junior Fel-
low, Harvard. Also associated with E. E. Wil-
liams as undergraduate student at Harvard.
Research interests include ecology in general.

ALBERT SCHWARTZ
Professor, Department of Biology, Miami-Dade
Community College, North Campus, Miami,
Florida 33167; Research Associate, Depart-
ment of Amphibians and Reptiles, Carmegie
Museum of Natural History, Pittsburgh. M.S.
(parasitology), University of Florida; Ph.D.
(mammalogy), University of Michigan. Profes-
sional associate of E. E. Williams for many
years. Research interests include taxonomy and

XV1

JON

Advances in Herpetology and Evolutionary Biology

zoogeography of amphibians, reptiles, birds
and mammals of the West Indies.

SEGER

Post-doctoral Fellow, Museum of Zoology,
University of Michigan, Ann Arbor, Michigan
48109. Ph.D. (biology), Harvard University
under E. E. Williams. Research interests in-
clude kin selection theory and evolution of
genetic systems and sex ratios, especially in
solitary Hymenoptera.

MICHAEL E. SOULE

Director, Institute for Transcultural Studies,
901 S. Normandie Ave., Los Angeles, California
90006; former Professor of Biology, University
of California, San Diego. M.A., Ph.D. (popula-
tion biology), Stanford University. Professional
association with E. E. Williams mainly through
G. C. Gorman. Research interests include varia-
tion in natural populations and conservation
biology.

SAMUEL F. TARSITANO

Adjunct Assistant Professor, Department of
Biology, Queens College, City University of
New York, Flushing, New York 11367. Ph.D.
(comparative and functional anatomy), City
University of New York. Association with E. E.
Williams indirect through M. K. Hecht. Re-
search interests include functional morphology
and vertebrate paleontology.

RICHARD THOMAS

Associate Professor, Department of Biology,
University of Puerto Rico, Rio Piedras, Puerto
Rico 00931. Ph.D. (zoology), Louisiana State
University. Professional associate of E. E. Wil-
liams for many years. Research interests in-
clude systematics of West Indian amphibians
and reptiles.

ROBERT L. TRIVERS

Professor of Biology, University of California,
Santa Cruz, California 95064; former Assistant
and Associate Professor, Department of Biol-
ogy, Harvard University. Ph.D. (biology),
Harvard University under E. E. Williams.
Regents post-doctoral fellowship, Smithsonian
Institution. Research interests include social
evolution in general.

LINDA TRUEB

Associate Curator, Division of Herpetology,

Museum of Natural History, and Associate
Professor, Department of Systematics and
Ecology, University of Kansas, Lawrence,
Kansas 66045. Ph.D. (zoology, herpetology),
University of Kansas. Professional associate of
E. E. Williams. Research interests include
evolutionary biology of amphibians with
emphasis on osteology and systematics of
anurans.

PAULO EMILIO VANZOLINI

Director, Museu de Zoologia, Universidade de
Sao Paulo, Sao Paulo, Brazil. Ph.D. (zoology),
Harvard University under A. S. Romer. Profes-
sional associate and collaborator with E. E.
Williams for more than thirty years. Research
interests include herpetology and evolution.

DAVID B. WAKE

Professor of Zoology and Director, Museum of
Vertebrate Zoology, University of California,
Berkeley, California 94720; President, Society
for the Study of Evolution. Ph.D. (biology),
University of Southern California. Professional
associate of E. E. Williams for many years. Re-
search interests include origin of evolutionary
novelty and adaptive radiation, evolutionary
biology of plethodontid salamanders, size and
shape in ontogeny and phylogeny, and func-
tional, developmental, and comparative
morphology.

ROGER CONANT Woop

Professor of Zoology, Stockton State College,
Pomona, New Jersey 08240; Member IUCN/
SSC Freshwater Chelonian Specialist Group.
Ph.D. (vertebrate paleontology), Harvard Uni-
versity under B. Patterson and E. E. Williams.
Post-doctoral fellowships at Harvard and Flori-
da Audubon Society. Research interests in-
clude evolution, morphology, ecology and
behavior of fossil and living turtles.

SuH YUNG YANG

Professor and Chairman, Department of Biol-
ogy, Inha University, Inchon, Korea. Ph.D.
(zoology), University of Texas. Association with
E. E. Williams indirect through G. C. Gorman.
Research interests include evolutionary biol-
ogy in vertebrates.

TABLE OF CONTENTS
SYSTEMATICS

ELIAS, PAUL, AND DAVID B. WAKE. Nyctanolis pernix, a new genus and
species of plethodontid salamander from northwestern Guatemala and

Cinigions,, IMIGSCGO, tec are nee ae rs ane eR) Pee ae I
INGER, ROBERT F. Larvae of southeast Asian species of Leptobrachium and
weptoonacwella(Anura: Pelobatidae). .....-....-.+-5+4-40.5e 20458. 13
DUELLMAN, WILLIAM E., AND LINDA TRUEB. Frogs of the Hyla columbiana
group: taxonomy and phylogenetic relationships. .................. 33
LYNCH, JOHN D. A new leptodactylid frog from the Cordillera Oriental of
Callomplotans . 0:0: 65.5: ce Renae rn een PRR Ones tis tee tin Se ehent rors) > 8 52

RHODIN, ANDERS G. J., AND RUSSELL A. MITTERMEIER. Description of
Phrynops williamsi, a new species of chelid turtle of the South American

P. geoffroanus complex. Bae at ln, AM OT A aie DE ON le AM One See 58
WOOD, ROGER CONANT. Kenyemys williamsi, a fossil pelomedusid turtle
from few OCENETOLNENYV As Adlcsiuds ica ees Nona de MERON ee 74
THOMAS, RICHARD, AND ALBERT SCHWARTZ. Variation in Hispaniolian
Spnacnrodaciylus (Sauria: Gekkonidae). ....:..2.-+.....-----0--::- 86

LAZELL, JAMES D., JR. Biogeography of the herpetofauna of the British
Virgin Islands, with description of a new anole _ (Sauria:

Tene raTGlAVS)), “So Blo Ee es ee ee a eee eRe eR ENN FORO EN, try Ok 99
VANZOLINI, P. E. Guiano-Brasilian Polychrus: distribution and speciation
(Sauria: Coun Tal Gl Yes) rece ra Votes ok eerie ated tliat nce ea 118
DIXON, JAMES R. Systematics of the Latin American snake, Liophis
Epinepnelus (Serpentes: Colubridae). .....2-...sea-.++++255---e-0- 1183

COMPARATIVE MORPHOLOGY

BELLAIRS, A. D’A. Partial cyclopia and monorhinia in turtles. ........ 150
BROADLEY, DONALD G. Neural pattern—a neglected taxonomic character in
the genus Pelusios Wagler (Pleurodira: Pelomedusidae). .......... 159

MCDOWELL, SAMUEL B. The genus Emydura (Testudines: Chelidae) in
New Guinea with notes on the penial morphology of Pleurodira. . 169
GAFFNEY, EUGENE S. The basicranial articulation of the Triassic turtle,
noganochelys: 5 ..0..25.:- Pes SOME neva et cieys cSt nek ee eNs 190
Moopy, SCOTT M. The rectus abdominis muscle complex of the Lacertilia:

terminology, homology and assumed presence in primitive iguanian
[Mizeracleys °) So Ae, ot es ee BNE ERE A OMe TUNED Cenc ete Alen LA eet Trae 195

Xvi

xviii Advances in Herpetology and Evolutionary Biology

GREER, ALLEN E. On the adaptive significance of the reptilian spectacle:

the evidence from scincid, teiid, and lacertid lizards. ............ 213
FRAZZETTA, T. H. Adaptation and function of cranial kinesis in reptiles:
a time-motion analysis of feeding in alligator lizards. ............. 222,
PETERSON, JANE A. The evolution of the subdigital pad in Anolis. I.
Comparisons among the anoline genera. ...................00005- 245
FORSGAARD, KARL. The axial skeleton of Chamaelinorops. ........... 284
PARSONS, THOMAS S., AND LUDMELA DJATSCHENKO. Variation in the left
lung and bronchus of Thamnophis sirtalis parietalis. ............. 296
ROSS, FRANKLIN D., AND GREGORY C. MAYER. On the dorsal armor of the
Crocodilia? ..si0nedgh es aoa ee ee 305
HECHT, MAX K., AND SAMUEL F. TARSITANO. The tarsus and metatarsus of
Protosuchus EVoVal thes jOAAS TheMyOlCRSIONS, 5..0ece00cens062-50025- 332
ZOOGEOGRAPHY
LAURENT, R. F. About the herpetofauna of central African montane
FOREST. esc n S4iuses Bete HU ide RRS. erage 350
KIESTER, A. ROSS. Zoogeography of the skinks (Sauria: Scincidae) of Arno
Atoll, Marshall Wolamdsi. acs 6:5 2 5 cae eee se Ei eee 359

ESTES, RICHARD. The fossil record and early distribution of lizards. .. 365
CONANT, ROGER. Commentary on a frog and lizard newly recorded from

central Durango,Me@xicos hai ..,4.60 ese ee eee 399
ESPINOZA, NELLY CARRILLO DE. List of Peruvian Anolis with distributional
data. (Sauria: Ieuamidae). ssc.0624e0lel eee Cee eee 406
KOOPMAN, KARL F. Two general problems involved in systematics and
Zoogeosraphy Ob Wats.a: vecw.ce lo eates siete cr oe ee 412
ECOLOGY
BROWN, WALTER C., AND ANGEL C. ALCALA. Modes of reproduction of
Philippine anurans. 4000552 8k es ee 416
MEDEM, FEDERICO. Reproductive data on Platemys platycephala
(Testudines: Chelidac)iin Colombiage.-se-eeee eee eee 429
PARKER, FRED. The prehensile-tailed skink (Corucia zebrata) on
Bougainville Island) Papua New, Guinea aero eee eee 435
ANDREWS, ROBIN M., A. STANLEY RAND, AND STELLA GUERRERO. Seasonal
and spatial variation in the annual cycle of a tropical lizard. ...... 44]

RAND, A. S., STELLA GUERRERO, AND ROBIN M. ANDREWS. The ecological
effects of malaria on populations of the lizard Anolis limifrons on Barro

Colorado Island, Panama... we cee eee 455
HERTZ, PAUL E. Eurythermy and niche breadth in West Indian Anolis
lizardsa reappraisal: B... 2. .2eu eee ee ee ee 472

HUEY, RAYMOND B. Natural variation in body temperature and physiological
performance in a lizard (Anolis cristatellus). ...............s+.+-> 484

a

TABLE OF CONTENTS

SCHOENER, THOMAS W., AND AMY SCHOENER. On the voluntary departure of
lizard seiromaveryasmalliaislands; .+...,.5 4005 6.2 54a ene 491

ROUGHGARDEN, JONATHAN, JOHN RUMMEL, AND STEPHEN PACALA.
Experimental evidence of strong present-day competition between the
Anolis populations of the Anguilla Bank—a preliminary report. ...499

MOERMOND, TIMOTHY C. Competition between Anolis and birds: a

MEASS CS SING IM MMR EOS ici See Nii ke stays Mave MIM caphtcne dg ls Palak a Snes me OO
MITTERMEIER, RUSSELL A., AND Marc G. M. VAN ROOSMALEN. A
Symecologicalustudy) of Surinam monkeys, ...........-.4...2++.-4) 521
BEHAVIOR
AUFFENBERG, WALTER. Courtship behavior in Varanus bengalensis (Sauria:
\VATAMICAS). sac 08 Coe Oe ee et nN Rr cece. 55)
JENSSEN, THOMAS A. Display behavior of two Haitian lizards, Anolis cybotes
MENA NOUUSHOUSELCMUS) a. tio i. g)4-s bein o Sa od Gab AEs He bee ee ee ODD:
HICKS, ROBERT A., AND ROBERT L. TRIVERS. The social behavior of Anolis
CCUCTICIOMI oc ah San Oe ee ee ee eee er ee gee 570)
EVOLUTION
SEGER, JON. Conditional relatedness, recombination, and the chromosome
ARUN De TMOLINTISE CRS SiN Ch attic. a iol? wesinuuons yt nimeem aaa ela ae OO
GANS, CARL. Is Sphenodon punctatus a maladapted relict? .......... 613
BLAKE, J. A. Chromosomal C-banding in Anolis grahami. ............ 621

GORMAN, GEORGE C., DONALD G. BUTH, MICHAEL SOULE, AND
S. Y. YANG. The relationships of the Puerto Rican Anolis: electrophoretic
AMGMICATVOLY PICIStUGICS. 6 sobs. dis ale tc Peeled ee ud epee needa de gasnde 626
HALL, WILLIAM P. Modes of speciation and evolution in the sceloporine
iguanid lizards. I. Epistemology of the comparative approach and

MmMKOGuctOnutorthe problem), {os sccce sacs case eoeene sae nmeeate es 643
HOLT, ROBERT D. Immigration and the dynamics of peripheral popula-

(BOING. din dh Sho By SSB oreo Rae Urea a Por eon tReet Teac ear 680
CONSERVATION

PLOTKIN, MARK J., FEDERICO MEDEM, RUSSELL A. MITTERMEIER, AND
ISABEL D. CONSTABLE. Distribution and conservation of the black
canimam (MCUCMOSWIEDUS MHEG), wiosccseccbssacc00mGdooscovccconcas 695

ROSS, JAMES PERRAN, AND JOSE ALBERTO OTTENWALDER. The leatherback
sea turtle, Dermochelys coriacea, nesting in the Dominican

EVE NOU COMME ie er ise dantcer acnosietrnse Mas nahin dln a Ae ec hy theacut aba pease 706

xix

{
f
Cw
Wes
S
" ations #2 oe
Pair. | a 7
+ ® Ane e ; j
i Dip wes a re 1
\ . rae
£ Ay NJ ee | AAR
' j ri, i H ier hakel tay
| et fe. in zy at
re; aig ll AN Clad, Ais a cage Chae
ii te 0 ap eS i,

oe 7
nae ipa wll, een
oe
, = eh ern ee Aad Zainal iat ‘heme, 3
e y(t, 0@ Gra ae oN ne Avwee.a- | :
ral je Res ee ; -
et Mok uw
2 ap rH Mer cae os 7 “a*
'v3a¢ ad eter. Yi

Nyctanolis pernix, A New Genus and
Species of Plethodontid Salamander from
Northwestern Guatemala and

Chiapas, Mexico

PAUL ELIAS!
DAVID B. WAKE?

ABSTRACT. A new _ bolitoglossine salamander,
Nyctanolis pernix, from the Cordillera de los
Cuchumatanes of Guatemala and neighboring
Mexico has been discovered. It differs from all
other neotropical plethodontids in its spotted color
pattern, long legs, and divided premaxilla. The
osteology of Nyctanolis is the most plesiomorphous
of any member of the supergenus Bolitoglossa. Ex-
cept for the apomorphies characterizing the super-
genus, Nyctanolis retains a morphology ancestral to
the entire tribe Bolitoglossini and one more gener-
ally primitive than found in the other supergenera
of the tribe, Batrachoseps and Hydromantes. The
morphological features of the new genus place it as
a sister group to the rest of the supergenus Bolito-
glossa. The apparently ancient lineage represented
by the new form occurs today in the oldest land-
positive area in all of Central America.

INTRODUCTION

The remarkable radiation of the Bolitoglossini
that has occurred in Mexico includes 7 genera
and 63 species as at present understood. ... Be-
yond question other genera and dozens of other
species remain to be discovered. . . .

—Smith and Smith, 1976

Our understanding of the evolutionary ©

history of tropical salamanders is still in

1,2Museum of Vertebrate Zoology, University of
California, Berkeley, California 94720, U.S.A.

its infancy, and undescribed species are
found with regularity. No new genera
have been described for thirty years
(Tanner, 1950), but despite this seeming
stability several of the presumed lineages
are poorly defined (Wake and Lynch,
1976). In the summer of 1974 the senior
author visited a remote area on the east-
ern slopes of the Sierra de los Cuchuma-
tanes, in extreme northwestern Guate-
mala. There, in a limited area, he dis-
covered five species of salamanders liv-
ing in sympatry in a cloud forest at inter-
mediate elevations. Three of these spe-
cies proved to be undescribed. In this
paper we present a description of the
most distinctive of the new species. It is
so unusual in its combination of ancestral
and derived morphology that it requires
its own genus. It is the only tropical
plethodontid that has two premaxillary
bones, and it differs from all other tropi-
cal salamanders as well in its body form
and behavior. This long-legged, large,
colorful salamander, with its long, whip-
like tail, and high level of scansorial ac-
tivity reminds us more of an anoline liz-
ard than a typical salamander.

We name the new genus for its anoline
aspect and nocturnal habits (Gr., nyktos,
night) and the species (L., quick, agile)
for its gymnastic behavior.

2 Advances in Herpetology and Evolutionary Biology

TAXONOMY AND MORPHOLOGY

Nyctanolis new genus

Type species. Nyctanolis pernix sp. nov.

Diagnosis. A plethodontid salamander
belonging to the subfamily Plethodon-
tinae, tribe Bolitoglossini, supergenus
Bolitoglossa; distinguished from all other
members of the supergenus by the pres-
ence of paired premaxillary bones and
extremely long tail, limbs, and digits.
The genus is easily distinguished from
the diminutive species of Parvimolge and
Thorius by its large size, and from the
elongate species of Lineatriton and
Oedipina by its long limbs and large,
broad head. Most species of both Chirop-
terotriton alpha and beta are much smal-
ler, but it is further distinguished from
the former by having a fifth distal tarsal
that is smaller than the fourth, and from
the latter in having a distinct lingual carti-
lage and a well-developed columella. It
differs from Bolitoglossa in having a dis-
tinct sublingual fold, nine tarsal elements,
a well-developed columella, and a lingual
cartilage. It differs further from Bolito-
glossa alpha in having a well-defined
tibial spur, and from Bolitoglossa beta in
having only a slightly constricted tail
base and unspecialized transverse proc-
esses on the first caudal vertebra. Nyct-
anolis most closely resembles species of
Pseudoeurycea, but is distinguished from
that genus by being more extreme in
limb, hand, and foot specializations and
by having a more fully developed colum-
ellar stylus and lingual cartilage, in addi-
tion to its unique premaxillae.

Nyctanolis pernix sp. nov.
Figures 1-4

Holotype. Museum of Vertebrate Zoology (MVZ)
134641, an adult female from Finca Chiblac, 10 km
(air) NE Barillas, Huehuetenango, Guatemala,

(91°16'W, 15°53’N), 1,370 m (4,500 ft) elevation, col-
lected 29 August 1975, by P. Elias and J. Jackson.

Paratypes. MVZ 131583-85, 134639-40, 134642-
44, 149370, 149372-73, 173062 (cleared and
stained), MCZ 100154, same data as holotype but
collected at different times by P. Elias, J. Jackson,
H. B. Shaffer, and A. Diaz. United States National
Museum 206925, cave near stream that empties the
main lake, Lagunas de Montebello, Chiapas, Mexi-
co, (91°32'W, 16°5'N), collected by Scott Belfit, 8
July 1972.

Diagnosis. A large, slender species
(standard length, SL, in four adult males
43.2-68.1, x 54.9; nine adult females
57.8-73.6, x 67.9) with a very long tail
(SL/tail length in three adult males is
0.73-0.84, x 0.79; three adult females
0.80-0.94, x 0.87) and long limbs and di-
gits (when appressed fore and hind limbs
overlap in four adult males by 34 costal
grooves, x 3.6; in eight adult females by
1.54.0 costal grooves, x 2.5). The head is
broad (SL/head width in four adult males
5.8-6.3, x 6.0, in nine adult females 6.0—
6.6, x 6.2) and relatively flat, and contains
large numbers of maxillary-premaxillary
(88-119, x 103 in four adult males; 82—
131, ¥ 110 in nine adult females) and
vomerine teeth (29-47, x 41.5 in four
adult males; 38-53, x 45.7 in nine adult
females). The species has highly distinc-
tive coloration, and typically animals are
shiny dark black with bright red spots on
the eyelids that become red-orange, then
orange, then yellow, and finally cream-
colored posteriorly on the limbs, body,
and tail.

Description. This large and _ long-
legged but slender species has a short
snout. Both males and females grow to a
large size. The nostrils and labial pro-
tuberances are small in both sexes but are
largest in adult males, whose snouts are
generally expanded forward to consider-
ably overhang the mandible. The mental
hedonic gland is large and pronounced in
adult males, attaining a width half that of
the entire head. The head is wide and
flattened, always markedly broader in
dorsal aspect than the broadest point on
the trunk. The ratio of maximum head

width to head depth at the angle of the

.
:

NEW CENTRAL AMERICAN SALAMANDER - Elias and Wake 3

jaw varies in adults from 2.1 to 2.3. A
deep unpigmented groove extends below
the eye, following its curvature, to near
the posterior border of the eye opening,
but does not communicate with the lip. A
marked angle in the line of the mouth
beneath the eye cocks the posteriormost
section of the lip gently ventrad to the
angle of the jaw. The large and promi-
nent eyes bulge upward from the flat-
tened head, especially in males, but do
not extend lateral to the jaw margins. An
indistinct postorbital groove extends pos-
teriorly and gently ventrad from behind
the eye as a shallow depression turning
sharply directly ventrad as a deep crease
which, passing behind the jaw articula-
tion, disappears below the level of the
mandible. No trace of a nuchal groove is
evident, but the gular fold is strongly de-
veloped and arches slightly forward at
the midline. Vomerine teeth increase in
number with increasing size (Table 1)
but there are many even in small indi-
viduals. These teeth are arranged in a
long gently arched row which extends
laterally to far beyond the small internal
nares, nearly to the jaw line. The many
maxillary teeth extend in a long row to
the interior angle of the jaw and slightly
beyond the posterior margin of the eye.
The maxillary teeth are smoothly con-
tinuous with the premaxillary teeth so
that there is no break except for a slight
discontinuity in the largest male. The
premaxillary teeth are numerous (up to
15) but are only slightly sexually differ-
entiated and do not pierce the upper lip
in males. The tongue pad is nearly round
and lies at the end of a pedicel. There is
no anterior attachment, and the tongue
has the boletoid form characteristic of the
supergenus. A large, fleshy sublingual

fold is present. The trunk and tail are

slender and cylindrical in this species
and no strong basal constriction is evi-
dent. The tail may be autotomized at any
point along its length. The tail is rela-
tively long. No postiliac gland is visible.
The limbs are extremely long, and when
fore and hind limbs are stretched along

the flank they overlap extensively. Hands
and feet are large and unwebbed, with
distally expanded, quadrangular digital
tips. The fingers are, in order of decreas-
ing length, 3,4,2,1; the toes, 3,4,2,5,1.
Measurements of the Holotype (in
mm). Head width 11.6; snout to gular fold
(head length) 15.2; head depth at pos-
terior angle of jaw 5.4; eyelid length 5.0;
eyelid width 3.7; anterior rim of orbit to
snout 4.4; horizontal orbit diameter 3.5;
interorbital distance 3.5; distance be-
tween vomerine teeth and parasphenoid
tooth patch 0.5; snout to forelimb 21.1;
distance separating internal nares 3.3;
distance separating external nares 3.4;
snout projection beyond mandible 0.3;
snout to posterior angle of vent (standard
length, SL) 70.5; snout to anterior angle
of vent 64.4; axilla to groin 36.9; tail
length 79.7; tail width at base 4.8; tail
depth at base 5.6; forelimb length (to tip
of longest toe) 21.6; hindlimb length
23.7; width of hand (across from tip of
innermost to tip of outermost toe when
spread) 7.6; width of foot 9.7.
Coloration (in alcohol). All ventral sur-
faces are slate black except palmar and
plantar surfaces and tips of the nasal cirri
which are less saturated with pigment
and thus appear lighter grey. One large
adult male (MVZ 134642) is uniformly
dark black dorsally as well as ventrally,
with the exception of a small, indistinct
spot middorsally. All other specimens
have essentially the same, almost gaudy
coloration. Bright, discrete spots extend
from eyelids to tail tip in these 13 ani-
mals. One of these has been cleared and
stained; variation in 12 other specimens
is as follows. A series of large, round, pale
yellow spots with irregular margins, each
about equal in diameter to the eye, is
present on the black dorsum. These spots
are almost symmetrically arranged. One
spot covers each eyelid in all animals.
One spot covers the dorsal surface of
each elbow in all but one animal, which
has a spot on one side only. All animals
have a large spot covering each knee. A
pair of spots lies over the shoulders in 11

Advances in Herpetology and Evolutionary Biology

-

“ad AQ0[0}{ ,

c'0- €T Os gL 16 oe LG CL G'LZ f VBSTET ZAW
cc SP 60T : 9°81 S61 69 16 CCl 81S Af GLE6PT ZAW
Sica 8€ POT e 661 £06 v8 GO1 Cel 19 4 G9DELT ZAW

2 OF LOT Fi VIG 0°SS 16 60I ec eZ9 A Sc6906 WNSN
0'S— LY cTT TOL 961 8°06 €'8 601 VST 9°19 A €8STET ZAW
Or- 1g 68 & 916 916 88 vO 1 S10 689 A ELE6rPL ZAW
ccs €¢ T€T : 661 GIG 86 VIT ec 6°69 A EPOPET ZAW
(Oi Ss GP al L'6L 916 L8G L'6 OTT GSI SOL A +1 VOPET ZAW
0'G- 67 OIT 6°68 L0G LCG 61 CTI T 91 LUZ A 6E9PET ZAW
SS eV LIT = G'S GSS 66 eC 6ST 9EL A OF9PET ZAW
tom LY Oot : (Bal T 41 Tg 69 86 CEP W OLE6FI ZAW
Or 63 88 €°89 O'9T io) | v9 98 e1l T0s W S8STET ZAW
CO) Ss cV col 69 O'8T 881 we C6 VET 08s W VPOPET ZAW
Oi cV 6IT V'98 9°06 T& £6 STI 891 189 W GPOFET ZAW

[eAroquy = Foquinyy requinyy weueyT wsueyT wsueyT PPM PPM WsueyT ywwusyT xog
quiry Woo], YOO], [RL quiry quiry 004 peeH peeH prepurys
SULIDUIO A Arle] [IxXeuloig 3104 pulpy
pue Arey Rey

“XINYUAd STIONVLOAN: NI NOILVIYVA “[T AIEAV

“LpOVEL ZAW ‘xiusad sjouejoAN jo adAyojoy 9} JO MAIA JesIOgq “| anbi4

WO e

NEW CENTRAL AMERICAN SALAMANDER °: Elias and Wake 5

animals, and the 12th has a single shoul-
der spot. Many individuals have one spot
over the pelvis (five have one, one has
two, and six are unspotted). Most indi-
viduals (nine) have some sort of spot,
variable in size and position, in the cau-
dosacral region. Full-sized (i.e., the size
of the eyelids) dorsal spots on the trunk
between shoulder and pelvis occur in all
individuals, and range in number from
two to eight. Included among the trunk
spots of four specimens is a pair placed
symmetrically behind the shoulder.
Three animals show variable numbers of
small, obscure spots on the trunk. Most
individuals lack complete tails, but large
spots occur only on the proximal one
third of the tail and the first third is com-
plete in 11 specimens. Of these 11, all
have some distinct tail spots, ranging
from one to eight in number. In addition
four of the seven with over one-half the
tail present show clusters of up to nine
small indistinct spots on the second half
of the tail. Miscellaneous small spots oc-
cur in many animals: seven have spots on
the temple; two have parietal spots; two
have paired wrist spots; one has a foreleg
spot; one has a flank spot; one has a man-
dibular spot.

Coloration (in life). These animals
were pattemed as they are in preserva-
tion, but the coloration of the dorsal spot-
ting differed. The spots on the eyelids
and neck are a deep crimson, those on the
elbows and knees are orange, and the
trunk and tail spots are light yellow, be-
coming cream at the tail tip.

Habitat. The series from the type lo-
cality was collected in a large clearing
surrounded by virgin cloud forest. The
locality is extremely humid as indicated

by rainfall records taken within 5 km of .

the spot. These records show rainfall of 5
to 6 m annually, but they were taken be-
low the cloudline and thus may under-
estimate precipitation at the type locality.

The animals at the type locality were
collected under both moss and bark on
felled cloud forest hardwoods. They
were active on rainy nights. Rarely were

more than two individuals taken per
man-day collecting.

Also occurring at the site were four
other salamanders: Bolitoglossa hart-
wegi, Bolitoglossa cuchumatana, a new
species of Bolitoglossa (Elias, in prepara-
tion), and another undescribed species
belonging to an as yet undetermined
genus of plethodontid salamanders. The
three Bolitoglossa were taken in the
same microhabitats as those in which
Nyctanolis was found, but the disturbed
nature of the habitat leaves ecological re-
lationships vague.

Various hylid and leptodactylid frogs
were also taken at the site as was one
snake (Leptodeira) and lizards of the
genera Anolis, Sceloporus, Scincella,
Barisia, and Lepidophyma.

Range. The species is known from the
type locality and one other locality about
20 km NW of Finca Chiblac just across
the border in Chiapas, Mexico.

Osteology. Information conceming
osteology has been derived from one
cleared and stained specimen (MVZ
173062), an adult female, and from radio-
graphs of all of the specimens. Dorsal and
ventral views of the skull are illustrated
(Fig. 2).

The relatively broad and low skull is
compact and well ossified, with well-
organized articulations and well-defined
bones. Premaxillary bones are distinctly
separate in all individuals in which they
can be seen; the smallest individual is
insufficiently ossified for the bones to be
resolved in radiographs. The premaxil-
lary bones are large, relative to the small,
fused bones characteristic of other mem-
bers of the supergenus Bolitoglossa.
They are well articulated with each
other, and are tightly articulated laterally
to the maxillaries, which dorsally overlap
the dental portions of the premaxillaries.
The premaxillaries of both males and fe-
males are part of a continuous maxillary-
premaxillary arcade, rather than being
discontinuously projected anteriorly as in
many other tropical salamanders. The
palatal process of each premaxillary bone

6 Advances in Herpetology and Evolutionary Biology

Figure 2. Dorsal (left) and ventral (right) views of the skull of Nyctanolis pernix, drawn from cleared and stained

specimen with aid of microprojector. Cartilage stippled.

Abbreviations: Fr, frontal; Ma, maxillary; Op, operculum; Os, orbitosphenoid; Ot, otic-occipital; Pa, parietal; Pf,
prefrontal; Pm, premaxillary; Ps, parasphenoid, Pt, pterygoid portion of palatopterygoid cartilage; Pv, posterior
vomerine tooth patch (outline only, teeth not drawn); Qa, quadrate, Sq, squamosal; Vo, vomer.

is narrow and well separated from the
vomer behind it. Frontal processes of the
premaxillaries are stout and well de-
veloped; they arise from a narrow base as
columnar structures, then immediately
diverge around the large, internasal fon-
tanelle. As the processes rise dorsally and
posteriorly they flatten and form the bor-
ders of the fontanelle. As they flatten
they become somewhat expanded. Later-
ally the processes closely abut the large
nasals, and posteriorly they broadly over-
lap the anterior extensions of the facial
portions of the frontals. The frontal proc-
esses do not contact each other behind
the fontanelle, but in some individuals
they come very close. The frontal proc-
esses terminate rather far forward, well in
advance of the orbit, and of the posterior
tip of the nasals. Nasals are large, but
they are nonprotuberant in all females
and in all but the largest male; in that

individual the premaxillary bones are
larger than in any other specimen and the
nasal bones extend anteriorly slightly
beyond the premaxillaries. The large
nasals are triangular in shape, and they
have strong, overlapping articulations
with the facial processes of the frontals.
The nasals slightly overlap the prefron-
tals, and are overlapped by the anterior
part of the facial processes of the maxil-
laries. The nasals are not domed, but are
rather flattened or slightly concave. No
septomaxillaries can be seen in any spe-
cimen (a slight shadow on one side of the
skull in radiographs of one of the smaller
specimens might conceivably be a sep-
tomaxillary). Prefrontal bones are dis-
crete but relatively small. The bones are
only about one-quarter or less the area of
the nasals. The prefrontals broadly over-
lap the facial parts of the frontals, and are
overlapped rather narrowly by the pos-

NEW CENTRAL AMERICAN SALAMANDER - Elias and Wake 7

terior and dorsalmost parts of the facial
processes of the maxillaries. The foramen
for the nasolacrimal duct lies at the ex-
treme anterolateral tip of the prefrontals
and is indented into the dorsal margin of
the facial processes of the maxillaries.
There is a slight indentation of the pos-
terolateral border of the nasals as well.
The nasolacrimal duct proceeds from the
orbit across the lateral parts of the pre-
frontals, which are depressed and almost
grooved. The maxillary bones are very
large and well developed. The bones ex-
tend posteriorly to terminate in sharp
points at the level of the posterior margin
of the eye. The maxillaries bear small but
well-developed, bicuspid teeth from the
premaxillary nearly to the posterior tip;
these teeth are about the same size as the
premaxillary teeth in females, but are
somewhat smaller in males. The facial
processes of the maxillaries are relatively
very large and well developed; they are
larger in area than the nasals. The palatal
processes are moderately well developed
and extend as shelflike processes to the
margins of the vomer bodies, against
which they tightly abut.

The vomers are large bones which are
well separated from each other by the
intervomerine fontanelle anteriorly. This
fontanelle contains a large glandular
mass, present also in other plethodontid
salamanders but frequently not so well
developed. The vomers abut dorsally and
posteriorly to this mass, and the posterior
parts of the two vomers are tightly articu-
lated. The preorbital processes of the
vomers are long and stout; they extend
beyond the lateral margins of the vomer
bodies, and bear teeth in a long, curving
series nearly to their tips, well beyond
the lateral border of the internal nares.

Frontals and parietals are well de-
veloped and firmly articulated with each
other in an extensive interlocking con-
tact. The paired elements are tightly arti-
culated to each other along the midline as
well. Anterior parts of the frontals are
moderately large and contribute impor-
tantly to the facial part of the skull. The

overlapping posterior lobes of the fron-
tals are relatively very large. The frontals
appear to be relatively narrow, but we
believe that this is simply an illusion that
results from the unusual (for a tropical
salamander) full development of the
outer bones of the skull. Parietals are
well developed, but they do not have the
distinct lateral spurs that are character-
istically present (but sometimes very
poorly developed) in tropical sala-
manders (Wake, 1966). The region of the
lateral spur is somewhat expanded and
ventrally directed, but is not spurlike.
The otic capsules are large and well de-
veloped, but they do not appear to be
relatively as large as in other tropical
genera, in which the other portions of the
skull are frequently reduced in size. Two
or three spinelike projections arise from
the anterodorsal parts of the capsules,
and are directed laterally; these form a
kind of rudimentary crest. The large
parasphenoid is narrow anteriorly, but
the orbitosphenoids are well separated
from each other. The tip of the para-
sphenoid is blunt. Posterior vomerine
teeth are bome in long, relatively narrow
patches that are well separated from each
other and from the anterior vomerine
teeth. The teeth in these patches are
somewhat smaller than the maxillary
teeth. In the cleared and stained speci-
men there are 81 (left) and 92 (right)
bicuspid, ankylosed teeth. The oper-
culum has a small but well-developed
stilus that is as large as this structure ever
becomes in any of the other tropical
genera (it is absent in the vast majority of
tropical species). Well-developed, stout
quadrates are connected to the otic cap-
sules by the very large, broad squamo-

-sals, and by the cartilaginous suspen-

sorium.

The lower jaw has a long, slender den-
tary bearing a very long series of small
teeth, about the size of those on the max-
illary. The prearticular is relatively small
and low.

The hyobranchial apparatus (Fig. 3) is
very generalized for the supergenus Boli-

8 Advances in Herpetology and Evolutionary Biology

toglossa (cf. Tanner, 1952; Lombard and
Wake, 1977). It is typical of the group;
there is no urohyal, and the radii are es-
sentially continuous with the basibran-
chial. There is a well-developed lingual
cartilage, but it is represented by what is
essentially a direct anterior continuation
of the basibranchial, although there is a
change from hyaline to fibrocartilage in
the joint region between the cartilage
and the basibranchial. We have also had
available one sectioned head of the spe-
cies, and detailed examination of it re-
veals that in all respects the hyobranchial
apparatus and associated musculature fits
the morphological features of Mode VI of
Lombard and Wake (1977). Specifically,
the entire apparatus is greatly elongated,
and the proportions of the elements are
unequal. The epibranchials are very
long, more than three times the length of
the basibranchial (minus lingual carti-
lage), and about six times the length of
the second ceratobranchials. The dia-
meters of the basibranchial, second cera-
tobranchial, and base of the epibranchial
are about equal, and all are greater than
that of the relatively slender first cerato-
branchial. The first ceratobranchial is not
so slender as is characterisitc of that ele-
ment in many members of the super-
genus Bolitoglossa. The radii are of
moderate length for a member of the
supergenus, but are relatively stout. The
ceratobranchials are relatively long, but
generalized in form; there is a distinct,
but narrow, flattened blade distally, and
there is no anterior filament. The point of
attachment of the suprapeduncularis
muscle is drawn medially into a broad-
based process.

Vertebral structure is generalized. In-
dividual vertebrae have centra with hol-
low, tapered, bony husks, and are joined
to one another by the spindlelike inter-
vertebral cartilages and the zygapoph-
yses. There is a single atlas, 14 trunk
vertebrae, one sacral vertebra, two cau-
dosacral vertebrae, and a variable num-
ber of caudal vertebrae. The trunk verte-
brae all have transverse processes that

Figure 3. Dorsal view of the cartilaginous hyo-
branchial apparatus of Nyctanolis pernix.

Abbreviations: Bb, basibranchial; CbI, first cerato-
branchial; CbII, second ceratobranchial: Ch, cerato-
hyal; Eb, epibranchial.

arise and remain separated until their
tips. These processes are relatively long,
and extend well beyond the limits of the
zygapophyses. The upper transverse
process is a little anterior in placement
relative to the lower one. Relatively long,
slender ribs are borne on all but the last
trunk vertebra; rib heads are relatively
widely separated. Spinal nerve routes are
similar to those reported by Edwards
(1976) for this group: a ventral root issues
from the neural arch of the atlas; dorsal
and ventral roots issue from separate
foramina in the anterior, pre-transverse
process part of the neural arch of the first
trunk vertebra; there is a very large fora-
men in the anterior, pre-transverse proc-
ess part of the neural arch of the second
trunk vertebra, and a set of dorsal and

NEW CENTRAL AMERICAN SALAMANDER °- Elias and Wake 9

ventral roots emerge through a common,
small foramen in the posterior, post-
transverse process part of the neural arch
of the same vertebra; dorsal and ventral
roots emerge through a common, small
foramen in the posterior, post-transverse
process part of the third and all succeed-
ing trunk vertebrae.

The first caudosacral vertebra has rela-
tively long, nearly straight transverse
processes. The second is a smaller verte-
bra, with processes that are much shorter
than those of the first; these processes
may be straight, but are more frequently
oriented anteriorly. The origin of the
processes on the first caudosacral verte-
bra is about midcentral, but those of the
second vertebra arise in advance of the
midpoint of the centrum. The second
caudosacral vertebra bears rudiments of
the haemal arch, not joined to each other,
on the posterior part of the ventral sur-
face of the centrum. The foramen for the
spinal nerves is located behind the pro-
cesses, about midcentrally.

The tail base region is not nearly as
specialized for tail autotomy as in most
other members of the supergenus (Wake
and Dresner, 1967), but fundamentally it
qualifies as a constricted-based tail even
if the constriction is only slight. The first
caudal vertebra is slightly, but not ob-
viously, shorter than the second. It differs
from the second caudal in having a com-
plete haemal arch that lacks an anterior
keel. It differs from the second caudo-
sacral immediately in front of it in having
transverse processes that arise at the very
anterior end of the vertebra, essentially
as extensions of the bases of the prezyga-
pophyses. The transverse processes are
short, and extend only a short distance
anterior and lateral to the edge of the
prezygapophyses. The musculature ex-
tending between the first caudal and the
second caudosacral vertebrae is visibly
shortened, relative to adjacent segments,
but the tail is slender and the shortened
segment is not obviously constricted.
The smallest juvenile lost its tail at the
base during capture, between the second

caudosacral and the first caudal verte-
brae. The skin of the shortened segment
has collapsed over the wound in the
preserved animal, in the typical wound-
healing specialization that characterizes
salamanders with constricted and slen-
der-based tails (Wake and Dresner,
1967).

Complete tails are rare. The only large
specimen with a complete tail has 39
caudal vertebrae, a very large number for
the supergenus, with the exception of the
exceptionally elongate species of Linea-
triton and Oedipina. Other large speci-
mens show signs of earlier breaks, always
away from the basal region. We cannot
rule out tail base breaks in juveniles,
with subsequent regeneration. One spe-
cimen has a tail broken at vertebra 39.
Another has a total of 45 vertebrae, re-
generated from vertebra 24. There are 49
vertebrae in the tail of another specimen,
regenerated from vertebra 24. The holo-
type has 34 caudal vertebrae, but its tip is
missing; further, it shows signs of an ini-
tial break at vertebra 18, and a subse-
quent break at vertebra 27. The caudal
vertebrae are narrow and elongate; trans-
verse processes are very short and are
little more than small projections of the
prezygapophyses. The spinal foramina
are located midcentrally.

The hands and feet are highly distinc-
tive (Fig. 4). The digits are very long,
with terminal expansion. All phalangeal
elements are well developed. The
phalangeal formula is 1, 2, 3, 2 for the
hand and 1, 2, 3, 3, 2 for the foot; this is
the primitive formula for the supergenus
Bolitoglossa. The long metapodial ele-
ments are cylindrical, as are all but the
terminal phalanges. Terminal phalanges
are specialized, and resemble those of

' Aneides lugubris (Wake, 1963); they are

greatly expanded distally, and the ex-
panded portion is slightly recurved and
nearly bifurcated in the larger speci-
mens. The expanded portion of each
terminal phalanx is distinctly flattened.
On the ventral surface of each phalanx,
near the base, a well-defined projection is

10 Advances in Herpetology and Evolutionary Biology

present which serves as the site of inser-
tion of a digital tendon. Even the phalanx
of the first digit displays some distal ex-
pansion.

The carpal and tarsal arrangements are
those characteristic of primitive pletho-
dontids. There are eight carpals and nine
tarsals. Distal tarsal 5 is smaller than dis-
tal tarsal 4 and does not articulate with
the centrale.

Limb bones are very elongate, slender,
and cylindrical. There is a well-defined,
distinct tibial spur near the proximal end
of the tibia, and the bone has a sharply
defined crest distally.

. Relationships. The tribe Bolitoglossini
is characterized by seven unique apo-
morphies relative to all other plethodon-
tids. All members of the tribe have 1) lost
the urohyal; 2) radii fused to the basi-
branchial; 3) long epibranchials relative
to the ceratobranchials; 4) a second cera-
tobranchial modified for force transmis-

Figure 4. Dorsal view of the left foot of Nyctanolis
pernix, cartilage stippled.

sion; 5) a cylindrical muscle complex
around the tongue; 6) lost the circum-
glossus muscle (all discussed in Lombard
and Wake, 1977); and 7) a juvenile otic
capsule configuration (Lombard, 1977).

The supergenus Bolitoglossa is charac-
terized by having a specialized autotomy
area in the tail base that follows two
caudosacral vertebrae (Wake and Dres-
ner, 1967), and by some modifications of
the throat and tongue musculature (Lom-
bard and Wake, 1977).

The morphology of Nyctanolis in-
cludes all of the unique apomorphies
characteristic of both the tribe Bolito-
glossini and the supergenus Bolitoglossa
and we therefore consider it to be a
member of both taxa.

In most of its characters Nyctanolis is
generalized within the supergenus Boli-
toglossa and is similar to generalized
members of related genera. Because of its
generalized structure the phylogenetic
position of the new genus relative to the
other members of the supergenus is sug-
gested by only two characters: absence of
septomaxillary bones, and presence of
paired premaxillaries. The absence of
septomaxillary bones from most speci-
mens of Nyctanolis (one of fourteen may
have one septomaxillary) is a derived
state that would place the new genus
somewhere among the other genera.
However, the erratic occurrence of these
bones in species in which they are typi-
cally found (Wake, 1966), combined with
the possible occurrence of septomaxilla-
ries in one Nyctanolis specimen, weak-
ens the value of this character. In having
paired premaxillaries Nyctanolis is more
primitive than any other member of the
supergenus Bolitoglossa. Among neo-
tropical salamanders only Nyctanolis has
the premaxillary divided. This division is
occasionally seen in an individual or two
of some other species (Wake, 1966), but
nowhere else is it characteristic of a
taxon. This plesiomorphy argues strongly
for the cladistic isolation of Nyctanolis;
the new genus appears to be a sister

NEW CENTRAL AMERICAN SALAMANDER: Elias and Wake 11

group to the rest of the supergenus (Fig.
5).

In many respects Nyctanolis preserves
more plesiomorphic character states than
any other member of the tribe Bolito-
glossini; only those characters which di-
agnose the supergenus Bolitoglossa dis-
tinguish Nyctanolis from the ancestral
state of all bolitoglossines. Members of
the genus Hydromantes have the most
primitive tail base region in the tribe, and
have paired premaxillaries and well-
developed septomaxillaries; however,
they have a host of apomorphous charac-
ters, including a highly specialized pro-
jectile tongue and associated features,
and they have lost prefrontals (Wake,
1966; Lombard and Wake, 1977). The
species of Batrachoseps mainly have
apomorphous characters, including only
four toes, widely separated frontals and

other genera
of supergenus
Bolitoglossa NY

Figure 5. Cladogram showing the relationships of the
members of the tribe Bolitoglossini. Principal syna-
pomorphies characterizing each branch are as follows:
a. Tribe Bolitoglossini, seven apomorphies discussed
in text. b. Radii lost. c. Reduction in number of caudo-
sacral vertebrae to two (three are still found in certain
Batrachoseps but in these same species animals with
two are common). d. Toe number reduced to four. e.
Supergenus Bolitoglossa, tail base with complex of
breakage specializations. f. See generic and species
diagnoses in text. g. Characterized within the super-
genus Bolitoglossa by fused premaxillae.

parietals, and a strongly developed la-
teral parietal spur. All Batrachoseps are
plesiomorphic relative to the other boli-
toglossine supergenera in retaining
genioglossus muscles as tongue retrac-
tors (although in a highly specialized
form, see Lombard and Wake, 1977).
While most Batrachoseps lack prefron-
tals and have but a single premaxillary, B.
wrighti gains prefrontals and divided
premaxillaries when it gets very large,
and B. campi has these characters almost
from the time of hatching (Marlow,
Brode, and Wake, 1979). Batrachoseps
has two or three caudosacral vertebrae
and a wound-healing specialization, and
in these respects is more apomorphic
than Hydromantes but less than super-
genus Bolitoglossa, all members of
which have two caudosacral vertebrae
and provision for autotomy of the tail at
the base. In fact, only in this respect, and
in the absence of well-developed septo-
maxillaries, is Nyctanolis less plesio-
morphic than B. campi.

A feature uniting Batrachoseps and the
supergenus Bolitoglossa is the reduction
in number of pairs of diploid chromo-
somes from 14 to 13; unfortunately, we
have no chromosomal information for
Nyctanolis.

Nyctanolis probably represents the
earliest surviving offshoot from the an-
cient ancestral stock of the supergenus
Bolitoglossa. From a biogeographic per-
spective this is especially interesting, for
this ancient group occupies not the nor-
thernmost part of the tropical distribution
of bolitoglossines, where the most gen-
eralized forms have been found previous-
ly, but the eastern slopes of the moun-
tainous core of Nuclear Central America.
This area is one of three foci of evolu-

‘tionary activity in tropical salamanders,

and geographic isolates associated with it
are thought to have been important in the
adaptive radiation of tropical pletho-
dontids (Wake and Lynch, 1976). The
area in which Nyctanolis survives, now
in a highly specialized ecological and

12. Advances in Herpetology and Evolutionary Biology

behavioral form, is the oldest land-
positive area in all of Central America
(Wake and Lynch, 1976; Rosen, 1978).

ACKNOWLEDGMENTS

Jeremy L. Jackson and H. Bradley
Shaffer provided assistance and com-
panionship during the field work of the
senior author in Guatemala. We thank
James F. Lynch, James Hanken, and
Arden H. Brame, Jr. for review of the
manuscript, and Gene M. Christman for
preparing the illustrations. We appreci-
ate the assistance of officials representing
INAFOR for permission to conduct field
work in Guatemala. The research has
been supported by the National Science
Foundation (current grant DEB-78
03008), the Museum of Vertebrate
Zoology, and the National Institutes of
Health (training grant 5-T32-GM07117).

For many years the work of Emest E.
Williams on the evolution of anoline liz-
ards has been an inspiration for our work
on the evolution of tropical salamanders.
It is a privilege to be able to describe this
anoline-like salamander in a volume
dedicated to Professor Williams.

ADDENDUM

Since preparation of this paper,
Nyctanolis pernix has been discovered at
a third locality. Jonathan Campbell has
collected one juvenile specimen (KU
189586) 2.4 mi. SE Purulha, Depto. Baja
Verapaz, Guatemala, 1,615 m elevation.

This is approximately 125 km ESE of the
type locality (by air).

LITERATURE CITED

EDWARDS, J. L. 1976. Spinal nerves and their bear-
ing on salamander phylogeny. J. Morphol.,
148: 305-327.

LOMBARD, R. E. 1977. Comparative morphology of
the ‘inner ear of salamanders (Caudata: Am-
phibia). Contrib. Vert. Evol., 2: viii + 140 pp.

LOMBARD, R. E., AND D. B. WAKE. 1977. Tongue
evolution in the lungless salamanders, family
Plethodontidae. J. Morphol., 153: 39-79.

MARLOw, R. W., J. M. BRODE, AND D. B. WAKE. 1979.
A new salamander, genus Batrachoseps, from
the Inyo mountains of California, with a dis-
cussion of relationships in the genus. Contrib.
Sci. Nat. Hist. Los Angeles Co., 308: 1-17.

ROSEN, D. E. 1978. Vicariant patterns and historical
explanation in biogeography. Syst. Zool., 27:
159-188.

SMITH, H. M., AND R. B. SMITH. 1976. Synopsis of
the herpetofauna of Mexico, Vol. IV. Source
analysis and index for Mexican amphibians.
North Bennington, Vermont, John Johnson.

TANNER, W. W. 1950. A new genus of plethodontid
salamander from Mexico. Great Basin Nat., 10:
37-44.

___. 1952. A comparative study of the throat mus-
culature of the Plethodontidae of Mexico and
Central America. Univ. Kansas Sci. Bull., 34,
pt. 2(10): 583-677.

WAKE, D. B. 1963. Comparative osteology of the
plethodontid salamander genus Aneides. J.
Morphol., 113: 77-118.

____. 1966. Comparative osteology and evolution of
the lungless salamanders, family Plethodonti-
dae. Mem. So. Calif. Acad. Sci., 4: 1-111.

Wake, D. B., AND I. G. DRESNER. 1967. Functional
morphology and evolution of tail autotomy in
salamanders. J. Morphol., 122: 265-306.

WAKE, D. B., AND J. F. LYNCH. 1976. The distribu-
tion, ecology, and evolutionary history of
plethodontid salamanders in tropical America.
Nat. Hist. Mus. Los Angeles Co., Sci. Bull., 25:
1-65.

Larvae of Southeast Asian Species of
Leptobrachium and Leptobrachella (Anura:

Pelobatidae)
ROBERT F. INGER!

ABSTRACT. Seven larval forms of Leptobrachium
Tschudi and Leptobrachella Smith (Anura: Pelo-
batidae) are identified from Southeast Asia. In
terms of their buccopharyngeal and external an-
atomy, they are divisible into three groups that cor-
respond to taxa proposed by Dubois (1980) on the
basis of adult morphology: Leptobrachium (Lepto-
brachium), Leptobrachium (Leptolalax), and
Leptobrachella. Larvae of Leptobrachium (Lepto-
brachium have spheroidal bodies, numerous rows
of denticles on both lips, no more than four pairs of
papillae medially on the ventral velum, and well-
developed glandular zones in front of the dorsal
velum. All of these larvae live exposed in pools of
small streams. Larvae of Leptobrachium (Leptola-
lax) have slender bodies, fewer rows of denticles,
more than eight pairs of papillae medially on the
ventral velum, no evident spicules, and reduced or
absent glandular zones. Larvae of Leptobrachella
mjobergi, the only larval form of this genus known,
are like the preceding group in velar papillae and
spicules and in reduction of the glandular zone.
However, they are more slender, completely lack
denticles, have only a single gill cavity on each side
and lack ruffled filter rows. Larvae of the last two
groups live in stronger currents than the first, and at
least three of them are known to use interstices of
stream bottoms. Limited analysis of food revealed
no association with morphology or habitat.

INTRODUCTION

Larvae of the pelobatid genus Lepto-
brachium constitute a conspicuous por-
tion of the tadpole assemblages in small
streams of Southeast Asia and the Greater
Sunda Islands. Most occur in relatively

1Field Museum of Natural History, Chicago, IIli-
nois 60605, U.S.A.

quiet reaches of these streams (see for
example Smith, 1917; Inger, 1966) and
have the slightly depressed, spheroidal
bodies we have come to associate with
tadpoles inhabiting such environments.
Others having more elongate bodies live
in currents (Smith, 1917) and in the in-
terstices of gravel on the bottom of riffles
(Inger, 1966). The latter group of larval
forms belong to species that Dubois
(1980) has assigned to the subgenus
Leptolalax, whereas the first group be-
long to species of the subgenus Lepto-
brachium (sensu Dubois, 1980). The two
groups of larvae also differ in the number
of rows of denticles and, as will be shown
below, in features of the buccopharyn-
geal cavity. This range of variation in
body form and denticular development is
extended by Bomean larvae assigned
here to Leptobrachella, which are even
more elongate than larval Leptolalax and
completely lack denticles. With material
assembled for this study, eight larval
forms are now known from the area south
of China (see Table 1).

Although descriptions of external fea-
tures of some of these larvae have ap-

- peared elsewhere (e.g., Smith, 1917; Tay-

lor, 1962; Inger, 1966), the literature does
not contain information on any given set
of characters for all these larval forms.
This paper attempts to correct this defi-
ciency. Descriptions of the bucco-
pharyngeal cavity in two larval forms of
Leptobrachium (Wassersug, 1980) sug-
gested a new source of taxonomic in-

14

Advances in Herpetology and Evolutionary Biology

TABLE 1. EXTERNAL FEATURES OF SOUTHEAST ASIAN LARVAE OF SPECIES OF LEPTOBRACHIUM AND LEPTO-

BRACHELLA.
Leptobrachium Leptobrachella
(Leptobrachium) (Leptolalax)
srt n
E 5 a 3S
2) oo fe) i
Bea Gi te ge rae re S e
ee Wiis ME eu ee is % 2
a 8 h Ss -e a 3 =
See eS Sy bb a &
Body shape
wide and deep, width = 0.60 HBL sei ciel Dehic Mata
slender, width < 0.55 HBL + + +
Oral disk
only upper lip notched + + 4+ ++ +
both lips notched + + +
number of divided rows of denticles
upper lip = 5 Se eee ae Paes
upper lip < 5 pe +
lower lip = 4 + + + + ++
lower lip < 4 + +
denticles absent +
denticles clawlike + + + + «+ + +
marginal papillae of upper lip
bases confluent + + +
bases separated + + + + +
inframarginal papillae, lower lip
present, with denticles a, ane
present, no denticles at + ae ate
Narial rim
one small middorsal projection + + + ++ + +
2-3 small middorsal projections +
6-9 small middorsal projections ate
4 subequal lobes around margin +
Spiracular tube
no free portion + + + +
free at tip + + +
free portion 2X opening diam. 4
Glandular patches
present sF ar +
absent + + + + +

*Exceeded in 1 specimen.

formation, which has been pursued in
this study. In addition, a start has been
made on analyzing food of some of these
larvae with the ultimate goal, not yet re-
alized, of relating diets to morphology
and habitat distribution.

It is a pleasure to dedicate this paper to
Dr. Ernest E. Williams who, despite his
refusal to work on tadpoles, has done so
much to stimulate study of morphology of
reptiles and amphibians.

MATERIALS AND METHODS

Descriptions are based upon speci-
mens preserved in ethanol and formalin.
Most are from the collections of Field
Museum of Natural History (FMNH),
though a few are from the Bernice P.
Bishop Museum (BBM) and the United
States National Museum of Natural His-
tory (USNM).

Complete descriptions are given for

one form of each of the three main types
of larvae: Leptobrachium montanum,
Leptobrachium gracilis, and Lepto-
brachella mjobergi. Descriptions of other
forms are abbreviated, emphasizing

SE ASIAN PELOBATID LARVAE: Inger 15

points of difference from the respective
basic type. Information on all is summar-
ized in Tables 1-3.

Body proportions, except where noted,
are given in terms of head-body length

TABLE 2. VENTRAL SURFACES OF BUCCOPHARYNGEAL CAVITY OF SOUTHEAST ASIAN LARVAE OF SPECIES OF

LEPTOBRACHIUM AND LEPTOBRACHELLA.

Leptobrachium Leptobrachella
(Leptobrachium) (Leptolalax)
(S| nN
SI 2 a z
ae BE Be ie = »
Bae ee en 88 3 S
2 a o 3 Q, < oO =
& a a a i) a, =|
Infralabial papillae
anterior—large, transverse + + + + ++ + + +
posterior—2 or 3 branches + +
—4 or more branches Sr a ee +
Lingual papillae (paired)
present ar ar ar ar + + AP
absent +
Buccal floor arena
long lateral papillae + + + + ++ + + +
papillae across posterior
present ar =P oF
absent + + + + +
central area
30 or more pustules + + + + + + +
papillae present + + +
Buccal pocket area
pre-pocket pustules or papillae
present =F ar + ale aF
absent + + +
post-pocket pustules or papillae
present =F sF =F sF +
absent * at a
Ventral velum
pustules, dorsal surface
2 groups a AF
1 group, > 20 + + + + +
1 group, < 20 ar
spicules, present + + + + ++
absent 1 1 ar
median papillae, 2-4 pairs a) ee oe pre ei
8 or more pairs ar ar ar
secretory pits, dorsal surface
on median papillae ae ee
on margin ar + + +
absent ate ate ala
gill cavities, 3 + + + + + + +
1 se
gill filter rows, present jee: Saag Ge + +P
absent ar

*Condition obscure in dissection.

16

Advances in Herpetology and Evolutionary Biology

TABLE 3. DORSAL SURFACES OF BUCCOPHARYNGEAL CAVITY OF SOUTHEAST ASIAN LARVAE OF SPECIES OF

LEPTOBRACHIUM AND LEPTOBRACHELLA.

Leptobrachium Leptobrachella
(Leptobrachium) (Leptolalax)
om n
S| 8 > 3
5 fe a
fee fe 2 s 5
¢ o as) = =) Oo e) Q
Ses eS S 3 2
eS 2g - & & Sb oy &
Prenarial papillae
present al + =P =F =F + sF
absent +
Narial walls, papillae
each with one papilla Ce ee ee +
one or both without papilla + +
Postnarial papillae
more than 2 pairs + + + ++ «+ + +
pustules only ate
Lateral ridge papillae
more than 2 branches + + + ++ +4 + +
only 2 branches at
Median ridge
semilunar + +
triangular + + +
rectangular +
divided to base at a
Buccal roof arena, central portion
with papillae + + + + +
with pustules + + + ++ «+ +
Glandular zone
extends across roof + + + + «+4
restricted, medial or lateral + =F
not evident +
Dorsal velum, papillae
fewer than 15 + + + + + +
more than 15 + +

(HBL). Terminology and sequence of
description of internal buccopharyngeal
anatomy follow Wassersug (1980).

TAXONOMY AND MORPHOLOGY

Leptobrachium montanum Fitzinger

Dr. Julian Dring (in litt.) has argued
convincingly that the populations in
Borneo and Thailand assigned to hasselti
differ sufficiently from topotypic (i.e.,
Javan) hasselti Tschudi to warrant full

specific status as L. montanum Fitzinger
and L. pullum Smith, respectively.

Larval material examined: Sabah—BBM 7362,
7365; FMNH_ 63531-41, 77220-22, 130857-59,
131244, 140216. Sarawak—FMNH 77517, 83017-18,
96022, 121858, 139375-80, 148286, 212388-91.

EXTERNAL FEATURES

Body oval, widest at level of spiracle;

maximum width about 0.60 of length;

somewhat flattened above, rounded be-
low.

Eyes dorsolateral, not visible from be-
low, oriented laterally, line connecting

pupils about 0.20 of HBL behind snout;
length of eye 0.09; interorbital 0.27.
Nostrils open, narial rim low with one
simple weak projection middorsally; line
connecting nostrils 0.10 of HBL behind
snout; internarial 0.20; eye-nostril 0.11.

Oral disk ventral, subterminal, width
0.23 of HBL; lips continuous around
margin except for narrow notch in center
of upper lip; thick, homogeneous papil-
lae in single continuous row around en-
tire margin, bases of papillae contiguous;
five to seven elongate, inframarginal
ridges in diagonal row lateral to lower
rows of denticles; largest of inframarginal
ridges with small denticles. Dental for-
mula in larvae state 26 or older usually
1:6+6/5+5:1 or 1:5+-5/5+5:I, rarely more
than six divided rows on upper lip or
fewer than five on lower; undivided rows
short, that of upper lip only half length of
outer lower row; denticles laterally com-
pressed, curved and pointed; dental
ridges high. Beaks very heavy, fully pig-
mented, coarsely serrated; median cusp
usually larger than adjacent ones.

Spiracle sinistral, slightly above mid-
lateral line, tube fused to body wall dor-
sally and ventrally its entire length;
opening 0.45 of HBL behind snout. Anus
dextral, tube attached to ventral fin.

Tail 1.50 to 1.75 of HBL; weakly con-
vex above and below, tapering to broadly
rounded tip; maximum depth just behind
mid-length, 0.52 of HBL, 0.27 of tail
length. Caudal muscle heavy, 0.33 of
HBL at end of body, 0.20 at maximum tail
depth. Dorsal fin origin at end of body;
fleshy in anterior fourth; margin rising
only slightly; maximum depth subequal
to muscle at point of maximum tail depth.
Ventral fin not as deep as dorsal except
near tip.

Head and body without asperities or
spines. Lateral line system visible on HB
and tail. Three whitish, subcircular glan-
dular areas ventrally on each side of HB;
each patch about size of pupil; two close
together ventrolaterally behind level of
eyes; third about twice its diameter an-
terolateral to base of anal tube.

SE ASIAN PELOBATID LARVAE: Inger 17

Young larvae (i.e., stage 25) light with a
few large dark spots on HB dorsally, usu-
ally with one large spot across root of tail
and a few spots on caudal muscle; no
spots on ventral surface of HB. Older lar-
vae becoming darker, in many HB
brownish dorsally and ventrally with ob-
scure darker spots dorsally; caudal
muscle and fins dusky with obscure
darker spots.

HBL in stages 26-28 is 11.1 to 17.3
mm, stages 35-37 is 18.2 to 27.5 mm;
maximum width of HB 0.59 to 0.64, mean
0.609; oral disk width 0.20 to 0.26, mean
0.231. Denticle formula (stages > 25)
upper lip I:6+6 (66 tadpoles), I1:6+5 (14),
1:5+5 (52), 1:5+4 (1); lower lip 6+6:I (1),
5+5:I1 (118), 5+4:1 (6), 4+4:I1 (8).

BUCCAL CAVITY

Based on FMNH 63537, stage 36, HBL 17.5 mm
and FMNH 139380, stage 37, HBL 20.5 mm.

Ventral surfaces. Two large infralabial
papillae or palps on each side; one just
inside lateral corner of beak compressed,
with four long, tapering, pustulose
branches; second medial and posterior to
first, with thick base and five to seven
unequal fingerlike projections, lateral
two longest. A pair of simple lingual pa-
pillae, half length of posterior infralabials
and medial to them. Tongue anlage a low
swelling. Buccal floor arena roughly
hexagonal; 10 to 13 large pointed papil-
lae on each lateral border, all except end
ones branched, posterior two-thirds of
central area with many (35-50) low pus-
tules. Buccal pocket transversely oval,
perforated; no pre-pocket papillae or pus-
tules. A row of six to seven small, slender
papillae beginning posteromesad from

_ buccal pocket and curving behind last

three buccal floor arena papillae.
Ventral velum wide, with obvious
spicules; dorsal surface between velar
margin and buccal floor arena with
numerous rounded pustules. Median
area of velum set off by notches and di-
vided into dorsal and ventral portions;
dorsal portion with two pairs of large pa-

18 Advances in Herpetology and Evolutionary Biology

pillae, median ones longer; all with small
knobs; ventral portion with one pair of
papillae subequal to lateral ones of dorsal
portion; all surfaces of median area with
glandular pits. Lateral sections of velar
margin with shallow curves marked by
short projections.

Three gill filter cavities; 7 to 11 filter
rows on filter plates; filter rows complexly
folded and dense, but filter canals not
roofed over.

Glottis on raised disk, just visible be-
tween median papillae of velum.

Dorsal surfaces. Prenarial arena nar-
row, long, with one pair of simple, slen-
der papillae. Nares transversely oval,
with thin walls; anterior wall with one
inwardly curved, pointed, pustulose pa-
pilla laterally; higher posterior wall with
similar papilla medially; relative heights
of narial papillae vary. Postnarial arena
with varying number of papillae; in one
specimen three long, slender, un-
branched papillae on each side; in
second specimen two oblique, parallel
rows of papillae, those of posterior row
longer and pustulose. Lateral ridge papil-
la long, with one short branch. Median
ridge triangular, margin pustulose. Buc-
cal roof arena with 8 to 10 long pointed
papillae laterally in staggered row, those
in center of row with short branches; cen-
tral area with many short papillae or pus-
tules, decreasing in height posteriorly. A
row of three to four short papillae lateral
to buccal roof arena. A transverse band of
secretory pits across roof between buccal
roof arena and dorsal velum; band widest
medially.

Dorsal velum with ca. 12 papillate pro-
jections on each side; medial ones long-
est. One large lateral pressure cushion
visible in one larva; area damaged in dis-
section in other.

Leptobrachium nigrops Berry and Hen-
drickson
Figures 1-5

Larval material examined: Sarawak—FMNH
77512-13, 77578, 148284-85; 151515-21, 158209-
14,

Figure 1.

Leptobrachium nigrops (FMNH 148285),
anterior portion of roof of buccal cavity and upper por-
tion of oral disk (x17). Anterior at top.

Abbreviations: LRP—lateral ridge papilla (slightly
damaged); MR, median ridge; NPA, anterior narial
papilla; NPP, posterior narial papilla; UB, upper beak;
ULP, upper marginal labial papilla.

EXTERNAL FEATURES

Very similar to L. montanum larvae in
general form and proportions, differing
mainly in coloration in life and in shape
of dorsal fin. Beak more coarsely serra-
ted, and marginal papillae of upper lip
more distinctly separated than in mon-
tanum (Fig. 1). Denticles (Fig. 2) as in
montanum.

Dental formula I:6+6/5+5:I except in
young larvae (stage < 26). Dorsal fin ori-
gin slightly behind end of body, fin rising
abruptly, margin more convex than in
montanum; otherwise tail form similar to
that of latter species. HB (in life) with
middorsal stripe of gold chromatophores;
in preservative HB straw-colored, with a
dark middorsal stripe of melanophores; a
dark interorbital band; a dark transverse
band or a pair of large, narrowly, sepa-

Figure 2. Leptobrachium nigrops (FMNH 77513),
denticles of lower lip (963).

rated, dark spots at rear of HB; ventrally
HB without markings.

HBL of two largest stage 25 tadpoles
17.5 and 18mm, two in stages 31-32 24.5
and 26.6 mm, one stage 38 22 mm.

BUCCAL CAVITY

Based on FMNH 77513, stage 25, HBL 14.8 mm
and FMNH 148285, stage 32, HBL 24.5 mm. Very
similar to L. montanum.

Ventral surfaces (Figs. 3, 4). Buccal
floor arena with 50 to 100 low pustules;
anterior to lateral papillae a low, thick,
transverse ridge bearing several short
papillae. Several short papillae and pus-
tules in front of buccal pockets.

Dorsal surfaces (Figs. 1, 5). Prenarial .

area with a large 3-branched papilla just
behind lateral corner of beak. Postnarial
arena with six to eight pustulose papillae
on each side. Lateral ridge papilla with
one short branch near base and two
longer terminal ones. An elongate oval
area of pustules and papillae lateral to
buccal roof arena. Band of secretory pits
widest laterally.

SE ASIAN PELOBATID LARVAE: Inger 19

Figure 3. Leptobrachium nigrops (FMNH 77513),
anterior portion of floor of buccal cavity (x42). Anterior
at top.

Abbreviations: ILPA, anterior infralabial papilla;
ILPP, posterior infralabial papilla; LP, presumed
lingual papilla.

Leptobrachium hendricksoni Taylor

Differences between larvae of hen-
dricksoni and montanum support the dif-
ferences between adults noted by Taylor
(1962).

Larval material examined: Malaya—BBM 6965,
7137; FMNH 134544. Sarawak—119898-99,
129014, 129016-7, 129019-20, 148282-83, 158215—
18.

EXTERNAL FEATURES

Very similar to larvae of montanum
and nigrops in general body form and
oral disk, but differing from them in colo-
ration and in absence of glandular
patches ventrally on HB.

Narial rim low, in Bornean larvae one
small middorsal projection flanked by

20 Advances in Herpetology and Evolutionary Biology

Figure 4. Leptobrachium nigrops (FMNH 148285),
rear portion of floor of buccopharnyngeal cavity (x 18).
Anterior at top.

Abbreviations: BFA, buccal floor arena; FC, filter cav-
ity; GL, glottis; VVMP, median papillae of ventral
velum; VVP, pustules of ventral velum.

several subsidiary ones giving impres-
sion of a low, pustulose triangular projec-
tion; Malayan larvae with simple, low
middorsal projection. Oral disk re-
sembling that of montanum; dental for-
mula usually 1I:7+7/6+6:I or 1:6+6/
6+6:I1; denticles shaped like those of
montanum. Tail form as in montanum.
No glandular patches evident ventrally
on HB. In preservative reddish brown
above and on tail, with numerous small
dark, round dots on all surfaces.

HBL of largest stage 25 tadpole 23.0
mm, stages 26-28 (8 individuals) 24.2 to
30.0 mm, stages 35-37 (6) 25.8 to
34.4 mm. Width of HB 0.54 to 0.66 (mean
0.600), eye length (stages > 25) 0.08 to
0.12 (mean 0.092) width of oral disk 0.19
to 0.23. Denticle formula (stages > 25)
upper lip I:7+7 (4 tadpoles) 1:6+7 (1),
1:6+6 (13); lower lip 6+6:I (13), 6+5:I
(1), 5+5:1 (4).

Figure 5. Leptobrachium nigrops (FMNH 148285),
left rear portion of roof of buccal cavity (x 18). Anterior
at top.

Abbreviations: BRA, buccal roof arena; DV, dorsal
velum; GZ, glandular zone.

BUCCAL CAVITY

The larva (FMNH 148282) described
by Wassersug (1980) under the name L.
hasselti is L. hendricksoni. A stage 32
larva (FMNH 148283), dissected in the
present study, agrees in almost all par-
ticulars with Wassersug’s description.
This specimen has a trilobate papilla
anterior to the infralabial papilla and
lateral to the three pustulations noted by
Wassersug. The buccal cavity is gener-
ally similar to that of L. montanum.

Ventral surfaces. Buccal floor arena
with 12 to 20 short papillae across its pos-
terior fourth; a few low pustules medial
to lateral papillae. About six pre-pocket
papillae.

Dorsal surfaces. Prenarial area with
two to three low pustules on each side.
Postnarial arena with three short papillae
on each side. Lateral ridge papilla bifur-

cate near tip. Median ridge semilunar;
margin with small projections. Buccal
roof arena medially with many pustules,
densest posteriorly. Distribution of pits
in transverse glandular zone as in nig-
rops. Dorsal velum with seven to nine
papillae on each margin.

Leptobrachium pullum Smith

Four samples of larvae from southern
Thailand (Nakkon Si Thammarat Prov-
ince, Ron Phi Bun—FMNH_ 173481,
173583, Phangnga Province, Lam
Pi—FMNH 206820; probably from Trang
Province—FMNH_ 172253) have the
general facies of larval L. montanum and
appear to belong to a single species,
which I tentatively assign to L. pullum
Smith (see above, p. 16). Although these
samples are from an area close to the
range of L. hendricksoni, which Taylor
(1962) reported from Yala Province,
Thailand, and not too distant from that of
L. nigrops in central Malaya, these puta-
tive pullum larvae differ from those of
hendricksoni and nigrops. The coarse
serration of their beaks, the thickness of
their labial papillae, and their coloration
are quite different from the conditions in
larval hendricksoni. The confluence of
the bases of their upper labial papillae, a
light triangular area at the root of the tail,
and the apparent absence of ventral
glandular patches distinguish them from
larval nigrops. In the last two characters
they also differ from larval Bornean mon-
tanum.

As noted earlier, J. Dring (in litt.) has
suggested that all Thai frogs previously
assigned to L. hasselti Tschudi are prob-
ably conspecific with L. hasselti pullum
Smith, which Dring believes warrants
full species recognition.

EXTERNAL FEATURES

Body form in general as in L. mon-
tanum, i.e., oval, flattened above, spher-
oidal below; narial rim as in montanum;
labial papillae thick, bases of those on

SE ASIAN PELOBATID LARVAE: Inger 21

upper lip confluent; dental formula most
often 1:6+6/5+5:I or 1:5+5/5+5:1; den-
ticles shaped as in montanum; beaks
coarsely serrated, upper with enlarged
median cusp. Tail shaped as in mon-
tanum. Glandular patches not evident
ventrally on HB.

Color (in alcohol) pale to dusky on HB
with many small, irregular dark spots
dorsally; a conspicuous triangular light
area over root of tail bounded anteriorly
by wide transverse dark band and pos-
teriorly by an oblique band, the two
bands joining laterally; ventrally HB
unmarked or with a few obscure spots in
anterior fourth; tail muscle with large
dark spots mostly in anterior half; fins
with smaller, less conspicuous spots.

HBL of largest stage 25 tadpole is 16
mm, stages 26-28 is 15.8 to 19.2 mm,
stages 35-37 is 21.5 to 24.7 mm; maxi-
mum body width 0.51 to 0.64 (mean
0.563); eye 0.09 to 0.11; width of oral disk
0.22 to 0.28 (mean 0.253); denticle for-
mula (stages > 25) of upper lip I:6+6 (9),
1:6+5 (1), 1:54+5 (5), of lower lip 5+5:I
(13), 44+.4:I (2).

BUCCAL CAVITY

Based on FMNH 172253, stage 36, HBL 22.0 mm
and FMNH 173481, stage 36, HBL 24.3 mm. Gen-
erally similar to L. montanum.

Ventral surfaces. Posterior infralabial
papillae with three major branches, each
further subdivided into pustulose twigs.
Central area of buccal floor arena with 40
to 50 low pustules.

Dorsal surfaces. Lateral ridge papilla
slender, unbranched. Median ridge low,
rectangular; margin sinuate; one low pro-
jection at lateral corner. Central area of

_ buccal roof arena with many low pus-

tules. A cluster of 8 to 10 papillae lateral
to posterior papillae of buccal roof arena.
Glandular zone widest laterally.

Leptobrachium sp.

One sample of two tadpoles from nor-
thern Thailand (Chiang Mai Province,

22 Advances in Herpetology and Evolutionary Biology

Doi Suthep—FMNH 212387) and
another slightly shriveled series of two
from Annam, Indo-China (FMNH 26505)
appear to represent another species of
the L. montanum group. The well-
preserved sample from Doi Suthep dif-
fers from those previously described in
this paper in having conical, denticle-less
inframarginal papillae laterally on the
lower lip and in having seven to nine
small projections on the dorsal rim of the
nares. The Annam larvae have the same
type of inframarginal papillae on the
lower lip, but their unsatisfactory state of
preservation makes count of narial pro-
jections uncertain, although they clearly
have more than one small projection.
Both series also differ from larval pullum
and montanum in having the bases of the
upper labial papillae slightly separated,
and from larval hendricksoni in colora-
tion, upper labial papillae, and in serra-
tion of the upper beak.

Because there still remains some doubt
about definition of species in the Thai
members of this group, these larval series
are not assigned a specific name.

EXTERNAL FEATURES

Body form and oral disk generally as in
L. montanum, though slightly more slen-
der; HBL at stage 30 is 26.0 mm, at stage
38 is 28.3 mm. Narial rim low with six to
nine low pustules dorsally. Papillae of
upper lip with bases not confluent; in-
framarginal papillae of lower lip conical,
without denticles; width of oral disk 0.27;
dental formula I:5+5/4+4:I; denticles as
in montanum; beaks coarsely serrated,
upper with enlarged median cusp. Tail
shaped as in montanum. Glandular
patches ventrally on HB as in montanum.
HB and caudal muscle with very fine
colorless asperities.

Color in formalin dark grayish brown
on HB and tail; HB without evident pat-
tern on spotting; indistinct dark spots or
bars on tail; fins with dense dusting of
melanophores.

BUCCAL CAVITY

Based on FMNH 212387, stage 30, HBL 26.0 mm.
Generally similar to L. montanum.

Ventral surfaces. Buccal floor arena
with about 40 low pustules. Dorsal sur-
face of ventral velum with 12 low papil-
lae or pustules; secretory pits not evident
on dorsal surfaces of velum.

Dorsal surfaces. Postnarial arena with
four slender papillae on each side. La-
teral ridge papilla with two anterior and
one posterior conical projections. Median
ridge triangular, margin with seven pro-
jections. Buccal roof arena medially with
dozen papillae anteriorly and equal
number of low pustules posteriorly.

Leptobrachium spp. (Malaya)

Two samples of larvae trom West
Malaysia, one from Johore and one from
Pulo Tioman, resemble larval L. mon-
tanum in form, but differ from each other
and other larvae described in this paper.
The only Leptobrachium known from the
Malay Peninsula to which larvae have
not been assigned is L. heteropus. No
species of Leptobrachium has been re-
ported from Pulo Tioman (Hendrickson,
1966). None of these tadpoles in these
two samples is developed to the point at
which diagnostic features of adults can
be discerned.

BBM 7162—Mawai Estate, ca. 2.5 km north of
Kota Tinggi, Johore, West Malaysia. Seven indi-
viduals, stages 25-39.

EXTERNAL FEATURES

Shape of oral disk as in L. montanum,
body more slender. HBL 15.6 to
18.5 mm, stages 36-39; maximum body
width 0.35 to 0.49. Nares large, narial rim
very low, with single very weak mid-
dorsal projection present in two. Papillae
of upper lip with bases separated; a few
rounded inframarginal papillae on lower
lip, several with small denticles; width of

oral disk 0.17 to 0.20; dental formula
3+3/34+3:1 (stages 31-39); beaks black in
marginal half; coarsely serrated, no en-
larged median cusp. Tail with straight
margins, tapering abruptly in final third.
Spiracular tube free at tip. Vent median,
opening long and exposed. Coloration
obscured in preservative.

BUCCAL CAVITY
Based on one tadpole at stage 38 (HBL 18.5 mm).

Ventral surfaces. Anterior infralabial
papillae well separated from beak, each
with narrow base and _ three short
branches; posterior infralabial papillae
with thick cylindrical bases, each with
three branches in terminal half. Buccal
floor arena oval, bounded laterally by a
partially doubled row of 12 long, tapering
papillae; no pustules or papillae in cen-
tral area. No pustules or papillae be-
tween buccal floor arena and margin of
ventral velum. Ventral velum with well-
developed spicules. Median fifth of
velum divided into dorsal and ventral
portions, each with two pairs of thick
papillae. Three subequal gill cavities; fil-
ter rows complexly branched; filter
canals canopied.

Dorsal surfaces. Prenarial arena as
wide as long; one pair of short papillae
pustulose. Nares slitlike, with low thick
walls, the posterior one drawn out later-
ally into short, thick, valvelike projec-
tion; no narial papillae. Postnarial area
with deep, narrow pit behind each naris,
pits narrowly separated; two short papil-
lae behind each pit. Lateral ridge papilla
long, slender, with three subequal
branches in distal half. Median ridge
semilunar, low, margin pustulose. Buccal
roof arena with 12 to 14 long papillae in

partly doubled row laterally; a single

pustule visible in central area. Three to
five secretory pits in extreme postero-
lateral corner of buccal roof. Surface from
center to rear of buccal roof “furry” but
without distinct pustules or papillae. Dor-

SE ASIAN PELOBATID LARVAE: Inger 23

sal velum deep; median half of left side
with five short, thick projections; right
side damaged.

BBM 6972—Ulu Lalang, Pulo Tioman. One indi-
vidual, stage 26.

EXTERNAL FEATURES

Body form as in L. montanum, HBL
22.6 mm, maximum width 0.73. Nares as
in L. montanum. Width of oral disk 0.32;
both lips with deep median notch; mar-
ginal papillae of upper lip with separate
bases; lower lip very broad anteroposter-
iorly, surface covered with pillowlike,
circular swellings lacking denticles;
swellings in three irregular rows, con-
tinuing around corners of disk to upper
lip where they become limited to single
row reaching median notch of lip. Dental
formula [:4+4/3+3:I, undivided upper
row one-fourth length of corresponding
lower row; beaks completely black,
coarsely serrated, no median cusp. No
glandular patches. The tail is damaged,
obscuring the shape, but the muscle is
heavy. HB and tail are dusky in formalin
without evident markings.

Leptobrachium gracilis Gunther
Figures 6-8

Larval material examined: Sabah—FMNH
130860, 130864, 13124546, 212381-82. Sarawak—
FMNH 77509, 77511, 77516, 139381, 146323,
146341.

EXTERNAL FEATURES

Head and body forming elongate oval,
flat above, slightly rounded below; max-
imum width about 0.50 of length.

Eyes dorsolateral, not visible from be-
low, oriented laterally, line connecting
pupils 0.20 of HBL behind snout; length
of eye varying with development of larva,
about 0.06 in stage 36; interorbital about
0.15. Nostrils open; narial rim low with
one low, simple projection mid-dorsally;

24 Advances in Herpetology and Evolutionary Biology

nostrils 0.08 behind snout, internarial
about 0.15; eye-nostril varying with de-
velopment, about 0.09 in stage 36.

Oral disk ventral, subterminal, cuplike;
width about 0.20; both lips with deep,
median indentation; small homogenous
papillae in single row around entire mar-
gin of lips; small round inframarginal
papillae flank lower dental rows; infra-
marginal papillae without denticles;
broad band of lower lip smooth between
dental rows and marginal papillae. Den-
tal formula variable, in stages > 25 upper
lip 1:3+3 to 1:6+6, lower lip usually
2+2:I; undivided row of lower lip longer
than first divided row; undivided row of
upper lip around one-third length of
lower one; denticles laterally com-
pressed, curved and pointed (Fig. 6).
Beaks heavy, fully pigmented, finely ser-
rated; upper beak without enlarged me-
dian cusp.

Spiracle sinistral, mid-lateral; with
raised ventrolateral edge, free only at
end; opening 0.50 behind snout. Anal tub
dextral, attached to ventral fin.

Tail about 2.0 of HBL, margin very
slightly convex, tapering gradually in last
third to rounded tip; maximum depth be-
fore mid-length, 0.24 of HBL, about 0.10
of tail length. Caudal muscle heavy, 0.26
of HBL at end of body, not tapering until
point of maximum tail depth. Dorsal fin
origin just beyond level of anal opening;
very low in proximal fourth; equal to
depth of muscle only in distal third. Ven-
tral fin deeper than dorsal only in proxi-
mal fourth.

Head and body without asperities or
glandular patches. Lateral line system
visible on HB and tail.

Coloration (in preservative) with light
or heavy dusting of melanophores dor-
sally on HB and proximal two-thirds of
caudal muscle; lower half of body, fins,
and distal third of caudal muscle without
pigment.

HBL of two largest stage 25 larvae is
19.3 and 20.0 mm, two in stages 35-37 are
21.6 mm; maximum width of HB 0.42 to
0.51; width of oral disk 0.17 to 0.22. Den-
ticle formula (stages > 25) of upper lip

Figure 6. Leptobrachium gracilis (FMNH 146323), denticles of upper lip (x2300).

I:6+6 (1 larva), 1:4+4 (5), 1:34+3 (1);
lower lip 2+2:I (6), 3+3:I (1).

BUCCAL CAVITY

Based on FMNH 146323, stage 35, HBL 21.6 mm.

Ventral surfaces (Fig. 7). Anterior in-
fralabial papilla a thick transverse flap
just behind lateral corner of lower beak;
separated from that of other side by about
half its length; lateral corner low, bearing
short, smooth, conical spur; rest of papilla
with thick pustulose projections; median
portion highest. Posterior infralabial pa-
pilla with thick base and three terminally
pustulose branches. A laterally curved
simple papilla between the infralabial
papillae. A pair of conical, smooth lingual
papillae between posterior infralabial
papillae, their bases continuous with
those of the latter. Tongue anlage low.
Buccal floor arena (Fig. 7) a wide rec-
tangle, bounded laterally by eight to nine
papillae; several of largest papillae with
confluent bases and three to four sharp,
posteriorly directed spurs; midline of
floor arena with three thick, short pus-
tulose papillae in center; about 40 low
pustules in transverse band across rear
third of floor arena, continuous with pus-
tulose area of velum, Buccal pocket an
inverted U-shaped slit; perforation not
evident, an oblique row of three smooth
pre-pocket papillae. Area behind pocket
without papillae or pustules.

Ventral velum (Fig. 7) wide, without
evident spicules; a deep median notch;
dorsal surface densely covered medially
with pustules except for bare strip in
mid-line; laterally dorsal surface of
velum smooth; margin without secretory
pits. Median area of velum divided into

dorsal and ventral portions; dorsal por-

tion with two pairs of papillae, outer pair
longer; ventral portion a forest of fila-
mentous papillae, at least 20 on each
side; no secretory pits on velar papillae.
Lateral area of velum sinuate; without
projections or marginal pits.

Three filter cavities, lateral one half

25

SE ASIAN PELOBATID LARVAE - Inger

Figure 7. Leptobrachium gracilis (FMNH 146323),
floor of buccopharyngeal cavity behind tongue anlage
(x 18).

Abbreviations: Anterior at top. BP, buccal pocket; other
abbreviations as in Fig. 4.

length of others; 8 to 11 filter rows on
filter plates; filter plates bordering mid-
dle cavity with a few dorsally directed,
branching projections (Fig. 7, lower left);
filter canals open.

Glottis hidden by fringing papillae of
velum.

Dorsal surfaces (Fig. 8). Prenarial
arena narrow, long; one pair of thick, low
papillae just behind center. Nares
obliquely oval, separated by length of
one opening; anterior wall thin, low, with
pustulose margin; posterior wall slightly
higher, with one long, smooth papilla
medially. Postnarial area with row of
three flattened, triangular papillae on
each side; central one of each set about
twice height of others. Lateral ridge pa-
pilla a simple flat, elongate triangle, bi-
furcate at tip. Median ridge directly be-

26

Figure 8. Leptobrachium gracilis (FMNH 146323),
roof of buccal cavity behind internal nares (x18). An-
terior at top.

Abbreviations: see Figs. 1, 5.

hind gap in postnarial papillae deeply
divided into three smooth, triangular
lobes. Buccal roof arena (Fig. 8) oval,
with about 30 smooth, unbranched,
pointed papillae. A transverse band of
pustules or low papillae behind roof
arena, extending beyond lateral margins
of roof arena; posterior border of band
curved and pointed forward at midline;
pustules along posterior border higher;
anterolateral corner of band with three to
four long papillae; about 150 pustules on
each side of midline. No secretory pits.
Dorsal velum with 18 to 20 filamentous
papillae on margin of each side; papillae
subequal to those of ventral velum.

Leptobrachium pelodytoides Boulenger

Dubois (1981) suggests that L. mini-
mum Taylor (type locality Doi Suthep,
Chiang Mai Province, Thailand) is con-

Advances in Herpetology and Evolutionary Biology

specific with L. oshanense Liu (type
locality Mt. Emei, Sichuan). Dring (in
litt.) has suggested that minimum is con-
specific with L. pelodytoides Boulenger
(type locality Karen Hills, Burma). The
material available for this study is not
adequate to resolve this issue.

The Thai larvae listed below differ
from our Chinese tadpoles of L. oshan-
ense Liu in two respects; in the former
the beaks are more coarsely serrated, and
the rows of denticles on the lower lip are
flanked by several rounded inframarginal
papillae, which are absent in the Chinese
series (see Liu, 1950: Fig. 42). The Thai
larvae have dark spots on the tail as fig-
ured by Smith (1917) in his description of
larval pelodytoides. Smith also illustrated
inframarginal papillae, though on both
lips. Tentatively, given these slight dif-
ferences between Thai and Chinese tad-
poles, I assign the former samples to
pelodytoides.

Larval material examined: Thailand—FMNH
174525, 212386, USNM 130402. In addition several

samples of larval L. oshanense Liu from China—
FMNH 24419, 49589.

EXTERNAL FEATURES

Body form similar to that of L. gracilis;
maximum width 0.50 to 0.55 of HBL; eye
length (stage 35) 0.09; interorbital about
0.20. Narial rim prominent, perimeter
divided into four subequal lobes. Oral
disk as in gracilis; marginal and infra-
marginal papillae as in gracilis; dental
formula [:2+2/2+2:I (4) or 1:3+3/2+3:1
(1), proportions of rows as in gracilis;
beaks coarsely serrated, median cusp of
upper beak not enlarged. Spiracular tube
as in gracilis. Tail shape as in gracilis. No
spinules or glandular patches evident.

Color (in preservative) brown with ir-
regular dark spots on caudal muscle and
dorsal fin; ventral fin colorless proxim-
ally, becoming brownish as dorsal fin in
distal half; ventrally HB without melano-
phores.

HBL of largest stage 25 is 13.0 mm,
stage 35 is 20.5 mm.

BUCCAL CAVITY

Based on FMNH 212386, stage 35, HBL 20.5 mm.
Generally similar to L. gracilis.

Ventral surfaces. Posterior infralabial
papillae with three major branches,
middle one complexly subdivided. Cen-
tral portion of buccal floor arena with ca.
100 low pustules, densest posteriorly.
Band of pustules on dorsal surface of ven-
tral velum without median interruption.

Dorsal surfaces. Prenarial area without
evident pustules or papillae. Anterior
narial wall with one long papilla later-
ally. Postnarial arena with three triangu-
lar, flattened papillae on each side. Later-
al ridge papilla with four or five divisions
of varying lengths. Median ridge semi-
lunar with 10 short marginal projections.
Central portion of buccal roof arena with
ca. 70 pustules. Glandular zone limited to
extreme posterolateral corner of roof.
Dorsal velum with 8-10 long flattened
papillae on each side.

Leptobrachella mjobergi Smith
Figures 9-10

A number of Bornean samples of
pelobatid larvae, erroneously assigned to
Leptobrachium gracilis (Inger 1966),
resemble that species in having deep
cuplike oral disks, but differ from that
form and all other larvae previously de-
scribed in this paper in lacking denticles
(Fig. 9). These larvae further differ from
gracilis in having more slender bodies
(width of HB in stages > 25 0.32 to 0.41,
in gracilis 0.43 to 0.50) and in having the
spiracular tube free of the body wall fora
distance equal to about twice the di-
ameter of the opening. In gracilis the
Spiracular tube is free only at the very tip.

Association of these larvae with Lepto-—

brachella is based on a stage 41 larva that
has the pointed digits peculiar to this
genus (Inger, 1966: Fig. 8A).

The samples listed below may com-
prise more than one species. The infra-
marginal papillae in those from eastern
Sabah (FMNH 77503-04) are restricted

SE ASIAN PELOBATID LARVAE: Inger 27

Figure 9. Leptobrachella mjobergi (FMNH 77503),
anterior portion of floor of buccal cavity (x44). Anterior
at top.

Abbreviations: |MP, inframarginal papilla of lower lip;
LB, lower beak; other abbreviations as in Figs. 1, 3,
and 4. Left ILPP cut and tip of right obscured by detri-
tus. Left edge of BFA damaged.

to the vicinity of the beaks leaving a
wider portion of the lips smooth than is
the case in the other samples. Also one of
these Sabah larvae (FMNH 77503, stage
36) differs from one from Mount Kina-
Balu (FMNH 130861) in the amount of
pustulation on the ventral velum within
the buccal cavity (see below). Although
only two larvae were dissected, this vari-
ation exceeds that observed in the small
samples of other species.

Larval material examined: Sabah—BBM 7363,
FMNH 77503-04, 130861-63. Sarawak—FMNH
77515, 157998, 21238485.

EXTERNAL FEATURES

Head and body extremely elongate, flat
above, rounded below; maximum width
about 0.35 of HBL.

28 Advances in Herpetology and Evolutionary Biology

Eyes dorsolateral, not visible from be-
low, oriented laterally; eyes 0.18 behind
snout; length of eye varying with de-
velopment, about 0.07 in stages 35-37;
interorbital about 0.10. Nostrils open; rim
not raised except for small middorsal pro-
jection; nostrils about 0.07 behind snout;
internarial about 0.10. .

Oral disk ventral, subterminal, cuplike;
width about 0.15; lower lip with deep
notch, upper with shallow notch; very
short, fine papillae in single row around
margin of lips; round inframarginal knobs
close to beaks (Fig. 9); area between
margin of lips and inframarginal papillae
variable. No denticles, no long dental
ridges (Fig. 9). Beaks fully pigmented,
finely serrated, median cusp of upper
beak slightly enlarged.

Spiracle sinistral, mid-lateral; spiracu-
lar tube free for distance 2 to 2.5 times
diameter of opening; opening usually
less than 0.50 behind snout. Anal tube
dextral, attached to fin.

Tail about 2.0 of HBL, margins roughly
straight most of length, tapering abruptly
near end to rounded point; maximum
depth at beginning of distal third, about
0.40 of HBL. Caudal muscle very heavy,
almost as wide as HB; not tapering ap-
preciably until distal third; three to five
times depth of fins at mid-length. Dorsal
fin origin behind HB; thick and low in
proximal half of tail; equal to depth of
muscle only in distal third. Ventral fin
deeper than dorsal in proximal half.

Color in preservative pale yellowish
brown without markings, in life pinkish
gray.

HBL of two largest in stage 25 are 12.5
and 13.8 mm, of three in stages 26-27 are
11.5 to 13.0 mm, of three in stages 35-37
11.5 to 13.2 mm; maximum width of HB
0.31 to 0.41; width of oral disk 0.13 to
Or.

BUCCAL CAVITY

Based on FMNH 77503, stage 36, HBL 11.5 mm,
and FMNH 130861, stage 30, HBL 13.8 mm. The
narrow heads made dissection of these larvae diffi-
cult and both were slightly damaged.

Ventral surfaces (Fig. 9). Anterior infra-
labial papilla immediately inside lower
beak, transversely oriented, thick, lowest
laterally and thickly pustulose in median
two-thirds; narrowly separated from flap
of other side. Posterior infralabial papilla
thick, flattened with two smooth, sub-
equal branches in distal half. Two or
three smooth conical papillae between
and almost confluent with the infralabial
papillae. No lingual papillae. Tongue
anlage low. Buccal floor arena trape-
zoidal, wider posteriorly, bordered
laterally by 10 to 12 long, slender, simple
papillae, bases contiguous; central area
of arena with three to four short, thick
papillae in mid-line anteriorly, followed
by an expanding triangular zone of longer
papillae and pustules; between this zone
and lateral papillae 20 to 30 low pustules
on each side. Buccal pocket a broad in-
verted U, perforate; three to five short
papillae anterior to pocket parallel to
buccal floor arena. Area behind pocket
with three to four papillae.

Ventral velum (Fig. 10) wide, without
evident spicules; a deep median notch;
dorsal surface with two large, rounded
groups of pustules narrowly separated in
mid-line, anterior pustules larger, ca. 75
in each group; laterally dorsal surface of
velum smooth in FMNH 77503, with low
pustules in FMNH 130861; margin di-
vided into two lobes on each side; margin
without secretory pits. Median lobe of
each side divided horizontally; dorsal
portion with two long papillae; ventral
portion with more than 10 filamentous
papillae. Lateral lobe with two to three
short marginal papillae.

Only one filter cavity on each side; no
ruffled filter rows.

Glottis in raised disk, under median
notch of velum.

Dorsal surfaces. Prenarial area long
and narrow, bounded at rear by row of
three to four short papillae. Nares
obliquely oval; narial walls thin, low;
both walls with one or two pustules, no
papillae. Postnarial area smooth or with
several indistinct pustules. Lateral ridge

Figure 10. Leptobrachella mjobergi (FMNH 77503),
median portion of right half of ventral velum (x82).
Anterior towards top.

Abbreviations: see Fig. 4.

papilla with two long, subequal
branches. Median ridge with three tri-
angular lobes, middle one slightly wider
in one specimen; margins without pus-
tules; one short papilla narrowly sepa-
rated from median ridge on each side.
Buccal roof arena occupying full width of
buccal cavity; with about 40 pointed pa-
pillae decreasing in size posteromedi-
ally; one near center bifurcate; rear of
arena with small, smooth triangular area

apointed forward in FMNH 130861. A |

large transverse band of postules behind
buccal roof arena; ca. 100 to 125 pustules,
largest ones posteriorly; rear boundary of
pustulose area arched forward medially.
Indistinct glandular zone behind center
of pustulose band.

Dorsal velum with 15 to 20 long papil-
lae on margin of each side.

SE ASIAN PELOBATID LARVAE: Inger 29

HABITAT DISTRIBUTION AND FOOD

All of the larvae considered here were
collected in streams, although that term
encompasses a variety of microhabitats.
Precise information on microhabitats is
available for only a portion of the speci-
mens examined and is summarized in
Table 4.

The results are not surprising. The
heavy-bodied forms are largely confined
to areas of slow or no current and the
slender-bodied ones (except for one
sample of Leptobrachium pelodytoides)
to areas of stronger current. At the time of
collection of the L. pelodytoides sample,
the stream had dried into a series of small
pools (notes of D. L. Damman). The tabu-
lation does not show, however, an impor-
tant difference between montanum,
hendricksoni, and nigrops, on the one
hand, and gracilis and Leptobrachella on
the other. Individuals of the former
group, though feeding on the bottom, are
usually exposed on the substrate or hid-
den among dead leaves covering the bot-
tom, whereas those of the second group
are rarely exposed on the substrate and
are usually within the interstices of the
bottom gravel or boulders. Liu (1950) ob-
served Leptobrachium oshanense, which
is generally similar to gracilis in body
form and oral apparatus, both in crevices
between rocks and on the surface of the
substrate.

Gut contents of Leptobrachium
montanum (3 individuals), pullum (2),
hendricksoni (2), and gracilis (2) were
studied by means of light microscope (in
Hoyers medium) and SEM. Only the
most anterior portion of the gut was used.
In all cases the most abundant type of
food was plant fragments—either parts of
leaves or vascular tubes. The SEM re-
vealed diatoms and fragments of hyphae
or other fungal parts in all except gracilis.
Larvae of montanum, reared in a field
laboratory, fed actively on dead leaves
and fragments of prawn, but did not at-
tack one another. Size range of food
fragments, selected as representative and

30 Advances in Herpetology and Evolutionary Biology

TABLE 4. HABITAT DISTRIBUTION OF LARVAE OF CERTAIN SOUTHEAST ASIAN SPECIES OF PELOBATIDS.

Quiet pools*
Species Ind.}~ Coll.§
Leptobrachium montanum 123 13
L. hendricksoni 15 2
L. nigrops - 66 3
L.sp 2, 1
L. gracilis
L. pelodytoides 2) iL
Leptobrachella mjobergi 1 1

*Areas of slow to moderate current.

Side poolst Shingle beds Riffles
Ind. Coll. Ind. Coll. Ind. Coll.
11 4 ® 2 PP 3

12 1
6 2
18 6
2) Dy 14 5

t Areas cut off from the current, usually by gravel bars and often the site of accumulations of dead leaves.

+ Number of individuals.
SNumber of collections.

TABLE 5. SIZE DISTRIBUTION OF REPRESENTATIVE FOOD FRAGMENTS FROM STOMACHS OF LARVAL LEPTOBRACHIUM.

Greatest dimension (mm) of food item

Species < 0.01 0.01-0.05
L. montanum 1 6
L. hendricksoni 1 5
L. pullum il 3
L. gracilis 0 0

measured on SEM photographs, shows
little differentiation among _ species
(Table 5). As this is not an exhaustive or
random sampling, rigorous statistical an-
alysis is not appropriate.

DISCUSSION

Except for the genus Megophrys, lar-
vae of Asiatic megophryine genera have
depressed, spheroidal bodies even
though all live in streams (Smith, 1917;
Pope, 1931; Liu, 1950; Inger, 1966). The
slender bodies of larval Leptobrachium
(Leptolalax) and Leptobrachella may
consequently be considered as a derived,
specialized condition. These two groups
also share other characteristics that dis-
tinguish them from all other non-
Megophrys megophryines; 1) expanded,
cuplike lips; 2) reduced or no spicules in
the ventral velum; 3) dense clusters of

0.06-0. 10 0.11-0.20 0.21-0.30 0.30+
0 2 1 4
0 2 1 3
5 2 it 1
4 3 4 2

filamentous papillae medially on the
ventral velum; and 4) reduction of the
glandular zone in front of the dorsal
velum (Tables 2, 3). As well-developed
glandular zones and spicules character-
ize not only other megophryines (includ-
ing Megophrys minor) but also Scaphio-
pus bombifrons (Wassersug, 1980),
character states (2) and (4) appear to be
derived. The cluster of filamentous velar
papillae (3) is unlike anything observed
by Wassersug (1980) in other larvae,
except Leptobrachium (Leptolalax)
oshanense. The expanded lips, though
their function is still not clear (Liu, 1950),
also represent a derived condition within
the subfamily. At this time only one of
these character states, the slender body
form which in several species is associa-
ted with the habit of using bottom inter-
stitial spaces, has an apparent functional
significance. The sharing of these de-
rived character states by all known larvae

of species of Leptobrachium (Leptolalax)
supports Dubois’s (1980) recognition of
this taxon.

Larvae of Leptobrachella have three
additional, functionally puzzling special-
izations: a single filter cavity on each
side, loss of the usual frilled gill filter
rows, and the absence of denticles. The
first seems to be a further development of
a condition in Leptobrachium gracilis,
which has one filter cavity much reduced
in size. These two larval forms utilize a
microhabitat different from those of other
tadpoles that have filter cavities reduced
in size (Wassersug, 1980: 106) or number
(Wassersug et al., 1981). The reduction
may be related to life in well oxygenated
waters. The other two specializations
suggest modified feeding behavior,
though no clear evidence for this specu-
lation is at hand. Loss of denticles has
occurred in larvae of diverse habits and
families (Mertens, 1960) and even within
the megophryines (Megophrys).

The tadpoles from Johore (BBM 7162)
are intermediate between forms assigned
in Table 1 to the subgenera Leptobra-
chium and Leptolalax. Character states
shared with the latter are: 1) slender
body form; 2) number of divided rows of
denticles less than four on both lips; and
3) restriction of the glandular zone at the
rear of the buccal roof. The first two
states, while presumably derived within
the megophryines, represent trends seen
frequently in other groups of tadpoles
and, in this case, may be instances of
parallelism. The third state, since it also
is a reduction, may be another parallel-
ism. Significantly, the Johore tadpoles
lack the one derived character that would
clearly ally them with Leptolalax, name-

ly, the numerous filamentous papillae at .

the center of the ventral velum. Instead,
they share the primitive condition of the
velum found in the subgenus Lepto-
brachium. The cylindrical, rather than
flaplike, infralabial papillae and the
absence of pustules down the center of
the buccal floor differentiate them from
both subgenera. The combination of pri-

SE ASIAN PELOBATID LARVAE: Inger 31

mitive and derived, reduced states makes
it difficult to place these larvae in rela-
tion to the others reviewed here.

Differences among the larvae de-
scribed in this paper in buccal anatomy
are not easily explained functionally as
all of those studied seem to feed on the
same kinds of material and on the same
size range. However, a more thorough
analysis of diets might suggest how the
array of large palps and papillae are in-
volved in processing food.

ACKNOWLEDGMENTS

I am grateful to Richard Wassersug for
advice and criticism, to Doyle L.
Damman for several series of larvae from
Thailand, and to Carol Small-Kaplan for
help with the SEM micrographs.

LITERATURE CITED

Dusois, A. 1980. Notes sur la systématique et la
répartition des amphibiens anoures de Chine
et des régions voisantes. IV. Classification
generique et subgenerique des Pelobatidae
Megophryinae. Bull. Mens. Soc. Linn. Lyon,
49: 469-482.

____. 1981. Notes sur la systématique et la réparti-
tion des amphibiens anoures de Chine et des
régions avoisinantes. V. Megophrys oshanensis
Liu, 1950 et Leptobrachium minimum Taylor,
1962. Bull. Mens. Soc. Linn. Lyon, 51:
183-192.

HENRICKSON, J. R. 1966. Observations on the fauna
of Pulau Tioman and Pulau Tulai. The Am-
phibians. Bull. Nat. Mus. Singapore, 34: 72-84.

INGER, R. F. 1966. The systematics and zoo-
geography of the Amphibia of Bomeo.
Fieldiana: Zool., 52: 1-402.

Liu, C. C. 1950. Amphibians of Western China.
Fieldiana: Zool. Mem., 2: 1—400.

MERTENS, R. 1960. Die Larven der Amphibien und
ihre evolutive Bedeutung. Zool. Anz., 164:
337-358.

Pore, C. H. 1931. Notes on amphibians from
Fukien, Hainan, and other parts of China. Bull.
Amer. Mus. Nat. Hist., 61: 397-611.

SMITH, M. A. 1917. On tadpoles from Siam. Jour.
Nat. Hist. Soc. Siam, 2: 261-275.

TayLor, E. H. 1962. The amphibian fauna of Thai-
land. Univ. Kansas Sci. Bull. 43: 265-599.
WASSERSUG, R. 1980. Internal oral features of larvae

of eight anuran families: functional, system-

32 Advances in Herpetology and Evolutionary Biology

atic, evolutionary and ecological considera- 1981. Adaptations for life in tree holes by
tions. Univ. Kansas Mus. Nat. Hist. Misc. Publ. rhacophorid tadpoles from Thailand. J. Her-
No. 68, 146 pp. petol., 15: 41-52.

WASSERBURG, R., K. J. FROGNER, AND R. F. INGER.

Frogs of the Hyla columbiana Group:
Taxonomy and Phylogenetic Relationships

WILLIAM E. DUELLMAN'
LINDA TRUEB?

ABSTRACT. Three species comprise the Hyla
columbiana group—H. carnifex on the Pacific
slopes of Ecuador and the northern part of the
Cordillera Central in Colombia, H. columbiana in

e upper Rio Cauca Valley in Colombia, and a new
species, H. praestans, in the upper Rio Magdalena
Valley in Colombia. Hyla variabilis Boulenger,
1896, is placed in the synonymy of H. columbiana
Boettger, 1892. The three species are readily dis-
tinguished by color pattern, morphometric features,
and mating calls. The Hyla columbiana group be-
longs to a complex of Neotropical hylids character-
ized by 30 chromosomes and reduced larval mouth-
parts.

INTRODUCTION

Among the approximately 385 genera
of Neotropical amphibians and reptiles,
three—Anolis, Eleutherodactylus, and
Hyla—contain more than 800 recognized
species. These three megagenera repre-
sent a significant component of the entire
Neotropical herpetofauna. The largest of
these is the leptodactylid frog genus
Eleutherodactylus, which contains ap-
proximately 360 species—115 in the
West Indies, 70 in Middle America, and
175 in South America. The most widely
distributed is the hylid frog genus Hyla;

although most of the 230 species are ~

Neotropical, including six in the West
Indies, there are also 20 species in the
Holarctic Realm. The 230 species of the

L2Museum of Natural History and Department of
Systematics and Ecology, University of Kansas,
Lawrence, Kansas 66045, U.S.A.

iguanid lizard genus Anolis are divided
about equally between the continental
Neotropics and the West Indies; only one
species is native to North America.

While we have attempted to sort out
relationships among the hylid frogs dur-
ing the preceding two decades, John D.
Lynch has applied himself to Eleuthero-
dactylus. Our current understanding of
the taxonomy, relationships, and_bio-
geography of the complex genus Anolis is
largely the result of the assiduous efforts
of Ernest E. Williams. In recognition of
his devotion to the study of one of these
megagenera and the frustrations associ-
ated with such endeavors, we dedicate
this modest effort and a new species of
Hyla to Emest E. Williams.

Throughout the tropical lowlands of
Middle and South America there are
many species of small, yellowish-tan or
brown Hyla that have pond-dwelling
tadpoles with xiphicercal tails and re-
duced mouthparts. All of these species
that have been studied karyologically
have 30 chromosomes. There are few
species of Hyla in the Andes, but some of
these (H. carnifex, H. columbiana, and H.
variabilis) resemble, superficially at
least, the lowland taxa.

In 1975 we discovered a previously
unknown species of small Hyla presum-
ably related to H. columbiana. This find
prompted our investigation of the tax-
onomy of the montane species and the
definition of a Hyla columbiana group.
In this paper we present the results of

34

these studies and propose a phylogenetic
arrangement of the group and its rela-
tives.

MATERIALS AND METHODS

Systematic aspects of this study are
based on the examination of 591 pre-
served specimens, 14 skeletons, 12 lots of
tadpoles, and | clutch of eggs belonging
to species in the Hyla columbiana group.
Recordings were made on a Uher-4000
tape recorder and analyzed on a Vibra-
lyzer (Kay Electric Company). All mea-
surements of morphological features and
calls were taken as described by Duell-
man (1970), and webbing formulae were
ascertained following Savage and Heyer
(1967). Tadpoles were staged according
to Gosner’s (1960) tables. Statistical an-
alyses of morphometrics were based on
measurements of uniformly preserved
specimens from the vicinities of the type
localities. Pairwise comparisons of mea-
surements and proportions were accom-
plished by one-way analyses of variance
(Sokal and Rohlf, 1969). Results of F_.
tests indicated normal distribution of the
data. In a multiple comparison of the
means, the Student-Newman-Keuls test
was used to determine levels of signifi-
cance.

All specimens are referred to by the
following abbreviations: AMNH = Amer-
can Museum of Natural History; ANSP =
Academy of Natural Sciences, Philadel-
phia; BMNH = British Museum (Natural
History); FMNH = Field Museum of
Natural History; INDERENA = Instituto
Nacional de Recursos Naturales, Bogota;
KU = Museum of Natural History, Uni-
versity of Kansas; LACM = Museum of
Natural History, Los Angeles County;
MCZ = Museum of Comparative Zo-
ology, Harvard University; NHMW =
Naturhistorisches Museum, Wien;
NHRM = Naturhistoriska Riksmuseet,
Stockholm; SMF = Senckenberg Mu-
seum, Frankfurt-am-Main; UMMZ =
Museum of Zoology, University of Michi-

Advances in Herpetology and Evolutionary Biology

gan; USNM = National Museum of
Natural History; ZMB = Zoologisches
Museum, Berlin; ZSM = Zoologisches

Sammlung, Munchen.

‘TAXONOMY

The Hyla columbiana Group

Definition. 1) Moderate sexual di-
morphism in size; snout-vent lengths of
males about 85% those of females; 2)
snout short, blunt; 3) tympanum visible;
tympanic annulus distinct or not; 4)
hands and feet moderately webbed; 5)
axillary membrane moderately extensive;
6) thoracic glands absent; 7) calcars and
tarsal tubercles absent; ulnar tubercles
present or absent; 8) males having single,
median, subgular vocal sacs and no
nuptial excrescences; 9) dorsum tan or
brown, with or without dark dorsal mark-
ings and pale dorsolateral stripes; 10)
pale labial stripe or spots present or not;
11) axilla and groin marked with yellow,
orange, red or bluish gray in life; 12) anal,
ulnar, and tarsal stripes absent; 13) iris
bronze to reddish copper; 14) quadrato-
jugal absent; 15) prevomers not articu-
lating with maxillary arch, bearing small,
posteromedially inclined dentigerous
processes; 16) crista parotica poorly to
moderately ossified; 17) zygomatic ramus
of squamosal extending about one-third
distance to maxillary arch; 18) pterygoid
poorly developed, anterior ramus articu-
lating weakly with unexpanded pars
palatina of maxillary at level of midorbit;
medial ramus widely separated from otic
region of neurocranium; 19) sacral dia-
pophyses moderately expanded (45-60°),
anterior edge forming acute angle with
longitudinal axis of body; 20) ilia lacking
obvious crests; preacetabular angle ob-
tuse; 21) tadpoles having ovoid bodies
and xiphicercal tails with moderately
deep fins barely extending onto body; 22)
larval mouths terminal with two rows of
small papillae ventrally, robust serrate
beaks, and one upper and two lower rows

HYLA COLUMBIANA GROUP: Duellman and Trueb 35

of denticles; 23) mating call consisting of
short, moderately pitched notes, fol-
lowed or not by shorter secondary notes;
24) diploid chromosome number 30.
Content. Three species, which may be
identified by the following key:

1. Broad pale dorsolateral stripe present; groin
bluish gray mottled with black in life ........
00:0 ONO ER Ronee tne ecoo00gpcccc0e Jas (IRTESUONS
— Pale dorsolateral stripe absent; groin yellow,
orange, or red in life ....:+......266...0005 2
2. Small yellow spots on upper lip; venter bright
Glilonne Tin INS 3 leelaoree eee eee H. carnifex
— No yellow spots on upper lip; venter creamy yel-
oxy tin INS okie H. columbiana

Distribution. Moderate elevations
(975—2,580 m) on the Pacific slopes of the
Cordillera Occidental in Ecuador and in
the upper Rio Cauca and Magdalena
valleys and the northern part of the Cor-
dillera Central in Colombia (Fig. 1).

Comment. The chromosome number is
known only for Hyla carnifex (KU
173114), and tadpoles are known only for
H. carnifex and H. columbiana. The
populations from the northern part of the
Cordillera Central, Colombia, are tenta-
tively assigned to H. carnifex.

Following is an account of each of the
species in the Hyla columbiana group.
The results of the analyses of the mor-
phometric data are presented in Table 1.
For illustrations of the adult and larva of
H. carnifex, see Duellman (1969).

Hyla carnifex Duellman

Hyla carnifex Duellman, 1969: 242 (Holotype.—
KU 117993 from Tandapi, 1,460 m, Provincia de
Pichincha, Ecuador); 1974: 7; 1977: 43.

Hyla bogerti Cochran and Goin, 1970: 261 (Holo-

type.—USNM 118731 from Medellin, Departa- -

mento de Antioquia, Colombia). Synonymy fide
Duellman (1974: 7).

Hyla charlesbogerti Goin, 1970: 788 (Substitute
name for Hyla bogerti Cochran and Goin, 1970).
Synonymy fide Duellman (1974: 7).

Diagnosis. 1) Size small, snout-vent
length in males 24.6 to 27.7 mm, in fe-
males 29.2 to 32.5 mm; 2) head short,
28.5 to 31.3% of snout-vent length; 3)

Figure 1.

Western Colombia and Ecuador showing
distributions of Hyla carnifex (®@), H. columbiana (@),
and H. praestans (a). The dashed line is the 1,000-m
contour; shaded areas are above 3,000 m.

snout round in profile; 4) interorbital dis
tance narrow, 29.5 to 35.4% of head width;
5) skin on dorsum finely shagreened,
nearly smooth; 6) ulnar tubercles distinct;
7) distal subarticular tubercle on fourth
finger strongly bifid; 8) subanal tubercles

36 Advances in Herpetology and Evolutionary Biology

TABLE 1. STATISTICAL COMPARISON OF MORPHOMETRIC CHARACTERS OF TOPOTYPIC MALES OF THREE SPECIES IN
THE HYLA COLUMBIANA GROUP.

First line = mean + | standard deviation; second line = observed range. Means of all characters of H.

columbiana significantly different (P < 0.05) from those of the other species; significant differences (P <

0.05) between H. carnifex and H. praestans are indicated by an asterisk*.

Character H. carnifex

N 20

Snout-vent length (SVL) 26.1 + 0.90*
24.6 = X70
Tibia/SVL 0.455 + 0.015
0.433 — 0.481
Foot/SVL 0.424 + 0.018
0.395 — 0.455
Head length/SVL 0.199 + 0.008
0.285 — 0.313
Head width (HW)/SVL 0.327 + 0.010
0.310 — 0.349
Interorbital/HW 0.332 + 0.015
0.195 — 0.354
Total prevomerine teeth 9.1 ste ee Le
0 = Ill

H. columbiana H. praestans

20 8

27.5 + 1.14 28.8 2 176%
25.8 — 99.3 26.0 —= ils)

0.482 + 0.017 0.455 + 0.016
0.454 — 0.515 0.424 + 0.471
0.456 + 0.014 0.424 + 0.026
0.430 — 0.479 0.372 — 0.452
0.339 + 0.013 0.346 + 0.009
0.314 — 0.356 0.334 — 0.360
0.312 + 0.010 0.331 + 0.006
0.290 — 0.332 0.324 — 0.340
0.416 + 0.032 0.319 + 0.021
0.358 — 0.471 0.300 — 0.347
Woo) sz L183 8.6 eel yoy) Ee
6 — — 10

10 6

large; 9) dorsum tan with brown markings
and usually with creamy yellow suborbi-
tal spot; 10) venter and vocal sac bright
yellow, with or without brown spots on
belly; 11) groin deep yellow orange; 12)
mating call consisting of primary note fol-
lowed by two or three secondary notes.

The presence of small, yellow spots on
the upper lip and the bright yellow ven-
ter immediately distinguishes H. carni-
fex from other members of the group.
Members of the Hyla parviceps group
have larger suborbital spots and have
white or gray venters and either orange
spots on the ventral surfaces of the
shanks or yellow spots on the anterior or
dorsal surfaces of the thighs. The dorsal
color pattern of H. carnifex resembles
that of H. minuta, a species with a creamy
yellow venter and white anal and heel
stripes.

Description. Body robust; head small,
noticeably narrower than body; snout
short, its length about equal to diameter
of eye, round in dorsal view and in pro-
file; nostril about four-fifths distance
from eye to tip of snout; canthus rostralis
rounded, barely evident; loreal region
slightly concave; lips moderately thick,
not flared; internarial area noticeably

depressed; interorbital area slightly con-
vex, much wider than eyelid; supra-
tympanic fold short, moderately weak,
obscuring upper part of tympanum; tym-
panic annulus indistinct; tympanum
separated from eye by distance slightly
greater than diameter of tympanum.
Axillary membrane short, extending
about one-fourth length of upper arm;
forearm moderately slender, bearing row
of distinct ulnar tubercles; fingers short,
fourth distinctly longer than second;
discs round, half again as wide as digits;
fingers webbed basally; webbing absent
between first two fingers; webbing for-
mula for other fingers II2—3- III3°-—2IV;
subarticular tubercles moderately large,
low, subconical; distal tubercle on fourth
finger strongly bifid; supernumerary
tubercles small, subconical, numerous on
basal segments of digits; palmar tubercle
flat, bifid; prepollical tubercle flat, ellip-
tical; nuptial excrescences absent. Hind
limbs moderately short, slender; calcars,
tarsal folds and tubercles absent; inner
metatarsal tubercle low, flat, elliptical;
outer metatarsal tubercle evident in some
specimens; toes moderately long, third
and fifth subequal in length; toes about
three-fourths webbed; webbing formula

HYLA COLUMBIANA GROUP: Duellman and Trueb 37

ee OT (144 —9-\l_— (1-—9)
IV(2-—2—IV;_ _subarticular tubercles
small, round; supernumerary tubercles
minute, present in single row on proxi-
mal segment of each digit.

Skin on dorsum weakly shagreened,
nearly smooth; skin on throat, belly, and
proximal posteroventral surfaces of
thighs granular; skin on other surfaces
smooth. Anal opening directed posteri-
orly at upper level of thighs; anal flap
short; subanal tubercles large. Tongue
elliptical, shallowly notched posteriorly,
and barely free behind. Dentigerous
processes short, posteromedially in-
clined, widely separated medially, be-
tween small ovoid choanae, bearing 34
to 5-6 teeth for totals of 7-11 (x = 9.2, 20
6 6) and 4—5 to 6-6 teeth for totals of 9-12
(x = 10.4, 15 22). Vocal slits short, ex-
tending from near midlateral base of
tongue towards angles of jaws. Vocal sac
large, single, median, subgular.

Coloration in preservative. Dorsum of
head, body, and limbs dark brown or
grayish brown with darker brown flecks;
middorsal dark brown blotch extending
from occiput or eyelids at least to sacral
region in all females (30) and most males
(83 of 100); venter cream with or without
dark brown spots; variation in 100 males
and 30 females—plain (68%, 27%), spots
on posterior part of belly (24%, 50%),
spots on belly and chest (8%, 13%), spots
on belly, chest, and throat (0%, 10%) (Fig.
2). Flanks dark brown, usually enclosing
cream lateral streak; groin mottled dark
brown and cream; posterior thighs brown
with longitudinal cream stripe; axilla,
webbing, and spots on upper lip cream;
variation in number of small spots on
upper lip in 130 adults—none 6.9%, one
70.0%, two 17.7%, three 3.1%, four 2.3%.

Coloration in life. Dorsum of head and
body pale creamy tan or grayish tan with
median brown or grayish-brown blotch
and numerous dark brown flecks; dorsal
surfaces of limbs, exclusive of thighs,
pale brown with or without faint darker
brown transverse bars; upper lip brown,
usually with small yellow spots, at least

one below eye; axilla, posterior flanks,
and anterior and posterior surfaces of
thighs deep yellow or orange bordered
by dark brown or black; ventral surfaces
of shanks, inner surfaces of feet, web-
bing, and ventral surfaces of upper arms
deep yellow or orange (deepest colors
only in larger individuals; webbing
orange-red in some large females); throat,
belly, and median part of flanks bright
yellow with or without faint brown, lav-
ender, or gray spots; iris gray, flecked
with reddish bronze.

Tadpoles. The tadpoles were de-
scribed and illustrated by Duellman
(1969). In structure and coloration they
are much like those of H. columbiana
(described and figured herein), except
that the upper beak has distinct lateral
processes and white flecks are absent on
the body and tail.

Mating call. The mating call of Hyla
carnifex consists of a monophasic pri-
mary note followed by two or three
shorter secondary notes—“Wraah-ack-
ack.” Analysis of recordings of two indi-
viduals shows that call groups are re-
peated at a rate of about 14 calls per
minute. Primary notes have durations of
0.20 to 0.36 sec. The fundamental fre-
quency is at about 500 Hz, and the domi-
nant frequency is the fifth or sixth
harmonic at about 2,500 or 3,000 Hz
(Fig. 3).

Distribution. This species is definitely
known from the Pacific slopes of the
Cordillera Occidental in Ecuador, where
it occurs at elevations of 1,140 to 2,150 m.
Northern populations assigned to this
species are known from elevations of
1,480 to 2,580 m in the northern part of
the Cordillera Central and the eastern
slopes of the Cordillera Occidental in
Departamento de Antioquia, Colombia

(Fig. 1).

Remarks. The status of the populations
in northern Colombia named as Hyla
bogerti by Cochran and Goin (1970) and
recognized as H. carnifex by Duellman
(1974) is questionable. Although many
specimens are available from Departa-

| TRUEB 1/990

Figure 2. Ventral color patterns. Hyla carnifex: A. KU 164301, 3. B. KU 164238, ?. Hyla praestans: C. KU
169578, 3. D. KU 169575, 3. Hyla columbiana: E. KU 169470, ¢. F. KU 169490, 5. G. KU 181165, d. H. KU

169472, °.

mento de Antioquia, Colombia, most are
faded and poorly preserved, thereby
making detailed morphometric compari-
sons useless. We have seen only one liv-
ing specimen from Antioquia, and it had
the distinctive bright yellow venter and
flank pattern of H. carnifex from Ecua-
dor. Although a gap of nearly 700 km
separates the Ecuadorian localities from

those in Colombia, we anticipate that the
species will be discovered along the in-
tervening Pacific slopes of the Cordillera
Occidental in Colombia. Until series of
fresh, well-preserved specimens and
recordings of mating calls are available
from localities in northern Colombia, we
can say only that living frogs from the
widely separated regions have the same

HYLA COLUMBIANA GROUP : Duellman and Trueb 39

KILOHERTZ

TIME IN SECONDS

Figure 3. Audiospectrograms of mating calls of: A. Hyla carnifex (KU Tape 1237, 19°C). B. H. columbiana (KU
Tape 1337, 17°C). C. H. praestans (KU Tape 1348, 17.5°C). Narrow band (45 Hz) analysis.

distinctive coloration; therefore, we as-
sign them to the same species.

Only a few specimens of H. carnifex
have been found by day—under the bark
of logs, in bromeliads, and in axils of

leaves of elephant-ear plants (Xantho-
soma). Calling males have been found
between January and September. Males
call from low (<50 cm), emergent herbs
and grasses in marshy, temporary pools

40

in clearings in cloud forest. Tadpoles
have been found in these pools from
April to July, and metamorphosing young
have been collected from May until
August. Five young with tail stubs of <2
mm from Tandapi, Ecuador, collected on
18 July 1970, have snout-vent lengths of
12.7 to 14.7 (x = 13.8) mm. Three young
collected on 3 April 1975, 5 km east-
southeast of Chiriboga, Ecuador, have
snout-vent lengths of 15.0 to 15.5
(x = 15.3) mm, and four obtained on 11
May 1975 at the same locality have snout-
vent lengths of 16.1 to 18.2 (x = 16.9)
mm. In recently metamorphosed young,
the dorsum is tan with a bronze tint; the
anterior and posterior surfaces of the
thighs and ventral surfaces of the shanks
are dark gray, and the throat and belly are
grayish white. In half-grown individuals
(20-22 mm) the belly is white with a faint
tint of yellow on the chin and chest; the
flanks and anterior and posterior surfaces
of the thighs are dark brown. A middorsal
dark blotch is evident in these individu-
als, but the bright yellow or orange colors
on the limbs and in the axilla and groin,
and the spots on the upper lip do not
develop until the frogs attain sexual
maturity (snout-vent lengths of >23 mm
in males and >27.5 mm in females).

Hyla columbiana Boettger

Hyla columbiana Boettger, 1892: 41 (Lectotype.—
SMF 2365 [designated by Mertens, 1967: 4] from
Popayan, Departamento de Cauca, Colombia);
Nieden, 1923: 261; Cochran and Goin, 1970: 264;
Duellman, 1977: 48.

Hyla variabilis Boulenger, 1896: 20 (Syntypes.—
BMNH 1947.2.13.15-21; FMNH 3565; MCZ
2606; UMMZ 46464, 51269, 58908 [8 specimens];
USNM 71115 from Cali, Departamento de Valle,
Colombia); Nieden, 1923: 262; Cochran and
Goin, 1970: 258; Duellman, 1977: 108. NEW
SYNONYMY.

Diagnosis. 1) Size small, snout-vent
length in males 25.8 to 29.3 mm, in fe-
males 30.6 to 35.4 mm; 2) head moder-
ately long, 31.4 to 35.6% of snout-vent
length; 3) snout round in profile; 4) in-
terorbital distance wide, 35.8 to 47.1% of

Advances in Herpetology and Evolutionary Biology

head width; 5) skin on dorsum smooth
with scattered, small tubercles; 6) ulnar
tubercles low, indistinct; 7) distal sub-
articular tubercle on fourth finger weakly
bifid; 8) subanal tubercles small; 9) dor-
sum tan with brown markings; 10) venter
creamy yellow with variable amount of
brown or black markings (flecks to reticu-
lations); vocal sac bright yellow in life;
11) groin yellow-orange to tomato-red in
life; 12) mating call consisting of groups
of 2 to 6 notes of equal duration.

The combination of small tubercles on
the dorsum and absence of distinct labial
spots and dorsolateral stripes distin-
guishes H. columbiana from other mem-
bers of the group. The deep yellow,
orange, or red axilla and groin and the
extensive flecking or reticulations on the
venter distinguish H. columbiana from
all other species of small Neotropical
Hyla.

Description. Body moderately robust;
head as wide as body; snout moderately
long, half again length of eye, round in
dorsal view and in profile; nostril about
two-thirds distance from eye to tip of
snout; canthus rostralis acutely rounded;
loreal region barely concave; lips thick,
not flared; internarial area slightly de-
pressed; interorbital area flat, nearly
twice width of eyelid; supratympanic
fold moderately weak, curving poster-
oventrally, obscuring upper edge of tym-
panum; tympanic annulus _ distinct;
tympanum separated from eye by dis-
tance only slightly greater than diameter
of tympanum.

Axillary membrane moderately long,
extending about two-fifths length of up-
per arm; forearm moderately robust; ul-
nar tubercles low, indistinct; fingers
short, fourth slightly longer than second;
discs round, slightly wider than digit;
fingers webbed basally; webbing absent
between first two fingers; webbing for-
mula for other fingers II2-—3~III2%—
21V; subarticular tubercles small, sub-
conical; distal tubercle on fourth finger
weakly bifid; supernumerary tubercles
small, subconical, numerous on _ basal
segments; palmar tubercle low, flat,

HYLA COLUMBIANA GROUP: Duellman and Trueb 41

weakly bifid; prepollical tubercle flat, el-
liptical; muptial excrescences absent.
Hind limbs moderately long, slender;
calcars, tarsal folds and tubercles absent;
inner metatarsal tubercle flat, narrowly
elliptical; outer metatarsal tubercle ab-
sent; toes moderately long, third and fifth
subequal in length; toes about three-
fourths webbed; webbing formula I(1-—-
1%)—21112—(2-2*T11(1t — 142)—21TV2—
(1-1*)V; subarticular tubercles small,
subconical; supernumerary _ tubercles
minute, subconical, numerous on proxi-
mal segments of each digit.

Skin on dorsal surfaces of head, body,
and limbs smooth with scattered, small
tubercles; skin on throat, belly, and
proximal posteroventral surfaces of
thighs coarsely granular. Anal opening
directed posteriorly at upper level of
thighs; anal flap short; subanal tubercles
small. Tongue ovoid, barely notched
behind, only slightly free posteriorly.
Dentigerous processes of prevomers
small, posteromedially inclined, moder-
ately separated medially, between pos-
terior borders of small, ovoid choanae,
bearing 3-3 to 5-5 teeth for totals of 6-10
(x = 7.3, 20 dd) and 4-5 to 6~7 teeth for
totals of 9-13 (¥ = 11.0, 3 2 2). Vocal slits
long, extending from midlateral base of
tongue to angles of jaws. Vocal sac large,
single, median, subgular.

Coloration in preservative. Dorsal sur-
faces of head, body, and limbs tan to
brown with three types of patterns of
darker brown markings in 40 recently
preserved specimens: 1) middorsal
blotch extending from eyelids to middle
of dorsum and a smaller blotch usually
evident in sacral region (50%); 2) narrow,
diffuse stripe middorsally (25%); and 3)
irregular flecks and/or _ reticulations
(25%). Dorsal surfaces of limbs plain,
flecked, or barred (in some specimens
having dorsal blotches). Pale canthal, la-
bial, and dorsolateral stripes absent; faint
dark canthal and postorbital stripes evi-
dent in some specimens. Venter creamy
tan with a variable amount of flecking and
reticulation in 40 specimens (Fig. 2): 1) no
flecks (35%); 2) flecks and reticulations

only in groin (30%); 3) reticulations in
groin; flecks on belly, especially pos-
teriorly (20%); and 4) reticulations in
groin and on belly (15%). All females and
some males in last category; females also
having reticulations on throat. Anterior
and posterior surfaces of thighs brown or
tan with or without irregular brown mark-
ings; webbing and discs brown.
Coloration in life. At night dorsum yel-
lowish tan; axilla, groin, proximal an-
terior surfaces of thighs, and inner sur-
faces of shanks pink to orange or yellow.
By day, dorsum yellow, bronze, tan, or
brown, with or without distinct brown
blotches; venter yellow, with or without
dark brown flecks and/or reticulations on
belly; axilla, groin, proximal anterior sur-
faces of thighs, and inner surfaces of
shanks deep yellow, orange, or red. Iris
reddish copper. In some individuals, a
faint dark canthal stripe and a dark brown
postorbital stripe extending onto the
anterior flanks are present; also, in some
individuals a diffuse creamy yellow
suborbital bar is present (Fig. 4).
Tadpoles. Among a series of tadpoles
(KU 170202) in developmental stages 25
through 41, a typical tadpole in stage 38
is described. Body length 13.2 mm; total
length 35.5 mm; body as wide as deep,
sides nearly parallel for most of the
length of body posterior to eyes, deepest
posteriorly; snout in dorsal view broadly
rounded, in _ profile more _ acutely
rounded; eyes large, directed laterally;
nostril about midway between eye and
tip of snout, directed anteriorly; spiracle
sinistral, directed posterodorsally just
below midline at point about two-thirds
length of body; anal opening dextral.
Caudal musculature moderately robust
anteriorly, tapering to attenuated point
posteriorly; fins deepest at about mid-

length of tail, tapering abruptly to xiphi-

cercal tail; depth of either fin at mid-
length of tail about equal to depth of
caudal musculature; dorsal fin barely
extending onto body (Fig. 5).

Mouth small (about one-fourth greatest
width of body), terminal, directed an-
teroventrally; upper lip bare; two rows of

42 Advances in Herpetology and Evolutionary Biology

Figure 4. Hyla columbiana, KU 169463, 6, 28.1 mm snout-vent length.

TRUEB 1980

Figure 5. Tadpole of Hyla columbiana, KU 170202, 35.5 mm total length.

small, pointed papillae laterally and ven-
trally; lateral labial folds absent; beaks
moderately massive, bearing fine serra-
tions; upper beak broadly arch-shaped
with no narrowed, lateral processes;
lower beak broadly V-shaped; one upper
and two lower complete rows of den-
ticles; upper row just inside upper lip;
second lower row having smallest den-
ticles.

In preservative, dorsum of body and
snout dull brown; sides and venter trans-
parent with bluish sheen and numerous
white flecks; caudal musculature orange-
tan; caudal fins transluscent; caudal
musculature and fins bearing minute
white flecks; in larger specimens (stage
30 and beyond), brown blotches on
caudal fins, becoming larger and darker
in larger tadpoles. In life, dorsum olive-

HYLA COLUMBIANA GROUP: Duellman and Trueb 43

green with broad, tan dorsolateral stripe;
tail cream with minute brown flecks in
small individuals and brown spots on fins
in larger individuals; venter silvery
white; iris pale bronze.

Mating call. The mating call of Hyla
columbiana consists of short series of
similar notes. Analysis of recordings of
five individuals reveals that each call
consists of groups of 2 to 6 (x = 3.6) notes
repeated at a rate of 12 to 16 (x = 14.4)
notes per minute. Each note has a dura-
tion of 0.1 to 0.3 (¥ =0.16) sec. The
fundamental frequency is at 240 to 320
(x = 265) Hz. The eighth or ninth har-
monic is emphasized, giving a dominant
frequency of 2,000 to 2,560 (x = 2,480)
Hz (Fig. 3).

Remarks. Examination of the type
specimens of Hyla columbiana and H.
variabilis and study of living and pre-
served specimens from the regions of the
type localities of the two named taxa, as
well as analyses of mating calls, reveal
the presence of only one species in the
upper Rio Cauca Valley. Cochran and
Goin (1970: 260, 266) noted colors of a
few living specimens from Cali and
Popayan; they emphasized that the axilla
was red in specimens from Popayan and
yellow in specimens from Cali. In a ser-
ies of 38 adults collected 8 km east of
Popayan, the color of the axilla varied
from deep yellow to orange and tomato-
red, in living frogs from Lago de Calima
(north of Cali) the axilla was pink.

Notable ontogenetic, sexual, and geo-
graphic variation exists in the amount of
dark pigmentation on the venter. Juve-
niles lack dark pigment ventrally, where-
as large adult females of all populations
have reticulate venters, a condition also
seen in many males from Cali and a few
males from Popaydan (Fig. 2).

We have observed this species only in
disturbed areas that formerly supported
cloud forest. Three individuals were
found in Agave plants by day. All others
were in shallow, grassy marshes or pools
at night. Males call from clumps of grass
in water or from grasses within 5 cm of

the surface. On 20 September 1974 we
found a single clutch of eggs attached to
grass on the surface of the water in a tem-
porary pool. Tadpoles were found in the
same pool.

Five recently metamorphosed young
(KU 170201) have snout-vent lengths of
13.7 to 18.6 (= 16.1) mm. In life, the
dorsum was bronze tan; the anterior and
posterior surfaces of the thighs were dark
brown. The upper arms and suborbital
bar were creamy yellow, and the throat
and belly were white. The iris was red-
dish bronze. In preservative, the juve-
niles are pale tan dorsally and creamy
white ventrally; there is no indication of
flecks or reticulations on the belly or in
the groin.

Distribution. Hyla columbiana _ is
widespread in the upper Rio Cauca Val-
ley, Colombia, where it occurs on the val-
ley floor at elevations of 975 to 1,000 m
and on the lower eastern slopes of the
Cordillera Occidental and western slopes
of the Cordillera Central to elevations of
2,250 m. It also occurs in the upper
drainage of the Rio Patia, where it is
known from elevations of 1,800 to 2,350 m
(Fig. 1).

Hyla praestans new species

Holotype. KU 169575, an adult male, from the
Parque Arqueologico San Agustin, 3 km southwest
of the village of San Agustin, Departamento de
Huila, Colombia (1°53'N, 76°16’W), 1,750 m, one of
a series obtained on 27 May 1975 by William E.
Duellman, John E. Simmons, and Linda Trueb.

Paratypes. KU 169574, 169576-80, MCZ 100216,
INDERENA (2 specimens) collected with the holo-

type.
Referred Specimens. ICN 7556-58 from 7.4 km

(by road) NW of San José de Isnos, Departamento
de Huila, Colombia, 1,970 m (John D. Lynch, per-
sonal communication).

Diagnosis. 1) Size small, snout-vent
length in males 26.0 to 31.5 mm, females
31.2 mm; 2) head moderately long, 33.4
to 36.0% of snout-vent length; 3) snout in
profile truncate, rounded above; 4) in-
terorbital distance narrow, 30.0 to 34.7%
of head width; 5) skin on dorsum finely

44. Advances in Herpetology and Evolutionary Biology

shagreened, nearly smooth; 6) ulnar
tubercles absent; 7) distal subarticular
tubercle on fourth finger strongly bifid;
8) subanal tubercles small; 9) dorsum tan
with creamy yellow labial stripe and
orange-tan dorsolateral stripe; 10) venter
and vocal sac creamy white with brown
flecks on belly; 11) groin pale bluish gray
mottled with black; 12) mating call con-
sisting of primary note followed by one or
two secondary notes.

Hyla praestans differs from the other
members of the Hyla columbiana group
by having pale labial and dorsolateral
stripes; the latter is characteristic of
females of many species in the Hyla
parviceps group, all of which have yel-
low spots on the thighs or orange spots on
the ventral surfaces of the shanks.
Another distinguishing feature of H.
praestans is the bluish-gray groin with
black mottling, a pattern occurring in
some populations of H. labialis, a much
larger frog that is uniformly dark green

dorsally.
Description. Body _ robust; head
slightly narrower than body; snout

moderately short, only slightly longer
than diameter of eye, round in dorsal
view, truncate and rounded above in
profile; nostril about two-thirds distance
from eye to tip of snout; canthus rostralis
rounded; loreal region slightly concave;
lips moderately thick, not flared; in-
ternarial area slightly depressed; inter-
orbital area flat, much wider than eyelid;
supratympanic fold short, weak, obscur-
ing upper part of tympanum; tympanic
annulus distinct; tympanum separated
from eye by distance one third greater
than diameter of tympanum.

Axillary membrane short, extending
about one-fourth length of upper arm;
forearm moderately slender, lacking
ulnar tubercles; fingers moderately long,
fourth slightly longer than second; discs
round, slightly wider than digit; fingers
webbed basally; webbing absent be-
tween first two fingers; webbing formula
for other fingers 112-—(3--3+)III3—
2¥%2IV; subarticular tubercles moderately

large, subconical; distal tubercle on |
fourth finger strongly bifid; supermumer- |
ary tubercles low, flat; palmar tubercle
low, flat, ovoid; prepollical tubercle flat,
elliptical; nuptial excrescences absent.
Hind limbs moderately long, slender;
calcars, tarsal folds and tubercles absent;
inner metatarsal tubercle small, flat, el-
liptical; outer metatarsal tubercle absent;
toes moderately long, fifth barely longer |
than third; toes about two-thirds webbed;
webbing formula I2-—2II1—(2 -2)II1(1-
144)—2- —2)IV(2- —24%2)—(1-14%)V;_ sub-
articular tubercles small, subconical; su- |
pernumerary tubercles absent.

Skin on dorsum weakly shagreened,
nearly smooth; skin on throat, belly, and |
proximal posteroventral surfaces of
thighs weakly granular; skin on other sur-
faces smooth. Anal opening directed pos-
teriorly at upper level of thighs; anal flap
short; subanal tubercles small. Tongue
elliptical, shallowly notched posteriorly,
barely free behind. Dentigerous proces-
ses of prevomers small, posteromedially
inclined, narrowly separated medially,
between small, ovoid choanae, bearing
3-3 to 5—5 teeth for totals of 6-10 (x = 8.6)
teeth in males. Vocal slits long, extending
from midlateral base of tongue to angles
of jaws. Vocal sac large, single, median,
subgular.

Coloration in preservative. Dorsal sur-
faces of head, body, and limbs brown
with darker brown flecks on body and
limbs and dark brown longitudinal
streaks of variable size and distinctness
on body. Narrow, cream to orange-tan
canthal line from tip of snout to eye;
broad, cream to orange-tan dorsolateral
strip from eye to groin; flanks dark
brown. Venter cream with variable
amount of brown flecking or spotting on
throat, belly, and thighs (Fig. 2); anterior
and posterior surfaces of thighs creamy
tan with dark brown mottling; webbing
and discs brown.

Coloration in life. At night, dorsum tan
with creamy yellow dorsolateral stripes.
By day, dorsum and flanks dark brown
with orange-tan dorsolateral stripes,

HYLA COLUMBIANA GROUP: Duellman and Trueb 45

creamy tan canthal stripes, and creamy
yellow labial stripe; groin mottled bluish
gray and black; posterior surfaces of
thighs dark brown; throat and_ belly
creamy white with brown flecks; ventral
surfaces of limbs tan; iris coppery brown
(Fig. 6).

Measurements of holotype (in mm).
Snout-vent length 28.3, tibia length 13.3,
foot length 12.6, head length 9.8, head
width 9.3, eye-nostril 2.5, interorbital dis-
tance 3.2, eyelid width 2.1, horizontal
diameter of eye 2.8, horizontal diameter
of tympanum 1.3.

Tadpoles. Unknown.

Mating call. The call of Hyla praestans
consists of a primary note followed by
one or two, shorter, secondary notes.
Analysis of a single recording shows that
the call groups are repeated at a rate of
4.5 calls per minute. The duration of the

primary note is 0.08 sec. The fundamen-
tal frequency is at 450 Hz, and the fifth
harmonic at 2,250 Hz is dominant (Fig.
&)),

Distribution. This species is known
from elevations of 1,750 to 1,970 m on the
lower eastern slopes of the Cordillera
Central next to the upper Rio Magdalena
Valley, Colombia (Fig. 1).

Remarks. The region of the type local-
ity is humid lower montane forest with
bamboo, Heliconia, and some bromeli-
ads. The exact type locality is a small,
temporary pond immediately behind
(+100 m) the Museo Arqueologico San
Agustin. Males were calling from a tangle
of brambles over (up to 1 m above) the
water at night following a light rain (13
mm).

Etymology. The specific name, praes-
tans, is Latin meaning preeminent or

/980

Figure 6. Hyla praestans, KU 169574, 3, 27.3 mm snout-vent length.

46

distinguished and is proposed in recogni-
tion of Ermest E. Williams’s recognized
preeminence in the field of herpetology.

PHYLOGENETIC RELATIONSHIPS

The phylogenetic relationships of
groups of taxa may be inferred by deter-
mining shared-derived characters in de-
fining monophyletic taxa; determination
of sequential sets of shared-derived
character states provides a stepwise as-
sociation of character states defining
branching sequences in a phylogenetic
arrangement and classification reflecting
these phylogenetic relationships. This
approach has become known as phylo-
genetic systematics (see Wiley, 1980, for
terms and methodology).

Although we can provide a reasonable
hypothesis of the phylogenetic relation-
ships of the Hyla columbiana group with
other groups of Hyla, the distribution of
character states among the species within
the group and the absence of meaningful
interpretation of evolutionary trends of
many characters, such as skin texture and
coloration, defies a reconstruction of the
phylogeny of the three species. Using 10
of the characters listed in the diagnoses
(all except snout-vent length and propor-
tion of head width) plus six cranial
characters (nature of frontoparietal fon-
tanelle, degree of ossification of sphen-
ethmoid, shape of parasphenoid alae, ser-
ration of coronoid, and amount of separa-
tion of prevomers and of palatines), we
found that there was little concordance in
shared character states. Hyla carnifex
shares six states with H. columbiana and
seven with H. praestans, which shares
only three with H. columbiana. Thus,
even a phenetic analysis reveals only that
H. columbiana has more unique charac-
ter states that either of the other two
species.

Various apparently related groups of
Neotropical hylid frogs have been de-
fined as a matter of convenience by
Bokermann (1964), Duellman (1970,
1974), and Duellman and Crump (1974).

Advances in Herpetology and Evolutionary Biology

Informally recognized assemblages of
species include the leucophyllata (6
species), marmorata (4), microcephala
(+12), minuta (2), and parviceps (7)
groups. For the purposes of the following
discussion, we add H. labialis and the
columbiana group.

In attempting to ascertain the phylo-
genetic relationships among _ these
groups, we utilized eight character states;
2 from adults, 5 from larvae and the
karyotype. These states were selected
because they are ones in which we place
confidence in our knowledge of phylo-
genetic trends within the Hylidae. Other
characters are highly variable within, as
well as between, groups, or are restricted
to members of only one group; therefore,
they are useless in determining inter-
group relationships. The characters, their
states, and direction of evolutionary
change (0 = primitive state; 1 = derived
state; 2 = secondarily or independently
derived state) are as follows:

A. Chromosome number. Hylid frogs
have diploid numbers of 22, 24, 26, 28, or
30 chromosomes. Bogart (1973) showed
that the primitive number of chromo-
somes in leptodactylids, the presumed
ancestors of hylids, is 26—a number
occurring in phyllomedusine, some
pelodryadine, and some amphignatho-
dontine hylids. Most hyline frogs have 24
chromosomes, but all species so far ex-
amined of the groups of Hyla listed above
have 30 chromosomes, a number un-
known in other groups of hylids. Bogart
(1973) and Morescalchi (1973) suggested
that the evolution of 30 chromosomes
occurred by centric fission, and Bogart
(1973) noted variable numbers of pairs of
telocentric chromosomes in those species
having a diploid number of 30. Because
this number is unique to the groups of
Hyla under consideration, we assume
that this chromosome number is a
uniquely derived character state in these
frogs. The coded character states and
polarity are:

0 = <30 chromosomes; | = 30 chromo-
somes

O— 1

HYLA COLUMBIANA Group: Duellman and Trueb 47

B. Quadratojugal arch. Primitively in
anurans, the quadratojugal forms a com-
plete arch between the quadrate and the
maxillary. This bone is reduced or absent
in a number of primitive and advanced
families (Lynch, 1973), including some
leptodactylids (Lynch, 1971) and some
hylids (Duellman, 1970). The presence of
a complete quadratojugal arch is primi-
tive (Lynch, 1973; Trueb, 1973). The
coded character states and polarity are:

0 = arch complete; 1 = quadratojugal

reduced or absent

0-1

C. Nuptial excrescences. Nuptial ex-
crescences on the prepollex of breeding
males usually are associated with species
that amplex in water, and the absence of
excrescences is characteristic of those
that amplex on land or on vegetation
(Parker, 1940; Lynch, 1971). Presence of
excrescences is considered to be primi-
tive. The coded character states and po-
larity are:

0 = excrescences present; 1 = excres-

cences absent

=> Il

D. Larval body shape. The general
configuration of the body of tadpoles is a
reflection of the major habitats in which
they develop. Tadpoles in torrential
streams tend to have depressed bodies,
whereas those in ponds generally have
ovoid bodies. Modifications of pond tad-
poles are associated with the level in
which they feed in the water column;
both surface-feeders and bottom-feeders
are more depressed in comparison with
midwater-feeders. In dorsal view, tad-
poles of the Hyla leucophyllata group
have a lateral constriction at midbody, re-
sulting in a violin shape. The coded
character states and polarity are:

0 = ovoid; 1 = violin; 2 = depressed

0O—-1,;0->2

E. Larval tail shape. The tadpoles of
most frogs in primitive and advanced
familes have a tail that terminates in a
rounded or acuminate tip; the caudal
musculature terminates anterior to the
margin of the fin. In some of the hylids
under consideration, the caudal muscula-

ture extends posterior to the margin of
the fins, which extend as a narrow fringe
along the terminal part of the muscula-
ture. This type of tail with a distal fila-
ment is termed xiphicercal and is a de-
rived feature. The coded character states
and polarity are:

0 = acute; | = xiphicercal

O— 1

F. Larval mouth position. The general-
ized tadpole mouth is directed antero-
ventrally. Stream-adapted tadpoles have
ventral mouths, and surface-feeding tad-
poles have mouths directed dorsally.
Some midwater tadpoles have mouths
directed anteriorly at the tip of the snout.
The coded character states and polarity
are:

0 = anteroventrally; 1 = anteriorly

0-1

G. Larval labial papillae. In pond-
dwelling tadpoles of the Leptodactylidae
and Hylidae, the mouth usually is bor-
dered ventrally by two rows of small
papillae. Within the groups under con-
sideration, we find one or two rows of
small papillae, one row of large papillae,
or no papillae. The large papillae seem to
represent a fusion of smaller papillae,
and the absence of papillae seems to re-
sult from the fusion of large papillae into
a dermal fold, as seen laterally in mem-
bers of the Hyla parviceps group. The
coded character states and polarity are:

0 =2 small rows; 1=1 small row;

2 = 1 large row; 3 = none

O—1;0-2—-3

H. Larval labial denticles. The stand-
ard denticle formula in tadpoles is two
upper rows and three lower rows (2/3).
This number is increased in many
stream-adapted tadpoles and to a lesser
extent in some pond-dwelling tadpoles.
Within the Hylidae reduction of the

number of rows of denticles is uncom-

mon in free swimming and feeding tad-
poles and is most extreme in the groups
considered here. The coded character
states and polarity are:

0 = 2/3; 1 = 1/2; 2 = 0/1; 3 = 0/0

QS l—=2—s

In order to ascertain the evolutionary

48

trends within the groups being consi-
dered, it is necessary to have an outgroup
for comparison. Ideally such an outgroup
should be the most closely related taxon,
but such a determination is not presently
possible because of our inadequate
knowledge of many Neotropical hylids.
Thus we have chosen to make the com-
parisons with Smilisca, a genus of six
generalized species of Neotropical hylids
(Duellman and Trueb, 1966). The charac-
ter states for the taxa are given in Table 2,
and the resulting phylogenetic arrange-
ment is shown in Figure 7.

Thus, if our assumption that the pres-
ence of 30 chromosomes defines a
monophyletic complex of hylid frogs is
correct, then it is evident that the major
evolutionary trends in this complex in-
volve larval structure, especially mouth-
parts. Wassersug (1980) emphasized that
larval buccal morphology in the Hyla
leucophyllata group was specialized for
macrophagy and that this specializaion

Advances in Herpetology and Evolutionary Biology

was best developed in the Hyla micro-
cephala group. The ecological and bio-
geographical significance of these modi-
fications are not well understood. In
those few species that have been studied
(Crump, 1974; Duellman, 1978), the re-
productive strategy is to deposit multiple
clutches of small eggs in, or on, vegetation
above, temporary ponds.

Throughout most of the lowlands in
tropical America, three or four species of
this complex occur in sympatry, but 11
species occur sympatrically in Amazon-
ian Ecuador. Only two of the groups
occur in the Andes. The three species of
the Hyla columbiana group have allo-
patric distributions in the western and
central cordilleras in Colombia and
Ecuador, whereas Hyla labialis inhabits
higher elevations (2,400-3,000 m) in the
Cordillera Oriental in Colombia and the
Mérida Andes in Venezuela. The diver-
sity of species in this complex in the
tropical lowlands may be the result of

0
ot? ro
gn? co ef? ny" ce?
. \ 0 0 C 0 0
po” oom ain’ wo ot yeu a
D(O—1)
a G6 (2-3)
DOS D(O—2)
G(0—2)
H(2—=3)
H(4—=2)
F (O—=1)
E (O—=1)
c (O—=1)
H(O—=1)
B(O—=1)
A (0-1)

Figure 7. Reconstruction of the phylogeny of groups of Neotrpical Hy/a with 30 chromosomes. Shifts in charac-
ter states are indicated by the horizontal bars and identified by the letters and numbers given in Table 2 and text.

HYLA COLUMBIANA GROUP: Duellman and Trueb 49

TABLE 2. CHARACTER STATES OF GROUPS OF HYLA
HAVING 30 CHROMOSOMES.

Characters
D E

>
jee)
an

Group

H. labialis 1
H. columbiana 1
H. minuta 1
H. marmorata 1
H. parviceps 1
H. leucophyllata 1
H. microcephala 1
Outgroup (Smilisca) 0

one ee
One eee ES Q
ONroOoCoCoCSo
eo ee)
(See) ue)
CWNNHKHOCOCOO!]A
SOWWWWNR Ee

late Cenozoic and Quatemary climatic-
ecological fluctuations in the tropical
lowlands that resulted in the altemating
fragmentation and reunification of forest
and nonforest habitats (Haffer, 1979;
Duellman, 1982a). On the other hand, the
allopatric distributions of the few species
in the Andes may reflect isolation of pre-
viously more widespread species with
the uplift of the Andes or as a conse-
quence of the vicissitudes of Pleistocene
climatic change in the Andes (Simpson,
1975, 1979; Duellman, 1979, 1982b).

ACKNOWLEDGMENTS

We are indebted to the following per-
sons for the loan of specimens or for the
provision of working space in their re-
spective institutions: Josef Eiselt, Alice
G. C. Grandison, the late Walter Hell-
mich, Jorge Hernandez C., W. Ronald
Heyer, Konrad Klemmer, Edmond
Malnate, Hymen Marx, Charles W.
Myers, Gunther Peters, the late James A.
Peters, Greta Vestergren, the late Charles
F. Walker, Emest E. Williams, and John
W. Wright. Our field work in 1974-75,
which was supported by grants from the
National Geographic Society (No. 1304)
and the National Science Foundation
(DEB 74-01998), profited from the able
assistance of Dana T. Duellman and John
E. Simmons. Collecting permits for
Colombia were generously issued by
Jorge Hernandez C. of INDERENA. We

are grateful to Guillermo Gutierrez, Di-
rector of the Parque Arqueologico San
Agustin, for his gracious hospitality.

APPENDIX: SPECIMENS EXAMINED

Only specimens of the Hyla colum-

biana group are listed.

Hyla carnifex (427). COLOMBIA: Antioquia:
Envigado, AMNH 39796, 39798-99; Finca San José
de Bella Vista, LACM 47036-52, 47139; Jerico, 1,970
m, KU 98139-41, 98142 (skeleton), MCZ 15076—100,
24888-93, UMMZ 90601, USNM 118243; La Ceja,
2,180 m, FMNH 63883; Medellin, 1,480 m, AMNH
1357, ANSP 25005, FMNH 17098-101, MCZ 7664—
65, UMMZ 56508 (38), 60251 (2), 92167, USNM
75989, 118729-31; Rio Porce, ANSP 25773; San José
de los Andes, LACM 72999; San Pedro de Osa, 2,450
m, AMNH 38713-14, 39153-77, ANSP 21029,
FMNH 30567-69, 63884, 63897, MCZ 24911-15,
UMMZ 71214, 71215 (2), 78301 (17), USNM 152021-
29; Santa Rosa, 2,580 m, AMNH 39453, 39458,
Yarumal, 2,265 m, USNM 152030-31. ECUADOR:
Azuay: Cuenca (? locality), AMNH 23703, SMF
2368, ZMB 28023. Imabura: Apuela, 1550 m, KU
117992, 13242841, 132544 (tadpoles). Pichincha: 5
km ESE Chiriboga, 2,010 m, KU 164302-05, 166189-
92, 166311-14 (skeletons), 173113-15; 4 km W
Chiriboga, 2,120 m, KU 14263844; 14 km W
Chiriboga, 1,960 m, KU 164306-07, 166193-95 (tad-
poles); 4 km NE Dos Rios, 1,140 m, KU 164283-301,
166188 (tadpoles); Hwy 28, 22 km from Hwy 30,
1,770 m, MCZ 91202; Miligal (? =Milligalli),
USNM-GOV 8041-47; 3.5 km NE Mindo, 1,540 m,
KU 164308-11; Tandapi, 1,460 m, AMNH 81399-
403, BMNH 1969.640-44, FMNH 170746-50, KU
109557 (skeleton), 111838-66, 111867-70 (skele-
tons), 111871-90, 112360 (tadpoles), 117993-8001,
132442, 136199-254, 138775_76 (tadpoles), 178720-
22, 180352 (tadpoles), MCZ 7506266, UMMZ
129016 (8), USNM 16654350; 6.3 km E Tandapi,
1,770 m, MCZ 94728; 9 km SE Tandayapa, 2,150 m,
KU 164312.

Hyla columbiana (180). COLOMBIA: Cauca: El
Tambo, 1,800 m, ANSP 25148, FMNH 54776, KU
145075-79, NHRM 2015 (11); road to Munchique,
2,350 m, KU 148493 (tadpoles); Popayan, 1,740 m,
ANSP 25677, FMNH 43933-80, 43982-91, 44050,
44109, 82003-07, SMF 2365, USNM 152081-82,
152146-51; 8 km E Popayan, 2,110 m, INDERENA
(2), KU 169464-98, 170110-12 (skeletons), 170201-
02 (tadpoles), 170203 (eggs); Quintana, FMNH

- 54719; 4 km S Silvia, 2,200 m, KU 169461-62; 6km S

Silvia, 2,250 m, KU 169463. Quindio: Boquia, 25 km
N Armenia, 1,710 m, KU 133449, 139524. Valle: Cali,
1,000 m, AMNH 10674-77, BMNH 1947.2.13.15-21,
FMNH 3565, MCZ 2606, NHMW 6198, 19438, SMF
2642-44, UMMZ 46464, 51269, 58908 (8), USNM
71115, ZSM 1182/0 (3); Candelaria, 975 m, USNM
151983-84; Cerro La Herrera, USNM 137770-72;

50 Advances in Herpetology and Evolutionary Biology

Finca Wesfalia, ZSM 315/1937; Lago de Calima,
1,350 m, KU 181165-66.

Hyla praestans (11). COLOMBIA: Huila: Parque
Arqueologico San Agustin, 3 km SW San Agustin,
1,750 m, INDERENA (2), KU 169574-80, 170113
(skeleton), MCZ 100216.

LITERATURE CITED

BOETTGER, O. 1892. Katalog der Batrachier-
Sammlung im Museum Senckenbergischen
Naturforschenden Gesellschaft in Frankfurt-
am-Main. Senckenberg. Natur. Ges. Ber.,
1892: 1-73.

BocakT, J. P. 1973. Evolution of anuran karyotypes,
pp. 337-349. In J. L. Vial (ed.), Evolutionary
biology of the anurans: Contemporary research
on major problems. Columbia, Univ. Missouri
Press.

BOKERMANN, W. C. A. 1964. Notes on tree frogs of
the Hyla marmorata group with description of
a new species (Amphibia, Hylidae). Sencken-
berg. Biol., 45: 243-254.

BOULENGER, G. A. 1896. Descriptions of new rep-
tiles and batrachians from Colombia. Ann.
Mag. Nat. Hist., (6) 17: 16-21.

CocuHRAN, D. M., AND C. J. Goin. 1970. Frogs of
Colombia. U.S. Natl. Mus. Bull., 288: 1-655.

Crump, M. L. 1974. Reproductive strategies in a
tropical anuran community. Univ. Kansas Mus.
Nat. Hist. Misc. Publ., 61: 1-68.

DUELLMAN, W. E. 1969. A new species of frog in the

Hyla parviceps group. Herpetologica, 25: 241—

247.

1970. The hylid frogs of Middle America.
Univ. Kansas Mus. Nat. Hist. Monogr., I:
a7).

____. 1974. A reassessment of the taxonomic status
of some Neotropical hylid frogs. Univ. Kansas
Mus. Nat. Hist. Occ. Pap., 27: 1-27.

__. 1977. Liste der rezenten Amphibien und
Reptilien, Hylidae, Centrolenidae, Pseudidae.
Das Tierreich, 95: 1-225.

___. 1978. The biology of an equatorial herpeto-
fauna in Amazonian Ecuador. Univ. Kansas
Mus. Nat. Hist. Misc. Publ., 65: 1-352.

____. 1979. The herpetofauna of the Andes: Patterns
of distribution, origin, differentiation, and
present communities, pp. 371-459. In W. E.
Duellman (ed.), The South American herpeto-
fauna: Its origin, evolution, and dispersal.
Univ. Kansas Mus. Nat. Hist. Monogr., 7:
1-485.

____. 1982a. Quatemary climatic-ecological fluctu-
ations in the lowland tropics: Frogs and for-
ests, pp. 389-402. In G. T. Prance (ed.). Bio-
logical diversification in the tropics. New York,
Columbia Univ. Press.

____. 1982b. Compresion climatica cuatemmaria en
los Andes: Efectos sobre la especiacion. Actas
VIII Congr. Latinoamer. Zool. (in press).

DUELLMAN, W. E., AND M. L. Crump. 1974. Specia-
tion in frogs of the Hyla parviceps group in the
upper Amazon Basin. Univ. Kansas Mus. Nat.
Hist. Occ. Pap., 23: 1-40.

DUELLMAN, W. E., AND L. TRUEB. 1966. Neotropical
hylid frogs, genus Smilisca. Univ. Kansas Mus.
Nat. Hist. Publ., 17: 281-375.

Goin, C. J. 1970. A new name for Hyla bogerti
Cochran and Goin. Copeia, (4)1970: 788.
Gosn_ER, K. L. 1960. A simplified table for staging
anuran embryos and larvae with notes on iden-

tification. Herpetologica, 16: 183-190.

HAFFER, J. 1979. Quaternary biogeography of tropi-
cal lowland South America, pp. 107-140. In W.
E. Duellman (ed.), The South American her-
petofauna: Its origin, evolution, and dispersal.
Univ. Kansas Mus. Nat. Hist. Monogr., 7:
1-485.

LyncH, J. D. 1971. Evolutionary relationships,
osteology, and zoogeography of leptodactyloid
frogs. Univ. Kansas Mus. Nat. Hist. Misc.
Publ., 53: 1-238.

____. 1973. The transition from archaic to advanced
frogs, pp. 133-182. In J. L. Vial (ed.), Evolu-
tionary biology of the anurans: Contemporary
research on major problems. Columbia, Univ.
Missouri Press.

MERTENS, R. 1967. Die herpetologische Sektion des
Natur-Museums und _ Forschungs-Instituts
Senckenberg in Frankfurt a. M. nebst einem
Verzeichnis ihrer Typen. 1. Senckenber. Biol.,
48: 1-106.

MorESCALCHI, A. 1973. Amphibia, pp. 233-248. In
A. B. Chiarelli and E. Capanna (eds.). Cyto-
taxonomy and vertebrate evolution. New York,
Academic Press.

NIEDEN, F. 1923. Anura I. Subordo Aglossa und
Phaneroglossa. Section 1, Arcifera. Das Tier-
reich, 46: 1-584.

ParkER, H. W. 1940. The Australasian frogs of the
family Leptodactylidae. Novit. Zool., 42:
1-106.

SAVAGE, J. M., AND W. R. HEYER. 1967. Variation and
distribution in the tree-frog genus Phyllo-
medusa in Costa Rica, Central America. Beitr.
Neotrop. Fauna, 5: 111-131.

SIMPSON, B. B. 1975. Pleistocene changes in the
flora of the high tropical Andes. Paleobiology,
1: 273-294.

____. 1979. Quaternary biogeography of the high
montane regions of South America, pp.
157-188. In W. E. Duellman (ed.), The South
American herpetofauna: Its origin, evolution,
and dispersal. Univ. Kansas Mus. Nat. Hist.
Monogr., 7: 1-485.

SOKAL, R. R., AND F. J. ROHLF. 1969. Biometry. San
Francisco, W. H. Freeman Co.

TRUEB, L. 1973. Bones, frogs, and evolution, pp. 65—
132. In J. L. Vial (ed.), Evolutionary biology of
the anurans: Contemporary research on major
problems. Columbia, Univ. Missouri Press.

WASSERSUG, R. 1980. Internal oral features of larvae
from eight anuran families: Functional, sys-

HYLA COLUMBIANA GROUP: Duellman and Trueb 51

tematic, evolutionary and ecological con- WILEY, E. O. 1980. Phylogenetic systematics and
siderations. Univ. Kansas Mus. Nat. Hist. Misc. vicariance biogeography. Syst. Bot., 5:
Publ., 68: 1-146. 194-220.

A New Leptodactylid Frog from the Cordillera

Oriental of Colombia
JOHN D. LYNCH'

ABSTRACT. Eleutherodactylus anolirex is named
from the northern Cordillera Oriental of Colombia
(2,830-3,400 m). This species was previously con-
fused with E. supernatis and E. vertebralis. It is
most closely related to E. devillei, E. supernatis,
and E. vertebralis, species of the western Andean
cordilleras of Colombia and Ecuador.

INTRODUCTION

Although Cochran and Goin (1970) re-
ported the leptodactylid frog from Eleu-
therodactylus vertebralis (Boulenger) to
be widely distributed in Andean Colom-
bia, Lynch (1980) showed E. vertebralis
to be a species of relatively restricted
distribution found on the high Pacific
slopes of the Andes in central Ecuador
and described E. supernatis from the
Cordillera Central of Colombia and the
northernmost Cordillera Real of Ecuador.
Cochran and Goin (1970) confused
several species with E. vertebralis. Ex-
cept for noting the misidentifications of
specimens from the Sierra Nevada de
Santa Marta, Lynch (1980) deferred
taxonomic comment on other purported
populations of “E. vertebralis.” Ade-
quate material is now available to de-
scribe the frog from the Cordillera
Oriental of Colombia reported by
Duellman and Simmons (1977) as E.
vertebralis.

1School of Life Sciences, University of Nebraska,
Lincoln, Nebraska 68588, U.S.A.

TAXONOMY AND MORPHOLOGY
Eleutherodactylus anolirex sp. nov.
Figure 1

Holotype. KU 168626, an adult female, obtained
18.5 km (by road) S Chitaga, Departamento de
Norte de Santander, Colombia, 2,850 m, on 23
August 1974 by William E. Duellman.

Paratypes. Topotypes, KU 168627-30; 32 km (by
road) S Chitaga, Depto. Norte de Santander,
3,400 m, KU 168631-39; Presidente, 23 km S Chi-
taga, Depto. Norte de Santander, KU 150727-30; 35
km (by road) ENE Bucaramanga, road to Cucuta,
Depto. Santander, 2,830 m, KU 132722-23; Paramo
del Almorzadero, Depto. Santander, KU 150731-32,
and 17 uncatalogued specimens in the Museo La
Salle (Bogota).

Diagnosis. 1) Skin of dorsum bearing
flattened warts (least obvious anteriorly),
that of venter coarsely areolate; dorso-
lateral folds most distinct on posterior
part of body; 2) tympanum prominent, its
length one-third to one-half eye length;
3) snout subacuminate in dorsal view,
round in profile; eye-nostril (E-N) < eye
length; canthus rostralis sharp; 4) upper
eyelid width narrower than interorbital
distance (IOD); low cranial crests pres-
ent; 5) vomerine odontophores low, oval
in outline; 6) males with vocal sac and
slits, non-spinous nuptial pads on
thumbs; 7) first finger shorter than
second; moderate pads on fingers I-IV;
8) fingers bearing lateral fringes; 9) ulnar
tubercles present; 10) small tubercle on
heel; elongate tubercle on inner edge of
tarsus; 11) two metatarsal tubercles,
inner oval, three to four times size of sub-
conical outer; numerous supemumerary
plantar tubercles; 12) toes bearing lateral

fringes; no webs; toe pads slightly smal-
ler than those of fingers; 13) brown above
with faint cream dorsolateral stripes; lip
bearing cream stripe; venter heavily
flecked with brown; groin and concealed
surfaces of limbs brown with small cream
spots; 14) adults moderate-sized, males
24.0 to 31.8 mm, females 35.5 to 40.0 mm
snout-vent length (SVL).

Eleutherodactylus anolirex is a mem-
ber of the vertebralis assembly of Lynch
and Duellman (1980), most similar to E.
briceni (Boulenger), but differs from all
other species of the assembly in that
males have vocal slits and a vocal sac.

Description. Head narrower than or
equal to width of body; head width (HW)
36.4 to 39.4% (x = 37.8 n= 14) SVL;
snout short, E-N 71.8 to 93.8% (x = 83.8,
n = 14) eye length; snout subacuminate
in dorsal view, rounded in lateral profile;
nostrils weakly protuberant, directed
dorsolaterally; canthus rostralis sharp,
feebly sinuous; loreal region nearly flat,
sloping abruptly to non-flared lips; upper
eyelid width 69.7 to 87.4% (x = 80.6,
n = 14) IOD; no tubercles on eyelids;
edges of frontoparietals weakly up-
turned, producing frontoparietal furrow;
supratympanic fold thick, glandular, ob-
scuring upper and posterior edges of
tympanic annulus: tympanum distinct,
superficial, round, length 37.8 to 47.7%
(x = 42.5, n = 8) length in males, 32.4 to
45.2% (x = 36.9, n = 6) in females; tym-
panum separated from eye by distance
equal to 1 to 1.5 tympanic diameters; two
prominent round postrictal tubercles;
choanae small, round, not concealed by
palatal shelf of maxillary arch when roof
of mouth is viewed from directly below;
vomerine odontophores low, medial and
posterior to choanae, separated medially
by distance twice the width of a tooth
clump, each bearing a clump of three to
five teeth; tongue longer than wide,
posterior margin notched; posterior one-
third of tongue not adherent to floor of
mouth; males with vocal slits, subgular
vocal sac.

Skin of dorsum appears smooth an-

cream,

NEW LEPTODACTYLID FROG: Lynch 53

teriorly but is pitted (here interpreted as
bearing low, flattened warts), whereas on
lower back and upper flanks warts are
more pungent; flanks areolate; low dorso-
lateral folds (aligned warts) most evident
posteriorly; similar fold above coccyx;
venter coarsely areolate; discoidal folds
present; not distinct; forearm bearing
ulnar tubercles, antebrachial largest;
palmar tubercle bifid, much larger than
oval thenar tubercle; numerous super-
numerary palmar tubercles; subarticular
tubercles round, moderately pungent;
fingers flattened, bearing lateral fringes;
fingers bearing broad discs on relatively
small, rounded pads; pads of fingers III
and IV larger than tympanum; first finger
shorter than second (when equally ad-
pressed, tip of I reaches base of disc of
II); male with diffuse non-spinous nup-
tial pads on dorsal surfaces of thumbs.

Heel bearing distinct but small tu-
bercles; an elongate tubercle on inner
edge of tarsus (or a fold along distal two-
fifths of tarsus); no tubercles on outer
edge of tarsus; outer metatarsal tubercle
subconical, one-third to one quarter size
of oval (length twice width) inner meta-
tarsal tubercle; numerous supemumerary
plantar tubercles arranged in rows; sub-
articular tubercles round, low; toes lack-
ing webbing (basal between IV and V)
but bearing distinct lateral fringes; pads
bearing broad discs, slightly smaller than
those of fingers: hind legs moderately
long; shank 48.4 to 56.1% (x = 52.0,
n= 15) SVL; heels overlapping when
legs are flexed and held at right angles to
sagittal plane; heel of adpressed hind leg
reaching eye.

Dorsum dark brown with pale cream
dorsolateral stripes; flanks brown with
numerous cream flecks; labial stripe
canthal-supratympanic _ stripe
brown to black; groin, anterior and pos-
terior surfaces of thighs, ventral surfaces
of shanks dark brown with small cream
flecks or spots; venter cream with dense
brown flecking to brown with numerous
minute cream flecks (general appearance
brown).

54 Advances in Herpetology and Evolutionary Biology

Figure 1.
equals 2 mm, for head 5 mm.

Unfortunately few color notes on living
specimens exist. Living specimens are
described as dark brown with a distinct
white stripe on the lip, venter gray-
brown; iris bronze with median reddish-
brown streak. The failure to mention the
pale dorsolateral stripes apparent in pre-
served examples may mean that the
stripes are not evident in living frogs.

Measurements of the holotype in mm.
SVL 35.8, shank 18.4, HW 14.0, head
length 11.6, upper eyelid width 3.8, IOD
4.4, tympanum length 1.4, eye length 4.2,
E-N 3.7.

Etymology. The trivial name is Latin
meaning King of Anoles and is used in
loose reference to Ernest E. Williams.
The name might also refer to the ease
with which frogs of the genus Eleuthero-
dactylus sit astride their lofty Andean

Hand and head of Eleutherodactylus anolirex sp. nov. (KU 168626, holotype). Scale for hand

thrones gazing down toward the Andean
slopes and lowland environments with
their populations of Emest’s favorite
animals, the anoles.

Relationships. Eleutherodactylus ano-
lirex is a member of the wnistrigatus
group as defined by Lynch (1976). This
species group is very large (Lynch listed
about 100 species) and is centered distri-
butionally in northwestern South Ameri-
ca, especially in the Andes of Colombia
and Ecuador (Duellman, 1979; Lynch,
1979). The high species densities in
Ecuador are partially real and partially a
product of intense systematic research
there between 1966 and 1978. Lynch and
Duellman (1980) advocated the use of as-
semblies (a subunit of a species group) in
discussing relationships and distribu-
tions within the wnistrigatus group.

NEW LEPTODACTYLID FROG: Lynch 55

200

KILOMETERS

3000 METERS & ABOVE

3 < 1000 METER CONTOUR

_ a : — 68 es

Figure 2. Distributions of members of the Eleutherodactylus devillei assembly. E. anolirex (4), E. briceni
(m), E. supernatis (@). The distributions of E. devillei (eastern Andean slopes, Ecuador) and E. vertebralis
(western Andean slopes, Ecuador) are outlined (see Lynch, 1980, for dot map).

56

Eleutherodactylus anolirex is a member
of Lynch and Duellman’s devillei as-
sembly. The five named species of the
assembly are all moderate-sized, cloud
forest-dwelling Eleutherodactylus with
moderately developed digital pads,
broad interorbital spaces, and low cranial
crests. The distribution of the five spe-
cies (Fig. 2) is disjunct and encompasses
both modes cited by Lynch (1981) for the
paramo and subparamo assemblies found
in northem Ecuador. Similar disjunct dis-
tributions seen in more vagile groups of
organisms often prompt evocations of
dispersal as a means to explain the bio-
geographic pattern. However, equally
likely explanations include 1) extinctions
of geographically annectant populations
and 2) inadequate collecting. Whatever
explanation is advocated must be con-
sistent with the phylogenetic statement
for the assembly.

My hypothesis of the relationships of
the five taxa is expressed in Figure 3. The
characteristics and polarities employed
are as follows: A—vocal slits (present =
0, absent = 1), B—eyelid tubercle (ab-
sent = 0, present = 1), C—separation of
vomerine odontophores (wide = 0, nar-
row = 1), D—heel tubercle (absent = 0,
present = 1), E—inner tarsal tubercle or
fold (absent = 0, present = 1), F—toe
tips black (no = 0, yes = 1), G—dorso-
lateral folds (present = 0, absent = 1),
H—canthus shape (sharp = 0, rounded =
1), I—occipital folds (absent = 0, pres-
ent = 1), J—outer tarsal tubercles (ab-
sent = 0, present = 1). In each case, the

Advances in Herpetology and Evolutionary Biology

Figure 3. Cladogram of relationships in the Eleuth-
erodactylus devillei assembly. Shifts from primitive to
derived character-states are indicated by lettered hori-
zontal lines (letters correspond to characters, see
Table 1 and text).

derived state is scored as 1] and the primi-
tive state as 0. The character-states for
each species are summarized in Table 1.

The resultant cladogram (Fig. 3) ap-
pears to be geographically consistent
(nearest relatives are geographically
proximate for at least the three westem
and southernmost taxa). Unfortunately,
further analysis is impeded by a dearth of
knowledge about the eleutherodactyline
frogs of the Cordillera Oriental of
Colombia and the Venezuelan Andes.
The frogs of the Ecuadorian Andes are
better known, but phylogenetic state-
ments about them are not available. The
alleged impoverishment of the Cordillera
Oriental is more apparent than real

TABLE 1. CHARACTER-STATES OF SPECIES OF THE ELEUTHERODACTYLUS DEVILLEI ASSEMBLY.

Character briceni anolirex
A iL 0
B 0 0
ic 1 0
D 0 i
E 0 1
F 0 0
G 0 0
H 0 0
I 0 0
J 0 0

devillei supernatis vertebralis

cooococororFCrF
SOR RR Ke KF OF Ee
ll ee ee

(Pedro M. Ruiz, personal communica-
tion) and for the present prevents further
analysis.

ACKNOWLEDGMENTS

Specimens were loaned and facilities
for study made available by Hermano
Daniel (Museo La Salle, Bogota), Wil-
liam E. Duellman (Museum of Natural
History, University of Kansas—KU),
Pedro M. Ruiz (Instituto de Ciencias
Naturales, Universidad Nacional de
Colombia, Bogota), and Emest E. Wil-
liams (Museum of Comparative Zoology,
Harvard University). Special thanks are
extended to Ken Miyata and to Susan and
Anders Rhodin for the opportunity to par-
ticipate in honoring a giant in our field.

LITERATURE CITED

COCHRAN, D. M., AND C. J. Goin. 1970. Frogs of
Colombia. U.S. Natl. Mus. Bull., (288): 1-655.
DUELLMAN, W. E. 1979. The herpetofauna of the
Andes, pp. 371-459. In W. E. Duellman (ed.),

NEW LEPTODACTYLID FROG: Lynch 57

The South American Herpetofauna: its origin,
evolution, and dispersal. Univ. Kansas Mus.
Nat. Hist. Monogr., (7): 1-485.

DUELLMAN, W. E., AND J. E. SIMMONS. 1977. A new
species of Eleutherodactylus (Anura, Lepto-
dactylidae) from the Cordillera Oriental of
Colombia. Proc. Biol. Soc. Washington, 90: 60-
65.

LyNCH, J. D. 1976. The species groups of the South
American frogs of the genus Eleutherodactylus
(Leptodactylidae). Univ. Kansas, Mus. Nat.
Hist. Occas. Pap. (61): 1-24.

___. 1979. The amphibians of the lowland tropical
forests, pp. 189-215. In Duellman, W. E. (ed.),
The South American herpetofauna: its origin,
evolution, and dispersal. Univ. Kansas, Mus.
Nat. Hist. Monogr. (7): 1485.

___. 1980. The identity of Eleutherodactylus ver-
tebralis (Boulenger) with the description of a
new species from Colombia and Ecuador (Am-
phibia: Leptodactylidae). J. Herpetol., 13:
411-418.

___. 1981. The eleutherodactyline frogs from the
high Andes of northem Ecuador and adjacent
Colombia. Univ. Kansas, Mus. Nat. Hist. Misc.
Publ. (72): 1-46.

LyNCH, J. D., AND W. E. DUELLMAN. 1980. The
Eleutherodactylus of the Amazonian slopes of
the Ecuadorian Andes (Anura: Leptodactyli-
dae). Univ. Kansas, Mus. Nat. Hist. Misc. Publ.
(69): 1-86.

Description of Phrynops williamsi,
A New Species of Chelid Turtle of the
South American P. geoffroanus Complex

ANDERS G. J. RHODIN'
RUSSELL A. MITTERMEIER?

ABSTRACT. Phrynops williamsi, sp. nov., is a dis-
tinctive member of the P. geoffroanus complex
(Pleurodira: Chelidae). The species is characterized
by heavy black facial bands with a separate, thick,
horseshoe-shaped band on the ventral surface of the
neck, fine well-delineated radiating carapace retic-
ulations, and no plastral markings. The skull differs
markedly from other species of the P. geoffroanus
complex and from most Phrynops in general. A
wide parietal roof, lack of exoccipital contact above
the foramen magnum, and widely divergent troch-
lear processes appear to be primitive features. Ro-
bust anterior maxillary triturating surfaces with lin-
gual ridges and a shovel-shaped mandible adapted
for bottom feeding appear to be unique derived
characters. The species has a limited distribution in
southern Brazil (Rio Grande do Sul and Santa
Catarina) and adjacent Uruguay.

INTRODUCTION

The South American side-necked tur-
tles of the family Chelidae are among the
most poorly known of all turtles. No
comprehensive review has ever ap-
peared, and much of the scattered litera-

'Department of Surgery (Orthopaedics), Yale
University School of Medicine, New Haven, Con-
necticut 06510 and Museum of Comparative Zo-
ology, Harvard University, Cambridge, Massachu-
setts 02138, U.S.A. (Present address: Orthopaedic
Associates, P.C., Nichols Rd., Fitchburg, Massachu-
setts 01420, U.S.A.)

2World Wildlife Fund—U.S. Primate Program,
1601 Connecticut Avenue, NW, Washington, D.C.
20009, U.S.A.

ture has suffered from inaccuracies based
on inadequate sample analysis. Elusive
and often rare in the wild, museum col-
lections usually include only haphazard-
ly or incidentally collected specimens.
Many descriptions of new taxa have been
based on single specimens, often without
comparative material at hand. No clear
definition exists of the valid taxa and
their distribution, variation, morphology
and ecology. Only Gaffney (1977) has
looked at the family in a relatively com-
prehensive manner, but his work dealt
with supraspecific relationships based on
cranial osteology and examined less than
half the known species.

In an effort to clarify the taxonomy of
the South American Chelidae, Emest E.
Williams and Paulo E. Vanzolini began
an extensive review of the family during
the late 1950’s, but abandoned the pro-
ject due to inadequate material. They
turmed their preliminary work over to us
several years ago, and this data base has
served as the foundation for our work. It
is therefore particularly fitting that the
first description of one of the new species
of chelid turtles to arise from this ma-
terial appear in a volume dedicated to Dr.
Williams. In addition, it is with great
pleasure that we name this new species
in honor of Ernest E. Williams, both in
appreciation of his wide-reaching contri-
butions to herpetology and the study of
turtles, as well as in gratitude for his un-

DESCRIPTION OF PHRYNOPS WILLIAMSI : Rhodin and Mittermeier 59

failing support, his friendship, and his
guidance in our studies of systematic bi-
ology over the years. This has been a
most rewarding relationship, undimin-
ished by our respective professional
divergences into the fields of orthopaedic
surgery and primate conservation, and
we take this opportunity to offer him our
thanks.

Both Wermuth and Mertens (1977) and
Pritchard (1979) recognize 18 taxa of
South American chelids, though they dif-
fer in their interpretation of subspecific
rank in three cases. Of the four currently
recognized genera (Chelus, Hydro-
medusa, Phrynops, Platemys), Phrynops
is the most complex taxonomically, with
eleven currently recognized taxa: P.
dahli, geoffroanus, gibbus, hilarii, hogei,
nasutus, rufipes, tuberculatus, tubero-
sus, vanderhaegei, and wermuthi. Three
of these taxa (P. geoffroanus, hilarii,
tuberosus) form a natural superspecies
complex based on external and osteologi-
cal similarities and will hereafter be re-
ferred to as the P. geoffroanus complex.
Briefly, the forms within the complex
share the general features of large, broad
shells with moderately wide heads, pari-
etal roof moderately broad, neurals usu-
ally numbering six or seven and contact-
ing the nuchal bone broadly, carapace
color either dark brown or reticulated
with radial black markings, plastron color
usually reddish, yellow or white and
either immaculate or with scattered dark
vermiculations or spots, head with charac-
teristic lateral black stripe from snout
through the eye and tympanic membrane
and extending along the lateral surface of
the neck, ventral surface of the neck with
either broad black bands, scattered thin-
ner vermiculations or spots.

During the course of our revision of the

South American Chelidae, we have per-

sonally examined 353 specimens of
members of the P. geoffroanus complex.
Of this series, 54 represent the readily
distinguishable P. hilarii, with its im-
maculate dark brown or gray carapace,
black-spotted white plastron, and re-

duced head markings. Of the 299 other
specimens of P. geoffroanus and P.
tuberosus, several distinct geographic
populations are identifiable. A complete
analysis of the systematic status of these
populations is beyond the scope of the
present paper, but will be presented in a
future publication. The most distinctive
of the populations is represented by six
specimens from Rio Grande do Sul and
Santa Catarina in Brazil. Though allo-
patric in respect to other P. geoffroanus
and tuberosus, the combination of exter-
nal and osteological features present in
this population warrant its recognition as
a new species within the P. geoffroanus
complex.

TAXONOMY AND MORPHOLOGY

Phrynops williamsi sp. nov.
Figures 1-6

Holotype. MCZ 64135, Rio Grande do Sul, Brazil,
collected by H. von lhering.

Paratypes. BMNH 84.2.5.1, 84.2.5.3, Rio Grande
do Sul, Brazil, collected by H. von Ihering; MZUSP
2675, Porto Alegre, Rio Grande do Sul, Brazil;
MNBJ 3146, Tubarao, Santa Catarina, Brazil; ZMB
6858, “Estancia Velha,’ Rio Grande do Sul, Brazil,
collected by R. Hensel (probably collected at
Pikada do Café, Rio Cadéa).

Type Locality. Since the holotype specimen has
no specific locality, we hereby designate the type
locality as Rio Cadéa, Rio Grande do Sul, Brazil,
where the species was first collected by R. Hensel
in 1865.

Synonymy

Platemys geoffreyana (partim) Hensel, 1868:350

Platemys geoffroyana (partim) Hensel, 1868:354;
Boulenger, 1885:191, 1886:424; Strauch, 1890:
104; Lema, 1958:11

Hydraspis geoffroyana (partim) Boulenger, 1889:
923; Siebenrock, 1904:23; Goeldi, 1906:751;
Siebenrock, 1909:576; Luederwaldt, 1926:433

Phrynops geoffroyana (partim) Vaz-Ferreira,
1955:xxv; Froes, 1957:19

Phrynops geoffroana geoffroana (partim) Mertens
and Wermuth, 1955:404; Vaz-Ferreira and Sierra
de Soriano, 1960:14

Phrynops _ geoffroanus

geoffroanus _ (partim)

60 Advances in Herpetology and Evolutionary Biology

Wermuth and Mertens, 1961:333,
Pritchard, 1967:234, 1979:787
Phrynops geoffroanus (partim) Freiberg, 1970:190,
1971:92, 1972:248, 1975:92; Lema and Fabian-
Beurmann, 1977:65
Phrynops geoffroana (partim) Achaval, 1976:26
Phrynops geoffroanus sspp. (partim) Mittermeier,
Medem, and Rhodin, 1980:15

1977:130;

Distribution. Low-lying areas (below
500 m elevation) of eastern coastal Santa
Catarina and Rio Grande do Sul in Brazil
as well as the northern half of Uruguay as
far west as the Rio Uruguay, possibly in-
cluding eastem portions of Entre Rios
and southeastern Corrientes in Argentina
as well as southwestern inland Rio
Grande do Sul (Fig. 9). Apparently ab-
sent from southern coastal Uruguay and
northwestern inland Rio Grande do Sul
in the upper Rio Uruguay and Rio Pelotas
drainages.

Diagnosis. A moderately-sized mem-
ber of the P. geoffroanus complex
characterized by three heavy black facial
stripes including a separate thick horse-
shoe-shaped stripe on the ventral neck,
well-delineated thin radiating carapacial
reticulations and no plastral markings.
Skull characterized by a wide parietal
roof, lack of exoccipital contact above the
foramen magnum, widely divergent
trochlear processes, and robust anterior
maxillary triturating surfaces with lingual
ridges and shovel-shaped mandible.

EXTERNAL MORPHOLOGY

Carapace. Carapace (Fig. 1) broadly
oval, almost subcircular in juveniles,
carapace length averaging 1.15 times
width, becoming relatively narrower
with increasing size, carapace length
averaging 1.33 times width in subadults
and smaller adults, 1.46 in larger adults
(see Table 1 for all measurements and
ratios of external features). Shell moder-
ately deep, length averaging 3.39 times
height in adults and subadults. No mar-
ginal flaring, recurving or broad expan-
sion. Mid-lateral marginals slightly nar-
rower than anterior and posterior ones.
Entire marginal rim mildly serrate in

juveniles, partially retained posteriorly
in subadults and adults but less distinct.
Small supracaudal notch. Nuchal
approximately twice as long as broad,
projecting slightly anterior to carapace
margin. Vertebrals generally wider than
long, including V4 which is notably
wide. Intercostal lateral seam contacts at
M2, 5, 7, 9, and 11. Vertebrals without
furrow, trough, keel or knobs in sub-
adults and adults. Very vague, low mid-
line bulge posteriorly on V5, extending
toward supracaudal notch, causing very
mild keeling at posterior end of shell,
more prominent in large adults than sub-
adults. Juveniles with low inconspicuous
vertebral bulges on V1-5 and costals.

Carapace color in preserved specimens
medium to light brown with extensive
thin black reticulations radiating in a
well-delineated radial pattern on all
vertebrals, costals and marginals. Center
of radiating pattern located at site of ori-
ginal juvenile scute: posteriorly in mid-
line on vertebrals, posteromedially on
costals, and posterolaterally on margin-
als. Pattern as distinct in large adults as in
juveniles. Black lines generally same
thickness or thinner than interspersed
brown base color. Color demarcation
sharp and clear, with pattern extending
fully to edge of marginal rim. Carapace
color in live specimens brown with black
reticulations and thin yellowish orange
carapacial edge (Vaz-Ferreira and Sierra
de Soriano, 1960).

Plastron. Plastron (Fig. 2) broad,
length averaging 1.82 times width, cara-
pace width averaging 1.60 times plastron
width, anteriorly truncate, posteriorly
slightly narrower. Anal notch moderately
deep. Very small axillary and inguinal
scutes present. Intergular short and
broad. Plastral seam length formula Fem
> Ig = An > Abd = Pect = Hum.

Plastron color in preserved specimens
yellowish brown, occasionally oxidized
to darker brown. Immaculate on all ven-
tral surfaces, though occasional large
specimens with indistinct dorsal type
color pattern on posterior ventral margin-

DESCRIPTION OF PHRYNOPS WILLIAMSI - Rhodin and Mittermeier 61

Figure 1.
radiating reticulations.

als. Plastron color yellow in live speci-
mens (Vaz-Ferreira and Sierra de Sori-
ano, 1960).

Head and Neck. Head width moderate-
ly narrow, becoming relatively narrower
as compared to carapace length with
allometric growth (Fig. 8). Neck length
fairly short.

Dorsal view of carapace of ZMB 6858, a paratype of P. williamsi. Note the clearly delineated fine

Head and neck with distinctive color

‘pattern (Fig. 3), primarily dark dorsally

and light ventrally, characterized by
three subparallel broad black bands. The
uppermost band serves as the ventral
border of the dark dorsal head and neck
pattern, extending from the nostrils
through the eye, through the upper one

62 Advances in Herpetology and Evolutionary Biology

Figure 2. Ventral view of BMNH 84.2.5.3, a paratype of P. williamsi. Indistinct reticulations can be seen on
marginals posterior to bridge.

DESCRIPTION OF PHRYNOPS WILLIAMS] -

Figure 3. Ventral and lateral views of head and neck
of MCZ 64135, the holotype of P. williamsi. Ventral
photograph slightly retouched to obscure small lacera-
tion present on specimen itself.

third to one half of the tympanum, and
then along the mid-lateral surface of the
neck, gradually fading caudally. The
lowermost band forms a posteriorly di-
rected horseshoe-shaped figure on the
ventral chin, extending anteriorly to the
base of the barbels, usually sharply dis-
continuous posteriorly, with an interrup-
tion at the level of the posterior border of
the tympanum, before continuing for a
short distance as short subparallel bands
or elongate spots. The intermediate band
extends caudally from the angle of the
jaws, serving as a continuation of thin
bands of dark pigment on the external
tomial surfaces of the inferior portion of
the maxillary and superior portion of the
mandibular horny sheaths. The band
then continues along the inferior border
of the tympanic membrane and ventro-
laterally along the neck, ending abruptly
at approximately the same level as the

Rhodin and Mittermeier 63

last band or spot in the lowermost band.
These three broad bands are usually
totally separate from one another. Of
seven specimens examined (including
specimens figured in the literature), none
had a connection between the upper and
intermediate bands. Three out of seven
specimens had very thin connections
between the lower and intermediate
bands posterior to the tympanum, two of
them only unilateral and one with thin
bilateral connections. One out of seven
specimens had a discontinuous ventral
horseshoe, narrowly lacking midline con-
tact under the chin. The dorsal head pat-
tern is relatively indistinct, composed of
a dark background with narrow indistinct
lighter lines subparallel to the uppermost
dark band.

Live color of dorsal surface of head
black with whitish lines, ventral surface of
chin and neck reddish yellow or yellow
with black bands (Vaz-Ferreira and Sierra
de Soriano, 1960).

Two barbels present on the ventral sur-
face of the chin, arising from the anterior
portion of the horeshoe band. Barbels
often unpigmented, though one some-
times the same color as the band.

Skin of top of snout, interorbital region
and middle third of top of head adherent
to bone, incompletely divided into ap-
parent scales. Remaining dorsal surface
of head (above temporal muscles) with
more regular polygonal scales. Skin of
dorsal aspect of neck plicate with only
occasional more or less well defined pap-
illae. Skin of ventral aspect of neck shal-

-lowly plicate, reticulate.

Limbs. Dorsal aspect of limbs with in-
distinct dark vermicular pattern. Ventral
aspect light colored, juvenile with few
scattered spots and vermiculations. En-
larged scales forming prominent swim
flap on free ulnar skin edge of anterior
limbs. Three pre-tibial scales enlarged,
with the distal one prominent and corni-
fied, resembling a blunt claw. No pre-
tibial flap or ischial tuberosities. Claws
numbering five on forelimbs, four on

hindlimbs. Color of limbs in life blackish

Advances in Herpetology and Evolutionary Biology

64

“UU UT S}USUWOINSBOUT [TY “([BUITUTUT) YApIMm prosA1a}d = MAIg “([ewWTUIW) UpIM jeeued = Md “YIPIM [eWGIO = MO

“WAPIM [BUqIOLayUI = MOT “(A[sor19ysod ‘ourpprut) ydep [[nys = qs ‘(Are]]Ixew) prim [[NyAs = WAAS ‘(ofuedurAy) yprm [[NAS = LAS “(ze[Iseqo][Apuoo) yyBu9] [[Nyxs = Ts “wSua] soedereo = 7D,
—_—_—_—_eerere—e—e———n—n—n eae CO COCO
9£°0 LOE 080 6S'0 880 68 T col 99'S O91 SCI 98 ODL GHI SHS GCE GH sco §97rS HNNG
6£°0 ORS 66.0 Is'0 080 LL I 61 T I9¢ 8CI SIT GL O9 8IT GO0G ODE BSE 10Z 8989 ANZ

IMS/Md CS/TS Mid/Md Md/MOI MO/MOI WA\S/TS LIAMS/TS TS/TD Mid Md MO MOI GS WMS IMS TIS TD LTT
uaunsads

» SOILVY OISVE GNV SLNAWHYNSVAW TIONS ISWVITIIM SAONAYH *G ATAV I,

‘(OLET “B1I9qTE1,4 ‘O96 ‘OURLIOS ap eLIOIG puke eIloLIag-zeA) oINjeI9}] IY} WOI; pap1osay {
“WU UT sySUEINsveUI [[Y “(orueduIA}) ypPIM peey = MH ‘Wp uogseld = Md “Wsue] uoysejd = 7g ‘yep soedereo = GD ‘WIpIM aoedeied = MD ‘IWSUa] soedeIeD = JD,

OST 033 Ose é {u/s SIeqto1y

er VIG GOES é 4261 H-OAZIN
681 rT GL'9 61 Sel GLE raul alia 6L €81 GSS é ESF HNNG
PST 09'T 6£°9 LVE 9E'T 08s ZI 90G OL 6LI €vG é 1'S3r8 HNN
98'T LOT 8e'9 GLE Ie GI€ 6 TAI PG PSI 10@ Pd 8Ss9 ANZ
ZS cc 139 CO's 0€'T 0°63 68 ZOL 6S Sel OST éd CSIF9 ZOW

cul LIT IFT 2 {E61 H-OAZIN
OL'T PST €8'S os'¢ 0€'T O'FS OL 6IT OF SOL OFT éd GL9Z dSAZW
€9'T 89'T I8'F CLS CIT 0'9T OF fete) 8G L9 LL Ant 9FIe [HUNIN

Md/'Id Md/MO MHI/TO dod/TD MO/TO MH Md Tal do MO 19 xag 1aquinn

uawiaad¢

x SOILVY OISVE CNV SLNAWHYNSVEW TYNYALXA ISWVITIIM SdONXYHZ “T ATAV LL

DESCRIPTION OF PHRYNOPS WILLIAMSI - Rhodin and Mittermeier

brown dorsally, reddish yellow or yellow
ventrally, with red webbing on feet and
red swim flap on forelimb (Vaz-Ferreira
and Sierra de Soriano, 1960).

Sexual Dimorphism and Size. Sexual
dimorphism not apparent from series ex-
amined. Large specimens with deep
shells, short tails and flat plastrons repre-
sent females. No obvious males with
shallow shells, long tails and concave
plastrons present in series, but Vaz-
Ferreira and Sierra de Soriano (1960) fig-
ure a 141 mm male with a long, thick tail.
The largest specimen examined mea-
sured 252 mm carapace length, but Vaz-
Ferreira and Sierra de Soriano (1960)
recorded a specimen measuring 305 mm
and Freiberg (1970) measured a 330 mm
individual.

OSTEOLOGY

Skull. Rather than describe the skull of
P. williamsi in detail, we prefer to dis-
cuss it in relation to other Phrynops and
chelid skulls. We have examined two
skulls of P. williamsi and have compared
them to 47 skulls of other Phrynops
(representing all known species), of
which 22 represent other members of the
P. geoffroanus complex. Most of this
comparative material represents our
unpublished data. However, Gaffney
(1977) has discussed and figured skulls of
P. hilarii (as “P. geoffroanus”) and P.
gibbus, and we provide figures of P. wil-
liamsi (Figs. 4, 5) and P. geoffroanus
(Figs. 5, 7) in this work. Other species
mentioned (e.g., P. rufipes, hogei, van-

derhaegei) will be fully discussed and -

figured in future publications. Table 2
lists all skull measurements and ratios for
the two skulls of P. williamsi examined.

A number of features distinguish the
skull of P. williamsi from the other mem-
bers of the geoffroanus complex. In fact,
it is among the most distinctive of all
Phrynops skulls in many respects.

The lateral extent of the parietal roof is
greater than in most other Phrynops, the
sides being nearly parallel. There is con-
siderably less posterior temporal emargi-

65

nation, with relative loss of overlap be-
tween the anterior and posterior tem-
poral emarginations (slight overlap in the
larger specimen). The temporal arch is
shorter and thicker than in other
Phrynops. There is very little lateral
maxillary protrusion, but two other
Phrynops (hogei and rufipes) show less
protrusion, and other young geoffroanus
complex species can have the same ex-
tent. This characteristic is ontogenetic-
ally variable and larger williamsi have a
slightly more developed protrusion.

The two processi trochlearis pterygoi-
dei show wide divergence from the lon-
gitudinal axis of the skull, making a ca. 70
to 90° angle with each other when
viewed from above. No other Phrynops
shows this condition, though hogei
reaches 50 to 60° and rufipes 60 to 70°. All
other Phrynops have an angle of 45° or
less. The maxillary triturating surface has
a moderately well-developed short lin-
gual ridge anterior to the choanae, a con-
dition not seen in any other Phrynops.
Compared to other geoffroanus complex
species, the anterior maxillary region of
williamsi is much more pronounced and
robust with a thicker, heavier maxilla, an
angular rather than rounded snout, a
wider maxillary triturating surface es-
pecially anteriorly and the presence of
short lingual ridges. The only other
Phrynops with prominent ventral snout
development are hogei and rufipes, with
gibbus and vanderhaegei showing an
intermediate condition.

The mandible of williamsi is very dis-
tinctive, with the triturating surface be-
ing markedly widened and extremely
flattened, coming very close to the hori-
zontal. This is totally unlike other geof-
froanus complex species or any other
Phrynops, including hogei and rufipes. It
most closely resembles the condition
seen in the South American pelomedu-
sine turtle Podocnemis sextuberculata.
This shovel-like appearance of the man-
dible may represent a specialized adapta-
tion to bottom feeding.

The exoccipitals of williamsi do not

66 Advances in Herpetology and Evolutionary Biology

Figure 4. Osteology of skull and mandible of P. williamsi (ZMB 6858). A. dorsal view; B. ventral view; C. lateral
view; D. posterior view; E. lateral view of mandible; F. medial and slightly dorsal view of right half of mandible.
Refer to Gaffney (1979) and Rhodin and Mittermeier (1976) for skull nomenclature.

meet above the foramen magnum, being
moderately separated by the supraoccipi-
tal. This feature is also shared by hogei
and rufipes. The same feature, though
less pronounced with only narrow sepa-
ration, is variably present in vander-
haegei, gibbus, hilarii and geoffroanus.

The normal condition in these species is
either a narrow or moderate exoccipital
contact. No other South American chelid
has separated exoccipitals, the feature
being shared only by the Australopapuan
genera Emydura, Elseya, and Pseudemy-
dura (Gaffney, 1977).

DESCRIPTION OF PHRYNOPS WILLIAMSI « Rhodin and Mittermeier 67

Figure 5. Dorsal views of mandibles of P. williamsi (left, BMNH 84.2.5.3) and P. geoffroanus (right, MZUSP 50;
Rio Grande, Sao Paulo, Brazil; carapace length 250 mm, skull condylobasilar length 44 mm). Note markedly
widened and flatter triturating surface in P. williamsi. Photographic enlargement slightly greater for P. geoffro-
anus.

The foramina nervi trigemini of wil-
liamsi face laterally, and are not visible
from the dorsal aspect of the skull. In
other geoffroanus complex species the
foramina face dorsally and are easily
visible from the dorsal aspect. The flat-
tened horizontal portion of the pterygoid
flare is relatively reduced in williamsi so
that only a very small portion of the
pterygoid is visible from the dorsal as-
pect of the skull. In other geoffroanus
complex species the pterygoid flare is
quite wide so that a broad expanse of
pterygoid is visible from the dorsal as-
pect.

The postorbital wall of williamsi has a
distinct sulcus jugalis and the medial por-
tions of the jugal and postorbital face
more posteriorly than all other Phrynops,
where this wall faces more dorsolaterally
and has a less well-developed sulcus
jugalis.

The features of widely divergent
trochleas, deep sulcus jugalis with over-

Figure 6. Neural bones of P. williamsi (BMNH 84.2. _ lying parietal roof, posteriorly facing pos-
2) torbital wall, shovel-like appearance of

68

Advances in Herpetology and Evolutionary Biology

Figure 7. Osteology of skull and mandible of P. geoffroanus (MZUSP 2680; Ilha Solteira, Rio Parana, Sao
Paulo, Brazil; carapace length 172 mm, skull condylobasilar length 33.7 mm). Views same as in Fig. 4.

the mandible and robust widened an-
terior maxillary triturating surface with
lingual ridges appear to all represent
feeding adaptations. The power of jaw
closure is heightened through increased
muscle mass in the anterior portions of
the jaw adductors combined with diver-
gent trochleas which improve leverage.

This increased muscle mass is most
prominent in the relatively enlarged area
of the postorbital wall and sulcus jugalis.
The modified triturating surfaces appear
adapted for bottom feeding and crush-
ing of large hard food such as snails or
small bivalves. Other Phrynops do not
possess this spectrum of characteristics

DESCRIPTION OF PHRYNOPS WILLIAMS! « Rhodin and Mittermeier 69

Figure 8. Head width in P. williamsi. Plot of carapace
length in mm (abscissa) versus carapace length di-
vided by head width (ordinate).

and many of them are known to be preda-
tory feeders. Unfortunately, nothing is
known of the feeding habits of williamsi.

The wide parietal roof, lack of exoccipi-
tal contact above the foramen magnum
and widely divergent trochlear processes
of P. williamsi appear to be primitive fea-
tures within the genus Phrynops. These
features show a similarity to the pelo-
medusine turtles of the genus Podoc-
nemis as well as the Australopapuan
chelids of the genera Elseya, Emydura,
and Pseudemydura (see Gaffney, 1977
and McDowell, 1983). The robust an-
terior maxillary triturating surfaces with
lingual ridges and shovel-shaped man-
dible may be primitive features, but more
likely represent unique derived charac-
ters within the genus Phrynops, showing
secondary convergence with Podocnemis
sextuberculata due to similar feeding
Strategy.

Cervicals. One specimen examined
has the typical central articulation pat-
tern of all Chelidae as described by Wil-
liams (1950): (2(3(4(5)6)7(8).

Neurals. Six large contiguous neurals
present (Fig. 6). N1 ca. twice as long as
N2-5. N6 half as small as N2-5. N1 contac-

ting nuchal widely, broadly rectangular.
N2-5 roughly hexagonal, tapering pos-
teriorly. N6 small and pentagonal, only
partly separating C6. Pattern identical in
the two specimens examined.

REPRODUCTION

No field data are available on eggs,
nests or hatchlings of P. williamsi. One
specimen, BMNH 84.2.5.3, a 252 mm fe-
male collected somewhere in Rio Grande
do Sul, contained nine white, oval, brittle
shelled oviductal eggs (four in the left
oviduct, five in the right). The average
egg size in this series was 33.3 x 27.0 mm
with length ranging from 32.9-34.2 mm
and width ranging from 26.7-27.6 mm. It
is unclear when the specimen was collec-
ted, but egg deposition probably occurs
in either November or December. Phry-
nops hilarii in northeastern Argentina
nests in November (Gallardo, 1980) and
in Rio Grande do Sul nests from late
October to early January, with the peak
activity occurring in November (Reischl
et al., 1979). Pseudemys dorbignyi
(Emydidae) from the same area has the
same nesting season but peaks primarily
in December (Reischl et al., 1979).

Clutch size in P. hilarii is apparently
somewhat larger than in P. williamsi. In
Rio Grande do Sul clutch size of P. hilarii
averages 11, with the incubation period
taking 105 to 140 days (Reischl et al.,
1979). Serrano (1977) notes that P. hilarii
in Rio Grande do Sul lays clutches
averaging 13.4 eggs (range 1-20) with
spherical eggs measuring 33+4 mm in

_diameter. Saporiti (1960) indicates that P.

hilarii in Buenos Aires lays 10 to 14
subspherical eggs averaging 32.9 x 30.8
mm in size. Cohen (personal communica-
tion) had a captive specimen lay 17 eggs
averaging 36.1 x 33.8 mm in size.

Few comparative data are available for
P. geoffroanus and tuberosus. Wied
(1825) noted that populations of P. geof-
froanus inhabiting the rivers of eastern
coastal Brazil (e.g., Rios Pardo, Jequitin-
honha, and Mucuri) laid from 12 to 18
spherical eggs from December through

70 Advances in Herpetology and Evolutionary Biology

February. Medem (1969) recorded three
nests of P. tuberosus from Colombia as
having from 10 to 18 subspherical eggs
with average size measuring 33.9 x 32.5
mm. We have examined a specimen of P.
geoffroanus from the Rio Tapajos, Para,
Brazil (MZUSP 2682) containing 13 sub-
spherical shelled eggs, average size mea-
suring 30.5 x 29.0 mm.

HABITAT

Vaz-Ferreira and Sierra de Soriano
(1960) obtained two specimens at Picada
del Negro Muerto, Uruguay, where the
Rio Cuareim flows relatively rapidly over
a rocky streambed. Hensel (1868) noted
that the species occurred in the “rushing
and stony brooks of the forest’ and was
not to be found in the slower lowland
rivers with muddy bottoms.

DISCUSSION

The species here recognized as Phry-
nops williamsi was first collected and
described as Platemys geoffreyana
(Schweigger, 1812) by Hensel (1868). His
description of an animal he collected in
1865 at Pikada do Café, Rio Cadéa, Rio
Grande do Sul fits perfectly our descrip-
tion of P. williamsi. Another specimen of
the same species may have been ob-
served at Estancia Velha, but Hensel’s
description of that animal is not as clear
and it may have been based on a speci-
men of P. hilarii. Hensel states that he
was able to obtain only a single specimen
of “P. williamsi’ on his trip. This speci-
men was collected at Pikada do Café, Rio
Cadéa and had a carapace length of 202
mm (Hensel’s measurement). The P. wil-
liamsi paratype allegedly collected by
Hensel at Estancia Velha (ZMB 6858) has
a carapace length of 201 mm, and if in-
deed collected by Hensel, probably
represents the original Rio Cadéa speci-
men, though now mislabeled as coming
from Estancia Velha.

The species was next collected by von
Ihering, whose specimens served as the

basis for Boulenger’s (1889) description
of Hydraspis geoffroyana (Schweigger).
One of these specimens is our P. wil-
liamsi holotype, two others are para-
types.

Clear descriptions of animals referable
to P. williamsi did not re-appear in the
literature until Vaz-Ferreira and Sierra
de Soriano (1960) figured and described
specimens of P. geoffroana geoffroana
from Uruguay. Some of these and others
as well were described and figured by
Freiberg (1970) and served as his basis
for the specific separation of P. hilarii
from P. geoffroanus based on their
Uruguayan sympatry.

The characteristics of P. williamsi dis-
cussed above readily distinguish it from
any other Phrynops, as well as from any
other member of the P. geoffroanus com-
plex. The black bands on the head and
neck coupled with the reticulate cara-
pace pattern immediately differentiate it
from P. gibbus, rufipes, hogei, nasutus,
dahli, wermuthi, vanderhaegei, and
tuberculatus. Phrynops hilarii is distin-
guished by its black spots on a white
plastron, immaculate carapace and
markedly reduced head and neck bands
(see photos in Freiberg, 1970, 1975).
Phrynops geoffroanus and P. tuberosus
represent two recognized populations of
a very wide-spread species complex with
marked geographic variation, consisting
of several previously unrecognized and
undescribed forms. In general, P. wil-
liamsi is distinguished from all of these
populations by the combination of fea-
tures listed. The most marked distinction
is in skull morphology, but this is a fea-
ture not always available for identifica-
tion purposes. The carapacial reticula-
tions of P. williamsi are thinner, finer and
more clearly well developed than in P.
geoffroanus or tuberosus where they
tend to be either broad and thick, irregu-
lar, unclear, or reduced. Most popula-
tions of P. geoffroanus and tuberosus
have plastrons with varying degrees of a
dark vermiculate pattem. However, some
also have immaculate plastrons, so this

DESCRIPTION OF PHRYNOPS WILLIAMSI - Rhodin and Mittermeier 71

Figure 9. Map showing distribution of P. williamsi (A), P. geoffroanus (@) and P. hilarii (€). Geographic ranges
of the latter two species only partially shown. Stippled areas correspond approximately to elevations greater than

500 m.

feature in itself is not diagnostic for P.
williamsi. The black head and neck
bands of P. geoffroanus and tuberosus
are almost always relatively thin and
usually confluent in several places. Some
populations have very thin bands, others
somewhat thicker ones, but none are as
prominent as in P. williamsi. Though
occasionally separated from each other,
the facial bands of P. geoffroanus and
tuberosus usually connect broadly either
in front of the tympanum (intermediate
and lower bands connecting), behind the
tympanum (upper and_ intermediate
bands connecting) or at several places

along their course. Many populations
have only two bands plus irregular ver-

miculations. A separate ventral horse-

shoe band is usually not present, though
some specimens of one population of P.
geoffroanus from eastern coastal Brazil
occasionally exhibit this character as
well.

Based on external morphology alone it
would be difficult to determine whether
P. williamsi is distinct from other P.
geoffroanus and P. tuberosus at a specific
or subspecific level. Though sympatric
with P. hilarii, it is allopatric with respect
to P. geoffroanus, the closest known

72 Advances in Herpetology and Evolutionary Biology

populations occurring approximately 300
km to the north in the Rio Pelotas drain-
age (Fig. 9). However, the intervening
region has not been well collected, and a
zone of sympatry may well exist. The dis-
tinct skull morphology of P. williamsi
argues strongly for specific status. The
unique combination of presumably
primitive and derived features is not
shared even by other Phrynops, let alone
other members of the P. geoffroanus
complex.

ACKNOWLEDGMENTS

We thank the following people for as-
sistance in various aspects of this work:
F. Achaval, E. N. Arnold, A. L. de Car-
valho, H. J. Cohen, G. Peters, S. D. Rho-
din, J. Rosado, P. E. Vanzolini and E. E.
Williams. [Illustrations were prepared by
J. Braslin and A. G. J. Rhodin. Rhodin’s
studies were partly funded by the Ameri-
can Philosophical Society, and Mitter-
meiers by the World Wildlife Fund—
WES:

APPENDIX: LOCALITY DATA

Italicized numbers represent speci-
mens of Phrynops williamsi examined or
confirmable literature records. BMNH =
British Museum of Natural History;
MCZ = Museum of Comparative Zo-
ology, Harvard University; MHNM =
Museo Nacional de Historia Natural,
Montevideo; MNRJ = Museu Nacional,
Rio de Janeiro; MZUSP = Museu de
Zoologia da Universidade de Sao Paulo;
MZVC-R = Museo de Departamento de
Zoologia Vertebrados, Facultad de Cien-
cias, Universidad de la Republica,
Montevideo; ZMB = Zoologisches Mu-
seum, Berlin.

BRAZIL: Rio Grande do Sul: MCZ 64135,
BMNH 84.2.5.1, 84.2.5.3, Boulenger 1885, 1886,
1889; Estancia Velha (29°39’S,51°11'W): ZMB 6858,
Hensel 1868; Pikada do Café, Rio Cadéa
(29°37'S,51°15’W): Hensel 1868; Porto Alegre

(30°03’S,51°10’'W): MZUSP 2675; Santa Catarina:
Tubarao (28°29'S,49°00’W): MNRJ 3146. URU-
GUAY: Artigas: Freiberg 1970; Picada del Negro
Muerto, Rio Cuareim (30°45'S,56°05’W): MZVC-R
192-3, Vaz-Ferreira and Sierra de Soriano 1960;
Salto de Agua del Penitente: MHNM 1582, Lema
and Fabian-Beurmann 1977; Paysandu: Arroyo
Chapicuy Grande (31°42'S,57°55'W): Freiberg
1970; Cerro Largo: Arroyo Yaguaron (31°55’S,
53°55'W): MHNM s/n; Rivera: Arroyo Cunapiru
(31°30'S,55°35’W): MZVC-R s/n; Rio Negro: Rincon
del Bonete, Represa del Rio Negro (32°52’S,
56°27'W): Freiberg 1970; Salto: Rio Arapey
(30°55'S,57°50’W): MHNM s/n.

LITERATURE CITED

ACHAVAL, F. 1976. Lista de las especies de verte-
brados del Uruguay. Reptiles. Montevideo:
Museo Nacional de Historia Natural, pp.
26-29.

BOULENGER, G. A. 1885. A list of reptiles and
batrachians from the Province Rio Grande do
Sul, Brazil, sent to the Natural-History Mu-
seum by Dr. H. von Ihering. Ann. Mag. Nat.
Hist., (5)15: 191-196.

____. 1886. A synopsis of the reptiles and batrachi-
ans of the Province Rio Grande do Sul, Brazil.
Ann. Mag. Nat. Hist., (5)18: 423-445.

____. 1889. Catalogue of the Chelonians, Rhyncho-
cephalians, and Crocodiles in the British Mu-
seum (Natural History). London: Trustees of
the Museum, pp. 311.

FREIBERG, M. A. 1970. Validez especifica de
“Phrynops hilarii’ (D. & B.) (Testudines,
Chelidae). Rev. Mus. Argent. Cienc. Nat.
Zool., 10: 189-197.

___. 1971. El] mundo de las tortugas. Buenos Aires:
Albatros, 139 pp.

____. 1972. Validez especifica de Phrynops hilarii
(D. & B.) (Testudines, Chelidae). Acta Zool.
Lilloano, 29: 247-260.

____. 1975. El mundo de las tortugas. 2a. Edicion.
Buenos Aires: Albatros, 134 pp.

Froes, O. M. 1957. Atualizacao da nomenclatura
dos quelonios Brasileiros. Iheringia Zool., 2:
1-24.

GaFENEY, E. S. 1977. The side-necked turtle family
Chelidae: a theory of relationships using
shared derived characters. Amer. Mus. Novit-
ates No. 2620, pp. 1-28.

____. 1979. Comparative cranial morphology of re-
cent and fossil turtles. Bull. Amer. Mus. Nat.
Hist., 164: 65-376.

GALLARDO, J. M. 1980. Estudio ecologico sobre los
anfibios y reptiles del noreste de la provincia
de Buenos Aires (Argentina). I. Reunion
Iberoamer. Zool. Vert., La Rabida, 1977(1980):
331-349.

DESCRIPTION OF PHRYNOPS WILLIAMSI - Rhodin and Mittermeier 73

GoELDI, E. A. 1906. Chelonios do Brazil (Jabotys—
Kagados—Tartarugas). Bol. Mus. Goeldi, 1906:
699-756.

HENSEL, R. 1868. Beitrage zur Kenntniss der
Wirbelthiere Sudbrasiliens. Arch. Naturg.,
1868(1): 323-375.

LEMA, T. DE 1958. Notas sobre os repteis do Estado
do Rio Grande do Sul—Brasil. I. Introducao ao
estudo dos repteis do Estado do Rio Grande do
Sul. Iheringia Zool., 10: 1-18.

LEMA T. DE AND M. E. FABIAN—-BEURMANN. 1977.
Levantamento preliminar dos repteis da regiao
da fronteira Brasil-Uruguai. Iheringia Zool.,
50: 61-92.

LUEDERWALDT, H. 1926. Os chelonios brasileiros.
Rev. Mus. Paulista, 14: 403-470.

McDowELL, S. B. 1983. The genus Emydura
(Testudines: Chelidae) in New Guinea with
notes on the penial morphology of Pleurodira.
This Volume, pp. 169-189.

MEDEM, F. 1969. Estudios adicionales sobre los
Crocodylia y Testudinata del alto Caqueta y
Rio Caguan. Caldasia, 10: 329-353.

MERTENS, R., AND H. WERMUTH. 1955. Die rezenten
Schildkroten, Krokodile und Bruckenechsen.
Zool. Jahrb., 83: 323-440.

MITTERMEIER, R. A., F. MEDEM, AND A. G. J.
RHODIN. 1980. Vermacular names of South
American turtles. Soc. Stud. Amph. Rept. Her-
petol. Circ., 10: 1-44.

PRITCHARD, P. C. H. 1967. Living turtles of the
world. Jersey City, T.F.H., 288 pp.

——. 1979. Encyclopedia of turtles. Hong Kong,
T.F.H., 895 pp.

REISCHL, E., C. O. D. C. DIEFENBACH, AND C. V.
TONDO. 1979. Ontogenetic variation of hemo-
globins from South American Chelonia (Rep-
tilia). Comp. Biochem. Physiol., 62B: 539-544.

RHODIN, A. G. J., AND R. A. MITTERMEIER. 1976.
Chelodina parkeri, a new species of chelid
turtle from New Guinea, with a discussion of

Chelodina siebenrocki Werer, 1901. Bull.
Mus. Comp. Zool., 147: 465-488.

SAPORITI, E. J. 1960. Iconografia de las tortugas que
vivieron y/o viven en el Jardin Zoologico de la
Ciudad de Buenos Aires. Diana; Revista Caza-
Tiro-Canofilia-Pesca, 236, 238, 240, 246:
40-46, 62-65, 68-70.

SCHWEIGGER, A. F. 1812. Prodromus monographiae
Cheloniorum. Konigsb. Arch. Naturwiss.
Math., 1: 271-368, 406-458.

SERRANO, L. L. 1977. Nota sobre uma membrana
extra-embrionaria precoce, no embriao de
Phrinops hilari Dumeril e Bibron, 1835 (nota
previa). Pesqui. Zool. (Porto Alegre), 29: 7-13.

SIEBENROCK, F. 1904. Schildkroten von Brasilien.
Akad. Wiss. Wien. Math. Naturwiss. KI.
Denkschr., 76: 1-28.

___. 1909. Synopsis der rezenten Schildkroten.
Zool. Jahrb. Suppl., 10: 427-618.

STRAUCH, A. 1890. Bemerkungen uber die Schild-
krotensammlung im Zoologischen Museum.
Mem. Ac. St. Petersb., (7)38: 1-127.

VAZ-FERREIRA, R. 1955. La fauna en el manuscrito
de Toller. In Toller, W. 1955. Viaje de William
Toller a la Banda Oriental y Rio de la Plata en
1715. Montevideo, Fac. Human. Cienc., 82 pp.

VAZ-FERREIRA, R., AND B. SIERRA DE SORIANO. 1960.
Notas sobre reptiles del Uruguay. Rev. Fac.
Human. Cienc. Montevideo, 18: 133-206.

WERMUTH, H., AND R. MERTENS. 1961. Schildkroten,
Krokodile, Bruckenechsen. Jena, Gustav
Fischer, 422 pp.

ey ANID) . 1977. Liste der rezenten Amphibien
und Reptilien: Testudines, Crocodylia,
Rhynchocephalia. Tierreich, 100: 1-174.

WIED, M. zu. 1825. Beitrage zur Naturgeschichte
von Brasilien. I. Band. Weimar, Landes, 614

pp.

WILLIAMS, E. E. 1950. Variation and selection in the
cervical central articulations of living turtles.
Bull. Amer. Mus. Nat. Hist., 94: 510-561.

Kenyemys williamsi, a Fossil Pelomedusid
Turtle from the Pliocene of Kenya

ROGER CONANT WOOD!

ABSTRACT. Kenyemys williamsi, a new genus and
species of pelomedusid turtle from the Pliocene of
northwestem Kenya, is described. Kenyemys was a
freshwater turtle whose relationships to other Afri-
can pelomedusids cannot at present be determined.

INTRODUCTION

During the mid 1960's, a series of Har-
vard paleontological expeditions to the
Turkana District of northwestern Kenya,
led by the late Professor Bryan Patterson,
recovered a large quantity of fossil chelo-
nian material. The majority of these
specimens represent two new types of
pelomedusid turtles, one common and
the other rare (Wood, 1971). This fest-
schrift provides an appropriate occasion
for formally naming the latter taxon in
honor of Ernest E. Williams.

SYSTEMATICS

Order Testudines

Suborder Pleurodira

Family Pelomedusidae

Kenyemys gen. nov.
Figures 1-5

Type species. K. williamsi sp. nov.

Distribution. Pliocene, northwestern
Kenya

!Faculty of Science and Mathematics, Stockton
State College, Pomona, New Jersey 08240, U.S.A.

Diagnosis. Differing from all other
members of the family by the following
combination of characters: 1) a series of
elongate tuberosities forming an inter-
rupted keel extending along the midline
rearward from the dorsal surface of the
second neural bone; 2) six neural bones
forming a continuous series, the anterior
end of the first abutting directly against
the rear margin of the nuchal bone and
the sixth one being heptagonal; 3) outer
corners of nuchal bone extending beyond
lateral margins of first vertebral scute; 4)
pentagonal shape of first vertebral scute;
5) only eighth and posterior part of
seventh pairs of pleural bones meeting at
midline of carapace; 6) anterior plastral
lobe truncated; 7) triangular intergular
scute not overlapping anterior end of
entoplastron and only partially separat-
ing the gular scutes along the midline
axis of the plastron.

Etymology. The generic name refers to
Kenya, the nation within whose borders
this turtle was found.

Discussion. Two pelomedusid sub-
families, the Podocneminae and Pelo-
medusinae, can be recognized. However,
their critical distinguishing characters
are based primarily on skull morphology
and cervical vertebra structure (Wood,
1971). Because these parts of the skeleton
are not known for Kenyemys williamsi,
its subfamilial allocation is not at present
possible.

In the past, many African pelomedu-
sids have been referred to the South
American genus Podocnemis. Such iden-

tifications have generally been based on
just a few shell characters (particularly
the shape and arrangement of the neural
bones as well as the presence of subcir-
cular, laterally-placed §mesoplastra)
which, by themselves, are generally of
limited taxonomic importance at the
generic level.

Invariably, when better material be-
comes available or more thorough studies
are carried out, these purported African
representatives of Podocnemis prove
referable to some other genus. Thus,
“Podocnemis” congolensis is now
Taphrosphys (Wood, 1975), “Podoc-
nemis’” antiqua is Shweboemys (Wood,
1970), and “Podocnemis” madagascari-
ensis, whose origins surely lie within
Africa on the basis of fossil material from
the Miocene of central Africa that I cur-
rently have under study, is now generally
accepted as representing a _ distinct
genus, Erymnochelys (Pritchard, 1979).
Moreover, it seems very possible that
shell material from the Oligocene Fayum
Depression of Egypt which has been
described as Podocnemis fajumensis (in-
cluding its synonym P. blanckenhorni;
see Wood, 1971) probably represents the
same taxon as the single skull which has
been designated the type of Dacquemys
(Williams, 1954).

The name Podocnemis, therefore, has
been used in Africa much as it has been
applied to other fossil pelomedusids
elsewhere (see, for example, Wood and
Gamaro’s 1971 description of ““Podoc-
nemis” venezuelensis), as a catch-all
taxon to which a variety of pelomedusids
have been referred more for convenience
than anything else. (Recently-collected
additional material of ‘“Podocnemis”
venezuelensis, for that matter, clearly
indicates that it is a new genus.) As the
fossil record of pelomedusids continues
to improve, it becomes increasingly ap-
parent that there is no compelling reason
to believe that the South American genus
Podocnemis (sensu strictu) has ever
reached other continents.

The suite of features which character-

NEW PELOMEDUSID TURTLE : Wood 75

izes the pelomedusid described here
clearly precludes its reference to Podoc-
nemis. Moreover, its shell is markedly
different from that of other African pelo-
medusids including the African pre-
cursors of Erymnochelys madagascarien-
sis (see Table 1).

On this basis, recognition of a new
genus seems appropriate.

Kenyemys williamsi sp. nov.

Type. National Museum of Kenya (NMK) LT 127,
a nearly complete but somewhat crushed shell.

Hypodigm. The type and NMK LT 128, the an-
terior half of a plastron.

Horizon and locality. Pliocene (Lo-
thagam—1 [Patterson, Behrensmeyer,
and Sill, 1970], now regarded as being
somewhere in age between 5 and 6 mil-
lion years B. P. [Maglio, 1974: Fig. 2]),
Lothagam Hill, southwestern Turkana
District, Kenya (approximately 2°53’N,
36°04’E).

Diagnosis. As for the genus.

Etymology. I take great pleasure in
naming this species in honor of Ernest E.
Williams, in recognition of his significant
contributions to knowledge of both fossil
and living turtles, not to mention herpe-
tology in general.

Description. Essentially all of the shell
of Kenyemys williamsi has been pre-
served. Despite some distortion resulting
from cracking and crushing, it is apparent
that the carapace was roughly oval in out-
line and probably moderately arched.
Restorations of the shell are shown in
Figures 3-5.

At the front of the carapace, the nuchal
bone is hexagonal and slightly broader
than long. There are six neurals arranged
along the midline in an uninterrupted
series. The fusiform first neural abuts
squarely against the rear margin of the
nuchal; posteriorly it is rounded to fit
into a semicircular notch at the front of
the second neural. Neurals two through
five are of approximately the same size

(thjpwuo.1qg

‘qd ul [euo

-sejued yda0xa)
IepnsueLy

(149ULOLD *G)
[euoseydoy
“(vohqy “S)
[euoseyued

[euose}
-uad 0} 1ejnsue
-LQ :o[/qeLiea

aynos Ie[ns

Jeuose}ued repnsueLy [euosedoy JepnsueLy -I9}UI Jo odeys (QT

pepunol

[eAo

G-% :oTqeurea

Jeuosezued

ou

sAeMye OU
yng Aj]jensn
S-L

[euos

-ejyued ATjensn

ysnoyye ‘aiqe

-LIBA yeEUMOIUIOS
ou

pepunor

[eAo

“AG

[euosejued

ou

sod

[euose}ued
ou

peyeounya

0} pepunor

a] qeurea
[eao

“AG

[euosydoy

ou

SoA

jeuoseyued
ou

pepunor
¢

“”l

[euosequed

ou

SoA

[euoseyued
ou

Dsnpawuojad §shjayoouwhigq {. stuausopog,, }sisuajosuoo

peyeounya
[eAo

nw“

[euose}ued

sod

ou

[euoseyued
ou

,Shuad0a1a4s

pepunos
ULIOJIP10O

“AG

[euosexoy

ou

SoA

Jeuoseyued

ou

pnbiyun

payeouny
[eao

“”l

[euosrjued

sodA

sod

jeuose}doy
sod

ISWDIPIUN

9qo|
[eyseyd 10119}
-ue jo odeys
sordrieo

jo odeys
jesAdeid

-ns pue [einou
Se] UIaMjOq
sovdeivo Jo oul]
-pluwl ye suljoou
sjeineyd jo
sired Jo Joquinu
ayNos [BIqa}I9A
ysi1y Jo odeys
demos
[PIQeHOeA ysITy
jo sulsireul
[ereye] puod
-eq pus}xe 9u0g
yeyonu saop
deuog jeyonu
jo pua 101190}3sod
ysutese A]}091
-Ip jnge suO0q
[eineu ysIlf soop
souog [eineu

jo Jaquinu [e}0}

9uo0g [einou

yse] Jo adeys
dquasoid jooy

YALOVYVHO

shydsosydn], shwaoqanys shwahuay

“SHTLYNL GISNGYNOTAd NVOIWAY UV TIINIS LYHMANOS YHHHLO JO ASOHL HLIM GHYVdNOD ISWVITIIM SAWHANAN AO SHALOVYVHO TITANS “7 ATA

Advances in Herpetology and Evolutionary Biology

NEW PELOMEDUSID TURTLE - Wood 177

‘snuas ay} 0} a[qeiayar Apreayo st sutApnys wre AMuaxino | yorum AaTpeA YI [PHueD Ss voupy Woy [eLo}eUr [Issoy “IoAaMOY ‘IeOseSepry|
‘sapjowausopod | 'g,, pue ‘sisuawn(nf -g,, “Wpwosq .-d

‘suuedaid Aue) ue | uonduosap asoym suaurtoads payodaiun A|snotAeid wos paatiap st UUIN]OO sty} OF eJep ayy Jo Weg
“TOULIOF OY} FO JeYA UY} Padtasoid Iayeq A[qeIapIsuoD SuIeq Jaye] aUyY JO [eUOyeUL ‘Noh

»

‘popydhsan |g

‘ayetidoidde suisas a1ay snues SIq} JO UOIsNOUr ‘souay{

>

0} peulzuoo pur o1dAjououl juasaid ye st snuas SIUL§
,, ‘POUSTUBAUOD JO sUTeU SY} 0} Salads Ino} Jayar AT@Ae}UD} It

t

qi] °S pure tawo019 “§ ‘paquiosep useq savy satoads om],

Se ——— eee

s}[Npe sinjzeuw
UI S}sIsiod
usayo uoyseyd
24} JO uoni0d
[eQus9 ay} Jo
uoyeyQsousy (e

119”
se sjeioumny at}
JO [[e 10 pred pue

skemye sie[ns

sok

popued
-xo uPyy
Joyjyei Ale

-Ia}e] pexoLys
-u0d ATYSITS
UdAa IO papIs

-[e] [ered

sjeineyd jo
ired yquaaas (8 =

Ajuo srejns
jo wed 101190}ue

Apysi[s ATUO ‘os
jt ‘you ATyensn

1194

se s[eioumny ou
jo [[@ 10 owos
UdAB IO ‘sIe[NS
JO []e ‘srepns
jo ped 101190}uUKe
A[UO oyeredas
ABUL :O][qeLIeA

dey]
-I19A0 }0U ABUL 10
ABUL :o[QeLiea

“UILI 19}NO 0}
0} [eT[ered
suluunI oqo]
[egsed 10110}
-sod jo apts
Jaye JO voy
-INS [B1IOSIA

uO 8BpLI MOT (q
2qo]|
Joliejue ueyy
Jesuo] sown
@ Ajysnor
2qoy| [ease|d

IOLI9}sod (e

AT9}9]d
-WI09 sTe1ouIny
pue sie[ns y}0q

SoA

pedoyjea

-op Apyeom
sesse.Qyng [euins

-UI pue [erxe (e

Ajay0 0
-uI09 sTeIouInYy
pure sirens y0q

SoA

([9ey B “IaA9
-Moy ‘J0U)
asp oUI[prul
snonuquoo
yun]q worry
ABMe BUI
-doys ‘paaino
ueyy Joyzel
yey speineyd (o
qo] 1OL1e}uUe
uvy} 1eSuU0]
souy YT
eqoy| jeyse|d
Iot1a}sod (q
OUI| Pru
UL SUQZOUL
woy s[euIs
-Ieul Jo 11ed
SIT} SUQUDA
-o1d ‘aoed
-e1ed JO UIs
-IBUL IOLIO}Ue
sayoeal anos
[PIQeHOA ysIF (e

souln

-9ul0s sTeloumny
jo pred 10L190}uUK
pue ‘poyeredas
7 F232) (600 Cofo)

ATUO
sre[ns fo

sAeMy[e srepns jred 10L19}Uue

sodA

ou

soiny
-eoy [eUOTIPpe

qoulfpru

UI SeyOs [eIOU
-ny JO repns ayy
Jaye a}eredas
Ie[Ns19}UI saop
duoyse[dojua
jo pue 10119}
-ue 0}U0 de]1aA0

(€T

(ZI

Je[Ns19zUI svop (TT

78 Advances in Herpetology and Evolutionary Biology

and roughly hexagonal, having antero-
lateral sides that are much shorter than
the postero-lateral ones. The sixth neural
is considerably longer than wide and in-
stead of being pentagonal, as is typically
the case for the last neural in podoc-
nemine pelomedusids, it is heptagonal.
In contrast to the common fossil pelo-
medusid from Lothagam, in which three
pairs of pleurals are involved, separation
between this terminal neural and the tri-
angular suprapygal is effected only by
the midline union of part of the seventh
and all of the eighth pair of pleurals. Ex-
tending along the antero-posterior ridge
of the carapace is a series of keels which
become progressively more prominent
towards the rear of the shell. The first of
these protrudes only slightly from the
surface of the second neural and the front
half of the third. The second, arising from
the posterior surface of the third neural
and continuing back onto the front half of
the fifth, has a higher crest than the pre-
ceding one, while the third and highest
of the keels extends from the rear portion
of the fifth neural back onto the eighth
pair of pleurals. A median bulge on the
suprapygal suggests that a fourth keel
may also have been present. Carapacial
keels are unusual features in pelomedu-
sid turtles. Immature examples of Podoc-
nemis unifilis and P. sextuberculata can
have quite prominent tubercular keels
(P. C. H. Pritchard, personal communi-
cation). Similar but less pronounced
keeling is also characteristic of adults of
various species of Pelusios (Loveridge,
1941; Pritchard, 1979).

Owing to damage at the posterior end
of the carapace, it is not possible to de-
termine the shape of the pygal, but it was
probably rectangular as in all other
pelomedusids. Nine peripheral bones
remain attached to the carapace on either
side and, in addition, two complete but
disarticulated peripherals as well as
several fragments have also been pre-
served, so that it appears likely that the
actual complement of peripherals was
the normal pelomedusid number of

eleven pairs. Flanking the neurals are
eight pairs of pleurals which decrease in
size from front to rear.

Except for the shape of the first verte-
bral, the scute pattern on the carapace of
Kenyemys williamsi is not exceptional.
The first vertebral was pentagonal, with
its apex directed forward and its lateral
sides more or less parallel. The lateral
ends of the first pair of marginal scutes
extended beyond the greatest width of
the first vertebral. The nuchal bone at its
greatest breadth extends beyond the
lateral boundaries of the first vertebral.

The anterior plastral lobe of Kenyemys
williamsi is truncated, with pronounced
notches occurring at the junction be-
tween the gular and humeral scutes along
the epiplastral margin to mark the posi-
tion where the sides begin to curve later-
ally. The small, triangular intergular
scute did not extend back onto the ento-
plastron in either of the two known speci-
mens of Kenyemys. The plastral scute
sulci and suture patterns conform to the
typical pelomedusid configuration.

Most of the pelvis is preserved in the
type but is somewhat crushed and dis-
torted. From what can be seen, it appears
to be similar to those of other podoc-
nemine pelomedusids. The upper half of
the acetabulum is formed by an excava-
tion in the outer surface of the ilium,
while the ischium and pubis each contri-
bute about equally to the formation of the
lower half of this socket. A low ridge on
the visceral surface of the plastron,
formed by projections running inward
from the base of the ischia, serves to con-
nect these elements at the midline.

Estimated measurements for the shell
are: length of carapace, 32 cm; length of
plastron, 28 cm. The plastral formula is:
abdominal>pectoral>femoral>anal>gu-
lar>intergular>humeral.

DISCUSSION

Kenyemys was clearly a fresh water
turtle. The sediments of the Lothagam-1

unit as well as the fauna recovered from it
both indicate a mixture of channel,
floodplain, and backwater depositional
environments (Patterson, Behrensmeyer,
and Sill, 1970). Most of the Lothagam
pelomedusids were recovered from
sandstones, the type of Kenyemys being
the only one to have been found in a
rather limey clay. This might indicate
that Kenyemys occupied a somewhat dif-
ferent kind of habitat than did the other,
much more abundantly represented
Lothagam pelomedusid. Whatever the
case, it seems reasonable to suppose that
these two species were broadly sym-
patric, as both pelomedusid taxa have
been discovered at essentially the same
stratigraphic level.

The relationship of Kenyemys to other
African pelomedusids is at present ob-
scure. Perhaps the most distinctive fea-
ture of its shell is the midline keel of the
carapace, a character not present on any
other adequately known podocnemine
from the continent or adjacent island of
Madagascar.

Rather tenuous evidence exists to sug-
gest that there may have been another
African fossil pelomedusid characterized
by a keeled carapace. An isolated nuchal
bone has been referred to Latisternon
microsulcae, the type of which is a single
left epiplastron from Olduvai Gorge,
Tanzania (Auffenberg, 1981). This nuchal
bone was found at a different site with-
in Olduvai Gorge than the type speci-
men, so the association of the fragments
is questionable. Whether or not the
association is correct, the fact remains
that a median keel is reported to rise from
the dorsal surface of the nuchal. The
available evidence is insufficient to
determine either the extent of this keel or
whether it was continuous or interrupted
along its length. What little is known,
does, however, clearly reveal differences
in detail, since the keel of Kenyemys first
appears farther back along the midline of
the carapace. Moreover, the proportions
of the nuchal bones in Kenyemys and
Latisternon differ markedly, that of the

NEW PELOMEDUSID TURTLE - Wood 79

former being broader than long whereas
that of the latter is considerably longer
than broad. No reason exists, therefore, to
believe that there is any close relation-
ship between these two taxa based on
their common possession of a carapacial
keel.

With respect to another notable feature
of the carapace of Kenyemys, only Stere-
ogenys libyca, among other African
podocnemines, also has a first vertebral
scute whose lateral borders lie within the
confines of the nuchal bone. But in other
features (e.g., number, shape, and ar-
rangement of neural bones, shape of an-
terior plastral lobe and patterns of scutes
thereon) S. libyca differs markedly from
Kenyemys and there is no reason to pos-
tulate a close relationship between these
species either.

The truncated anterior plastral lobe of
Kenyemys is superficially similar to those
of several other African fossil pelomedu-
sids (e.g., “Podocnemis” podocnemoides
from the late Eocene of Egypt, “Podoc-
nemis” fajumensis from the Oligocene of
Egypt, “Podocnemis” aegyptiaca from
the Miocene of Egypt, both species of
Stereogenys and_ Latisternon micro-
sulcae from the Pleistocene of Tanzania)
but differs in detail. For example, the in-
tergular scute of “Podocnemis” podoc-
nemoides, while triangular, is relatively
large and extends well back onto the sur-
face of the entoplastron. Moreover, the
gulars are small, triangular scutes con-
fined to the epiplastra (Wood, 1971: Fig.
5). “Podocnemis” fajumensis and “Po-
docnemis” aegyptiaca have virtually
identical anterior plastral lobes. In both,
the triangular intergular scute apparently
varied in size, in some cases being con-

. fined to the median portions of the epi-

plastra while in others extending slightly
or even considerably back onto the ento-
plastron. Further, in all cases, the lateral
extent of the anterior truncation is not
proportionately so broad as in Kenyemys
(Wood, 1971: Figs. 7, 8, 10). The lateral
extent of the truncated part of the anterior
lobe of Latisternon is also considerably

80 Advances in Herpetology and Evolutionary Biology

less than in Kenyemys. Moreover, Lati-
sternon’s intergular was pentagonal and
reached back onto the entoplastron,
while its gulars were triangular and con-
fined to the surface of the epiplastra.
Both species of Stereogenys also had
large intergulars, pentagonal in S. libyca
and heptagonal in S. cromeri. Finally, for
all of these turtles, other characteristics of
the shell preclude the likelihood of any
particularly close relationship.

In sum, no probable precursors of
Kenyemys williamsi can as yet be identi-
fied, nor is it at present possible to de-
termine whether this species gave rise to
any descendants.

ACKNOWLEDGMENTS

I am indebted to the late Bryan Patter-
son, who not only made it possible for me
to study the material described in this
paper but who also provided me the op-
portunity to be a member of his paleonto-
logical expeditions to Kenya during the
summers of 1965, 1966, and 1967. Emest
Williams deserves much of the credit
(perhaps blame might be a more appro-
priate term) for encouraging my interest
in turtles while I was a graduate student
at Harvard; I thank him not only for his
freely given advice but also for access to
the resources of his herpetology depart-
ment at the MCZ. The type specimen of
Kenyemys williamsi was collected in the
field during the summer of 1967 and then
subsequently prepared in the laboratory
by Arnie Lewis, whose work in both
places is unsurpassed by anyone else I
have ever met. The illustrative skills of
Laszlo Meszoly are responsible for the
shell reconstructions depicted in Figures
3-5. Peter Pritchard was kind enough to
review a manuscript version of this
paper.

Professor Patterson’s 1967 expedition
to Kenya was supported by National

Science Foundation (NSF) Grant GA-
425. Several grants from the National
Geographic Society enabled me to study
relevant collections of fossil and recent
turtles in Africa, Europe, and the United
States. An NSF Science Faculty Profes-
sional Development Award for the
academic year 1980-81 enabled me to
prepare this paper for publication. Dur-
ing this period I was a visiting research
scientist with the Florida Audubon So-
ciety, for whose hospitality I am most
grateful.

LITERATURE CITED

AUFFENBERG, W. 1981. The fossil turtles of Olduvai
Gorge, Tanzania, Africa. Copeia, (3)1981: 509-
522.

LOVERIDGE, A. 1941. Revision of the African ter-
rapin of the family Pelomedusidae. Bull. Mus.
Comp. Zool., (6)88: 467-524.

MaGLIo, V. J. 1974. Late Tertiary fossil vertebrate
successions in the northern Gregory rift, east
Africa. Ann. Geol. Surv. Egypt, 4: 269-286.

PATTERSON, B., A. K. BEHRENSMEYER, AND W. D.
SILL. 1970. Geology and Fauna of a new Plio-
cene locality in northwestem Kenya. Nature,
226, nr. 5249: 918-921.

PRITCHARD, P. C. H. 1979. Encyclopedia of Turtles.
Neptune, New Jersey, T.F.H. Publications,
895 pp.

WILLIAMS, E. E. 1954. New or redescribed pelo-
medusid skulls from the Tertiary of Africa and
Asia (Testudines, Pelomedusidae), 1. Dac-
quemys paleomorpha, new genus, new species
from the lower Oligocene of the Fayum,
Egypt. Breviora Mus. Comp. Zool. No. 35, pp.
1-8.

Woop, R. C. 1970. A review of the fossil Pelomedu-
sidae (Testudines, Pleurodira) of Asia. Brevi-
ora Mus. Comp. Zool. No. 357, pp. 1-24.

___.. 1971. The fossil Pelomedusidae (Testudines,
Pleurodira) of Africa. Ph.D. thesis, Harvard
University.

___. 1975. Redescription of “Bantuchelys” con-
golensis, a fossil pelomedusid turtle from the
Paleocene of Africa. Rev. Zool. Africaine, 89:
127-144.

Woop, R. C., AND M. L. Diaz DE GAMERO. 1971.
Podocnemis venezuelensis, a new fossil pelo-
medusid (Testudines, Pleurodira) from the
Pliocene of Venezuela and a review of the his-
tory of Podocnemis in South America. Breviora
Mus. Comp. Zool. No. 376, pp. 1-23.

eee ae

NEW PELOMEDUSID TURTLE - Wood 81

Figure 1. Carapace of Kenyemys williamsi (type specimen, NMK LT 127).

9

Advances in Herpetology and Evolutionary Biology

Figure 2. Plastron of Kenyemys williamsi (type specimen, NMK LT 127).

|
|

NEW PELOMEDUSID TURTLE - Wood

Figure 3. Reconstruction of the carapace of Kenyemys williamsi.

83

84 Advances in Herpetology and Evolutionary Biology

.

Figure 4. Reconstruction of the plastron of Kenyemys williamsi.

85

NEW PELOMEDUSID TURTLE - Wood

“MAIA [esO}e| Ul ISWeI/IM SAWAaAUaY JO |]eYUS BY} JO UOHONsSUODaY “G aunbi4

Variation in Hispaniolian Sphaerodactylus

(Sauria: Gekkonidae)

RICHARD THOMAS!
ALBERT SCHWARTZ?

ABSTRACT. A review of the gekkonid species
Sphaerodactylus altavelensis assesses variation and
clarifies the nomenclature. Four subspecies (one
described herein) are recognized. A second species,
apparently related to S. altavelensis, is described as
S. williamsi.

INTRODUCTION

The West Indian island of Hispaniola
is the center of radiation of members of
the notatus group of the gekkonid lizard
genus Sphaerodactylus. The difficilis
complex (of the notatus group) is the
most widespread and diverse; this com-
plex is named for S. difficilis Barbour,
the first species in the complex to be de-
scribed. The last revision of the complex
was that of Shreve (1968); since that time
we have named four new species (S.
cryphius, S. ocoae, S. nycteropus, S.
zygaena; Thomas and Schwartz, 1977;
Schwartz and Thomas, 1977), and re-
moved a subspecies of S. difficilis as a
separate species (S. randi; Schwartz,
1977). This then is our fourth paper deal-
ing with the Hispaniolan difficilis com-
plex. There remain eight nominal species
in the complex (Schwartz and Thomas,
1975); among these species we have
made nomenclatural changes that require
justification. The present paper justifies

1Biology Department, University of Puerto Rico,
Rio Piedras, Puerto Rico 00931.

2Miami-Dade Community College, North Cam-
pus, Miami, Florida 33167, U.S.A.

these changes and assesses the variation
in one of the remaining species, S. al-
tavelensis Noble and Hassler and de-
scribes another. Our arrangement is due
primarily to having collected and ex-
amined in detail many more specimens
than were available to Shreve (1968) and
to having visited many areas in His-
paniola whose Sphaerodactylus fauna
had been either poorly known or al-
together unknown. In the following ac-
counts, measurements and counts in
parentheses represent mean values.

TAXONOMIC ACCOUNTS

Sphaerodactylus altavelensis Noble and
Hassler

Definition. A species of Sphaerodac-
tylus with large acute, strongly keeled,
flattened, imbricate dorsal scales, axilla
to groin 20 to 36; no area of middorsal
granules or granular scales; dorsal body
scales with four to six hair-bearing scale
organs around apex and distal edge, each
with one hair. Dorsal scales of tail
keeled, acute, imbricate, flat-lying; ven-
tral scales of tail smooth, rounded, en-
larged midventrally. Snout blunt, not
depressed or decurved; snout scales
moderate in size, rounded, keeled, juxta-
posed, cobblelike to weakly raised on
posterior edge, or weakly imbricate; 2
postnasals; 0 to 3 internasals; upper la-
bials to mid-eye 3; dorsal head scales

HISPANIOLIAN SPHAERODACTYLUS - Thomas and Schwartz 87

small, rounded to subhexagonal, keeled,
juxtaposed and cobblelike to weakly
raised on posterior edge; temporal scales
broad, rounded, keeled, raised, weakly
imbricate; first infralabial subrectangular
to subtriangular; gular scales between in-
fralabial rami rounded to hexagonal,
moderately-sized, smooth to keeled, jux-
taposed to weakly imbricate; central gu-
lars small, acute, imbricate, smooth to
keeled; some gular scales keeled in
nearly all specimens; chest scales
smooth; ventral scales smooth, rounded,
imbricate, axilla to groin 22 to 32; scales
around midbody 38 to 62; fourth toe
lamellae 6 to 13; escutcheon with broad
central area and extensions onto thighs to
underside of knee area (3—8 Xx 11-27).

Color pattern in life weakly dichro-
matic; males with yellow throat ground
color and reduced pattern, occasionally
unicolor. Pattern of paired light areas on
upper surface of head, followed by
second pair on occiput and neck; dark
scapular patch and scapular ocelli sur-
rounded by continuous light border con-
cave anteriorly and convex posteriorly;
dorsal body pattern of dark spotting or
lines or combination of both; venter pale
with dark lines; throat with dark lines or
mottling; sacral pattern of dorsolateral
light and dark lines; and tail with dark
middorsal zone and series of irregular
ocelli. We have examined 847 S. alta-
velensis.

Basic pattern. In order to facilitate
descriptions of variation in the subspe-
cies of S. altavelensis, we will describe in
some detail here the basic pattern for the
species. The fullest pattern development
is found in some females of two of the
subspecies. For convenience of discus-

sion the basic pattern is broken down.

into the following elements:

Anterior cephalic figure. A posteriorly
directed, bilobed, light figure outlined in
dark pigment extending to about the
level of the ear opening; the breadth
covers most of the dorsal surface of the
head; the lines forming the lateral edges
originate as superior postocular stripes.

Posterior cephalic figure. A bipartite,
dark-edged figure in the occipitonuchal
area having posteriorly directed lobes,
the anterior margins being formed by the
posterior edges of the anterior cephalic
figure.

Scapular figure. A complex figure con-
sisting of 1) a black, roughly rectangular
scapular patch enclosing; 2) a pair of pale
ocelli; and 3) a narrow light border of
greater extent than the patch (i.e., includ-
ing an area of ground color peripheral to
the patch), in roughly the shape of an in-
verse concave-convex trapezoid, and
formed of a dark-edged light line.

Body pattern. About ten longitudinal
dark lines occur on the dorsum (3) and
sides (2) and venter (about 5); the dorsal
series extends from the scapular figure to
the sacral region; the lateral lines extend
from the post-axillary region to the groin,
and the ventral lines, which are faintest,
extend irregularly from about midbody to
the groin and vent. The lineate pattern is
rarely as fully expressed as described,
but many females show indications of it.
Often, however, the dorsal body pattern
is one of small dark spots about one to
three scales in area, irregularly distribu-
ted. The spotted pattern seems to be de-
rived from fragmentation of the linear
one, in which even those individuals
with spotted patterns have lines in the
sacral region and on the venter.

Sacral figure. A pair of dorsolateral
stripes on each side form the dark edges
of a light dorsolateral sacral zone, which
extends beyond the sacral region and
fades on the posterior body and base of
the tail.

Tail pattern. Six very dark longitudinal
lines around the circumference form a
lineate pattern: a dorsolateral pair of
lines which is invaded by several pairs of
pale ocelli, a lateral pair of lines, and a
ventrolateral pair; the spaces between
the dorsolateral and lateral, and the
lateral and ventrolateral lines are very
light, forming a contrastingly striped
lateral tail pattern. The lateral and ven-
trolateral stripes with their pale inter-

88 Advances in Herpetology and Evolutionary Biology

spaces continue anteriorly onto the pos-
terior surface of the thigh to form a promi-
nent zonation there.

Male pattern. Males lose the basic pat-
tern as they mature and assume either a
nearly uniform brown or a uniformly
spotted pattern. Faint remnants of the
basic pattern may persist, and in a few
individuals (and in one subspecies) ap-
parently mature males retain rather fully
developed basic patterns. The scapular
patch is present in males of three sub-
species (see below). Some males show an
intensification of the head markings in
which the edges of the markings become
darkened and intensified with eventual
loss of the lighter areas resulting in a
vermiculate pattern.

Distribution. Isla Alto Velo; on the

y

GY,
A. BREVIROSTRATUS oO

Figure 1.

mainland of Hispaniola, from the Llanos
de Azua in the east, west continuously to
the Baie de Port-au-Prince in xeric to
semi-xeric situations, along the north
coast of the basal Tiburon Peninsula to
the vicinity of Petit-Goave and south
through the Vallée de Trouin (presump-
tive) to the vicinity of Cayes Jacmel on
the south coast. A population occurs in
the vicinity of Jérémie on the distal
Tiburon Peninsula, and another on tiny
Ile a Cabrit in the Golfe de la Gonave
(Schwartz, 1979). On the north island,
Sphaerodactylus altavelensis is known
from the coast north of Port-au-Prince to
Pierre Payen and inland in the vicinity of
Lascahobas and in the northwest from
Gonaives to Port-de-Paix and Ennery in
Haiti (Fig. 1).

I
4 YY y: J
" Gy wf
NS A y cae wh

LUI

\
“/ ona
4 Pe. a
al
> 7 A. ENRIQUILLOENSIS

7
ly

Q fo 20 30 40
KM
@A. ALTAVELENSIS

Map of western and central Hispaniola, showing the known distributions of S. altavelensis (@ ) and S.

williamsi (&). Subspecies of S. altavelensis are noted and their ranges shaded.

HISPANIOLIAN SPHAERODACTYLUS - Thomas and Schwartz 89

Sphaerodactylus altavelensis altavelensis
Noble and Hassler
Figures 2A, 3A

Sphaerodactylus altavelensis Noble and Hassler,
1933. Amer. Mus. Novitates, 652: 7.

Type locality. Isla Alto Velo, Republica Domini-
cana.

Holotype. AMNH 51488.

Definition. An insular subspecies of
Sphaerodactylus altavelensis character-
ized by large size, high midbody scale
counts (44-62), and very bold contrasting
basic pattern (females).

Distribution. Isla Alto Velo, Republica
Dominicana.

- Variation. We have examined 22
specimens. Largest male (AMNH 51490)
28 mm snout-vent length (SVL), largest
female (MCZ 45947) 29 mm SVL; dorsal
scales between axilla and groin 26 to 36
(30.1); ventral scales axilla to groin 22 to
32 (27.9); scales around midbody 44 to 62
(51.7); supralabials to mid-eye 3/3; inter-
masals 1 (N=21), 3 (N=1); fourth toe
lamellae 8 to 11 (mode 10); throat scales
keeled; escutcheon 3 to 6 X 22 to 27.

Males have a uniform grayish ground
color and a nearly uniform scattering of
dark brown spots over the dorsal surface
of the head and body; the scapular patch
and ocelli are absent or reduced to
minute ocelli connected by a small
scapular patch that does not include the
ocelli. Most females show a bold de-
velopment of the basic pattern in which
the anterior and posterior cephalic fig-
ures are heavily outlined in dark pig-
ment; the scapular patch is large and en-
closes the pair of ocelli, and the light
scapular figure border is markedly con-

cave-convex producing a comic-mask

mouth appearance (Fig. 3A). The dorsal
body pattern of females is heavily mot-
tled or vermiculate with a tendency
towards formation of transverse dark
markings reaching an extreme in AMNH
51478. Some females show a simplifica-
tion and fragmentation of pattern that ap-

proaches the spotted male pattern; the
anterior and posterior cephalic figures
are fragmented; the scapular patch and
ocelli are reduced; and the scapular fig-
ure border is absent. Lateral and sacral
lines are prominent in consonance with
the boldness of the rest of the pattern;
ventral lines are present as in the other
populations of the species.

Remarks. Shreve (1968) regarded S.
altavelensis as a species confined to Isla
Alto Velo and suggested that it was re-
lated to S. difficilis randi (= S. randi;
Schwartz, 1977). The bold scapular figure
and dorsal spotting of some randi popula-
tions do bear a resemblance to altavelen-
sis. However, the trilineate head pattern
of randi plus the presence of a com-
pletely expressed basic pattern such as is
found in Shreve’s S. brevirostratus (and
the lack of these pattern elements in
randi) leave no question of the relation-
ship of the Alto Velo sphaerodactyls to
brevirostratus. In fact, the scapular pat-
tern of S. randi only superficially re-
sembles that of S. altavelensis. Shreve
examined only the two MCZ paratypes of
altavelensis, in which the pattern is not
so well expressed as in some of the other
specimens.

Sphaerodactylus a. altavelensis is
separated from other populations of the
species by not only a water gap but also,
as far as is known, by a significant geo-
graphic hiatus that includes all of the
Peninsula de Barahona. The nearest
records are the region just to the south of
the city of Barahona on the east, and the
region of Cayes Jacmel to the west. Al-
though the species may occur farther east
along the south coast of Haiti than Cayes
Jacmel, it apparently does not extend to
the region of Pedernales on the Domini-
co-Haitian border—an area that has been
relatively well-collected and from which
other species of Sphaerodactylus are
known. Another similarly relict member
of the Alto Velo fauna is Leiocephalus
vinculum altavelensis Noble and Hassler
(Schwartz, 1967).

90

Advances in Herpetology and Evolutionary Biology

Figure 2. Dorsal views of subspecies of S. altavelensis and S. williamsi as follow: A. S. a. altavelensis (AMNH 51485). B. S. a. enriquilloensis (ASFS

V23288). C. S. a. brevirostratus (ASFS X2376). D. S. a. lucioi

(MCZ 156208—holotype). E. S. williamsi (MCZ 156209—holotype).

HISPANIOLIAN SPHAERODACTYLUS « Thomas and Schwartz 91

Figure 3. Diagrammatic dorsal views of cephalic and scapular patterns of S. altavelensis (see text for discus-
sions of variation), as follow: A. S. a. altavelensis. B. S. a. enriquilloensis, C. S. a. brevirostratus. D. S. a. lucioi.

Specimens examined. REPUBLICA DOMINICANA: Isla
Alto Velo: AMNH 51472-86, AMNH 51489-91,
MCZ 4594647, AMNH 51488, ASFS V26910.

Sphaerodactylus altavelensis enriquilloen-
sis Shreve
Figures 2B, 3B

Sphaerodactylus brevirostratus enriquilloensis
Shreve, 1968, Breviora, 280:14.

Type locality. 4 km E La Descubierta, near Lago
Enriquillo, Independencia Province, Republica
Dominicana.

Holotype. MCZ 57846.

Definition. A subspecies of Sphaero-
dactylus altavelensis characterized by
lack or very weak development of the
scapular patch and ocelli.

Distribution. The Valle de Neiba east
of the Dominico-Haitian border, south
around the eastern edge of the Sierra de
Baoruco to slightly beyond the city of

Barahona, eastward to the Llanos de ~

Azua (17 km E Azua), south to Punta
Martin Garcia and north at least to the
Azua-San Juan province border (south-
east of Guanito). A specimen from near
Vallejuelo within the Sierra de Neiba is
tentatively referred to this subspecies
and may indicate a more widespread

occurrence of the species in the Valle de
San Juan and associated areas.

Variation. We have examined 408 spe-
cimens and have taken full or partial
counts on 314 of these. Largest males 26
mm, largest female 28 mm; dorsal scales
axilla to groin 23 to 31 (26.1); ventral
scales axilla to groin 23 to 32 (28.0);
scales around midbody 38 to 50 (46.0);
supralabials to mid-eye usually 3/3,
rarely 3/4 or 4/4; internasals 1 (N=205), 2
(N=28), 3 (N=1); fourth toe lamellae 7 to
11 (mode 9 and 10); throat scales keeled;
escutcheon 4 to 8 X 11 to 27.

The anterior and posterior cephalic
figures are absent in virtually all males,
although some show a residual trace of
them. Scapular patches and ocelli are
present (not necessarily in the same
specimen) in scattered males (the fre-
quency is less than it is for the same pat-
tern elements in females). Ocelli in
males are very reduced, often to the size
of a single scale; and the patch, if present,
may be no more than a small area of black
pigment between the two ocelli.

The anterior cephalic figure is well
developed in females; the posterior fig-
ure is not so strongly developed and is
frequently represented by only a pair of

92

light crescents in the occipital region
demarcating the posterior end of the pos-
terior figure (Fig. 3B). Eight of 86 Valle
de Neiba females and juveniles have pos-
terior extensions of the anterior cephalic
figure from roughly the middle of each
lobe, which may fuse with the posterior
figure or fade out in the parietal region.
Although absent or very reduced, scapu-
lar ocelli and patches are present in a
small proportion of all specimens
throughout the range of the subspecies.
For example, 17 of 86 Valle de Neiba
females and juveniles have discernible
ocelli; among these, some lack scapular
patches of any noticeable extent, and
others have them weakly developed, two
having what might be termed moderate
development of the patch. Eight of the
ocellate females from the Valle de Neiba
(28 females) occur in the Las Clavellinas
series. None of the Duvergé vicinity
specimens has ocelli, and only one of the
females from Mella (28 females from both
localities combined) has ocelli. Thus the
frequency of scapular ocelli seems lower
in the females from the southem side of
the Valle de Neiba, although the fre-
quency of scapular ocelli, even though
very reduced in expression, is as high or
higher (15 of 50) in the Barahona-and-
vicinity sample as in the Valle de Neiba
one. In the long series (159 specimens)
from the northeastern part of Barahona
Province, indications of ocelli are found
on but two females (although six males
have faint small ocelli). The scapular fig-
ure border is complete or nearly com-
plete in 22 of 86 Valle de Neiba females
and juveniles; others have some indica-
tion of this feature, most frequently the
concave anterior border. Of the five (of
50) females from the Barahona region
having indications of the scapular figure
border, it is complete in only one. Ap-
proximately the same situation holds for
the more eastern population of S. a. en-
riquilloensis: indications of the scapular
figure border are seen, but in no speci-
mens are they complete. Dorsal body
coloration in the subspecies varies from

Advances in Herpetology and Evolutionary Biology

nearly unicolor to heavily spotted or
vermiculate with dark brown; dorso-
lateral and lateral stripes extend the
length, or nearly the length, of the body
in some specimens. Ventral lines are
present to a variable extent in nearly all
specimens.

The 13 Punta Martin Garcia specimens
are noteworthy for their virtual lack of
any pattern development in males, fe-
males, and juveniles. Two show ghost
remnants of the posterior cephalic figure
and scapular figure border, and others
show brief ghost traces of the posterior
cephalic figure; spotting and flecking is
very reduced, although ventral lines are
present.

The single specimen from 7 km NW
Vallejuelo (between the north and south
ranges of the Sierra de Neiba) clearly
shows the anterior cephalic figure, traces
of the posterior figure, and the anterior
and posterior edges of the scapular figure
border; there is a faint smudge that may
represent the scapular patch and a small
light marking that gives the appearance
of a diverticulum of the anterior edge of
the scapular figure border, but which
may be a scapular ocellus. The dorsum is
mottled with irregular dark markings
tending to form transverse light bars in
the middorsal zone between indications
of dorsolateral lines (a pattern sometimes
seen in other altavelensis); ventral lines
are present but faint. We tentatively as-
sign this specimen to the subspecies en-
riquilloensis, but the final allocation of
the population represented by it must
await the acquisition of more specimens.

Specimens examined. REPUBLICA DOMINICANA:
Provincia de Independencia: Boca de Cachoén, ASFS
V4381-83, ASFS V39707; 2 km E La Descubierta,
ASFS V45007-08; 5 km N Jimanr, ASFS V35480-
502; 10.4 km NE Jimanr, ASFS X9488-501; 5 km SE
La Florida, ASFS V23303-08; 9 km W Duverge,
ASFS V20508; 7.0 km W Duverge, ASFS V4401-06;
6.7 km NW Duverge, ASFS V4343-47; 6 km NW
Duvergé, ASFS V17170-71; La Source, about 5 km
W Duvergée, ASFS V20803-06; just E Duverge,
ASFS V23276-302; outskirts of Mella, ASFS
V30671-89; 8 km N Colonia Mixta, ASFS V35508; 3
km SE Angostura, ASFS V41761-64; 11 km SE

HISPANIOLIAN SPHAERODACTYLUS : Thomas and Schwartz 93

Angostura, ASFS V41299-307; Provincia de Ba-
oruco: 1.4 km WNW Las Clavellinas, ASFS V30377-
85; Jaragua, ASFS X9787; 5.4 km ENE Neiba, ASFS
V228-43; 5.9 km ENE Neiba, ASFS V30334-70; 4
km SW Galvan, ASFS V40753-57; Provincia de
Barahona: 11 km NE Cabral, ASFS V35509-10; 7.5
km E Cabral, ASFS X9623-25, ASFS V219-20, ASFS
V350, ASFS V1450; 5 km S Cabral, MCZ 51820-21;
5 km N Barahona, ASFS V20515-37; Barahona,
ASFS X9451, ASFS X9525; Barahona, southem out-
skirts, ASFS V30988-1011, ASFS V31013; 2 km SE
Barahona, ASFS X9520; 4.8 km S Barahona, ASFS
V14058, ASFS V14060, LDO 7-5344-50, LDO 7-
5352-54, LDO 7-5361-63, LDO 7-5365-66, LDO 7-
5522: 0.6 km NW Palo Alto, 31 m, ASFS V30543-
72; 4.2 km NW Palo Alto, 31 m, ASFS V30530-40,
ASFS V31515; Fondo Negro, ASFS X9708; 1 km
NE Fondo Negro, ASFS V35667-78; El Iguito, 2.6
km NE Fondo Negro, ASFS V30495-522, ASFS
V31429; 3.2 km NE Fondo Negro, ASFS X9678-
81; 1.1 km NW Puerto Alejandro, ASFS V30637-47;
west side, Punta Martin Garcia, ASFS V117-29;
Provincia de Azua: Barreras, ASFS V31176-78; 7.4
km NE Barreras, ASFS V31035-36; 19.8 km SE
Guanito, 275 m, ASFS V31335-37; 15.2 km E Azua,
ASFS X8095-101, ASFS V19352-54; 17 km E Azua,
ASFS V21101; Provincia de San Juan: 7 km NW
Vallejuelo, 793 m, ASFS V303.

Sphaerodactylus altavelensis brevirostra-
tus Shreve
Figures 2C, 3C

Sphaerodactylus brevirostratus brevirostratus
Shreve, 1968, Breviora 280:10.

Type locality. 5 km S Dufort, south of Léogane,
Département de |’Ouest, Haiti.

Holotype. MCZ 63234.

Definition. A subspecies of S. alta-
velensis characterized by well-developed
anterior and posterior cephalic figures in
females; scapular figure present as a
black patch and ocelli in all females and
the majority of males; scapular figure
border present in all individuals but in-
complete in most.

Distribution. Haiti, from the region of
Lascahobas and St. Marc on the north
island, south to the Cul de Sac Plain onto
the north slopes of the Massif de la Selle
(Morne Il’Hopital) in the vicinity of
Pétionville (and apparently as high as
Furcy) and west along the base of the
Tiburon Peninsula at least to between
Petit-Goave and Grand-Goave, and south
to the south coast of the peninsula to the

vicinity of Cayes Jacmel. A population at
Jérémie is assigned to this subspecies; it
averages lower in scale counts but agrees
in coloration. Although collecting on the
Tiburon Peninsula has failed to reveal
any altavelensis between Jacmel and the
Petit-Goave records and those from
Jéremie, we would not be surprised at
the eventual discovery of populations in
this area.

Variation. We have examined 394
specimens and taken counts on 103 of
these. Largest males 25 mm SVL, largest
female 28 mm dorsal scales between
axilla and groin 20 to 29 (24.7); ventral
scales axilla to groin 22 to 32 (26.2); mid-
body scales 38 to 48 (42.9); supralabials
to mid-eye 3/3 (85), 3/4 (2), 4/4 (1); in-
ternasals 0 (3), 1 (83), 2 (7); fourth toe
lamellae 5 to 11 (mode 9 but 10 nearly
equimodal); throat scales keeled; es-
cutcheon 4 to 8 X 11 to 24.

Males are gray to gray-brown or brown
and may or may not be spotted or flecked
with dark brown; heads may be unicolor
in combination with a spotted dorsal
body pattern or may have contrasting
vermiculate patterns (few specimens);
the scapular patch is reduced compared
with that of females and may be nearly
pin-head in size or absent. The anterior
cephalic figure is weakly developed in
most males; the posterior cephalic figure
is very weakly indicated or absent with
the exception of the males having well-
developed basic patterns (Fig. 2C).

Most females have well-developed
anterior cephalic figures and posterior
figures (Fig. 3C) but show considerable
variation in expression. All females have
at least some indication of a scapular
patch and ocelli; the extent of the patch
varies from reduced (about four scales in
width by three to four scales in length at
greatest length) to fully developed. Fully
developed patches may be 12 scales wide
and six scales in length. In the reduced
form the ocelli occur at the lateral edges
of the patch and may not be completely
included in it.

In life, the ground color of S. a. brev-

94 Advances in Herpetology and Evolutionary Biology

irostratus is tan to brown with dark
brown markings. Anterior and posterior
cephalic figures are paler (central por-
tion) than the rest of the ground color.
Scapular ocelli are white or whitish;
venters are gray with brown lines. Males
have yellow or yellowish heads and
throats.

Specimens examined. HAITI. Dépt. de IlArti-
bonite: Pierre Payen, 14.4 km S St. Marc, ASFS
V39457, ASFS V43777; Dépt. de ’Ouest: 3.5 km SW
Lascahobas, 275 m, ASFS V26539, ASFS V26592;
10.1 km SE Montrouis, ASFS V39461-63; Trou For-
ban, USNM 117157-58, USNM 118894; 3.5 km SE
Trou Forban, ASFS X3994-99; Ile-a-Cabrit, ASFS
V40441-43; Sources Puantes, USNM _ 117159-64;
Tabarre, 5.1 km SE Francois Duvalier airport, ASFS
V36220; 20.2 km SE Mirebalais, 366 m, ASFS
V35702; 6.2 km NW Ganthier, ASFS X2088-100,
LDO 7-5799-800, LDO 7-5825-39, LDO 7-5842-43,
LDO 7-5853-54, LDO 7-5801; 5.6 km E Croix des
Bouquets, ASFS X2101; 12.6 km E Croix des
Bouquets, ASDS V36906; 13.1 km E Croix des
Bouquets, ASFS V42563-64, ASFS V8139-42, ASFS
V8298-303; 13.4 km E Croix des Bouquets, ASFS
V40464, ASFS 44852-55; Téte Source, 1.4 km NNE
Thomazeau, ASFS V8177-78; Source Fond Parisi-
en, ASFS V36974-77; Pétionville, ASFS V13688;
Morne Calvaire, 1.6 km SW Pétionville, ASFS
X1302; Furey, AMNH 124158; Diquini, ASFS
2375-79; Mariani, 11.2 km E Gressier, ASFS
V8462; 6.4 km SW Léogane, ASFS V8316; 1.6 km
SE Fauché, ASFS V37270-73; + 6.4 km SE Fauché,
ASFS V37267; 9.9 km W Fauché, ASFS X2051;
beach just W Grand-Goave, ASFS V36569-71, ASFS
V42557-61; 8.8 km W Grand-Goave, ASFS V45218-
346; 4.0 km E Petit-Goave, ASFS V43311-16, ASFS
V43605-711; 7.5 km E Petit-Goave, ASFS V43317,
ASFS V43586-89, ASFS V44073-83; 1.6 km N
Jacmel, ASFS V9798; La Fond, near Jacmel, MCZ
64822; less than 1.6 km W Cayes Jacmel, ASFS
V9704-09; Dépt. du Sud: beach area within 1 km E
Jéremie, ASFS V25217-29; ca. 5 km SE Jérémie,
ASFS V9392; Pistache, SW Jérémie, ASFS V9126-
28.

Sphaerodactylus altavelensis lucioi new
subspecies
Figures 2D, 3D

Holotype. MCZ 156208, adult male, from Terre
Sonnain, 1.6 km N Les Poteux, 132 m, Département
de l’Artibonite, Haiti, one of a series collected 9
July 1978 by native collectors. Original number
ASFS V46397.

Paratypes. ASFS V46398-404, same data as holo-
type; ASFS V40263-65, same locality as holotype, 9
August 1977, native collectors; ASFS V49900-01,

Balladé, 8.8 km S Port-de-Paix, Dépt. du Nord-
Ouest, Haiti, 18-19 July 1979, native collectors;
ASFS V46984-86, Balladé, 8.8 km S Port-de-Paix,
Dépt. du Nord-Ouest, Haiti, 22 July 1978, native
collectors; ASFS V47797-98, 1.9 km W Ennery, 336
m, Dépt. de |’Artibonite, Haiti, 5 August 1978,
native collectors; ASFS V50141-43, 1.9 km W
Ennery, 336 m, Dept. de I’ Artibonite, Haiti, 22 July
1979, native collectors; ASFS V46664-66, Gonaives,
Dépt. de l’Artibonite, Haiti, 12 July 1978, native
collector.

Definition. A subspecies of Sphaero-
dactylus altavelensis characterized by a
prominent anterior cephalic figure in
both sexes, a very obscure to absent pos-
terior cephalic figure in both sexes (Fig.
3D), and scapular figure present as a
black patch and usually two white ocelli
in almost all specimens of both sexes, the
patch without edging.

Distribution. Northeastern Haiti, from
the vicinity of Port-de-Paix in the north to
Gonaives in the south, and inland as far
as the vicinity of Ennery.

Description of holotype. Adult male;
23 mm SVL, tail length 21 mm, tip re-
generated; dorsal scales axilla to groin 29,
ventral scales axilla to groin 27, scales
around midbody 41; fourth toe lamellae
9; escutcheon 5 x 22. Snout moderately
blunt in dorsal aspect, very slightly de-
pressed; snout scales moderately-sized,
somewhat flattened, smooth, not imbri-
cate; 1 internasal; 2 postnasals; supra-
labials to mid-eye 3/3; dorsal head scales
small, elongate, keeled, almost imbricate
to juxtaposed; temporal scales small,
oval, keeled, subimbricate. First infra-
labial broader anteriorly than posteriorly,
roughly rectangular in shape; gular series
between infralabial rami large, juxta-
posed, smooth; central gulars small,
smooth, subimbricate; chest and ventral
scales smooth, with about | lateral row of
throat scales on each side keeled.

Dorsal ground color in life dark brown,
dorsum finely peppered with slightly
darker (grayish brown) scales; anterior
cephalic figure buff, outlined with very
dark brown to black, bilobed and promi-
nent; posterior cephalic figure reduced to
a series of dark flecks or spots; scapular

HISPANIOLIAN SPHAERODACTYLUS - Thomas and Schwartz 95

patch present and prominent, not bor-
dered, black with a pair of included
white ocelli, and constricted medially;
tail vaguely lineate dorsally and without
obvious ocellate pattern; venter creamy
white, with chin and throat rather
densely dotted with dark brown; a series
of about four or five fragmented and ir-
regular dark longitudinal lines on venter.

Variation. Of the 23 specimens, the
largest male is 24 mm SVL, largest fe-
males 24 mm SVL; dorsal scales axilla to
groin 22 to 29 (27.1), ventral scales axilla
to groin 24 to 29 (26.7); midbody scales
37 to 47 (42.5); supralabials to mid-eye
3/3; internasal 1; fourth toe lamellae 8 to
13; throat scales smooth in all but three
specimens, although a few lateral throat
scales in lateral sequence may be keeled,
whereas all other throat scales on the
same specimen are smooth.

The dorsal ground color in all speci-
mens was brown, dark brown, or almost
black. Even darker dorsal body pattern
elements are usually more or less iso-
lated dark scales which give a dotted pat-
tern. These dots may at times be aligned
into a very vague series of dorsal lines.
The anterior cephalic figure is bilobed
and always prominent, whereas the pos-
terior cephalic figure is absent and repre-
sented merely by a series of dark frag-
ments. The black scapular patch is pres-
ent but variable in extent in all speci-
mens but one, although it was recorded
as present in life on that specimen. Two
white ocelli are present and may be in-
cluded in the dark patch or occasionally
lie just peripheral to it. Unregenerated
tails continue the more or less lineate
dorsal pattern, although the dorsal sur-
face of the tail may have merely a random
series of dark scales without any lineate
effect. Sacral stripes are present in some
specimens, absent in others. Throats are
dotted with dark brown and venters
either rather lineate with dark or not.

Females are colored and patterned
dorsally like males; the anterior cephalic
figure is prominent and the scapular
patch is variable in expression, with

ocelli present in all specimens. Juveniles
are patterned like females. Venters are
like those of males except that they seem
slightly more strongly lineate and there is
a clearer indication of ventrolateral longi-
tudinal stripes than in males (Fig. 3D).
Comparisons. Sphaerodactylus a. luc-
ioi differs from the other subspecies of S.
altavelensis in smaller size (although it is
closest to brevirostratus), in having the
black scapular patch and ocelli present in
both sexes (the patch without a clear out-
line edging) and in the expression of the
anterior cephalic figure and absence of
the posterior figure in both sexes. The
other subspecies usually have the throat
scales keeled, whereas S. a. lucioi usu-
ally has them smooth, with at best a few
lateral keeled scales. The mean of dorsal
scales (27.1) is greater than those of en-
riquilloensis and brevirostratus (26.1
and 24.7), less than that of altavelensis
(30.1). The mean of midbody scales is
less (42.5) than that of any other subspe-
cies (42.9 to 51.7), and that of ventral
scales (26.7) is intermediate between that
of brevirostratus (26.2) and altavelensis
and enriquilloensis (27.9 and 28.0).
Remarks. We have no data on the pre-
cise niche which S. a. lucioi occupies.
The general habitats are quite variable.
The type locality is xeric, with cacti and
Agave as typical plants, on a gravelly and
rocky substrate. The specimens from
Gonaives presumably came from the
vicinity of the city itself; Gonaives lies in
a xeric region. Extending inland from
Gonaives is the valley of the Riviere
d’Ennery. The Ennery locality for the
specimens of S. a. lucioi is mesic and at
an elevation of 336 m; this locality is es-
pecially interesting in that there is a
combination of a mesic interior and xeric
exterior (= coastal) faunas there. The
locality at Balladé is extremely mesic, in
fact a coffee-cacao grove. Syntopic con-
geners of S. a. lucioi vary with locality: at
the type-locality they are found with S.
cinereus Wagler sensu Graham and
Schwartz, 1978 (common) and S. aster-
ulus Schwartz and Graham (rare). At

96 Advances in Herpetology and Evolutionary Biology

Ennery they were also taken with S. dif-
ficilis Barbour and S. cinereus in about
equal frequency. Balladé yielded only S.
difficilis in addition to S. a. lucioi. At any
locality, the numbers of specimens of
other species far outnumbered those of S.
a. lucioi. Whether this is due to its scarc-
ity or to its relatively smaller size is un-
known; collecting was presumably non-
selective.

We are pleased to name this sub-
species in honor of John C. Lucio, who
accompanied the junior author in Haiti
for two successive summers (1978-79)
and who contributed greatly to the suc-
cess of both trips.

Sphaerodactylus williamsi new species
Figure 2E

Holotype. MCZ 156209, adult female, from 12.2
km W Ca Soleil, Département de |’ Artibonite, Haiti,
taken by native collector on 14 July 1978. Original
number ASFS V46794.

Definition. An apparently small spe-
cies of Sphaerodactylus (only known
specimen 22 mm SVL), with small scales
(33 keeled dorsal scales and 33 smooth
ventral scales between axilla and groin,
52 scales around midbody), keeled throat
scales; no area of middorsal granules or
granular scales; dorsal body scales with
three or four hair-bearing scale organs,
each with one hair, on the posterior
periphery of the scale, one organ at the
apex; no head or scapular pattern or patch
or ocelli, but dorsum with a series of four
well-defined pale buffy lines alternating
with broader dark brown lines, plus two
pale dorsolateral lines; venter pale,
throat with a few fine dark brown stipples
and very vague indication of ventral
longitudinal lines, most prominent pos-
teriorly; head small and relatively nar-
row.

Distribution. Known only from the
type locality (Fig. 1).

Description of holotype. Gravid fe-
male; 22 mm SVL and 17 mm tail length
(mostly regenerated); dorsal scales

keeled and imbricate, small, axilla to
groin 33, ventral scales axilla to groin 33,
scales around midbody 52, fourth toe
lamellae 10. Snout attenuate, head small
and relatively narrow; snout scales
moderately large, juxtaposed, smooth
anteriorly and keeled posteriorly; 1 in-
ternasal; 2 postnasals; supralabials to
mid-eye 3/3; dorsal head scales small,
keeled, subimbricate; temporal scales
small, juxtaposed, weakly keeled. First
infralabial broader anteriorly than pos-
teriorly, rectangular; gular series be-
tween rami large, smooth, juxtaposed;
central gulars very small, smooth, im-
bricate, grading quickly to much larger
keeled throat scales; chest and ventral
scales smooth.

Dorsal ground color in life dark brown
with four longitudinal pale buffy lines
(Fig. 2E); head and neck scales without
cephalic or nuchal pattern but with iso-
lated scattered buffy scales; a pair of
slightly more prominent pale buffy lines
on the sides; venter creamy white, with a
few scattered dark brown dots on the
throat and vague indications on the pos-
terior portion of the belly of dark longi-
tudinal lines; no scapular patch or ocelli.

Comparisons. Sphaerodactylus wil-
liamsi seems most closely related to S.
altavelensis; the basic dorsal color (dark
brown) is similar in S. williamsi and S. a.
lucioi. But direct comparison of the
single williamsi with many comparably
sized female altavelensis of any sub-
species shows the much narrower and
more acuminate snout of the former.
Although some _ S. altavelensis are
vaguely lineate, none shows the distinct
lineation of S. williamsi. The total ab-
sence of cephalic patterns (except for the
scattered pale buffy scales) and of a
scapular patch and ocelli are also distinc-
tive, especially in contrast to local S. a.
lucioi which, in both sexes, have promi-
nent anterior cephalic patterns and
scapular patches and ocelli in both sexes.
The high counts of the holotype of S.
williamsi (dorsals 33, ventrals 33, mid-
body 52) are all greater than the highest

HISPANIOLIAN SPHAERODACTYLUS - Thomas and Schwartz 97

counts in local S. a. lucioi (29, 29, 47); the
keeled throat scales of S. williamsi like-
wise are distinctive. The scale counts fall
within (dorsals, midbody) or just above
(ventrals) those of S. altavelensis; but it is
the far-removed subspecies S. a. alta-
velensis, not the closer brevirostratus,
enriquilloensis, or lucioi, whose upper
parameters exceed or almost equal the
counts of S. williamsi. The smaller scales
of S. williamsi are obvious upon inspec-
tion and comparison with similarly sized
S. a. lucioi, with which S. williamsi
might be confused.

Remarks. The type locality of S.
williamsi is a coastal oasis; the general
region is called Lapierre. The surround-
ing area is xeric and barren (cacti, Agave),
in strong contrast to the luxuriant growth
in the oasis which is fed by a small
source. Associated congeners here are S.
asterulus and a second, as yet unnamed,
species. Both these congeners are pre-
sumably more abundant in the adjacent
xeric areas than within the oasis; this is
certainly true for S. asterulus. Haitians
living within the oasis are relatively in-
dependent and are not easily encouraged
to collect; this may account for the lone
specimen of S. williamsi (and perhaps
even for the lack of S. altavelensis, which
occurs only 19 km to the east at the type-
locality of lucioi and 10 km to the south
east at Gonaives, across the Baie de
Gonaives). On the other hand, it is pos-
sible that S. williamsi is here at the very
periphery of its range, which may en-
compass the herpetologically largely
unknown Presqu’ile du Nord-Ouest.
From that peninsula proper, S. shrevei
Lazell and S. elegans MacLeay are
known (both from Mole St. Nicholas, far

to the northwest and on the northem side

of the Massif du Nord-Ouest). We call at-
tention to this phenomenon once more:
the line of demarcation between the
peninsular fauna and the “mainland”
fauna appears to be sharp. In the matter
of Sphaerodactylus, S. cinereus and S.
altavelensis are moderately common on
the “mainland,” but appear not to occur

on the peninsula. Sphaerodactylus
asterulus is peninsular but a southern
cognate of the related S. shrevei in the
north. Sphaerodactylus difficilis occurs
only along the northern slopes of the
Massif du Nord-Ouest, as far west as
Bombardopolis and also on the adjacent
“mainland.” The ecological relationships
among this suite of congeners are indeed
puzzling, and their geographical rela-
tionships are just beginning to be clari-
fied.

We take very great pleasure in naming
this new species in honor of Ernest E.
Williams, who, over the years, has been a
constant source of friendship, advice,
cooperation, and loans of pertinent
materials for our research in the Antilles.

ACKNOWLEDGMENTS

Most specimens examined for the
present paper are in the Albert Schwartz
Field Series (ASFS); many of them were
collected under the sponsorship of Na-
tional Science Foundation Grants G-7977
and B-02603 to the junior author. We
thank the following field workers, both
friends and students, who have helped
make such a mass of material available to
us: Patricia A. Adams, Robert K. Bobilin,
Jeff R. Buffett, David A. Daniels, James
R. Dennis, Danny C. Fowler, Eugene D.
Graham, Jr., Ronald F. Klinikowski, Mark
D. Lavrich, David C. Leber, James K.
Lewis, John C. Lucio, Gary C. Mosely,
James W. Norton, Lewis D. Ober, S.
Craig Rhodes, James A. Rodgers, Jr.,
Bruce R. Sheplan, William W. Sommer,
Michael H. Strahm, Oscar Vargas, and C.
Rhea Warren. Specimens, in addition to
those in the ASFS, have been examined
in the following collections, to whose
curators we are grateful for loans: Ameri-
can Museum of Natural History
(AMNH—Richard G. Zweifel and
George W. Foley); Museum of Compara-
tive Zoology (MCZ—Emest E. Wil-
liams); National Museum of Natural
History (USNM—George R. Zug). Lewis

98 Advances in Herpetology and Evolutionary Biology

D. Ober (LDO) also allowed us to ex-
amine material in his personal collection.
Finally, we are grateful to Fred G.
Thompson of the Florida State Museum
for allowing the junior author to retain
one specimen of S. a. altavelensis. The
illustrations are the work of David C.
Leber and Alvis Gineika, to whom we are
grateful, and the junior author.

LITERATURE CITED

GRAHAM, E. D., JR., AND A. SCHWARTZ. 1978. Status
of the name Sphaerodactylus cinereus Wagler
and variation in “Sphaerodactylus stejnegeri”
Cochran. Florida Sci., (4)41: 243-521.

NOBLE, G. K., AND W. G. HassLer. 1933 Two new
species of frogs, five new species and a new
race of lizards from the Dominican Republic.
Amer. Mus. Novitates, 652: 1-17.

SCHWARTZ, A. 1967. The Leiocephalus (Lacertilia,
Iguanidae) of Hispaniola. II. The Leiocephalus
personatus complex. Tulane Stud. Zool., 14(1):
1-53.

____. 1977. The geckoes (Sauria, Gekkonidae) of the
genus Sphaerodactylus of the Dominican
Peninsula de Barahona, Hispaniola. Proc. Biol.
Soc. Washington, (2)90:243-254.

____. 1979. The herpetofauna of Ile a Cabrit, Haiti,
with the description of two new subspecies.
Herpetologica (3)35: 248-255.

SCHWARTZ, A., AND R. THOMAS. 1975. A check-list of
West Indian amphibians and reptiles. Carne-
gie Mus. Nat. Hist. Spec. Publ., 1: 216 pp.

AND __. 1977. Two new species of
Sphaerodactylus (Reptilia, Lacertilia, Gek-
konidae) from Hispaniola. J. Herpetol. (1)11:
61-68.

SHREVE, B. 1968. The notatus group of Sphaero-
dactylus (Sauria, Gekkonidae) in Hispaniola.
Breviora Mus. Comp. Zool., 280: 1-28.

Tuomas, R., AND A. SCHWARTZ. 1977. Three new
species of Sphaerodactylus (Sauria: Gek-
konidae) from Hispaniola. Ann. Carnegie Mus.
Nat. Hist., (4)46: 33-43.

Biogeography of the Herpetofauna of the
British Virgin Islands, with Description of a
New Anole (Sauria: Iguanidae)

JAMES D. LAZELL, JR."

ABSTRACT. Specimens and field data were collected
for 44 of 46 named islands in this group. Published
lists provide additional data for 24 of the islands
visited and both that were not. The islands consis-
tently have more species than areas, elevations, and
distances from propagule sources might lead one
to predict. There seems to be a basic, minimal
number of three species per island, regardless of
physical parameters. These are generally an Anolis,
a member of the family Gekkonidae, and some
other reptile. Even tiny rocks and overwashed bars
usually support at least an Anolis. One such, Carrot
Rock, 1.2 ha and 26 m high, supports a remarkable
new species of the alpha section of Anolis. It is a
member of the cristatellus group sensu Williams
attaining very large size, combining two scale
characters: 1) dorsals small; 2) digital lamellae
count high; and three color characters: 1) chin tri-
colored; 2) latero-dorsal trunk and neck boldly pat-
terned; 3) a dark bar or blotch on each side of the
sacrum. Islands, species, areas, heights, and dis-
tances are tabulated and discussed.

INTRODUCTION

The British Virgin Islands constitute
the fragmented northeastem extremity of
the Greater Antilles. For the most part
they are old, folded, and buckled conti-
nental strata of grano-diorites, basalt,
conglomerates, and slates and shales

(Martin-Kaye, 1959). The nether isle of -

Anegada and some strata scattered on
some of the other cays are oceanic lime-
stones deposited during Pleistocene in-

1The Conservation Agency, 8 Swinburne Street,
Conanicut Island, Rhode Island 02835, U.S.A.

terglacials. These same interglacial seas,
standing dozens of meters higher than
the sea today, loosened, rolled, and
tumbled the rocks in many areas, leaving
piles of huge boulders—some as big as a
house—signalled today by names like
Fallen and Broken Jerusalem.

As sea level fell following the Sanga-
mon Interglacial, some hundred thou-
sand years ago, more and more land was
exposed. At the Wurm glacial maximum,
the entire Puerto Rico Bank was dry land
all the way to Anegada. In the last ten
thousand years the rising sea has refrag-
mented the land into fifty or more
islands, rocks, and cays. Since the Wurm,
sea level has probably never stood higher
than it does today; it continues to rise
(Morris et al., 1977).

The British Virgin Islands are an arti-
ficial entity, separated from their
American neighbors by a line passing
between Pelican (British) and Flanagan
(U.S.) Islands on the south, curving west
through The Narrows between the
Thatches (British) and St. John (U:S.),
and running westward into the North
Atlantic south of the Tobagos (British).
My study of these islands involved 40
days of field work in March and April,
1980. Not counting rocks with no more
than herb-stage vegetation, I tallied 46
islands suitable for reptiles and/or am-
phibians (see Fig. 1). I visited all of these
except Jost Van Dyke and Mosquito, both
large islands, often and easily visited,
and both with good species lists. It is un-

100

BRITISH VIRGIN ISLANDS

TOBAGO:

JOST VAN DYKE:
Great

Little

Little
Cr OREEN
iss Se SANDY SPI

TORTOLA

Advances in Herpetology and Evolutionary Biology

/0km
aah mange CLELP HELE
CANN pee _NECKER
EUSTATIA
aga cockroach: \| ee Te)
Le SABA ROCK
North %
D065.
South—s3q_
SCRUB & S5-Greot
—~west

QS MARINA
IN—BEL LAMY

—) aN
SANDY JERUSALEM.
Dee C—Fallen
GREAT. THATCH, aN e sued
\ o
i DEAD MANS & 2S rsiiie aoe
LITTLE NS AEN ay ee ae
FRENCHMANS Ss COORER
PELICAN |— Sagi
ee c
OLE LSS | Ncarror rock
X 4
Hf NORMAN
Figure 1. The British Virgin Islands. Four nameless cays are numbered. (Modified from Marler, 1973.)

likely I could have expanded either list,
except by collecting the gecko, Hemi-
dactylus mabouia, which I presume is
present, if unrecorded, on both.

My count of 46 is too low. An unnamed
but well-vegetated island of more than a
hectare lies off Key Point, south of the
western peninsula of Peter Island. Most
people assume it to be part of that island,
but I am assured (Marler, personal com-
munication) that it is disjunct. Two rocks
south of Dead Man’s Chest appear to
have shrub-stage vegetation as viewed
from the heights of Peter Island. A rock
separated from the southwest tip of
Norman Island requires investigation. A
separated, southern chunk of the Carrot
Rock scarp should be climbed. There
may be more.

Within historic times (though no one

living seems sure just when) “Pelican
Cay” (not Pelican Island) was amalga-
mated with Little Jost Van Dyke by a
(now) well-vegetated tombolo of cobbles
and sediment; it lost its status as an is-
land. In temporal similarity (and uncer-
tainty), however, Green Cay was frag-
mented in two; its erstwhile southern
terminus is today fully disjunct and
called Sandy Spit. That is one of the
smaller land areas on earth with two
species of native reptiles. Man’s activi-
ties have connected Wickhams Cay and
Nannie Cay to the Tortola mainland with
solid fill; these are not given separate
consideration. Frenchmans Cay was
separated from Tortola historically by
mangrove swamp. Fill was built up to
make a causeway from Tortola almost to
the cay, and a channel ca. 2 m wide was

dredged (and bridged); Frenchmans Cay
has been considered here as a separate
island.

An island measuring ca. 50 x 100 m,
showing the symbol for scrub vegetation,
is a prominent feature on Directorate of
Overseas Surveys (1959) map. of
Anegada, labelled “Little Anegada.” It is
imaginary. The real Little Anegada is a
tiny (ca. 3 X 5 m), wave-washed block of
limestone, ca. 30 cm above water at mean
low tide. It is vegetated only with the
maritime plants Susuvium portulacas-
trum and Rhizophora mangle (one). “In
before time,” states Mr. Lance Vanter-
pool (personal communication), “it was
no bigger, but it was more prominent
because here were not all these mang
groves. The more recent map, Director-
ate of Overseas Surveys (1977) quite cor-
rectly shows the region as mangrove con-
tinuous with Anegada itself. The place
where the imaginary “Little Anegada”
was located in 1959, is, quite correctly,
open water.

Much herpetological work had been
done in these islands before I arrived.
Lists were available, prepared by
Maclean et al. (1977) from Schwartz and
Thomas (1975) and Philibosian and
Yntema (1976, 1977, 1978). Heatwole
(1976) had examined Anolis cristatellus
wileyae from eight additional islands.
Nevertheless, there were no data for 18
islands. I obtained 13 new records for 10
islands with extant lists, and 44 new rec-
ords altogether. Nevertheless, I must
have failed to find many populations, and
there is undoubted necessity for far more
work.

I missed populations for three major
reasons: 1) I had to average over a cay per

day in my field inventory; this is obvious-’

ly far too little time for even the tiniest
land areas. 2) I was on the ground at the
height of the dry season in an unusually
arid year; many species were cryptic, es-
pecially by midday, and some may have
been in brumation. 3) I am me. I have
poor vision and a gimp leg, and some say
I move more slowly than I did a quarter-

BRITISH VIRGIN ISLANDS : Lazell 101

of-a-century ago—and climb less high.
And, I do not like to hunt frogs too small
to eat (all in these islands are that). How-
ever, from a theoretical point of view, the
fact that I missed lots of populations will,
I hope to demonstrate, prove trivial.

THE FAUNA

Maclean et al. (1977) list 23 species of
reptiles and amphibians occurring in the
British Virgin Islands. Of these, at least
one, Anolis cuvieri, is not known from
specimens; Schwartz and Thomas (1975)
note that its existence on Tortola has
remained “unverified for over a cen-
tury.” Perhaps it should be deleted, but I
saw fine habitat for this species in the big
ghuts draining Mount Sage, on Tortola,
and on the rugged eastern slopes of ad-
jacent Beef Island. Anolis cuvieri could
occur here.

Another species, the tortoise Geo-
chelone carbonaria, is a suspected intro-
duction. It is today unknown from Peter
Island, whence listed, and rarely seen on
Tortola. It is reported to occur on Virgin
Gorda. Tortoises of one sort or another
were widespread in the pre-Columbian
Antilles (Auffenberg, 1967, 1974); no one
has done the sort of study of geographic
variation in G. carbonaria needed to
demonstrate its true status, spurious or
otherwise. I leave it in.

I have previously (Lazell, 1973) dis-
cussed the notion that Iguana iguana was
introduced to these islands, the northern
limit of its range. The I. iguana seen and
collected during the present investiga-
tion support my earlier view: they are
native.

On three rock cays, where I believe
reptiles probably occur, I failed to obtain
specimens: Carval, North Cockroach, and
the Indians. All support sea grape
(Coccoloba uvifera) clumps and other
vegetation. I believe Anolis (probably A.
cristatellus wileyae) live on them, but
weather conditions prevented me finding
the animals on North Cockroach. I was

102

unable to climb to the best vegetation on
Carval and the Indians.

Mirecki (1977) led a group of four or-
nithologists and a botanist to the islands
in 1976. I visited all the islands they
failed to, and 11 more they did not con-
sider. However, Dr. J. D. H. Smith (per-
sonal communication) even went onto
rocks with a few sedges where he reports
seeing no lizards.

The results of all this are presented in
Table 1, where the islands are ranked in
order according to area. Despite the
numerous new island records, the fauna
presented no surprises whatsoever until
4 April, 1980, when I climbed the sheer
face on the northeast side, tunnelled
through the Coccoloba, and entered the
strange world of Carrot Rock. Apart from
the previous three weeks, I had been out
of Anolis work in the Antilles for over a
decade, but I there saw my next new one:

Anolis ernestwilliamsi* sp. nov.
Figures 2, 3

Holotype. MCZ 158395, an adult male, J. Lazell
coll. 4 April, 1980.

Type locality. Carrot Rock, south of Peter Island,
British Virgin Islands, 18°19'45’’ N, 64°34'18’' W,
Caribbean Sea.

Diagnosis. An alpha section Anolis of
the cristatellus group sensu Williams
(1976:17), attaining very large size (to at
least 82 mm SVL), combining the follow-
ing two scale characters: 1) scales small,
34 to 45 (av. 40) middorsals contained in
the standard distance at midbody; 2) digi-
tal pads large, males with 24 to 27 lamel-
lae, females with 20 to 22 lamellae, under
the second and third phalanges of the
fourth toe; and combining the following
three color characters: 1) chin tricolored,
with a bold, reticulate or barred pattern
of near-black to grey on lighter blue-grey,

1This name commemorates the man who, in 1958,
agreed to pay my way back to Dominica if I could
prove that island was occupied by one, not four,
species of Anolis.

Advances in Herpetology and Evolutionary Biology

with egg-shell white to creamy blotches.
In the male the dark chin elements ex-
tend onto the anterior throat fan; in the
female they extend posterior to at least
the level of the eye. 2) Latero-dorsal
trunk and neck boldly patterned with
near-white to yellow spots and marbl-
ings, coalescing to form stripes, and iso-
lating spots of darker pigment ventro-
laterally. 3) A dark bar or blotch on each
side of the sacrum, set off posteriorly by a
pale, ashy or tan border. In the females
the middorsal stripe is irregularly edged,
often fragmented, and in poor contrast to
the bold latero-dorsal pattern.

Description of the type. MCZ 158395
is an adult male 81 mm SVL, with a
standard distance 13.8 mm. There are 42
middorsals, 33 ventrals, and 53 dorsals
(eight rows lateral to the middorsal line)
contained in the standard distance at
midbody. There are 26 lamellae under
phalanges II and III of the fourth toe.
There is an enlarged, double row of
bluntly tectiform middorsals. The ven-
trals are smooth posteriorly and become
tectiform to keeled anteriorly (chest re-
gion). There are seven loreal rows (left
side); seven scales border the _ inter-
parietal; there are seven plates across the
snout between second canthals; there are
22 mental rows between third infra-
labials.

In life the type was somber, dark grey-
brown with olive tints on the limbs, tail,
and lateral body. Near-white to cream-
yellow spots coalesced to form streaks
and marbling which became especially
bold ventrolaterally. Slaty to sooty
blotches and bars were scattered over the
dorsal surfaces and were boldly set off on
the ashy venter. There were yellow tones
ventrally on the limbs and abdomen. The
throat fan was deep crimson with a dark
green center and scales that could change
from white to grey. The anterior 30° of
the extended throat fan was boldly reticu-
lated with slate- and blue-grey with
cream white. This tricolored pattern
covers the entire chin region back to the
jowls.

103

BRITISH VIRGIN ISLANDS : Lazell

‘ZOveSl ZOW ‘odAyesed ayews) ynpe ue ‘anoge ‘seees! ZOW ‘edAy eu “spue|s] UIB4IA YsHg “YOY JOUED WO. “AOU ‘ds ISWeI/IMJSOUIA SI]OUY

Rate Lk,
%, teNV ga sise =),4,

+>)

a

“MO|aq
"g aunbi-

104

Advances in Herpetology and Evolutionary Biology

TABLE 1. HERPETOFAUNA OF THE BRITISH VIRGIN ISLANDS AS KNOWN IN APRIL 1980.

1a1and s1jouy *

uoidouayzvd snjhzopposavyds *

inwa, ofng

avuD1yI0I snjhzapposayjnayy *
snpnooidos snphzoppvoay f, *
ISUDYJINJSIULI SIJOUY *
sisuapijup snjphyovpo.sayynayy *
DIADUOGIDI BUO]AYIOA) *
SISUGUOW sazDAIIAT *
IZLUDMNYIS snjhzovposaynayy °

snnsixa s1ydory °

siuqv}iqip snjhyovpojdaT
Dipsysauaf Duapgs1ydwuy

DUDNS. DUDNS]

sisuansoziod srydory *
snjjayojnd sijouy °

pinoqvwu snjfijovpiwarzy *

tauvo]s pvinqvyw

SN]NID1YS SIJOUY *
jnsxa Dalawy °
sidajo1sovu snjhzopposanyds °*

$N]]2JDISILI SIOUY *

6
8
ZL
©)
Sg
4
€
G
I

CG

sinsuid puvnzy] -

GL
TI
‘OL
ipipyoar. sdojyahy °

setoads jo 1loaquinny

Ul -UONBAI[ YH

ey :eoly

Island or Cay

a.
O
O

@

@

@® oO

O O

@ n

@® oO

Oo © ©

@o oO

@@o oO @ O
@0oo000e@eo0oo0oo0d

@@oOo Ox xK XK O
OOOOO Oo © O @
@0000 0C@8 OO xO
(ox Nex X Nex 3 x Mae x Baa i )
S@@0O00e0eOOKOOXK Xe
(ae eek Xa SO Been

MAHODHODDOHMNNONAK+
a44 o
TNQMOEYE STO TAI
AgGtADLYDODNMNODOoOtIANH
Ya) HORN RAN Raa ae aS
HASDOSCANKKKEANMNMNKKAS
TESPTAEQSHOVWASCDOKrES
spice) eal Ee) 3 (63) ES) XI) PI =
aN
co)
oO fe)
wd ©
ae) 2) oe
@ a ®
ao) c E S c oO Ay
Se 2S Cab ae & = =
CHEF be SEE OR OOS ar)
BEEZ Ses ssereess
= chet 1S) i]
Bes Sa mouvZovuvavalvng
ANMNHNOrFORDOHANDNtINO
Sessa

105

- Lazell

BRITISH VIRGIN ISLANDS

“Q0UdLINDDO a[qeuoysenb =d

‘pepe [Oo JOU jNq UI Aq uv9s = §

‘out Aq payoe][oo = x

eee oe ee ee eee

‘oul Aq pauLiyyuos pue (9761) afomyeayy Aq past] aphapin +9 srjouy =
“Oul Aq Uu9v—dSs JOU nq “(LL6B1) 1]? nie) “uBapoRr Aq pest] sotoeds = oO
‘our Aq pouLzuos pur (17167) 77 2a ‘uRafoRy Aq payst] setoads = @
x 1 6 OF TO ywdg Apues oF
oI 10 YyoRoryo0D) YON GF
XO) Ga SiG Woy eqes FP
i 6 suvIPU] CP
“x ft tw FO yoRoryooD YyNOS Zp
x & 6 OF LO Aer Aureyjaq TF
H I ¥ 80 B0q [P9S sey OF
9¢ 80 TRAreD 6E
xX IT 86 60 woyesniof usx01g gE
Ok SG Sr OT Aed vuLeyy Le
IP Ge Sl yoy WOUeD 9E
Ha 1) eee 61 BOC [B28 ISOM GE
xX xX © 6¢ 8 ueoled Pe
@©@@eFfr @% gE Aer Apues ¢¢
@®eck se wg ABD U22ID GE
@®eer zs 9g2L yoy punoy TE
XX. X. AUCe cS eyeIsny OF
@er oF ZB wafesniof UsT[ey 6G
x Se 6reenel 80d 3S9M 8G
@e x esc 99 FI ysoyD suey peaqd LZ
S X H cM ce. cr B0q 281095) 94
e x He€ T€ 9 aouewley) a[HIT GZ
S\@) Denes 9Sa) V1 yng FS
Xe EX Se, 166. 6S oseqoy, spuy €%
SSF OO 126 yoreyL eAIT GS
x XG OX BS IS FS suvulyouely 1Z
@eees cv og IOYION 0G
Me XU KsiGee GREECE 80q 3215 61
oO O O § Gs Ms oyNbsoW ST
XooN Hiss 106 69 ayAq ueA SOf BPHIT LT

106

The sacro-iliac region was pigmented
on each side in a rough triangle of sooty
color, set off posteriorly by ashy-white.

Variation. Ten paratypes, six males
and four females, were collected by me
and George Marler on Carrot Rock, 18
April, 1980. The males are MCZ 158396-
8 and 158400-1; the females 158402-4. A
male, MCZ 158399, and a female, MCZ
158405, have been deposited in the Brit-
ish Museum (BMNH 1980. 1-.2).

The largest adult male, MCZ 158396,
was 82 mm SVL, fresh. The smallest,
judged adult in the field on the basis of
territorial defense, courtship, and dis-
play, was MCZ 158399: 71 mm SVL. Two
juvenile males, MCZ 158400-1, measure
63 and 59 mm SVL, respectively. Fe-
males measure 52 to 60 mm SVL; MCZ
158402 is the largest. Two 52 mm females
are apparently immature. One, MCZ
158404, shows some beginnings of
ovarian activity on both sides, with larger
yolks present on the right, but seems

Figure 3. Head shape in two Anolis. A. A. ernestwil-
liamsi sp. nov., type-specimen, MCZ 158395, from
Carrot Rock. B. A. cristatellus wileyae, type-specimen,
MCZ 34792, from Culebra.

Advances in Herpetology and Evolutionary Biology

never to have laid an egg. The other,
MCZ 158405, is definitely immature. A
57 mm female, MCZ 158403, has a well-
developed ovarian egg on the left and an
apparently recently spent oviduct on the
right; she was mature.

The standard distance varied from 17.0
to 18.3 (av. 17.4) percent of SVL in males,
and from 17.2 to 19.0 (av. 18.0) in fe-
males. The difference is not significant
with the small sample sizes available.
Except for lamellae and ventrals, dis-
cussed below, meristics were combined
for both sexes.

Dorsals counted in the standard dis-
tance at midbody eight rows lateral to the
midline varied from 45 to 53 (av. 49). Mid-
dorsals in the same distance varied from
34 to 45 (av. 40). There were 6 to 8 (av.
7) loreal rows, 5 to 7 (av. 6) postrostrals,
20 to 28 (av. 25) scales between third in-
fralabials, 8 to 11 (av. 10) suboculars, 5 to
8 (av. 7) scales across the snout between
second canthals, and 12 to 18 (av. 15)
circumparietals. I do not know how
Heatwole (1976) quantified forearm
scales, but I can see no obvious differ-
ence between the new species and A.
cristatellus in this regard, so have omit-
ted the character.

Sexual dimorphism in lamellae count
is striking. Males have 24 to 27 (av. 26),
females 20 to 22 (av. 21). There is also
strong dimorphism in ventral scale size,
quantified as midventrals in the standard
distance at midbody. Males have 28 to 34
(av. 31), females 23 to 27 (av. 25). All
specimens are plotted for these counts in
Figure 4.

There is variation in the placement of
the spots, streaks, mottlings, and marbl-
ings on the dorsal and ventro-lateral sur-
faces. However, these are always bold
and prominent. Color change is from
darker to lighter, and often enhances the
markings in the disturbed extreme.
There is real variation from a browner
extreme (MCZ 158396) to a greyer ex-
treme (MCZ 158399). The light pattern
elements may approach white (MCZ
158400) or be quite bright yellow (MCZ

Lamellae

15

SEanEEEEEEEE ELEnDELEEEPRaRUDDLaRRRL aE EERE cee
20 25 30 35 40
Ventrals in St. D.

Figure 4. Sexual dimorphism in Anolis ernestwil-
liamsi sp. nov. from Carrot Rock. @ = type-specimen.
See text.

158397). In both males and females the
throat fan is deep crimson with a green
center and white to grey scales. The
female fan is quite small. The large male
fan is variably invaded by the tricolor
chin pattern elements anteriorly. In MCZ
158397, only about 20° of the extended
fan is grey, blue, and cream. In MCZ
158396 nearly 40° of the fan is so pat-
termed.

The sooty sacral blotch is prominent on
all specimens, varying slightly in size
and considerably in shape. The irregular,
ashy middorsal stripe fades with age. The
two smallest males retain it. Probably
females retain portions of it throughout
life, but in the large female (MCZ
158402) it was fading to obsoletion.

Comparisons. Heatwole (1976) pro-
vides an exhaustive account of geo-
graphic variation in Anolis of the crista-
tellus group on the Puerto Rico Bank and

its satellites. He did not separate the -

sexes even with respect to lamellae
count, and I do not find sexual di-
morphism as pronounced in A. crista-
tellus—even A. c. wileyae—as in ernest-
williamsi. I counted 30 female wileyae
and got 16 to 19 lamellae (av. 17): no
overlap with female ernestwilliamsi. In
males overlap occurs only in the Ramos,

BRITISH VIRGIN ISLANDS - Lazell 107

Isleta Marina, and Villa del Mar popula-
tions where there may be 25, 24, or 25
lamellae, respectively. Two of these
Puerto Rican coastal cays, Ramos and
Isleta Marina, are populated by A. c.
cristatellus < wileyae intergrades which
cannot look much like A. ernestwilliamsi.
At both Isleta Marina and Villa del Mar
the scales are much larger than in A.
ernestwilliamsi.

Anolis cristatellus has a shorter snout,
on the average, than does A. ernest-
williamsi; in males it is 15.3 to 17.8%
SVL (av. 17.0) and in females 16.4 to 18.2
(av. 17.2). Once again, the difference
between sexes is insignificant, but the
difference between species is significant
at the 95% level of confidence. This dif-
ference does bias counts higher for A.
ernestwilliamsi whenever the standard
distance is used, but the scale size differ-
ence is real nonetheless, as can be seen
from comparing like-sized individuals
(e.g., MCZ Z-08348: A. c. wileyae; MCZ
158399: ernestwilliamsi).

Heatwole (1976:10) comments on the
cline in scale size seen in A. c. wileyae:
as one proceeds eastward from Puerto
Rico through the Virgins, the anoles have
lower counts—therefore larger scales. A.
ernestwilliamsi, residing about 76% of
the way along that cline, stands in bold
contrast to it, with the smallest scales
seen in any member of the complex.

I am not enamored of size as a taxo-
nomic character in reptiles. Nevertheless
I am compelled to admit the importance
of size in the relationships of A. ernest-
williamsi. I measured 135 females of A. c.
wileyae, and the largest I can find are 50
mm SVL (MCZ Z-08400, George Dog;
MCZ 60635, Palominos, E of Puerto
Rico; and MCZ 127725, Seal Dog). Some
females of A. c. wileyae are producing
eggs at 32 mm SVL: MCZ 35734,
Vieques. Females of A. ernestwilliamsi
from Carrot Rock may still be immature
at 52 mm SVL (see above). Thus there is
no evidence of overlap of sizes of mature
females.

There is some overlap in sizes of males

108

that seem mature. Thus, large males of A.
c. wileyae, MCZ Z-08348, from Broken
Jerusalem, and MCZ 128484, from
Culebra, are both 72 mm SVL, just larger
than MCZ 158399, the smallest mature
ernestwilliamsi. One male wileyae, MCZ
Z-08429, from South Cockroach is a giant
for his kind, measuring 75 mm SVL; he
has 31 middorsals and 23 lamellae.
MCZ Z-08429 has the shortest snout
measured, 15.3% SVL, and was nearly
patternless brown in life. None of the
large wileyae approaches the color
characters of ernestwilliamsi; none has
more than 33 middorsals in the standard
distance or 23 lamellae.

The three color characteristics which
taken together are diagnostic of ernest-
williamsi can all be seen singly or weakly
developed in occasional specimens of A.
c. wileyae. MCZ 138543 (Estate Tutu, St.
Thomas) combines a bold chin pattern
with sacral marking like that of ernest-
williamsi; however, it lacks the dorsal
and latero-ventral pattern. Conversely,
MCZ Z-08409 (Eustatia) has fine trunk
pattern but lacks the tricolored chin and
sacral blotch. An adult male from Beef
Island, MCZ Z-08436, was less boldly
marked but a close approach to the color
characters of ernestwilliamsi in life. The
chin pattern was bicolor—dark grey and
pale blue-grey—however, and there are
only 33 middorsals and 22 lamellae. The
closest approach to ernestwilliamsi I can
find is MCZ 35952, from Culebra. This
specimen is faded and may not have had
a tricolor chin. Nevertheless, the pattern
elements that remain closely resemble
those of ernestwilliamsi in all three
respects. The specimen just overlaps
ernestwilliamsi in middorsal count with
34, but has only 23 lamellae. It is a seem-
ingly adult male, but very small: 48 mm
SVir

There is a strong distinction in head
shape of A. ernestwilliamsi relative to all
A. cristatellus, but difficult to quantify.
Basically the head of the Carrot Rock
species is longer, lower in profile, the
rostral bulge is longer, lower, and more
gently tapered, and the angle the rostral

Advances in Herpetology and Evolutionary Biology

makes with the mouth is oblique, not
rectilinear. These differences are shown
in Figure 3.

In summary, Anolis ernestwilliamsi is
closely allied to Anolis cristatellus and
shares its presacral vertebral count of 23,
as do monensis, desechensis, and_scrip-
tus. A. ernestwilliamsi differs from all of
these, and from Puerto Rican A. c. cris-
tatellus in throat fan color, just as does A.
c. wileyae. From this last form, its closest
neighbor, A. ernestwilliamsi is distinct in
combination of color characters, small
scales, high lamellae counts, large size,
and head shape. For those of us who like
quick, quantitative ways to separate spe-
cies, A. ernestwilliamsi may be separated
from A. cristatellus by simply dividing
the lamellae count into one hundred
minus the middorsal count. In Figure 5
I have plotted this against snout-vent
length to graphically portray the distinc-
tion.

Habitat and behavior. Carrot Rock is
steep-to and cliffed for virtually all of its
perimeter. The vegetated top of the ca.
1.2 ha islet is a canted plateau rising from
ca. 6 m above sea level on the windward
side to 27.6 m on the lee. The surface is a
deceptively uneven boulder field.

The deception is perpetrated by the
vegetation, which covers the surface
rather evenly. The dominants are the
polygonaceous sea grape, Coccoloba
uvifera, and leguminous vines. Main

‘stems of these plants may reach 30 cm and

5 cm, respectively. The plants grow pros-
trate and sprawling over high points on
the boulder substrate and canopy pockets
or low areas rather like caclin bush
(Clusia mangle) does in the high Lesser
Antilles (Lazell, 1972). There are small
areas with soil in which grow grasses and
sedges, especially near the crest of the
cay.

The anoles live primarily on the rocks
(to a lesser extent on the vegetation, es-
pecially sea grape trunks) under the
canopy. This is a shady, dimly lit zone,
cooler and somewhat protected from the
full force of the tradewinds.

Anolis ernestwilliamsi is quite com-

+ $+
Gas
tTr+e+
een oy
3.5 O
oO
@)
Oo
IO™,
Sex
a ()
SS oh)
S 3.0 @ ? %, © (oue)
| @ fe) °
8 age
é
2.5
&
&
é
2.0 é

Figure 5. A graphic depiction of the morphometric
distinction between Anolis ernestwilliamsi sp. nov.
(solid symbols), and A. cristatellus wileyae (males, O ;
females +). M = middorsals counted in the standard
distance. L = subdigital lamellae.

mon in its habitat. Sitting quietly in a
pocket below the vegetation, I could
count two to four in ca. 100 m?. I estimate
about two-thirds of the island provides
good habitat, so a population of two to
three thousand individuals is not un-
reasonable.

My observations indicate a balanced
sex ratio. Both sexes bob and fan in the
manner of the widespread A. c. wileyae.

Apparently sexual maturity is attained at.

sizes larger than 63 mm SVL in males and
52 mm SVL in females. The largest fe-
male, MCZ 08460, 60 mm SVL, has a
large egg in the left oviduct and a smaller
one in the right. This implies the alter-
nating oviduct, single egg laying strategy
common in Antillean anoles (Lazell,
1972). The 57 mm female (MCZ 158403)
was yolking up a large egg on the right

BRITISH VIRGIN ISLANDS - Lazell 109

and seemed to have just laid an egg from
the left at time of capture.

Small islands in the British Virgins
(like those elsewhere in the Antilles) are
occasionally burned over and frequently
used for raising such livestock as goats.
No evidence of fire or goats was visible
on Carrot Rock, and its difficulty of ac-
cess may eliminate the temptation to so
abuse it. While I have no evidence that
either burning or goat browsing extir-
pates lizards, I would certainly regard
either or both with grave trepidation on
Carrot Rock. In view of Carrot Rock’s
remarkable nesting seabirds (Lazell et
al., 1982)—none of which utilize the
vegetated upland—and Anolis ernest-
williamsi, this islet should be set aside as
a sanctuary or preserve. Limited collect-
ing of the anole, especially by the tech-
nique of noosing individuals around the
edge of the vegetated top, can do no seri-
ous harm, but such collecting should be
regulated (see Lazell, 1980a).

Discussion. Heatwole’s (1976) taxo-
nomic conclusions are difficult for me to
understand. It is unclear to me why the
karyotypic difference between A.
monensis and A. cristatellus augurs for
the species status of A. desechensis
(Heatwole, 1976:13). In any case, karyo-
typic differences are not always indica-
tive of species status in Anolis (or other
vertebrates): Hall (1974). All the scale
counts provided for desechensis are in-
cluded within the range of variation of
Puerto Rican cristatellus, leaving no
device for separating desechensis from
the yellow-fanned form in southwestern
Puerto Rico. Scale characters are not
specified (on p. 7 Heatwole alludes to a
difference in postrostrals but his Figure 4
belies it), and probably only modal.

I agree with Heatwole (1976:12-13)
that A. c. wileyae is an indivisible entity.
The statistically significant differences
Heatwole found fall far short of taxo-
nomic significance. I find it impossible to
frame diagnoses that would separate 75%
of Vieques specimens, for example, from
those in the Anegada population—at the
opposite extreme of the range. (I have

110

examined 21 Anegada specimens: U.S.
National Museum of Natural History
140309-10; MCZ 12145-56 and 35704-
10.)

Against this background it is easy to
defend Anolis ernestwilliamsi as a full
species. Even granting that the minor
karyotypic difference between cristatel-
lus and monensis bequeaths species
status to them, and assuming that ernest-
williamsi has the cristatellus-like com-
plement of 27 (as one suspects desechen-
sis does), ernestwilliamsi is far more dif-
ferent from any of these previously
named forms than they are from each
other. It does not fit into the pattern of
geographic variation shown by the inter-
grading forms of Anolis cristatellus, and
geographically proximate populations of
A. c. wileyae show no hint of character
approach.

I fought my way through dense scrub
on Peter Island to reach Stoney Bay, the
southern edge of which is the closest
land to Carrot Rock, and there collected a
series of typical A. c. wileyae (MCZ
158927-31). This most proximate habitat
is a xeric woodland dominated by
gumbo-limbo (Bursera simarouba), figs
(Ficus sp.), and mangroves (Rhizophora
mangle and Avicennia rhizophorarum).
A bay head barrier of cobbles, coral, and
slabs that look like limestone fronts the
sea. Anoles were scarce, but there were
geckos (Sphaerodactylus macrolepis),
iguanas (I. iguana), skinks (Mabuya
sloanei), and kestrels (Falco sparverius).
Standing on the cobbles I could identify
to species the tropic birds (Phaethon
aethureus) wheeling over their nest sites
on Carrot Rock, a scant 450 m away. I
wondered if the tropic birds ever
alighted over here, or the kestrels over
there. So far as I know, I am the only
living thing which has ever been both
places.

Anolis ernestwilliamsi is an autochtho-
nous endemic on its peculiar little cay. It
must have evolved very rapidly (see
Lazell, 1972: 102-103) in a few thousand
years of isolation.

Advances in Herpetology and Evolutionary Biology

Electrophoretic, karyotypic, and etho-
logical studies of Anolis ernestwilliamsi
are certainly called for. On present evi-
dence, its status as a full species seems to
me more secure than that of monensis or
desechensis. Granting that tastes differ on
these points, dialogue and exposition
will prove useful to those of us particu-
larly interested in insular speciation and
geographic variation.

OTHER FORMS

I suggest some nomenclatural changes
not commonly in use today. Previously
(Lazell, 1973) I commented on the trivial
nature of the putative distinctions be-
tween “Cyclura,” “Brachylophus,” and
Iguana. Examination of specimens re-
veals that even these differences are
modal; I refer to all species in the com-
plex as Iguana.

Maglio (1970) regarded the Caribbean
colubrids as “oversplit.”” I concur. The
entire complex of “Alsophis,” “Dromi-
cus,’ “Arrhyton,’ and _ Liophis are
closely comparable to Anolis or Eleu-
therodactylus (or Iguana sensu lato) in
diversity. Differences in dentition, cra-
nial bone proportions, coloration, and
hemipenes are, I suggest, reflections of
character divergence among very closely
allied species. Certainly the two Virgin
Island species are congeneric; they seem
a near perfect analog of the sympatric
lizard pair Anolis cristatellus and A.
pulchellus. The oldest name for this
assemblage of snakes seems to be
Liophis; it is used herein.

The species and their disributions are
tabulated (Table 1). I have previously
(Lazell, 1980a) listed four origins for
members of this herpetofauna: 1) Species
widely distributed on the greater Puerto
Rico Bank stranded today in the Virgins
by post-Wurm sea level rise. 2) Species
coming from the west (Greater Antilles
and other Puerto Rico Bank islands)
across water. 3) Species coming from the
east and south (Lesser Antilles) across
water. 4) Autochthonous endemics

whose original ancestors may have come
by any of the first three methods.

I cannot envision a way to separate
examples of the first and second cate-
gories. I suspect most of the herpetofauna
belongs to the first category, but it is truly
unfortunate that I cannot perceive ex-
amples of the second, for they would be
most interesting.

Four species (ca. 17%) came by the
third method from the Lesser Antilles:
Hemidactylus mabouia, Thecadactylus
rapicaudus, Iguana iguana, and Geo-
chelone carbonaria.

Five species (ca. 25%) are endemic to
the islands east of Puerto Rico at full
species level and are not known to have
ever occurred on Puerto Rico: Amphis-
baena fenestrata, Sphaerodactylus par-
thenopion, Iguana pinguis (?), Anolis
ernestwilliamsi, and Eleutherodactylus
schwartzi. There are bones of an Iguana
near pinguis from Puerto Rico. Of these
forms the last four are today known only
from the British Virgin Islands (Lazell,
1980a). All of these species seem most
likely to have been derived from Greater
Antillean (in fact Puerto Rican) stocks. If
one assumes that speciation began after
the retreat of the Wurm, the processes
have been rapid indeed. This is not, of
course, necessarily the case, but seems
most likely for at least Anolis ernest-
williamsi.

THEORIES OF BIOGEOGRAPHY

Two disparate and sharply contrasting
theories of island biogeography have
emerged in recent years. The first, pub-
lished by MacArthur and Wilson (1967), is

essentially abiological. It attempts to ex- |

plain faunas (or floras) entirely in terms
of mensurable physical and/or temporal
parameters such as island area, elevation,
distance from a source of propagules, or
length of time of island separation. The
implicit philosophical basis of this theory
is that biological phenomena and situa-
tions, like those in chemistry and

BRITISH VIRGIN ISLANDS - Lazell Lil

physics, should be reducible to symbols
and numbers, which, in turn, can be ar-
ranged in mathematical formulae that
“predict” (or postdict) similar biological
phenomena or situations.

A necessary adjunct to the MacArthur
and Wilson view is that time of island
separation cannot matter very much be-
cause a large fauna stranded by rising sea
level must necessarily dwindle by extinc-
tions to the “right” number of species for
an island with similar physical para-
meters which was never part of a larger
land area. This notion has been hotly
debated (see Simberloff, 1976, and works
cited therein).

It is interesting that the MacArthur and
Wilson theory germinated from Darling-
ton’s (1957) little rule of thumb which
stated that Antillean herpetofaunas in-
crease by roughly a factor of two for every
increase in island size of a factor of ten.
MacArthur and Wilson explicitly be-
lieved that, by measuring other parame-
ters and coalescing them into ever more
complex formulae, the roughness in Dar-
lington’s rule could be eliminated: the
biota of an island should be calculated
with little more difficulty than Boyle cal-
culated pressure, given volume and
temperature.

Lack’s (1976) view of islands is con-
siderably more complicated. While he
did not doubt the importance in a general
way of obvious physical features like
island size, he was deeply impressed by
biotic factors as well. He was impressed
by the fact that, cumbersome as the for-
mulae became, they still failed to “‘pre-
dict” (or, correctly, postdict to the time
the data were collected) the right num-
bers in many cases. Most importantly,
however, Lack was impressed by the
resilience of insular faunas: apart from
the artificial and edificarian catastrophes
wrought by man, and often in spite of
them, the predicted extinction rates were
simply way out of line with apparent
reality. Far from becoming extinct or re-
placed by new colonizers, insular popu-
lations often tended to expand their

112

niches and resist propagule input for so
long that evolution carried them to the far
realms of the bizarre. The dodo, Lack
might have said, was just a pigeon which
lived alone too long—but very, very long
indeed. Many recent data on. population
genetics within the well-studied species
Felis catus support this view of impene-
trable matrices: Blumenberg (1977);
Todd and Blumenberg (1978).

Lack’s view may have grown from his
field work in the Galapagos. Insular
fragmentation may not result in dimin-
ished faunas at all. Quite the contrary, it
may well lead to spectacular radiations.

The number of forms of finches, or
tortoises, or lizards increases (very
roughly) as a function of the number of
islands. Since rising sea level increases
the number of islands, the fact that it
decreases their size may utterly fail to
reverse the trend of species proliferation.
Anolis ernestwilliamsi, possibly Iguana
pinguis, and the subspecies of Typhlops
richardi and Liophis portoricensis are
examples of this trend in the British Vir-
gins.

ANALYSIS

When a person has a poor ear for music he will
flat and sharp right along without knowing it. He
keeps near the tune, but it is not the tune.

—Mark Twain, 1895

Of the 43 islands with presently de-
monstrated herpetofaunas, only two—
Great Tobago (two species; 88.6 ha) and
West Dog (one species; 12.5 ha) fit the
area-species curve of MacArthur and
Wilson (1967:Fig. 2). All the rest have
“too many’ species. Tortola, Virgin
Gorda, and Peter Island, for example, all
have about three times the predicted
number. Extrapolating the MacArthur
and Wilson curve indicates that islands of
less than ten hectares should have no
species at all. More than 30% of the Brit-
ish Virgin Islands diverge from
MacArthur and Wilson’s predictions,

Advances in Herpetology and Evolutionary Biology

therefore, by an infinite factor. The rela-
tionship of area to number of species is
shown in Figure 6.

Using the basic formula of MacArthur
and Wilson (1967:8), but with new num-
bers, we find:

S = 1.63 A023

Spearman r = 0.797, p = 0.01, which is
reasonable: larger islands do tend to have
more species. E:xtrapolating this curve
into the realm of no species present indi-
cates that at least one species should be
present on any island of ca. 50 m? or
greater. While this prediction will be
most interesting to test, it is certainly less
unreasonable than the previous 10 ha
prediction.

The value Z = 0.233 is far lower than
MacArthur and Wilson cited for any other
fauna except land and fresh-water birds
in the West Indies (1967:9). As Z ap-
proaches zero, the importance of area
diminishes (at Z = O the formula be-
comes S = C).

Using the new, “improved” formula,
14 islands (nearly a third: 32 percent)
have “too many’ species and 11 islands
(more than a quarter: 26 percent) have
“too few.” Therefore, 58% of the islands
fail to fit even the new predictions. It is
important to understand that a biogeo-
graphic formula with a positive correla-
tion of 80% is still wrong most of the
time.

The cases involving too few species
are, I must admit, probably the result of
collecting failure. Great Thatch, Scrub,
Great Tobago, West Dog, and Pelican are
all “missing” two species. Cooper, Little
Jost Van Dyke, West Seal Dog, East Seal
Dog, Carrot Rock, and Broken Jerusalem
are all “missing” one species each. Of
these only Carrot Rock was examined
carefully by me. Nevertheless, I believe
additional search at a wetter season will
reveal at least one more species there,
and I believe all these islands are good
for more species than are known at pres-
ent. Finding the “missing” species will

BRITISH VIRGIN ISLANDS - Lazell 113

20 °
@

10
” @
o 5 ° oe ee @
= eo @ e@ C)
Qo © em @ eo © ee
” =

2 e e e e e Wea

oe
1 e eo e e Secon
0.1 1 10 100 1,000 10,000
Hectares

Figure 6. Area-species relationships. Dots represent the 43 British Virgin Islands with presently known herpeto-
faunas. The solid line is the curve shown by MacArthur and Wilson (1967: Fig. 2). The dashed line is the
extrapolation of this curve to the point where no species would be predicted present by MacArthur and Wilson

(1967).

have a profound effect on the formula and
the next generation of predictions. The
value of C will be elevated; therefore the
predicted size for a minimal area island
harboring one species will be even less
than 50 m?. The value of Z will be re-
duced. Therefore the importance of area
in the calculation will diminish.

Islands with too many species are far
more embarrassing for advocates of a
MacArthur and Wilson approach to bio-
geography. Of the islands with only one
species “too many” the error is only
about one-fifth for Mosquito and Necker
(each with five species). On Fallen
Jerusalem the error approaches 30% and
on Round Rock it is more than a third
(both have four species). For Saba Rock
and Sandy Spit the failure of prediction is
about 50%.

Of islands with two species
many, Great Camanoe and Guana are
not egregiously in error at about 20 to
25%. Sandy Cay (with four species) is
about half off predicted, but Marina Cay
(with five species on one hectare) is
“wrong by a factor of more than two-
thirds.

The larger islands are little closer to

“too-

predicted values. All have counts 36 to
39% too high. This means that Salt Island
has three species “too many,” Peter has
four, Virgin Gorda six, and Tortola has
seven species more than predicted by the
formula. Errors of this magnitude and
frequency render the formulation useless
to me as a practicing field biologist.

There is a weaker but “good” correla-
tion of elevation (E) to species number:
Spearman r = 0.586, p = 0.01. Using both
kinds of data and searching for the best
formulation in terms of least root mean
square error (Smith, 1977: 257), one
gleans:

S = C,A+ C3E + C3

Where: C, = 0.00198 C, = 0.016
C3 = 2.1 RMS = 1.65.
This formula is strongly appealing

mathematically because it represents a
canted, but flat, plane rising in three
dimensions. The fact that it rises from
A = O, E = O, S = 2~an island with
neither area nor elevation, but with two
(sodden) species—is a mathematically

114

inconsequential problem easily relega-
ted to some lizard hunter for solution.

An interesting feature of this formula is
that it values area less by an order of
magnitude than it does elevation. This
reverses the order of correlation pro-
duced when each factor was indepen-
dently compared to species number via
Spearman rank coefficients. —

For islands > 100 ha the percent error
for this formula is > 30 for only two is-
lands: Peter (ca. 47% or five species “too
many ) and Great Thatch (78% or two
species “too few’’). Virgin Gorda has ca.
19 percent, or three species “too many.”
Of the twelve largest islands, the re-
mainder are either off by one or two spe-
cies (seven) and low percentages (10 to
30%) or are right on: two, Beef and
Ginger Islands. In my opinion, both Beef
and Ginger are undercollected and will
prove to support several additional spe-
cies in each case.

Of the 18 medium-sized (> 10, < 100
ha) islands, 13 have numbers of species
differing from the prediction. The worst
is West Dog, ca. 200% off with two spe-
cies “too few.” Great Tobago is more
than 100% off also, and “missing” two
species. Necker Island is more than 40%
off with two species “too many,” and
Scrub errs about the same extent with
two “too few.” Salt Island is about 40%
“wrong, with three species “too many.”
Buck, Little Camanoe, George Dog,
Dead Mans Chest, and Eustatia are right
on. The rest (eight) are off by one spe-
cies; this amounts to an error of more
than a third for Little Jost Van Dyke and
Frenchmans Cay (both “too few’’). For
the remainder the percentage error is
lower.

Of the 13 small islands (< 10 ha), ten
have numbers of species differing from
the prediction. Five of these have errors
between 100 and 200%; of course the
number of species “wrong” is usually
just one, but both Carrot Rock and West
Seal Dog are “missing” two species.
Sandy Cay has two species “too many”
and Marina Cay is the champion with

Advances in Herpetology and Evolutionary Biology

three species “too many.” Bellamy Cay,
Saba Rock, and Sandy Spit are right on.

Thus, using the new formula we get a
much tighter fit in terms of percentage
error in the larger islands (especially),
but for 33 of the 43 islands—more than
three-quarters—the formula is still
wrong. As before, errors of “too few” will
predictably be corrected by better hunt-
ing.

In attempting to find a fit that avoided
the problem of A = O, E = O, S = 2 we
fed the computer data for Carval, Indians,
and North Cockroach, which have fine
areas and elevations, but no known spe-
cies. We also fed it the open ocean at
A = 0, E = 0, S = 0. The computer could
deal with this in the linear equation, but
it did not help. It increased RMS but did
not bring C,; = S below two significant
figures. In exponential arrangements the
log of zero cannot compute, of course.
Giving tiny, false values permitted com-
putation, but no exponential arrange-
ment, with or without the S =O data
points, was as good as the linear one.

A slight further refinement can be
made to this formula utilizing the fact
that there is some relationship between
area and elevation (i.e., big islands tend
to be higher than small islands):

S= C,A + CE a C3AE a5 Cy

Where C, = 0.0025 Cz = 0.019
C3 == IS x lg Cy = 1.8
RMS = 1.60

The effect of the admittedly tiny, nega-
tive value of C, is to dip the plain for
large areas and high elevations. The big-
gest and smallest islands are now right
on. Discounting islands with “too few”
species as examples of collector failure,
however, there are still 13 islands—ca.
30%— with “too many.” Peter Island is
the worst by a factor of nearly 50% and
five species. Virgin Gorda, Salt, Necker,
and Marina Cay all have an excess of
three species. Great Camanoe and Sandy
Cay are off by two. Norman, Prickly Pear,

Mosquito, Little Camanoe, Fallen Jeru-
salem, and Round Rock are all over-
stocked by one species.

The computer still asks us to believe
that the open ocean has two species (at
A=0, E=0, C,=1.8, which is two).
Perhaps it wants us to acknowledge the
existence of marine turtles. In all the
formula provides the wrong answer for 31
of 43 islands, or 72%.

Increasing the number of points to 47
by feeding in the three S = 0 islets and
the open ocean brought C, to one—which
is only half as insane, one might say—but
increased RMS to 1.63. There was no
improvement: the formula was. still
wrong for three-quarters of the islands.

CONCLUSIONS

Statistics is a game played by compu-
ters which are just as intelligent as bowl-
ing balls. My acquaintance is not large,
but I have yet to meet a computer which
knew that 16 (the number of known spe-
cies on Virgin Gorda) is not at all the
same thing as 16.00000. I believe that if
mathematically inclined biogeographers
would be more sanguine about “good”
correlations and actually calculate the
percentage error between the observed
species numbers and their formulas’
“predictions” (and pay attention to which
percentage errors amount to species
numbers) they would take a far less en-
thusiastic view of computer time and
costs. As I have pointed out before
(Lazell, 1976, 1979), to determine the
number of species on an island one must
at the very least go to the island and
count.

One striking fact was available to
MacArthur and Wilson (1967) before they
published: I had already shown that
Sombrero—anchor point in Darlington’s
(1957) table from which MacArthur and
Wilson selected their data (1967: Fig.
2)—had three species present (Lazell,
1964). Most West Indian islands, in fact,
have at least three species comprising a

BRITISH VIRGIN ISLANDS - Lazell 115

herpetofauna standardly involving a
member of the genus Anolis, a member of
the family Gekkonidae, and some other
form.

The Rule of Three in the Virgin Islands
holds for 22 islands over three orders of
magnitude in area: Table 2. All of these
islands support Anolis c. wileyae. All but
three (the American islands of Leduck
Cay, Flanagan, and Salt Cay)—86%—also
support the gecko Sphaerodactylus m.
macrolepis. The teiid lizard Ameiva
exsul is present on 15, or 68%. Other
species making up the complement of
three include a second Anolis (A. stratu-
lus): five for ca. 23%; a skink (Mabuya
sloanei): two for nine precent; a snake
(Liophis portoricensis): two for nine per-
cent; and a second gecko (Hemidactylus
mabouia): one for less than five percent.

The Rule of the Irreducible Anole
states simply that any West Indian islet,
cay, rock, or spit which supports more
than herb-stage vegetation—i.e., bushes
or shrubs—will also support a member of
the genus Anolis. This is not to be con-
fused with Marler’s (1980) Mexican bush
rule, albeit not wholly dissimilar. In my
work in the Lesser Antilles (Lazell, 1972)
I found but one islet—Green Cay off
Saba—that had bushes (two Coccoloba
uvifera) but no Anolis I could find. I
believe every case of an islet in the Brit-
ish Virgins where I failed to find Anolis c.
wileyae (Carval, Indians, North Cock-
roach) is simply that: where I failed.

I predict that subsequent work in these
islands will add at least seven cays to
those fitting the Rule of the Irreducible
Anole (the three I failed to find Anolis on
plus four unnamed ones). I predict that
many islands now thought to fit the Rule
of Three will tur out to have additional
species, and many where less than three
are known today will move up to fill the
ranks. In short, I believe the fauna will
certainly prove to be richer than is now
known.

No extinction or extirpation need take
place in the British Virgin Islands if rea-
sonable conservation measures are en-

116

Advances in Herpetology and Evolutionary Biology

TABLE 2. THREE SPECIES ISLANDS IN THE VIRGINS.

Anolis c. Sph. m.

Island wileyae
Rotto Cay, A
Steven Cay, A
Henley Cay, A
Cas Cay, A
Leduck Cay, A
Cockroach I., A
Congo Cay, A
Flanagan, A
Eustatia

10 Dead Mans Chest
11 George Dog

12 Little Camanoe
13 Buck

14 Little Tobago

15 Little Thatch

16 Frenchmans

17 Salt Cay, A

18 Great Dog

19 Little Jost Van Dyke
20 Scrub

21 Great Thatch

22 Hans Lollik, A

OBONHDUIPRWNW-e

Po PK PS PK PM PK PS PS PS OS OS OK ON PS OS XS SS OO
KP

macrolepis

Ameiva

exsul Other species Area (ha)
XxX 0.8
Anolis stratulus 2.4
Hemidactylus mabouia 5.3

xX 7.1
X Anolis stratulus 183
Liophis portoricensis Toth

X 10.3
X Anolis stratulus 10.5
X 10.5
xX 14.1
X 15.4
X 16.2
X 17.0
Mabuya sloanei 22.3

Anolis stratulus 23.5

xX 24.1
xX Mabuya sloanei 25.6
xX 30.3
X 57.2
Xx 88.6
Anolis stratulus 112

Liophis portoricensis 128

A = American islands with data from Maclean et al. (1977).

acted. I have already listed a number of
endangered, threatened, and/or declin-
ing species and outlined plans to con-
serve them (Lazell, 1980a). In every case
the decline, threat, or endangerment is
strictly artificial: the direct result of
human activities. In no case is a possible
extinction or extirpation the foreseeable
result of any natural phenomenon. Even
tiny, overwashed Sandy Spit appears
likely to accrete sediment on growing
coral about as rapidly as sea level rises. I
expect it to retain its two species and—
with better search—prove to fit the Rule
of Three, at least.

ACKNOWLEDGMENTS

I am most grateful to Gregory Mayer
and Christopher Lutman for assistance in
analyzing the data. William Geizentan-
ner generously prepared Figures 4, 5,

and 6. Numi Spitzer made the map, Fig-
ure l.

Numerous people assisted me in the
field; at risk of omitting someone, I es-
pecially thank George and Luana Marler,
Robert Chipley, John D. H. Smith,
Gerald Christian, Leonard Dawson,
Mary Randall, Herman and Marilyn
Groezinger, Clement and Gracita Faulk-
ner, Lance Vanterpool, Wallace Vanter-
pool, Paul Backshall, Grant Bongiorno,
Hank Valdes, George and Jane Mitchell,
Rowan Roy, Norman Thomas, Ishma
Christopher, and the staff of the Peter
Island Yacht Club. Robert Jenkins and
Robert Creque handled the administra-
tive end; Jose Rosado curated the collec-
tion.

My work was funded by the Depart-
ment of Natural Resources, Government
of the British Virgin Islands, and the
International Program of The Nature
Conservancy.

LITERATURE CITED

AUFFENBERG, W. 1967. Notes on West Indian tor-
toises. Herpetologica, 23: 3444.

____. 1974. Checklist of fossil tortoises (Testudini-
dae). Bull. Florida State Mus., 18: 121-251.

BLUMENBERG, B. 1977. Genetic differences and
selection in domestic cat populations of the
United Kingdom and former British colonies.
Theoret. Appl. Genetics, 49: 243-247.

DARLINGTON, P. J. 1957. Zoogeography: the geo-
graphical distribution of animals. New York,
John Wiley & Sons.

DIRECTORATE OF OVERSEAS SURVEYS. 1959.
Anegada. D.O.S. 346 (E 837). London, Edward
Stanford Ltd.

____. 1977. Anegada. E 803 (D.O.S. 346P). London,
Edward Stanford Ltd.

HALL, W. P. 1974. Report in E. E. Williams (ed.).
The Second Anolis Newsletter, Mus. Comp.
Zool: 1-2.

HEATWOLE, H. 1976. Herpetogeography of Puerto
Rico. VII. Geographic variation in the Anolis
cristatellus complex in Puerto Rico and the
Virgin Islands. Univ. Kansas Mus. Nat. Hist.
Occ. Pap., 46: 1-18.

Lack, D. 1976. Island biology illustrated by the
land birds of Jamaica. Studies in Ecology 3,
Berkeley, Univ. California Press.

LAZELL, J. D. 1964. The reptiles of Sombrero, West
Indies. Copeia, 1964: 716-718.

—___.. 1972. The anoles (Sauria: Iguanidae) of the
Lesser Antilles. Bull. Mus. Comp. Zool., 143:
1-115.

___. 1973. The lizard genus Iguana in the Lesser
Antilles. Bull. Mus. Comp. Zool., 145: 1-28.

___. 1976. This broken archipelago. New York,
Quadrangle, Harper and Row.

—__. 1979. Deployment, dispersal, and adaptive
strategies of land vertebrates on Atlantic and
Gulf barrier islands. Proc. First Conf. Sci. Res.
Nat. Parks, 1: 415-419.

—__. 1980a. Report. British Virgin Islands, 1980.
Privately printed.

__.1980b. Lesser Antillean Anolis (Sauria:
Iguanidae) in the British Museum. J. Herpe-
tol., 14: 194-195.

LAZELL, J. D., R. CHIPLEY, AND G. MARLER. 1982.
Contribution to the ornithology of the British
Virgin Islands: in press.

MACARTHUR, R., AND FE. O. WILSON. 1967. The
theory of island biogeography. New Jersey,
Princeton Univ. Press.

MACLEAN, W. P., R. KELLNER, AND H. DENNIS. 1977.
Island lists of West Indian amphibians and

BRITISH VIRGIN ISLANDS - Lazell Lily

reptiles. Smithsonian Herp. Info. Service, 40:
EAT,

MaGLio, V. J. 1970. West Indian xenodontine
colubrid snakes: their probable origin, phy-
logeny, and zoogeography. Bull. Mus. Comp.
Zool., 141: 1-54.

MARLER, G. 1973. The undersea world of the British
pen Islands. Tortola, Caribbean Printing Co.

td.

———. 1980. In Mexico, no matter how small it is, if
it’s got a bush, that bush will be full of crabs.
Field Cat., 19: 390-391.

MakrTIN-KaYE, P. 1959. Reports on the geology of
the Leeward and British Virgin Islands. St.
Lucia, Voice Publ. Co.

MIRECKI, D. N. 1977. Report of the Cambridge
ornithological expedition to the British Virgin
Islands. Cambridge, Bluebell.

Morris, B., J. BARNES, F. BROWN, AND J. MARKHAM.
1977. The Bermuda marine environment.
Bermuda Biol. Sta. Spec. Publ., 5: 1-120.

PHILIBOSIAN, R., AND J. A. YNTEMA. 1976. Records
and status of some reptiles and amphibians in
the Virgin Islands. I. 1968-1975. Herpeto-
logica, 32: 81-85.

ANID) . 1977. Annotated checklist of the
birds, mammals, reptiles, and amphibians of
the Virgin Islands and Puerto Rico. Fredrik-
sted, St. Croix, Information Services.

TAN . 1978. Records and status of some
reptiles and amphibians in the Virgin Islands.
II. 1975-1976. Herpetologica, 34: 47-51.

SCHWARTZ, A., AND R. THOMAS. 1975. A check-list of
West Indian amphibians and reptiles. Carne-
gie Mus. Nat. Hist. Spec. Publ., 1: 1-216.

SIMBERLOFF, D. 1976. Species tumover and equi-
librium island biogeography. Science, 194:
572-579.

SMITH, J. M. 1977. Scientific analysis on the pocket
calculator. Second edition. John Wiley & Sons,
New York.

Topp, B. B., AND B. BLUMENBERG. 1978. Mutant
allele frequencies and genetic distance rela-
tionships in domestic cat populations of Lower
Egypt and the eastern Mediterranean.
Theoret. Appl. Genetics 52: 257-262.

Twain, M. 1895. Fenimore Cooper's literary of-
fenses. In C. Neider (ed.), The complete
humorous sketches and tales of Mark Twain.
Garden City, N.Y., Doubleday.

WILLIAMS, E. E. 1969. The ecology of colonization
as seen in the zoogeography of anoline lizards
on small islands. Quart. Rev. Biol., 44:
345-389.

____. 1976. West Indian anoles: a taxonomic and
evolutionary summary 1. Introduction and a
species list. Breviora Mus. Comp. Zool. No.
440, pp. 1-121.

Guiano-Brasilian Polychrus: Distribution and
Speciation (Sauria: Iguanidae)

P. E. VANZOLINI!

ABSTRACT. A historical résumé and formal syn-
onymies are presented for Polychrus marmoratus
and P. acutirostris. The two species are compared
from the viewpoint of scale characters, color pat-
tern, body proportions, and some aspects of ecology
and behavior. Their distribution is analyzed and
possible mechanisms of speciation discussed.

INTRODUCTION

Polychrus is a well-defined lizard
genus, but the relationships between its
species are not simple. The two forms
that occur in the Guiano-Brasilian region,
P. marmoratus (L., 1758) (Fig. 1) and P.
acutirostris Spix, 1825, present very
interesting problems of geographical-
ecological distribution and consequently
of speciation mechanisms. The data at
hand permit a statement of such prob-
lems, and some measure of speculation.

HISTORICAL RESUME

The history of Polychrus, as that of any
common genus, is long, voluminous, full
of irrelevance and redundancy, as well as
of overlooked valuable information. At
the species level the main taxonomic
references are:

Lacerta marmorata was diagnosed by
Linné (1758: 208) in the tenth edition of
the Systema Naturae. He had seen
specimens, and described them well, in
the Amoenitates Academicae (Amphibia
gyllenborgiana and Museum Principis),

1Museu de Zoologia, Universidade de Sao Paulo,
Sao Paulo, Brazil.

Figure 1. Polychrus marmoratus from Puruzinho, Rio
Madeira, Amazonas, Brazil. Photo by W. R. Heyer.

and in the Museum Regis; in the Systema
he also cited a Seba plate. In view of the
existence of actual specimens, this Seba
citation as well as others included in the
Amoenitates, but not in the Systema, may
well be dismissed.

In his review of the Linnean materials
in Uppsala, Lonnberg (1896: 14) stated
that the specimen described in Museum
Principis was still extant and was without
doubt Polychrus marmoratus as then
accepted, i.e., in the concept of Boulen-
gers catalogue. The specimen from the
Amphibia gyllenborgiana, however, had
disappeared. The Stockholm specimen
(Museum Regis) had already been lost by
1802 (Andersson, 1900: 10). Holm (1957),
reviewing the Uppsala materials, found
four Amphibia gyllenborgiana speci-
mens of Polychrus overlooked by
Lonnberg (1896), and considered them as
undoubted marmoratus. This name is
then solidly founded on specimens.

Linné, following Seba, initially gave
“Gallaecia” (Galicia, in northern Spain)
as the type locality of L. marmorata,
broadening it to “Hispania” in the Sys-
tema. Gmelin (1789: 1065), in what is
called the thirteenth edition of the Sys-
tema, added “America” to the distribu-
tion.

The first figure I found of Lacerta
marmorata is Lacépede’s (1788) Plate 26,
a rough but perfectly recognizable wood-
cut. His description is adequate, and was
adopted by subsequent writers, who also
copied the figure.

Daudin (1802: 433) gave a good de-
scription and recorded the _ species,
which he called Agama marmorata, from
Surinam. I believe it is convenient to
adopt this, and
where the species is common, as the type
locality, as proposed by Hoogmoed
(1973); this is of course very plausible for
a Linnean species.

Cuvier (1817: 41) erected the genus
Polychrus, monobasic (monotypic at
creation), for Lacerta marmorata, cited
from Lacépéde and thus recognizable

beyond doubt.

indeed Paramaribo,,

POLYCHRUS -: Vanzolini 119

Raddi (1820: 32; 1822: 58) listed
Agama marmorata Daudin from Rio de
Janeiro; this is the first mention of a
specimen from the Atlantic forest, and
the first reference to its being a forest
animal.

Wied collected P. marmoratus, but did
not refer to it in his “Reise” (1820). How-
ever, he made his notes available to
Schinz who, in 1822 (p. 65), described,
without locality, “Polychrus virescens Px.
Max.” In the Beitrage, Wied (1825: 120)
gave the exact locality of virescens, “Villa
Vicoza am Flusse Peruhype,” the present
Nova Vicosa, Bahia; he had meanwhile
come to the conclusion that virescens was
a synonym of marmoratus. Wied’s
specimen is not cited by Burt and Burt
(1931) in their report on the collection of
the American Museum of Natural His-
tory, where Wied’s surviving materials
are kept. In the Abbildungen (1822-1831)
Wied included a plate, published in
1829, with a good figure of the specimen,
labelled as P. marmoratus.

In 1821 the Isis von Oken (pp. 337-
342; see Vanzolini, 1977: 26) published
an advertisement, signed by Wagler, of a
forthcoming book. I have not been able to
find any further reference to this work,
but in the pages following the advertise-
ment there is an anonymous review of
the sample fascicle, certainly by Oken
himself, mentioning a new generic name,
Psilocercus, proposed by Wagler, appar-
ently without further explanation, for
Lacerta marmorata. This name seems to
have never appeared again in the herpe-
tological literature. It is, in any case, a
clear junior synonym of Polychrus
Cuvier, 1817.

Spix (1825: 14, Pl. 14) included Poly-
chrus marmoratus from Rio de Janeiro
among his new Brasilian species, as he
did to a few other forms already known:
this is not to be construed as the descrip-
tion of new taxa. On the next page and
plate Spix described the new species
Polychrus acutirostris, from “sylvis
Bahiae.” I take the word Bahia to mean
the state (then province), not the hom-

120

onymous city, as the latter is always men-
tioned by Spix as “urbs.” Spix’s descrip-
tion, and especially the plate, are suffi-
cient to identify the form. In fact, H. Boie
(1826: 119) challenged the new species,
saying that it was merely P. marmoratus
stripped of its epidermis, but Spix (1826:
603) replied quite adequately, stating
that his was a good specimen and men-
tioning the shape of the snout and the
absence of a gular crest as differential
characters.

Fitzinger (1826), in the Neue Classifi-
cation, cited Polychrus marmoratus,
proposed, as he was wont to do, a nomen
nudum (Polychrus geometricus, “patria
ignota’) and placed acutirostris in his
new genus Ecphymotes, which other-
wise contained E. plica (= Plica plica), E.
undulatus (Anisolepis undulatus), and E.
pictus (Enyalius pictus). In his review of
Spix’s lizards, published after or about
the time of Spix’s death, Fitzinger (1827)
maintained the same scheme.

Spixs materials were later reviewed
again by W. Peters (1877: 410), who con-
firmed the identification of marmoratus
and the validity of acutirostris.

Delaporte (1826), the future Count of
Castelnau, described Polychrus fascia-
tus, without definite locality, but thought
“with good reason” to come from either
the Philippines or the Moluccas.
Dumeril and Bibron (1837: 69) noted that
the type was a stuffed P. marmoratus on
which a vertebral stripe had been
painted. The specimen has not been
cited either by Duméril and Duméril
(1851) or by Guibé (1954) as extant in the
Paris Museum collection.

In his Icones, Wagler (1833a: Pl. 12)
figured and described two species. The
first was Polychrus virescens Wied in
Schinz, to which he attributed Wied’s
and Spix’s Atlantic Forest specimens.
The other, P. strigiventris, he with some
hesitation described as new. Both plates
are very good and clearly of marmoratus.
P. acutirostris was mentioned as a good
species.

Advances in Herpetology and Evolutionary Biology

Wagler (1833b), in a paper on Seba’s
lizard plates, published a synopsis of
Polychrus, including: (i) acutirostris
Spix; (ii) marmoratus Linné; (iii)
geometricus Fitzinger (in doubt; not
diagnosed and so still a nomen nudum);
and (iv) neovidanus, new. The latter was
based on a figure of Seba (Vol. 2, Plate 76,
Fig. 4) and on Spix’s marmoratus. The
name should be attached to Spix’s speci-
men, which has a definite locality (Rio de
Janeiro), and was later examined by W.
Peters; thus neovidanus definitely ap-
plies to the Atlantic forest form.

Wiegmann (1834: 16) in the Herpe-
tologia Mexicana described a new spe-
cies from “Brazil,” Polychrus anomalus,
very recognizably acutirostris. Gray
(1845: 184) proposed the new genus
Sphaerops for P. anomalus.

Hallowell (1845) described from
“Colombia” (actually Venezuela: “with-
in 200 miles from Caracas’’) Leiolepis
auduboni, recognized by Roze (1958),
who saw the presumptive type, as Poly-
chrus marmoratus. Berthold (1846) de-
scribed P. gutturosus from Popayan,
Colombia.

The next significant contribution, and a
very important one, was made by Rein-
hardt and Lutken (1861). They had both
species, marmoratus without prove-
nance data, and acutirostris (under the
name dnomalus) from several unspeci-
fied localities in Minas Gerais, but also
definitely from Lagoa Santa, the first well
defined locality for the species. They
presented an excellent differential di-
agnosis, which has helped to orient the
present work, and noted that the genus
Sphaerops could not subsist on the basis
of Gray’s diagnosis, although it might
eventually come to be justified on other
grounds.

Boulenger (1885: 98-100), in the Brit-
ish Museum Catalogue, using good ma-
terials of marmoratus (including one
specimen from Bahia) and only one
acutirostris from “Brazil,” presented a
reasonable key, and adequate descrip-

tions, but was not as incisive as Reinhardt
and Lutken (l.c.).

Following Boulengers Catalogue
came the usual rash of papers identifying
specimens in collections from several
parts of South America, some fortunately
with field data. Cope (1899: 20) in a bare
list of “a collection of Reptilia from
Argentina’ cited one specimen of
“Polychrus angustirostris Wagl.,”’ an
obvious lapsus calami that should not be
taken into account.

The first interesting contribution of
this period was Peracca’s (1897) observa-
tion that two young marmoratus from
Amazonian Ecuador had smooth ventrals.
Boulenger (1908: 113), using this charac-
ter as diagnostic, described Polychrus
liogaster, from two localities: Provincia
del Sara in Bolivia and “Chancamayo” in
Peru (actually Chanchamayo, beyond
doubt since Schunke was the collector;
Vanzolini, 1978).

In 1910 Werner described P. femoralis
from Guayaquil, Ecuador, and Boulen-
ger, in 1914, P. spurrelli, from the Choco,
in Colombia.

In 1924 Noble described the new
genus and species Polychroides peruvi-
anus. The genus was later synonymized
with Polychrus by Gorman, Huey, and
Williams (1969).

Mertens (1930: 269) reviewed the dif-
ferences between P. marmoratus and
acutirostris. He was the first to mention
what the accumulation of locality records
already permitted to infer: that P.
acutirostris is an open formation form,
and marmoratus a forest one. He also
discussed the meaning of differences in
color pattern, metachrosis, and relative
eye size in terms of the ecological prefer-
ences.

Burt and Burt (1931), in their report on
the American Museum collection, pre-
sented a key to the genus and made some
confusing statements. They initially gave
the range of marmoratus as “northeast-
erm South America” and that of acuti-
rostris as “southern Brazil, Uruguay, and

- tries)

POLYCHRUS - Vanzolini 121

Argentina,” all of which goes against
most of the previous records in the litera-
ture for both species. Next they referred
to the “intermediate region of Sao
Paulo,’ which does not make sense
either in view of the literature or of their
own distributional data. They also recog-
nized the trans-Andean P. spurrelli as a
probable race of marmoratus, and de-
termined as such a Manaus specimen,
which I have seen and is plain mar-
moratus.

In their Check List, Burt and Burt
(1933) assembled all species so far de-
scribed, except gutturosus, as subspecies
of Polychrus marmoratus, with the fol-
lowing geographical scheme: acutiros-
tris, southern Brasil, Uruguay, Paraguay
and eastern Bolivia; femoralis, southern
Ecuador; liogaster, Bolivia and Peru;
marmoratus, northeastern South Amer-
ica; spurrelli, Colombia and northwest-
ern Brasil. This scheme is just as bad as
the preceding one.

Parker (1935: 515), in his herpetology
of Guyana, criticized the Burts’s scheme,
made some comments on the several
forms, and proposed the presently accep-
ted arrangement; his key, adapted from
the Burts’s and not very good, has sub-
sequently been the basis for identifica-
tion, and was adopted, in a simplified
version, by Peters and Donoso-Barros
(1970).

Amaral (1937), in his Check-list, pro-
posed a new and totally gratuitous distri-
butional pattern: Polychrus m. marmora-
tus, “equatorial and northeastern dis-
tricts (from Amazonas to Pernambuco)’;
P.m. acutirostris, “western, southem and
southeastern districts (from Bahia to Rio
Grande do Sul and neighboring coun-

I (Vanzolini, 1948: 386) commented
that all materials then at my disposal
could be unambiguously assigned either
to acutirostris or marmoratus (curiously
enough, I failed to say why) and, unaware
of Mertens (1930), remarked on the
ecological distribution (open formation

122 Advances in Herpetology and Evolutionary Biology

versus forest). Schmidt and Inger (1951:
451), in their turn not aware of my paper,
also came to the conclusion that acuti-
rostris was a savanna (= open formation)
animal.

Peters (1959: 119) commented on the
difficulty of separating marmoratus and
liogaster in Amazonian Ecuador. In 1967
he gave a key to Ecuadorian forms, based
on Parker (1935), stressing body squama-
tion and shape of the canthus rostralis.

SYNONYMY

Polychrus marmoratus (L., 1758)

Lacerta marmorata Linnaeus, 1758: 208.

Polychrus (marmoratus): Cuvier, 1817: 41.

Psilocercus marmoratus: Wagler in Oken, 1821:
341.

Polychrus virescens Schinz, 1822: 65. Brasil (Wied
leg.).

Polychrus fasciatus Delaporte, 1826: 110. Philip-
pine Islands or Moluccas.

Polychrus strigiventris Wagler, 1830: Pl. 12 and
text. No locality given.

Polychrus neovidanus Wagler, 1833: 897. Rio de
Janeiro.

Leiolepis auduboni Hallowell, 1845: 246. “Colom-
bia”: within 200 miles of Caracas.

Polychrus acutirostris Spix, 1825

Polychrus acutirostris Spix, 1825: 15, Pl. Sylvis
Bahiae.

Polychrus anomalus Wiegmann, 1834: 16. Brasil.

Sphaerops anomalus: Gray, 1845: 154.

EXTERNAL MORPHOLOGY

Pholidosis (Hoogmoed, 1973: Figs.
29a—d; Vanzolini, Ramos-Costa, and Vitt,
1980: Figs. 71-74). The species of Poly-
chrus are very similar in external appear-
ance. They are characteristically arbori-
colous, with a laterally compressed body;
the tail is long, prehensile and has all the
scales strongly keeled, the keels forming
longitudinal ridges.

The head is prismatic in shape, with a
definite, if rounded, canthus rostralis.
The upper head scales are irregular,
without even proper supraocular semi-

circles or well differentiated supraorbit-
als. The gulars are arranged in character-
istic longitudinal rows of narrow and
elongate scales. The skin between the
rows permits a considerable amount of
stretching, the whole functioning as a
throat fan.

The eyelids are also extremely char-
acteristic. They are formed by concentric
rows of granules, the upper and lower
lids being fused together in front and
behind, the seams remaining evident.

The dorsal scales, especially on the
anterior half of the body are irregular,
arranged in irregular oblique rows; on
the flanks much wrinkled skin appears
between the rows, allowing for stretching
in display. The ventrals are lanceolate,
keeled or smooth.

Femoral pores are always present, but
vary much in degree of development, as
well as in number (11-24 on a side). At
times they are inconspicuous, or much
deformed.

Although marmoratus and acutirostris
are very similar in physiognomy, they
differ in important respects in their ex-
ternals, as first noted by Reinhardt and
Lutken (1861).

The snout of acutirostris, as the name
indicates, is narrow and pointed, com-
pared to that of marmoratus. This is quite
obvious at inspection, but I have not
been able to devise a morphometric
comparison that would clearly demon-
strate the fact.

The scales of the top of the head of
marmoratus are larger and nearly
smooth, those of acutirostris smaller and
of a very irregular texture.

The nostril, which is large, opens on
the middle of a single scale, and its posi-
tion differs in the two species. In
marmoratus the nasal scale rests on the
middle of the first labial, at most on the
suture between the first and the second.
In acutirostris it is on the suture between
the second and third labials, or squarely
on the third. Correspondingly, there is
one scale between nasal and rostral in
marmoratus, two in acutirostris.

As noted by Mertens (1930), the open-
ing between the partly fused eyelids is
relatively larger in marmoratus. This is
not an easy character to measure, but the
opening in this species approaches 50%
of the diameter of the orbit, being 30% or
less in acutirostris.

In the gular region there are two very
noticeable differences. Starting a little
behind the post-symphysials, there is in
marmoratus a series of 10 to 15 thin
raised scales (the “crista palearis” of the
older authors); these scales are pliable
but form a very obvious crest, no sign of
which is seen in acutirostris. In the latter
the rows of narrow elongate gulars abut
on the labials; in marmoratus there are in
between two or three rows of flat, well
imbricate scales.

The middorsals of marmoratus are
lanceolate, regular, and have a low, even
keel, ending in a small swelling that may
approximate a mucro. The scale rows are
less regularly arranged toward the
sacrum, but the shape of the individual
scales varies little. In acutirostris the
scales are irregular, very frequently
truncated, with blunt, even swollen
keels, or keelless; towards the back they
may be practically smooth.

Much importance has been attached in
the literature to the scales on the anterior
part of the flanks. In typical cases
marmoratus has more regular scales,
with much less exposed skin between the
rows. There is much variation in this
character, though, and its use in keys has
led to confusion.

The ventrals of marmoratus are
regular, and have a low keel on the distal
end, ending in a moderate mucro. In
acutirostris they are narrower, with a
complete high keel, but a less evident
mucro. The general aspect is of a smooth
surface in marmoratus, a ridged one in
acutirostris.

A very interesting difference, appar-
ently first noticed by Reinhardt and
Lutken (1861), is in the shape of the
femoral pores. In marmoratus each pore
typically opens on the middle of a flat

POLYCHRUS - Vanzolini 123

scale; in acutirostris it notches the edge
of the scale. The basic pattern is always
obvious, but there is some variation in
both species. In marmoratus this is not
much: a few pores may sit on the edge of
the scale. In acutirostris there is much
more variation. In some extreme cases
the pore-bearing scale is conical, raised,
volcanolike; the pore, with prominent
secretion plugs, is completely encircled,
but a suture marks the place of the notch.
In other cases the notch is open, narrow
and deep, the scales assuming an irregu-
lar V shape; plugs of secretion may be
present or not. In other cases the emargi-
nations are so feeble that it is under-
standable that some authors have con-
sidered the pores as absent.

Color pattern. It appears that P.
marmoratus has no sex dimorphism in
color. There is considerable variation,
even within single localities, but a gen-
eral pattern is clearly discernible. I shall
describe a live, healthy adult female from
Oriximina, Para (not all authors have fol-
lowed the same scheme in describing
pattern, but all descriptions agree
broadly).

The dorsal parts of the resting lizard
were predominantly brown, the flanks
barred with black, green and yellow, and
the under parts greenish. The top of the
head was havana brown with darker
flecks, turning greenish towards the
snout. The sides of the head were green.
There were two black lines starting from
the middle of the eye and running, re-
spectively, towards the temporal edge
and the corner of the mouth (these are
very obvious in preserved animals).

The scapular region was dark green
with dark brown spotting. On each flank
there were six sets of oblique bars. The
anterior element of each set was a thin

_ black stripe meeting its fellow on the

middle of the back. Each black stripe
melted posteriorly into a green zone; the
green zones grew broader (from six to
twelve scales wide) towards the sacral
region. The skin between the green
scales was black. Behind each green

124

band there was a yellow one, two to four
scales wide, with orange skin between
the scales. On the middle of the back
there were short dark bars between those
that ran down the flanks. The members
were greenish brown or vice-versa. The
ventral parts were lettuce green, the
color extending onto the base of the tail,
the remainder of which was brown. The
skin between the gulars was gray.

The tongue was orange, the mouth gray
and the throat black. In a specimen de-
scribed by Hoogmoed (1973) the whole
mouth was black and the tongue pale
orange.

Changes in the emotional or reproduc-
tive state of the animal cause the pattern
to brighten or fade, from bright green to
somber brown. The underparts may be-
come whitish or light tan. Authors that
have described metachrosis in this ani-
mal are Peracca (1895), Cott (1926),
Mertens (1930), and Hoogmoed (1975).

Polychrus acutirostris (color Plates
XXI and XXII in Vanzolini, Ramos-Costa,
and Vitt, 1980) is a generally gray animal,
with more or less distinct oblique bars in
different tones of gray and brown, with
wavy and frayed edges. The skin is the
color of the surrounding scales. The two
black lines found in P. marmoratus, radi-
ating from the eye, also appear in acu-
tirostris, but are usually very thin, at
times almost imperceptible. I have not
noted the color of the inside of the
mouth.

A most important difference between
the two forms is the fact that acutirostris
presents sex dimorphism: the male in
nuptial livery has two deep black spots
surrounded with yellow on the anterior
flank (Plate XXII in Vanzolini, Ramos-
Costa, and Vitt, 1980).

Size and body proportions. Both spe-
cies show sex dimorphism in snout to
vent length, the females being larger, as
noticed by Beebe (1944) and commented
upon by Underwood (1962) and by Vitt
and Lacher (1981). There seem to be no
interspecific differences in body length.
My largest marmoratus are, male

Advances in Herpetology and Evolutionary Biology

126mm, female 144. Vitt and Lacher
(1981) record respectively 124 and
145 mm for acutirostris.

The two species differ in relative tail
length (as noted by Wagler, 1830;
Wiegmann, 1834; Mertens, 1930) and in
hind limb length (Boulenger, 1885;
Mertens, 1930). The differences are quite
real: for the same snout to vent length
marmoratus has higher values of both
measurements. The matter, however, is
complicated by the presence of sex di-
morphism in tail length in marmoratus
and of geographic variation within both
species; this should be the subject of
specific research.

ECOLOGY AND BEHAVIOR

Habitat. It has been generally recog-
nized that P. marmoratus is associated
with rain forests, and the distributional
data (see Fig. 2) permit no other interpre-
tation. However, it is not absolutely cer-
tain what kind of habitat it prefers. It has
been found inside the forest, on low trees
and vines (Duellman, 1978; personal ob-
servation), and at least once in a crown
situation (Rand apud Rand and Hum-
phrey, 1968), but very frequently in the
ecotone (Rand and Humphrey, 1968;
Crump, 1971; Hoogmoed, 1973; Dixon
and Soini, 1975; Duellman, 1978), in
deciduous forest (Test, Sexton, and
Heatwole, 1966), in secondary growth
(Crump, 1971) and associated with water
courses (Hoogmoed, 1973; Duellman,
1978).

The irregularity with which this lizard
is taken makes me think it is primarily a
crown animal, which at times comes
down to the ecotone, which is ecologi-
cally similar to the crown environment,
and where the animal is of course much
more conspicuous.

Polychrus acutirostris, on the contrary
has been collected in all types of open
formations: cerrados, caatingas, fallow
land, etc. It is usually found on low tree
branches (Vitt and Lacher, 1981), but I

&

have captured one specimen on the
ground.

It would thus seem that the essential
ecological difference between the two
species lies in the association with forest
(marmoratus) as opposed to open forma-
tions (acutirostris), not in a preference
for shade instead of exposed situations, as
supposed by Mertens (1930, 1970).

Locomotion. The environments in
which the two species have been col-
lected have one thing in common: the
substrate is predominantly twigs. It is
thus not surprising that the two species
are very similar in their locomotor be-
havior, which is rather striking. P. mar-
moratus has been studied in detail by
Boker (1929, 1935), and many authors
have figured and described the curious
stances this lizard takes in the vegetation,
especially on thin twigs.

One primary adaptation is the grasping
capability of the feet (Boker, 11.cc.).
Movement along or between branches
may be remarkably slow (thence the
Brasilian common names “sloth beast”
and “blind lizard’) or occasionally by
means of very agile leaps. The animal
may go through long periods of immobili-
ty, hanging in grotesque positions, with
one or more limbs, or the tail, completely
loose. On the ground the lizard may also
be exasperatingly slow or unexpectedly
agile (Test, Sexton, and Heatwole, 1966;
Nicéforo, 1930; personal observation).

There is no comparable study of acu-
tirostris in the literature, but Cott (1926),
who wrote on marmoratus, noted that a
specimen of acutirostris from the cerrado
of Minas Gerais behaved in exactly the
same way. I have had the opportunity of
observing both marmoratus and acu-
tirostris and, at the level of incidental
observation, saw no major differences in
locomotor behavior.

Food. Polychrus marmoratus has been
said in the literature to be “omnivorous ’’;
this is too broad a term, but in reality it
has been known to eat small arthropods
(insects of several orders, spiders) and
vegetable food (leaves, fruits, seed—most

POLYCHRUS + Vanzolini 125

probably also flowers) (Beebe, 1944;
Underwood, 1962; Rand and Humphrey,
1968; Hoogmoed, 1973; Duellman,
1978). P. acutirostris (see Vitt and
Lacher, 1981 for a detailed study) has
fundamentally the same habits.

Agonistic display. Full agonistic dis-
play in Polychrus in general (Breder,
1946; Underwood, 1962; Gorman, Huey,
and Williams, 1969) comprises lateral
flattening of the body, both of the dewlap
and of the thorax (the latter aided by air
sacs connected to the lungs), metachro-
sis, wide opening of the mouth, lunging
and eventually biting. The same ele-
ments, and additionally bobbing are de-
scribed for P. acutirostris by Vitt and
Lacher (1981).

DISTRIBUTION

Materials. 1 found in the literature 111
definite locality citations of P. mar-
moratus and acutirostris (my own papers
not included) which seemed reliable
enough for mapping. I did not include
localities in the Andes of Ecuador and
Colombia because I feel that the situa-
tion there is not well understood (Peters,
1959; Gorman, Huey, and Williams,
1969; materials under study by myself).
These areas, in any case, are not relevant
to the present argument.

Discounting from the 111 citations the
cases in which the same materials were
cited twice and the localities represented
by more than one collection, the litera-
ture yielded 95 individual localities.

In our own collection we have 364 P.
acutirostris and 125 marmoratus, re-
spectively from 114 and 45 localities.
Some of our localities have already been
cited in the literature. We therefore end
up with 129 localities for P. acutirostris

and 102 for marmoratus.

In drawing the map (Fig. 2), some areas
were too densely covered for adequate
representation: I had to omit many locali-
ties in the Distrito Federal of Venezuela,
in the Peninsula de Paria, in northern
Guyana, around Paramaribo, in the

Advances in Herpetology and Evolutionary Biology

Figure 2. Approximate distribution of Polychrus marmoratus (@) and P. acutirostris (O).

Iquitos area and especially in the state of
Sao Paulo.

An important qualification refers to the
southern limit of acutirostris. Koslowsky
(1898) cites the species for Argentina
(Gran Chaco, Misiones, Corrientes); he
obviously saw specimens, but the refer-
ences are vague. The only actual Argen-
tinian specimens mentioned in the litera-
ture seem to be those from the Para-
guayan border collected by the Deutsche
Gran-Chaco-Expedition (Hellmich, 1960)
and one from Misiones (no further data)
cited by Burt and Burt (1931). The oc-
currence of the species in Uruguay is
likewise in doubt, no specimens being
mentioned by Vaz-Ferreira and Sierra de
Soriano (1960) and Lema and Fabian-

Beurmann (1977). Finally, I know of no
specimens from the three southernmost
states of Brasil, Parana, Santa Catarina,
and Rio Grande do Sul.

Description. Inspection of the map
shows one first important feature: the two
species are parapatric. In fact, there are
no intergrading specimens and no cases
of sympatry. The complementarity be-
tween open and closed formations is
practically perfect. The apparent mixture
of localities in northeastem Brasil is due
to the clearing of the Atlantic forest for
agriculture. P. acutirostris very quickly
occupies such spaces; this can also be
very clearly seen in the state of Sao
Paulo.

A second feature is that the distribution

of marmoratus is disjunct. It comprises
two extensive continuous areas, one in
the hylaea, up to the Caribbean coast, the
other in the Atlantic forest, from Recife,
in Pernambuco, to Sao Sebastiao, in Sao
Paulo. Between the two areas there is a
broad gap in northeastern and central
Brasil.

On the contrary, P. acutirostris occurs
continuously from the northernmost cer-
rados and caatingas (Maranhao to Mato
Grosso) to the Chaco, but is not known
from the great savanna isolates of the
North.

DISCUSSION

The problem posed by Polychrus acu-
tirostris is by which mechanisms a spe-
cies adapted and restricted to open for-
mations arises from a forest-dwelling
ancestor with which it maintains a para-
patric relationship.

Vanzolini and Williams (1981) have
proposed a variation of the orthodox
model of geographic speciation, the
“vanishing refuge,’ to account for such
cases. The model starts with a wide-
spread forest form at the humid peak of a
climatic cycle. Deterioration of the cli-
mate brings about the dissection of the
continuous forested areas by inroads of
open formations. The forest isolates, ini-
tially large, progressively shrink in area.
Some of them last the whole of the dry
leg of the climatic cycle, constituting
refuges in which forest fauna survives
and undergoes geographic differentia-
tion. Other forest isolates dwindle and
finally vanish; normally the forest fauna
would disappear with the refuge, but
under certain circumstances, some such
populations might survive, adapted to
open formations and isolated from their
relatives still faithful to the forest and in-
habiting refuges. The populations surviv-
ing in open formations would be geo-
graphically isolated and under strong
selective pressures, i.e., in ideal condi-
tions for fast differentiation.

POLYCHRUS +: Vanzolini 127

Let us see to what extent a simple
model of this type would contribute to
explain the appearance of Polychrus
acutirostris. I begin by taking as putative
parent the geographically most plausible
relative, P. marmoratus.

The main prerequisite for a scenario
such as the one outlined above is that the
population outliving the vanishing
refuge be pre-adapted to life in open
situations. In the paper in which the
model was first proposed (Vanzolini and
Williams, 1981) and in others (e.g., Wil-
liams and Vanzolini, 1980) examples are
given of incidental populations of forest-
dwelling, umbrophilous species, living
in ecotone or edge situations. In the case
of P. marmoratus ecotonal populations
seem to be at least extremely frequent, as
stressed by authors working in the most
diverse areas: the Napo in Ecuador
(Duellman, 1978), the Iquitos region in
Peru (Dixon and Soini, 1975), eastern
Amazonia (Rand and Humphrey, 1968;
Crump, 1971) and Surinam (Hoogmoed,
1973).

If the model is indeed viable, it is hard
to see how Polychrus could avoid it.

The nature of the differences between
marmoratus and acutirostris is not really
essential to the argument, but affords a
measure of support to it. The majority of
these differences—body proportions,
roughness of ventrals, color pattern—are
obvious adaptations to life in cerrados
and caatingas.

In these two types of plant formations
the twig and crown niche is not other-
wise occupied. The only lizard of similar
ecology sympatric with P. acutirostris is
the iguanid Anisolepis grillii, but the
area of sympatry is the “cultural steppe”
of the state of Sao Paulo, clearly under
human influence. Another possibly

’ syntopic species, Anolis meridionalis, a

much smaller animal, is infrequent in the
cerrados and absent from the caatingas.
Karyotypes. There is a fair amount of
cytological work on Polychrus. The situa-
tion is highly interesting: “Polychrus
remains a puzzle within the iguanids.

128

Now that three species have been
studied it is quite clear that they are close
to each other inter se, but distinctly dif-
ferent from anoles and_ anoline
genera...” (Gorman, Huey, and Wil-
liams, 1969: 7). The three species then
studied were P. marmoratus, femoralis
and peruvianus. Later the karyotype of
acutirostris was described by Peccinini
(1970), under the name marmoratus
(specimens from Sao Paulo) and by
Becak et al. (1972) under the name Poly-
chrus sp. (In fact, it was the uncer-
tainty in identifying these specimens
with the help of current literature that led
me to start this review; the problem of
identification turned out to be exceed-
ingly trivial, but the distribution captivat-
ing).

Paull, Williams, and Hall (1976: Table
4, p. 24) call Polychrus “the one egre-
gious example of chromosomal diversity
in a small genus” and thus summarize the
situation: the diploid number varies from
20 (acutirostris) to 26 (femoralis) to 28
(peruvianus) to 30 (marmoratus).

Gorman, Atkins, and Holzinger (1967)
demonstrated that the mechanism of sex
determination in P. marmoratus is of the
type X,X,Y/X,X,X,X,. Peccinini, Frota-
Pessoa, and Ferrari (1971) showed that in
P. acutirostris (also under the name
Polychrus sp.) there was a particular case
of the same mechanism (“pseudo
XO/XX’’), in which the Y chromosome is
indistinguishable from X,. This is again
strong indication that chromosomal re-
arrangements are frequent and obviously
important in Polychrus.

It is possible, then, that we have
chromosomal lability and extrinsic bar-
riers to gene flow acting concurrently, a
most favorable conjunction for fast rates
of evolution. This is the mechanism I
propose for consideration.

SOME OPEN QUESTIONS

This is a preliminary and speculative
work that leaves many problems to be

addressed.

Advances in Herpetology and Evolutionary Biology

An important problem is the nature of
P. marmoratus, which brings forth the
vexing question of disjunct populations
that cannot be sorted out morphologi-
cally. In reality, we are talking about
evolution at the species level, but our
concept of species is extremely pragmatic
and its application never more than pro-
visional.

Another problem is that of time scale.
If there has been differentiation within
Polychrus dependent on climatic cycles,
then it is certainly a Quaternary event,
and this should be taken into account in
discussions of evolutionary rates con-
cerning chromosomal rearrangements
(Williams and Hall, in Paull, Williams,
and Hall, 1976).

That acutirostris does not reach the
open formations north of the Amazon is
probably due to its having arisen south of
it. There are many open formation forms
(e.g., Crotalus, Tropidurus) that occur on
both sides of the valley, and even in
small enclaves of open formations within
the hylaea. Others (e.g., the extremely
successful bird Cariama) do not. Why
this should be so remains a puzzle.

Finally, it becomes mandatory to ex-
plore in depth the differentiation of
Polychrus in the western end of its range,
where the influence of Andean refuges
(Haffer, 1969; Vanzolini and Williams,
1970) is apt to be strongly felt, and where
the relationships between the several
forms are anything but clear. There is in
fact no reason to believe a priori that at
any given time an animal group is sub-
jected to only one speciation mechanism.

ACKNOWLEDGMENTS

This is one of the very few pieces of
research in the last thirty-odd years
which I did not do in close rapport with
Ernest E. Williams. That the stamp of our
collaboration is clearly on it enhances my
pleasure in this dedication. I hope
Ernest, while as usual disbelieving the
results, will in some measure enjoy in the

paper the elements of uncertainty, eco-
tones, karyotypes and some contrariness.

LITERATURE CITED

AMARAL, A. 1937. Check-list of the Lacertilia of
Brazil. C.R. XII Congr. Int. Zool. (Lisbonne
1935), 3: 1733-1743.

ANDERSSON, L. G. 1900. Catalogue of the Linnean
type specimens of Linnaeus’s Reptilia in the
Royal Museum in Stockholm. Bih. t. K.
Svenska Vet.-Akad. Hand. 24 (4) art. 1: 29 pp.

BECAK, M. L., W. BECAK, AND L. DENARO. 1972.
Chromosome polymorphism, geographical
variation and karyotypes in Sauria. Caryologia,
25(3): 313-326.

BEEBE, W. 1944. Field notes on the lizards of Kar-
tabo, British Guiana, and Caripito, Venezuela.
IJ. Iguanidae. Zoologica, New York, 29: 195-
216.

BERTHOLD, A. A. 1846. Ueber verschiedene neue
oder seltene Reptilien aus Neu-Granada und
Crustaceen aus China. Abh. K. Ges. Wiss.
Goettingen, 3: 1-16.

BOKER, H. 1929. Einige Ergebnisse einer biolo-
gisch-anatomisch Forschungsreise nach Nord-
Brasilien. Anat. Anz., 67 (Erganzungsheft): 9-
20.

___. 1935. Einfuhrung in die vergleichende bi-
ologische Anatomie der Wirbeltiere. 2 vol-
umes. Jena, Gustav Fischer.

Bolg, H. 1826. Bemerkungen uber die von Hr. von
Spix abgebildeten brasilianischen Saurier. Isis
von Oken, 18: 117-120.

BOULENGER, G. A. 1885. Catalogue of the lizards in
the British Museum (Natural History). Second
edition. Vol. 2. London, Trustees of the British
Museum, xiii + 497 pp., pls.

___. 1908. Descriptions of new South-American
reptiles. Ann. Mag. Nat. Hist., 1(8): 111-115.

—__. 1914. On a second collection of batrachians
and reptiles made by Dr. H.G.F. Spurrell,
F.Z.S., in the Choco, Colombia. Proc. Zool.
Soc. London, 1914: 813-817.

Burt, C. E., AND M. D. BurT. 1931. South American
lizards in the collection of the American
Museum of Natural History. Bull. Amer. Mus.
Nat. Hist., 61: 227-395.

Burt, C. E., AND M. D. Burt. 1933. A preliminary
check list of the lizards of South America.
Trans. Acad. Sci. St. Louis, 28(1—2): 1-104.

BREDER, C. M., jr. 1946. Amphibians and reptiles of
the Rio Chucunaque drainage, Darien, Pana-
ma, with notes on their life histories. Bull.
Amer. Mus. Nat. Hist., 86(8): 375-436.

Core, E. D. 1899. Catalogue of a collection of Rep-
tilia from Argentina. Sci. Bull. Philadelphia
Commerc. Mus., 1: 19-21.

Cort, H. B. 1926. Observations on the life-habits of
some batrachians and reptiles from the lower
Amazon; and a note on some mammals from

POLYCHRUS - Vanzolini 129

Marajo Island. Proc. Zool. Soc. London, 1926:
1159-1178.

Crump, M. 1971. Quantitative analysis of the eco-
logical distribution of a tropical herpetofauna.
Univ. Kansas, Mus. Nat. Hist. Occ. Pap. 3: 62
pp.

Cuvier, G. 1817. Le Reégne Animal, distribué
d’apres son organisation. Tome II. Paris,
Deterville, xviii + 532 pp.

DaupDINn, F. M. 1802. Histoire Naturelle générale
et particuliere des reptiles. Vol. 3. Paris, F.
Dufart, 452 pp. (For date see Vanzolini, 1977).

DELapPoRTE, F. 1826. Description d’une nouvelle
espece de reptile du genre Marbré (Poly-
chrus). Ann. Sci. Nat., Paris, 9: 110-111.

DIxoN, J. R., AND P. SoINI. 1975. The reptiles of the
Upper Amazon Basin, Iquitos Region, Peru. I.
Lizards and amphisbaenians. Milwaukee Publ.
Mus. Contrib. Biol. Geol, 4: 58 pp.

DUELLMAN, W. E. 1978. The biology of an equa-
torial herpetofauna in Amazonian Ecuador.
Univ. Kansas Mus. Nat. Hist., Misc. Publ. 65:
352 pp.

DUMERIL, A. M. C. AND G. BIBRON. 1837. Erpétolo-
gie générale. Vol. 4. Paris, Librairie Encyclo-
pédique de Roret, ii + 571 pp.

DUMERIL, C., AND A. DUMERIL. 1851. Catalogue
méthodique de la collection des Reptiles
(Muséum d Histoire Naturelle de Paris). Paris,
Gide et Baudry, iv + 224 pp.

FITZINGER, L. 1826. Neue Classification der Rep-

tilien ... Wien, J. G. Heubner, 66 pp.

1827. Recension des Spixischen Ejidechs-
enwerkes. Isis von Oken, 20: 741-750.
GMELIN, J. F. 1789. Caroli a Linné Systema

Naturae ... editio decima tertia, aucta, refor-
mata, cura... Tomus I Pars III: 1035-1514
Lipsiae: G. E. Beer. (For date see Hopkinson,
1908.)

GorMaAN, G. C., L. ATKINS, AND T. HOLZINGER. 1967.
New karyotypic data on 15 genera of lizards in
the family Iguanidae with a discussion of
taxonomic and_ cytological implications.
Cytogenetics, 6: 286-299.

GorMAN, G. C., R. B. HUEY, AND E. E. WILLIAMS.
1969. Cytotaxonomic studies on some unusual
iguanid lizards assigned to the genera
Chamaeleolis, Polychrus, Polychroides, and
Phenacosaurus, with behavioral notes. Brevi-
ora Mus. Comp. Zool. No. 316, pp. 1-17.

Gray, J. E. 1845. Catalogue of the specimens of liz-
ards in the collection of the British Museum.
London, Trustees of the British Museum,

' Xxvili + 289 pp.

GuIBE, J. 1954. Catalogue des types de lezards du
Muséum National d’Histoire Naturelle.
Bayeux, Colas, 119 pp.

HAFFER, J. 1969. Speciation in Amazonian forest
birds. Science, 165 (3889): 131-137.

HALLOWELL, E.. 1845. Description of reptiles from
South America, supposed to be new. Proc.
Acad. Nat. Sci. Philadelphia, 2: 241-247.

HELLMICH, W. 1960. Die Sauria der Gran-Chaco

130

und seiner Randgebiete. Abh. Bayer. Akad.
Wiss. (N.F.), 101: 1-131.

HoLm, A. 1957. Specimina Linnaeana i Uppsala
bevarade zoologiska sammlingar fran Linnés
tid. Uppsala Univ. Arsskr., 1957 (6): 1-68.

HOOGMOED, M. S. 1973. Notes on the herpetofauna
of Surinam IV. The lizards and amphisbaeni-
ans of Surinam. The Hague, W. Junk, ix + 419

____. 1975. De hagedissen en wormhagedissen van
Suriname IT: de leguanen, de skinken. Lacerta,
33 (6): 87-104.

HOPKINSON, J. 1908. Dates of publication of the
separate parts of Gmelin’s edition (13th) of the
“Systema Naturae’ of Linnaeus. Proc. Zool.
Soc. London, 1907: 1035-1037.

KosLowsky, J. 1898. Enumeracion sistematica y
distribucion geografica de los réptiles argen-
tinos. Rev. Mus. La Plata 8: 161-200, pls.

LACEPEDE, B. G. E. L., COMTE DE, 1788. Histoire
naturelle des quadrupedes ovipares et des
serpens. Tome premier. Paris, Imprimerie du
Roi, xviii + 651 pp.

LeEMA, T., AND M. E. FABIAN-BEURMANN, 1977.
Levantamento preliminar dos répteis da regiao
da fronteira Brasil-Uruguai. Iheringia (Zool.),
Porto Alegre, 50: 61-92.

LINNE, C. 1758. Systema Naturae, per regna tria
naturae ... Editio decima, reformata.
Holmiae, Laurentii Salvii (Cura Societatis
Zoologicae Germanicae iterum edita, 1894.
Lipsiae, Wilhelm Engelmann), 824 pp.

LONNBERG, L. G. 1896. Linnean type-specimens of
birds, reptiles, batrachians and fishes in the
Zoological Museum of the R. University in
Upsala. Bih. t. K. Svenska Vet.-Akad. Handl.,
22 (4) art., 1: 45 pp.

MERTENS, R. 1930. Zur Biologie der Iguanidengat-
tung Polychrus G. Cuvier. Zool. Gart., 3 (9-10):
269-279.

____. 1970. Reptilienauge und Umwelt. Natur und
Museum, 100 (10): 435-446.

NICEFORO Maria. 1930. Los reptiles de Villavicen-
cio en el museo de La Salle. Rev. Soc. Colomb.
Ci. Nat., 19 (105): 40-54.

NOBLE, G. K. 1924. New lizards from northwestern
Peru. Occ. Pap. Boston Soc. Nat. Hist., 5: 107—
113.

PARKER, H. W. 1935. The frogs, lizards and snakes of
British Guiana. Proc. Zool. Soc. London, 1935:
505-530.

PAULL, D., E. E. WILLIAMS, AND W. P. HALL. 1976.
Lizard karyotypes from the Galapagos Islands:
chromosomes in phylogeny and evolution.
Breviora Mus. Comp. Zool. No. 441, pp. 1-31.

PECCININI, D. 1970. Variacao dos cromossomos do
lagarto Polychrus marmoratus (Sauria, Iguan-
idae) de diferentes localidades (nota pre-
liminar). Rev. Brasil. Biol., 30 (1); 1-4.

PECCININI, D., O. FrRoTA-PESSOA, AND I. FERRARI.
1971. Sex determination of the “pseudo

Advances in Herpetology and Evolutionary Biology

XO/XX” type in the Brazilian lizard Polychrus
sp. (Sauria, Iguanidae). Caryologia, 24 (1): 129-
139.

Preracca, M. G. 1895. Viaggio del dott. Alfredo
Borelli nella Repubblica Argentina e nel
Paraguay. X. Rettili ed anfibi. Boll. Mus.
Torino, 10 (195): 32 pp.

PERRACCA, M. G. 1897. Viaggio del Dr. Enrico Festa
nell’Ecuador e regioni vicine. IV. Rettili. Boll.
Mus. Torino, 12 (300): 20 pp.

PETERS, J. A. 1959. Notas miscelaneas sobre saurios
del Ecuador. Ci. Nat., Quito, 2 (3): 118-124.

PETERS, J. A., AND R. DONOSO-BarRos. 1970. Cata-
logue of the Neotropical Squamata. Part II.
Lizards and amphisbaenians. Bull. U.S. Nat.
Mus., 297: viii + 293 pp.

PETERS, W. 1877. Uber die von Spix in Brasilien
gesammelten Eidechsen des Koniglichen
Naturalien-Kabinets zu Munchen. Monatsber.
Akad. Wiss. Berlin, 1877: 407-414.

RApDDI, G. 1820. Di alcune specie nuovi di rettili e

piante brasiliane. Atti. Soc. Ital. Sci., Modena,

18: 1-39, pls.

1822. Continuazione della descrizione dei
rettili brasiliani indicati nella memoria inserita
nello secondo fascicolo delle memorie di fisica
del precedente volume XVIII. Atti Soc. Ital.
Sci., Modena, 19: 56-73.

RAND, A. S., AND S. S. HUMPHREY. 1968. Interspeci-
fic competition in the tropical rain forest:
ecological distribution among lizards at Belém,
Para. Proc. U.S. Nat. Mus., 125 (3658): 17 pp.

REINHARDT, J., AND C. LUTKEN. 1861. Bidrag til
Kundskab om Brasiliens Padder og Krybdyr.
Forste Afdeling, Paddeme og Ogleme. Vid.
Medd. Nat. For. Copenhagen, 1861: 143-242,
pls.

RoZE, J. A. 1958. On Hallowell’s type specimens of
reptiles from Venezuela in the collection of the
Academy of Natural Sciences of Philadelphia.
Notulae Naturae, 309: 4 pp.

SCHINZ, H. R. 1822. Das Thierreich eingetheilt
nach dem Bau der Thiere als Grundlage ihrer
Naturgeschichte und der vergleichende An-
atomie von dem Herm Ritter von Cuvier. Vol.
2: 1-189. Stuttgart and Tubingen: J. G. Cotta.

SCHMIDT, K. P., AND R. INGER, 1951. Amphibians
and reptiles of the Hopkins-Branner Expedi-
tion to Brazil. Fieldiana Zool., 31 (42): 439-
465.

SpIx, J. B. 1825. Animalia nova sive Species novae
lacertarum quas in itinere per Brasiliam annis
MDCCCXVII-MDCCCXX jussu et auspicius
Maximiliani Josephi I Bavariae Regis suscepto
collegit et descripsit Dr. J. B. de Spix. Lipsiae,
T. O. Weigel, 26 pp., pls.

___.. Berichtigung der Bemerkungen des Herm Dr.
Boie in Leyden ueber die von Herm Dr. Spix
abgebildet Saurier (Isis 1826 Heft 1. S. 117).
Isis von Oken, 18: 601-604.

TEST, F. H., O. J. SEXTON, AND H. HEATWOLE. 1966.

Reptiles of Rancho Grande and vicinity,
Estado Aragua, Venezuela. Misc. Publ. Mus.
Zool. Univ. Michigan 128: 63 p.

UNDERWOOD, G. 1962. Reptiles of the Eastern
Caribbean. Caribbean Affairs (N.S.), 1: 192 pp.

VANZOLINI, P. E. 1948. Notas sobre os ofidios e
lagartos da Cachoeira de Emas, no municipio
de Pirassununga, estado de S. Paulo. Rev.
Brasil. Biol., 8 (3): 377-400.

____. 1977. An annotated bibliography of the land
and fresh-water reptiles of South America. Vol.
1 (1758-1900). Sao Paulo, Museu de Zoologia
da Universidade de Sao Paulo, iv + 186 pp.

___. 1978. An annotated bibliography of the land
and fresh-water reptiles of South America. Vol.
2 (1901-1975). Sao Paulo, Museu de Zoologia
da Universidade de Sao Paulo, 316 pp.

VANZOLINI, P. E., A. M. M. RAMOoS-CosTA, AND L. J.
VitT. 1980. Repteis das caatingas. Rio de
Janeiro, Academia Brasileira de Ciéncias, 161
pp., pls.

VANZOLINI, P. E., AND E.. E. WILLIAMS. 1970. South

-American anoles: the geographic differentia-
tion and evolution of the Anolis chrysolepis
species group (Sauria, Iguanidae). Arq. Zool.

S. Paulo, 19 (1-4): 298 pp., pl.

AND . 1981. The vanishing refuge: a
mechanism for ecogeographic speciation.
Papéis Avulsos Zool., S. Paulo, 34 (23): 251-
256.

VAZ-FERREIRA, R., AND B. SIERRA DE SORIANO. 1960.
Notas sobre reptiles del Uruguay. Rev. Fac.
Human. Ci. Montevideo, 18: 133-206.

VitT, L. J., AND T. E. LACHER jr. 1981. Behavior,
habitat, diet, and reproduction of the iguanid

POLYCHRUS - Vanzolini 131

lizard Polychrus acutirostris in the caatinga of
northeastern Brazil. Herpetologica, 37 (1): 53—
63.

WAGLER, J. 1821. Ankundigung: Die Amphibien.
Isis von Oken, 8: 337-342.

___. 1830. Naturliches System der Amphibien ...
Munchen, Stuttgart und Tubingen, J. G. Cotta,
vi + 354 pp.

____. 1833a. Descriptiones et icones amphibiorum.

Pars tertia. Monachii, Stuttgartiae et Tubingae,

J. G. Cotta, (For date see Vanzolini, 1977).

1833b. Deutung die in Sebas’s Thesauro

Rerum Naturalium T. 1. et 2. enthaltenen

Abbildungen von Lurchen, mit _ kritische

Bemerkungen. Isis von Oken, 1833 (9): 896—

905.

WERNER, F. 1910. Ueber neue oder seltene Rep-
tilien des Naturhistorischen Museums in
Hamburg II. Eidechsen. Mitteil. Naturhist.
Mus. Hamburg, 27: 1-46.

WIED-NEUWIED, MAXIMILIAN, PRINZ ZU. 1820. Reise
nach Brasilien in den Jahren 1815 bis 1817.
Vol. 1. Frankfurt a.M., H. L. Bronner.

____. 1825. Beitrage zur Naturgeschichte von Bra-
silien. Vol. 1. Weimar, G.R.H.S. priv. Landes-
Industrie-Comptoirs.

____.. 1822-1831. Abbildungen zur Naturgeschichte
Brasiliens: Weimar. 2 volumes (For dates see
Vanzolini, 1977).

WIEGMANN, A. F. A. 1834. Herpetologia Mexi-
cana... Berlin, C. G. Luderits, vi + 54 pp.

WILLIAMS, E. E., AND P. E. VANZOLINI. 1980. Notes
and biogeographic comments on anoles from
Brasil. Papéis Avulsos Zool. S. Paulo, 34 (6):
99-108.

Systematics of the Latin American Snake,
Liophis epinephelus (Serpentes: Colubridae)

JAMES R. DIXON!

ABSTRACT. Liophis epinephelus is discussed in
light of the 14 names associated with the species.
Four currently recognized species, L. bimaculatus,
L. pseudocobella, L. fraseri, and L. albiventris are
reduced to subspecies of L. epinephelus. The
known variation and distribution of L. epinephelus
is presented; its possible origin is discussed con-
sidering past historical events.

INTRODUCTION

The widespread and highly variable
Latin American snake, Liophis epi-
nephelus, is the most enigmatic of all
Liophis species. It has the greatest alti-
tudinal distribution of the genus, sea
level to 3,400 m. The species is primarily
Trans-Andean, occurring in the lowlands
of Panama, Colombia, and Ecuador; but
also in foothills and mountains of Costa
Rica, Panama, Colombia, Venezuela,
Ecuador, and Peru (Figs. 1-4). Each
mountain range and/or major vegetation
community may contain a population of
L. epinephelus with somewhat different
characteristics than adjacent ones. This is
evidenced by the various color pattern
morphs from the northern, northwestern,
and western Andes, the Trans-Andean
lowlands of Colombia and Ecuador, low-
lands of Panama, and mountains of Costa
Rica (Figs. 5, 6). These morphs resulted
in published descriptions of Liophis epi-
nephelus (Cope, 1862), L. reginae alb-
iventris (Jan, 1863), L. reginae quadri-

1Department of Wildlife & Fisheries Sciences,
Texas A&M University, College Station, Texas
77843, U.S.A.

lineata (Jan, 1863), L. alticolus (Cope,
1868), Zamensis ater (Gunther, 1872), L.
fraseri (Boulenger, 1894), L. bimaculatus
(Cope, 1899), L. opisthotaenia (Boulen-
ger, 1908), L. pseudocobella (Peracca,
1914), L. cobella alticolus (Amaral, 1931),
L. taeniurus juvenalis (Dunn, 1937), and
L. bimaculatus lamonae (Dunn, 1944).
Some quantitative characters (preoculars,
ventrals) associated with changes in ele-
vation resulted in the descriptions of L.
bipraeocularis (Boulenger, 1903) and L.
epinephelus ecuadorensis (Laurent,
1949).

Of the 14 published names, five are
currently recognized as full species: al-
biventris, bimaculatus, fraseri, pseudo-
cobella, and epinephelus. In this paper, I
reduce the first four to subspecies of epi-
nephelus.

A definition of L. epinephelus includes
most quantitative and color pattern char-
acteristics of all species of Liophis. Scale
rows are typically 17-17-15, smooth, one
apical pit; usually one preocular (occa-
sionally two); two postoculars (occasion-
ally one or three); one loreal; one anterior
temporal (occasionally none); two pos-
terior temporals (occasionally one or
three); fourth and fifth supralabial enter-
ing orbit, infrequently third and fourth or
10 other combinations; supralabials nor-
mally eight, occasionally nine other
combinations; infralabials typically ten,
occasionally 11 other combinations
(Tables 1, 2). The number of ventrals
vary from 128 to 191; subcaudals 44 to 80;
maxillary teeth 17 to 29; total length/tail
ratios .17 to .27; hemipenial length 7 to 13

SYSTEMATICS OF LIOPHIS EPINEPHELUS - Dixon

Figure 1.

subcaudals; total maximum length 775+
mm, 2°, 700+ mm, 6¢ (both with in-
complete tails).

Color patterns vary by population.
Some are banded red and black dorsally
and ventrally; some are striped pos-

133

Distribution of Liophis epinephelus juvenalis (black circle), L. e. epinephelus (open circles), and their
intergrades (open circles with X) in Costa Rica and Panama. Insert illustrates the zone of contact between the two
subspecies along the Costa Rica/Panama border. The dashed line represents the approximate position of the
1,500 m contour line.

teriorly with two or four black stripes on
an olive, green, or brown ground color;
some are totally leaf green dorsally and
pale yellow ventrally; others may have
combinations of black and yellow or red
checkered venters with two black dorsal

134 Advances in Herpetology and Evolutionary Biology

Figure 2. Distribution of L. epinephelus and its subspecies in the vicinity of the Andes of Colombia and western
Venezuela. Asterisk = L. e. opisthotaenia; Black square = L. e. lamonae; Black star = L. e. epinephelus; Black
circle = L. e. bimaculatus; Black eight point star = L. e. pseudocobella; Black circle with a white star = sympatry
of two and four-lined patterns and possible intergrades between the subspecies bimaculatus and lamonae; Open
square with a black dot = intergrades between the subspecies pseudocobella and lamonae; Open square with
open star = intergrades between subspecies pseudocobella and epinephelus; Open square with black star =
intergrades between the subspecies pseudocobella and bimaculatus; stippled zone represents elevations above
1,400 m.

SYSTEMATICS OF LIOPHIS EPINEPHELUS : Dixon 135

Figure 3. Ecuadorian distribution of L. epinephelus fraseri (black square), L. e. lamonae (open circle with black
dot), L. e. albiventris (black circle), L. e. bimaculatus (open circle with black triangle), L. e. epinephelus x e.
albiventris (open circle with black star), and L. e. bimaculatus x e. fraseri or e. albiventris (half white/black circle).
Dotted line represents the approximate 1,500 m contour.

136 Advances in Herpetology and Evolutionary Biology

Figure 4. Distribution of samples of L. epinephelus by latitude in Latin America. Numbers in circles correspond
to map code numbers and subspecies given in Table 2 and Figure 5.

SYSTEMATICS OF LIOPHIS EPINEPHELUS - Dixon

137

TABLE 1. VARIATION IN SUPRALABIALS, INFRALABIALS, AND SUPRALABIALS ENTERING ORBIT OF L. EPINEPHELUS.

SUPRALABIALS BS | CO en Te 7s 8S eo 68) 1070
e. juvenalis 49 1 1 1
e. juvenalis X epinephelus 33
e. epinephelus 99 4
e. pseudocobella 2 29
e. opisthotaenia 1 8 9 14
e. bimaculatus 1 1 8 5 22
e. albiventris 1 3 86 1
e. lamonae 1 1 1 39
e. fraseri 1 1 2 6 33

INFRALABIALS 7-7 7-8 7-9 8-8 89 8-10 9-9 9-10 9-11 10-10 10-11 11-11
e. juvenalis 1 5 42 2 3
e. juvenalis X epinephelus 2 28 B IL
e. epinephelus 2 1 5 ja 76 2} 2
e. pseudocobella 11 4 1 13 1
e. opisthotaenia By 19 w 1
e. bimaculatus 1 1 1 6 3 7 5 4
e. albiventris 1 ® 3 10 0 66 2 1
e. lamonae 1 3) 5) G 7% 17 il
e. fraseri Bi > IG) wal 12

3+4 3+4+5 4+5 4+5 4th

SUPRAL/ORBIT 3rd 3+4 445 4+5 34+44+5 4+5 4+5+6 4+5+6 5+6 4+5 5+6
e. juvenalis 48 2 1 1
e. juvenalis X epinephelus 33
e. epinephelus 87 3
e. pseudocobella 29
e. opisthotaenia 6 9 2 14
e. bimaculatus LB 4 29 1
e. albiventris Il 92 1
e. lamonae 3 46
e. fraseri 4 6 36

nape spots, obscure dorsal diagonal dark
markings, and black tail stripes. Some
high altitudinal populations tend to have
75% of the venter black. Intergrade zones
between populations typically have
combinations of patterns and these fre-
quently aid in allocating the intergrades.

ALLOCATION OF NAMES

Rather than discussing each subspe-
cies individually, I have discussed them
collectively under an intraspecies varia-
tion section. There are eight recogniz-
able taxa associated with the species epi-

nephelus. To avoid confusion and proper-
ly allocate all published names from the
past 100 years, the following synonymies,
available names, and distributions are
listed for the subspecies of L. epi-
nephelus.

1) Liophis epinephelus epinephelus Cope
1862

Lower elevations of mountains of
western Panama east to Colombian
lowlands and most Colombian inter-
Andean valleys to elevations of
1,500 m; also south to extreme nor-
thern Ecuador along Pacific coast.

138 Advances in Herpetology and Evolutionary Biology
TABLE 2. RANGE OF VARIATION OF VENTRALS AND SUBCAUDALS OF L. EPINEPHELUS BY LATITUDE.
LATITUDE N R M SE N R M SE MAP CODE
10-11° N (J)* 18 139-148 143.9 0.57 18 5262 545 0.68 la
(E) 4 138-152 144.8 3.00 4 60-66 63.5 1.50 lb
9-10° N (J) 34 139-152 145.7 0.50 30 #£5i1-61 56.5 0.42 2a
(E) 35 133-142 136.9 0.60 33 5467 460.0 0.54 2b
(O) 29 142-152 147.0 053 29 5378 64.7 0.92 2e
8—9° N (J) 2 139-140 139.5 — 2 59-63 61.0 — 3a
(J/E) 33 132-148 140.0 0.50 30 5362 583 0.50 3b
(E) 13 136-147 139.7 0.94 14 55-64 60.6 0.64 3c
(E) 3 134-144 138.0 — 3 5666 60.0 — 3d
6-7° N (E) 7 132-141 137.9 1.10 7 61-69 643 1.00 4a
(L) 5) 141-150 144.8 1.60 5 od461 574 1.10 4b
(P) 38 143-151 146.4 0.35 35 4854 £512 0.31 4c
3-6° N (E) 24 128-149 137.1 082 23 4972 62.6 1.10 da
(L) 3 148-153 150.3 — 3 53-55 54.0 — 5b
(P) 1 149 — — 1 45 — — 5c
(B) 1 171 — — 1 68 — — 5d
45° N (E) wy 142 — — 2 61-67 464.0 — 6a
(L) 15 146-156 150.0 0.96 15 57-67 63.4 0.70 6b
(P) vif 135-146 141.8 1.50 5 47-53 48.8 1.10 6c
(B) 30 165-184 174.2 0.90 24 5680 63.3 0.97 6d
34" (E) B) 134-143 138.6 1.60 5 6471 ~& 68.6 1.00 Ta
(L) 1 156 — — 1 64 — — 7b
2-3° N (L) 2 147-149 148.0 — 2 59 — — 8a
(P) 9 149-158 151.8 1.00 9 51-57 34.0 0.78 8b
(B) 1 169 — — 1 62 — — 8e
1-2° N (E) 2 141-151 147.5 — 2 66-70 68.0 — 9
Q-1° N (A) 20 145-164 152.3 1.10 19 55-70 461.4 1.00 10a
(L) 1 146 — — 1 68 — — 10b
(B) 3 185-188 186.0 _ 1 69 — = 10c
0-1° S (A) 39 143-165 151.7 0.80 36 5368 59.9 0.65 lla
(L) 2) 154-156 155.0 — 2 60-61 460.5 — 1lb
2-3° S (A) 6 143-154 149.2 1.60 6 55-66 60.1 1.70 12a
(L) 2 143-146 144.5 — 2 51-57 # 54.0 — 12b
34° S (A) 1 141-148 143.7 1.10 7 5468 61.1 1.70 13
4-5° § (F) 28 151-162 155.5 0.50 28 #£=61-74 ~ #467.7 ~~ 0.70 14
7-8° S (F) 11 153-164 159.2 1.00 10 61-76 67.2 1.40 15
IB? & (F) 2 145-154 149.5 — 2 53-56 545 — 16
*Letter in parenthesis indicates first letter of subspecies name to which the sample is referred.
A = albiventris; B = bimaculatus; E = epinephelus; F = fraseri; J = juvenalis; L = lamonae; O = opisthotaenia; P = pseudocobella; J/E =
juvenalis * epinephelus. (See Fig. 4 for map localities.)
2) Liophis ephinephelus albiventris Jan western Andean slopes of southern

1863

Liophis reginae quadrilineata Jan 1863
Opheomorphus alticolus Cope 1868
Zamensis ater Ginther 1872

Western Ecuador from sea level to
2,600 m in the Andean valleys.
Liophis epinephelus fraseri Boulen-
ger 1894

Middle elevations of the eastern and

4)

Ecuador, south to central Peru.

Liophis epinephelus bimaculatus Cope
1899

Liophis bipraeocularis Boulenger 1903
Leimadophis epinephelus ecuadorensis
Laurent 1949

High (2,600-3,300 m) Andean valleys
of western Venezuela, central
Colombia, south to northern Peru.

SYSTEMATICS OF LIOPHIS EPINEPHELUS + Dixon

5) Liophis epinephelus
Boulenger 1908
Merida region of Venezuela and
Paramo de Tama region of Venezuela
and Colombia.

6) Liophis epinephelus pseudocobella
Peracca 1914

Liophis cobella alticolus Amaral 1931
Middle elevations of central and

western portions of Colombian
Andes south to Ecuador border.

7) Liophis epinephelus juvenalis Dunn
1937
Middle slopes of most Costa Rica
mountains southeast to western
Panama.

8) .Liophis epinephelus lamonae Dunn
1944

opisthotaenia

Eastern slopes of Andes of Colombia
from 1,500 to 2,600 m, southward to
east central Ecuador.

Liophis reginae albiventris and_L.
reginae quadrilineata represent color
pattern morphs of west Ecuadorian L.
epinephelus; the morphs show a complex
pattern of sympatry and allopatry in
Ecuador. About 98% of western Ecua-
dorian lowland epinephelus have im-
maculate cream or yellow venters and a
black lateral posterior body stripe ex-
tending to the tip of the tail. Occasionally
juveniles and adults possess paraverte-
bral black stripes beginning at midbody
or more posteriorly, in addition to the
lateral black stripe characteristic of the
quadrilineata morph. With increasing
altitude, the frequency of four black
stripes does not increase, whereas the
number of ventrals do. Liophis alticolis is
identical to L. albiventris with the excep-
tion of the abnormal single postocular.
Zamensis ater, purportedly from Biscar,
Algeria, represents “blackened” speci-
mens of Liophis albiventris, probably
from western Ecuador.

Liophis bimaculatus from the Sabana
region of Bogota, Colombia, normally has
four posterior black stripes, occasionally

139

two, infrequently none. The venter may
be immaculate, dusky, black and yellow
checkered, or almost entirely black.
White ventered bimaculatus are very
similar to albiventris in color pattern but
are distinguished by their high numbers
of ventrals and persistence of four black
stripes.

Liophis fraseri and L. bimaculatus
lamonae represent allopatric populations
of L. epinephelus from alternate slopes of
the middle elevations of the eastern,
western, and/or southern flanks of the
Andes.

The holotype of L. bipraeocularis is a
two preocular variation of L. e. bimacu-
latus, and the holotype of L. epinephelus
ecuadorensis is a pattenless example of
L. e. bimaculatus.

Liophis opisthotaenia and L. taeniurus
juvenalis are examples of two (semi-
isolated) mountain populations of L. epi-
nephelus to the west and east of the
nominate subspecies’ major distribution.

Liophis pseudocobella and L. cobella
alticolus represent the same population
of L. e. pseudocobella from near Medel-
lm, Colombia.

ZOOGEOGRAPHICAL COMMENTS

The evidence at hand suggests there
were two separate invasions of pro-
epinephelus stock in the western and
central Andes of Colombia. The pro-
epinephelus stock probably crossed the
Andes in the Quaternary, at or near the
Huancabamba Deflection zones of nor-
thern Peru. This invasion through the
Andes left some populations behind that
became isolated and/or adapted for
higher elevations. Those that passed
through the portal and distributed them-
selves along the Pacific lowlands moved
northward into favorable environments,
eventually reaching northern Costa Rica.
The harsh deserts to the south prevented
distribution in that direction. Those
populations that remained at or near the
upper slopes of the Andes moved both

140

north and south, but at a slower rate than
the Pacific lowland population. The ex-
tremely cool environments of the upper
slopes probably delayed adaptation to
and reproduction in such environments.
By the time the upper slope populations
reached the highlands of northem
Colombia, the lowland Pacific popula-
tions had established themselves along
the middle slopes, eventually evolving
into the pseudocobella and juvenalis
races of the western slopes of Colombia
and Costa Rica. The lowland populations
adapted to the warmer wet and dry en-
vironments, eventually evolving into the
epinephelus (warm-wet) and albiventris
(warm-dry) subspecies.

The populations of the higher slopes
evolved into a two and four-lined pattern
population; those that continued to ex-
pand their range into the very high slopes
evolved into the four-lined subspecies
with high numbers of ventrals (bimacu-
latus), and those that remained along the
higher middle slopes evolved into the
subspecies fraseri, lamonae and opistho-
taenia. The latter two races were prob-
ably isolated by the Tachira Depression
in northem Venezuela and evolved allo-
patrically.

INTRASPECIES VARIATION

The eight populations are recognized
as subspecies by a combination of color,
pattern, squamation, and to some extent,
maxillary teeth (Fig. 7). Each population
exhibits a unique feature which may
represent a major environmental adapta-
tion. The major shift in squamation, such
as the number of ventrals and subcau-
dals, or meristic features such as tail
length, are associated with cooler en-
vironments (higher elevations) and Plei-
stocene orogeny that resulted in geo-
graphic isolation. For example, L. e.
juvenalis and L. e. pseudocobella possess
a very similar color pattern of red and
black bands encircling the body and the
absence of a posterior lateral black line

Advances in Herpetology and Evolutionary Biology

that extends to the tip of the tail. There
are 32 to 49 (x 41.4) black body bands in
juvenalis and 28 to 42 (x 37.4) in pseudo-
cobella. However similar they are in
color pattem, number of ventrals, maxil-
lary teeth and tail/total length ratios, the
two taxa are still significantly distinct in
the number of subcaudals (Table 3, Fig.
8). Liophis e. pseudocobella occurs~be-
tween 1,500 and 2,000 m along the mid-
dle slopes of the central and western
Andes of Colombia; L. e. juvenalis occurs
between 650 and 2,100 m along both
slopes of the mountains of Costa Rica and
western Panama.

There is little doubt that L. e. epi-
nephelus gave rise to both populations. L.
e. epinephelus occurs in the lowlands be-
tween e. juvenalis and e. pseudocobella
and intergrades with juvenalis along the
lower slopes of the mountains of eastern
Costa Rica and western Panama and with
pseudocobella along the western and
eastern slopes of the Andes of Colombia
(Figs. 1, 2, 5d). The color pattern of indi-
viduals from the southern regions of Car-
tago and Punta Arenas Provinces of Costa
Rica are distinctly banded but have less
numerous black ventral marks and the
subcaudal area is red, with only a few
black flecks or spots laterally. A black
lateral tail stripe is evident above the
anus and continues to the tip of the tail.
These individuals appear to have about
75% of the juvenalis features. Individuals
from Boquete region of Chiriqui, Pan-
ama, have completely red or orange red
subcaudals without black edging, the
dorsal banding is less distinct posteriorly
and the black ventral marks vary from
almost absent to moderately dense. A
black lateral caudal stripe is present in all
specimens and occasionally begins in
front of the vent. These individuals are
almost perfect intermediates between
juvenalis and epinephelus. Similar
intermediates between epinephelus and
pseudocobella occur along the upper
slopes of the Andean foothills near
Pueblo Rico and San Antonio, Pacific
Colombia (also see Dunn, 1944).

SYSTEMATICS OF LIOPHIS EPINEPHELUS - Dixon

141

TABLE 3. VARIATION OF L. EPINEPHELUS SUBSPECIES FOR VENTRALS, SUBCAUDALS, AND TAIL/TOTAL LENGTH RATIOS

(LATTER CHARACTER BASED ON ADULTS ONLY).

VENTRALS
N Range M SE

e. juvenalis 51 139-152 145.0 0.4
e. pseudocobella 50 135-158 1466 0.6
e. epinephelus 103. 128-152 138.4 0.4
e. albiventris 93 141-165 150.7 0.5
e. fraseri 43 143-164 153.8 1.3
e. lamonae 48 141-157 149.7 0.7
e. opisthotaenia 32 142-152 147.0 0.5
e. bimaculatus 41 162-191 1749 1.0

Liophis e. epinephelus from the low-
lands of Panama and western Colombia
have moderately defined black and red
bands from the nape to midbody with the
posterior part of the dorsum obscurely
marked. The venter and subcaudals are
orange or reddish. The black lateral
stripe begins just anterior to the vent and
continues to the tip of the tail. This pat-
tern is typical of all samples of e. epi-
nephelus in lowland Panama, Colombia,
and extreme northwestern Ecuador. In
some areas of its range (Cali, Pamplona,
Colombia), e. epinephelus reaches an ele-
vation of 1,800 m, where it does not con-
tact another taxon of epinephelus.

Liophis e. epinephelus probably inter-
grades with L. e. lamonae along the up-
per slopes of the Rio Cauca and Rio
Magdalena valleys. I have examined a
single specimen from the state of Boyaca,
Colombia, that shares characters of both
taxa.

Color pattern morphs in the Andes of
Colombia and Ecuador are complicated
because there are three additional taxa
that are parapatric with each other and
infrequently intergrade with e. epi-
nephelus and e. pseudocobella. These
taxa, e. lamonae, e. fraseri, and e. bi-
maculatus (Figs. 5, 6), are similar in ap-
pearance. They tend to lose all indica-
tions of black dorsal banding in adults
except for an occasional pair of black
spots on the nape and have two to four

TAIL/TOTAL
LENGTH RATIOS (%)

Range M SE N Range M SE

SUBCAUDALS

45 51-62 560 04 25 189252 215 0.2
46 45-57 512 04 14 17.1-209 192 0.3

49-72 614 05 74 19.0260 23.1 0.2
83 52-70 603 05 63 19.1-242 220 02
43 51-75 666 08 17 185266 23.7 0.6

51-67 59.2 0.7 22 20.2241 222 02
29 53-78 64.7 09 17 202-256 236 0.3
34 4480 632 10 17 189240 213 03

well-defined black stripes (two lateral
and/or two dorsolateral). These stripes
may extend from midbody or the pos-
terior one-third of the body to the tip of
the tail, with the two dorsolateral stripes
pieteine as one on the median line of the
tail.

Both the two and four-lined patterns
may be found in the same population, but
usually one has a higher frequency. The
population of the Sabana region of
Bogota, Colombia and areas between
2,000-3,400 m are usually four-lined,
whereas populations to the east, south,
and west at elevations of 1,300-2,900 m
are usually two-lined. The black lateral
line is usually broad, occasionally occur-
ring on part of the second, usually all or
part of the third, and one-half to nearly all
of the fourth scale rows. The lateral black
line is usually bordered above by an ill-
defined yellowish line about one to two
scale rows wide, from midbody to the tip
of the tail. The middorsal coloration is
usually dark olive or dark brown, with or
without a black line edging the union
between the sixth and seventh (occa-
sionally the fifth and sixth) scale rows.

-The ventral color is highly variable from

almost immaculate yellow or salmon with
a few scattered black spots to almost
completely black with the lateral edges
salmon or yellow. The usual condition is
a yellow/black checkered pattern of
about equal density. The checkered ven-

142 Advances in Herpetology and Evolutionary Biology

Figure 5. Subspecies of Liophis epinephelus. A. Cotypes of L. bimaculatus (AMNH 17532, 17604) “Colombia.”
B. L. pseudocobella (BMMH 1910.7.11.25) Pueblo Rico, Colombia. C. Syntype of L. opisthotaenia (BMNH
1946.1.4.55) Merida, Venezuela. D. Intergrade between L. e. epinephelus and L. e. pseudocobella (BMNH
1910.7.11.26) slopes above Rio San Juan, Colombia. E. Holotype, L. taeniurus juvenalis (ANSP 3687) San Jose,
Costa Rica. F. Holotype, L. epinephelus (ANSP 3688) Truando, Colombia.

SYSTEMATICS OF LIOPHIS EPINEPHELUS - Dixon 143

Figure 6. Subspecies of L. epinephelus. A and C represent L. e. bimaculatus; B and D, L. e. albiventris; E and F,
L. e. fraseri. A. Holotype of L. bipraeocularis (BMNH 1946.1.4.93) Facatativa, Colombia. B. Syntype of Zamensis
ater (BMNH 1946.1.21.52) Biscra, Algeria (in error). C. Holotype of L. e. ecuadorensis (MRHN 3267) “Ecuador.”
D. Liophis e. albiventris (KU 164215) 9.5 km NW Nono, Ecuador. E. Holotype of Liophis bimaculatus lamonae
(MLS no number, photo courtesy of Charles Myers) Sonson, Colombia. F. Holotype of Liophis fraseri (BMNH
1946.1.6.63, photo courtesy of British Museum of Natural History) “West Ecuador.”

144

ter is normal for L. e. fraseri and L. e.
lamonae. The two-lined patter is preva-
lent between elevations of 1,600-2,200 m
in the eastern, central, and westem
Colombian Andes and the four-lined pat-
tern between 2,200-3,400 m in the east-
ern Andes of Colombia and the central
Andes of Ecuador and Peru.

Liophis e. bimaculatus is primarily a
high altitude taxon (mean elevation of
2,600 m, Figs 2, 5a) with 162 to 191 (x
174.9) ventrals, whereas L. e. fraseri
usually occurs at lower elevations (mean
elevation of 1,900 m) with 141 to 164 (x

. albiventris
eee es
30
20
uy L.e. epinephelus
8 a ee
10
0 Ie a L.e. fraseri
10
2 x Oe aes L.e. juvenalis
=)
= 10
>
Z 4 me NS L.e. lamonae
10
A we L.e. opisthotaenia
__ ole
10
| = Aa L.e. pseudocobella
10
al si eee L.e. bimaculatus
SEES a ES

17 18 19 20 21 22 23 24 25 26 27
MAXILLARY TEETH

Figure 7. Variation in the numbers of maxillary teeth
in L. epinephelus. The irregular black line represents
the number of individuals in a particular subspecies
with a particular number of maxillary teeth. The hori-
zontal line represents the total range of variation, verti-
cal line, the mean; open rectangle one standard devia-
tion on either side of the mean; black rectangle two
standard errors on either side of the mean. Sample
sizes are given in Table 2 and Figure 5.

Advances in Herpetology and Evolutionary Biology

152.8) ventrals. Liophis e. bimaculatus, e.
lamonae, and e. fraseri tend to have 21 to
30% of their populations with seven
supralabials and 76 to 80% with nine or
less infralabials, respectively. In addi-
tion, approximately 20% of L. e. bi-
maculatus populations have two pre-
oculars on each side. The latter taxon
appears to have the highest number of
ventrals in Ecuador with a slight de-
crease both north and south of the Ibarra
region. On the other hand, the number of
ventrals of L. e. lamonae averages about
150 in Colombia and 149 in northem
Ecuador, while e. fraseri averages 155
ventrals in southern Ecuador, and de-
creases to a mean of 151 in central Peru
(Fig. 8).

The nominate taxon intergrades with
L. e. albiventris in the lowlands of
southwestern Colombia and northwest-
ern Ecuador. Two specimens from the
lowlands of northwestern Ecuador have
the dorsal and ventral pattern of L. e. epi-
nephelus but with the number of ventrals
normally found in L. e. albiventris. Typi-
cally patterned individuals of L. e. albi-
ventris (Fig. 6d) occur within 70 km of
these two specimens, suggesting that the
zone of intergradation may be narrow.
The majority of the specimens of L. e.
albiventris (91%) occur above 500 m ele-
vation, and most are found between 1,000
and 2,000 m along the westem face of the
Andes in Ecuador (Fig. 3). Liophis e.
albiventris has an immaculate cream to
yellow venter and subcaudal area. Only
six of 93 specimens had any indication of
black marks on the ventrals.

The dorsum of adult L. e. albiventris is
leaf green with none to few obscure black
marks on the nape and/or anterolaterally
on the trunk. Juveniles (150 to 190 mm
total length) are similar but have a black
nape band or two black nape spots fol-
lowed by a narrow white or cream collar.
The light collar and dark nape spots dis-
appear as the individual reaches 230 to
260 mm in total length. Young L. e.
fraseri and e. lamonae have similar nape
markings that fade almost as rapidly, but

SYSTEMATICS OF LIOPHIS EPINEPHELUS - Dixon

145

@oervrnoooo wu uy
N

N==0000

SSeS

VENTRALS

SUBCAUDALS

Figure 8. Variation in the numbers of ventrals and subcaudals in L. epinephelus. Numbers and letters in left
hand column refer to subspecies, latitude, and geographic location (see Table 2, Fig. 3). Horizontal line = range;
vertical bar = mean; black rectangle = two SE on either side of mean; open rectangle = one SD on either side of

mean.

occasional adults retain some indication
of the juvenile pattern. All L. e. albiven-
tris have a black lateral posterior stripe
that continues to the tip of the tail. Infre-
quently, young and adults have black
paravertebral posterior stripes that con-
verge into one dorsomedian tail stripe.
Juveniles and hatchlings occasionally
have a diffuse light stripe immediately
above the black lateral posterior stripe,
resembling the typical pattern seen in L.
e. fraseri and e. lamonae except for the
immaculate venter.

Liophis e. albiventris and possibly L. e.

lamonae seem to merge with populations -

of L. e. bimaculatus at elevations be-
tween 2,600-2,800 m in northern Ecua-
dor. The number of ventrals increase
from an average of 150 between 1,000-
2,000 m to 156 between 2,000-2,600 m
and 174 ventrals above 2,800 m. Liophis

e. albiventris also intergrades with L. e.
fraseri in the vicinity of Celica, southern
Ecuador, where the latter taxon switches
from having an eastern slope distribution
to a western one (Fig. 3). An individual
from near Celica, 1,900 m, has a color
pattern similar to most L. e. fraseri and
the squamation of L. e. albiventris.
Specimens from Alamor, 1,100 m, some
10 km northwest of Celica, are typical L.
e. albiventris. The Alamor locality seems
to be the southern limit of the distribu-
tion of L. e. albiventris, whereas L. e.
fraseri continues southward along the
middle slopes of the westem face of the
Andes to above Trujillo, Peru, switching
again to the eastern slopes and continu-
ing southward to Ayacucho, Peru. The
two specimens of L. e. fraseri from cen-
tral Peru are allopatric from other
samples by 325 km. These two differ in

146

having almost black venters, white sub-
caudals, few ventrals and subcaudals
(Sample 16 of Fig. 8). The dorsal body
pattern resembles L. williamsi from
Venezuela, and one of the two specimens
has a similar head pattern. However, L.
williamsi has no scale row reductions
while L. fraseri has a 17-17-15 pattern.
With additional! material, the central Peru
population may prove to represent a new
taxon. .

The only subspecies of L. epinephelus
that appears to be allopatric is the Vene-
zuelan L. e. opisthotaenia. This popula-
tion occurs from the upper drainage of
the Rro Chama, 2,000 m, just east of
Merida, west-southwest to the vicinity of
Las Delicias, 2,000 m, Paramo de Tama,
on the Colombian border. Although there
is no evidence of intergradation of this
race with adjacent subspecies, it may oc-
cur because the subspecies crosses the
Tachira Depression that separates two
major Andean formations.

Most Liophis e. opisthotaenia (Fig. 5)
have immaculate venters (26 of 32). Six
show scattered to many black marks on
the lateral edges of the ventrals. The dor-
sum is olive green, usually with two ob-
scure black nape spots and occasionally
dorsolateral dark streaking anteriorly.
The juvenile pattern is similar to that of
other L. epinephelus. The black postocu-
lar bar is usually present, similar to most
individuals of L. e. bimaculatus. The
basic body pattern is very similar to that
of L. e. bimaculatus, L. e. fraseri, and L.
e. lamonae. However, occasional indi-
viduals resemble L. e. albiventris, except
for the slightly longer tail and usually 21
or fewer maxillary teeth (Fig. 7).

KEY TO THE SUBSPECIES OF LIOPHIS
EPINEPHELUS

The following key is provided to aid in
the identification of the taxa of this dif-
ficult species. The characters used are
those that best exemplify the differences
between the subspecies, but the geo-

Advances in Herpetology and Evolutionary Biology

graphic area of the samples is also
important.

1. Lateral black tail stripe absent, dorsal and ven-
tral surfaces of body and tail red and black
banded .cccc.8 ccd. ee ee 2

— Lateral black tail stripe present, dorsum may be
banded, spotted, flecked, or almost unicolor,
belly checkered with black or not

2. Caudals usually 54 or less (Colombia) ...-...

Pera evade Senna siesaenme aes Soar A orn 6 e. pseudocobella

— Caudals usually 54 or more (Costa Rica) .....

ae Go ASE EO aS SOE Hod ba'o e. juvenalis

3. Ventrals usually 165 or less ............... 4

— Ventrals usually 165 or more (high Andes,
Venezuela to Peru) .......... e. bimaculatus

4. Anterior half of dorsum with or without black
flecks, streaks, or dark spots, never banded in
adults, ventrals 143 or more

— Anterior fourth of dorsum banded with black or
brownish-black bands, venter immaculate
white, yellow or reddish, ventrals usually 143
or less (Panama to Ecuador)... e. epinephelus

5. Venter usually immaculate white, yellow, or
pinkish

— Venter yellow and black checkered

Dorsum leaf green, posterior dorsolateral black

stripe usually absent (coastal and foothill zones

of western Ecuador) ........... e. albiventris

— Dorsum olive, olive brown, or grayish brown,
posterior dorsolateral black stripe usually pres-
ent (Paramo de Tama, east to Trujillo, Vene-
ZUElay Gh cae San ae ee eee e. opisthotaenia

7. Ventrals range from 141 to 156 (x 150), subcau-
dals 51 to 67 (x 59) (Andean slopes of central
Colombia to central Ecuador) ....e. lamonae

— Ventrals range from 151 to 164 (x 157), subcau-
dals 61 to 76 (x 67) (Andean slopes of southern
Ecuador to central Peru) .......... e. fraseri

>

ACKNOWLEDGMENTS

My interest in South American herpe-
tology was stimulated by the astute
Ernest E. Williams, who on several occa-
sions presented a view different from my
own, and to my dismay, more frequently
correct. His expertise, thoughtfulness,
and willingness to help has been a rock
upon which herpetologists have obtained
a better understanding of evolutionary
principles. Thanks, Emest, for the more
than 25 years of correspondence and
friendship.

I heartedly thank those curators who
willfully, but sometimes grudgingly, al-
lowed me access to their specimens. I

SYSTEMATICS OF LIOPHIS EPINEPHELUS + Dixon

offer my condolences to those curators
who did not, for their Liophis specimens
may forever remain misidentified. The
collections examined, their acronyms and
respective curators are: American Mu-
seum of Natural History (AMNH), R. G.
Zweifel, C. M. Myers; Academy of Na-
tural Sciences, Philadelphia (ANSP), E.
Malnate; British Museum (Natural His-
tory) (BMNH), A. G. C. Grandison, A. F.
Stimson; California Academy of Sciences
(CAS), A. E. Leviton, R. C. Drewes; Uni-
versity of Southern California Costa Rica
Series (CRE), J. M. Savage; Collection of
Vertebrates, Universidad Los Andes,
Merida (UAM), J. Pefaur; Field Museum
of Natural History (FMNH), R. Inger, H.
Marx; private collection of Jay M. Savage
(JMS); University of Kansas Museum of
Natural History (KU), W. E. Duellman;
Los Angeles County Natural History
Museum (LACM), J. W. Wright; Louisi-
ana State University Museum of Zoology
(LSUMZ), D. A. Rossman; Museo de His-
toria Natural La Salle, Caracas (MHNLS),
A. Paolillo; Harvard University Museum
of Comparative Zoology (MCZ), E. E.
Williams; Museum d’Histoire Naturelle,
Geneva (MHNG), V. Mahnert; Museum
d Histoire Naturelle, Paris (MNHP), R.
Roux-Esteve; Institut Royal des Sciences
Naturelles de Belgique (MRHN), G. F.
de Witte, J. P. Gosse; University of Cali-
fornia Museum of Vertebrate Zoology
(MVZ), D. B. Wake; Universidade de Sao
Paulo Museum de Zoologia (MZUSP), P.
E. Vanzolini; University of Texas Natural
History Museum (TNHC), R. F. Martin;
University of Illinois Museum of Natural
History (UIMNH), D. F. Hoffmeister;
University of Michigan Museum of Zo-
ology (UMMZ), A. G. Kluge, R. A. Nuss-
baum; National Museum of Natural His-
tory (USNM), W. R. Heyer, G. Zug; and

Federico Medem Collection (WWL), W. -

W. Lamar.

I thank Michael McCoid, Jack Sites,
Jr., and Robert Dean for reading all or
parts of the manuscript. I especially
thank the administration of Texas A&M
University for allowing me the time and

147

funds to examine some of the type ma-
terial in European Museums. I especially
acknowledge the perseverance of my
wife, Mary, who, for the first time, found
out what it was like to take squamative
data from snakes.

APPENDIX: SPECIMENS EXAMINED

Liophis epinephelus albiventris. ECUADOR:
doubtful or no specific localities, AMNH 17498,
17634, 49104, 35807, UIMNH 63524, UMMZ
88970, USNM (GOV) 7316-17, 7291, BMNH
1946.1.21.51-53, MHNG 1367.47; Alamor, AMNH
18320, 22230, 22095, 22231, 22235, 56168; AI-
luriguin, USNM (GOV) 7429-30; Balzapamba,
UMMZ 88959; Balzar, USNM (GOV) 7330, 7423;
Banos, UMMZ 88962-63; Camino de Gualea,
USNM _ un-numbered; Chaucha Valley, UMMZ
63121; 8 km W Chiriboga, USNM (JAP) 4109-10; 4
km W Chiriboga, KU 142807; 7.7 km E Chiriboga,
KU 142805; 3.7 km E. Dos Rios, KU 142807; 14 km
NE Dos Rios, KU 164214; Guayaquil, USNM
62792; Hacienda Carcelén, 15 km NE Quito,
USNM (GOV) 7341-42; Intac, BMNH 78.1.25.52;
Lita, USNM (GOV) 7425; Llambo, USNM (no num-
ber); Los Rios, Hda. Clementina, MHNG 1337.36;
10 km SE Machala, AMNH 113021; Mapoto,
UMMZ 88960-61; Milligalli, MCZ 8394-95, USNM
(GOV) 7424; Milpe, USNM (JAP) 9024; Mindo,
BMNH 1916.5.23.3-4, USNM (GOV) 7426-27; 4 km
N Mindo, USNM (JAP) 8278; 1 km E Mindo, USNM
(JAP 8273; 1.5 km E Nanegal Chico, USNM (JAP)
6285; 9.5 km NE Nono, KU 164215; 13 km NW
Nono, KU 158534; Olalla, BMNH 82.7.24.2; near
Oyacachi, USNM (GOV) 7396; below Pacto USNM
(GOV) 7422; Paramba, BMNH 98.4.28.68; 56 km N
Quevedo, KU 152605; Quito, BMNH 72.2.26.7;
near mouth of Rio Aguas Claras, USNM (GOV)
7333; 24 km S Rio Baba, UIMNH 92256; Rio
Blanco, 20 km from Mindo, USNM (JAP) 7331 Rio
Mindo, USNM (JAP) 7428; Rio Mulaule, USNM
(GOV) 7299-300; Rio Saboya, BMNH 1940.2.20.30;
Rio Silanchi, USNM (GOV) 7819; San José de
Chimbo, AMNH 17498, 17634; Santo Domingo
AMNH 27144, BMNH 1918.1.15.3; 5-10 km SSW
Santo Domingo de los Colorados, AMNH 110588,
UIMNH 92256, USNM (GOV) 6006; between km
markers 38/29, NNW Santo Domingo de los Colo-
rados, USNM (JAP) 1927, Tandapi, KU 112272-73,
121319-21; 9 km E Tandapi, AMNH 113020; Tan-
dayapa, USNM (GOV) 7329: Tanti, MCZ 8400;
Tumbaco, USNM (GOV) 7332.

Liophis epinephelus bimaculatus. COLOMBIA:
no specific locality, AMNH 27615, 17532, 17604;
Bogota, AMNH 7648, 14029-31, 14035, 24227,
24232, 24251, 67151, 67157-58, 91813, CAS-SU
8274, BMNH 1919.3.6.21, MCZ 17133-34, MVZ
71534, USNM 95180; Bolivar, Vereda de San Juan

148

MZUSP 6131; Facatativa, BMNH 1946.1.4.93; La
Calera, MVZ 71535-36; La Uvita, MCZ 66382;
Moscopan, UMMZ 121039; 23 km E Pacho, AMNH
104679: Pueblo Viejo, MCZ 6551; Usaquén, LACM
28856; Villavicencio, (in error?) BMNH 1915.3.
11.3; Vitelma, AMNH 67152-56. ECUADOR: no
specific locality, MRHN 3267; Ibarra, BMNH 96.4.
28.74-75; Otovalo, KU 135182. PERU: Cajamarca,
UMMZ 59846; Huagal, UMMZ 59147. VENE-
ZUELA: 17 km N. Guaraque, UAM 1160.
Liophis epinephelus epinephelus. COLOMBIA:
no specific locality, AMNH 177556-58; Andagoya,
BMNH 1915.10.21.13-14, 1916.4.25.7, MCZ 32733-
35, UMMZ 48409, 121050, 121057; 100 km N
Bogota, MHNG 1367.55; Buenaventura, USNM
151724, 154037; 13 km from Buenaventura, FMNH
165441, 165451, 165453, 165564; Cali, USNM
151749; Cano Docordo, CAS 119599; Chigorodo,
AMNH 97276; Cisneros, FMNH 43742-44:; Con-
doto, BMNH 1913.11.12.48, 1914.5.21.43-45; Im-
bilt, FMNH 166024; Pueblo Rico, FMNH 54940,
54950-56; Rio San Juan, USNM 72364, 72368,
159487-88: Rio Sinu, UMMZ 126722; San Sebas-
tian de Rabago, CAS 113880; Santa Rosa de Osa,
AMNH 35784; Sierra Negra, USNM 117505; Tru-
ando, ANSP 3688; near Turbo, AMNH 94343-44,
USNM 158946; Vista Nieve, UMMZ 63776 (3).
ECUADOR: Borbon, BMNH 1902.5.27.19; near
Esmeraldas, USNM (OVS) 981. PANAMA: no
specific locality, FMNH 6113, 31211-13, 31721,
31725, 31728, 83546, 154513-14, 154527, USNM
13565; UMMZ 90338; Darien, AMNH 36205; Agua
Clara, MCZ 45372-73; Albrook Air Force Base,
TNHC 23713-15; Barro Colorado Island, AMNH
80037, 77575, FMNH 178599, KU 75707-08, 80595,
LACM 9278-79, MCZ 20593, UMMZ 63745, 76022,
124160-61, 124186, USNM 203834; near Cam-
arones, KU 110712; Camp Chagres, KU 75709;
Cana, USNM 50118-20; Cana Mines, UMMZ 83157
(6), 117703; Cerro la Campana, KU 75710; Chagres
River, FMNH 154489, MCZ 34378, 37107-11,
UMMZ 76721 (3), 83145; El Valle Rio Anton,
AMNH 76018, 108706, 108968, FMNH 68074, KU
110720-21; Empire, KU 110717, USNM 54228,
60028; Esperanza Ridge, MCZ 42721; Fort Clayton,
KU 110714-16, UIMNH 41737; Fort Frank S Todd,
USNM 140690-91; Fort Gulick, KU 110719; Fort
Kobbe, CAS 85222; Fort Randolph, MCZ 20592;
Fort Sherman, MCZ 22239, 22242, USNM 192283;
Gamboa, AMNH 71672, 73323, KU 110713, MCZ
23972; Gatuncillo and Chagres fork, MCZ 37872;
Las Cumbres, KU 110710-11, 110724; near Limon,
MCZ 23984; Madden Dam, MCZ 38231, UMMZ
76022, 76724 (2); Minas de Cana, MCZ 42752;
Pihuila-Iucuti fork of Rio Tuira, MCZ 37119-20;
upper Rio Chiriqui, AMNH_ 114320-22; Rio
Grande-Rio Cocle, MCZ 37879; Rio Sardinillo, KU
110718; Rio Tiura at Rio Mono, KU 110725; San
Pablo, MCZ 3930-31, USNM 120826-27; Three

Advances in Herpetology and Evolutionary Biology

Falls Creek, AMNH 65300; Venado Beach, USNM
193374, MCZ 38215.

Liophis epinephelus fraseri. ECUADOR: “West
Ecuador,” BMNH 1946.1.6.63; 10 km N Celica,
MCZ 93590; Loja, BMNH 1935.11.3.67-68, KU
125415; 2 km E Loja, KU 121322-23; 12.2 km S
Loja, KU 141269-87, 141289; 6.5 km N Loja, KU
142809-11 7.3 km N Loja, KU 142808; Rio Piuntza,
KU 147199; San Ramon, USNM (GOV) 7328.
PERU: 15 km E Canchaque, LSUMZ 32554—59;
Chanchamayo, FMNH 40639; MHuanacabaniba,
MCZ 17398; 45 km NE Tarma, MCZ 45902;
Trujillo, USNM_ 192303 (in error?); just below
Yuracyacu, on Tambo-Valle del Apurimac path,
LSUMZ 27138.

Liophis epinephelus juvenalis—COSTA RICA:
no specific locality, CRE 2996; vic. Alajuelita, CRE
8483-85; Barba, AMNH 17281; Capellades, CRE
7057; Cartago, CRE 6401, UMMZ 131415; Cer-
vantes de Cartago, CRE 6393; Cinchona de Sara-
piqui, CRE 6427, KU 31899, 31901, 35519-21,
63826; Concordia, KU 63827; between Curridabat
and San Antonio de Desamparados, CRE 2725; El
Angel, UMMZ 136999; E] Silencio, CRE 3054-55;
E] Tablazo de la Carpintera, CRE 3031; Finca El
Helechales, CRE 8278; Guadalupe, KU 31900;
Guayapiles, UMMZ 83181; Isla Bonita, FMNH
10124344, 188808, KU 25750-52: La Estrella,
MCZ 28072; La Gloria, CRE 3012; La Palma, CRE
243, Limon, AMNH 5304; Monteverde, CRE 3178;
Navarro, MCZ 15297-99, UMMZ 74302; 1.5 km
NW Pastora, CRE 34; Puerto Veijo, CRE 3030 (in
error?); Puriscal Centro, Santiago, CRE 6335; Rama
sur Rio Las Vueltas, KU 103885; Rio Claro, JMS
1763; Rio Coton, KU 100637; Rio Torres in Gua-
delupe, CRE 908, 7170, KU 31901; Sabanilla, JMS
1160; San Antonio de Belem, MCZ 32041; San Ig-
nacio de Acosta, CRE 3015 (2); San Isidro de Coro-
nado, MCZ 55116, FMNH 37231; San Isidro del
General, UMMZ 117729; San Jose, AMNH 17298,
17377, ANSP 3687, CRE 6336, 7003; San Pedro, in
San Jose, CRE 2956, 6421, 6428-29, UMMZ
125508, 131414, 131416-20; San Pedro de Montes
de Oca, CRE 6331; San Ramon de Tres Rios,
LSUMZ 15655; 4.5 km S San Vito de Java, LACM
114075; 6 km S San Vito de Java, LACM 114074;
Tapanti, CRE 6314; vic. Tapanti, 0.8 km from
Puente de Rio Grande, CRE 3246; Tres Rios, CRE
2946; Turrialba, MCZ 55025; Varablanca, KU
125479; near Villa Quesada, CRE 707; Volcan Poas,
east slope, UMMZ 117828 (3). PANAMA: Santa
Clara, AMNH 80085.

Liophis epinephelus lamonae. COLOMBIA: no
specific locality, MHNG 840.59; Aguadita, AMNH
35241; Bogota, AMNH 24236 (in error ?); Chusaca,
KU 124939; Las Palmas, FMNH 63759; near Man-
zanares, WWL 747-48, 780-82, 802-06, 989-90,
1001, 1006; Medellin, AMNH 35492, 35721; Pen-
silvania, AMNH 35250; Rio San Joaquin, FMNH

SYSTEMATICS OF LIOPHIS EPINEPHELUS + Dixon

54957; San Agustin, FMNH 69666; San Francisco,
AMNH 106660; San Pedro, AMNH 35745; Sinar,
MCZ 150201, MZUSP 6128; Villavicencio, BMNH
1915.3.11.2. ECUADOR: no specific locality,
AMNH 2793, 35980, 35897-99, BMNH 1946.1.6.63;
Baeza, USNM (JAP) 5758; Cayambe, UIMNH
82483 (in error ?); La Alegria, USNM (JAP) 4501; La
Bonita, USNM (JAP) 4502-05, 4917; Mt. Tungura-
hua, FMNH 36623, KU 121329; ca. Sevilla de Oro,
USNM (JAP) 6691.

Liophis epinephelus opisthotaenia. VENE-
ZUELA: Betania, Paramo de Tama, MCZ 112415;
Chama, BMNH 1905.6.30.59-60; Mérida, BMNH
1905.5.31.56-58, 1908.5.29.146-147, 1912.11.1.73—
76, 1946.1.4.54-55, UAM 602, 771, 1046-48, 1050-
51, 1162, 1697, 1710, 2318; Misisi, 14 km E Trujillo,
USNM 162827; Rio Albarregas, BMNH 1903.6.30.
37-38, USNM 162822; Rio Albarregas y Milla,
AMNH 13418-19; San Cristobal, MHNLS 1096.

Liophis epinephelus pseudocobella. COLOM-
BIA: no specific locality, AMNH 35263-65, 35267—
76; 10 km NW Andes, LACM 72760; Barbacoas,
FMNH 54938; Bogota, MCZ 19210-11 (3) (in error);
Jerico, AMNH 35246, FMNH 27015-16, UMMZ
84088, 78280 (2), USNM 86820; Medellin, AMNH
39539-44, 3555356, 35560-65, 35623, 35627-28,
35639, 35672, 4492; Pasto, AMNH 91814; Popayan,
FMNH 54931-35, 54985-95, MHNG 1532.13; 15
km N Popayan, MVZ 68705; Pueblo Rico, BMNH
1910.7.1.25; Rio Cofre, KU 140411; 4 km NW San
Antonio, MVZ 68697-703; ca. 5 km NW San Jose,
AMNH _ 106655; Sonson, FMNH 63780, MCZ
21991.

Liophis epinephelus epinephelus x e. juvenalis.
PANAMA: no specific locality, AMNH 75637;
Boquete, CAS 71340; north of Boquete, CAS 78922—
36; Chiriqui, BMNH 93.11.22.5-10; El Hato,
FMNH 68072-73, 154479-80; El Volcan, AMNH
63596, 75633, KU 75711, 110722-23.

Liophis epinephelus epinephelus x e. lamonae.
COLOMBIA: Junin, LACM 36780.

Liophis epinephelus bimaculatus x e. pseudo-
cobella. COLOMBIA: 13 km SE Pedrancha, AMNH
106659.

Liophis epinephelus epinephelus x e. pseudo-
cobella. COLOMBIA: slopes above Rio San Juan,
BMNH 1910.7.11.26.

149

LITERATURE CITED

AMARAL, A. DO. 1931. Studies of neotropical ophidia
XXIII. Additional notes on Colombian Snakes.
Bul. Antivenin Inst. Am., 4: 85-89.

BOULENGER, G. A. 1894. Catalogue of the snakes in
the British Museum (Natural History). Vol. 2.
London, Trustees of Museum, 382 pp.

___. 1903.Descriptions of new snakes in the col-
lection of the British Museum. Ann. Mag. Nat.
Hist., 12(7): 350-354.

__. 1908. Descriptions of new South American
reptiles. Ann. Mag. Nat. Hist., 1(8): 111-115.

Cope, E. D. 1862. Synopsis of the species of Holco-
sus and Ameiva, with diagnosis of new West
Indian and South American Colubridae. Proc.
Acad. Nat. Sci. Phila., 13: 60-82.

__. 1868. An examination of the Reptilia and Ba-
trachia obtained by the Orton Expedition to
Ecuador and the upper Amazon, with notes on
other species. Proc. Acad. Nat. Sci., 20: 96-140.

____.. 1899. Contributions to the herpetology of New
Granada and Argentina. Philadelphia Mus. Sci.
Bull., (1): 1-22.

DuNN, E. R. 1937. New or unnamed snakes from
Costa Rica. Copeia, 1937: 213-215.

____. 1944. A revision of the Colombian snakes of
the genera Leimadophis, Liophis, Lygophis,
Rhadinaea, and Pliocercus, with a note on
Colombian Coniophanes. Caldasia, 11(10):
479-495.

GUNTHER, A. 1872. Seventh account of new species
of snakes in the collection of the British Mu-
seum. Ann. Mag. Nat. Hist., 9(4): 13-37.

JAN, G. 1863. Enumerazione sistematica degli ofidi
appartenenti al gruppo Coronellidae. Arch.
Zool. Anat. Fisiol., 2(2): 213-330.

LAURENT, R. 1949. Notes sur quelquies reptile ap-
partenant a la collection de |’Institute Royal
des Sciences Naturelles de Belgique. Bull.
Inst. Roy. Sci. Nat. Belgium, 25(9): 1-20.

Peracca, M. G. 1914. Reptiles et batrachiens de
Colombia. In O. Fuhrmann and E. Mayor
1914, Voyage d’Exploration Scientifique en
Colombie. Mem. Soc. Neuchatel, 5: 1—1090.

Partial Cyclopia and Monorhinia in Turtles

A. D’A. BELLAIRS!

ABSTRACT. Three albinoid embryos of the logger-
head turtle (Caretta caretta) with partial cyclopia
and one with a single nostril are described. The two
specimens sectioned show many other abnormali-
ties of the head including virtual absence of the
midline components of the anterior chondro-
cranium.

INTRODUCTION

The congenital malformation of cyclo-
pia is well known in man and other
mammals (Schwalbe and Josephy, ca.
1913; Willis, 1958) and has also been
reported in birds (Romanoff, 1972), in
reptiles (Ewert, 1979), and in amphibians
and fish (Adelmann, 1936). It is char-
acterized by the fusion or partial fusion of
the eyes in the midline and by abnormal-
ities of the brain and craniofacial struc-
tures. In amniotes the nose is often repre-
sented by a proboscis-like structure
above the single or partially fused eyes.
The malformation has been produced
experimentally by various chemical and
physical procedures (Johnston et al.,
1977), but may also occur spontaneously
in the absence of any obvious environ-
mental cause.

Among reptiles, cyclopia appears only
to have been reported in chelonians
(Ewert, 1979), though Bellairs (1965) de-
scribed a Siamese twin embryo of
Lacerta agilis showing a _ kind of
““pseudocyclopia’ caused by fusion of the
two heads creating a third, single eye in
the midline. Although Ewert figured

1$t. Mary’s Hospital Medical School, London,
W:2,, UK

cyclopic near-hatching embryos _ of
Chrysemys picta and Pseudemys scripta
elegans, the anatomy of the condition
remains undescribed. The present con-
tribution seems appropriate in this col-
lection of essays in honor of Dr. Emest E.
Williams. who has made _ important
studies on the normal anatomy of che-
lonians.

MATERIALS AND METHODS

The four specimens described are very
late embryos of the Atlantic loggerhead
turtle, Caretta caretta caretta, with yolk
sacs attached. Three specimens show
partial cyclopia, while the fourth shows a
malformation which I have termed
monorhinia (single nose), though I know
of no precedent for this term. These spe-
cimens, kindly sent me by Dr. P. C. H.
Pritchard of the Florida Audubon Soci-
ety, were obtained in the course of a
study of the effects of environmental
conditions on the hatchability of turtle
eggs (McGehee, 1979). Both unsuitable
temperatures and mechanical disturb-
ance during incubation are known to
have deleterious effects upon develop-
ment (Ewert, 1979), and were among the
variables investigated. It is by no means
certain, however, that the malformations
were caused by the experiments, since
the majority of eggs were unaffected by
them, and abnormal individuals also oc-
cur in natural nests. Specimens B and C
were preserved intact while the heads of
A and D were serially sectioned and
stained with Masson and Casson’s triple

PARTIAL CYCLOPIA AND MONORHINIA IN TURTLES - Bellairs

stains. Some normal material was also
available for comparison.

EXTERNAL APPEARANCE
Figures 1, 2

All four specimens were albinoid.
None of them possessed an egg-caruncle,
a defect which in itself might have pre-
vented hatching. Apart from showing
duplication of some of the epidermal
scutes (notably the vertebrals in speci-
mens B and C) and the reduction of pig-
mentation, these turtles show no obvious
abnormalities outside the head region.

Specimen A. Carapace-length 36 mm.
Egg turned through 360° four hours after
laying in McGehee’s experiments.

There is a single slitlike orbital open-
ing within which the eyes are barely
visible externally. Above this opening is
a well-developed proboscis 2.5 mm long,
bearing a single narial opening at its tip.
The upper jaw is truncated, and the ap-
parently normal lower jaw projects for-
ward well beyond it. A small portion of
the hindbrain is herniated from the back
of the head.

Specimen B. Carapace-length 36 mm.
Egg incubated at 32°C. instead of the
optimum 30°. The eyes are not markedly
reduced in size but are fused or con-
tiguous medially. The proboscis is turned
back over the head. The upper jaw is
truncated.

Specimen C. Carapace-length 40 mm.
Egg turned through 360° 48 hours after
laying. As in B, the eyes are fused or con-
tiguous medially. The lower jaw projects
anteriorly and dorsally, covering the
truncated upper jaw. The proboscis is
directed forward, as in A.

Specimen D. Carapace-length 40 mm. -

Egg turned through 360°/40 hours after
laying. The eyes are not markedly re-
duced and hardly closer together than
normal. The upper jaw is shortened,
though not to the same extent as in the
previous specimens. There is no probos-

151

cis, but there is a single median nostril
between the eyes.

MICROSCOPIC FINDINGS

Specimen A. Study of sections shows
that the proboscis is traversed by a single
nasal tube which widens out posteriorly,
forming a single blind sac which has no
connection with the mouth or pharynx. A
similar absence of the choanal opening is
mentioned by Schwalbe and Josephy (ca.
1913) in cases of mammalian cyclopia.
The nasal tube is lined anteriorly by
pseudostratified squamous epithelium.
Posteriorly the nasal sac contains areas of
olfactory epithelium with associated
Bowman’s glands. The nasal tube and sac
are partly invested by a tube of cartilage
which represents the nasal capsule (Fig.
3A-C). A small process projects dorsally
from the floor of the capsule, as if it were
an attempt to form a nasal septum (Fig.
3C); however, it shows little resemblance
to the normal structure.

The bones are difficult to interpret.
The bones within the base of the probos-
cis and around the nasal tube and sac
might be the prefrontals as labelled here
(Fig. 3); these are normally the most an-
terior bones of the cranial roof since there
are no nasals in Recent cryptodiran
Testudines. The identity of the more
lateral bones which extend ventral to the
partly fused eyes (Fig. 3C) is problemati-
cal; their relations are very different from
those of normal frontals. An alternative
possibility is that the bones around the
nasal tube and sac are abnormally large
frontals which have grown anteriorly and
ventrally, while the problematical bones
are prefrontals which have been dis-
placed ventrolaterally, still retaining
their normal contact with the maxillae.
The maxillae themselves are partly fused
in the midline and the premaxillae ap-
pear to be absent, as in most cases of
mammalian cyclopia (Schwalbe and
Josephy, ca. 1913).

Olfactory nerves arise from the pos-

Figure 1.
right.

terior region of the nasal sac (Fig. 4A),
pass back as a large bundle beneath the
forebrain and appear to end _ blindly
among the eye muscles. These nerves
arise from neuroblasts in the olfactory
epithelium, and it would seem that in
this specimen they have grown back but
have failed to enter the olfactory bulb in
the normal fashion. The brain itself is
grossly malformed and difficult to inter-
pret. There appears, however, to be only
one olfactory bulb (possibly fused bulbs),
and the cerebral hemispheres are more or
less fused. The forebrain is supported by
wings of cartilage which probably repre-
sent the planum supraseptale. There is
no indication of the interorbital septum,
and no parts of the trabeculae can be
identified.

The eyes are considerably smaller than

Advances in Herpetology and Evolutionary Biology

Abnormal late embryos of Caretta. A, B, and C (partially cyclopic) and D (with monorhinia) from left to

normal and are fused medially (Figs. 3C
4A). Each eye consists of a much folded
layer of retinal tissue surrounded by a
layer of pigment, probably of retinal ori-
gin. It is interesting that no lenses are
present. The lens is normally induced by
the optic vesicle or cup, and it would
seem that in this specimen there was a
failure of induction, either because of the
small size of the vesicles, or their failure
to invaginate. This last possibility is sug-
gested by the folded appearance of the
retinae.

The two eyes are surrounded by a
single capsule of rather thick scleral car-
tilage. Between the eyes and the probos-
cis is a flattened gland which discharges
anteriorly through a long duct beneath a
“palpebral” fold of tissue dorsal to the
eyes. Probably this represents either an

PARTIAL CYCLOPIA AND MONORHINIA IN TURTLES - Bellairs

153

Figure 2. Abnormal late embryos of Caretta. A (left) and B are at top, C (left) and D at bottom. A, B, and C are
partially cyclopic, proboscis is turned back in B. D has monorhinia.

Harderian or a lacrimal gland. Massive
but disorganized eye muscles are at-
tached to the scleral cartilage.

A remarkable feature is the presence of
a third eyelike structure which is con-
nected with the two fused eyes by a nar-
row stalk which issues from a cleft in the
median ventral aspect of the scleral car-
tilage (Figs. 3C, 4A). Anteriorly the stalk
is pigmented; posteriorly it expands into

a dilated vesicle composed of apparently
retinal tissue. The vesicle extends cau-
dally for some distance and ends blindly

‘without reaching the brain. It seems pos-

sible that this vesicle was derived from a
single optic stalk which has become di-
lated and differentiated into retinal tis-
sue, but which has lost its original con-
nection with the brain.

The specimen shows many other cra-

154 Advances in Herpetology and Evolutionary Biology

Figure 3. Transverse sections through front of head of partially cyclopic specimen A of Caretta. A. Through
proboscis. B. Through base of proboscis. C. Through nasal sac and fused eyes. A strand of retinal pigment
associated with the “third eye” is issuing through the cleft in the scleral cartilage.

Abbreviations: b, bone of problematic identity; e, eye; g, eye gland; |j, lower jaw; mx, maxilla; nc, nasal capsule
(cartilaginous); ns, nasal sac; nt, nasal tube; p, pigment; pf, ? prefrontal; re, retina; sc, scleral cartilage.

PARTIAL CYCLOPIA AND MONORHINIA IN TURTLES - Bellairs

nial malformations. These are more ex-
tensive than the external appearance of
the head would suggest. They include
extreme narrowing of the pharynx, ab-
sence of inner ear structures (presup-
posing absence of otocysts), and absence
of obvious pituitary elements. This last
abnormality has been reported in other
embryos with cranial malformations
(Bellairs, 1965; Johnston et al., 1977).

Specimen D. This specimen has a
single tubular nasal vestibule (Fig. 4B)
which leads back into the main part of the
more or less single nasal sac. The most
anterior part of this sac is paired.
Throughout most of its extent, however,
it is only partly divided by a broad but
incomplete septum which lacks a car-
tilaginous component (Fig. 4C). As in
specimen A, the sac ends blindly pos-
teriorly and there is no internal choana.
Both vestibule and main nasal sac are
more or less surrounded by a single car-
tilaginous nasal capsule.

The front of the upper jaw has a sharp,
beaklike appearance in transverse sec-
tion (Fig. 4C.) There are no premaxillae,
but maxillae, prefrontals and frontals are
recognizable. In the absence of choanae,
the bones of the palate are abnormal and
the vomers cannot be identified. The
palatal shelves of the maxillae form a
median symphysis throughout much of
their length (Fig. 4C,D) and articulate
posteriorly with the palatines (Fig. 4E). A
median diverticulum from the mouth,
anterior to and quite separate from the
supposed pituitary (see below), extends
upwards between the palatines (Fig. 4E).
It is possible that this diverticulum is a
vestige of the choana.

The olfactory nerves pass caudally to
enter the olfactory bulbs in more or less
normal fashion. The nerves are, however,

displaced to one side by a large cyst, ©

presumably pathological, which lies
above the posterior part of the nasal sac
(Fig. 4C). The cerebral hemispheres are
partly fused with each other and obvi-
ously abnormal.

The eyes are quite separate and rela-

155

tively normal in general appearance
(though not perhaps in detailed struc-
ture), and possess lenses. The two optic
nerves, however, appear to be directly
continuous with each other across the
midline (Fig. 4E), and do not form a
chiasma. They have no connection with
the brain. This would suggest that the
proximal part of the optic stalk was ori-
ginally single, and that this portion dis-
appeared at some later stage of develop-
ment.

No neural components of the pituitary
can be identified. There is, however, a
median diverticulum from the pharynx
which extends dorsally to end within the
substance of the ossifying basisphenoid.
It seems possible that this represents the
pharyngeal component of the pituitary, or
Rathke’s pouch.

A remarkable feature of this specimen
is the deficiency of the midline structures
of the anterior part of the chondrocra-
nium, i.e., the structures which are partly
or completely of trabecular origin. As
previously mentioned, there is no car-
tilaginous median nasal septum of nor-
mal type. The lateral walls of the pos-
terior part of the nasal capsule, however,
become approximated almost in the mid-
line for a short distance behind the blind
nasal sac, and give the appearance of a
paired septum. More posteriorly, there is
no interorbital septum, although the eyes
are not especially close together. The two
Harderian glands, which normally lie
one on either side of the interorbital sep-
tum (Fig. 4F), are contiguous with each
other (Fig. 4D).

The anterior orbital cartilages are rep-
resented by a well-developed planum
supraseptale which lies in the normal
position beneath the forebrain. The
paired parts of the trabeculae, which
normally flank the pituitary fenestra, are
not apparent, and there is no fenestra.
Instead there is a pointed median rod of
cartilage which projects anteriorly from
the front of the parachordal or basal plate.
It is partly ossified and merges posterior-
ly with the basisphenoid ossification

156 Advances in Herpetology and Evolutionary Biology

ob
\

XQ Ly
\ WX
: /
Gale

Figure 4. A. Transverse section through head of partially cyclopic specimen A of Caretta showing fused eyes
and‘‘third eye.” Level is posterior to Fig. 3C. B—E. Transverse sections through head of specimen D of Caretta
with monorhinia. B. Through front of snout. C. Through nasal sac showing incomplete septum. D. Through middle
of eyes showing absence of interorbital septum. E. Through posterior parts of eyes showing optic nerve. F.
Transverse section through interorbital septum of normal Caretta embryo.

Abbreviations as in Fig. 3, plus new terms: br, brain; cs, conjunctival space; cy, cyst; d, diverticulum from mouth
(not pituitary); e 3, “third eye”; fr, frontal; Hg, Harderian gland; is, interorbital septum; j, jugal; le, lens; m, eye
muscles; nse, nasal septum; ob, olfactory bulb and peduncle; p, parietal; pf, prefrontal; po, postorbital; ps, planum
supraseptale; pt, palatine; so, scleral ossicle; |, olfactory nerves; II, optic nerve.

PARTIAL CYCLOPIA AND MONORHINIA IN TURTLES - Bellairs

developing in the basal plate on either
side of the notochord. This rod might
represent either a taenia intertrabecularis
or an intertrabecula; these structures are
characteristic of the developing skulls of
some Testudines (Bellairs and Kamal,
1981). Another possibility is that it repre-
sents the posterior parts of the trabecu-
lae, which have undergone abnormal fu-
sion in this region since abnormal tra-
becular fusion has been noted in am-
phibian embryos with experimentally
produced cyclopia (de Beer, 1937). The
more posterior parts of the skull of this
specimen, the otic capsule and inner ear
structures, show no gross malformations.

DISCUSSION AND CONCLUSIONS

Of the two specimens examined histo-
logically, A is by far the more abnormal,
showing partial fusion of the eyes with
presence of a third eyelike structure, a
well-developed proboscis as in typical
mammalian cyclopia, absence of the
pituitary and of inner ear structures, as
well as other gross abnormalities which
affect almost every part of the head. It is
impossible from external examination to
state whether similar abnormalities are
present in the unsectioned specimens B
and C.

Specimen D, though not actually cy-
clopic, seems to show a less extreme ex-
pression of the cyclopic trend. Although
the eyes are quite separate, the two nasal
sacs and their capsules appear to have
fused into a largely single structure. As in
A, the premaxillary component of the up-
per jaw is missing and there is virtual ab-
sence of the midline elements of the
anterior chondrocranium.

The ontogenetic development of cy-

clopia and related conditions seems to _

stem from some deficiency of the front of
the neural plate (or the neural tube) and
of associated mesoderm manifesting it-
self at a very early stage in development.
Cyclopia has been produced experi-
mentally by surgical procedures de-

157

signed to produce such defects (Johnston
et al., 1977). Coalescence or approxima-
tion of the eyes and of the nasal placodes
might be expected to occur if there was a
reduction in the amount of tissue present
in the anterior midline of the early em-
bryo. Moreover, such approximation of
paired structures might well prevent the
tissues of the frontonasal process (which
forms the anterior and median part of the
face) from growing ventrally and giving
rise to the premaxillary region of the
upper jaw. Such material, prevented from
reaching its normal destination, may pro-
duce instead an abnormal, proboscis-like
structure.

Deficiencies of the cranial neural plate
would also involve deficiencies of the
cranial part of the neural crest. It is well
known that neural crest tissue normally
gives rise to branchial arch elements, in-
cluding the trabeculae, and to other parts
of the facial skeleton (Le Liévre, 1978). It
is thus possible to explain the absence of
nasal and interorbital septa, which are at
least partly of trabecular origin (Bellairs
and Kamal, 1981), as well as the more
posterior parts of the trabeculae which
normally remain paired.

The albinoid condition of these em-
bryos suggests some generalized, rather
than a purely cranial, deficiency of the
neural crest. It is noteworthy that Cald-
well (1959) found albinoid embryos with
malformations of the jaws and eyes in
natural nests of Caretta, though cyclopia
is not explicitly mentioned in his ac-
count.

ACKNOWLEDGMENTS

I am most grateful to P. C. H. Pritchard
for supplying these specimens and rele-
vant information about them, to A.
McGehee for allowing me access to data
in her unpublished thesis, and to M. W.
Ferguson, A. D. Hoyes, and G. M.
Morriss-Kay for helpful comments. My
thanks are also due to B. G. H. Martin and
to the Audio-visual Department of St.

158

Mary’s Hospital Medical School for the
photographs.

LITERATURE CITED

ADELMANN, H. B. 1936. The problem of cyclopia. Q.
Rev. Biol., 11: 161-182 and 284-304.

BELLAIRS, A. D’A. 1965. Cleft palate, microphthal-
mia and other malformations in embryos of
lizards and snakes. Proc. Zool.:Soc. Lond., 144:
239-251.

BELLAIRS, A. D’A., AND A. M. KAMAL. 1981. The
chondrocranium and the development of the
skull in Recent reptiles. In C. Gans and T. S.
Parsons (eds.), Biology of the Reptilia. London
and New York, Academic Press, 11: 1-263.

CALDWELL, D. K. 1959. The Atlantic loggerhead sea
turtle, Caretta caretta caretta (L.), in America.
III. The loggerhead turtles of Cape Romain,
South Carolina. Bull. Fla. St. Mus. Biol. Sci.,
4(10): 319-348.

DE BEER, G. R. 1937. The development of the ver-
tebrate skull. Oxford, Clarendon Press.

EWERrT, M. A. 1979. The embryo and its egg: de-
velopment and natural history pp. 333-413. In
M. Harless and H. Morlock (eds.), Turtles. Per-
spectives and research. New York, John Wiley.

JOHNSTON, M. C., G. M. Morriss, D. C. KUSHNER,
AND G. J. BINGLE. 1977. Abnormal organogene-
sis of facial structures, pp. 421-451. In J. G.
Wilson and F. C. Fraser, (eds.), Handbook of
teratology. New York, Plenum Press.

Advances in Herpetology and Evolutionary Biology

LE LIEvRE, C. S. 1978. Participation of neural crest-
derived cells in the genesis of the skull in
birds. J. Embryol. Exp. Morphol., 47: 17-37.

MCGEFHEE, M. A. 1979. Factors affecting the hatch-
ing success of loggerhead sea turtles. Master’s
thesis, University of Central Florida, 1-252.

McGnratTH, P. 1981. Absence of the presphenoid in
relation to the neural and facial abnormalities in
human cyclopia. J. Anat. 133: 473-474.

ROMANOFF, A. L. 1972. Pathogenesis of the avian
embryo. New York, Wiley-Interscience.

SCHWALBE, E.., AND H. JOSEPHY. ca. 1913. Die Miss-
bildungen des Kopfes. II. Die Cyclopie pp.
205-246. In Die Morphologies der Missbil-
dungen des Menschen und der Tiere. (1906—
1958) G. Fischer, Jena, 3 Teil, Kapitel 5.

WILLIS, R. A. 1958. The borderland of embryology
and pathology. London, Butterworth.

ADDENDUM

An association between craniofacial
defects and albinism in turtles has also
been noted by Ewert (1979) and by J. D.
Miller (personal communication) and
may reflect neural crest deficiency.
McGrath (1981) reports absence of the
presphenoid in cases of human cyclopia.
This bone would probably correspond
with part of the unossified interorbital
septum of reptiles.

Neural Pattern—A Neglected Taxonomic
Character in the Genus Pelusios Wagler
(Pleurodira: Pelomedusidae)

DONALD G. BROADLEY!

ABSTRACT. Variation in the pattem of neural bones
in the genus Pelusios is documented and shown to
be a useful taxonomic character at the species level
with respect to both Recent and fossil material.
Neural pattern is also documented for Pelomedusa
subrufa.

INTRODUCTION

The taxonomy of the African hinged
terrapins of the genus Pelusios Wagler
has long been a source of controversy,
and numerous taxonomic characters have
been employed by various workers. In
the last comprehensive revision of the
Recent species of the genus, Loveridge
(1941) recognized only four species, i.e.,
P. adansonii (Schweigger), P. gabonensis
(Dumeril), P. suwbniger (Lacépede), and
P. sinuatus (A. Smith). Muller and
Hellmich (1954) subsequently reinstated
P. niger (Duméril and Bibron) as a valid
species. Laurent (1956, 1964, 1965) resur-
rected two more species, castaneus
(Schweigger) and bechuanicus Fitz-
Simons, from the synonymy of P. sub-
niger and also described three new spe-
cies, nanus, carinatus, and williamsi
(with a subspecies lutescens) from Cen-

tral Africa. He also revived the taxa derbi- -

anus (Gray) and rhodesianus Hewitt as
subspecies of P. castaneus and described
a new subspecies P. c. chapini. Raw

INational Museum, P.O. Box 240, Bulawayo,
Zimbabwe.

(1978) restored P. rhodesianus to specific
rank when he found it it be sympatric
with P. castaneus in Zululand. Bour
(1978) proposed that the name P. cas-
taneus castaneus should be restricted to
the West African populations previously
included under P. c. derbianus and he
erected the name P. castaneus kapika for
the Malagasy populations. Broadley
(1981) confirmed the status of P. rho-
desianus as a full species, sympatric with
P. castaneus castanoides Hewitt in
Zululand, where the two species are
readily distinguished on neural pattern.

Laurent (1965) described P. williamsi
as a full species because its yellowish
western race lutescens is sympatric with
the blackish P. castaneus chapini in the
Lake Albert (Lac Mobutu Sese Seko)
basin. Broadley (1981) noted sympatry
between blackish P. rhodesianus and yel-
lowish (but usually appearing black
through soil staining) P. w. williamsi at
Entebbe, Uganda. The neural patterns of
the few specimens examined from this
region suggest that P. williamsi is actu-
ally conspecific with P. castaneus and the
subspecies chapini is conspecific with P.
rhodesianus.

A review of the six southern African
species of Pelusios (Broadley, 1981)
generally confirms the taxonomy estab-
lished by previous authors (Laurent,
1956, 1964, 1965; Raw, 1978). It also
shows that the pattern of neural bones in
the carapace provides a useful taxonomic
character neglected by previous workers

160

on this genus. In the present paper, the
survey of neural patterns is extended to
cover the remaining species of Pelusios,
plus Pelomedusa subrufa (Lacépede).
This paper is dedicated to Dr. Emest
E.. Williams, who made a major contribu-
tion to our understanding of the evolu-
tion of the genus Pelusios in his 1954

paper.

MATERIALS AND METHODS

The initial review of neural patterns in
Pelusios was based on osteological speci-
mens in the National Museum collection.
The neural arrangement was _ also
checked on many complete specimens
(both wet and dry) by stripping the ver-
tebral shields from the carapace, as
neural patterns cannot be identified on
X-rays. With the cooperation of several
Curators, I was able to extend the in-
vestigation to additional specimens on
loan from other institutions.

The nomenclature of bones of the shell
follows Zangerl (1969). Neurals are num-
bered from the front N1 to N8.

Most abbreviations for institutional
prefixes are indicated under Acknowl-
edgments. UM and NMZB (formerly
NMSR) indicate two sets of catalogue
numbers in the Department of Herpe-
tology at the National Museum of
Zimbabwe. ¢ indicates extinct forms.

NEURAL BONES IN THE PLEURODIRA

Zangerl (1948) noted that the Chelidae
differed from the Pelomedusidae in
showing a strong tendency towards re-
duction of neural bones. The Neotropical
Phrynops nasutus has only four neurals
(N1, N6, N7, and N8 absent; Boulenger,
1889: Fig. 58) and Platemys platycephala
lacks neurals. Chelodina oblonga, iso-
lated in the southwest corner of Australia,
is the only Australasian chelid with five
to eight neurals (Burbidge et al., 1974).
Although the other taxa normally lack

Advances in Herpetology and Evolutionary Biology

neurals, one to four vestigial and usually
isolated neurals are occasionally present
in three other species of Chelodina and
two species of Elseya (Rhodin and Mit-
termeier, 1977).

Recent Pelomedusidae do not show
such striking reductions in neurals,
Podocnemis has seven (rarely 6), Erym-
nochelys six, Pelomedusa seven (rarely 6
or 8), and Pelusios five to eight (rarely 4).
However, the Pliocene Podocnemis
venezuelensis lacked neurals (Wood and
Gamero, 1971) as did the Eocene Eu-
sarkia rotundiformis Bergounioux of
Tunisia (Moody, 1972).

Although no species of Pelusios has
less than four neurals, and there is intra-
specific variation in the number and size
of those present, neurals may be lost at
either or both ends of the series, so that
sympatric taxa may be distinguishable on
neural pattern, e.g., P. rhodesianus, P.
castaneus castanoides, and P. sinuatus in
Zululand. Neural patterns are apparently
nonadaptive and are therefore particu-
larly useful taxonomic characters at the
species level.

NEURAL PATTERNS IN PELOMEDUSA
AND PELUSIOS

Boulenger (1889) stated that in the
Pleurodira the neural series rarely con-
tains more than seven bones. A single
‘pygal’ (=suprapygal) is present, and in
none of the Recent forms does it make
contact with the neurals. He illustrated
the neural pattern of a Sternothaerus
derbianus (= Pelusios c. castaneus) with
a continuous series of eight neurals,
separated from the single suprapygal
(Boulenger, 1889: Fig. 47).

The first author to report a reduction in
the number of neural bones in Pelusios
was Hewitt (1933), who illustrated a para-
type of P. sinuatus leptus with only five
neurals, N1 failing to contact the nuchal
and N5, N7, and N8 absent. He also
recorded a Zululand specimen of P. sinu-
atus zuluensis Hewitt with a continuous
series of seven neurals.

In his diagnosis of the family Pelo-
medusidae, Zangerl (1948) noted the
presence of six to eight neural bones.

Broin (1969) illustrated the variation in
neural pattern in nine Recent and fossil
specimens of P. sinuatus, including the
type of P. rudolphit Arambourg. These
have five to seven neurals: all lack N8,
most lack N7, one lacks N6, and five have
N5 reduced or absent, so that N6 is iso-
lated.

In a review of the genus Pelusios south
of Latitude 8°S Broadley (1981) checked
the neural arrangement in 136 specimens
from this region. In order to expand the
study to cover the remaining species in
the genus I have examined an additional
33 specimens from north of Latitude 8°S
and incorporated data from the literature
(Hewitt, 1933; Broin, 1969). For compari-
son, neural patterns for 25 specimens of
Pelomedusa subrufa were also recorded.
The neural patterns found in Pelomedusa
and Pelusios are described below and
summarized in Table 1.

Pelomedusa subrufa (Lacépede)

Twenty-five specimens were ex-
amined. N8 is absent in 20 and reduced
in three others; two specimens also lack
N7. N1 is narrowed anteriorly in some of
these specimens, but it is reduced and
widely separated from the nuchal in
FMNH 17160 (Zangerl, 1948).

Genus Pelusios

P. rusingaet Williams

The type and only known specimen of
this species is the earliest known shell
belonging to the genus Pelusios (Wil-
liams, 1954, Pl. 1). The neural pattern is
complete, consisting of an unbroken
series of eight neurals, N1 in good con-
tact with the nuchal and N8 well sepa-
rated from the single suprapygal (Fig.
AY):

NEURAL PATTERN IN PELUSIOS - Broadley

161

P. gabonensis (Dum6éril)

Five specimens have been examined.
BM 1912.6.27.38 has a similar arrange-
ment to P. rusingae. BM 1912.6.27.40 has
N1 reduced, but N8 elongate and in con-
tact with the suprapygal. Three others all
lack N1, but the first of these specimens
has a small N9 (Fig. 1B).

P. adansonii (Schweigger)

All seven specimens examined lack N1
(Fig. 1C). BM 1900.9.22.7 also lacks N8,
but BM 1900.9.22.6 has a small N9.

P. nanus Laurent

Neural patterns were determined on
four specimens. AMNH 50757 has eight
neurals, but N8 is reduced in size and
meets N7 at a point (Fig. 1D); three
others lack N8. N1 always tapers an-
teriorly, meeting the nuchal at a point or
just failing to make contact.

P. niger (Duméril and Bibron)

Three specimens were examined. Two
have NI reduced or absent and N8
elongate and in contact with, or narrowly
separated from, the suprapygal (which is
divided into three in one specimen, Fig.
1E). BM 1927.9.27.237 lacks N1 and N8.

P. subniger (Lacépéde)

Of 25 specimens examined for neural
pattern, 23 have a continuous series of
eight neurals, Nl in good contact with
the nuchal, N8 well separated from the
suprapygal. Three of these specimens

(UM 850; 3181-2) have two superposed

syprapygals (Fig. 1F). Two specimens
(AM 5432; UM 9720) lack N8.

P. bechuanicus bechuanicus FitzSimons

All 13 specimens examined have a con-
tinuous series of eight neurals separated

162

from the suprapygal (Fig. 1G). In UM
9718 and UM 29157, N1 is reduced and
does not contact the nuchal; in addition,
the latter specimen has N8 transversely

divided.

P. bechuanicus upembae Broadley

Both specimens examined have eight
neurals, but MRAC 11460 has N8 trans-
versely divided.

P. rhodesianus Hewitt

Twenty-eight specimens were ex-
amined. Nine have a continuous series of
eight neurals with N1 in good contact
with the nuchal and N8 contacting the
suprapygal (Fig. 1H), 15 others have N8&
separated from the suprapygal. Two have
N7 and N§8 reduced in size and isolated,
another has only N8 isolated. In MRAC
16962 N8 is absent.

P. castaneus castaneus (Schweigger)

Five specimens were examined, in-
cluding the specimen illustrated by
Boulenger (1889: Fig. 47). All have a con-
tinuous series of eight neurals with N1 in
good contact with the nuchal and N8 well
separated from the suprapygal (Fig. 2A).

P. castaneus williamsi Laurent

Two specimens from the type locality
were examined. One lacks N1, and in
both specimens N§8 is elongate and only
narrowly separated from the suprapygal
(Fig. 2B).

P. castaneus lutescens Laurent

The paratype examined has seven
neurals, Nl absent and N8 reduced in
size.

P. castaneus castanoides Hewitt

Eighteen specimens were examined.
The number of neurals varies from five to

Advances in Herpetology and Evolutionary Biology

eight, but Nl, N7, and N8 are always

reduced in size or absent. The Malagasy
population (P. castaneus kapika Bour)
seems to be indistinguishable from main-
land populations. Geographical variation
is as follows:

Malawi: N1, N7 and N8 absent in six

specimens.

Mocambique: One specimen has N1
reduced, N7 absent and N8 minute
(Fig. 2d).

Zululand: Five neurals in three speci-
mens, six in two (Fig. 2C), and seven
in two.

Madagascar: One specimen has N1,
N7, and N8 absent.

Seychelles: Three specimens have
seven or eight neurals, with N1 and
N7 small to minute, N8 minute or
absent.

P. carinatus Laurent

The paratype examined has eight
neurals, but N7 and N8 are both very
small and N8 is isolated (Fig. 2E).

P. sinuatus (A. Smith)

Neural patterns were examined in 55
specimens and complete patterns for an
additional four Recent and four fossil
specimens are illustrated by Broin
(1969). There are four to seven neurals,
N8 always absent. N1 is reduced (not
contacting the nuchal) in six specimens,
absent in only one (UM 32996). N5 is
absent in five, reduced in 13 and divided
in one (Fig. 2G); N6 is absent in five and
reduced in four; N7 is absent in 52 speci-
mens and reduced in four others.

Specimens with seven neurals (Fig.
2F) are more common in the south. A
reduction to five (rarely 4) neurals seems
to be restricted to populations in Lake
Tanganyika and the Luangwa Valley in
eastern Zambia (Fig. 2H), which attain
the maximum sizes recorded for any
Pelusios (maximum carapace length 465
mm in Lake Tanganyika; Witte, 1952).

Usually it is N5 that is missing or re-
duced, less commonly N6, rarely both
-(MCZ 48019). It should be noted that
three specimens from the early Pleisto-
cene of Omo, north of Lake Turkana,
dated at 3.75 to 1.8 million years B.P., all
have NOS absent (type of P. rudolphit
Arambourg) or reduced and N7 and N8
absent (Broin, 1969). A specimen from
Bed I, Pleistocene of Olduvai (BM
R5761) also has six neurals, with N5
reduced and NI not contacting the
nuchal. The Recent Somali specimen of
P. sinuatus illustrated by Broin (1969)
has N5 reduced, but another from Uebi
Shabeli River (MF 20647) shows no
reduction of N5.

PHYLOGENY OF PELUSIOS

The earliest known fossil Pelusios shell
is the type specimen of P. rusingaet,
described from the early Miocene of
Rusinga Island, Lake Victoria, Kenya, by
Emest E. Williams (1954). The other
Miocene species, P. blanckenhornit
Dacque, is known only from a skull and
hence could equally prove to be a
Pelomedusa (Williams, 1954; Wood,
1974). P. dewitzianust von Reinach is
known only from fragments from the
middle Pliocene of Egypt. P. rudolphit
Arambourg from the early Pleistocene of
Omo, Ethiopia, appears to be indistin-
guishable from Recent P. sinuatus
(Broin, 1969).

Williams (1954) suggested that small
lateral mesoplastra are primitive for the
Pelomedusidae, as this condition is
found in all known Cretaceous pelo-
medusids. This in turn implies that the
large mesoplastra with median contact in
Pelusios is a secondarily derived condi-
tion associated with the development of
the plastral hinge. Pelomedusa subrufa,
the only other African pelomedusid spe-
cies, retains small lateral mesoplastra.

In the first couplet of his key to the
genus Pelusios, Loveridge (1941) sepa-
rated the species adansonii and ga-
bonensis, which have the anterior lobe of

NEURAL PATTERN IN PELUSIOS - Broadley

163

the plastron more than twice the length
of the suture between the abdominal
shields, from subniger and sinuatus, in
which the anterior plastral lobe is less
than twice the length of the abdominal
suture. Williams (1954) endorsed this
division into two species groups when he
described P. rusingaet.

The four species of what may be called
the P. gabonensis group form an evolu-
tionary sequence based on shape of
mesoplastra. In the Miocene species P.
rusingaet the mesoplastra are strongly
tapered medially both anteriorly and
posteriorly, hardly meeting. In P.
gabonensis the mesoplastra are moder-
ately tapered medially both anteriorly
and posteriorly, forming oblique sutures
with both hyo- and hypoplastra. In P.
adansonii the mesoplastra are tapered
only posteriorly, anteriorly forming a
straight transverse hinge with the hypo-
plastra. In P. nanus the mesoplastra are
only slightly tapered posteriorly.

In the P. subniger group the meso-
plastra are not tapered, forming straight
transverse sutures with both hyo- and
hypoplastra. The taxonomy of this group
has always presented problems because
of the shortage of clear-cut morphological
characters.

P. niger has a distinctive skull with a
narrow snout and well-developed beak
(Boulenger, 1889: Fig. 46; Muller and
Hellmich, 1954: Figs. 3, 4; Gaffney,
1979: Figs. 132, 133). Although Wood
(1974) placed this species in the P.
gabonensis group, it has a short anterior
plastral lobe, the mesoplastra do not
taper medially, and it is a relatively large
species (four adults in the British Mu-
seum are 250 to 277 mm in carapace
length). P. subniger and P. bechuanicus

may be distinguished on the relatively

large head size (Broadley, 1981). The
other species in this group are largely
distinguished on the proportions of the
epidermal shields and_ coloration
(Laurent, 1956, 1964, 1965; Broadley,
1981), but neural patterns provide a use-
ful new character.

164 Advances in Herpetology and Evolutionary Biology

E

Ht

Figure 1. Neural patterns in Pelusios. A. P. rusingaet (Type—Rusinga Island, Lake Victoria). B. P. gabonen-
sis (CM 39662—Nyabessem, Cameroun). C. P. adansonii (UM 33631—Bahr el Ghazel, Sudan). D. P. nanus
(AMNH 50757—Chitau, Angola). E. P. niger (BM 1974.3012—Nko, Nigeria). F. P. subniger (UM 850—Lake
Kariba, Zimbabwe). G. P. bechuanicus (UM 32985—Kasane, Botswana). H. P. rhodesianus (Type—Mpika
District, Zambia). Thick lines indicate sulci between epidermal shields, thin lines indicate sutures between bones.
The horizontal lines equal 1 cm to scale.

NEURAL PATTERN IN PELUSIOS : Broadley 165

rie

Figure 2. Neural patterns in Pelusios. A. P. castaneus castaneus (UM 33495—Ghana). B. P. castaneus wil-
liamsi (UM 33165—Kaimosi, Kenya). C. P. castaneus castanoides (TM 52160—Lake St Lucia, Zululand). D. P.
castaneus castanoides (UM 27828—Tando, Mocambique). E. P. carinatus (UM 33158—Eala, Zaire); F. P.
sinuatus (UM 12076—Lundi River, Zimbabwe). G. P. sinuatus (UM 5278—Bumi Confluence, Lake Kariba, Zim-
babwe). H. P. sinuatus (AM 5432—Mpika District, Zambia). Conventions as in Fig. 1.

166

Advances in Herpetology and Evolutionary Biologi
8 Y SY

TABLE |. AFRICAN PELOMEDUSIDAE: DEVIATIONS FROM THE ANCESTRAL CONDITION OF 8 NEURALS, THE FIRST IN
CONTACT WITH THE NUCHAL, THE LAST SEPARATED FROM THE SUPRAPYGAL.

Nl

ae)
S
No. of 5
TAXON N _neurals* Bo
PELOMEDUSA
subrufa 25 (6)7(8) _—
PELUSIOS
rusingaet 1 8 =
gabonensis 5 7-8
adansonii U (6)7(8) =
nanus 4 7(8) =
niger 3 6-8 1
subniger 25 (7)8 —
bechuanicus bechuanicus 13 8 2
bechuanicus upembae 9) 8 —
rhodesianus 28 (7)8 —
castaneus castaneus 5 8 —
castaneus williamsi 9) 7-8 =
castaneus lutescens 1 TW =
castaneus castanoides 18 5-8 6
carinatus 1 8 —
sinuatus 63 (4)5-7 5

N5 N6 N7 N8
z g e is
O Ma) Ma) Roly ear 4
<¢ «= e@-2w?e-2z ge
= RE ee nee
tk ae Jee
Tome ER a
Se EA SS eee
g Lo 2S 2 ee ee
we Lk i re
2. 2 a. 3 See
las 2 2. 2 Eee
1 2 oS aS 2a]
m2 = & + 2) 5 >6o eee
we a ee ee Ea eee
Li 10. 54.4 4+ 4 33

*Rare conditions are placed in parentheses. Numbers of specimens with various neurals enlarged (i.e., N8 contacting or nearly contacting

suprapygal), reduced in size or absent, are indicated for each taxon.

Zangerl (1948) considered that the
primitive pelomedusid carapace had
eight neural bones, the first adjoining the
nuchal. The limited number of neural
patterns examined for the P. gabonensis
group indicate a primitive condition for
P. rusingaet with no reduction in neurals;
P. gabonensis and P. adansonii usually
lack N1, whereas in P. nanus N1 is nar-
rowed anteriorly and N8 is reduced or
absent, placing this species on a different
evolutionary line.

In the P. subniger group, the two
broad-headed species P. subniger and P.
bechuanicus retain the full complement
of 8 neurals, but P. niger has N1 reduced
or absent.

The P. rhodesianus/P. castaneus as-
semblage exhibits a variety of patterns,
but in the latter species there seems to be
a stepped east-west cline in reduction of
neurals, with N1 being lost first, then N7

and N8 simultaneously. The neural pat-
tern of P. castaneus castanoides, with N1,
N7, and N8 all reduced or absent, is very
distinctive.

The P. carinatus neural pattern (N7
and N8 reduced) appears to represent a
stage in the evolution of the P. sinuatus
condition (N8 absent, other neurals lost
in the sequence N7, N65, rarely N6). The
two species are allopatric, with P. carin-
atus in the Zaire (Congo) basin, and P.
sinuatus widespread in eastern Africa,
but formerly extending westward to
Chad (Broin, 1969).

It is hoped that this brief review of
neural patterns in Pelusios will encour-
age future workers to explore the range of
variation in the poorly known northern
forms and also the fossil material from the
fluvial and lacustrine deposits east of
Lake Turkana (Rudolf) (Behrensmeyer,
1975).

ACKNOWLEDGMENTS

I am grateful to the following col-
leagues for assistance: P. H. Skelton and
J. C. Greig of the Albany Museum,
Grahamstown (AM); W. D. Haacke and
Ms. L. Wessels of the Transvaal Museum
(TM); D. F. E. Thys van den Audenaerde
of the Musee Royal de |’Afrique Cen-
trale, Tervuren (MRAC); R. G. Zweifel of
the American Museum of Natural History
(AMNH); C. J. McCoy and A. V. Bianculli
of the Carnegie Museum (CM); E. E.
Williams and J. P. Rosado of the Museum
of Comparative Zoology (MCZ); R. H.
Parker of FitzSimons’ Snake Park, Dur-
ban (DSP); Lynn Raw of Pietermaritz-
burg (LR); C. J. McCarthy of the British
Museum Natural History (BM); J. P.
Gosse of the Institut Royal des Sciences
Naturelles de Belgique (IRScNB); and B.
Lanza of the Museo Zoologico de ‘La
Specola’, Florence (MF).

I wish to thank M. L. Angell for pho-
tographic assistance, A. Spector for the
donation of X-ray photographs, and M.
Clift and B. L. Bennefield for typing
various versions of the manuscript.

I am grateful to J. B. Iverson for com-
menting on the first draft of this paper
and to the reviewers for their helpful
suggestions.

APPENDIX: SPECIMENS EXAMINED
FOR NEURAL PATTERN

Pelomedusa subrufa. GHANA: NMZB_ 6004.
ZAMBIA: NMZB 2629. MALAWI: UM 3184,
33022. BOTSWANA: UM 7430, 116034, 14708,
29152-6, 32968. ZIMBABWE: UM 3174, 5277,
20584, 32950, 32983, 33021, 33029, 33130. SOUTH
AFRICA: NMZB 6020; UM 334234.

GENUS PELUSIOS

Pelusios gabonensis. CAMEROUN: BM 1912.6.
27.38 and 40; CM 39662-3. GABON: UM 33431.

Pelusios adansonii. NIGERIA: BM 1970.1797.
SUDAN: BM 1900.9.22.6 and 7; UM 33631; USNM
75099, 75105, 75107.

Pelusios nanus. ANGOLA: AMNH 50757, 50760;
UM 33262. ZAMBIA: NMZB 3958.

Pelusios niger. GHANA: BM_ 1927.9.27.237.
NIGERIA: BM 1974.3012-3.

NEURAL PATTERN IN PELUSIOS : Broadley

167

Pelusios subniger. ZAMBIA: AM 5144, 6574,
6876; KM—; NMZB 3150; TM 38181; UM 6574.
MALAWI: MCZ 51100; UM 25462. BOTSWANA:
TM 46229. ZIMBABWE: UM 849, 850, 3181, 3182,
9720, 14517, 20370, 32966, 32986, 32987, 33170.
MOCAMBIQUE: UM 9713, 9714, 28345, 28842.

Pelusios bechuanicus bechuanicus. BOTS-
WANA: UM 12938, 16191, 22338, 29150, 29151,
29157, 32984, 33317. ZAMBIA: UM 791, 9718,
10673, 10677, 15594.

Pelusios bechuanicus upembae. ZAIRE (Shaba
Province): MRAC 11460 (paratype); TM 38178
(holotype).

Pelusios rhodesianus. UGANDA: AM—(2); BM
1906.5.30.1; UM 33389. ANGOLA: AMNH 50752,
50753. ZAIRE: MRAC 16962, 16964, 16965; UM
33392. ZAMBIA: AM—(holotype), 5432 (2); NMZB
2625, 2626; UM 47, 49. ZIMBABWE: UM 9715,
14516, 21654, 33033, 33038. MOCAMBIQUE: UM
9717. ZULULAND: TM 52341, 52342. NATAL
(Durban): DSP 86, 87; UM 33621.

Pelusios castaneus castaneus. “WEST AFRICA’:
BM 63.10.8.5. SENEGAL: CM 24785. GHANA:
UM 33495. NIGERIA: AM 7282; BM 1974.3014.

Pelusios castaneus williamsi. KENYA: TM
16433; UM 33165 (paratype).

Pelusios castaneus lutescens. ZAIRE: CM 62248

(paratype).
Pelusios castaneus castanoides. MALAWI: AM
T15, T16, T18, T19; UM 32990, 33390.

MOCAMBIQUE: UM 27828. ZULULAND: LR
1038, TM 13433 (holotype), 45644, 48318, 52160,
52544: UM 33634. MADAGASCAR: UM 33166.
SEYCHELLES: AM—; BM 1911.4.7.2; TM 49338.

Pelusios carinatus. ZAIRE: UM 33158 (paratype).

Pelusios sinuatus. SOMALIA: MF 20647.
TANZANIA: MCZ 30013, 48018, 48019. ZAIRE:
BM 1953.1.11.48; IRScNB 3, 33, 99. ZAMBIA:
AM—(holotype of P. s. leptus Hewitt), 5432, 5794,
6696, IRScNB 7143; NMZB 3151; UM 762, 17528,
23583, 32996, 32998, 33028. MALAWI: UM 4851,
25461, 33031, 33040. ZIMBABWE: NMZB 1210,
3700, 3751; UM 3178, 5278, 5856, 5867, 9799,
12076, 32975, 32976, 33029, 33030, 33032, 33039,
33386-8. MOCAMBIQUE: UM 3157, 10343, 27585,
30436. BOTSWANA: UM 11255. TRANSVAAL:
AM 7361; TM 15076. ZULULAND: TM 34682,
34684.

LITERATURE CITED

BEHRENSMEYER, A. K. 1975. The Taphonomy and
Paleocology of Plio-Pleistocene Vertebrate
assemblages east of Lake Rudolf, Kenya. Bull.
Mus. Comp. Zool., 146: 473-578.

BOULENGER, G. A. 1889. Catalogue of the Chelo-
nians, Rynchocephalians and Crocodiles in the
British Museum (Natural History). London,
British Museum (Nat. Hist.), 311 pp.

168

Bour, R. 1978. Les tortues actuelles de Madagascar
(République malagache): liste systématique et
description de deux sous-especes nouvelles
(Reptilia—Testudines). Bull. Soc. Etud. scient.
Anjou N.S., 10: 141-154.

BROADLEY, D. G. 1981. A review of the genus
Pelusios Wagler in southern Africa (Pleurodira:
Pelomedusidae). Occas. Pap. Natl. Mus.
Zimbabwe, B., Nat. Sci., 6: 635-683.

BROIN, F. DE. 1969. Sur la présence d’une Tortue,
Pelusios sinuatus (A. Smith) au Villafranchien
inférieur du Tchad. Bull. Soc. géol. Fr., 11:
909-916.

BURBIDGE, A. A., J. A. W. KIRSH, AND A. R. MAIN.
1974. Relationships within the Chelidae
(Testudines: Pleurodira) of Australia and New
Guinea. Copeia, 1974: 392-409.

GaFFNEY, E. S. 1979. Comparative cranial mor-
phology of Recent and fossil turtles. Bull.
Amer. Mus. Nat. Hist., 164: 65-376.

HEwiIrTT, J. 1933. Descriptions of some new reptiles
and a frog from Rhodesia. Occ. Pap. Rhod.
Mus., 1: 45-50.

LAURENT, R. F. 1956. Contribution a | Herpetologie
de la Région des Grands Lacs de |’Afrique
centrale. Annls Mus. r. Congo belge, Sér. 8vo,
48: 1-390.

—__. 1964. Reptiles et amphibiens de |l’Angola
(troisieme contribution). Publcoes cult. Co.
Diam. Angola, 67: 1-165.

____. 1965. A contribution to the knowledge of the
genus Pelusios (Wagler). Annls Mus. r. Afr.
cent., Sér. 8vo, 135: 1-33.

LOVERIDGE, A. 1941. Revision of the African ter-
rapin of the family Pelomedusidae. Bull. Mus.
Comp. Zool., 88: 462-524.

Advances in Herpetology and Evolutionary Biology

Moopy, R. T. J. 1972. The turtle fauna of the Eo-
cene phosphates of Metlaoui, Tunisia. Proc.
Geol. Ass., 83: 327-335.

MULLER, L., AND W. HELLMICH. 1954. Zur Kenntnis
einiger Pelusios—Arten (Testudines). Veroff.
zool. St. Samml. Munch., 3: 51-79.

Raw, L. R. G. 1978. Taxonomic notes on the hinged
terrapins, genus Pelusios, of Natal. Durban
Mus. Novit., 11: 287-294.

RHODIN, A. G. J., AND R. A. MITTERMEIER.~1977.
Neural bones in chelid turtles from Australia
and New Guinea. Copeia, 1977: 370-372.

WILLIAMS, E. E. 1954. A new Miocene species of
Pelusios and the evolution of that genus.
Breviora Mus. Comp. zool. No. 25, pp. 1-7.

WITTE, G.-F. DE. 1952. Amphibians and Reptiles.
Explor. hydrobiol. Lac. Tanganyika (1946—
1947), 3(3): 1-22.

Woop, R. C. 1974. The systematics, evolution and
zoogeography of African turtles. Natn. geogr.
Soc. Res. Rep. 1967 projects: 301-306.

Woop, R. C., AND M. L. D. DE GameEro. 1971.
Podocnemis venezuelensis, a new fossil pelo-
medusid (Testudines, Pleurodira) from the
Pliocene of Venezuela and a review of the
history of Podocnemis in South America.
Breviora Mus. Comp. Zool. No. 376, pp. 1-23.

ZANGERL, R. 1948. The vertebrate fauna of the
Selma formation of Alabama. Part II. The
Pleurodiran turtles. Fieldiana, Geol. Mem., 3:
17-56.

____.. 1969. The turtle shell, pp. 311-339. In C. Gans
(ed.), Biology of the Reptilia. 1, Morphology A.
London & New York, Academic Press.

The Genus Emydura (Testudines: Chelidae) in
New Guinea with Notes on the Penial
Morphology of Pleurodira

SAMUEL B. MCDOWELL!

ABSTRACT. Two species of Emydura occur in New
Guinea: E. dentata (synonyms: Elseya dentata,
Emydura novaeguineae, Elseya novaeguineae) and
E. australis (synonyms: Emydura kreffti, E. sub-
globosa, E. albertisii). The latter is closely related
to E. dentata and probably derived from it. E. dent-
ata probably also lies close to the ancestry of E.
latisternum (northeast Australian), which in turn,
appears closely related to E. macquaria (southeast
Australian) and Pseudemydura umbrina (westemm
Australian). Emydura dentata is the most primitive
and pelomedusid-like of Australian Chelidae. The
other Australo-New Guinean chelid genus, Chelo-
dina, is not morphologically close to Emydura, but
instead seems closer to South American forms. The
penis of Pleurodira is intermediate between the
protrusile erectile organ of Cryptodira and cloacal
eversion as seen in caecilians. The pleurodiran
penis, formed of an almost cartilagelike corpus fi-
brosum tightly bound to a highly expansible corpus
cavemosum, warps when the corpus cavernosum
expands while the corpus fibrosum remains of con-
stant size. This warping doubles the penis over and
acts as a lever to evert the cloaca as a sheath around
the base of the protruded penis. This is interpreted
as the primitive form of the penis for Testudines.
Penial morphology offers useful taxonomic charac-
ters in Pleurodira, and the two New Guinea species
of Emydura differ in penial structure.

INTRODUCTION

It seems fitting that any commemora-
tive volume to Emest E. Williams should
include some discussion of turtles, and
since it was during his curatorship of her-
petology at the Museum of Comparative

1Department of Zoology and Physiology, Rutgers
University, Newark, New Jersey 07102, U.S.A.

Zoology that the MCZ acquired the
largest collection of New Guinea reptiles
in the world, it seems equally fitting to
discuss New Guinea turtles. Accordingly,
I here report on the side-neck terrapin of
the genus Emydura in New Guinea,
based mainly on those in the collection of
the American Museum of Natural History
(AMNH). I have purposely omitted use of
the MCZ collection for framing the hy-
pothesis presented below, so that the
MCZ collection may be used for the test-
ing of this hypothesis.

Prior to Goode (1967), the taxonomic
arrangement and diagnostic characters
used for Emydura (and the supposedly
distinct genus Elseya) followed the lines
laid down by Boulenger (1889) and modi-
fied to accommodate additional named
forms by de Rooij (1915). This arrange-
ment recognized: Elseya, with one spe-
cies, E. dentata, characterized by a longi-
tudinal ridge on the maxillary triturating
surface; and Emydura, for the remaining
forms (other than Pseudemydura
umbrina of Westem Australia, not de-
scribed until 1901), characterized by lack
of a ridge on the triturating surface of the
maxillary bone.

Goode _ (1967)

radically redefined

-Emydura and Elseya, the latter diag-

nosed by a homy shield on the crown of
the head which developed at an early
age, and by presence of erect tubercles
on the dorsal surface of the neck; Em-
ydura was restricted to those forms with
soft skin on the top of the head (except in

170

some old adults) and with low and incon-
spicuous tubercles on the neck. By this
redefinition, the species novaeguineae
and latisternum were transferred to EI-
seya and Emydura was restricted to E.
kreffti, E. australis, and E. macquaria
(the type). Goode also much simplified
the taxonomy by reducing many names to
synonymy (my own findings support all
of Goode’s synonymizing, and I would go
still further).

Burbidge, Kirsch, and Main (1974)
examined osteological features and used
serological tests with elaborate statistical
analysis of their serological results.
These serological comparisons had the
defect that only a small number of indi-
viduals (sometimes only one) repre-
sented each (supposed) species in the
comparison and thus, it is impossible to
say what might be the range in genetic
diversity within a single deme, much less
within a broadly distributed species.
However, the results of Burbidge et al.
are extremely important in giving an in-
dependent estimate of relationships be-
tween the taxa. All previous authors used
a series of morphological characters tradi-
tionally used in taxonomic studies of
Testudines, but for which the utility in
determining biological relationships was
largely untested. Burbidge et al. found
Elseya (sensu Goode) and Emydura
closely related to one another and quite
different from Chelodina. Pseudemydura
umbrina was well isolated from both the
Elseya-Emydura group and from Chelo-
dina but, to judge from their graphs (their
Figs. 8, 9), closer to Elseya (sensu Goode)
than to anything else. Their initial com-
putation of similarity on the basis of ser-
ology showed close relationship between
all Emydura (sensu Goode), but the
second computation (removing distor-
tions produced by gross differences be-
tween groups in the first computation)
showed Elseya dentata and E. novae-
guineae close to each other and E. lat-
isternum well separated. To judge from
their graphs, an Emydura australis from
Wyndham, Western Australia, and an

Advances in Herpetology and Evolutionary Biology

Emydura “subglobosa” from Lake Mur-
ray, Western Province, Papua New
Guinea, came out extremely close, while
an E. kreffti from Woodstock, Queens-
land was close but well separated and E.
macquaria from Patho, Victoria was still
farther separated. These results sup-
ported the general features of Goode’s
arrangement, but not all details of his
taxonomy: Goode considered the New
Guinea Emydura (“E. subglobosa’’) as a
synonym of E. kreffti, but suspected that
E. australis might not even belong in the
genus Emydura.

It is my own finding that if certain
morphological traits (particularly, form of
the horny rhamphotheca, details of base
of skull, form of ossified hyoid, coloration
of soft parts) are given high weight, and
certain other characters (particularly,

details of external shell structure) are ~

given low weight, then a morphological
grouping emerges that is fully consistent
with Burbidge, Kirsch, and Main’s sero-
logical groupings.

It also would appear from my material
that “Elseya” dentata and “Elseya
novaeguineae’ are so closely related that
no character will consistently separate
them and, since they are allopatric, there
is no evidence that any speciation has
occurred and these two named _ taxa
should be considered a single species,
Emydura dentata. Similarly, there seem
to be no constant characters for distin-
guishing Emydura australis, E. “sub-
globosa” and E. “kreffti,” and all three
names appear to refer to a single species,
Emydura australis.

THE NEW GUINEA SPECIES OF
EMYDURA

Study of the American Museum of
Natural History (AMNH) Emydura from
New Guinea convinces me that two spe-
cies occur there, each also occurring in
Australia:

1) Emydura dentata (Gray) (Elseya
dentata and Elseya novaeguineae of

NEW GUINEA EMYDURA AND THE PLEURODIRAN PENIS - McDowell

Goode [1967], which see for full refer-
ences). Waigoe I. and nearly all of New
Guinea (at least as far east as Popondetta,
Northern Province, Papua New Guinea),
but not known from the Huon Peninsula
or Milne Bay Province (I cannot find a
definite record for the Port Moresby re-
gion); and Victoria River, northem Aus-
tralia to the Cairns and Gayndah regions
of Queensland.

2) Emydura australis (Gray) (Emydura
kreffti and Emydura australis of Goode
[1967], which see for full references).
Southern coast of New Guinea from
Mimika and Setekwa Rivers (Irian Jaya)
to the Aird Hills (Gulf District, Papua
New Guinea) and the elevated region
around Port Moresby (Central District);
tropical and subtropical Queensland,
Northern Australia, and northern West-
ern Australia.

These two species, as here expanded,
are largely sympatric and clearly distinct;
but not all the characters supposedly dis-
tinguishing them will work on my ma-
terial, and some features suggest these
two forms are most closely related to each
other. The special differences and re-
semblances need discussion.

Head pattern (see Fig. 1). Both New
Guinea species differ from other Aus-
tralo-New Guinean short-necked chelids
in having a pale mark on the upper tem-
poral region (1). (The numbers in paren-
theses refer to correspondingly num-
bered characters depicted in the figures.)
In E. australis this mark is always con-
spicuous and in sharp contrast to its
general surroundings. In E. dentata from
New Guinea the mark is not very distinct
and appears to be only one of several
paler mottlings of the head (2), and (fide
Goode, 1967) in Northern Australian E.
dentata this mark may merge with a
whitening of the entire temporal region.
In E. dentata this pale temporal mark is
continued through the eyelid to join its
counterpart as a transverse pale bar or
arch (3) between the fronts of the orbits.
In E. australis, this anterior continuation
of the pale temporal mark is either absent

171

(Western and Northem Australia) or ex-
tends forward on the snout to meet its
counterpart near the nostrils (Queens-
land and New Guinea).

Head scutellation (see Fig. 1). As
pointed out by Goode (1967), E. dentata
resembles the Australian E. latisternum
(and also P. umbrina) in having well-
defined convex scutes on the temporal
region (4), whereas E. australis re-
sembles the Australian E. macquaria in
having the temporal scutes flat and im-
perfectly defined from one another. If
there is any difference between E. den-
tata and E. australis in the thickness of
the cornification on the crown, the differ-
ence is too subtle for me to discern.

Neck tubercles. Iam unable to find any
constant difference between E. dentata
and E. australis in the development of
tubercles on the neck; E. dentata from
Papua New Guinea north of the central
watershed have the neck tubercles less
well developed than in the E. australis
from Central District, Papua New
Guinea, shown in Figure 1. (Further, all
northern New Guinea E. dentata have
rounded, rather than pointed, neck
tubercles and would key out as “Elseya
dentata’ rather than “Elseya novae-
guineae’ by Goode’s [1967] key.)

Penis (see Fig. 2). The terminology of
the penis used here is explained below
(pp. 180-185). In E. australis (Fig. 2A,C)
the penis is relatively longer and nar-
rower than in E. dentata (Fig. 2B). This
is most conspicuous in the proportions
of the free tip, which is longer than
wide in E. australis, but as wide as long
in E. dentata (5). In E. australis there
is a row of horny papillae (6) on the
anterior plica externa, sometimes on the
posterior plica externa, as well, but in E.
dentata these papillae are absent. (The

‘single Australian male of E. dentata

examined, BM [18]75.5.1.8, Gayndah, is
similar in penial structure to New Guinea
specimens; the penis of E. latisternum is
similar to that of E. dentata; I have not
seen the penis of E. macquaria or
Pseudemydura umbrina.)

f,
Be tad

itd

Sd
ieee)
Tiber

Heads of New Guinea Emydura. A. E. dentata, AMNH 59927. B. E. australis, AMNH 59910. Numbers

refer to text discussion.

Figure 1.

173

- McDowell

NEW GUINEA EMYDURA AND THE PLEURODIRAN PENIS

f

Q

I

unprotruded, AMNH

I

Figure 2. Penes of Emydura. A. E. australis, unprotruded, AMNH 69578. B. E. dentata

103629. C. E. australis, protruded, AMNH 57898. Numbers refer to text discussion.

Abbreviations: ape, anterior plica externa; ecl, everted cloacal lining; ft, free tip; pi, plica interna; pm, plica

medialis; ppe, posterior plica externa; sc, sinus communis.

174

Chelid foramen (see Fig. 3). Emydura
dentata and E. australis agree with each
other and differ from other Chelidae (ex-
cept Chelodina novaeguineae) in the
closure of the foramen (7) just antero-
lateral to the carotid foramen. In other
Chelidae this foramen is patent and
transmits a small vein from the eusta-
chian tube (which has its pharyngeal ori-
fice immediately ventral to the foramen)
to the vena capitis lateralis within the
cavum epiptericum. Siebenrock (1897)
erroneously identified this foramen as

Ili ATION

0
= ay,
“r=
&
T=
Hiway

nN
ll

if
|

=
=

wy

Advances in Herpetology and Evolutionary Biology

containing the palatine branch of the
internal carotid, but the palatine artery of
Chelidae branches from the carotid with-
in the carotid canal of the pterygoid, as in
other Testudines.

Triturating surfaces and associated
skull characters (see Fig. 4). In all E.
dentata examined, excepting AMNH
66733 (a hatchling from Lae, still with
egg-caruncle), the horny sheath of the
maxilla has a longitudinal row of tu-
bercles along the center of the triturating
surface (8), sometimes accompanied by

cal AN AN

Figure 3. Ear regions (not to same scale) of Emydura and Pseudemyaura. A. E. dentata, AMNH 62612. B. E.
australis, AMNH 104011. C. E. latisternum, AMNH 27287. D. E. macquaria, AMNH 66737. E. P. umbrina, WAM
R29338. Number (7), chelid foramen (see text discussion).

NEW GUINEA EMYDURA AND THE PLEURODIRAN PENIS - McDowell 175

Figure 4. Skulls of Emydura dentata (A, B, C, all AMNH 62612) and E. australis (D, from AMNH 57586; E, F, G,
from AMNH 104011), showing horny maxillary sheath (A, D, E), palatal view, horny sheath removed (B, F): and
medial view of mandible as if cut on symphysis (C, G). Numbers refer to text discussion.

176

an additional, more medial row of tu-
bercles. In E. australis the homy tritura-
ting surface may be worn smooth (as in
Fig. 4E), but usually has randomly scat-
tered small tubercles (which may, how-
ever, suggest two rows in juveniles, as in
Fig. 4D). In the largest E. dentata skull
examined (AMNH-~ 62611, Bemhard
Camp, Idenburg River, Irian Jaya; condy-
lobasal length 44 mm) there is a distinct
crest on the maxilla underlying the cen-
tral tubercle row (9). In three Papua New
Guinea skulls with condylobasal lengths
of 39 to 43 mm (AMNH 92675, Lae;
104007, Abam, Oriomo River; 99613,
Ambunti, West Sepik Province); there is
a feeble, but discernible, maxillary crest,
although smaller skulls (condylobasal
length 18 to 28 mm) show no crest. In E.
australis the crest is absent in even the
largest skulls examined (AMNH 104011,
Oriomo River, Papua New Guinea;
condylobasal length 43 mm; 108957,
Darwin area, Northem Australia, 46 mm).

The Australian E. dentata examined,
BM [18]75.5.1.8 (adult male, spirit speci-
men), BM[18]76.5.4.7 (juvenile female,
spirit), and BM[18]77.5.19.77 (large adult
skeleton) were all from Gayndah, south-
eastern Queensland. The juvenile (ex-
ternally measured headlength 31.4 mm;
carapace length 124.6 mm) has the ridge
on the triturating surface represented
only by a row of tubercles on the
rhamphotheca, without modification of
the underlying maxillary bone. The adult
skeleton (condylobasal length 44.2 mm),
which was the basis for Boulenger’s
(1889) diagnosis and figure of Elseya,
shows a distinct ridge on the maxilla, as
do New Guinea skulls of the same size.

Although the difference is slight in
small juveniles, E. australis has broader
triturating surfaces than those of E. dent-
ata, this being very marked in the largest
skulls (AMNH 104011 and _ 108957,
where I can detect no difference be-
tween the New Guinea specimen [“sub-
globosa’’| and the Darwin area specimen
[“australis’]). As a consequence, the
median posterior emargination of the
upper horny jaw sheath is longer than its

Advances in Herpetology and Evolutionary Biology

width in E. australis, but no longer than
wide in E. dentata (10). Further, the
choanae are more posterior in E. aus-
tralis than in E. dentata, with the an-
terior choanal rim opposite the center of
the eye in E. australis and opposite the
anterior margin of the eye in E. dentata.
Additionally, in E. dentata, the crests of
the prefrontals separating the olfactory
and orbital chambers are visible from
below through the choanae (11), but in E.
australis the triturating surfaces conceal
these crests. The vomer is stouter in E.
australis than it is in E. dentata, forming
a strong supporting strut for the crushing
triturating surface. In E. australis the
vomer conceals the median contact of the
palatines from beneath (12), whereas the
palatine contact is exposed ventrally in E.
dentata (the value of this character is put
in question by variability in E. latister-
num, where AMNH 69576, Pascoe River,
Queensland, has the interpalatine suture
exposed, but AMNH 26287 [Ravenshoe,
Queensland] and 69572 [Iron Range,
Queensland] have the contact con-
cealed).

The broadening of the mandibular
triturating surface in E. australis results
in an horizontal platform behind the
(rounded) symphysial hook (13), whereas
the triturating surface is confined to the
sloping posterior face of the (pointed)
symphysial hook in E. dentata.

In the Australian E. latisternum and P.
umbrina, the triturating surfaces are nar-
row and smooth, with the choanae an-
terior in position. E. macquaria has scat-
tered inconspicuous tubercles on triturat-
ing surfaces that are narrower than in E.
australis but broader than in the other
forms.

Hyoid. Emydura dentata agrees with
E. australis (and E. latisternum) in hav-
ing ossification of the body of the hyoid
confined to the region from the attach-
ment of the first ceratobranchial to the
attachment of the second ceratobranch-
ial. In E. macquaria and P. umbrina os-
sification extends farther anteriorly,
reaching the rim of the hyoid fenestra.

Shell. I am unable to find any com-

NEW GUINEA EMYDURA AND THE PLEURODIRAN PENIS

pletely reliable extemal character of the
shell distinguishing adult Emydura
dentata from E. australis, and both spe-
cies seem quite variable in shell struc-
ture. In both, the intergular may meet the
pectorals and completely separate the
humerals, as in Pseudemydura (seen in
E. australis: AMNH 59052, Mafulu, and
57586, Daru I.; in E. dentata: AMNH
66734, 66736, Lae). Both usually have a
nuchal (absent in 3 of 45 New Guinea E.
dentata but absent in all 3 Australian
specimens seen), an axillary (absent in 4
of 44 E. dentata and 5 of 15 E. australis)
and an inguinal (absent in 7 of 15 E. aus-
tralis). Contrary to Goode’s (1967) char-
acterization of “kreffti,” all my E. aus-
tralis have the carapace distinctly flared
over the hind legs and widest across the
eighth marginals except for AMNH
104339 (Manning Creek near Mt. Ber-
nard Homestead, Western Australia),
which has the carapace unflared and
widest across the seventh marginals.
Most E. australis show the transparency
of the carapacial scutes noted by Goode
(1967) for “kreffti,” but this is even more
marked in most New Guinea E. dentata.
I have been unable to find any constant
differences between the Emydura avail-
able to me in the proportions of the inter-
gular.

Juvenile E. australis (AMNH 57586,
Daru I., carapace length 90 mm; 57587,
same, 109 mm; 92967, Fly R., 108 mm;
104339, Western Australia, 105 mm)
agree with adults in having a smooth rim
of the carapace. But in E. dentata, the
edge of the carapace is conspicuously
serrated in 17 specimens with a carapace
of 58 to 92 mm (except AMNH 99621, W.
Sepik Prov., 79 mm), serrated or not in 10
specimens (including a strongly serrated

Queensland specimen, 2 strongly ser-_

rated New Guinea specimens) with cara-
pace 104 to 126 mm, smooth or only
faintly serrated in 17 specimens with
carapace 156 to 252 mm (including a
nearly smooth Queensland specimen;
most nearly “serrate” is AMNH 104837,
Western Prov., 184 mm). The three small-
est specimens (AMNH 66733, 66735,

- McDowell aa

99622) have a nearly smooth carapace rim
(carapace length 31 to 52 mm).

E. latisternum, supposedly the closest
relative of E. dentata, is less like New
Guinea E. dentata than are New Guinea
E. australis in several shell features: the
nuchal is absent (a vestige, excluded
from margin, in AMNH 69576, Pascoe
River, Queensland); the inguinal is ab-
sent (vestige on right in AMNH 69573,
Pascoe River); axillary absent except in 3
of 13 specimens (all three, AMNH
69569-71, from Iron Range, Queens-
land); the carapace is conspicuously ser-
rated at all ages; and the bridge is much
narrower than New Guinea E. australis
and E. dentata. With B = width of left
plastral bridge and P = maximum length
of plastron (measurements to nearest
mm), regression equations are as follows:

New Guinea E. dentata (N = 43) B =
leet Ook (Sae—slk6) hs

New Guinea E. australis (N = 11) B=
=H ae Oustele (Sj, = DIL)

. Queensland E. latisternum (N =
B = —5.9 + 0.30P (SB = 1.8)

13)

RELATIONSHIP OF EMYDURA DENTATA
AND E. AUSTRALIS

Based on my investigation, Emydura
australis (including “kreffti’ and “sub-
globosa’’) seems the closest relative of E.
dentata (the type species of “Elseya’).
Because of this, the generic recognition
of Elseya seems unwarranted and does
not appear to express the apparent phy-
logenetic relationships of these turtles. I
therefore synonymize Elseya with
Emydura.

Burbidge et al. (1974: 401, 402) give
stereo pair graphs of the relationships of
Australo-New Guinean chelids, first on
the basis of a correlation matrix of all
their serological data, then with data on
anti- Elseya” latisternum sera omitted.
In both graphs the cluster of E. australis
(including “E. kreffti? and “E. sub-
globosa’) seems nearest to that for E.
dentata (“Elseya dentata’ and “E.

178

novaeguineae ), a result in accord with
my morphological findings, provided
great weight is given to the pale temporal
spot and to occlusion of the chelid
(eustachian tube vein) foramen.

The serological findings of Burbidge et
al. do not appear to support Goode’s
(1967) division into Elseya and Emydura
at the generic level. However, when anti-
“Elseya’ latisternum serum data were
omitted, E. latisternum fell close to
“Elseya dentata” and “E. novwae-
guineae,’ a result in accord with my own
observations of penial structure (unfortu-
nately, I do not have data on the penis of
E. macquaria or P. umbrina). That their
preliminary analysis showed E. lat-
isternum to be quite distant from all
other Emydura (sensu lato) may not be
the result of experimental artifact, for it
agrees with my morphological findings,
that E. latisternum differs sharply from
E. dentata in having an open chelid
foramen, in lacking a pale temporal mark,
lacking axillary and inguinal scutes and
in having smooth homy triturating sur-
faces at all ages. The resemblances be-
tween E. dentata and E. latisternum
stressed by Goode (1967) in his concept
of “Elseya” may simply be primitive
characters retained by these species but
lost in the species constituting Goode’s
Emydura. This is supported by the fact
that P. umbrina, omitting its unique fea-
tures, would key out as “Elseya.” Al-
though paleontological evidence is lack-
ing, the large number of unique special-
izations and the isolated distribution
(around Perth) of P. umbrina suggest that
it separated phylogenetically from other
Emydura-like terrapins long ago and
might be expected to retain character-
istics of early, ancestral Emydura.

In both graphs given by Burbidge et
al., E. macquaria seems isolated, but is
nearer to E. australis than to E. dentata
(in keeping with the morphology of the
triturating surfaces and the flat skin of the
temporal region). In the first graph, in-
cluding data on anti-E. latisternum fac-

Advances in Herpetology and Evolutionary Biology

tors, E. macquaria agrees with E. lat-
isternum in position on the left-right axis,
but widely separated on other axes. My
own observations suggest that E. lat-
isternum is closely related to E. mac-
quaria. Of 13 E. latisternum examined,
the four smallest (AMNH 69567, 69573—
4, 69576; carapace length 97 to 144 mm)
show a strongly defined ventrolateral
white stripe on the neck, cited by Goode
(1967) as a characteristic of E. macquaria,
while Goode notes that older specimens
of the latter have a thickly cornified
crown (as in E. dentata, E. latisternum,
and P. umbrina). In having a patent
chelid foramen and a dark and unicolor
temporal region, E. latisternum and E.
macquaria resemble each other (and P.
umbrina) and differ from E. dentata and
E. australis.

Pseudemydura umbrina agrees with E.
macquaria in the extensive ossification
of the body of the hyoid and in its
southem distribution, although P.
umbrina is southwestern and E. mac-
quaria southeastern in distribution.
Pseudemydura is strikingly characterized
by several unique features. These in-
clude: extreme flattening of the skull that
causes the parietal to rest upon the
quadrate (but without an interdigitating
suture); contact of the supratemporal
(“squamosal” of most authors) with the
transversely expanded supraoccipital;
and constant enlargement of the inter-
gular to separate the humerals. In spite of
these peculiarities, it seems closely re-
lated to Emydura and I am not certain
that full generic (as opposed to sub-
generic) rank is necessary for Pseud-
emydura. The graphs of Burbidge et al.
show P. umbrina far separated from all
Emydura, but the nearest approach is to
E. dentata, a species showing separation
of the humerals by the intergular as a
variation (see above) and with narrow
triturating surfaces (as in P. umbrina). I
believe this is best explained by con-
sidering E. dentata the most primitive
species of Emydura.

NEW GUINEA EMYDURA AND THE PLEURODIRAN PENIS

THE BEARING OF EMYDURA DENTATA
ON CHELID PHYLOGENY

In the classification of Gaffney (1975),
the Infraorder Pleurodira contains two
families, Pelomedusidae and Chelidae.
The former seems the more primitive in
retaining mesoplastra in the shell and a
squamosal (=quadratojugal of most
authors) in the skull. In Pelomedusidae
(Podocnemis unifilis, Pelomedusa sub-
rufa, Pelusios derbianus examined) the
retractor penis muscles arise from the
sacral region, as in Cryptodira and prob-
ably the primitive condition. However,
in Chelidae the retractor penis is reduced
and arises from the transverse ventral
musculature anterior to the cloaca
(Emydura dentata, E. australis, E. lat-
isternum, Phrynops gibbus, P. geof-
froanus, Platemys platycephala, Chelus
fimbriatus, Chelodina novaeguineae, C.
expansa [=C. “rugosa’| examined). On
the other hand, in having nasal bones
(except in Chelus) and in usually having
a suturally distinct splenial, the Chelidae
seem more primitive than the Pelomedu-
sidae. (The splenial is variable within
New Guinea Emydura dentata and was
absent in three skulls examined [AMNH
92965 and 92971, Fly R., and 104007,
Oriomo R.] but present in five others.)

The genera Emydura and _ Pseud-
emydura are apparantly more primitive
than other Chelidae, more like the Pelo-
medusidae in several features: 1) the
dentaries are fused at the symphysis; 2)
the supraoccipital forms the dorsal rim of
the foramen magnum and is the most
backwardly projecting bone of the skull;
3) the anteriormost part of the temporal
fossa is roofed over by a transverse shell
of parietal from the ventral surface of
which short and vertical jaw adductor

fibres arise (seen also in Platemys; in-

other Chelidae the jaw adductor is
formed entirely of long and stretchable
fibres, allowing a wide gape of the mouth
at the expense of lessened crushing force
of the jaws); 4) there is a scar on the

- McDowell 179

pharyngeal surface of the pterygoid bone
for the ventral pterygoideus muscle (this
muscle seems to be absent in other
Chelidae except Platemys); 5) the flange
of the quadrate suspending the eus-
tachian tube is extended posteromedial
to the tube to meet the cartilaginous
border of the fenestra ovalis (not ex-
tended beyond the eustachian tube in
other Chelidae; Pelomedusa and Pelusios
are like Emydura, but in Podocnemis and
its close relatives, this flange is still bet-
ter developed as a firm brace against the
basisphenoid and basioccipital); and 6)
the cervical vertebrae have normal zyga-
pophysial articulations, oriented horizon-
tally (cylindroid or vertically oriented in
other Chelidae).

Among Emydura, E. dentata shows
special similarity to Podocnemis and its
close relatives in having a longitudinal
ridge on the maxillary triturating surface,
present as a tuberculate keel on the
horny sheath in all but hatchlings, and
with bony ridging of the maxilla in large
adults. E. dentata seems unique among
Chelidae in its ridged triturating surface
and, as the most Podocnemis-like of
Chelidae, seems the most primitive spe-
cies of the family structurally, except in
lacking neural bones (true of all Em-
ydura). As in the other Emydura ex-
amined (E. australis, E. latisternum), the
cloacal bursa of E. dentata is formed as in
Podocnemis: a bilobate sac with a single
median dorsal aperture into the cloaca
and with a lining of coarse papillae. The
simplest explanation for special re-
semblances between Podocnemis and E.
dentata is that these are primitive fea-
tures for Pleurodira that have been lost or
modified in other forms.

For geographical reasons the Australo-
New Guinean genus Chelodina might be
expected to show special relationship to
Emydura and Pseudemydura, but no
morphological evidence corroborates
this. Rather, as shown convincingly by
Gaffney (1977), Emydura and Chelodina
seem at opposite extremes, with the

180

South American Chelidae serving as
intermediates. Chelodina (C. longicollis,
C. novaeguineae, C. expansa [C.
“rugosa |, C. steindachneri examined)
has all of the specializations that set
South American chelids apart from Em-
ydura and Pseudemydura (persistent
suture between dentaries, supraoccipital
excluded from foramen magnum and
exceeded by exoccipitals in backward
extent, vertical cervical zygapophysial
articulations, etc.). In addition, Chelo-
dina adds some peculiarities of its own:
1) fusion of the frontals (perhaps unique
in Testudines); 2) only four claws on
hand (seen also in the South American
Hydromedusa); 3) greatly elongated cer-
vical vertebrae (as in the South American
Chelus and Hydromedusa); 4) total re-
duction of roofing of the temporal fossa
(approached by some South American
forms, e.g., Hydromedusa, and the exact
opposite of Emydura and_ Pseud-
emydura); and 5) cloacal bursae paired,
completely separated, and with smooth
lining (most nearly approached by
Chelus, where the lining is smooth and
the bursae fused only at the point of entry
into the cloaca). The only special re-
semblance between Emydura and
Chelodina, absence of neural bones, is
also seen in some South American forms
(e.g., Platemys) and is probably conver-
gent, since Burbidge et al. (1974) report
that Chelodina oblonga has five to eight
neural bones. Two possible interpreta-
tions of these morphological resem-
blances and differences are:

1) Chelodina arose in Australia from
Emydura through a series of intermedi-
ates that have become extinct. However,
one of these intermediates invaded South
America to give rise to the South Ameri-
can Chelidae. Any special resemblances
between some South American forms and
Chelodina are fortuitous.

2) Some primitive Emydura-like form
invaded South America from Australia,
and gave rise to the South American
chelid radiation, including a long-necked
line (Hydromedusa and Chelus). A primi-

Advances in Herpetology and Evolutionary Biology

tive member of the long-necked line re-
invaded Australia and gave rise to Chelo-
dina. Special resemblances between
Chelodina and certain South American
Chelidae are the result of close common
ancestry.

The latter alternative seems now to
have greater evidence, but the first alter-
native cannot yet be excluded. Real fos-
sils, rather than hypothetical common
ancestors, will have to be brought into
the discussion before the decision can be
considered a realistic examination of
nature rather than a statement of the rules
of irreversability and minimum coinci-
dence that we impose (as a necessary
control) on evolutionary hypotheses.

THE MORPHOLOGY OF THE PENIS OF
PLEURODIRA

Although Zug (1966) has described the
penis of cryptodires and shown the use-
fulness of penial morphology in the clas-
sification of that group, there appears to
be no account of the morphology of the
penis of Pleurodira, aside from
Schmidtgen’s (1907) account of the un-
protruded organ of Chelodina longicollis.
(Goode [1967] has figured what I would
consider the partial protrusion of the
penis of Emydura australis.) As noted in
the previous section of this paper, the
penis appears to offer very useful taxo-
nomic characters within the pleurodires.
Further, the morphology of the pleu-
rodiran penis is quite different (but not
fundamentally) from that of cryptodires
and suggests how the protrusile penis of
cryptodires might have arisen from an
ornamented cloacal evert, such as that of
living gymnophione Lissamphibia (and,
presumably, the earliest reptiles).

As in Cryptodira, the most ventral por-
tion of the penis of Pleurodira is formed
by the corpus fibrosum. The corpus fi-
brosum of pleurodires differs from that of
cryptodires in not being cavernous or
erectile in texture; rather, it is dense and
cartilagelike in consistency (but not hy-

NEW GUINEA EMYDURA AND THE PLEURODIRAN PENIS

aline cartilage). For most or all of its
length, the corpus fibrosum of pleu-
rodires is attached to the floor of the
cloaca. This attachment is complete in
Podocnemis, Chelodina, and Phrynops
geoffroanus. The extreme posterior end
of the corpus fibrosum projects as a short
free tip (as in Emydura dentata and E.
latisternum, as described in the previous
section) in Pelusios, Chelus, and Phry-
nops gibbus. There is a longer free pos-
terior tip (as in Emydura australis, as
described in the previous section) in
Pelomedusa. In Platemys the free tip of
the corpus fibrosum is much longer than
in the other forms examined, but still
does not include the portion of the organ
bearing the seminal groove.

The very spongy and obviously erec-
tile tissue of the corpus cavernosum is
only sparsely distributed over the corpus
fibrosum of pleurodires, as compared
with cryptodires, a point noted by
Schmidtgen (1907: 391) for Chelodina
longicollis. However, the corpora caver-
nosa are continued forward into the body
cavity as a pair of very large proximal
bulbs of the cavernosum that lie anterior
(proximal) to the corpus fibrosum. The
paired and narrow peritoneal canals
(essentially similar to those of crypto-
dires) pass dorsomedial to the bulbs of
the cavernosum to lie close to, and paral-
lel with, the seminal groove.

The seminal groove is simple in all the
pleurodires examined and runs _pos-
teriorly from the urodaeal chamber of the
cloaca along the dorsal surface of the
penis to about the middle of the corpus
fibrosum. The seminal groove lies in a
longitudinal trough of cavernosum tissue,
but this is not a conspicuous mass of
cavernous tissue (there seems to be just

enough to form lips defining the seminal -

groove). I feel reasonably confident that
this trough of erectile tissue defining the
seminal groove is homologous to the
plica interna of Cryptodira, in the ter-
minology of Zug (1966). I am also confi-
dent of the homology of the corpus fi-
brosum, the corpus cavernosum (as an

- McDowell 181

entity—not its parts), the seminal groove,
and the peritoneal canals of pleurodires
with the correspondingly named parts of
the cryptodiran penis. Beyond these, I
have very little confidence in the
homologies of pleurodiran to cryptodiran
penial structures, because of the very dif-
ferent appearance of the organ in the two
groups (particularly when protruded),
and my use of Zug’s terminology must be
considered very tentative.

The distal (posterior) end of the semi-
nal groove opens (in the unprotruded
organ) into a U-shaped basin that lies dis-
tal as well as lateral to the seminal groove
and plica interna. Zug (1966) finds three
separate and paired pits or sinuses in
cryptodires, but I find only this U-shaped
depression in pleurodires, although this
depression may show pocketlike exten-
sions that would seem to correspond to
the clearly defined sinuses of crypto-
dires. I term this single and U-shaped
depression of pleurodires the sinus
communis. The main portion of caver-
nous tissue of the pleurodiran penis lies
on the outer rim of the sinus communis;
thus the sinus communis isolates the
plica interna (and seminal groove) from
the remainder of the corpus cavernosum.
The anterior (i.e., proximal) ends of the
sinus communis lie near the posterior
(i.e., distal) ends of the paired peritoneal
canals and it is tempting (but without any
supporting developmental evidence) to
regard the sinus communis as a depres-
sion resulting from the collapse of an
underlying posterior continuation (and
median fusion, distal to the tip of the
seminal groove) of the peritoneal canals.

Schmidtgen (1907) found the perito-
neal canal to end blindly in all Tes-
tudines he studied (both male and fe-
male), but the point of approximation of
the tip of the peritoneal canal to the sur-
face of the penis to be marked by a pale
spot in Trionyx. I also have failed to find
any patent opening of peritoneal canal,
but found a yellow spot in Podocnemis.
Thus, present evidence (including all of
my own observations) favors Schmidt-

182

gens belief that the peritoneal canal
ends blindly and that reports of an open
pore are based on artifacts, but this ques-
tion must still be considered open. It is
possible that a patent pore may be of
seasonal occurrence, or perhaps only
develops during copulation. A pore
might be confined to females (with a
clitoris very similar to the penis of males,
but much smaller), since my own and
Schmidtgen’s dissections and histologi-
cal sections have been carried out on
males (this appears to be the case with
Zug’s dissections, as well). If the peri-
toneal canal is ever open to the outside,
then the sinus communis might serve as
the receiving chamber for whatever fluid
might issue from this pore. It should be
remembered that the precise stimulus to
ejaculation by the copulating male is
unknown for any turtle.

The cavemous tissue forming the outer
rim of the sinus communis is differenti-
ated at its middle—that is, at the level of
the distal end of the seminal groove and
plica interna—into a distinct mound,
usually (Platemys is the exception) bear-
ing a horn-like prominence. Tentatively,
I identify this mound as the homologue
in Pleurodira of the plica medialis (Zug’s
terminology) of testudinoids and term the
hornlike prominence the cornu of the
plica medialis. In the Pelomedusidae
examined, the corpus fibrosum has a
distinct lateral lobe supporting the plica
medialis and its comu and the cornu is
outwardly directed in the unprotruded
organ. The cornu of Podocnemis appears
to be bifid, with a longer and more acute
anterior lobe exactly opposite the tip of

Advances in Herpetology and Evolutionary Biology

the seminal groove, and with a blunter
posterior lobe on the lateral border of the
corpus fibrosum distinctly posterior (dis-
tal) to the level of the tip of the seminal
groove (but this posterior lobe perhaps
belongs to the posterior plica externa, to
be discussed below). In Pelomedusa there
is a single comu of the plica medialis,
supported by a homlike process of the
corpus fibrosum and extending antero-
laterally in the unprotruded organ, but an
oblique fleshy fold extends postero-
medially from the apex of the comu to
end blindly in the sinus communis, well
posterior (distal) to the tip of the seminal
groove. This oblique ridge may be the
homologue of the posterior lobe of the
cornu of the plica medialis of Pod-
ocnemis; the sinus communis forms a
deep, posteriorly directed pocket be-
tween the posterior end of this oblique
ridge and the posterior plica externa (this
pocket perhaps represents the “posterior
sinus of the glans” of testudinoids, in
Zug's terminology). In Pelusios the plica
medialis is a longitudinally ovoid swel-
ling lateral to the distal end of the semi-
nal groove and plica interna, with a blunt
free posterior end that probably repre-
sents the cornu.

In Chelidae, the corpus fibrosum does
not have any supporting lobe for the plica
medialis or its cornu. In Platemys, the
plica medialis is simply a low, longitudi-
nal ridge without a projecting commu. In
the others examined, the cornu is a quite
distinct free process, but it lies (in the
unprotruded organ) in the sinus com-
munis, with its free end directed medi-
ally (rather than laterally as seen in

=

Figure 5. Penis of: A. Podocnemis unifilis (AMNH 87966), unprotruded. B. Pelomedusa subrufa (AMNH 50744),
unprotruded. C. Pelusios derbianus (AMNH 50746), unprotruded. D. Chelodina expansa (“rugosa”) (AMNH
86548), unprotruded. E. Phrynops gibbus (AMNH 61521), protruded. F. Platemys platycephala (AMNH 61529),

protruded.

Abbreviations as in Fig. 2, as well as: alc, anterior lobe of cornu of plica medialis; c, tip of cornu of plica medialis;
plc, posterior lobe of cornu of plica medialis; psg, posterior sinus of glans; ssg, sinus of seminal groove; ys, yellow
spot marking termination of coelomic canal. The plica interna is present in F, but concealed by the plicae mediales

and everted cloacal lining.

183

- McDowell

NEw GUINEA EMYDURA AND THE PLEURODIRAN PENIS

bat

OS wT
LM

==

AAU et SP

RD Rp HH)
. ", d i
4
a aS —=* = =
NN | SB
hein le .
ira)
a °
ae eres 2
@

5

Z 3
a So (Sg ses
a shee

ee

hy
t

=a

(Masss ier

184

Pelomedusa and Podocnemis or dorso-
laterally in Pelusios). However, in the
naturally protruded organs of Phrynops
gibbus and Emydura australis, the comu
is rotated to extend laterally, and this is
probably true of the other Chelidae, as
well (manipulation of preserved chelid
penes, by bending them at the level of
the tip of the seminal groove, as in the
naturally protruded specimens available,
produces the lateral rotation of the
cornu). Probably the comu of the plica
medialis functions as a holdfast to pre-
vent dislodgement of the penis before
the completion of copulation.

In most pleurodires, the apparent
homologue of the cryptodiran plica ex-
terna is divided by the plica medialis into
two separate portions: 1) the plica ex-
terna anterior, proximal to the plica
medialis and with its posterior (distal)
end joined to the anterior end of the plica
medialis; and 2) the plica externa pos-
terior, lying on the lateral edges of the
most posterior (distal) part of the corpus
fibrosum, with its anterior (proximal) end
joined to the plica medialis. All Chelidae
examined and Pelomedusa show this
linear order of corpus cavernosum lobes
on the lateral rim of the sinus communis,
arranged (from anterior to posterior);
plica extema anterior, plica medialis,
plica externa posterior. In Pelusios, how-
ever, the plica externa anterior and plica
externa posterior join lateral to the plica
medialis. I am uncertain about the cor-
rect interpretation of Podocnemis, where
the “posterior lobe of the cornu of the
plica medialis” is joined basally to the
plica externa anterior, lateral to the “an-
terior lobe of the cornu of the plica
medialis.” If this “posterior lobe of the
cornu’ is really a lateral wing of the
proximal part of the plica externa pos-
terior, then Podocnemis would be like
Pelusios and cryptodires in having the
two portions of the plica externa continu-
ous lateral to the plica medialis.

In all pleurodires examined except
Platemys, the proximal end of the plica
externa posterior has a blunt lobe that

Advances in Herpetology and Evolutionary Biology

lies in the more distal (posterior) part of
the sinus communis of the unprotruded
organ. In at least Phrynops gibbus and
Emydura (probably in the others, as
well), this lobe swings to a ventrally
directed vertical position in the pro-
truded organ. The lobe probably func-
tions, along with the cornu of the plica
medialis, as a holdfast during copulation
(in Phrynops gibbus and P. geoffroanus
there are coarse and hard denticulations
on this lobe of plica externa posterior,
further suggesting that it acts as a hold-
fast).

In all the pleurodires examined except
the juvenile Chelus (the adult agreed
with the majority), a paired anterior
extension of the sinus communis extends
proximally between the plica interna and
the plica externa anterior. This would
seem to represent the “sinus of the semi-
nal groove” of cryptodires, using Zug’s
terminology. In Podocnemis, there is a
shallower and much less clearly defined
depression (continuous with the sinus
communis) between the plica medialis
and plica externa anterior. This perhaps
represents the “anterior sinus of the
glans” of Zug’s terminology for cryp-
todires.

Zug’s terminology was intended for the
description of only a small terminal part
of the cryptodiran penis, the “glans.”
Much of the length of the cryptodiran
penis, quite apart from the proximal
bulbs of the corpora cavemosa, is formed
by a shaft region in which the only struc-
ture formed by the corpus cavemmosum is
a paired seminal ridge (Zug’s term),
flanking the seminal groove. In Pleuro-
dira, the penis has no shaft and the entire
structure, apart from the bulbs of the cor-
pora cavernosa, seems to represent the
“glans” of the cryptodiran organ (in this
respect it resembles the clitoris of
cryptodires).

I have seen the naturally protruded
penis of only Emydura australis,
Phrynops gibbus, and Platemys platy-
cephala among pleurodires, but some
morphological points lead me to believe

NEW GUINEA EMYDURA AND THE PLEURODIRAN PENIS - McDowell

that the organ must function in a similar
manner in the other Pleurodira. In the
naturally protruded pleurodiran penis
seen, the organ is doubled over, so that
the tip of the seminal groove and plica
interna lie at the tip of the organ, flanked
by the plicae mediales, with the plica
externa posterior now inverted to lie on
the ventral side. The tip of the corpus
fibrosum, directed posteriorly in the un-
protruded state, is now pointed forward
because of the doubling-over of the organ
that has taken place at the level of the
plicae mediales (slightly behind the level
of the tip of the seminal groove of the
unprotruded organ, with some enlarge-
ment of the plicae internae that bear the
tip of the seminal groove accounting for
the extreme terminal position of the tip of
the seminal groove in the fully protruded
organ). That the seminal groove extends
only about halfway along the penis in all
pleurodires examined can easily be ac-
counted for if doubling-over of the penis
is characteristic of pleurodires generally.

The protruded penis is only a part of
the copulatory organ of pleurodires. At
least to judge from Emydura, Phrynops,
and Platemys, the entire copulatory
organ of pleurodires is an eversion of the
cloacal lining, to which the penis is en-
tirely, or almost entirely, attached. The
dorsal cloacal lining is drawn out with
the penis and forms a fold over the semi-
nal groove that is a nearly complete roof
for the groove. Thus, although the penis
of Cryptodira is extruded (Zug’s term) to
form the copulatory organ as in cro-
codilians, the copulatory organ of Pleu-
rodira is a cloacal evert as in Gymn-
ophiona, with the penis forming only a
specialized part of this evert.

The mechanism of penial protrusion in
Pleurodira. So far as I can determine
from the study of dead material, the penis
of pleurodires acts as a lever to evert the
cloaca; protrusion of the penis must
necessarily produce eversion of the
cloaca because the penis is attached to
the cloacal wall for almost its entire
length. The mechanism for protrusion of

185

the penis, in turn, would seem to be warp-
ing. The penis is formed of two layers
that are very tightly bound together: a
ventral layer, the corpus fibrosum, that is
flexible but probably unexpandable be-
cause it is densely fibrous without blood
spaces; and a dorsal layer of erectile tis-
sue, the corpus cavemosum, that has
large blood spaces and can increase in
size if the rate of blood flow into it ex-
ceeds the rate of outward blood flow.
When the corpus cavernosum expands
and the corpus fibrosum does not, the
penis must bend like a heated thermo-
couple, with the more elongate corpus
cavernosum on the outside of the curva-
ture and the shorter corpus fibrosum on
the inside of the curve.

Because the blood spaces within the
corpus cavernosum are connected with
one another, they would make up collec-
tively a fluid chamber subject to Pascal’s
law, with pressure exerted equally in all
directions. Therefore, the pressure at the
proximal bulbs of the corpus cavemosum
would be equal to that at the flexure-
region of the penis, and pressure on these
bulbs from the surrounding coelomic
fluid could increase the difference in
length between corpus cavernosum and
corpus fibrosum (thus increasing flexure
of the entire organ), as well as pressing
the entire penis—and cloacal lining
bound to the penis—outward. I do not
know of any special mechanism for in-
creasing coelomic pressure, although
contraction of the transverse musculature
in the region of the cloaca could do so.

The cryptodiran penis is freed at its
posterior end from the cloacal wall, so
that the cloaca is not everted when the
penis protrudes, and penial protrusion
appears to depend mainly or entirely on

_ the enlargement of the organ through dis-

tension of cavernous spaces by blood.
The difference from pleurodires is not
quite absolute, since in a number of cryp-
todires preserved with the penis pro-
truded, I have found a small eversion of
the cloacal lining at the base of the organ,
although not enough tissue is everted to

186

roof the seminal groove and there is a
vast difference in degree of cloacal ever-
sion from that observed in pleurodires;
examples of cryptodires with a small
eversion of the cloacal lining at the base
of the protruded penis are: Trionyx ater
(AMNH_ 88879); Platysternon mega-
cephalum (AMNH 30123); Clemmys
guttata (AMNH 7453). Because the
corpus fibrosum is not as expansible as
the corpus cavernosum in cryptodires, a
number of protruded cryptodiran penes
examined show a warping that causes the
distal tip to be bent downward (but not
the doubling over of the organ observed
in pleurodires); examples are: Clemmys
guttata (AMNH 7453), Ocadia sinensis
(AMNH_ 30190), Kinosternon hirtipes
(AMNH 71363). In Trionyx, the four
terminal branches of the seminal groove
(characteristic of the family) are con-
tained in four corresponding lobes of the
corpus fibrosum, with corpus caver-
nosum tissue firmly attached to the ven-
tral side of each lobe; accordingly, the
warping produced by the differential ex-
pansion of the two tissues causes the
sulcus-bearing lobes to arch upward at
their tips (this gives some support to the
mechanism here suggested).

I have not seen a protruded chelonioid
penis and the only chelonioid penis
examined (Chelonia mydas, AMNH
107114) suggests that the mechanism
may be somewhat different from that in
other Cryptodira. The free tip of the
organ, formed from the corpus caver-
nosum and bearing the seminal groove,
lies above (but free of) a distal (posterior)
projection of corpus fibrosum that is en-
tirely attached to the cloacal lining. I am
unable to guess whether: a) penial pro-
trusion is entirely the result of enlarge-
ment of the corpus cavernosum, with the
corpus fibrosum acting as an immobile
base; or b) full protrusion involves an
external (backward) displacement of the
corpus fibrosum, which, because of its
total attachment to the cloacal lining,
would drag the cloacal lining with it and
produce an extensive cloacal eversion,
perhaps approaching that of pleurodires.

Advances in Herpetology and Evolutionary Biology

The phylogenetic position of the
pleurodiran penis. A number of mor-
phologists, but most unequivocally
Tonutti (1932), have presented argu-
ments that the most primitive copulatory
organ among tetrapods is the eversible
cloaca of male caecilians (=Gym-
nophiona). The cleidoic egg, usually con-
sidered the distinctive specialization of
the earliest reptiles, implies the exist-
ence of internal fertilization, and thus
makes it likely some sort of copulatory
organ was present. The only living
Rhynchocephalian, Sphenodon, has no
copulatory organ and is alleged to evert
the cloaca (but I have been unable to find
a circumstantial account of copulation).
Most birds lack a copulatory organ, but
some (Struthionidae, Rheidae*, Casu-
ariidae* including Dromaeus, Apterygi-
dae, Tinamidae, Cracidae*, Anatidae*,
Otididae, Burhinidae, Threskiornithidae,
and Phoenicopteridae) have a solid pro-
trusible penis on the floor of the cloaca
(this penis, in turn, contains an eversible
sac, the Blindslauch, in the families
marked with an asterisk [*] above).
Probably the absence of a penis in most
birds is a secondary reduction. Crocodi-
lia, cryptodiran turtles and mammals also
have a solid and protrusible penis formed
from the cloacal floor. Squamata, on the
other hand, have a paired eversible sac
(the hemipenis) formed from the pos-
terolateral cloacal wall. A considerable
part of the attractiveness of considering
the caecilian cloacal evert as the primi-
tive tetraped copulatory organ stems from
the fact that by specializations that are
easily imagined, the simple cloacal evert
can give rise to the different kinds of
copulatory organs known in amniotes,
but it is difficult to imagine how the
squamatan hemipenis could either arise
from or give rise to the median pro-
trusible penis of crocodilians, cryp-
todires, or mammals.

The copulatory organ of at least
Emydura, Phrynops, and Platemys (and
probably all Pleurodira) is a cloacal evert,
essentially like that of caecilians but with
the simple holdfast welts and tubercles of

NEW GUINEA EMYDURA AND THE PLEURODIRAN PENIS : McDowell

the caecilian evert replaced by a unified
ventral structure, the penis, containing a
differentiated corpus fibrosum, paired
corpus cavernosum, and paired coelomic
diverticula. It is relatively easy to derive
the cryptodiran penis from that of pleu-
rodires, and thus from a cloacal evert; and
so, the pleurodiran penis gives some
support to the general theorem of com-
parative anatomists that the primitive
copulatory organ of tetrapods is a cloacal
evert. If this phylogenetic interpretation
is accepted, then the resemblance in
general penial structure between croco-
dilians and cryptodires must be con-
sidered convergent, since this interpreta-
tion would mean the protrusible (as op-
posed to eversible) organ of cryptodires
arose within the Testudines, after the
many specializations (e.g., shell, middle
ear, lack of lachrymal duct) that rule out
Testudines as ancestors of any archo-
saurian group. By this interpretation,
conversion of a cloacal evert to a pro-
trusible organ took place at least twice,
and quite possibly more often than that.

This interpretation is not bothersome
to me, but it should be remembered that
there is no proof the phylogeny did not
go the other way, from a protrusible cryp-
todire-like penis to a secondarily de-
veloped cloacal evert in Pleurodira. It is
very unlikely that any documentation
from the fossil record will ever indicate
the direction of evolution of the penis in
Testudines. This is the fundamental dif-
ficulty in the use of soft tissue anatomy in
higher classification. Though the charac-
ters of the reproductive system, circula-
tion, and pathways in the central nervous
system may have been much more im-
portant to the organisms than the details
of skeletal structure, and may have been

the most important factors in determining .

the success or failure of evolutionary
lineages, they are rarely testable against
the fossil record. This much does seem to
fit with what can be tested (at least for
general plausability): 1) pleurodires and
cryptodires diverged very early in the
history of Testudines and this divergence
is associated among living forms with

187

marked differences in the copulatory
organ (that is, structure of the copulatory
organ proves its usefulness as a taxo-
nomic character in turtles by indicating a
difference between two groups that are
known from other evidence to be quite
different); 2) just as Zug’s (1966) survey
of the penis of cryptodires divides that
group concordantly with Gaffney’s (1975)
division of the cryptodires on skull ana-
tomy, there appear to be penial differ-
ences between the two (osteologically)
defined families of living pleurodires;
and 3) species can be sorted out—at least
in some cases, such as the New Guinea
species of Emydura—by differences in
penial morphology. It seems a reasonable
expectation, then, that penial anatomy
will be as useful to the taxonomy of
Testudines, at all levels, as has been the
anatomy of the hemipenis to squamatan
taxonomy.

ACKNOWLEDGMENTS

I am most grateful to E. S. Gaffney
(Vertebrate Paleontology) and R. G.
Zweifel (Herpetology) for access to
material in their care at the American
Museum of Natural History (AMNH),
including certain specimens on loan to
them, from the Western Australian
Museum (WAM) and the Institut Royale
des Sciences Naturelles, Bruxelles
(IRSN). I am further indebted to A.
Grandison of the British Museum (Na-
tural History) (BM) for use of material
in her care.

APPENDIX: SPECIMENS EXAMINED

EMYDURA AND PSEUDEMYDURA EXAMINED

Emydura dentata. IRIAN JAYA: Kurik, near
Merauke: IRSN unnumb., 3: Bemard Camp:
AMNH 62611-13: “Dutch New Guinea”: AMNH
62043. PAPUA NEW GUINEA: (Western
Province): Fly River: AMNH 9296466, 92968-74;
above d’Albertis Junction: AMNH 59927; Abam,
Oriomo R.: AMNH 104007-08; Zim, Oriomo R::
AMNH 104009; Wipim: AMNH 104837-38; (West
Sepik Province): Ambunti: AMNH 99612-14,

188

99618-20: Wagu: AMNH 99615-16, 99621-22;
Hunstein R.: AMNH 99617; Lumi: AMNH 100002-
04; (Madang Province): near Alexishafen: Peabody
Mus. Field no. 0103: (Morobe Province): Oomsis
Creek: AMNH 92676; Lae: AMNH 66733-36,
92675, 103629; Wau: AMNH 103630. QUEENS-
LAND: Gayndah: BM [18]75.5.1.8, [18]76.5.4.7,
[18]77.5.19.77.

Emydura australis. PAPUA NEW GUINEA:
(Western Province): Daru: AMNH 57586-88; Fly
River: AMNH 92967; Zim, Oriomo R.: AMNH
104839: Oriomo, Oriomo R.: AMNH 104011; (Cen-
tral Province): Mafulu: AMNH 59050-52, 59910;
Baroko: AMNH 57898. Australia: (Wester Austra-
lia): Manning Creek: AMNH 104339; (Northern
Australia): Mainoru Station: AMNH 87878; Darwin
area: AMNH_ 108957; (Queensland): Wenlock:
AMNH 69578.

Emydura latisternum. AUSTRALIA (Queens-
land): Larradeenya Creek: AMNH 69567; Brown’s
Creek, Pascoe R.: AMNH 69573-77; Iron Range:
AMNH 69569-72; Speewah: AMNH 69566; near
Ravenshoe: AMNH 27287-88.

Emydura macquaria. AUSTRALIA: (Queens-
land): Brisbane: AMNH 103701; (Victoria): Patho,
20 mi. W of Echuca: AMNH 108961; Nancy R.::
AMNH 103702; 40 mi. SE of Mildura; AMNH
110488; (South Australia): Cooper’s Creek, near
Innamincka: AMNH 110486-87. No data: AMNH
77637, 77648.

Pseudemydura umbrina. AUSTRALIA (Westem
Australia): Ellenbrook Reserve: WAM R29338;
Warbrook (Twin Swamp) Reserve: WAM R29341;
Bullsbrook Reserve: WAM R21859.

Osteological Material. (*=hyoid examined as well
as skull; **=complete skeleton; CBL=condylo-
basal length in mm.)

Emydura dentata. AMNH 99618 (CBL 18.3);
AMNH 92971 (CBL 20.0); *AMNH 92965 (CBL
25.6); IRSN [McD tag 46] (CBL 28.0); **AMNH
92675 (CBL 39.0); AMNH 104007 (CBL 40.8;
*AMNH 99613 (CBL 43.0); AMNH 62612 (CBL
44.0); **BM(NH) [18]77.5.19.77 (CBL 44.2); also
hyoid only on AMNH 100003.

Emydura australis. AMNH 57586 (CBL 19.8);
AMNH 59090 (CBL 31.7): AMNH 104011 (CBL
42.7); **AMNH 108957 (CBL 45.6); also hyoid only
on AMNH 92967.

Emydura latisternum. AMNH 69576 (CBL 35.4);
*AMNH 27287 (CBL 56.6); also hyoid only on
AMNH 69567 and 69566.

Emydura macquaria. **AMNH_ 103701 (CBL
38.5 [rearticulated]); **AMNH 103702 (CBL 45.6);
**AMNH 77636 (CBL 46.3); also hyoid only on
AMNH 108961 and 110487.

Pseudemydura umbrina. WAM R29338; *WAM
R29341. ([Disarticulated], but smaller than any E.
macquaria or E. latisternum examined and smaller
than most E. dentata and E. australis examined):
WAM R21859.

Advances in Herpetology and Evolutionary Biology

PLEURODIRAN PENES EXAMINED

Pelomedusidae: Podocnemis unifilis (AMNH
87966; Brazil: Mato Grosso: confluence of R.
Araguaia and R. Tapirapa). Pelomedusa subrufa
(AMNH 50744: Tanzania: Chai Camp). Pelusios
derbianus (AMNH 50746; Angola: Huambo).

Chelidae: Chelodina novaeguineae (AMNH
57589; Papua New Guinea: Westem Province:
Mabaduane. [No no., Bruxelles IRSN]; Indonesia:
Irian Jaya: Kurik.). C. expansa [“rugosa”’] (AMNH
86548, 86549; Queensland: Normanton, lower
Norman R.). Chelus fimbriatus (AMNH 107378
[adult] and 76190 [juvenile]; no data). Phrynops
geoffroanus (AMNH 90674; Peru: Cuzco, R.
Apurimac). P. gibbus (AMNH 64721; Guyana:
Kartabo [organ unprotruded]. AMNH 60521;
Guyana: Parabam [naturally protruded organ)).
Platemys platycephala (AMNH 61529; Guyana:
Shudikarwan R. [naturally protruded organ]). Also
Emydura australis (AMNH 57898, 59090-1, 57588,
104339, 69578). E. dentata (AMNH 103629, 104008,
104837, BM [18]75.5.1.8). E. latisternum (AMNH
69566, 69576).

LITERATURE CITED

BOULENGER, G. A. 1889. Catalogue of the chelo-
nians, rhynchocephalians, and crocodiles in
the British Museum (Natural History). New
Edition. Trustees, British Museum, x + 311
pp., 6 pls.

BURBIDGE, A. A., J. A. W. KIRSCH, AND W. R. MAIN.
1974. Relationships within the Chelidae
(Testudines: Pleurodira) of Australia and New
Guinea. Copeia, 1974: 392-409.

GaFFENEY, E. S. 1975. A phylogeny and classification
of the higher categories of turtles. Bull. Am.
Mus. nat. Hist., 155(3): 387-436.

____. 1977. The side-necked turtle Family Chelidae:
a theory of relationships using shared derived
characters. Amer. Mus. Novitates, No. 2620, pp.
1-28.

GOoDE, J. 1967. Freshwater tortoises of Australia
and New Guinea (in the family Chelidae).
Lansdowne Press, x + 154 pp.

Roo, N. DE. 1915. The reptiles of the Indo-
Australian Archipelago. I. Lacertilia, Chelonia,
Emydosauria. E. J. Brill, xiv + 384 pp.

SCHMIDTGEN, O. 1907. Die Cloake und ihre Organe
bei den Schildkroten. Zool. Jahrb. (Abt. f.
Anat.), 24: 354-414, Taf. 32-33.

SIEBENROCK, F. 1897. Das Kopfskelet der Schild-
kroten. Sitzungsber. Akad. Wiss. Wien (math.-
naturw. cl.), 106: 1-84, taf. ivi.

TONUTTI, E. 1932. Vergleichend-morphologische
Studie uber die Phylogenie des Enddarmes
und des Kopulationsorganes der mannlichen

New GUINEA EMYDURA AND THE PLEURODIRAN PENIS - McDowell 189

Amnioten, ausgehend von den Gymnophio- ZuG, G. R. 1966. The penial morphology and the
nen. Morphol. Jahbr. (Jahrb. Morphol. und relationships of cryptodiran turtles. Occ. Pap.
Mikr. Anat. 1°t Abt.), 70: 101-130, Taf. 1. Mus. Zool. Univ. Mich. No. 647, pp. 1-24.

The Basicranial Articulation of the
Triassic Turtle, Proganochelys

EUGENE S. GAFFNEY!

ABSTRACT. The Triassic turtle Proganochelys
quenstedti (Keuper, Trossingen, West Germany)
has a movable basicranial articulation formed by the
basicranial tubercle of the basisphenoid medially
and the basicranial recess of the pterygoid and/or
epipterygoid laterally. This is interpreted as the
generalized amniote condition, while the akinesis
of all remaining turtles (the Casichelydia) is inter-
preted as a shared derived character testing their
monophyly.

INTRODUCTION

Emest E. Williams performed a singu-
lar service for turtle fanciers when he
revealed that the skull morphology of the
Triassic turtle Proganochelys “...ap-
proaches the primitive reptilian condi-
tion and is separated by a very sizeable
morphological gap from such modem-
ized turtles as the Jurassic ones we have
described” (Parsons and Williams, 1961:
91). Previous to Williams’s revelation the
only information about Triassic turtle
skulls came from Jaekel’s (1916) descrip-
tion of a partially crushed skull. Jaekel
understated the degree of damage to his
specimen and produced a palatal res-
toration showing an apparently akinetic
skull with a broad, median tooth plate.
Williams examined a specimen in Stutt-
gart that was well preserved and clearly
showed significant differences from
Jaekel’s descriptions. Williams recog-
nized that these differences gave a new

1American Museum of Natural History, New
York, New York 10024, U.S.A.

basis for interpreting the Triassic turtles
and argued that, except for the shell,
Proganochelys was a relatively general-
ized reptile.

The Stuttgart specimens are now the
focus of a renewed effort to study the
Triassic turtles, and it is singularly ap-
propriate to pay homage to Ernest E.
Williams’s many contributions to chelo-
nian systematics by dedicating this paper
to him. Although the major descriptive
work on Proganochelys is still in prepa-
ration, I thought it would be useful to
describe one particular morphologic fea-
ture of the skull that has figured in dis-
cussions concerning the oldest turtles.

AVAILABLE MATERIAL AND PREVIOUS
WORK

To the best of my knowledge, there are
three Triassic turtle skulls known at
present. All three are, in my opinion, the
same taxon, Proganochelys quenstedti,
and all are middle Keuper in age. In
order of discovery, they are as follows:

1) Museum fur Naturkunde, Berlin. Collected in
1912, Halberstadt, East Germany. Consists of skull,
lower jaws, cervical vertebrae, shoulder girdle,
carapace, and plastron. Type of Stegochelys (Trias-
sochelys) dux (Jaekel), described and figured in
Jaekel (1916), skull also figured in Gaffney (1979b:
Figs. 117-124).

2) Staatliches Museum fur Naturkunde in Stutt-
gart, 15759. Collected in 1927, Neuhaus near Aix-
heim, West Germany. Consists of skull, lower jaws,
forefoot, miscellaneous fragments of limbs and cer-
vicals, and a partial carapace. Skull figured in
Berckhemer (1931: 7, 1951: 36) and Kuhn (1956:
69).

3) Staatliches Museum fur Naturkunde in Stutt-
gart, 16980. Collected in 1932, Trossingen, West
Germany. Consists of a nearly complete skeleton.
Skull along with shell and part of the skeleton fig-
ured in Berckhemer (1938: Pl. 4), skull also figured
in Parsons and Williams (1961: Pls. 5, 6), and in
Gaffney (1979b: Figs. 114, 115).

Jaekel (1916) published information on
the basicranium of Proganochelys
(“Triassochelys’’), but in retrospect it has
proven to be misleading. Although
Jaekel’s specimen from Halberstadt is in
general well preserved (the cervicals, for
example, are still the best known), the
skull has been subjected to lateral crush-
ing. The anterior, denticle bearing por-
tions of the pterygoids have been pushed
together at the midline, producing what
Jaekel interpreted as a median toothbear-
ing element covering the area where the
basipterygoid articulation might be
found, resulting in an apparently akinetic
skull. Although this reconstruction was
reproduced from time to time (cf. Romer,
1956), it did not stimulate very much
discussion or comment.

The discovery of the Aixheim speci-
men in 1927 would not have significantly
aided basicranial studies, even had it
been described, because it is also later-
ally crushed and lacks the posterior por-
tion of the braincase. Nonetheless, it is
the only specimen preserving the major
part of the epipterygoid, and it reveals a
number of other important features. The
discovery of the best preserved Triassic
turtle skull was the result of a combined
University of Tubingen-Staatliches Mu-
seum fur Naturkunde, Stuttgart, effort to
collect Plateosaurus skeletons from a
quarry in Trossingen. Near the end of the
1932 season they discovered three turtle
skeletons. Only one of the specimens,
SMNS 16980, has a skull, but it is rela-
tively well preserved. These three spe-
cimens were prepared at Stuttgart under
the direction of Dr. F. Berckhemer, and
were later in the care of Professor K.
Staesche. It was during Staesche’s stew-
ardship of the material in the 1950’s that
Ernest E. Williams examined the skull

PROGANOCHELYS - Gaffney ‘191

and presented his conclusions (Parsons
and Williams, 1961) that the cranium
possessed many generalized amniote
features. In 1975 I presented a cladistic
view of his information and figures and
argued that Proganochelys was the sister
group to all other turtles. The nature of
the basicranial articulation was dubious
from available photographs, and I sug-
gested that the articulation might actually
be fused rather than movable (Gaffney,
1975: 424). The actual material, however,
had still not been studied in detail, so it
was with great elation that in late 1978 I
was given permission to work on the
Stuttgart material.

DESCRIPTION

In the generalized amniote condition,
such as seen in Captorhinus, the pala-
toquadrate is articulated to the braincase
by means of a movable basipterygoid
articulation. In turtles of the suborder
Casichelydia this kinesis is lost, and
extensive sutural contact between the
pterygoid and basisphenoid obliterates
the articular area. As I have discussed in
detail elsewhere (Gaffney, 1975, 1979b),
the basicranium of turtles is an important
source of characters differentiating the
two main infraorders of casichelydian
turtles, Cryptodira and _ Pleurodira.
Embryonic stages of turtles have rudi-
ments of a palatobasal articulation in the
form of the cartilago articularis and the
processus’ basipterygoideus (Kunkel,
1912), demonstrating that the general-
ized amniote condition can also be found
in turtles. Pleurodires have a medial
process of the quadrate that forms the
main brace to the braincase, whereas in

_cryptodires the pterygoid extends pos-

teriorly to form the brace between brain-
case and quadrate. In adults of both
groups the basipterygoid articulation is
fused and the area of the cranio-quadrate
space is nearly obliterated by the broad
sutural contact of the braincase (prootic,
opisthotic) and palatoquadrate (quadrate,

192

pterygoid). The akinetic condition of
casichelydian turtles, therefore, not only
includes fusion of the basipterygoid ar-
ticulation, but also the development of
these subsidiary contacts anterior, dorsal,
and posterior, to the position of the ar-
ticulation.

In known captorhinids (Captorhinus,
Eocaptorhinus, see Heaton, 1979; War-
ren, 1961) the basicranial articulation is
formed between the epipterygoid and
basisphenoid. The epipterygoid has a
well-developed ventral expansion that
sends a curved process into a pocket of
the pterygoid so that the actual articula-
tion is borne by the epipterygoid.

In Proganochelys the epipterygoid is
poorly known; the most complete one is
in SMNS 15759, but crushing has dis-
torted its position, and the ventral limits
are broken. However, in SMNS 16980, on
the left side, there is what appears to be a
fragment of curved bone closely applied
to the pterygoid, and I interpret this as
the epipterygoid contribution to the ar-
ticulation (see Fig. 1). The sutures on the
ventral surface are not clear but seem to
indicate that the epipterygoid forms most

Advances in Herpetology and Evolutionary Biology

of the articular area. It is possible, how-
ever, that the posterior part of the articu-
lation is actually bome on the pterygoid.

The basisphenoid of Proganochelys has
well-developed, paired basipterygoid
tubercles that extend ventrolaterally (see
Fig. 2), rather than anterolaterally as in
Captorhinus (Price, 1935). The articular
surface is nearly flat and faces antero-
laterally to fit into the basicranial recess.
The tubercles are posterior to the dorsum
sellae and sella turcica rather than an-
terior as in Captorhinus. In another
paper (Gaffney, 1979a) I argued that a
structure on the basisphenoid of Meso-
chelys was not a remnant basipterygoid
process as identified by Evans and Kemp
(1975) because it is posterior to the dor-
sum sellae rather than anterior as in
Captorhinus. This argument no longer
has validity because the tubercles are
also posterior in Proganochelys thus re-
opening the question of the identity of
these structures in Mesochelys and
Glyptops (I hope to deal with this prob-
lem elsewhere).

The pterygoid of Proganochelys has a
basicranial recess that probably includes

Figure 1.

Proganochelys quenstedti, SMNS 16980, stereo photograph of ventral view of skull.

Figure 2. Proganochelys quensteadti, SMNS 16980,
ventral view of basicranium.

at least part of the epipterygoid, but be-
cause the suture is dubious, particularly
in the ventral view, I am unable to de-
scribe the portions formed by each bone.
The recess has the form of a trough that is
oriented anterodorsomedially and close-
ly fits the opposing tubercle on the basi-
sphenoid. The general form of the recess
is similar to Captorhinus, but the recess
faces more posteriorly in Proganochelys.
The question of actual movement in the
skull of Proganochelys must await the
complete description of the specimens,
but it is clear that some sort of movement
was possible in the basicranial articula-
tion.

The akinesis of all turtles except
Proganochelys results from a _ broad
sutural fusion of the basisphenoid and
pterygoid that I have interpreted (Gaff-
ney, 1975) as a shared derived character
testing the monophyly of the Casichely-
dia. The recognition of the generalized
amniote condition of the basicranial ar-
ticulation in Proganochelys supports the
argument that it (as the suborder Pro-
ganochelydia) is the sister group of the
Casichelydia.

PROGANOCHELYS - Gaffney —193

ACKNOWLEDGMENTS

The opportunity to study the Triassic
turtles in Stuttgart is due to the coopera-
tion and support of Rupert Wild, Curator,
and Bermhard Ziegler, Director, Staat-
liches Museum fur Naturkunde in Stutt-
gart (SMNS). Wild spent many hours
aiding me and my colleagues, and I am
very grateful for his help and hospitality.
I would also like to thank Karl-Heinz
Fischer, Museum fur Naturkunde, Ber-
lin, for giving me access to the Jaekel
Triassic material in East Berlin. Otto
Simonis, Department of Vertebrate
Paleontology, American Museum of
Natural History (AMNH), performed the
superb preparation necessary to expose
the basicranial morphology of the Stutt-
gart Proganochelys skull. Chester Tarka,
also of the Department of Vertebrate
Paleontology, AMNH, took the photo-
graphs. This paper is part of a project
funded by National Science Foundation
Grant DEB 8002885.

LITERATURE CITED

BERKHEMER, F. 1931. Die Saurierfunde aus den
Keuperablagerungen der Baar. Tuttlinger
Heimatblatter, Heft 14, 12 pp.

__. 1938. 25 Jahre Verein zur Forderung der
Wiurtt. Naturaliensammlung in Stuttgart. Stutt-
gart, Ernst Klett, pp. 1-18.

___. 1951. Die Spracher der Steine. Verlag “Die
Schonen Bucher,” Reihe C, “Natur” Band 2,
Stuttgart, pp. 9, 36.

EVANS, J., AND T. S. Kemp. 1975. The cranial mor-
phology of a new lower Cretaceous turtle from
southern England. Palaeontology, 18: 25-40.

GAFFNEY, E. S. 1975. A phylogeny and classification
of the higher categories of turtles. Bull. Amer.
Mus. Nat. Hist., 155(5): 387-436.

____. 1979a. The Jurassic turtles of North America.
Bull. Amer. Mus. Nat. Hist., 162(3): 91-136.

____. 1979b. Comparative cranial morphology of
recent and fossil turtles. Bull. Amer. Mus. Nat.
Hist., 164(2): 65-375.

HEATON, M. J. 1979. Cranial anatomy of primitive
captorhinid reptiles from the late Pennsylva-
nian and early Permian Oklahoma and Texas.
Oklahoma Geol. Surv., Bull. 127, pp. 1-84.

JAEKEL, O. 1916. Die Wirbeltierfunde aus dem
Keuper von Halberstadt. Serie II. Testudinata,
Teil 1, Stegochelys dux, n.g., n. sp. Palaont.
Zeitschr., 2: 88-214.

194 Advances in Herpetology and Evolutionary Biology

Kuun, O. 1956. Deutschlands vorzeitliche Tierwelt
(2nd ed.). Munchen, Bayerischer Land-
wirtschaftsverlag, 126 pp.

KUNKEL, B. W. 1912. The development of the skull
of Emys lutaria. J. Morphol., 23: 693-780.

PARSONS, T. S., AND E. WILLIAMS. 1961. Two Juras-
sic turtle skulls: a morphological study. Bull.
Mus. Comp. Zool., 125: 43-107.

Price, L. I. 1935. Notes on the braincase of
Captorhinus. Proc. Boston Soc. Nat. Hist.,
40(7): 377-386.

RoMER, A. S. 1956. The osteology of the reptiles.
Univ. Chicago Press, 772 pp.

WARREN, J. W. 1961. The basicranial articulation of
the early Permian cotylosaur, Captorhinus. J.
Paleont., 35(3): 561-563.

The Rectus Abdominis Muscle Complex of the
Lacertilia: Terminology, Homology, and
Assumed Presence in Primitive

Iguanian Lizards
SCOTT M. MOODY!

ABSTRACT. Camp (1923) divided the lacertilians into
two groups, Autarchoglossa and Ascalabota, on the
basis of the presence or absence, respectively, of
the Muscle rectus abdominis superficialis. His
recognition of the “superficialis” is both anatomic-
ally and terminologically incorrect. The M. rectus
abdominis has only three parts in the Autarcho-
glossa: medialis, lateralis, and pyramidalis. In the
Ascalabota, it has only two parts: medialis and
pyramidalis. Nearly all genera of the iguanian fami-
lies Agamidae, Chamaeleonidae, Uromastycidae,
and Iguanidae (including Oplurinae) were sur-
veyed for the third part of the M. rectus abdominis.
I confirmed the earlier observations by anatomists
of a small muscle lying superficial to the M. pec-
toralis and inserting into the skin at the axilla in the
uromastycids Leiolepis and Uromastyx and the
primitive agamid Physignathus, and I additionally
observed it in the agamid Hydrosaurus. Observed
developmental variations, comparative morphology
and innervation demonstrate that this muscle is part
of the M. pectoralis and is named the M. pectoralis
pars cutaneous. Camp believed this muscle to be
instead a vestigial M. rectus abdominis lateralis and
considered this evidence that the Ascalabota were
evolutionarily advanced lizards which had lost the
autarchoglossan attributes of terrestrial slinking
made possible by this muscle. My reanalysis sug-
gests instead that the Ascalabota are primitive and
that the ancestral lacertilians possessed only two
parts of the M. rectus abdominis, with the third part
evolving later with the ancestor of the Autarcho-
glossa.

Department of Zoology and College of Osteo-
pathic Medicine, Ohio University, Athens, Ohio
45701, U.S.A.

INTRODUCTION

The goals of this paper are to: 1) review
the terminology and anatomy of the ven-
tral and lateral trunk musculature, with
emphasis on the M. rectus abdominis
complex, among lacertilians on the basis
of both original dissections and transla-
tions of the often cited but seldom read
publications of the comparative anato-
mists Gadow, Maurer, Hoffmann, and
Furbringer working in the 19th century
of Germany; 2) survey the lizard infra-
order Iguania (Agamidae, Chamaeleo-
nidae, Uromastycidae, and Iguanidae)
defined by Camp (1923) and widely be-
lieved to be monophyletic (Underwood,
1971; Northcutt, 1978) for a double-
layered M. rectus abdominis; 3) deter-
mine the homology of the small cu-
taneous muscle slip lying lateral to the
M. pectoralis which is present in the
primitive agamids Physignathus and
Hydrosaurus (infrequently observed)
and uromastycids Leiolepis and Ur-
omastyx and was assumed by Camp
(1923) to be a vestigial superficial layer of
the M. rectus abdominis; and 4) assess
the value of the variable M. rectus ab-
dominis as a phylogenetic character in
hypothesizing cladistic relationships
among lizard families because this
character was used by Camp (1923) for
the first dichotomy in his classification of
lizard families into higher taxa.

196

TERMINOLOGY

Herpetologists and reptilian anato-
mists currently follow Camp’s (1923:
377-378) arrangement of the M. rectus
abdominis:

The Rectus abdominis muscle in autarchoglossid
lizards can usually be subdivided into four
parts ... 1) the Rectus profundus or main median
portion of the Rectus, segmented . . . 2) the Rec-
tus medianus, segmented, often intimately con-
nected with the skin over the whole belly .. . 3)
the Rectus lateralis, unsegmented, and forming
the lateral continuation of the Rectus medianus,
4) the Rectus internus, unsegmented, arising on
the pubis and passing forwards sometimes as far
as the sternum and dorsal to the Rectus pro-
fundus ...the medianus and lateralis may to-
gether be called the Rectus superficialis.

Camp attributed the observation of
these four divisions to Maurer (1896).
However, careful reading of Maurer’s
monograph reveals that: 1) Maurer
recognized only three parts of the M. rec-
tus abdominis; 2) the M. rec. abd. me-
dianus and M. rec. abd. profundus! dis-
tinguished by Camp are in fact the same
muscle; and 3) the M. rec. abd. medianus
and M. ree. abd. lateralis are not embry-
onically or anatomically primarily con-
tinuous. Gadow (1882) earlier also recog-
nized only three divisions of the M. rec-
tus abdominis.

Maurer (1896: 228-229, my translation)
summarized the variation of the M. rectus
abdominis among lacertilians:

The M. rectus abdominis of lizards is complica-
ted...in Lacerta a M. rec. abd. medialis and
lateralis can be distinguished. The latter is mis-
sing in Sphenodon. Both extend completely
together from the pelvis to the posterior edge of
the sternum. Here the medial part terminates, as
does the entire M. rectus abdominis in Sphen-
odon. However the M. rec. abd. lateralis of lizards
is continued further anterior as a ventral slip

1For sake of brevity, the name M. rectus ab-
dominis will be abbreviated to M. rec. abd. when
preceding the descriptors superficialis, lateralis,
profundus, etc.

Advances in Herpetology and Evolutionary Biology

superficial to the M. pectoralis and is overlapped
along its lateral edge by the M. obliquus ab-
dominis externus superficialis. The boundary
between the M. rec. abd. medialis and lateralis is
easily recognized if the M. rec. abd. lateralis is
first severed and reflected from its lateral edge.
The M. obliquus externus profundus can then be
seen to meet along a line with the M. rec. abd.
medialis, just as in Sphenodon. This line de-
marcates the boundary between both M. rectus
abdominis portions, as well as the lateral-most
boundary of the entire M. rectus abdominis of
Sphenodon ... The superficial layer of M. rectus
abdominis connects with the integument without
aid of inscriptional ribs or gastralia... along the
margins of the large transverse series of ventral
scutes, by inserting into the skin by fibers coming
from both the M. rec. abd. medialis as well as the
M. rec. abd. lateralis muscles. Because of this
manner of attachment, the M. rectus abdominis is
necessarily segmented, similar to Sphenodon
which is segmented by inscriptional ribs, but in
the case of the lizard by a series of fibrous
bundles inserting into the skin. However these
bundles are comprised only of the superficial
layer of fibers. Therefore in Lacerta superficial
and deep portions of the M. rec. abd. medialis
may be distinguished, but they are continuously
adherent to each other. The M. rec. abd. lateralis
of Lacerta is connected with the skin to an even
greater extent than the M. rec. abd. medialis, es-
pecially at its anterior end which ends bluntly at
the base of the neck as a skin muscle. A triangular
muscle (M. rec. abd. pyramidalis after Maurer, M.
rec. abd. internus after Gadow) develops deep to
the M. rec. abd. medialis and immediately an-
terior to the pelvis with fibers radiating anteriorly
and medially and inserting into the linea alba.

I have examined within the Autarcho-
glossa, a scincid Tiliqua scincoides
(UMMZ 65915), the lacertids Lacerta
agilis (UMMZ 119991), L. lepida
(UMMZ 72473), L. viridis (UMMZ
65756), and Podarcis muralis (UMMZ
65947), a teiid Ameiva ameiva (personal
collection) and a cordylid Cordylus gi-
ganteus (personal collection), and have
confirmed the observations of Maurer
and Gadow. Ventral and anterolateral
views of the musculature of Ameiva
ameiva are illustrated in Figures 1 and 2.

Throughout his paper Maurer unfor-
tunately changes his terminology and
switches back and forth from formal
latinized names to descriptive anatomical

|

a i in (ie ae

M. RECTUS ABDOMINIS OF LACERTILIANS - Moody

Lz
\\\\l
\\\

NY 4
\

iy {/ 1
Nill Ufa
Yel ha MAL |
EEA nf

<

abd lat

=.
KEK

ext obl \\)‘

SS

pect

——

|
5 WS I i
\iiCGH AHA
ina hi
i) i
myccom V7) (ja
win uh
hh i i \WE= =
vil a
i i, h =
abd lat i i) | } | \ i =e —
Th if \ =
bt | A \\\
i | NN NN i} = bd
HU) Hi WN MM a
i AWS Pyr
HTN
i
Figure 1. Ventral view of the trunk musculature of

Ameiva ameiva (Teiidae). Specimen is superficially
dissected on the right with a window cut in the M. rec-
tus abdominis, and deeply dissected on the left with
portions of the M. obliquus abdominis externus and M.
obl. abd. internus removed at different levels to reveal
underlying layers.

Abbreviations: abd lat, M. rectus abdominis lateralis;
abd med, M. rectus abdominis medialis; abd pyr, M.
rectus abdominis pyramidalis; ext obl, M. obliquus
abdominis externus; int obl, M. obliquus abdominis
internus; myocom, Myocommata; pect, M. pectoralis;
trn abd, M. abdominis transversus.

terms. These changes were probably the
cause of Camp’s misinterpretation of the
anatomical arrangement and homology of

197

the different parts of the M. rectus ab-
dominis. For example, Maurer describes
the transverse series of fibrous bundles
as originating from the superficial layer
of the M. rec. abd. medialis and attaching
to the skin. Camp interprets these col-
lectively as a distinct muscle, the M. rec.
abd. superficialis medianus, and assigns
the name M. rec. abd. profundus to the
deeper fibers which do not connect with
the skin and are not segmented. How-
ever, Maurer (1896: 229) emphatically
points out that these parts are not separ-
able as distinct muscles: “..die aber
kontinuirlich zusammenhangen,” (“how-
ever they are continuously adherent to
each other,” my translation).

Camp may also have been easily con-
fused by Maurer’s (p. 206) description of
Lacerta, which I translate as follows:

In Lacerta the M. rectus abdominis extends from
the pelvis to the neck region as the M. rec. abd.
superticialis. It is comprised of two distinguish-
able parts which I have referred to as the M. rec.
abd. medialis and lateralis. The M. rec. abd.
medialis extends from the pubic symphysis to the
posterior end of the stemum. Superficial fibers
attach this layer to the integument, inserting
along the posterior edges of the scales. Along its
entire lateral margin it meets the M. pectoralis
anteriorly...and the M. obliquus abdominis
extemmus profundus posteriorly. The superficial
M. rec. abd. medialis is the only muscle present
in Sphenodon, which is also connected with the
M. pectoralis and M. obliquus abd. ext. pro-
fundus. The superficial M. rec. abd. lateralis is
absent in Sphenodon. In Lacerta it is attached
directly to the M. rec. abd. medialis along its
border, and extends from pelvis to throat as a
similarly wide band, which arcs laterally along its
middle part. It extends from the Ligamentum
ileo-pubicum anteriorly and laterally, overlies
(superficially) the M. obliquus abd. ext. pro-
fundus, and is overlain by the M. obliquus abd.
ext. superficialis along its lateral border. It con-
tinues forward to overlay the ventral surface of
- the M. pectoralis, nearly approaches the midline
of the M. pectoralis, and terminates in an apo-
neurosis which attaches to the integument along
the border of the anterior most large transverse
ventral scutes. Consequently, this M. rec. abd.
lateralis is intimately connected with the integu-
ment, especially the anterior part which overlies

198

the M. pectoralis, and is entirely a cutaneous
muscle. Superficial fibers of the posterior part of
the M. rec. abd. lateralis insert into the skin but
deeper fibers comprise a smooth (continuous)
muscle belly, without tendinous inscriptions, and
is rather unsegmented as Gadow and others have
already demonstrated.

Maurer discussed the M. rec. abd.
medialis and lateralis collectively as the
superficial M. rectus abdominis but not
as a formally named M. rec. abd. superfi-
cialis. Additionally, he did not describe a
M. rec. abd. profundus, which would
have been necessary as a contrasting
member of the pair of muscles. Appar-
ently, he considered the M. rec. abd.
pyramidalis (Camp’s internus) as the
deep muscle in contrast to his collective
superficial M. rectus abdominis.

It is additionally evident that Maurer
(p. 219, my translation) intended to
recognize only three parts of the M.
rectus abdominis from his description of
the scincid Tiliqua sp:

pect

abd lat

IN

Advances in Herpetology and Evolutionary Biology

One recognizes namely ...that three proper,
separable M. rectus abdominis parts lie near one
another, that the M. rectus medialis is divided
once again into medial and lateral parts. Directly
before the pelvis, a triangular muscle is ob-
served...

Note that the division of the M. rectus
medialis is not into formally named
muscles but only two anatomical regions
which are defined in relationship to the
position of the medial edges of the M.
rec. abd. lateralis and M. obliquus abd.
ext. profundus.

Camp (p. 378) also described the M.
rec. abd. lateralis as the lateral continua-
tion of the M. rec. abd. medianus (be-
lieved by him to lie superficial to the M.
rec. abd. profundus). Referring to the M.
rec. abd. lateralis, he wrote: “.. . accord-
ing to Maurer (1898) it arises in the
embryo of Lacerta as the lateral portion
of the superficial layer of the M. rectus
abdominis.” Camp may have based his
conclusions on the text figures in Maurer

rib erec sp

latis
ext obl

ext obl

myocom

cos ext

Figure 2. Anterolateral view of the trunk and thoracic musculature of Ameiva ameiva (Teiidae). Windows have
been cut in the M. obliquus abdominis externus and the M. intercostalis externus to show deeper layers.

Abbreviations: cos ext, M. intercostalis externus; cos int, M. intercostalis internus; erec sp, M. erector spinae;
latis, M. latissimus dorsi, and rib; and see Figure 1.

M. RECTUS ABDOMINIS OF LACERTILIANS - Moody

rather than the text. In adult Lacerta spp.
the M. rec. abd. lateralis is anatomically
adjacent to the superficial fibers of the M.
rec. abd. medialis, but Camp misinter-
preted both the anatomical positions of
the three named parts of the M. rectus
abdominis (see above) and the embry-
onic origin of these parts. A modified and
semischematic illustration of the late
developmental stages of the trunk
musculature of Lacerta agilis (Maurer,
1898: Fig. 23) is presented in Figure 3.

In his review of the embryonic de-
velopment of the ventrolateral trunk
muscles of lacertilians, Maurer (1898: 55—
56, my translation) observed:

cos int brev “| \f ‘ cos ext brev

cos int long :
cos ext long

it SDP Ia a 13

abd trn

Le ro:

ext obl pro
int obl
perit ext obl sup
rib skin
abd med WZ.
Ww abd lat
‘Yi
IY

Figure 3. Transverse section of a 31 day old embryo
of Lacerta agilis (Lacertidae). Illustration is semis-
chematic and modified from Maurer (1898: Fig. 23).

Abbreviations: abd lat, M. rectus abdominis lateralis:
abd med, M. rectus abdominis medialis; abd trn, M.
abdominis transversus; cos ext brev, M. intercostalis
externus brevis; cos ext long, M. intercostalis externus
longus; cos int brev, M. intercostalis internus brevis:
cos int long, M. intercostalis internus longus; ext obl
pro, M. obliquus abdominis profundus; ext ob! sup, M.
obliquus abdominis externus superficialis; int obl, M.
obliquus abdominis internus; perit, Peritoneum, rib,
and skin.

199

The M. rectus abdominis differentiates from a
uniform Anlage but in a different manner than in
the Urodeles. I distinguish a segmented M. rec.
abd. medialis that develops from the caudal bor-
der of the stemum to the pubic symphysis. The
M. rec. abd. lateralis is unsegmented and extends
anteriorly from the pubis to the throat region,
where it extends over the ventral surface of the
M. pectoralis major muscle and ends in the
integument. The M. pyramidalis differentiates
from the M. rec. abd. medialis and arises from the
pubic symphysis. It is not similar to the iden-
tically named mammalian muscle, but is located
instead deep and adjacent to the peritoneum.
The M. rec. abd. medialis extends laterally to join
with the ventral margins of the M. intercostalis
externus and internus because these parts are
embryonically related.

I summarize the nomenclatural history
of the M. rectus abdominis and its parts
in Table 1 and suggest that this muscle
be recognized as a tripartite muscle. I
further suggest that the terms M. rec. abd.
medialis and M. rec. abd. lateralis be
recognized and used rather than the M.
rec. abd. profundus and M. rec. abd.
superticialis. This follows either older or
more frequently used terminology, more
accurately depicts the anatomical posi-
tions of the two muscle parts, and avoids
confusion with the incorrect muscle divi-
sions recognized and named by Camp
(1923). The M. pyramidalis is considered
to be embryonically and functionally a
part of the M. rectus abdominis.

HISTORICAL REVIEW

Morphological variations of the pec-
toral girdle musculature (Furbringer,
1900; Sukhanov, 1961 and 1976) and the
ventrolateral trunk musculature (Maurer,
1896; Gadow, 1882; Camp, 1923) have
provided several characters frequently
used for hypothesizing the phylogenetic
relationships of lizard families. In
Camp’s (1923) “Classification of the
Lizards” three of the 34 character com-
plexes reviewed were of the muscula-
ture. English speaking herpetologists
rely heavily on this monograph as the
standard reference for lacertilian com-

200

Advances in Herpetology and Evolutionary Biology

TABLE 1. NOMENCLATURAL HISTORY OF THE M. RECTUS ABDOMINIS COMPLEX.

M. RECTUS ABDOMINIS MEDIALIS

(Maurer, 1896, 1898)

= “ventralis” Gadow, 1882; Hoffman, 1890

= “abdominis” Mivart, 1867, 1870; Sanders, 1870, 1872, 1874
= “profundus” Camp, 1923

= “externus”’ Byerly, 1925

= “superticialis medianus” Camp, 1923

M. RECTUS ABDOMINIS LATERALIS

“suprapectoralis”
“superticialis lateralis”

M. RECTUS ABDOMINIS PYRAMIDALIS

(Gadow, 1882; Hoffman, 1890;
Furbringer, 1900)

Rudinger, 1868; Furbringer, 1869, 1987

Camp, 1923

(Gorski, 1852; Stannius, 1856; Mivart, 1867, 1870;

Maurer, 1896, 1898;

Sanders, 1870, 1872, 1874; Maurer, 1896, 1898)

= intemus

Gadow, 1882; Hoffman, 1890; Camp, 1923; Byerly,

1925

= “triangularis”’

parative myology, primarily because
Camp summarized earlier German
works.

Subsequent to Camp (1923) there have
been several comparative and functional
studies of the limb and girdle muscula-
ture of lizards possessing snakelike
locomotory adaptations, but comparative
and phylogenetic studies of the muscula-
ture of nonslinking tetrapod lizards have
been few in number. This is unfortunate
because these lizards (especially the
Iguania) are morphologically more simi-
lar to the fossil ancestral lacertilians of
the Permian and Triassic (Robinson,
1967) and are presumably more primitive
than other extant lizard families
(Sukhanov, 1976; Estes, 1983; Moody, in
preparation). Cladistic analysis of the
character states derived from the present
restudy of the M. rectus abdominis sup-
ports this view (see discussion section).

Furbringer (1869) named the cutane-
ous muscle lying superficial to the M.
pectoralis as the M. suprapectoralis. Al-
though he _ studied several autarch-
oglossan lizards with a fully developed
M. rec. abd. lateralis, he was probably
strongly influenced by the condition of
this muscle in Uromastyx, which he had
primarily described. As will be discussed
later, Furbringer was correct in believing
this muscle to be part of the M. pec-
toralis.

Maurer, 1896

Mivart (1867, 1870) pioneered the
study of lizard musculature with su-
perbly illustrated and described dissec-
tions of Iguana tuberculata (=I. iguana)

(Iguanidae) and Chamaeleo_ parsonii
(Chamaeleonidae). Sanders (1870, 1872,

1874) published three detailed studies of

Platydactylus japonicus (=Gecko japon-
icus) (Gekkonidae), Leiolepis belliana
(Uromastycidae), and Phrynosoma cor-
onatum (Iguanidae). DeVis (1884) pub-
lished a dissection of Chlamydosaurus
kingi (Agamidae). Shufeldt (1890) thor-

oughly described the musculature of -

Heloderma suspectum (Helodermati-
dae). All but one of these studies, which
were published in English, were con-
cerned with four lizard families which
belong to Camp’s defined suborder

Ascalabota. This taxon is characterized |

by simple ventrolateral trunk muscula-
ture. The German comparative myo-
logical publications,

primarily representatives of the lizard
families belonging to Camp’s suborder
Autarchoglossa, the sister taxon of Asc-

alabota. These lizards possess a more |

complex ventrolateral musculature.

The earliest German study was that of |
Gorski (1852) who concentrated on the |

pelvic girdle. Stannius (1856) reviewed

lizard anatomy and Riudinger (1868) |
produced a comparative study of the fore- |

which appeared |
during the same time period, included |

M. RECTUS ABDOMINIS OF LACERTILIANS - Moody

limb musculature of reptiles and birds.
Three European lizards Anguis fragilis
(Anguidae), Chalcides chalcides (Scinc-
idae), and Ophisaurus apodus (An-
guidae) were included in these works.
Furbringer (1869, 1875, 1900) produced a
classic series of papers describing and
illustrating the skeleton, nerves, and
muscles of the lacertilian pectoral girdle,
especially that of Uromastyx aegyptius
(Uromastycidae). Gadow (1882) com-
pared the ventrolateral musculature of 16
lizard species representing the families
Varanidae, Teiidae, lLacertidae, and
Scincidae of the Autarchoglossa, and the
families Iguanidae, Gekkonidae, and
Chamaeleonidae of the Ascalabota.
Hoffman (1890) reviewed these investi-
gations, added original observations, and
listed synonymous names for different
muscles.

Maurer (1896) grossly compared the
abdominal musculature of Lacerta agilis
L. viridis, Podarcis muralis (Lacertidae),
Tiliqua sp. (Scincidae), Chamaeleo sp.
(Chamaeleonidae), and Sphenodon
punctatus (Sphenodonidae). This was
followed in 1898 by a study of the em-
bryonic origin of the ventrolateral trunk
musculature of Lacerta agilis. As I dis-
cussed earlier, this study developed ac-
curate terminology and determined the
homology of the parts of the M. rectus
abdominis.

The phylogenetic significance of the
different anatomical arrangements of the
M. rectus abdominis in lacertilians was
not appreciated until Camp (1923) sum-
marized the literature, made original ob-
servations, and hypothesized a_phy-
logenetic classification of the lizard fami-
lies. Presence or absence of his M. rectus
abdominis “superficialis” (including his
named parts medianus and lateralis) was
selected (pp. 297-298, 355, 385 415) as
the major character in creating his phy-
logenetic classification. He cladistically
divided the Lacertilia into two taxa and
assigned to them the subordinal names
Autarchoglossa and Ascalabota. The
former is characterized by presence of

201

the M. rectus abdominis “superficialis’’;
the latter by its absence. He also noted
the presence of a vestigial M. rectus ab-
dominis “‘superficialis” in Leiolepis bell-
iana, and Uromastyx hardwicki (Ur-
omastycidae), and Physignathus lesueurii
(Agamidae).

Prior to Camp’s monumental work, the
M. rec. abd. “superficialis’” had been de-
scribed and figured (but unfortunately
not labeled) for only one agamid, Ur-
omastyx spinipes (=U. aegyptius) (Fur-
bringer, 1875: 715, Figs. 63, 73). Moody
(1980) placed Leiolepis and Uromastyx in
the separate family Uromastycidae. I
have translated Furbringer (p. 715) as
follows:

Compared with each other (Uromastyx aegyptius
and the scincids Macroscincus coctei, Scelotes
sp., Chalcides chalcides), the lateral part of the
superficial layer of the M. pectoralis will be over-
lain by a small slip of muscle (M. suprapectoralis)
which originates either from the posterolateral
area of the M. pectoralis itself or alternatively
from the M. obliquus abd. externus and extends
forwards to just posterior to the axilla (Uromastyx)
or the area of the clavicle (scincids) where it in-
serts as an aponeurosis into the skin.

Camp (p. 380) incorrectly cited Furbrin-
ger as having examined Uromastyx
hardwicki.

Camp (p. 380) cites Sanders (1872) as
having described the M. rec. abd. “super-
ficialis” in the uromastycid Leiolepis
belliana. However, Sanders never men-
tioned this muscle slip, although the
adjacent M. rec. abd. medialis (Camp’s
“profundus ) was described, and the M.
rec. abd. “superficialis’” was unambigu-
ously figured, though not labeled (Sand-
ers: Fig. 2). Evidently Camp personally

observed the M. rec. abd. “superficialis”’

in Physignathus lesueurii as an accom-
panying literature citation is lacking.
Whether he also examined specimens of
Uromastyx hardwicki and Leiolepis bell-
iana cannot be proven because they were
neither figured nor identified by mu-
seum catalogue numbers.

202

STATEMENT OF PROBLEM

There has been considerable dis-
agreement among herpetologists, both in
literature and especially in private dis-
cussions, concerning whether Camp
believed the Ascalabota to be primitive
or derived relative to the Autarchoglossa
and whether the Ascalabota (Gekkota and
Iguania) is a monophyletic or poly-
phyletic suborder. The source of the dis-
agreements is a contradiction between
Camp’s cladogram (p. 333) and numerous
statements in his text. The cladogram
suggests that the Iguania is primitive and
the sister group to all remaining lacerti-
lians. However in his classification (pp.
297-298) and summary (pp. 416-417)
Camp divides the Lacertilia into the cur-
rently widely used suborders Ascalabota
and Autarchoglossa. In this scheme
Camp states that the Gekkota and Iguania
are sister taxa and uses the absence of the
M. rec. abd. “superficialis” as their pri-
mary definitive character.

Conceming the direction of evolution
of the one or two-layered M. rectus ab-
dominis, it is quite clear what Camp be-
lieved. He wrote (p. 385) that:

It is apparent . . . that the Rectus superficialis is a
morphologically widespread and embryologically
precocious muscle. Its absence in Ascalabota
cannot well be due to primitiveness of that group
but more probably to the fact that all forms in the
group except certain primitive agamids have lost
the muscle, while it remains and develops in the
Autarchoglossa. This would seem to imply that
the pro-Sauria may have carried the body close to
the ground and that this mode of locomotion has
been correlated with the development of “slink-
ing’ musculature in the belly-wall.

Lizards today which retain this mode of locomo-
tion retain the musculature, while the group that
seems to have early adopted arboreal life has
apparently lost the musculature.

In his summary, Camp wrote (p. 420)
that:

The superficial layer of the rectus is regarded as a
primitive crawling muscle. Its presence in certain
primitive agamids tends to show this and its ab-

Advances in Herpetology and Evolutionary Biologu
§ SY

sence in all other Ascalabota is considered sec-
ondary...

Within the Ascalabota, Camp believed
that the Agamidae (and Uromastycidae)
was evolutionarily derived from within
(not as sister taxa) the Iguanidae (p. 333).
However, this assumption is contradicted
by his observation that Leiolepis bell-
iana, Uromastyx hardwicki, and Phy-
signathus lesueurii possess the M. rec.
abd. “superficialis” because these spe-
cies are supposedly representative of
primitive iguanians which have retained
vestigially the M. rec. abd. “superfi-
cialis” from their autarchoglossan-like
ancestor. Camp’s argument that this
muscle is a plesiomorphic character for
the lacertilians is not congruent with the
presence of the derived character states
such as acrodont dentition, simplified
dermocranium and pectoral girdle, lack
of inscriptional abdominal ribs in both
the Agamidae and Uromastycidae and
presence of the epiotic foramen in the
Agamidae (Moody, 1980).

Also inexplicable is Camp’s defense of
the dichotomy of the Autarchoglossa and
Ascalabota when the above three species,
presumed by him (pp. 416-417) to repre-
sent the ancestral clade of the Agamidae,
exhibit the “wrong” character state.

Camp dissected the musculature of
only a few species and since his work, the
M. rectus abdominis has been studied
phylogenetically only in the Gekkota
(Kluge, 1976). Camp examined six
agamid genera from a total of 52 recog-
nized genera, two of two uromastycid
genera, 19 of 57 iguanid genera, and one
of four chamaeleonid genera. I inform-
ally recognize the Madagascan genera
Oplurus and Chalarodon as the sub-
family Oplurinae in the Iguanidae on the
basis of the ulnaris nerve pathway, ab-
dominal incriptional ribs, dermocranium,
hemipenes, scalation and vicariant zoo-
geography. Camp examined one of two
oplurine genera. The two iguanids
Hoplocercus and Morunasaurus con-
sidered by Etheridge (personal com-

M. RECTUS ABDOMINIS OF LACERTILIANS - Moody

munication) to be primitive were not ex-
amined. If additional agamid and ur-
omastycid species had been examined by
Camp, then he would have discovered a
character state distribution of the M. rec-
tus abdominis which would not have
been congruent with the genera Ur-
omastyx and Physignathus as recognized
at the time of his work (see later discus-
sion).

If the Agamidae, Chamaeleonidae, and
Uromastycidae were evolutionarily de-
rived from within the Iguanidae as Camp
suggested, then at least some of the
primitive iguanids should be expected to
also possess the M. rec. abd. “super-
ficialis.”” Of course, independent loss of
the muscle after the phyletic evolution of
the above three families may explain this
character state distribution. Subse-
quently, I surveyed all four families of
the Iguania exhaustively, but many more
species per genus for the Agamidae were
examined because the M. rec. abd.
“superticialis’ was known to be already
present and because the family was in
sore need of taxonomic revision.

The morphological nature of the M.
rec. abd. “superficialis” in Uromastyx,
Leiolepis and Physignathus is substan-
tially different than the equivalent
muscle in the Autarchoglossa. Camp be-
lieved that the muscle found in these
genera was a vestigial remnant. A goal of
this study is to determine whether this
homology is correct.

ANATOMICAL OBSERVATIONS

For the Agamidae, I examined 177 of
the circa 300 species which represent 48
of the 52 recognized genera and sub-
genera. The taxonomic nomenclature fol-
lows Moody (1980). The four genera not
examined (Harpesaurus, Mictopholis,
Thaumatorhynchus, undescribed gen.
nov.) are known only from type speci-
mens. Harpesaurus is closely related to
Pseudocalotes, and a new genus is being
described for “Cophotis’ sumatrana

203

(Moody and Bohme, in preparation).
Both genera were examined. Mictopholis
austeniana is related to Salea according
to Smith (1935) and I examined this
genus. Thaumatorhynchus brooksi is
closely related to Aphaniotis which I
examined. The undescribed gen. nov.
(Arnold, in preparation) is closely related
to Phrynocephalus which I sampled.

For the Uromastycidae, I examined 19
of the 20 species representing the only
genera, Uromastyx and Leiolepis. Only
Uromastyx dispar was not examined
because the single known specimen, the
holotype, is only a skin and skull.

For the Chamaeleonidae, I examined
19 of the circa 100 species representing
the four genera Bradypodion, Brookesia,
Chamaeleo, and Rhampholeon. Six spe-
cies representing four of the nine species
groups of Chamaeleo inhabiting Mada-
gacar and seven species of the four
groups inhabiting Africa were examined.
My nomenclature follows Hillenius
(1959), Raw (1976), and Klaver and
Bohme (in preparation).

For the Iguanidae excluding Opluri-
nae, I examined 86 of the circa 630 spe-
cies which represent all 53 currently
recognized genera (Etheridge, in prepa-
ration). Two recently described mono-
typic genera from South America, Vil-
cunia Donoso-Barros and Cei (1971) and
Pelusaurus Donoso-Barros (1973), were
not available for study. Moreover, their
status as valid genera is in dispute. The
original authors hypothesized relation-
ships to Liolaemus and Proctotretus, but
Etheridge (personal communication) has
inferred instead from the descriptions,
relationship to the Liolaemus-Cteno-
blepharis complex. In either case these
three related genera were examined.
Most of the species groups of the two
large genera Anolis and Sceloporus, fol-
lowing the respective arrangements by
Etheridge (1959) and Smith (1939) were
sampled. I also examined 14 of the 19
species of the three primitive iguanid
genera Morunasaurus, Enyaliodes, and
Hoplocercus. For the Oplurinae, I ex-

204

amined six of the seven species belong-
ing to the two genera Oplurus and
Chalarodon of Madagascar.

Camp's conclusion that within the in-
fraorder Iguania, only three genera pos-
sess the M. rec. abd. lateralis was con-
firmed. All five species of Leiolepis, both
species of Physignathus and all species
of Uromastyx, except U. asmussi and U.
loricatus, possess this muscle. In addi-
tion, one specimen of Hydrosaurus pus-
tulatus (personal collection), a close rela-
tive of Physignathus, has a tiny muscle
slip on only one side. Six other examined
specimens representing all three species

of Hydrosaurus totally lacked the
muscle.
All remaining examined agamids,

chamaeleonids, iguanids, and oplurines
completely lack a lateral or superficial
second layer of the M. rectus abdominis
or a vestigial muscle slip overlying the
M. pectoralis.

Presence or absence of the muscle slip
in question was determined by making a
paraventral incision through the skin
along the entire length of the pectoral
girdle and stemum and by reflecting the
skin laterally. This muscle is a thin rib-
bonlike slip lying superficial and lateral
to the large fanshaped M. pectoralis. This
muscle slip is oriented obliquely with its
insertion ventromedial to the axilla and
its origin the dorsolateral corner of the
thoracic region.

The insertion of this muscle anteriorly
is a broad tough sheet of fibrous connec-
tive tissue which firmly attaches the
muscle to the skin medial to the forearm,
which can be detached only by tearing or
cutting. The posterior origin of this
muscle is not attached to the skin. In-
stead, the fibers merge imperceptively
with fibers of three muscles (M. obliquus
abd. externus, M. pectoralis, and M. rec.
abd. medialis) at the area of common
intersection which overlies the point of
articulation of the last xiphisternal rib
with the last thoracic rib. The fibers of
these four merging muscles have a sub-
parallel orientation in this common area.

Advances in Herpetology and Evolutionary Biology

This muscle is unsegmented, unlike
the M. rec. abd. medialis which is vari-
ably segmented in agamids and _ ur-
omastycids with cartilaginous myo-
commata. The muscle is thin but dis-
tinctly several fibers thick, the thickness
in adults being 10 to 20% of the width of
the muscle. The width is one third to one
half of the distance measured from mid-
sternum to the glenohumeral articula-
tion. It is approximately equal in width
throughout its length. Anteriorly the fi-
bers are oriented at approximately a 40
degree angle with the adjacent fibers of
the M. pectoralis. This muscle in Phys-
ignathus cocincinus, together with the
M. pectoralis and M. rectus abdominis
medialis, is illustrated in Figure 4. The
configuration of this muscle slip is nearly
identical in Uromastyx, Leiolepis, and
Hydrosaurus.

The skin overlying the M. pectoralis
and M. rec. abd. medialis of all iguanians
is attached to the musculature only by
loose connective tissue and can be
peeled away easily. This is in contrast
with most autarchoglossan lizards in
which the skin is firmly attached seg-
mentally by both fibrous connective tis-
sue and multiple muscular slips. In Ur-
omastyx, Leiolepis, Hydrosaurus, and
Physignathus, only the anterior end of
the muscle slip is firmly attached to the
skin.

The M. rec. abd. medialis is wide and
moderately thick in agamids, uromasty-
cids, iguanids, and oplurines, but con-
siderably reduced in the compressed
chamaeleonids. It extends from the an-
terior margin of the pubis and extends
broadly to the xiphisternum. It is com-
prised of longitudinal fibers in a single
muscle layer which may be variously
segmented by transverse cartilaginous
myocommata in the agamids and ur-
omastycids and by both myocommata and
inscriptional ribs (sometimes but not
always congruent) in the iguanids, oplu-
rines, and chamaeleonids. Muscle slips
do not insert into the overlying ventral
skin at regular intervals. Laterally this

ee ee ee ee ee ee

M. RECTUS ABDOMINIS OF LACERTILIANS - Moody

pect hum

l pect cut

abd med

Figure 4. Anteroventral view of the superficial trunk
and thoracic musculature of Physignathus cocincinus
(Agamidae). The M. pectoralis pars cutaneous was left
adherent to the reflected skin on the right side. This
muscle slip was dissected from the skin on the left side
and oriented in correct anatomical position alongside
the M. pectoralis pars humeralis.

Abbreviations: abd med, M. rectus abdominis medi-
alis; pect cut, M. pectoralis pars cutaneous; pect hum,
M. pectoralis pars humeralis.

muscle abuts the M. obliquus abdominis
externus within the same fascial layer.
The M. obliquus abdominis externus
does not have separate superficial and
deep layers, except that some fibers over-

lap superficial and some deep to the.

abutting M. rectus abdominis medialis.

The only resemblances between the
muscle slip overlying the M. pectoralis in
Uromastyx, Leiolepis, Physignathus, and
Hydrosaurus and the M. rec. abd. la-
teralis of autarchoglossan lizards are: 1)
inserts into the skin; 2) overlies the M.
pectoralis; and 3) some fibers are contin-
uous with the M. rec. abd. medialis.
However, these resemblances must be
qualified and the differences enumer-
ated.

First, only the anterior end terminates

205

in an anastomosis with the skin. This
muscle does not have segmentally ar-
ranged slips of fibers inserting into the
skin as in Autarchoglossa. Second, this
muscle overlies only the M. pectoralis
and does not overlie any part of the M.
rec. abd. medialis or M. obliquus abd. ex-
ternus as in the Autarchoglossa. Third,
although it is true that some fibers of this
muscle are continuous with the antero-
lateralmost fibers of the M. rec. abd.
medialis, most are continuous with the
posterolateralmost fibers of the M. pec-
toralis. Moreover, in both the Ascalabota
and Autarchoglossa the M. pectoralis, M.
rec. abd. medialis, and M. obliquus abd.
extemmus imperceptively merge at their
common intersection. Fourth, in these
primitive iguanians, this muscle is nar-
row and located laterally. In the Autarch-
oglossa, it is usually very wide and com-
pletely covers the M. pectoralis, and may
extend to the throat.

Structurally, this muscle appears to be
simply a small lateral slip, surrounded
with epimysium, that has detached from
the main part of the M. pectoralis and has
inserted into the skin just ventromedial
to the glenohumeral joint. This muscle is
lacking in three of six Uromastyx hard-
wicki specimens examined, and presum-
ably develops late into the adult condi-
tion. Uromastyx asmussi and U. loricatus
usually lack this muscle, but in U. as-
mussi (ZFMK 7925) a thickened band of
connective tissue with a few detectable
muscle fibers was found. U. loricatus
(CAS 86468) has a small slip of muscle
identical to that found in other species,
but six other specimens lacked the
muscle. In one specimen of U. acanth-
inurus (personal collection) an interest-
ing developmental variation was dis-
covered (Fig. 5). Approximately one

‘quarter of the central fibers of the M. pec-

toralis inserted into the skin at the same
location where a lateral slip of muscle
would insert, and the lateral slip was
missing. These structural variations sug-
gest that this muscle slip is derived from

_ the M. pectoralis.

206

Figure 5. Anteroventral view of the superficial trunk
and thoracic musculature of Uromastyx acanthinurus
(Uromastycidae). Only in this specimen was the M.
pectoralis pars cutaneous found to arise from the
middle of the body of M. pectoralis pars humeralis,
rather than existing as a separate slip lateral to the
main muscle body.

Abbreviations: see Figure 4.

Finally, several specimens were dis-
sected in order to trace the innervation of
the muscle slip in question. The nerve
was found to be a branch of a large nerve
trunk of the brachial plexus in the axilla,
and the large nerve entered the M. pec-

toralis, in Leiolepis belliana (ZFMK
13156), Uromastyx ornatus (ZFMK
8576), and Physignathus cocininus

(ZFMK 21459). The latter specimen
demonstrated best the ramification of the
nerve into the muscle. Because the M.
rectus abdominis is innervated by tho-
racic intercostal and lumbar nerves, in-
nervation of this muscle by a branch of the
nerve to the M. pectoralis from the bra-
chial plexus clearly demonstrates correct
homology with the M. pectoralis. I have
named this muscle slip simply as the M.
pectoralis pars cutaneous in contrast with

Advances in Herpetology and Evolutionary Biology

the principal muscle body M. pectoralis
pars humeralis.

DISCUSSION

Camp’s (1923) suborders (Autarch-
oglossa, M. rec. abd. lateralis present;
Ascalabota, M. rec. abd. lateralis absent)
have remained relatively unchanged
since their establishment (Underwood,
1971) even though the Xantusiidae and
Pygopodidae were respectively transfer-
red from the Scincomorpha and An-
guimorpha to the Gekkota (Kluge 1967,
1976). Recently, the suborder Ascalabota
has been challenged as a polyphyletic
taxon because of numerous convergent
adaptations of the Iguania and Gekkota
(Sukhanov 1961, 1976; Northcutt, 1978;
Estes, 1983; Moody, in preparation).
Even Camp indicated that the Gekkota
were more closely related with the Aut-
archoglossa with his dendrogram (p. 333)
but argued for the Ascalabota in his clas-
sification and character synopsis (pp.
296-298). A cladistic analysis of Camp’s
characters has demonstrated that the data
available to him would support his den-
drogram, not his classification (Moody, in
preparation).

Sukhanov (1961, 1976) was the first to
rigorously point out the inconsistencies
between classification and dendrogram
in Camp (1923). He also argued that the
Gekkota and Scinco-anguimorpha are sis-
ter taxa on the basis of both traditional
morphological characters and his obser-
vations on the musculature of the pec-
toral girdle and forelimb. Similar phy-
logenetic hypotheses have been sug-
gested by Estes (1983) and Presch (per-
sonal communication). Northcutt (1978)
also derives the Gekkota from the ances-
tral lineage of Scincomorpha and An-
guimorpha and does not hypothesize a
close relationship with the Iguania.

Camp (p. 420) argued that the Autarch-
oglossa was primitive and that the M. rec.
abd. lateralis would be present in primi-
tive lizards because of their life style of

M. RECTUS ABDOMINIS OF LACERTILIANS - Moody

terrestrial and burrowing habits, i.e.,
creeping, crawling, and slinking. He then
argued that the Ascalabota were ad-
vanced lizards because of their life style
of arboreality and saxicolity, i.e., running,
jumping, and climbing. I would suggest
that the prejudice of progressive evolu-
tion may have been involved unwittingly
in Camp’s evaluation, i.e., advanced liz-
ards would resemble primates. Support
for this suggestion comes from the fact
that if one examines Camp’s predictions
for direction of character state evolution
separately for his many characters, in a
majority of cases he hypothesizes that the
state present in the Ascalabota is primi-
tive. But when he summarizes the overall
evolution of the lizards, they become
advanced (Moody, in preparation). Ac-
cording to Camp, the Gekkota and
Iguania shared an ascalabotan ancestor
which had lost the M. rec. abd. lateralis;
however, the muscle was somehow re-
tained in Uromastyx, Leiolepis, and
Physignathus. Camp used the presence
or absence of this muscle as the primary
diagnostic character for classifying liz-
ards, yet he ranked the muscle 22nd out
of 34 evaluated characters for its value as
a conservatively predictive phylogenetic
character, or in his terminology, its
“paleotelic value.” Perhaps this incon-
sistency gave rise to the contradiction
between his classification and his den-
drogram.

If Camp’s classification is followed,
then the M. rec. abd. lateralis must be
lost at least six times. Once in the Gek-
kota because the primitive Xantusiidae
possesses the muscle (Kluge, 1976). Be-
cause primitive agamids and uromasty-
cids possess the muscle, this requires a
second loss for the Iguanidae, a third loss
for the Chamaeleonidae and a fourth loss
for the advanced Agamidae. Because two
species of Uromastyx lack the muscle, a
fifth loss can be argued. Finally, a sixth
loss is for Hydrosaurus. That the losses in
the Iguanidae, Chamaeleonidae and
Agamidae are all independent is clear
because Camp (pp. 333, 417) believed

207

that the Agamidae (and Uromastycidae)
evolved from within the necessarily
paraphyletic Iguanidae.

The discovery with this paper that the
observed muscle in Uromastyx, Lei-
olepis, Physignathus, and Hydrosaurus is
not homologous with the M. rec. abd.
lateralis of the Autarchoglossa and
Xantusiidae solves the problem of too
many evolutionary losses. If the Gekkota
are more closely related to the Scinco-
morpha and Anguimorpha because
Xantusiidae possesses the M. rec. abd.
lateralis, then only one loss must be
hypothesized. If the Iguania is a rela-
tively primitive lizard group, then they
and ancestral lacertilians lacked the M.
rec. abd. lateralis. In the common ances-
tor of uromastycids and _ primitive
agamids, the M. pectoralis pars cutane-
ous evolved, but was once again lost in
the advanced agamids, two species of
Uromastyx, and Hydrosaurus.

This reanalysis of the M. rectus ab-
dominis complex in the Lacertilia, cor-
rections of its terminology and homology,
and brief review of alternative phy-
logenetic hypothesis differing from
Camp's classification, should encourage
systematic herpetologists to abandon his
classification, but accept his dendogram
except for minor shifts of families already
discussed above. However, the barrier of
tradition and expert opinion is often dif-
ficult to overcome as Rosen et al. (1981)
have so eloquently demonstrated in their
presentation of evidence that the lung-
fishes are the sister group of the tetra-
pods.

It is interesting that Camp (1923) dis-
covered correctly the three iguanian
genera possessing the M. pectoralis pars
cutaneous, his vestigial M. rec. abd.
“superticialis,’ not only because so few

‘genera were examined (a probablility of

circa 1% using the hypergeometric dis-
tribution with nonreplacement sampling)
but also because two of these genera,
Uromastyx and Physignathus, as then
recognized, were not monophyletic.

At the time of Camp’s work, Phy-

208

signathus included three species from
Australia (gilberti, longirostris, tem-
poralis) together with the originally
described P. lesueurii and P. cocincinus.
However, the former three are unrelated
to Physignathus and lack the compressed
swimming tail, M. pectoralis pars cu-
taneous, large lacrimal, and cross-shaped
interclavicle which Physignathus pos-
sesses. The scalation and body habitus of
these three species relate them with the
Amphibolurus muricatus species group
and Chlamydosaurus. Cogger (1975)
resurrected the genus Lophognathus for
these species and Moody (1980) dis-
cussed numerous characters separating
these genera.

Also at the time of Camp’s work,
Aporoscelis (Boulenger, 1885) included
two species, princeps and _ batilliferus,
and was considered a sister taxon of Ur-
omastyx. Anderson (1896: 34) recognized
that: “Uromastix batilliferus Vaill. from
its dentition and the form of the body, is
unquestionably a member of the genus
Agama, but with the tail of an Ur-
omastix.” His conclusions were not ac-
cepted until much later, when Parker
(1942) reassigned princeps to Uromastyx
and placed batilliferus in the new genus
Xenagama, though without discussing
the morphological criteria for the reas-
signments. Parker’s decision is substanti-
ated because U. princeps possesses the
specialized cranial, skeletal, dentitional,
and scalational features characteristic for
Uromastyx, (Moody, 1980), as well as
having the M. pectoralis pars cutaneous.
Xenagama batillifera lacks the special-
ized skull and dentition and the cutane-
ous muscle. On the basis of scalation and
skeleton, Xenagama and Stellio are sister
genera and related to Agama. Although
the tails of Xenagama have very large
spiny tubercles, they are not similar to
the tails of Uromastyx.

The M. pectoralis pars cutaneous of the
four iguanian genera is amazingly similar
in structure and presumably in function
to the thoracic-axillary portion of the M.
panniculus camosus, which is also de-

Advances in Herpetology and Evolutionary Biology

velopmentally related to the M. pec-
toralis, in mammals. Various cutaneous
muscles are known in a few amphibians
and reptiles (Romer, 1970). In humans, a
slip of the M. pectoralis (embryonically
part of the M. panniculus carnosus) fuses
with the adjacent M. rectus abdominis
and M. obliquus abdominis externus. It
may insert in the brachial fascia inde-
pendent of the main tendon of the M.
pectoralis (Anson, 1966). In the rhesus
monkey, the M. panniculus carmosus of
the lateral thorax is remarkably similar to
the M. pectoralis pars cutaneous, defi-
nitely effects movement of skin over the
glenohumeral joint and helps the M. pec-
toralis in depressing the forelimb from an
elevated position (Hartman and Straus,
1933).

ACKNOWLEDGMENTS

I appreciate the permission granted to
me to make small incisions in specimens
loaned to me by the following curators
and collections: A. G. C. Grandison
(BMNH), A. Leviton (CAS), J. Wright
(LACM), H. Marx (FMNH), E. E. Wil-
liams (MCZ), A. E. Greer (MVZ), B.
Lanza (MZUF), O. Rieppel (NHMB), T.
Fritts (SDNHM), H. Wermuth (SMNS),
A. G. Kluge (UMMZ), G. Zug (USNM),
G. Storr (WAM), and W. Bohme (ZFMK).
I especially thank Alan Leviton for re-
checking my observations on Uromastyx
loricatus and Joseph Eastman for con-
firming other observations and for pro-
viding comparative information on
mammals. Joseph Eastman, Arnold
Kluge, and Wolfgang Bohme construc-
tively criticized the manuscript. I thank
Paris Mihely for preparing the figures.

Travel money to visit the USA herpe-
tological collections was provided by a
Rackham Graduate School Block Grant,
University of Michigan. A fellowship
from the Deutscher Akademischer
Austauschdienst of Bonn, Germany,
allowed me to visit all major herpetologi-
cal collections in Europe. The College of

M. RECTUS ABDOMINIS OF LACERTILIANS - Moody

Osteopathic Medicine, Ohio University,
provided funds for the final morphologi-
cal comparisons and for the illustrations.

APPENDIX: SPECIMENS EXAMINED

Specimens examined are listed below
in alphabetical order within each family.
The museum collection acronyms follow
the published list by the Joint Herpeto-
logical Resources Committee (Herpeto-
logical Review, 11(4), 1980: 93-102).

IGUANIDAE: Amblyrhynchus cristatus, FMNH
19752, UMMZ 50122; Anisolepis undulatus, MCZ
123812; Anolis angusticeps, UMMZ 115617; A.
bimaculatus, UMMZ 60215; <A. carolinensis,
UMMZ 86466; A. chrysolepis planiceps, UMMZ
57704; A. coelestinus, UMMZ 110211; A. crist-
atellus, UMMZ 73703; A. equestris, UMMZ 58531;
A. fuscoauratus, UMMZ 123913; A. grahami,
UMMZ 85883; A. latifrons, UMMZ 48408; A. onca,
UMMZ 54806; A. petersi, UMMZ 91202; A. sagrei,
UMMZ 85888; A. valencienni, UMMZ 85913;
Aperopristis catamarcensis, MCZ 96710;
Aptycholaemus longicauda, BMNH_ 1946.8.9.2;
Basiliscus basiliscus, UMMZ 78493; B. galeritus,
UMMZ 51276; Brachylophus fasciatus, UMMZ
59772; Callisaurus draconoides, UMMZ 134018;
Chamaelinorops barbouri, MCZ 146951; Cha-
maeleolis chamaeleontides, MCZ 38419, C. porcus,
UMMZ 98014; Conolophus subcristatus, UMMZ
50121; Cophosaurus texanus, UMMZ_ 125303;
Corythophanes cristatus, UMMZ 46446; Crot-
aphytus collaris, UMMZ 81931; Ctenoblepharis
adspersus, MCZ 145040; Ctenosaura pectinata,
UMMZ 118975; Cyclura cornuta stejnegeri, UDMMZ
73899; Diplolaemus bibroni, MCZ 14918; Dips-
osaurus dorsalis, UMMZ 105886; Enyaliodes
heterolepis, MCZ 112347, MCZ 6983; E. l. laticeps,
MCZ 12438; E. laticeps festae, MCZ 43752; E.
oshaughnessyi, MCZ 29297; E. palpebralis, MCZ
84035; Enyaliosaurus quinquecarinatus, UMMZ
82376; Enyalius bilineatus, UMMZ 65946; E.
brasiliensis, UMMZ 108627; E. iheringii, UMMZ
108629; Gambelia wislizenni, UMMZ 61997;
Holbrookia maculata, UMMZ 122308; Hoplocercus
spinosus, UMMZ 108259; Iguana iguana, UMMZ
98389; Laemanctus serratus, UMMZ 110779; Lei-

ocephalus carinatus, UMMZ 85927; Leiosaurus -

belli, UMMZ 89655; Liolaemus alticolor walkeri,
UMMZ 133307; Morunasaurus annularis, USNM
6275; M. groi, MCZ 34875; Ophryoessoides or-
natus, UMMZ 90802; Petrosaurus mearnsi, MCZ
100068; Phenacosaurus richteri, UMMZ 65203;
Phrynosoma cornutum, UMMZ 136151; Phy-
maturus palluma, MCZ 2033; Platynotus semi-
taeniatus, MCZ 126647; Plica plica, UMMZ 84744,

209

Polychrus acutirostris, UMMZ 68076; P. per-
uvianus, UMMZ 107615; Pristidactylus scapulatus,
MCZ 33584; Proctotretus azureus, MCZ 133257; P.
pectinatus, MCZ 66989; Sator angustus, UMMZ
124060; Sauromalus obesus, UMMZ 90706; Scel-
oporus formosus, UMMZ 118798; S. graciosus,
UMMZ 70585; S. grammicus disparilis, UMMZ
79442; S. horridus albiventris, UMMZ 101975; S.
poinsetti, UMMZ 118930; S. scalaris, UMMZ
71451; S. undulatus, UMMZ 97528; S. utiformis,
UMMZ 115571; S. variabilis, UMMZ 105424,
Stenocercus guentheri, MCZ 146926; Strobilurus
torquatus, UMMZ 115646; Tropidurus peruvianus,
UMMZ 83102; Uma notata, UMMZ 67367; Ur-
acentron flaviceps, MCZ 37270; Uranoscodon
superciliaris, MCZ 123748; Urosaurus bicarinatus,
UMMZ 115037; Urostrophus torquatus, MCZ
123750; U. vautieri, MCZ 132256; Uta stansburi-
ana, UMMZ 76480. OPLURINAE: Chalarodon
madagascariensis, MCZ 11507, 11538; Oplurus
cyclurus, MCZ 11718, 37189; O. fierinensis, MCZ
11627; O. grandidieri, MCZ 7735; O. quadri-
maculatus, MCZ 67945; O. saxicola, MCZ 49659;
CHAMAELEONIDAE: Bradypodion pumilus,
UMMZ 60476; B. ventralis UMMZ 61531,
Brookesia stumpffi, UMMZ 65554; B. superciliaris,
UMMZ 65555; Chamaeleo africanus, UMMZ
113491; C. bitaeniatus, UMMZ 70271, C. brev-
icornis, UMMZ 59763; C. chamaeleon, UMMZ
1134934; C. fallax, UMMZ 65571; C. gracilis,
UMMZ 116897; C. jacksonii, UMMZ 61184; C.
lateralis, UMMZ 71611; C. namaquensis, UMMZ
65566; C. oustaleti, UMMZ 59764; C. owenii,
UMMZ 60477; C. tigris, UMMZ 65733; C. ver-
rucosus, UMMZ 65746; Rhampholeon boulengeri,
UMMZ 60479; R. spectrum, UMMZ 35619.
AGAMIDAE: Acanthosaura armata, CAS 73721;
A. crucigera, MVZ 79065; A. lepidogaster, MVZ
23431; Agama (Agama) agama, MVZ 97473; A. (A.)
anchietae, MVZ 75527; A. (A.) atra, MVZ 69241; A.
(A.) caudospinosa, MCZ 34957; A. (A.) doriae, MCZ
15829; A. (A.) hartmanni, MCZ 8809; A. (A.) his-
pida, MVZ 69252-53; A. (A.) kirkii, MCZ 67813; A.
(A.) mossambica, MCZ 50192; A. (A.) planiceps,
MVZ 9810; A. (A.) rueppelli, MCZ 49120; A. (A.)
sankaranica, MVZ 75522; A. (A.) weidholzi, MCZ
44311; Amphibolurus barbatus, CAS 77596; A.
caudicinctus, LACM 55105, 55117; A. clayi, LACM
90023, 55046; A. cristatus, LACM 55179, 55182; A.
diemensis, MVZ 81662; A. femoralis, MCZ 86572;
A. fionni, MCZ 10187; A. fordi, MVZ 81669; A.
inermis, CAS 84102; A. isolepis, MVZ 81666; A.
maculatus, MVZ 77631; A. microlepidotus, CAS
94360; A. minimus, MCZ 32969; A. minor, CAS
16214; A. muricatus, CAS 34024, 77645; A. nobbi,
CAS 16215; A. ornatus, LACM 55030; A. pictus,
CAS 16215; A. reticulatus, MVZ 78129; A. rufescens,
MCZ 38595; A. scutulatus, LACM 55011, 55013;
Aphaniotis acutirostris, CAS 64262, A. fusca, MCZ
38973; Brachysaura minor, MCZ 7211; Bronch-
ocela cristatellus, CAS 8447, 8470, 60546; B.
jubatus, CAS 55125-26; B. marmoratus, CAS

210

15457; Caimanops amphiboluroides, LACM 55285;
Calotes calotes, MCZ 39828; C. elliotti, MCZ 6207;
C. emma, MCZ 45268; C. jerdoni, MCZ 7204; C.
maria, MCZ 7203; C. mystaceus, CAS 73730; C.
nigrilabris, MCZ 4119; C. rouxii, CAS 104298; C.
versicolor, CAS 18084, 98885; Ceratophora stod-
dartii, SDNHM 25450, MCZ 116378; C. tennentii,
MCZ 4126: Chelosania brunnea, WAM 11808,

11810; Chlamydosaurus kingii, MVZ 77635;
Cophotis ceylanica, BMNH_ 72.1.26.7, MCZ
118596; Coryophophylax subcristatus, MCZ

27094; Dendragama boulengeri, FMNH 109794;
Diporiphora australis, MVZ 74732; D. bilineata,
MVZ 81663; D. winneckei, LACM 55329, 55331;
Draco blanfordii, MCZ 39089; D. fimbriatus, MCZ
39095; D. formosus, MVZ 145268; D. lineatus
beccarii, MCZ 25354; D. maculatus, MVZ 23401;
D. maximus, MCZ 39100; D. quinquefasciatus,
MCZ 39126; D. spilopterus, MCZ 44141; D.
taeniopterus, MVZ 15980; D. volans, MCZ 79697;
Gonocephalus bellii, MCZ 15176; G. beyschlagi,
MCZ 25911; G. chamaeleontinus, MCZ 27008; G.
grandis, MVZ 111826; G. interruptus, CAS 25858;
G. liogaster, MVZ 111835; G. miotympanum, MCZ
11265; G. robinsonii, MCZ 39172; G. semperi, CAS
23551; G. sophiae, CAS 19870; Hydrosaurus
amboinensis, SDNHM 25983, MCZ 9329; H. pus-
tulatus, CAS 24607, 28171; H. weberi, ZFMK
13993, MCZ 4574; Hylagama borneensis, FMNH
131540, 138196; Hypsilurus (Hypsilurus) binota-
tus, MCZ 130351, 137283; H. (H.) boydii, MVZ
77648; H. (H.) bruijni, MCZ 126436; H. (H.)
dilophus, CAS 64251; H. (H.) godeffroyi, MVZ
74903; CAS 54718, MCZ 4706; H. (H.) nigrigularis,
CAS 135207; H. (H.) papuensis, MCZ 44182; H. (H.)
spinipes, CAS 84031; Hypsilurus (Arua) auritus,
MCZ 96203; H. (Arua) modestus, CAS 13552;
Japalura brevipes, MVZ 111835; J. dymondi, CAS
63926; J. flaviceps, MCZ 12469; J. planidorsata,
MCZ 7197; J. polygonata, CAS 16262; J. splendida,
MCZ 20700; J. swinhonis, MVZ 23391; J. tricari-
nata, MCZ 58288; J. variegata, MCZ 7196; J. yun-
nanensis, CAS 31435; Lophocalotes ludekingi,
MCZ 38974; Lophognathus gilberti, CAS 77637-8;
L. longirostris, CAS 94365, MCZ 35196; L. tem-
poralis, MVZ 81675, MCZ 22906, 137294; Lyri-
ocephalus scutatus, MCZ 7155; Moloch horridus,
MVZ 82832, MCZ 6980; Oriocalotes paulus,
BMNH 70.11.29.43; Otocryptis wiegmanni, CAS
16872; Phoxophrys borneensis, MCZ 43487; P.
nigrilabris, MCZ 9036; Phrynocephalus arabicus,
MCZ 44759; P. axillaris, MCZ 7226; P. clarkorum,
CAS 120236; P. forsythii, MCZ 22174; P. guttatus,
CAS 99719; P. helioscopus, CAS 105806; P. inter-
scapularis, CAS 120107; P. luteoguttatus, CAS
120004; P. maculatus, MCZ 7225; P. mystaceus,
CAS 99716-7; P. ornatus, CAS 96273; P. przewal-
skii, MCZ 22177; P. reticulatus, CAS 99722; P.
rossikowi, CAS 94490; P. scutellatus, CAS 97978; P.
versicolor, CAS 99721; P. vlangalii, MCZ 22181;
Physignathus cocincinus, MCZ 20285, 39180,
39182; ZFMK 21459: P. lesueurii, UMMZ 65490,

Advances in Herpetology and Evolutionary Biology

CAS 77644, 77647, MCZ 10176, 35212; Psam-
mophilus blanfordanus, CAS 94353, 104104; P.
dorsalis, MCZ 7209; Pseudocalotes floweri, MCZ
39068; P. tympanistriga, MCZ 7647, 22160; Ptyc-
tolaemus gularis, BMNH_ 1940.6.3.71-2; Salea
anamallayana, CAS 104246-7; Salea horsfieldii,
CAS 94351, MCZ 7198; S. kakhiensis, MCZ 43055;
Sitana ponticeriana, MCZ 7153; Stellio agrorensis,
MCZ 7217; S. atricollis MCZ 18590; S. caucasica,
CAS 120747; S. cyanogaster, MCZ 50551; S.
erythrogaster, CAS 111930; S. himalayana, CAS
115934; S. lehmanni, CAS 94489; S. melanura,
MCZ 50192; S. nupta, CAS 86332; S. stellio, MVZ
97851, 97853; S. stoliczkana, CAS 99713; S.
tarimensis, MCZ 33577; S. tuberculata, CAS
90711; Trapelus (Trapelus) agilis, MCZ 45627; T.
(T.) flavimaculata, CAS 97587; T. (T.) impalearis,
MCZ 29957; T. (T.) mutabilis, MVZ 97477, 97480;
T. (T.) persicus, CAS 84611; T. (T.) rubrigularis,
MCZ 7213; T. (T.) ruderata, MCZ 9810; T. (T.)
savignii, MVZ 97482-83; Trapelus (Pseudo-
trapelus) sinaita, MCZ 103840; Tympanocryptis
lineata, MVZ 81663; T. tetraporophora, UMMZ
132278; Xenagama batillifera, MCZ 49121; X.
taylori, MCZ 49215-6. Also a new genus is being
described for “Cophotis” sumatrana, ZF MK 20790.

UROMASTYCIDAE: Leiolepis b. belliana,
ZFMK 13156, MCZ 122289; L. belliana ocellata,
MCZ 43088; L. guttata, MCZ 16656; L. peguensis,
BMNH 74.4.29.624; L. r. reevesii, MCZ 25185; L.
reevesii rubritaeniata, MCZ 74102; L. triploida,
MCZ 103696—7; Uromastyx a. acanthinurus, CAS
123485, 123489; U. acanthinurus werneri, MCZ
27384; U. aegyptius, MVZ 97471, ZFMK 2704; U.
asmussi, NHMB 14352, BMNH 74.11.23.9, ZFMK
7925; U. benti, BMNH 99.12.13.51, 1938.2.1.47; U.
geyri, BMNH 1932.9.10.15, CAS 103072; U. hard-
wickii, CAS 99787, ZFMK 8616-7; U. loricatus,
NHMB 3491, BMNH 1933.4.1.25, SMNS 607, 609,
CAS 86468, 86470, 102480; U. macfadyeni, BMNH
1937.12.5.117, BMNH 1937.12.5.120, 1946.8.14.54;
U. microlepis, ZFMK 20267, BMNH 1920.3.20.2,
1971.749; U. ocellatus, BMNH _ 97.10.28.210,
1937.8.38, SMF 10402; U. ornatus, SMF 10404,
ZFMK 8576, MVZ 97472; U. philbyi, MCZ 44760,
BMNH 1953.1.8.52; U. p. princeps, MCZ 70621,
BMNH_ 1931.7.20.276; U. princeps _ scortecci,
MZUF 113; U. thomasi, BMNH_ 1956.1.16.8,
1971.1355.

LITERATURE CITED

ANDERSON, J. 1896. Herpetology of Arabia. London,
122 pp.

ANSON, B. J. 1966. Editor, Morris’ Human Anatomy.
12th edition. McGraw-Hill Co.

BOULENGER, G. A. 1885. Catalogue of the lizards in
the British Museum (Natural History), Second
Edition, Volume I. London, 436 pp.

BYERLY, T. C. 1925. The myology of Sphenodon

M. RECTUS ABDOMINIS OF LACERTILIANS - Moody

punctatum. Univ. Iowa Studies Natural His-
tory, 11(6): 1-51, pl. i-iv.

Camp, C. L. 1923. Classification of the lizards. Bull.
Amer. Mus. Nat. Hist., 48: 289-481, fig. A-L
and J-112.

Coccer, H. G. 1975. Reptiles and Amphibians of
Australia. Reed, London, 608 pp.

DEVis, C. W. 1884. Myology of Chlamydosaurus
kingi. Proc. Linn. Soc. New South Wales, 8:
300-320.

Donoso-Barros, R. 1973. Un nuevo saurio de
Bolivia. Neotropica, 19(60): 132-134.

DONOSO-BARROS, R., AND J. M. CEI. 1971. New liz-
ards from the Volcanic Patagonian Plateau of
Argentina. J. Herpetol., 5: 89-95.

ESTES, R. 1983. The fossil record and early distribu-
tion of lizards, pp. 365-398. In Anders G. J.
Rhodin and Kenneth Miyata (eds), Advances in
Herpetology and Evolutionary Biology: Essays
on Honor of Emest E. Williams. Cambridge,
the Museum of Comparative Zoology, 1983.

ETHERIDGE, R. 1959. The relationships of the
Anoles (Reptilia: Sauria: Iguanidae): An
interpretation based on skeletal morphology.
Doctoral Dissertation. University of Michigan
at Ann Arbor, 236 pp.

FURBRINGER, M. 1869. Die Knochen und Muskeln
der Extremitaten bei den schlangenahnlichen
Sauriem. Inaug. Dissert. Berlin. Wilh.
Engelmann, Leipzig, pp. 1-58.

. 1875. Zur vergleichenden Anatomie der
Schultermuskeln. Vol. III, Chap. IV, Saurier

und Crocodile. Morphol. Jahrb., 1: 636-816,

pls. xxiii-xxvii.

1900. Zur vergleichenden Anatomie des
Brustschulterapparates und der Schulter-
muskeln. Jena. Zeitschr., 34: 215-718, pls. xiii-
XVii.

Gapow, H. 1882. Untersuchungen wber die
Bauchmuskeln der Krododile, Eidechsen und
Schildkroten. Morphol. Jahrb., 7: 57-100, pl. i.

Gorski, C. 1852. Uber das Becken der Saurier.
Inaug. Dissert. Dorpat.

HARTMAN, C. G., AND W. L. STRAuS, JR. 1933. The
anatomy of the Rhesus Monkey. New York,
Hafner Co.

HILLENIUS, D. 1959. The differentiation within the
genus Chamaeleo Laurenti, 1768. Beaufortia,
8(89): 1-92.

HOFFMAN, C. K. 1890. Dr. H. G. Bronn’s Klassen
und Ordnungen des Thierreichs, wissen-
schaftlich dargestellt in Wort und Bilt. Band

VI, Abt. III. Reptilien (Eidechsen II). Leipzig,

443 and 1399 pp.
KLUGE, A. G. 1967. Higher taxonomic categories of
gekkonid lizards and their evolution. Bull.
Amer. Mus. Nat. Hist., 135: 1-60.

1976. A reinvestigation of the abdominal
musculature of Gekkonoid lizards and its bear-
ing on their phylogenetic relationships.
Herpetologica, 32: 295-298.

Maurer, F. 1896. Die ventrale Rumpfmuskulatur

211

einiger Reptilien. Eine vergleichend-
anatomische Untersuchung. Festschrift fur
Gegenbaur, 1: 183-256, pls. i-iv.

—___. 1898. Die Entwicklung der ventralen
Rumpfmuskulatur bei Reptilien. Morphol.
Jahrb., 26: 1-60, pls. i-iii.

MiIvakT, ST. G. 1867. Notes on the myology of Iguana
tuberculata. Proc. Zool. Soc. Lon; 1867: 766—
797.

___. 1870. On the myology of Chamaeleo parsonii.
Proc. Zool. Soc. London, 1870: 850-890.

Moopy, S. M. 1980. Phylogenetic and historical
biogeographical relationships of the genera in
the family Agamidae (Reptilia: Lacertilia).
Doctoral Dissertation, University of Michigan
at Ann Arbor, xv + 373 pp.

NORTHCUTT, R. G. 1978. Forebrain and midbrain
organization in lizards and its phylogenetic
significance. In, N. Greenberg and P. D.
MacLean (eds.) Behavior and Neurology of
Lizards. NIMH, Rockville, Maryland.

PARKER, H. W. 1942. The lizards of British Somali-
land. Bull. Mus. Comp. Zool., 91: 1-101.

Raw, L. R. G. 1976. A survey of the dwarf chame-
leons of Natal, South Africa, with descriptions
of three new species (Sauria: Chamaeleonidae)
Durban Mus. Novit., 11: 139-161.

ROBINSON, P. L. 1967. The evolution of the Lacer-
tilia. Colloques int. Cent. natn. Rech. Scient.
No. 163, pp. 395-407.

RoMER, A. S. 1970. The vertebrate body, fourth ed.
Philadelphia, Saunders Co.

ROSEN, D., P. FOREY, B. GARDINER, AND C. PATTER-
SON. 1981. Lungfishes, Tetrapods, Pale-
ontology, and Plesiomorphy. Bull. Amer. Mus.
Nat. Hist., 167: 159-276.

RUDINGER. 1968. Die Muskeln der verdem Ex-
tremitaten der Reptilien und Vogel. Hol-
landsche. Maatsch. Wetensch. Haarlem, 4: 1—
187.

SANDERS, A. 1870. Notes on the myology of Platy-
dactylus japonicus. Proc. Zool. Soc. London,
1870: 413-426.

____. 1872. Notes on the myology of Liolepis belli.
Proc. Zool. Soc. London, 1872: 154-183.
___.. 1874. Notes on the myology of Phrynosoma
coronatum. Proc. Zool. Soc., London, 1874:

71-89.

SHUFELDT, R. W. 1890. Contributions to the study of
Heloderma suspectum. Proc. Zool. Soc. Lon-
don, 1890: 148-244.

SmiTH, H. M. 1939. The Mexican and Central
American lizards of the genus Sceloporus.
Fieldiana, Zoology, 26: 1-397, 31 pls.

SMITH, M. A. 1935. The Fauna of British India, in-
cluding Ceylon and Burma. Reptilia and Am-
phibia. Vol. II. Sauria. London, Taylor and
Francis, Ltd., 440 pp.

STANNIUS, H. 1856. Handbuch der Anatomie der
Wirbelthiere. In Siebold und _ Stannius,
Handbuch der Zootomie, Part II, Die Wirbel-
thiere, 2nd ed. Berlin, 270 pp.

22 Advances in Herpetology and Evolutionary Biology

SUKHANOV, V. B. 1961. Nekotorye voprosy filogenii i xvii. In Camp’s classification of lizards. Soc.
sistemy Lacertilia (sue Sauria). Zool. Zhumal Stud. Amphib. Rept. Fascimilae Reprint |
60(1): 73-83. (In Russian). Series.

____. 1976. Some problems of the phylogeny and WERMUTH, H. 1966. Liste der rezenten Amphibien
systematics of the Lacertilia. Translation of the und Reptilien: Chamaeleonidae. Das Tier-
Russian original. S. Moody and G. Zug, eds. reich (Berlin), 83: i-x, 1-37.

Smithsonian Herpetological Information Ser- ___. 1967. Liste der rezenten Amphibien und
vice No. 38, pp. 1-15. Reptilien: Agamidae. Das Tierreich (Berlin),

UNDERWOOD, G. 1971. A-modem appreciation of 86: i-xiv, 1-127.

Camp’s “Classification of the lizards,” pp. vii—

On the Adaptive Significance of the Reptilian
Spectacle: the Evidence from Scincid, Teiid,

And Lacertid Lizards
ALLEN E. GREER'

ABSTRACT. In scincid, teiid, and lacertid lizards
the spectacle, a permanent transparent scale cover-
ing the eye, is associated with species of small size
and diurnal habits inhabiting relatively dry habi-
tats. Presumably the spectacle reduces evaporative
water loss from the eye. A similar morphology and
ecology may have been associated with the evolu-
tion of the spectacle in other squamates (lizards and
snakes). The morphology of the squamate eyelid
may be tracking, evolutionarily, the interplay be-
tween body size and habitat.

INTRODUCTION

Many squamate reptiles (lizards and
snakes) have a clear, disc-shaped scale
permanently covering the eye. This
structure, known as a spectacle or brille,
appears to have evolved at least eight
times in scincids (see below) and at least
once in each of the following six taxa—
teiids (Gymnophthalmus and Micr-
ablepharus), lacertids (Ophisops),
gekkonids and pygopodids, xantusiids,
amphisbaenians, and snakes. The spec-
tacle has always been assumed to have
great systematic importance in squa-
mates and in the past was the hallmark
for vast polyphyletic assemblages, e.g.,
the subfamily “Ophiopthalmes” (Du-

meril and Bibron, 1836) and the genus .

Ablepharus (Boulenger, 1887).
Comparative morphological (Bellairs
and Boyd, 1948; Fuhn, 1969; Greer,

1The Australian Museum, Sydney, New South
Wales, 2000 Australia.

1974; Smith, 1935) and embryological
(Schwartz-Karsten, 1933; Neher, 1935)
studies suggest that the spectacle prob-
ably evolved in the following sequence:
1) the lower eyelid moveable and en-
tirely covered by small scales, 2) the eye-
lid moveable but with either a few cen-
tral transparent scales or “windows” or
only a single large scale or “window”
that fills most of the eyelid, and 3) the
large, clear eyelid fused in the raised
position to give the spectacle? (Fig. 1).
Four hypotheses have been advanced
to explain the adaptive significance of the
spectacle or its precursor, the windowed
eyelid. 1) The spectacle has no adaptive
significance on the evidence of its having
no discernable ecological correlates
(Smith, 1935, 1939; Bellairs, 1970). 2)
The spectacle protects the eye from phy-
sical damage by foreign bodies on the
evidence of a supposed correlation with
deserticolous (wind-blown sand), sub-
terranean (soil particles) or noctumal
(reduced visual acuity) habits (Walls,
1934, 1942). 3) The windowed eyelid
provides facultative protection from

2A fourth condition which need not concern us
here may occur in extreme burrowers: the eye is
vestigial and the “eyelid” is a covering scale similar
to the surrounding head scales. Examples are most
amphisbaenians, dibamids, a few scincids and
scolecophidians. Little work has been done on the
precursors of this condition, but my own experi-
ence, mainly with scincids, suggests that it can
evolve from any of the other three conditions.

214

Figure 1. Close-up photographs of the heads of three
different scincid lizards to show the three main struc-
tural stages in the evolution of the spectacle in lizards
and snakes. Bottom: lower eyelid moveable (shown
here raised) and scaly (Sphenomorphus tympanum);
middle: lower eyelid moveable (shown here raised) but
with a clear “window” (Carlia amax), and top: the large
clear spectacle, permanently covering the eye (Cryp-
toblepharus plagiocephalus).

strong solar radiation (Plate, 1924;
Mertens, 1954) on the evidence that a)
certain windowed-eyed lizards—and as it
happens species that also have some
pigmentation in the eyelid—have been
observed to raise their eyelids in strong
sunlight (Mertens, 1954) and b) the only
two anoline lizards with windowed eye-

Advances in Herpetology and Evolutionary Biology

lids (as opposed to scaly eyelids)—and
also with some pigment in the eyelid—
are suspected of having crepuscular,
cave-dwelling habits and are inferred to
protect their presumably light sensitive
eyes by raising their eyelids in daylight
(Williams and Hecht, 1955). 4) Both the
windowed eyelid and the spectacle re-
duce evaporative water loss from the eye
(Amold, 1973) on the evidence of a) an
experimental observation that corneal
transpiration may account for over 20% of
total transpirational loss in a small lizard
at normal activity temperatures (Reich-
ling, 1957; also see Mautz, 1980) and b)
anecdotal but informed observation that
both eyelid types are widespread in
small, sun-dwelling lizards (Amold,
1973).

The purpose of this paper is fourfold.

First, to test a prediction of the evapora-
tive water loss hypothesis using the three
squamate families in which the spectacle
occurs in lineages of low taxonomic rank
and in which, therefore, its original adap-
tive associations may be especially clear.
Second, to show that the evaporative
water loss hypothesis either contradicts,
or explains more parsimoniously, the ob-
servations advanced in support of the
other hypotheses. Third, to suggest that
the evaporative water loss hypothesis has
heuristic value in thinking about the
evolution of those groups in which the
spectacle characterizes lineages of high
taxonomic rank and in which, therefore,
its original adaptive significance may be
obscure. And fourth, to present a more
general hypothesis about the evolution of
all three major eyelid types in squamates.

RESULTS

The evaporative water loss hypothesis
predicts that taxa with a spectacle will be
as small or smaller than their nearest re-
latives with moveable eyelids and that
they will occur in habitats that are as dry
or drier. This prediction arises in part
from the classic allometric relationship
between eye size and body size: in a

series of similarly shaped but different
sized animals smaller forms have rela-
tively larger eyes (Walls, 1942; Rensch,
1959; Gould, 1966). This allometric rela-
tionship has been verified for scincids.
Regression analysis of a log—log plot of
eye surface area versus body weight in 20
diurnal, terrestrial scincids represented
by single adult individuals with com-
plete tails resulted in a regression co-
efficient (0.27; 95% confidence inter-
val = + 0.09) well below even isometric
conditions (0.68) under which, of course,
eye surface area would still not increase
as fast as body volume. Regression analy-
sis was by Bartlett's method as described
by Simpson, Roe, and Lewontin (1960).

The predicted morphological and
ecological associations of the spectacle
are especially clear in the eight scincid
and single teiid and lacertid lineages in
which this eyelid type has evolved (Fig.
2). These lineages are described briefly
below.

1) The Australian scincid genera
Morethia (8 species) and Proablepharus
(3 species) both have a spectacle which
they probably inherited from a common
ancestor (Greer, 1980). Both genera are
diurnal and terrestrial. Morethia occurs
in open habitats throughout Australia
except for the mesic southeast (Smyth,
1972; Storr, 1972; Greer, 1980), and
Proablepharus occurs in open habitats in
the arid and semi-arid interior and the
seasonally dry north (Storr, 1975; Greer,
1980). The taxon most closely related to
these two genera together is the tri-
lineatum species group of Leiolopisma (3
species). This group has windowed eye-
lids, diurnal and terrestrial habits and a
disjunct distribution in the mesic south-
west and southeast corners of Australia

(Greer, 1980). Morethia ranges in size .

from 33 to 58 mm, Proablepharus from 32
to 45 mm and the trilineatum species
group from 69 to 80 mm.! The two spec-

1The measure of species size employed through-
out is the maximum known snout-vent (SVL)
length. Tables giving maximum SVL, sample size,

REPTILIAN SPECTACLE : Greer 215

tacled genera are clearly both signifi-
cantly smaller (P < 0.001) and more xeric
adapted than their window-eyed rela-
tives. .

2) The Australian scincid genus Lerista
(37 species) occurs throughout most of
the continent but is centered over the
arid and semi-arid interior and the sea-
sonally dry north (Greer, 1967, 1979b;
Storr, 1971, 1976c). The genus is crypt-
ozoic and semifossorial. Nine species
have a spectacle, and 27 have windowed
eyelids (one species has vestigial eyes
and is therefore excluded from the analy-
sis). No attempt has been made to detect
a habitat difference between the two
groups, but in size the spectacled species
are significantly smaller (30-61 mm) than
their window-eyed relatives (51-104
mm) (P << 0.001). Storr (1971) clearly
understood the relationship between the
spectacle (his “ablepharine eye’) and
body size in Lerista when he said: “The
smallest species... [of the elegans spe-
cies group of Lerista] have acquired an
ablepharine eye (i.e., the lower eyelid is
immoveable and entirely transparent), a
condition for which there is evidently
strong selection pressure in litter inhabi-
ting skinks whose snout-vent length
averages less than 40 mm.”

3) The spectacled scincid genus Crypt-
oblepharus (37 species) occurs in two
disjunct areas: the southwest Pacific in-
cluding Australia and the western Indian
Ocean (Mascarene Islands, Madagascar,
and associated islands and the east Afri-
can coast). In most areas it occurs only
along open coasts (a dry habitat in terms
of potable water) (Mertens, 1931, 1958,
1964; Brongersma, 1942; Canaris and
Murphy, 1965), but in Australia it occurs
in a variety of habitats throughout most of
the arid, semi-arid, and seasonally dry

and sources of information for each species in every
taxon discussed are available from the author and
the Department of Herpetology, Museum of Com-
parative Zoology upon request. The distribution of
species sizes between taxa were compared using
the Mann-Whitney U Test as described by Siegel
(1956). The level of significance throughout is 0.05.

216 Advances in Herpetology and Evolutionary Biology
=. —— (3) LEIOLOPISMA
Ee ——--- —-—--— (8) MORETHIA ¢.001
—— —t— — 33) PROABLEPHARUS
ote es Sy a eee
2). eee : A ere (27) } LeRISTA «001
—2_ (7) CYANURA SP. GROUP OF EMOIA
5 3C ---Y— —— (37) CRYPTOBLEPHARUS «.00!
ic)
€ 4- —7— (2) NOTOSCINCUS
Faz
=v
s ii Sys 5) (2) } NOVAEGUINEAE SP. GROUP OF CARLIA car
S — Z — — (5) MENETIA .
2
© —_$_2____—___ (5) PANASPIS
tay SES J-——-— (7) P. CABINDAE + AFROABLEPHARUS <0)
eS) a (5) SCELO
LOTES
@ 70 —-— —-27----—-------— (4) TYPHLACONTIAs 08
£
——_—_—"—_ 17) SCINCELLA
g le | SSeS SS Z— — —— (5) ABLEPHARUS ».05
N ANOTOSAURA, BACHIA, HETERODACTYLUS,
JS of —— ee (23) IPHISA + TRETIOSCINCUS 206)
J—-— (7) GYMNOPHTHALMUS + MICRABLEPHARUS ;
Ee Lease
Olmak, es eee LT _!2) _ 6) CABRITA »>.05

OPHISOPS

Maximum Snout-vent Length of the Species (mm)

Figure 2. Ranges and means of maximum snout-vent lengths for the species in the scincid (1-8), telid (9), and
lacertid (10) lineages discussed in the text. In each lineage taxa with a permanent spectacle covering the eye are
indicated by a dashed line and their closest relatives with a moveable eyelid by a solid line. Numbers in parenthe-
sis are sample sizes. Numbers to the far right are P values (Mann-Whitney U Test) associated with the size
differences between the species with the two different eyelid types in each lineage. Data for individual species are

available from the author upon request.

areas (Storr, 1976a; Covacevich and
Ingram, 1978). Cryptoblepharus is most
closely related to the cyanura species
group of the genus Emoia (7 species)
which is widespread in the southwest
Pacific but extends only to the tip of Cape
York Peninsula in Australia (Greer,
1974). It is diurnal, terrestrial to semi-
arboreal, and occurs in habitats ranging
from moist, closed forest to open coasts
(Webster, 1969; McCoy, 1980). Crypt-
oblepharus is significantly smaller than
the cyanura species group of Emoia (34—
54 mm versus 45-61 mm; P << 0.001)
and appears to occur in habitats that are
as dry (e.g., open coasts) or drier (e.g., the
interior of Australia).

4) The spectacled, Australian scincid
genus Notoscincus (2 species) is diurnal

to crepuscular, terrestrial, and distributed
over the “less humid parts of northem
Australia” (Storr, 1974). The genus is a
member of the largest of the three skink
lineages in Australia (136 species out of a
total of 242), but otherwise its relation-
ships are obscure (Greer, 1979a). It is,
however, the smallest taxon in its lineage
in Australia (36-41 mm).

5) The novaeguineae species group of
the scincid genus Carlia (5 species) is
diurnal and ranges from southern New
Guinea and extreme eastern Australia as
far south as Sydney. The largest species
(42 mm) occurs in moist closed forest to
seasonally dry woodland and has a win-
dowed eyelid. The other four species, all
Australian, occur in seasonally dry open
forest to open woodland. The second

30

20

NUMBER OF SPECIES
S)

~ 20

IS 40" 6l SAalOS aa24.

l45 |66

REPTILIAN SPECTACLE : Greer Q17

SPECTACLE
N=83

WINDOWED EYELID
N=354

Ie Sp
262 3I9

SCAMS EYED
N= 48

y ara 7 are
Iis¢ 208 229 250 274 25 gsr
283 343

MAXIMUM SNOUT-VENT LENGTH (3mm intervals)

Figure 3. Frequency distributions of maximum snout-vent lengths for all skink species with one of the three basic

eyelid types: scaly, windowed, and spectacled.

largest of these (38 mm) has a windowed
eyelid; the others have a spectacle. The
novdeguineae species group's nearest
relative appears to be the diumal, spec-
tacled Australian genus Menetia (5 spe-
cies) (Greer, 1974; Ingram, 1977) which
occurs throughout the arid and semi-arid
interior of Australia and the seasonally
dry north (Storr, 1976b, 1978; Rankin,
1979) in shrubland and open forest. No
Menetia exceeds 38 mm. Taking these
two groups together, it is evident that
there is no significant size difference
between the spectacled and window-
eyed species (P < 0.07) but the former
occur in habitats that are generally drier.

6) The subsaharan African scincid sub-
genus Panaspis (6 species) is diurnal and

_ African

terrestrial. Five species have windowed
eyelids; four of these occur in moist
closed forest and the fifth occurs in
wooded savanna (Fuhn, 1972; Perret,
1973, 1975). A sixth species has a spec-
tacle and inhabits dense humid forest,
mangrove forest, and savanna at the edge
of forest (Fuhn, 1970). This latter species
is structurally similar to the hypothetical
ancestor of the spectacled, subsaharan
subgenus Afroablepharus (6
species) (Fuhn, 1970) which is diurnal
and terrestrial and associated with
savanna (Perret, 1975). The seven spec-
tacle-eyed species in these two groups
together are significantly smaller than
the five window-eyed species (24-55 mm
versus 43-68 mm; P < 0.01) and appear to

218

occur in generally drier habitats (savanna
versus forest).

7) The spectacled, south African scin-
cid genus Typhlacontias (4 species) is
fossorial and occurs in the aeolian sands
of the northern Namib Desert (3 species)
and in a relatively moist area of the
northern Kalahari (1 species) (D. G.
Broadley, in litt.). Typhlacontias is most
closely related to the south African genus
Scelotes (15 species) (Greer, 1970) which
has a moveable lower eyelid that is either
completely scaled or with a window. The
genus is cryptozoic to fossorial and is
restricted to the more mesic southern and
eastern parts of south Africa (D. G. Broad-
ley, in litt.). The genera do not differ
significantly in size (83-130 mm _ for
Typhlacontias versus 36-120 mm_ for
Scelotes; P >> 0.05), but Typhlacontias
appears to occur in generally drier habi-
tats.

8) The spectacled scincid genus Able-
pharus (5 species) is diurnal and terres-
trial and distributed over southeast
Europe and southwest Asia in Mediter-
ranean, prairie, steppe, and desert
biomes. Ablepharus is most closely re-
lated to the window-eyed genus Scin-
cella (Greer, 1974) which is diurnal and
terrestrial and distributed disjunctly in
central and south Asia and southeast
North America and Mexico in various
forest biomes. In size Ablepharus does
not differ significantly from neighboring
central Asian Scincella (7 species) (36-60
mm versus 47-65 mm; P > 0.05), but on
the basis of the biomes occupied, it ap-
pears to occur in generally drier habitats.

9) The only spectacled taxa among
teiids are the tropical American genera
Gymnophthalmus (6 species) and Micr-
ablepharus (1 species). The relationships
of Micrablepharus are poorly known but
are thought to lie close to Gymno-
phthalmus (W. Presch, personal com-
munication). The closest relatives of
these two genera are the window-eyed,
tropical American genera Anotosaura (3
species), Bachia (15 species), Heter-
odactylus (2 species), Iphisa (1 species),

Advances in Herpetology and Evolutionary Biology

and Tretioscincus (2 species) (Dixon,
1973, 1974; Presch, 1980). Both groups
occur in a variety of habitats, but in size
the spectacled taxa are significantly smal-
ler (41-48 mm) than their window-eyed
relatives (39-114 mm but only one spe-
cies less than 50 mm) (P < 0.001).

10) The only spectacled taxon of lacer-
tids is the diumal, terrestrial genus
Ophisops which ranges from west India
through southwest Asia and southeast
Europe to north Africa. Ophisops appears
to be most closely related to the diurnal,
terrestrial genus Cabrita which occurs in
central and eastern India (Smith, 1935).
The two genera do not differ significantly
in size (36-65 mm for Ophisops versus
44-57 mm for Cabrita; P >> 0.05), but it
is significant that both genera are small
and that the spectacled Ophisops ranges
over more generally arid country.

DISCUSSION

Before commenting on the spectacle in
other squamates, I critically review the
three other hypotheses that have been
advanced to explain the adaptive signifi-
cance of the spectacle and, where appro-
priate, compare them to the evaporative
water loss hypothesis. The hypothesis
that the spectacle has no adaptive signifi-
cance can be rejected, because, asser-
tions to the contrary notwithstanding
(Smith, 1935, 1939), there are clear eco-
logical associations with the spectacle.
The hypothesis that the spectacle pro-
tects the eye from physical damage
(Walls, 1934, 1942) is unsatisfactory for
two reasons. First, it offers three unre-
lated associations (burrowing, noctur-
nality, and life in desert environments) to
explain the adaptive significance of the
spectacle, whereas the evaporative water
loss hypothesis suggests only two related
associations. Second, two of its associa-
tions (burrowing and nocturnality) are
not convincing as there are many excep-
tions (e.g., many burrowing and noctur-
nal skinks have no spectacles), and the

third association (life in desert environ-
ments) is accounted for by the evapora-
tive water loss hypothesis. The hypothe-
sis that the windowed eyelid protects the
eye from solar radiation (Plate, 1924;
Mertens, 1954; Williams and Hecht,
1955) can be criticized on the grounds
that it confounds a modification that
reduces the amount of solar radiation
entering the eye, i.e., the eyelid pigment!
with a modification that permits vision
while the eyelid is temporarily closed to
reduce evaporative water loss, i.e., the
spectacle? The two modifications are
often associated but their roles are quite
distinct.

Unfortunately, the morphological and
ecological associations of the spectacle
are not as apparent in other squamates as
they are in scincids, teiids, and lacertids,
perhaps because subsequent evolution
has obscured them. The evidence on the
evolution of the spectacle is so telling in
the three lizard families discussed here,
however, that it may be useful to begin
using it in assessing the origin of the
spectacle in other groups. This may be
especially useful in those groups such as
non-eublepharine geckos and_ snakes
where the spectacle probably attended
the evolution of the group and where
there are several conflicting hypotheses
as to the origin of the group. In such con-
siderations, the two most important con-
clusions that can be drawn from the evi-
dence provided by scincids, teiids, and
lacertids are 1) the spectacle has evolved
many different times and 2) it has
evolved in relatively small forms (gener-

1A study of eyelid pigmentation in lizards would
be interesting in its own right. With regard to its
possible function as a sun screen, I note here that
window-eyed skinks living in alpine habitats in
southeast Australia, e.g., Leiolopisma greeni and L.
pretiosum, have intensely pigmented eyelids.

2Note that small, window-eyed skinks almost in-
variably close their eyelids after a brief period
when placed in an environment of relatively dry air
but subdued light intensity as, for example, on an
open desk in a dimly lit room.

REPTILIAN SPECTACLE - Greer 219

ally 65 mm or less in the lizards dis-
cussed here) living in relatively dry habi-
tats.

Finally, it may be useful for the pur-
poses of further testing to state the evapo-
rative water loss hypothesis in its logi-
cally broadest terms: the morphology of
the squamate eyelid (scaly, windowed, or
spectacled), through the intermediate
parameter of potential evaporative water
loss from the surface of the eye, may be
tracking, in an evolutionary sense, the
interplay between body size (as it affects
the eye surface area/body volume ratio)
and habitat (both temporal and spatial).
Hence, just as spectacles are generally
associated with small size and xeric habi-
tats, scaly eyelids should be associated
with large size and mesic habitats, and
windowed eyelids associated with in-
termediate size and intermediate (in
terms of moisture) habitats. Indeed, that
size alone accounts for much of the varia-
tion in eyelid type is evident from the
fact that in skinks, the largest and most
diverse family of lizards, the size dis-
tributions of the species with the three
different eyelid types are in the order
scaly > windowed > spectacled (P <
0.001 for each pairwise comparison;
Kolmogorov-Smimov Two Sample Test
as described by Siegel [1956]; Fig. 3).

ACKNOWLEDGMENTS

I thank the following people for pro-
viding me with summaries of the ecology
and/or size of groups about which I know
relatively little: J. P. Bacon, D. G. Broad-
ley, W. C. Brown, R. I. Crombie, J. R.
Dixon, W. D. Haacke, M. S. Hoogmoed,
A. G. Kluge, W. P. Maclean, J. L. Perret,
W. Presch, J. P. Rosado, R. Sage, R.

Thomas, P. E. Vanzolini, L. J. Vitt, and A.

Schwartz. I also thank C. Gans, S. J.
Gould, G. Underwood and E. E. Wil-
liams for critical readings of various
drafts and P. Greer for preparing the
graphs. Work for this report was sup-
ported in part by the Australian—

220

American Educational Foundation,
Canberra, A.C.T. (Fulbright Program).

LITERATURE CITED

ARNOLD, E. N. 1973. Relationships of the palaearctic
lizards assigned to the genera Lacerta, Alg-
yroides and Psammodromus (Reptilia: Lacer-
tidae). Bull. Br. Mus. nat. Hist., 25(8): 289-366.

BELLAIRS, A. D’A. 1970. The Life of Reptiles. New
York, Universe Books, 2 vols, 590 pp.

BELLAIRS, A.D’A., AND J. D. BoyD. 1948. The lacrimal
apparatus in lizards and snakes.—I. The brille,
the orbital glands, lachrymal canaliculi and
origin of the lachrymal duct. Proc. zool. Soc.
Lond., 117: 81-108.

BOULENGER, G. A. 1887. Catalogue of the Lizards in
The British Museum (Natural History). Vol. 3.
London, British Museum (Natural History),
O75 pp.

BRONGERSMA, L. D. 1942. Notes on scincid lizards.
Zool. Meded. Leiden., 24(1-2): 125-152.
Canaris, A. G., AND D. G. MurpHy. 1965. A scincid
reptile feeding primarily on marine crustacea,
with a note on its parasites. J. E. Africa nat.

Hist. Soc., 25(2): 129-130.

COVACEVICH, J., AND G. J. INGRAM. 1978. An unde-
scribed species of rock dwelling Crypt-
oblepharus (Lacertilia: Scincidae). Mem. Qd
Mus., 18(2): 151-154.

Dixon, J. R. 1973. A systematic review of the teiid
lizards, genus Bachia, with remarks on Heter-
odactylus and Anotosaura. Univ. Kans. Mus.
Nat. Hist. Misc. Publ. No. 57, 47 pp.

___. 1974. Systematic review of the lizard genus
Anotosaura (Teiidae). Herpetologica, 30(1):
13-18.

DuMERIL, A. M. C., AND G. BIBRON. 1836. Erpe-
tologie Générale ou Histoire Naturelle Com-
plete des Reptiles. Vol. 5. Paris, 854 pp.

Fuun, I. E. 1969. The “polyphyletic” origin of the
genus Ablepharus (Reptilia, Scincidae): a case
of parallel evolution. Z. f. zool. Systematik u.
Evolutionsforschung., 7(1): 67-76.

——_. 1970. Contribution a la systématique des
lygosomines Africains (Reptilia, Scincidae). I.
Les especes attribuées au genre Ablepharus.
Rev. Roum. Biol. -Zool., 15(16): 379-393.

___. 1972. Révision du phylum forestier du genre

Panaspis Cope (Reptilia, Scincidae, Ly-
gosominae). Rev. Roum. Biol. -Zool. 17(4):
PASTE AE

GOULD, S. J. 1966. Allometry and size in ontogeny
and phylogeny. Biol. Rev., 41: 587-640.
GREER, A. E. 1967. A new generic arrangement for
some Australian scincid lizards. Breviora Mus.

Comp. Zool. No. 267, 19 pp.

___. 1970. The systematics and evolution of the
subsaharan Africa, Seychelles, and Mauritius
scincine scincid lizards. Bull. Mus. comp.
Zool., 140(1): 1-23.

Advances in Herpetology and Evolutionary Biology

___. 1974. The generic relationships of the scincid
lizard genus Leiolopisma and its relatives.
Aust. J. Zool., Suppl. Ser. No. 31, 67 pp.

__. 1979a. A phylogenetic subdivision of Austra-
lian skinks. Rec. Aust. Mus., 32(8): 339-371.

____. 1979b. A new species of Lerista (Lacertilia:
Scincidae) from northem Queensland, with
remarks on the origin of the genus. Rec. Aust.
Mus., 32(10): 383-388. :

___. 1980. A new species of Morethia (Lacertilia:
Scincidae) from northem Australia, with
comments on the biology and relationships of
the genus. Rec. Aust. Mus., 33(2): 89-122.

INGRAM, G. J. 1977. Three species of small lizards—
two of them new. Genus Menetia (Lacertilia:
Scincidae) in Queensland. Victorian Nat.,
94(5): 184-187.

Mautz, W. J. 1980. Factors influencing evaporative

water loss in lizards. Comp. Biochem. Physiol.,

67A: 429-437.

McCoy, M. 1980. Reptiles of the Solomon Is-

lands. Wau Ecology Inst. Handb. No. 7, Wau,

Papua New Guinea, 80 pp.

MERTENS, R. 1931. Ablepharus boutonii (Desjardin)
und seine geographische Variation. Zool. Jb.
(Syst.), 61(1/2): 61-210.

___. 1954. Biologische Fahrten in Sudwest-Afrika.
3. Im Gebiete des Brandberges. Natur u. Volk,
84(6): 184-195.

____. 1958. Neue Ejidechsen aus Australien. Senck.
biol., 39(1/2): 51-56.

___.. 1964. Weitere Mitteilungen uber die Rassen
von Ablepharus boutonii, Il. Zool. Anz.,
173(2): 99-110.

NEHER, E. M. 1935. The origin of the brille in Cro-
talus confluentus lutosus (Great Basin Rattle-
snake). Trans. Am. ophthal. Soc. 33: 533-545.

PERRET, J. L. 1973. Contribution a l’étude des
Panaspis (Reptilia, Scincidae) d’Afrique oc-
cidentale avec la description de deux especes
nouvelles. Rev. Suisse Zool., 80(2): 595-630.

__. 1975. La différenciation dans le genre Pan-
aspis Cope (Reptilia, Scincidae). Bull. Soc.
Neuchatel. Sci. Natur. 98: 5-16.

PLATE, L. 1934. Allgemeine Zoologie und Abstam-
mungslehre. Vol. 2. Jena. (Not seen by me;
cited by Williams and Hecht, 1955.)

PRESCH, W. 1980. Evolutionary history of the South
American microteiid lizards (Teiidae: Gymn-
ophthalminae). Copeia, 1980(1): 36-56.

RANKIN, P. R. 1979. A taxonomic revision of the
genus Menetia (Lacertilia, Scincidae) in the
Norther Territory. Rec. Aust. Mus., 32(14):
491-499.

REICHLING, H. 1957. Transpiration und Vorzugs-
temperatur mitteleuropaischer Reptilien und
Amphibien. Zool. Jb. (Zool.), 67: 1-64.

RENSCH, B. 1959. Evolution above the Species
Level. London, Methuen, 419 pp.

SCHWARTZ-KARSTEN, H. 1933. Uber Entwicklung
und Bau der Brille bei Ophidiem und Lacer-
tiliern und die Anatomie ihrer Tranenwege.
Morph. Jb., 72: 499-540.

SIEGEL, S. 1956. Nonparametric Statistics for the
Behavioral Sciences. New York, McGraw-Hill
Book Co., 312 pp.

SIMPSON, G. G., A. ROE, AND R. C. LEWONTIN. 1960.
Quantitative Zoology. 2nd ed. New York;
Harcourt, Brace and World, Inc., 440 pp.

SmitH, M. A. 1935. The Fauna of British India.
Reptilia and Amphibia. Vol. II.—Sauria.
London, Taylor and Francis, 440 pp.

___. 1939. Evolutionary changes in the eye cover-
ings of certain lizards. Proc. Linn. Soc. Lond.,
1939: 190-191.

SmyTH, M. 1972. The genus Morethia (Lacertilia,
Scincidae) in South Australia. Rec. S. Aust.
Mus., 16(12): 1-14.

Storr, G. M. 1971. The genus Lerista (Lacertilia,
Scincidae) in Western Australia. J. Proc. Roy.
Soc. West. Aust., 54(3): 59-75. Bera

____. 1972. The genus Morethia (Lacertilia, Scin-
cidae) in Western Australia. J. Proc. Roy. Soc.
West. Aust., 55(3): 73-79.

1974. The genus Notoscincus (Lacertilia,
- Scincidae) in Western Australia and Northem
Territory. Rec. West. Aust. Mus., 3(2): 111-114.

___.. 1975. The genus Proablepharus (Scincidae,
Lacertilia) in Wester Australia. Rec. West.
Aust. Mus., 3(4): 335-338.

REPTILIAN SPECTACLE - Greer DOI

__. 1976a. The genus Cryptoblepharus (Lacer-
tilia, Scincidae) in Wester Australia. Rec.
West. Aust. Mus., 4(1): 53-63.

___.. 1976b. The genus Menetia (Lacertilia, Scin-

cidae) in Western Australia. Rec. West. Aust.

Mus., 4(2): 189-200.

1976c. Revisionary notes on the Lerista

(Lacertilia, Scincidae) of Wester Australia.

Rec. West. Aust. Mus., 4(3): 241-256.

___. 1978. Taxonomic notes on the reptiles of the
Shark Bay region, Western Australia. Rec.
West. Aust. Mus., 6(3): 303-318.

WALLS, G. L. 1934. The significance of the reptilian
“spectacle.” Amer. J. Ophthal., 17: 1045-1047.

____. 1942. The Vertebrate Eye and Its Adaptive
Radiation. Bloomfield Hills, Michigan; Cran-
brook Inst. of Science, 785 pp.

WEBSTER, T. P. 1969. Aspects of the morphological
and ecological variation in the cyanura group
of the lizard genus Emoia (Sauria: Scincidae)
in the Solomon Islands. Unpubl. B.A. Honors
Thesis, Harvard University, 159 pp.

WILLIAMS, E. E., AND M. K. HECHT. 1955. “Sun-
glasses’ in two anoline lizards from Cuba.
Science, 122: 691-692.

Adaptation and Function of Cranial Kinesis
In Reptiles: A Time-Motion Analysis of
Feeding in Alligator Lizards

T. H. FRAZZETTA!

ABSTRACT. Cranial kinesis is widespread among a
great many vertebrates, and should never be neg-
lected in any study of the morphological changes in
vertebrate skulls. Its adaptive role is not related to
an increase in gape. Lizards possess amphikinetic
(two kinetic joints) skulls, and thus differ from other
kinetic forms. Feeding movements of a number of
alligator lizards (Gerrhonotus multicarinatus) were
studied by means of high speed cinematography.
The attack upon prey may be divided into three
sequential phases. Almost always the mandibular
tip is accurately positioned very near the prey,
while the upper jaw is less precisely placed at vary-
ing distances to it. Once the mandibles are located
properly, relative to the prey, the upper jaw closes
down to engage the quarry from above, very shortly
after the mandibles engage it. Limitations in the
ability to precisely position both upper and lower
jaws at the same time enable kinesis to have an
adaptive role. In Gerrhonotus, probably in many
other lizards, and possibly in several other tetrapod
groups, cranial kinesis may be important in allow-
ing upper jaw engagement with prey without the
need for an extremely large change in momentum of
the attacker's head and forebody.

INTRODUCTION

KINETIC SKULLS

“Cranial kinesis” is a term introduced
by Versluys in 1910 to designate the con-
dition of many vertebrate skulls that
possess a functional, transverse hinge
across the skull roof. The effect of this,
generally, is to permit the upper jaw

!Department of Ecology, Ethology and Evolu-
tion, University of Illinois, Urbana, Illinois 61801,
U.S.A.

elements to move upward and downward
relative to the braincase.

The placement of the transverse hinge
varies among vertebrate groups. In many
early tetrapods the location was pos-
terior, where the upper, rear corners of
the dermatocranium movably join the
endocranium. When the kinetic axis is
thusly placed, the condition is termed
“metakinetic.” In lizards especially,
there exists a second kinetic axis across
the middle of the skull—just above the
posterior margin of the orbit—occupying
a ‘“mesokinetic” position. Apparently
most lizards have both kinetic axes, and
can be described as “amphikinetic.”
These three kinetic designations were
provided by Versluys in his early papers
(1910, 1912).

Some vertebrates, such as snakes and
birds, possess a single kinetic joint
located dorsally just anterior to the orbit.
This condition is “prokinesis.”

Kinetic skulls ordinarily depend upon
a sliding joint between braincase and
palate. Very often we can see, in a wide
variety of kinetic skulls, that this joint is a
diarthrosis connecting the pterygoids and
a pair of prominently developed basi-
pterygoid (basitrabecular) processes of
the braincase. A notable exception to this
particular structural configuration is
found in birds where, instead of an in-
volvement with knoblike basipterygoid
processes, the pterygoid and palatine
bones are movably coupled and, to-
gether, form a slider which can traverse a

KINESIS IN ALLIGATOR LIZARDS : Frazzetta

longitudinal, bony track on the brain-
case’s ventral surface.

Not all moveable upper jaws are prop-
erly termed “kinetic.” The complex,
protrusible upper jaws of teleosts, de-
scribed by many workers (e.g., Schaeffer
and Rosen, 1961; Alexander, 1967; Liem,
1970) do not involve a transverse hinge
(nor a functionally complementing brain-
case-palate modification). Some ganoid
fishes (e.g., Lepisosteus, Arnold Brody,
personal communication) can produce
displacements of upper jaw elements
which, again, are not accurately de-
scribed as kinetic movements. Sharks can
shift their jaws forward and downward
relative to the braincase.

The earliest jawed vertebrates, the
acanthodians and the placoderms, seem
not to have had kinetic skulls, although
the arthrodire placoderms were specially
modified to lift the entire head (hence tilt
the upper jaws) relative to the shoulder
girdle (Heintz, 1932).

Moveable upper jaws, especially
kinetic ones, were hardly exceptional
throughout vertebrate evolution. From
our mammalian vantage point, it perhaps
seems “proper that a skull should be
constructed with a rigid association be-
tween neurocranium and upper jaws,
allowing mandibular displacements
alone to perform jaw movements. It is,
however, the late therapsids and mam-
mals that were, and are, largely excep-
tional.

There is an extensive literature on the
presence of kinesis in many living and
fossil vertebrates. A good summary is
provided by lIordansky (197la) in
Russian. Less synoptic references docu-
menting kinesis—but in English—in-
clude Moy-Thomas and Miles (1971) and
Thomson (1966) for crossopterygian and
paleoniscoid fishes; Jarvik (1954) for the
structure of the skull in Ichthyostegalia;
Panchen (1964), Pfannenstiel (1932), and
Romer and Witter (1942) for various
labyrinthodonts; Romer (1956) for the
anatomy of reptile skulls; and Bock
(1964) for birds.

A kinetic skull seems to have been a

. interestingly

223

widespread, primitive feature of tetra-
pods that was lost in a number of evolu-
tionary lineages, while retained and
often embellished in others. Thus many
of the main, evolutionary avenues, lead-
ing to major groups, are sequentia of
kinetic forms. A number of comparative
anatomists still write about kinesis as
though it were a fascinating morpholog-
ical quirk which appeared here and there
in vertebrate history. On the contrary,
kinetic capabilities were of great impor-
tance. The compelling adaptive advan-
tage they conferred often must have
proscribed new cranial modifications that
would have disrupted the integration of
anatomical pieces that comprise kinetic
machinery. Hence, investigations on the
changes in cranial form are incomplete
when the matter of kinesis is neglected.

ADAPTATION IN THE KINESIS OF LIZARDS

Because the present paper deals with
adaptation, I should state that I do not
believe that all evolutionary change is
directly adaptive, nor that diverse animal
groups are usually very different in
fundamental, epigenetic ways, nor that
the evolution of the relatively few funda-
mental differences which have appeared
have come about necessarily in accord
with the common statistical models. But I
reaffirm that selective advantage, repre-
sented tangibly as adaptive form, is the
driving and preserving agent in evolu-
tion, even though much inadaptive bag-
gage may be carried along in a successful
evolutionary line, and even though
superior adaptations may compare tem-
porarily to sloppily engineered devices
(see Frazzetta, 1975).

The kinetic mechanism of lizards is
complicated, involving
several bony links, check ligaments, and
muscles. It is more intricate, apparently,
than the simpler, solely metakinetic
skulls of the great majority of predatory,
kinetic tetrapods.

Today, the living tetrapod whose skull
most closely represents this primitive,
metakinetic condition may not, ironical-

224

ly, be kinetic at all. Sphenodon puncta-
tus, the sole surviving rhynchocephalian
possesses a pair of well-developed ptery-
goid-basipterygoid joints, a firm sugges-
tion of a metakinetic hinge structure, and
the levator pterygoid and_ protractor
pterygoid muscles (Fig. 1). Functionally,
these muscles can serve no other purpose
than to activate the kinetic apparatus. But
they are sparely represented in Spheno-
don, consisting only of a few muscle
fibers.

Versluys (1910) described this condi-
tion, and it was noted again by Poglayen-
Neuwall (1953) and by Ostrom (1962).
I have seen it in an adult specimen
which I dissected. The adult skull
is, however, functionally akinetic be-
cause bones are shaped in a manner to
crowd into spaces necessary for kinetic
movements. Both Versluys and Ostrom
surmise that juvenile specimens are
kinetic, but lose the capacity (possibly
due to allometric bone growth) toward
adulthood. Assuredly this is conjecture.
In any case, the general cranial form of
Sphenodon can represent a good model
of the primitive, metakinetic configura-
tion.

Figure 2 shows a primitive type based
roughly on Sphenodon. Figure 2A repre-
sents the skull “at rest,” while 2B depicts
the major part of the dermal skull includ-
ing the upper jaws rotated upward
(protracted) about the metakinetic hinge.
The drawing shows an exaggerated dis-
placement to clarify the action. During
protraction the pterygoids slide on the
basipterygoids in an arc. The arc’s radius
is the distance, projected upon any para-
sagittal plane, from the metakinetic axis
to the pterygoid-basipterygoid joint.

Protraction occurs by the contraction of
the levator pterygoid and _ protractor
pterygoid muscles, which draw the
pterygoid bones forward and upward
relative to the braincase. Because the
pterygoids are firmly associated with the
dermal palate and ectocranium, the up-
per jaws are lifted.

Retraction, the pulling back of the

Advances in Herpetology and Evolutionary Biology

dermal cranium relative to the neuro-
cranium, restores the rest position. It is
accomplished by any or all jaw closing
muscles that originate on the dermal
skull, and by extrinsic head muscles (not
represented in Fig. 1) that run from the
posterior mandibular surfaces to the neck
region.

In lizards there are usually two kinetic
joints, an amphikenetic condition that is
presumably derived from a purely meta-
kinetic skull (although Robinson [1967]
makes the surprising statement that
lizards arose from akinetic reptiles, and
that the metakinetic hinge was a saurian
invention). One of the two joints occu-
pies the primitive, metakinetic location.
The other is mesokinetic. Figure 3 shows
a Gerrhonotus multicarinatus skull, with
musculature represented. Here, again,
there is a pterygoid-basipterygoid joint,
but its conformation is specialized. The
double-jointed kinesis demands that the
pterygoids slip longitudinally, not along
an arc’s curve. We see in amphikinetic
lizard skulls a basic four-bar linkage, one
corner of which is “pinned” to the meta-
kinetic axis (Frazzetta, 1962). Figure 4
shows a Gerrhonotus skull in outline,
with the cranial elements reduced to
mechanical links.

The anterior link, the muzzle unit, is
formed of a rigid, triangular frame, the
posterior strut of which is a typical bar in
the four-bar mechanism. The braincase,
shown as a rigid triradiate structure, joins
the four-bar frame by a rotational joint
whose axis is collinear with the meta-
kinetic hinge. Elsewhere, the four-bar
linkage is tied to the braincase by strong,
elastic ligaments. The epipterygoid does
not change the kinematic movement
patterns of the linkage, but acts as a
limiter and stabilizer. I have discussed its
role, briefly, in 1962, and shall provide a
future paper on its function.

In amphikinetic lizards, protraction in-
volves the slight depression of the dermal
cranium about the metakinetic hinge (the
opposite of what occurs in more primitive
types), and an elevation of the anterior

KINESIS IN ALLIGATOR LIZARDS - Frazzetta 22.5

p fa Ir
Me Affi tere MUN Mites on

uscres

1- depressor mandibulae
\1 2- pterygoideus

35 pseudotemporalis

4- adductor externus group
5- w uw a
6- adductor posterior
7-protractor pterygoid
Sr ze 8-levator pterygoid

Figure 1. Skull of Sohenodon punctatus. Dorsal, ventral, and lateral views A, B, C.D, lateral view with temporal
arcades removed to show placement of intrinsic jaw muscles.

Abbreviations: ar, articular; bo, basioccipital; bp, basipterygoid process; bs, basisphenoid; co, coronoid; de,
dentary; ec, ectopterygoid; eo, exoccipital; ep, epipterygoid; fr, frontal; j, jugal; |, lacrimal; max, maxilla; na, nasal;
pa, parietal; pal, palatine; pf, prefrontal; pm, premaxilla; po, postorbital; pof, postfrontal; pop, paroccipital process;
pr, prootic; pt, pterygoid; g, quadrate; s, stapes; sm, septomaxilla; soc, supraoccipital; sq, squamosal; st, supra-
temporal; sur, surangular; vo, vomer.

226

Advances in Herpetology and Evolutionary Biology

Figure 2. Diagram of kinetic movements in an hypothetical metakinetic (monokinetic) reptile. A. At rest or re-
tracted. B. Protracted with mandibular depression. Shaded area represents endocranium lying within ectocranium
(unshaded). Metakinetic pivot indicated by circled cross.

dermal skull (the muzzle unit) by upward
rotation about the mesokinetic axis. The
overall effect is nonetheless, a raising of
the upper jaw relative to the neuro-
cranium (Fig. 5).

In 1966 Iordansky published an excel-
lent study on the kinetics of lizard skulls.
He argued convincingly that amphi-
kinesis, because it permitted a reverse
angulation between muzzle and basal
units during powerful retraction, could
place forces on the prey between upper
and lower jaws whose opposing vectors
were more nearly collinear. In this man-
ner, fewer unbalanced forces would
exist, and the lizard would tend not to
lose its meal by jaw forces that wedged
the prey forward, inadvertently driving it
from the mouth.

Although Iordansky accepts the basic
four-bar linkage plan for most lizards, he
raises a point concerning Varanus. He is
accurate when he indicates that the lower
strut of the triangular muzzle unit is, it-
self, interrupted by at least one, rota-
tional turning-pair joint. However, this
joint resists flexing by strong connective
tissue attachments. As the skull is pro-
tracted, the joint allows very little
angular displacement. When forcibly
protracted, near the anterior limit of
kinetic displacement, the intramuzzle

joint will flex a little. The effect is a pro-
tracting movement, based mainly on the
initial four-bar linkage, which if forced
further, will be added to slightly by intra-
muzzle movement. This small amount of
additional movement does not alter sig-
nificantly the fundamental, four-bar
kinematics in such forms as Varanus.
Nevertheless, the possibility of an adap-
tive role should not be dismissed.

I do, however, disagree with Iordansky
in his claim (1966, 1970) that the ptery-
goideus muscle cannot act as a kinetic
retractor. So long as the lower jaw cannot
transmit a potentially nullifying force on
the upper jaw, either directly or through
prey held in the mouth, the tension
produced by an activated pterygoideus
muscle will result in an unbalanced force
on the muzzle unit. It tends to pull the
muzzle backward and downward for
precisely similar mechanical reasons that
it tends to pull the mandible forward and
upward.

This functional point, concermming the
pterygoideus muscle, leads to a compli-
cating issue unknown to either Iordansky
or myself in the 1960’s or early 1970's.
Throckmorton’s work (1978) on an
herbivorous lizard, and Smith’s (1980) on
a carnivorous species show that a portion
of the ptergoideus actually aids in mandi-

KINESIS IN ALLIGATOR LIZARDS - Frazzetta DONT

Muscles

1 depressor mandibulae
Z tery ordeus

3 eaeneals

4 adductor externus group
5 protractor pterygord
6 levator pterygoid

Figure 3. Skull of Gerrhonotus multicarinatus. (Modified from McDowell and Bogert, 1954, and from several skulls
at hand). Conventions as in Figure 1.

Figure 4. Linkage diagram of skull in Gerrhonotus. Braincase is represented as triradiate construction. Units of the
four-bar linkage are shown, as is the epipterygoid bone.

Figure 5. Diagram of Gerrhonotus skull showing A, skull at rest (portion of mandible broken out in A to expose
basal unit); B, skull protracted with mandibles depressed; C, skull fully retracted with jaws clamping prey.

Compare Figure 2.

bular depression; another portion, a
major one, apparently works as a mandi-
bular elevator.

I concluded in 1962 that kinesis could
be explained by its provision in allowing
the upper jaws to close downward as the
mandibles move upward, to insure a
nearly simultaneous contact with the
prey. This would avoid premature colli-
sion by the lower jaw alone, before the
prey was secured, and prevent the quarry
from being deflected out of seizing range
by the very apparatus designed to
capture it.

Iordansky (1966) took issue with this
by raising the reasonable and challeng-
ing question: why is a moveable upper

jaw necessary when, in fact, movement of
the entire head will position the upper
jaw relative to prey during its capture?
Thus, Iordansky rejects the proposal that
kinesis is a device specifically evolved
because of an advantage in moving the
upper jaws downward to match the up-
ward movement of the mandible when
prey is seized.

The challenge is meaningful because,
obviously, whole head movements can,
indeed, position the upper jaws. Yet, I
note the intuition of an earlier student of
cranial kinesis (in birds), William
Beecher who, in 1951, made the observa-
tion that a skull without a kinetic upper
jaw would compare with a hand without

KINESIS IN ALLIGATOR LIZARDS - Frazzetta

moveable fingers, but having a moveable
thumb. The present paper concentrates
on this particular question, by means of a
detailed time and motion analysis of
lizard feeding.

Since Iordansky’s 1966 paper, a num-
ber of papers relating to saurian kinesis
have appeared. These include Callison
(1967), Carroll (1977), Gomes (1974),
Gomes and Gasc (1973), Gow (1975),
Haas (1973), Iordansky (1970, 1971a,
1971b), MacLean (1974), Rieppel (1978,
1980), Robinson (1967, 1973), Russell
(1964), Smith (1980), and Throckmorton
(1976, 1978, 1980). Many of these attempt
an adaptive explanation for kinesis.

Additionally, in the past decade and
one half, a number of general works on
vertebrates allude to cranial kinesis in
lizards or in other tetrapods. Often these
allusions include a brief, adaptive com-
ment. Although it may be difficult to be
very confident about what the adaptive
significance of kinesis is, in some cases a
stronger confidence can be experienced
in determining what the kinetic role is
not.

Authors will state that cranial kinesis
increases the gape, a view rejected by
myself (1962), by Iordansky (1966), and
by Rieppel (1978). It should be obvious
why, in a purely metakinetic skull, there
can be no effect on the gape whatsoever.
In amphikinetic lizards, mandibular de-
pression through rotation about the
quadratomandibular joint can proceed
only so far until the tympanum is so tight-
ly sandwiched between the quadrate and
the mandible’s retroarticular process that
the middle ear is at risk. Several liga-
ments in the lizard skull limit depression
of the mandibles relative to the basal and
muzzle units; hence, their affect is to
allow the quadrate-retroarticular angle to
diminish only so far. Because of that, the
quadrate and mandible can be modelled
as a single rigid unit solidly joined at this
_ limiting angle. As protraction takes place,
the muzzle is lifted, but so also is the
mandible. Measurements reveal that the
gape is actually reduced in protraction.

229

Perhaps it can be seen intuitively that in
protraction of the four-bar mechanism,
the shorter quadrate will rotate forward at
a faster rate than will the longer, orbital
strut of the muzzle unit (see Frazzetta,
1962).

There may be some kinetic skulls, in
some vertebrates, that are constructed to
increase gape. Many bird skulls may be
so constructed (Bock, 1964). However,
where gape increase can be shown to
exist, caution should be respected in the
direction of the simplistic conclusion that
gape enlargement is, itself, the adaptive
role. Some explanations based on gape
can be shown to lead to the unconvincing
conclusion that an animal is adapted to
seize prey larger than it can swallow.

Suggestions that kinetic jaw structure
provides for shock absorption are probab-
ly erroneous in all published cases (see
Frazzetta, 1962). A future paper will
examine the issue in further detail.

Several papers (e.g., Carroll, 1977;
Robinson, 1967, 1973) have suggested
that kinesis is explainable as a
mechanism for swallowing. They pro-
pose that kinetic jaw movements move
captured prey back into the mouth by
alternating protraction and retraction. It
is hard to visualize this for the purely
metakinetic skulls of lizard ancestors.
Amphikinetic lizard skulls have the
mechanical capacity to draw the upper
jaw’s tip slightly backward relative to the
mandible. Observations on swallowing
in several lizards, however, show that
prey is moved backward into the gullet
by means of the tongue (Frazzetta, 1962),
and/or by inertial feeding (Gans, 1961,
1969). The enhancement of swallowing
ability provided by amphikinesis, either
in tongue or inertial methods, seems
negligible.

MATERIALS AND METHODS

Approximately two dozen Gerrhonotus
multicarinatus were studied by means of
high speed cinematography. Several

technological excellence of the Vanguard

230 Advances in Herpetology and Evolutionary Biology
other lizard species of the genera
Lacerta, Eumeces, Sceloporus, and

Gekko were filmed as a means to com-
pare similarities and differences between
Gerrhonotus and several other forms.

A Hycam, high speed _ camera,
equipped with its standard electronic
speed control gave excellent service. It
was used in conjunction with a Milimite
timing light generator, and an Angenieux
12-120 mm zoom lens with a viewing
tube. Occasionally a +1 or +2 diopter
close-up lens was applied to the zoom
lens.

Illumination was provided by four
Colortan, spot-flood movie lamps to
which were affixed heat absorbing glass
filters to protect the animals. Light levels
were measured with a Honeywell spot-
meter.

Approximately twenty, one-hundred
foot rolls of 16 mm Kodak film, of the
types Ektachrome B, Tri-X reversal, and
Four-X reversal, were exposed at speeds
ranging from 250 to 500 fps. Of this, about
18 separate feeding acts were useful. Of
the 18, only 8 were analyzable in fine
deatil.

All of the useful film was studied by
means of a stop-motion, variable-speed,
remote-controlled LW projector. Sequen-
tial tracings were made of the filmed
feedings from projections reflected from
a front-surfaced mirror angled at 45° to a
table top. Overlays of consecutive trac-
ings permitted determination of skull
movements during the lizards’ acquisi-
tion of prey. Eight of the filmed se-
quences were appropriate for more
detailed study by means of a Vanguard
Analyzer. This machine permits analysis
of point displacements, from one frame to
the next, relative to two coordinate axes
in measurements of inches x 107%. It
also allows measurement of angular dis-
placements between consecutive frames.

Continuously locatable points were
selected on the lizard heads to permit
point displacement analysis, or to define
cranial axes whose angular changes could
be measured. Despite the considerable

Analyzer, its actual use in this study
proved difficult. Slight head twistings by
the lizards as they lunged, an occasional
blurred frame, an unexpected, momen-
tary highlight which masks the sought-
after point, were among a number of un-
welcome factors. Hence, each point and

each angle was measured ten times for all

eighteen filmed sequences. In those
cases where there was a lack of strong
consistency in a set of ten readings, the
sequence was rejected from this aspect of
the study.

The prey offered was either a meal-
worm or a cricket, restrained by a fine,
cotton thread which was snipped after
the capture. Occasionally prey would be

suspended above the substrate, to simu- |

late a slow-flying quarry (which I have
seen Gerrhonotus capture), or instances
where the prey is on a leaf, twig or grass
stem above the ground.

Dissections were performed on three
preserved Sphenodon punctatus, and
fresh Gerrhonotus which died during
captivity.

THE PATTERN OF ATTACK
ON THE PREY

There are three discernable phases to
the attack on prey by a lizard. The first is
the rush, or the chase, where the attacker
dashes up to the prey—or pursues it—to
within lunging distance.

The second phase is the lunge, or what
I prefer to call the delivery because the
capturing apparatus, the jaws, are deliv-
ered to within seizing distance. Then the
seizure or grasp is the third and final
phase in the capture.

The second and third phases were

studied, and are represented in the trac-

ings and the Vanguard analysis. Figure 6
shows a typical delivery and grasp pat-
tern when prey is on a flat substrate
shared by the lizard.

With elevated forebody, the neck and
head arch above the ground a short dis-

KINESIS IN ALLIGATOR LIZARDS -

Frazzetta 231

Figure 6. Sequence of tracings of Gerrhonotus specimen Ill, film 2, from projections. “T” designations show

elapsed time in seconds from the first frame shown.

tance from the prey toward the lizard.
The head is turned downward, its axis
roughly in line with the prey.

The delivery phase begins with the
mouth closed, and occurs by coordinated
movements of the front limbs, forebody,
neck and cervicocranial joint. As the
delivery progresses, the mandibles are
gradually depressed so that the degree of
mouth opening increases more or less
steadily as the jaws approach their target.

Soon after mandibular depression
begins, kinetic protraction becomes pro-
gressively conspicuous. The muzzle is
seen to lift relative to the more posterior

regions of the head, and the leading,
vertical edge of the eardrum—which cor-
responds to the longitudinal axis of the
quadrate, to which it is fastened—be-
comes gradually angled forward from a
point just above it.

The tip of the lower jaw is almost
always placed in light contact with the
ground, as close to the prey as possible
without actually touching it. The preci-
sion of this placement is impressive.

At the time the lower jaw tip is placed,
the upper jaw is clear of prey contact by a
substantial margin of distance, and the
kinetic mechanism is usually protracted

232

to its maximum degree. Often, imme-
diately following this mandibular place-
ment on the substrate, the entire head is
turned further downward to close some
of the distance between the upper jaw
and prey surface. Almost always, as soon
as mandibular contact with the ground is
established, the head undergoes rapid,
kinetic retraction.

At this moment, the muzzle begins to
descend, relative to the more posterior
region of the head, and the tympanum’s
leading edge starts its backward swing.
Frequently the eye is seen to sink deep
into the orbit, presumably a result of con-
traction of the levator bulbi, a weak
kinetic retractor (Frazzetta, 1962) which
in Gerrhonotus inserts on the pyriform
membrane of the palate and on tissues
surrounding the eye (Gomes, 1974).

During these movements of the upper
jaw, the mandibular tip moves toward the
prey along the substrate. This movement
is caused only partially by rotation of the
mandible about the quadratomandibular
joint. It is also effected by pushing the
entire head forward as it tilts downward
while the neck arches higher upward.
The lizard now has positioned itself
further over the prey. The lower jaw
touches the prey first. The upper jaw,
which has been descending toward the
prey, may quicken its movement after the
lower jaw has made contact with the
prey. The upper jaw engages the prey a
tiny fraction of a second after the mandi-
bles’ contact. Significantly, the prey itself
is hardly moved from its position by the
seizing action. At the grasp, or just after
it, retraction is usually strong enough to
carry the muzzle downward past its rest
position, to thus hold the prey as
Iordansky has postulated (1966).

When prey is taken above the sub-
strate’s surface, the success probability of
capture appears reduced. Sometimes,
this was caused by artificial circum-
stances, such as the suspending thread
being touched by the lizard’s jaws which
swung the prey out of seizing range. Per-
haps, in other instances, there might have

Advances in Herpetology and Evolutionary Biology

been greater movement of the prey when

suspended, despite our efforts to hold the |

thread still; the lizards’ aim might thusly
have been less accurate. But beyond
these factors, the impression is that the
required upward angle of thrust toward
the prey, and the prey’s position closer to
the upper than to the lower jaw, were
awkward circumstances that the captor
had to overcome.

Figure 7 shows a successful, above-
ground capture. In Figure 8 the upper
jaw of the lizard bumps the suspending
thread and deflects the prey. Because
above-ground attacks have a greater
variety of factors that the lizard must deal
with, they tend themselves to be more
diverse, and difficult to typify according
to a single, general pattern.

The delivery phase begins with mouth
closed and, as before, mandibular de-
pression grows progressively as the prey
is neared. Soon the kinetic apparatus
begins a steady protraction, reaching a
maximum shortly before the jaws are to
close on the prey. In both Figure 7 and
Figure 8, the tips of the mandibles
approach closer to the target than does
the upper jaw. This is the usual case in
above-ground attacks, as it is when the
prey is located on the substrate, as in
Figure 6.

When the jaws close, the head may
turn downward on the neck (Fig. 8D-E)
or it may not (Fig. 7). Jaw closure is
produced by elevating the mandible
while, at this time, the skull undergoes
rapid kinetic retraction. Retraction occurs
in Figure 7, but it is but slightly notice-

able there. It is conspicuous in Figure 8, |

and begins (Fig. 8D-E) as soon as the
thread is touched. In this sequence the
lizard longitudinally twists the head,
possibly in a vain attempt to avoid the

thread.

VANGUARD ANALYSIS

Capturing prey on the ground is differ-
ent enough from taking it in the air that

KINESIS IN ALLIGATOR LIZARDS - Frazzetta

233

T=2 4/283

C

(nerves

~ —_-

Figure 7. Sequence of tracings of Gerrhonotus specimen Il, film 1, from projections. “T’ designations show

elapsed time in seconds from the first frame shown.

convenience enlarges if the two types of
attack are regarded separately. The same
points and axes, however, are studied for
both capture modes.

The points and axes are shown in
Figure 9. The purpose in choosing such
points as the upper and lower jaw tips (uj
and lj) is clear. Ear top (et) is a point that
corresponds closely to the joint connect-
ing quadrate, parietal unit, and braincase,
and lies close to the cervicocranial articu-
lation. It is a useful point to keep track of.

Figure 10 shows the Vanguard analysis
of the same sequence represented in
Figure 7. In Figure 10, however, data
points are shown for every frame, from

the moment the mouth begins to open,
early in the delivery phase, to the final
seizing of the prey. On any of the twelve
curves, each dot is the relevant data point
for that frame. Hence, the abscissa is time
divided into increments of the reciprocal
of camera frame rate, in this case 1/287.0
of a second, progressing from left to right.

The ordinate units show displacement

differences. For data points whose dis-

placements are measured as _ linear
changes projected upon a set of vertical
and horizontal axes on the Vanguard
machine (not on the axes of Figure 10),
the horizontal (X) and vertical (Y) dis-
placements are each graphed separately

234 Advances in Herpetology

at

and Evolutionary Biology

Figure 8. Sequence of tracings of Gerrhonotus specimen |, film E, from projections. “T” designations show

elapsed time in seconds from first frame shown.

here, against the time-related abscissa. A
scale of linear displacements is shown in
the figure.

Point data (upper jaw point x and y, ujx,
ujy; lower jaw point x and y, ljx, ljy; ear
top point x and y, etx, ety) are each ar-
ranged so that an increase in either
horizontal or vertical displacement is
indicated by an ascension of sequential
data points relative to the ordinate axis of
the figure.

Angular data is shown for parietal top
axis, pt, and is graphed so that a tilting
upward of the head is indicated by a rise
in the points shown. Angles between

pairs of several axes (snout top axis with
parietal top axis, snf-pt; muzzle unit
upper axis with cranial axis, mzu-cr;
muzzle unit lower axis with cheek axis,
mzl-ch; ear axis with cranial, @-cr) reveal
kinetic flexing of the skull. The graphed
points rise when protracting movements
are indicated, and fall in retraction.
The mandibular axis’ angle with the
ear axis, m-€, shows the turning of the
mandible upward or downward in its
rotation about its joint with the quadrate.
In several of the eight Vanguard
analyses (Figs. 10-13) there are occa-
sional inconsistencies. For example, the

KINESIS IN ALLIGATOR LIZARDS - Frazzetta

upper jaw_,\ eK
point

lower jaw
point

pariet al XO

235

pone

Cranial axis (e Gr top point

Figure 9. Head of Gerrhonotus showing points and axes used in Vanguard analysis.

mzl-ch angle sometimes corresponds
poorly with other kinetic indicators.
The sequences all start at time-zero
when the mouth first starts to open.
Hence, the delivery phase has already
begun shortly before the graphs begin.

PREY CAPTURED ON THE SUBSTRATE

Typically the ear top point shows a
rather smooth, continuous path in both x
and y displacements. This is seen in
Figure 10. The jaw tips, too, run a fairly
continuous course up to the time when
the mandibular tip is placed upon the
substrate. From that time on, there is
little or no vertical change in the lower
jaw tip (ljy in Figs. 10, 11A,B,C) because
it slides—or nearly does—along the sub-
strate toward the prey. In the cases
shown in Figures 10 and 11B, the hori-
zontal motion of the lower jaw tip (ljx) is
erratic as the moment of prey capture is
approached. In two other cases of capture
on the substrate (Fig. 11A,C) the lower
jaw tip moves in a smoother manner.
During this period, the upper jaw move-
ments—both horizontal and vertical (ujx

and y)—are reasonably smooth in three of
the four substrate captures; the exception
is the sequence shown in Figure 10.

Tilting of the head, shown by angular
changes in the parietal top axis (pf), is
different in each of the four substrate
captures. In Figure 10 the head tilts
sharply downward just before the
mandibular tip touches the substrate, and
then rotates further downward at varying
rates thereafter. In Figure 11A the down-
ward tilt begins after the mandibles come
into contact with the prey and, as before,
proceeds at a discontinuous rate. How-
ever, in the captures shown in Figure
11B and C, the head begins to turn up-
ward from the time of mandible-substrate
contact, and then turns downward later.

Lower jaw depression, relative to the
quadrate (m-é), is mostly a smooth, con-
tinuous motion except in Figure 11B. It
reaches its greatest relative depression
just before mandibular contact with the
substrate (except in Figure 11C). Eleva-
tion may be smooth or not.

The four kinetic angle differences are
the most difficult to measure on the
Vanguard Analyzer. Consideration of the

2:36

m2 cae eae erx
SO ne

GC ean o Sees

72870 |*_-~ ety

26mm_.~

a Wei.

oe i

ee iy ujx
ee po.

Figure 10. Vanguard analysis of Gerrhonotus speci-
men Ill, film 2. In this figure, and in Figures 11, 12 and
13, Roman numeral gives specimen number, Arabic
number or letter gives film number, fraction (here, 1/
287.0) is time interval between successive frames
(hence, dots), and mm measurement is image size of
head on analyzer screen, measured from eartop point
to upper jaw tip point. Vertical line with asterisk is
scale: 2.54 mm (image size) for linear data, 2.5° for
angular data.

Abbreviations in graph title: AGC, above ground cap-
ture; AGC*, above ground capture but prey very nearly
missed; AGM, above ground attempt but prey actually
missed; GC, capture on ground.

Abbreviations at graph’s bottom: \j-g, lower jaw
touches ground; |j-p, lower jaw touches prey; uj-p, up-
per jaw touches prey; es, eye sinks into orbit; ps, prey
securely seized by both upper and lower jaws. Other
abbreviations and explanations in text.

Advances in Herpetology and Evolutionary Biology

sequences illustrating substrate-based
capture, indicates that kinetic protraction
reaches its maximum at, or shortly before,
the moment of mandibular contact with
the ground. Most times (Figs. 10, 11A,B)
protraction seems to sharpen suddenly
just before retraction. This is corrob-
orated in the several traced sequences
unsuitable for Vanguard analysis (see
explanation above).

Prey seizure occurs after a retraction of
the kinetic apparatus. As already men-
tioned, the muzzle-cheek angle (mzl-ch)
corresponds poorly with other kinetic
indicators (expecially in Figure 11A). I
cannot offer an explanation for this.
There are at least two _ possibilities.
Accurate measurement of the angle may
be more difficult than that in the other
kinetic determinations. Or, small adjust-
ments of elements at the joints during
their rotation may show up conspic-
uously in this measurement. If so, the
effect could frustrate application of a
model which is simplified to regard the
moveable articulations between cranial
units as mechanical, turning-pair, planar
joints.

PREY CAPTURED ABOVE THE SUBSTRATE

It is a little harder to find common
denominators in head and jaw movement
patterns when prey is taken above the
substrate, than when taken upon the
substrate. However, in broader aspects,
some rough paradigms do exist.

As before, the ear top point (et) moves
continuously without noticeably sharp
breaks in rate or direction. The tips of
both upper and lower jaws move only
slightly less smoothly than the ear top
(Figs. 12, 13). The parietal top axis (pt),
which reveals the tilting of the head, in
all four cases tilts upward at first, and
then turns downward. The _ reversal
occurs just before, or when, the lower jaw
touches the prey. When prey is taken on
the substrate, the mandibles typically are
placed close to the prey, and then touch it
a tenth of a second or less before the

237

KINESIS IN ALLIGATOR LIZARDS : Frazzetta

‘uolyeue|dxe 10) 0} einBi4 88g ~punos6 ay) UO Seunydeo Bulnjoaul seouenbes wh; sesy) Jo sasAjeue pyenBue, *|| asnbi4

>
| rig = po\
ay fee Doe ee ee eN go
oe} SS ODED © ba 0 z : oN ee
™

: S59 = e~ v= f= 2 aU Os, em . 5 wee
OH Os 00 ict ce ae j2-9 *~ <a
42-49%, : a \ pf 8 \ a or “4
—- — : 8 8 BX 5 8 a ° f *
See 2 e 5 e

bas \ No a -

o—

ao-nzurs/ Bee , cee a8 eee TN Se te ae HERMIES de Naat : Vy), aie

d ; Sp Siete ee See : i Om) Bie : NGS ee Seer,
/ = : NS : \ ee. a = eat : Os At \
: Pe e ee = e —.
ean! . so -nzw™~. ae oS \

CIOS cso os 0sgee Se ee ae ys NGS eee cee cat ae

Seo 4a-4US ™.

= : pee Kin ate :
em , : ~—e—.— 5 ete ete : : ° : 5 aq oat
x/n Sloe : SOS S005 Ga. 1d : pena ioc eee ease eo : even pee
~ s—_ — EF s. v
OSC Go P=O= 9 ae Saag BC Ta e-°
Se a Beg Te OS a et et et 4 ="
I= 050505 Osoom Kit 2 ox, 57 La] F oS
— e— © 6 —e —0 — 8 ~s — _ I, Bet Ss 4 oe eee:
Ary: 1 Oss9. “ce xX Nie e_ es Se NO
: = = : : SH 0S05050502 005 6500 : : ao
i A, _e Semea se aes S' < 7 ee
LPP, SUF Sa a oe 1S Ee) ge
: eerie OO Os 6 : ; ee Sake eS Q : Uf DG
wie me : B72 SS oF we cae
oe eae, go ayers OSG pe a on Sa Wa NY
soe Boe -¢ +| I eee e PSOecegen ce 0 Ae: Va Oe
S00 : Bue A/a AE SOS0=0=0 = MU B2020 = . + “NY
este 3 Weer ww x : Pca X/, Si om Bape a ae
ae os 9b = z| peD Eee ie
: of oe Aja: Soe 77

Cae eo S'S6b/1 ME Bo WIS ° Obed a Gg ‘9'gbe/!
pea ees Se Peo Se
x/) 2 99 Ne 29 a JTBE
SUS Oye 62 MIXX ee eh BS oie 8c IAXX aie: Wesel

sre eealiean Ss a one xa

238

1/283.0 gs 2 eR

—e
e-?

EP AGG

20mm
[eee

e
—e—
=0=0
e—e—e—e-°-* SO

°m-e

ae e-°~e-?
eo—e—

=o
ee —e- ~

s=050

Figure 12. Vanguard analysis of Gerrhonotus speci-
men Il, film 1. See Figure 10 for explanation.

upper jaw makes prey contact. Prey taken
above the substrate is, as in the substrate-
based captures, approached and touched
by the lower jaw before the upper.

The angular change between mandible
and ear (quadrate) axis is much like those
seen in the previous set of examples.
Here, however, in two cases (Fig. 13B, C)
the angle begins to close well before con-
tact is made between mandibles and
prey.

In Figure 12, retraction begins imme-
diately after mandible-prey contact. The
other three cases (Fig. 13) appear to
reveal retraction beginning at various
times before the prey is touched by the
lower jaw.

Advances in Herpetology and Evolutionary Biology

DISCUSSION

The delivery phase involves a smooth,
continuous approach of the neck and
neck-cranial joint. The jaw tips them-
selves are also reasonably flowing in
their motions; their smooth movements
may become broken as the prey is closely
approached. Angular changes in the tilt
of the head vary from case to case: some-
times they are continuous, while at other
times they seem quite erratic.

Apparently the combination of move-
ments involving the neck-cranial joint,
the neck and forebody and front limbs,
the angulation of the head, the depres-
sion of the lower jaw relative to the
quadrate, and kinetic rotations can often
control the jaw tip displacements to
negotiate an even, unbroken motion.
Unquestionably, Iordansky (1966) is cor-
rect in that tiltings of the entire head can
bring the jaws close to coincidence with
the prey. There is, however, more. The
mandibles are always first to come near
the prey, and they are first to make con-
tact with it. Only afterward does the
upper jaw become involved. While the
lower jaw is rapidly being positioned
near the prey, the margin of distance
between prey and upper jaw is usually
moderate to fairly great.

Great precision in the aiming of the
mandibular tip, relative to the preys
location, is seen in most cases, especially
those where the prey is taken on the
substrate. At the moment of this accurate
mandibular placement, the upper jaw—
separated as it is from the prey by a con-
siderable distance—seems, in contrast,
imprecisely or but roughly positioned.
Very soon after the mandibles have been
accurately placed near the target, the
upper jaw moves quickly downward to
cooperate with the mandibles in making
the grasp. This downward motion of the
upper jaw may be effected by a variety of
factors (head tilting, etc.), but always in-
cludes kinetic retraction.

Prey is seized only when the skull has
undergone much retraction, and we thus

etta 239

“7
AA

- Fra

KINESIS IN ALLIGATOR LIZARDS

‘uoleue|dx9 410} 0} ainBi4 8aS ‘punos6 anoge Aaid ainjdeo 0} sydweye Bulajoaul Seouenbes wily 9914) Jo siskjeue psenbue,

Be : A Ya 20/1,
NG : NS Ses ES.2 o— YI-/ZU 4 Oe ta 0-8 e= oR” \ Yi fh :
Deere | ee ea ae gay YO-[ZUh—

Seu ‘

BE Vo Woe: a-u/ wi * Nee —
a” ee — Neo 4I-NZU ar SA : PE \ Se o- Dee

t \ So past — BN Se = 49 -NZU 9 ;

Sie [Vo ENG: *—e_e-*-° pore dd- Jus © a ae iN —— Sas ¥e

re ner say ~ et e-u. pete eSNG

Ue Ss ee ea, :
Ne * a ot ene #8 Kjjaes : Oso o~ ON oe cane
ry SS 07" \ =] \ Y ca
o-W/ \ aU cee xX Sees

es a Se Se eee BY if we

o—'J— o— see ee o—. 2a aa fe e C4 2 ee
ee ee a is
hee eorex Mar he * Kip" J. Va ee
Pee Kya Se wwy2 ‘2°88z/] AI), et

"el ounbiy

Aya oe eo +| ieee Po eee tne xin! / Y. wWge

X49 , WOV Kis a
| ie }

oe wwgz ‘2882/! ae ee #09 ‘aI Ce a bOLI/I
x09V

x/° 2| AX

240

can note on behavioral grounds—as we
had, previously, on mechanical ones—
that the adaptive meaning of kinesis can-
not be the increase of gape for the seizing
of larger prey.

It appears that the lizards in this study
do not act to set both upper and lower
jaws relative to the prey with equal
accuracy. There is a strong, perhaps ir-
resistable temptation to interpret the
films as revealing an inability of the
lizards to carefully guide both upper and
lower jaws simultaneously. Hence, the
lizard must pay attention to the jaw posi-
tion sequentially, “choosing” to regard
the lower jaw first for the best of mech-
anical and practical reasons.

Lizards live in our world of significant,
gravitationally induced forces. Much of
their food will lie wpon a surface, regard-
less how irregular or complex this surface
may be. An obvious disaster in prey cap-
ture would be the ramming of the mandi-
bular tip into the substrate. Potential
injury would, I think, be less important
than the sudden, unplanned-for jolt that
would be suffered by the head and jaws,
and the subsequent disruption of the
complexly controlled delivery phase of
prey capture. Equally unwelcome would
be the effect of misplacing the mandi-
bular tip behind a tiny hillock of earth, a
pebble, a twig, anything that would form
a barrier between a rapid, simple move-
ment of the lower jaw to the prey.

Because of these considerations, if the
lizard cannot concentrate equally on both
upper and lower jaws at the same time, it
should devote its first attention to the
mandibles. It follows that, because a
lizard so often deals with quarry upon a
substrate, its morphological and_ be-
havioral systems should be adjusted for
the sequence: lower jaw, then upper. As
a result, even prey taken off the ground,
where the sequence is less important,
will be approached by the mandibles
first.

The fact that the lizards seemed not to
aim with both upper and lower jaws
simultaneously has a possible parallel in

Advances in Herpetology and Evolutionary Biology

the grasping operation of a human hand.
Beecher’s (1951) comparison between
the hand and the kinetic bird skull illu-
minates the hand as a natural model of
the amphikinetic lizard skull. In this con-
text the wrist may be seen as representa-
tive of the lizard’s head-neck joint; the
hand, from wrist to the distal ends of
metacarpals II through V, compares with
the posterior skull half; digits II through
V together form the “upper jaw,” with
the “mesokinetic hinge” at the meta-
carpal-phalangeal joints; and the thumb
simulates the mandibles.

Among several important discrepan-
cies between the hand and the amphi-
kinetic lizard skull is that, of course, the
guidance system (eyes, mainly) travels
with the jaws when prey is grasped.
Nonetheless, some of the similiarities
between hand and lizard skull suggests
that the hand is a more accurate mechan-
ical analogy than would be a shop-crafted
structural model.

When grasping small objects on a desk
top, my intuitive impression is that I
exercise little conscious direction to the
details of placing thumb and fingers rela-
tive to the target. But I also notice that
while my thumb tip becomes set very
close to the subject, my fingers are quite
far away, but descend rapidly to com-
plete the grasp.

In an attempt to slightly objectify the
hand model of the kinetic skull, I en-
listed the aid of two laboratory assistants
who were told only to capture a ping
pong ball, thrown bouncingly toward
them along a table top, by thumb below
and fingers above. Those exercises were
performed before the high speed camera,
set at an average speed of 390 fps. Two
one-hundred foot rolls of Tri-X film were
thusly exposed. Not all capture attempts
were successful, and a number of misses
were recorded.

Figure 14 shows a typical (successful)
capture. The hand and fingers are posi-
tioned by movements of forearm and
hand rotations about the wrist. The
“gape” between thumb and fingers in-

KINESIS IN ALLIGATOR LIZARDS -:

creases slightly as the target comes
nearer. The thumb seems carefully
placed in close proximity to the ball. Very
soon afterwards the upper fingers—and
the hand—move downward. By the time
first contact is made by the thumb, the
upper fingers are so close that they are a
tiny fraction of a second away from reach-
ing the target themselves. Notice that be-
cause of the thumb’s first contact, the ball

T=10/390 } )
\

T=15/390

ee
T=20/380~

A SOR
2a
Ss

Figure 14. Sequence of tracings from projections of
filmed capture of ping pong ball by hand. “T’ designa-

tions show elapsed time in seconds from first frame
shown. Explanation in text.

T=22/390

Frazzetta 241

begins to bounce off of it, but that the
proximity of the upper fingers is suffi-
cient to arrest this momentary deflection,
and to contain the ball. During the
capture, although the upper fingers are
brought close to the target, they are not so
close that they may accidentally touch—
thus deflect and possibly lose—the object
during the time that attention is being
focused elsewhere (the thumb).

Kinetic lizards may deal with similar
problems in prey capture. Ideally, both
upper and lower jaws should be posi-
tioned precisely only a very small dis-
tance away from the prey so seizing can
occur immediately at the instant that the
jaws begin to close. No creature, how-
ever, is so accurate. In the lizards
studied, this inaccuracy seems much
greater in upper jaw than in lower jaw
placement. Its deficiency can partially be
compensated by certain kinds of adaptive
mechanisms which can act to dilute the
failure that inaccuracy may cause.

The kinetic jaw machinery can func-
tion as such a compensator. But so,
presumably, could an akinetic upper jaw
(lordansky, 1966). Hence, in assessing
the possible adaptive role of kinesis, it
becomes necessary to compare kinetic
with akinetic skulls.

Figure 15 illustrates a lizardlike reptile
taking prey from the ground. The
animal’s position is close to the end of the
delivery phase, and near the beginning of
the seizing phase. When delivery first
began the head and jaws were located at
some distance “behind” the prey. The
delivery brings the jaws rapidly to either
the place where the prey was located
when this phase began, or to the esti-
mated new location of an active prey as
gauged by its movement when delivery
was initiated.

Possibly some changes in aim can be
made during delivery, as a means of
“tracking” elusively moving prey.
Whether or not this occurs, the move-
ments during this phase must bring the
jaws to a seizing position. In Figure 15,
the jaws are positioned for seizing. A

242

vector is shown representing the momen-
tum of head and forebody, produced by
the thrusting effects of limb, trunk and
cervicocranial musculature. The syner-
gistic actions of this musculature result in
a smoothly, coordinated delivery of jaws
to prey.

Seizing of the prey requires an upper
jaw movement which can be shown (as in
Figure 15) vectorially as the required
change in momentum of the head. Data
on the lizards have shown, already, that
head tiltings can occur suddenly, during
the delivery phase, thus changing the
head’s momentum. This change is most
probably caused by _ cervicocranial
musculature.

But a change of momentum at the
instant of seizing the prey has different
implications than such a change occur-
ring earlier, in the delivery phase. This is
because successful seizing depends upon
a reasonably precise placement of the
jaws upon the prey’s surface, with the
correct magnitude and development of
mouth-closing forces. The momentum
change in the seizing phase has to be
rapid, complete, and its scheduling well
timed. In an akinetic skull, it requires a
break from the functional integration and
coordinated, sequential action of the
considerable musculature which func-
tionally is carrying out the delivery.

A deleting of some muscle group from
the control of the delivery, to enlist it
suddenly in the seizing phase, all within
hundredths of a second to effect a preci-
sion grasp of elusive—or potentially
elusive—prey, could, despite these tick-
lish factors, result in a successful capture.
But it would seem that much greater
assurance of success can be had through
kinetic movement. The capacity of any
type of kinetic system, enabling the up-
per jaw to snap down on the prey without
usurpation of the delivery musculature,
can close the distance between prey and
upper jaw established during mandibular
placement, without the impairment of
other facets of the attack procedure.

In the lizard examples observed, the

Advances in Herpetology and Evolutionary Biology

consistent, carefully timed retraction just
as prey is seized, and the relative incon-
sistency of head tilt scheduling, suggests
that retraction lacks a very constrained,
operational connection with any specific
kind of head or neck displacements.
Kinesis frees the performance of the
upper jaw from a rigid coupling with
movements and attitudes of other por-
tions of the head and of the forebody.
Perhaps the range of kinetic move-
ment, from the position of the upper jaw
in protraction, to that when it is retracted
at the moment of seizing prey, is a
measure of the inexactness of upper jaw
placement. If a lizard, while concentrat-
ing on the placing of the mandibles,
inaccurately positions the upper jaw,
through head and neck movements,
farther away from the prey—by a distance
s—than its upper jaw can readily traverse

Figure 15. Diagram of generalized reptile capturing
prey upon ground. Vector R shows momentum of head
and neck immediately prior to capture. Vector S shows
direction upper jaw must take to engage prey surface.
Magnitude of S is adjusted to make vector C as short
as possible. Vector C is the minimum change in
momentum required to capture prey if the skull were
akinetic. Further explanation in text.

i}
|
|

KINESIS IN ALLIGATOR LIZARDS - Frazzetta

by means of those same sources of move-
ment, the kinetic range should be,
approximately, s.

The parameter s need not be very
great. Kinetic ranges in some lizards are
small. Crompton (1955) in a study of
kinesis in some mammal-like reptiles,
decided that their kinesis had no adap-
tive implications because the kinetic
range was so restricted. However, even a
very small kinetic range can spare the
head a very large momentum change
(Fig. 15).

In many akinetic predators, head
movements are a large part of both the
delivery phase and the seizing phase.
Kinetic predators, such as Gerrhonotus,
can reduce the importance of head move-
ments in the seizing function. In these
forms, head movements, if they occur
beyond the delivery phase, act as an
operational link between delivery and
seizing, because the final grasp belongs
to the functionally independent kinetic
jaws.

ACKNOWLEDGMENTS

I am grateful to many colleagues who
assisted me throughout this study.
Prominent among them are Emest E.
Williams of Harvard’s Museum of Com-
parative Zoology, who loaned specimens,
including two of the preserved
Sphenodon, and discussed this work
during its preliminary phases; George
Zug provided study skulls from the col-
lections of the U.S. National Museum,
and Hyman Marx loaned material from
Chicago’s Field Museum; Richard Estes
offered critical comments on this study in
its early days; Dennis W. Nyberg,
Timothy C. Moermond, Kenneth V.
Kardong, Carlos F. Pinkham, F. William
Burley, and the late Sandra Harness
provided discussion and laboratory assis-
tance. The work was carried out with the
support of National Science Foundation
Grant GB 5831.

243

LITERATURE CITED

ALEXANDER, R. MCN. 1967. The functions and
mechanisms of the protrusible upper jaws of
some acanthopterygian fish. J. Zool., Lond.,
151: 43-64.

BEECHER, W. J. 1951. Adaptations for food getting in
the American blackbirds. Auk, 68: 411-440.

Bock, W. J. 1964. Kinetics of the avian skull. J.
Morphol., 114: 1-42.

CALLISON, G. 1967. Intracranial mobility in Kansas
mosasaurs. U. Kans. Paleont. Contr., (26): 1-15.

CARROLL, R. L. 1977. The origin of lizards, pp. 359-
396. In S. M. Andrews et al. (eds.), Problems in
vertebrate evolution. Linn. Soc. Symp. Ser.

CROMPTON, A. W. 1955. A revision of the Scalo-
posauridae with special reference to kinetism
in this family. Navors. Nas. Mus., 1: 149-183.

FRAZZETTA, T. H. 1962. A functional consideration
of cranial kinesis in lizards. J. Morphol., 111:
287-320.

____. 1975. Complex adaptations in evolving popu-
lations. Sinauer Assoc., Sunderland, Mass.
Gans, C. 1961. The feeding mechanism of snakes
and its possible evolution. Am. Zool., 1: 217-

DENT

____. 1969. Comments on inertial feeding. Copeia,
4: 855.

GomgEs, N. 1974. Anatomie comparée de la muscula-
ture trigéminale des lacertiliens. Mém. Mus.
National d’Hist. Natur., ser. A, Zool., 90:
1-107.

GoMEs, N. M. B., AND J. P. Gasc. 1973. Etude
biomécanique du mouvement de fermeture de
la mandibule chez Ophisaurus apodus (Sauria,
Anguidae). Papeis Avulsos d. Zool., 27: 1-25.

Gow, C. E. 1975. The morphology and relationships
of Youngina capensis Broom and Prolacerta
broomi Parrington. Paleont. afr., 18: 89-131.

Haas, G. 1973. Muscles of the jaws and associated
structures in the MRhynchocephalia and
Squamata, pp. 285-490. In G. Gans and T. S.
Parsons (ed.), Biology of the Reptilia, 4, NY,
Acad. Press.

HEINTZ, A. 1932. The structure of Dinichthys: A
contribution to our knowledge of the Arthro-
dira. Am. Mus. Nat. Hist., Bashford Dean
Mem. Vol., 4: 115-224.

IORDANSKY, N. N. 1966. Cranial kinetism in lizards.
Contribution to the problem of the adaptive
significance of the skull kinetism. Zool.
Zhurn., 45: 1398-1410. (In Russian).

____. 1970. Structure and biomechanical analysis of
function of the jaw muscles in the lizards. Anat.
Anz., 127: 383-413.

____. 197la. Cranial kinesis in the evolution of
reptiles. General Biol., 32: 348-365. (In Rus-
sian).

—____. 1971b. A contribution to the functional analy-
sis of the skull of lizards (Lacertilia). Peculiari-
ties of the construction of the skull related to

244

cranial kinesis. Zool. Zhurn., 50: 724-733. (In
Russian).

JaARVIK, E. 1954. On the visceral skeleton in
Eusthenopteron with a discussion of the
parasphenoid and palatoquadrate in fishes. K.
Svenska. Vetens.-Akad. Handl., (4)5: 1-104.

LieM, K. F. 1970. Comparative functional anatomy
of the Nandidae (Pisces: Teleostei). Fieldiana,
Zool., 56: 1-166.

MacLean, W. P. 1974. Feeding and locomotion
mechanisms of teiid lizards: functional
morphology and evolution. Papeis Avulsos d.
Zool., 27: 179-213.

McDOowELL, S. B., AND C. M. BOGERT. 1954. The
systematic position of Lanthanotus and the
affinities of the anguinomorphan lizards. Bull.
Am. Mus. Nat. Hist., 105: 1-142.

Moy-TuHomas, J. A., AND R. S. MILES. 1971. Paleo-
zoic fishes. Philadelphia, Saunders Co.

OstTroM, J. H. 1962. On the constrictor dorsalis
muscles of Sphenodon. Copeia, 1962: 732-735.

PANCHEN, A. L. 1964. The cranial anatomy of two
Coal Measure anthracosaurs. Phil. Trans. Roy.
Soc. Lond., Ser. B., 247: 593-637.

PFANNENSTIEL, M. 1932. Der kinetische Schadel
karbonischer Stegocephalen. Fortsch. Geol. U.
Palaeont. (Deecke-Festschrift), 11: 278-305.

POGLAYEN-NEUWALL, I. 1953. Untersuchungen uber
die Trigeminusmuskulatur von Hatteria.
Zeitschr. Wiss. Zool., 157(1-2): 57-76.

RIEPPEL, O. 1978. Streptostyly and muscle function
in lizards. Experientia, 34: 776-777.

___. 1980. The trigeminal jaw adductor muscula-
ture of Tupinambis, with comments on the
phylogenetic relationships of the Teiidae
(Reptilia, Lacertilia). Zool. J. Linn. Soc., 69:
1-29.

Advances in Herpetology and Evolutionary Biology

RoBINsoN, P. L. 1967. The evolution of the Lacer-
tilia. Probl. Actuels de Paleont. (Evol. d.
Vertebr.) Colloq. Intern. D. Centr. Nat'l. de la
Recherche Scient., 163: 395-407.

____. 1973. A problematic reptile from the British
Upper Trias. J. Geol. Soc. Lond., 129: 45779.

RoMER, A. S. 1956. Osteology of the reptiles. Uni-
versity of Chicago.

Romer, A. S., AND R. V. WITTER. 1942. Edops, a
primitive rhachitomous amphibian from the
Texas Red Beds. J. Geol., 50: 925-960.

RUSSELL, D. A. 1964. Intracranial mobility in
mosasaurs. Postilla, 86: 1-19.

SCHAEFFER, B., AND D. E. ROSEN. 1961. Major adap-
tive levels in the evolution of the actinoptery-
gian feeding mechanism. Am. Zool., 1: 187—
204.

SMITH, K. K. 1980. Mechanical significance of strep-
tostyly in lizards. Nature, 283: 778-779.
THOMSON, K. S. 1966. Intracranial mobility in the

coelacanth. Science, 153: 999-1000.

THROCKMORTON, G. S. 1976. Oral food processing in
two herbivorous lizards, Iguana iguana
(Iguanidae) and Uromastix aegyptius
(Agamidae). J. Morphol., 148: 363-390.

____.. 1978. Action of the pterygoideus muscle dur-
ing feeding in the lizard Uromastix aegyptius
(Agamidae). Anat. Rec., 190: 217-222.

____. 1980. The chewing cycle in the herbivorous
lizard Uromastix aegyptius (Agamidae). Arch.
Oral Biol., 25: 225-233.

VERSLUYS, J. 1910. Streptostylie bei Dinosauriern.

Zool. Jahrb. Abt. f. Anat., 30: 177-258.

1912. Das Streptostylie-Problem und die
bewegung im Schadel bei Sauropsida. Zool.

Jahrb., Suppl. 15, 2: 545-716.

The Evolution of the Subdigital Pad in Anolis.
I. Comparisons Among the Anoline Genera

JANE A. PETERSON!

ABSTRACT. The morphology of the subdigital pad
in each of the three anoline genera, Chamaeleolis,
Chamaelinorops, and Phenacosaurus, is compared
to that of Anolis species with similar body size and
where possible similar structural habitat. Chamae-
leolis and Phenacosaurus exhibit gross pad and seta
morphologies that are very similar to those of
Anolis. The gross morphology of the Chamaelin-
orops pad resembles that of several derived and
more terrestrial anole species, particularly A.
auratus, but the structure of the setae and spines is
dramatically different from that of any other anoline
examined. It is not clear whether the Chamae-
linorops morphology represents a highly special-
ized derivative of an Anolis-like pad complex or an
independent evolutionary experiment in subdigital
pads. The comparisons among Phenacosaurus,
Chamaeleolis, and a number of the more primitive
anole species suggest that the major elements of the
pad complex—the morphotypic series in fine struc-
ture, including the setae, and lamellar scales form-
ing a pad—are homologous among these anolines
and that the pad complex appeared quite early
in the anoline radiation, before Chamaeleolis di-
verged.

INTRODUCTION

Anolis and its close relatives, Chamae-
leolis, Chamaelinorops, and Phenaco-
sadurus, constitute a highly successful
radiation of arboreal iguanid lizards. The
four genera are unique among iguanids, a
group which includes many arboreal
forms, in possesing a subdigital adhesive
pad. The pad consists of laterally ex-
panded flexible scales, the lamellae. The
external surface of the lamellae is

1 Department of Biology, University of California,
Los Angeles, Los Angeles, California 90024, U.S.A.

covered with microscopic hairs or setae
(the presence of setae in Chamaelinor-
ops, Chamaeleolis, and Phenacosaurus
has previously been reported by
Maderson, 1970). The precise mech-
anism of grip is unknown, but the setae
appear to be the major agent (see Hiller,
1968 and Peterson and Williams, 1981 for
discussion of setal function in Anolis).

The pad complex appears to be a key
innovation (sensu Liem, 1973) in the
successful radiation of the group. Com-
pared to animals that must rely on their
claws and toe position for grip, anoles
appear to have greater range of useable
locomotor substrates. They can grip or
adhere to a perch of almost any diameter
(or radius of curvature), spatial orienta-
tion, or surface texture (e.g., surfaces as
smooth as glass or as rough as bark). It
may become possible to state the rela-
tionship between the pad complex and
physical aspects of the habitat more
precisely in the future, but the adhesive
pad seems to be among the morpholog-
ical adaptations that permit anoles to
adopt highly acrobatic locomotor be-
havior and to exploit the spatial heteroge-
neity of arboreal niches.

In large part through my association
with Ernest E. Williams I have become
interested in the origin and adaptive
diversity of the pad complex. In this
study Chamaeleolis, Phenacosaurus, and
Chamaelinorops are compared with
Anolis. Each of the non-Anolis genera
combines primitive osteological features
with unique specializations (Williams,
1977), and two or possibly all three of the

246

genera diverged from the Anolis stock
relatively early (Fig. 1). Chamaeleolis
and Chamaelinorops may represent a
much earlier invasion of the Greater
Antilles than does Anolis itself. If they
are relics of an early radiation in which
pad morphology, as seen in Anolis, was
not fully established, some features of the
pad complex, such as setal stalk diameter
and the density of setae on the lamellae,
may have evolved in parallel in two or
three lineages. Differences in pad
morphology among Chamaelinorops,
Chamaeleolis, and Anolis could reflect
alternate pathways in pad evolution.
Phenacosaurus is distinct from Anolis,
but it may be derived from very early
South American anoles (Lazell, 1969;
Etheridge, 1960). It could represent yet a

% S
a @ \y
x e @ y e
S > % \) N)
S
S cS & & &
\S NY WW
* Q Q G GQ
inferiorly casque head Jextreme inter-zygapophysial
directed 2 sternal ribs | casque head plates on dorsal
transverse shortened vertebrae
processes= digits
pseudo-
diapophyses
angular lost splenial
MULTIPLE CHANGES : anananioet
WITHIN THESE GROUPS
inscriptional ribs 5:2 to
2:2 (in betas even 1:3) inscriptional
ribs 4:3
inscriptional
ribs 4:2
inscriptional
ribs 4:2 or 3:3

caudal
transverse
processes
lost

splenial and angular present
laterally directed caudal
transverse processes

dewlap and toe pads present
q inscriptional ribs 5:2

Advances in Herpetology and Evolutionary Biology

third or a fourth lineage in which the pad
complex was perfected.

The comparisons are designed to
determine whether each of the genera
can be distinguished from Anolis. In
order to minimize the effect of differ-
ences in body size or habitat preference
on the comparisons, each genus is com-
pared with Anolis species which are
similar in habitat preference and body
size. Ideally the comparisons would also
minimize the phylogenetic distance
between the Anolis species and the
anoline genus, but in practice there are
too few data to identify nearest neighbors
within Anolis. Chamaeleolis, a large,
canopy-dwelling form, is compared with
a crown-giant anole ecomorph, A.
cuvieri. Chamaelinorops is a small and

a-Anolis Phenacosaurus

Northern South

America
Aptycholoemus |
Anisolepis B-Anolis
Chamaelinorops
Urostrophus Hispaniola
wo
0)
Se
eC
S Chamaeleolis
Cuba

anolines

Enyalus

preanolines:
Diplolaemus
Lelosaurus
Aperopristis

Pristidactylus

basal iguanids
Hoplocercus
Morunasourus
Enyalioides

b

Figure 1. The relationships among the anoline genera according to A) Williams (1977) and B) Etheridge [after
Paull, Williams and Hall (1975)]. The osteological and external characters on which the phylogeny is based are
included in A; B is based on osteological and chromosomal data.

EVOLUTION OF THE SUBDIGITAL PAD - Peterson

relatively slow-moving (personal obser-
vation) animal which is often found in
leaf litter. There is no completely suit-
able comparison within Anolis, but in
size, color, body shape, and cryptic be-
havior, Chamaelinorops most closely
resembles some of the twig-dwarf anole
species, and it is compared with A.
sheplani and A. occultus. These species
are also of interest because they are
thought to be among the most primitive
forms in the West Indian anole radiation
(Schwartz, 1974; Williams, 1976). Addi-
tional comparative data are included for
A. chrysolepis and a number of other
more terrestrial anole species. Phena-
cosaurus is compared with A. valen-
cienni. The two forms are similar in body
size and shape and in their preference for
slow movement on small diameter
perches. A. valencienni, however, may
be a relatively derived anole, and it is not
closely related to the mainland species
thought to be close to the ancestors of
Phenacosaurus. The comparison is sup-
plemented with data from the more
primitive West Indian species.

The morphological comparisons be-
tween Anolis species and each of the
genera concentrate on several specific
issues: 1) phalangeal proportions and the
distribution of scales relative to the
phalanges; 2) external morphology of the
lamellae and _ generalized subdigital
scales; and 3) the dimensions, spacing,
and distribution of setae. 4) Lastly, setae
appear to be derived from the spinules
which characterize the epidermis of
many iguanids (Ruibal and Ernst, 1965;
Ruibal, 1968; Maderson, 1970). The two
morphologies can be linked through a
morphotypic series (Olson, 1975) of
transitional forms which actually exist in
the basal or proximal portion of the
lamellae (Peterson and Williams, 1981;
Fig. 2; see discussion). Intergeneric
differences in the character of the series
could provide some of the most interest-
ing insights into the patterns of evolution
associated with the origin of the pad
complex.

247

MATERIALS AND METHODS

Table 1 lists the specimens examined
along with their body size (measured as
snout-vent length) and digital measure-
ments.

Scales and phalanges of the fourth digit
are numbered from distal to proximal
following Boulenger (1885) and systema-
tic convention rather than anatomical
convention (e.g., phalanx i in our termin-
ology is phalanx iv in anatomical nomen-
clature). The claw base is included in
phalanx i scale counts and measure-
ments. Following Peterson and Williams
(1981), lamellar scales are defined as
those having a thin, frayed free margin
and a width to length ratio greater than
one.

The distribution of scales relative to
the phalanges was determined by flexing
and extending the metatarsal-phalangeal
and interphalangeal joints. A scale adja-
cent to the iii/iv interphalangeal joint was
counted as a phalanx iii scale if it moved
with phalanx iii. This definition may shift
the phalangeal boundaries slightly
proximal to the transverse plane through
the joint, but it decreases ambiguity in
the count and emphasizes functional
associations. The i/ii interphalangeal
joint is synonymous with the distal
border of the pad in most anoles. The dis-
tal pad scales are often closely packed
and project out over the proximal phalanx
i region. These pad scales are included
with the phalanx ii scales if their bases
are attached to phalanx ii or to the inter-
phalangeal joint. In most anoles the pad
of the fourth digit lies under phalanges ii
and iii, and the scale count for these
phalanges is just one scale greater than
the total number of lamellae (the most
distal pad scale is non-lamellar; Table 1).
There are, however, significant differ-
ences in the distribution of lamellae
relative to the phalanges. These are de-
scribed in the text, and a separate count
of lamellar scales is included in Table 1
for species in which lamellae do not

Advances in Herpetology and Evolutionary Biology

248

(O1'0) S¢"0 (98°0) Fr'0 610 (T€) OF (Vaud g'),) LIU Eg 9 ae) UIUL ()'6) 6070 ZOW
snULapOLaZaYy SNINDSOIDUaYd
“WIL OE BO8TOT ZOW “MW O'6E CSE d “WW JE OTE d ‘WUC’ PE BBE od ‘SUsUIOEds [eUOnIPpY
(S0°0) 0S'0 L¢'0 ITO LI ONG Of COCO NY CL Oe 8969ST ZOW
$n}jNII0 syouy
(90°0) 19°0 LS'0 0) 0) 91 BEVIN 86 Cees ponents (eke, Tc00FT ZOW
(G0°0) £9'0 LS'0 L0°0 91 CO OG 86 OO) renee tes OI Tv9Scl ZOW
tupjdays syouy
"WU EF 16986 ZOW ‘ueurtoeds [euonIppy
(70'0) OF'0 (FS'0) 870 810 (9) L (Wau ¢*[) WILT QT GG ES EMBO OPTS 6169ST ZOW
(€0'0) 3E°0 (230) SE°0 810 (L)6 (UaUL (QT) UU YT 9G BENOEL Yt Oe: 8éLLS ZOW
(70°0) GF'0 (€Z'0) 860 810 (9) 6 (unuA G'T) UIU 1°g 93 wu 19 UI G"9F €Z69ST ZOW
(F0'0) TF'0 (96°0) €€°0 810 (1) 6 (UuuL g*Q)) WILT ET LG AG. ES 66LLS ZOW
(70'0) 9F'0 (G3'0) 6€'0 1c0 (1) 8 (aun Q"T) WILE PT 8G ENON SY IGG P6SPL ZOW
(€0" 0) 70 (130) Z€'0 LT 0 (9)8 (aun G°0) WUT g’T 86 ay OG 80L9GI ZOW
(F0'0) FEO (8Z°0) ZE0 1c0 (1) 8 (Wau GT) WILL ()'T GG AS, Benner(g) Ilts, TSsscT ZOW
1unogivg sdosouyanupy )
‘WU TET PL6SE ZOW
“WU TET LL6GE ZOW “WU GZT 8L6GE ZOW “WU GZT TL6SE ZOW “WU OZI SL8T9 ZOW ‘Wu ZT SL6GE ZOW ‘sueuttoeds jeuonIppy
(80°0) Tr'0 €S'0 810 T€ COMnONS ) 8S UIUL 8° GG UIUL PCT 9L6SE ZOW
(80'0) 170 €S'0 0G 0 T€ CUE GG) 8S EMDES YEG Uru GGT OL6SE ZOW
(80°0) 8€°0 vS'0 00 €€ UIUL Q'O)T 19 UU 8° eee SIL TL6SE ZOW
11010N9 SIOUY
(S0°0) OF'0 vS'0 €1 0 It EES LS CSA Mone SII 6S8P ZOW
sno1od syoajapwpy)
(90°0) 3F'0 6s'0 €10 T€ WENO Nts} 09 EE CIOG SE Tee6s ZOW
GO'0) Tv'0 vS'0 G10 (63 BONO Oe ES) 6S RO Weaner LON *C€6L9 ZOW
SapluoajavUDyO s1joajavULDY D
yysu9] SOTBIS [P}0} yysua'y (sayeos ped) (ped VISIp WAP VISIp WP yysue] uowitoeds
yUdA-ynNoUS Saeos ped qU9A-JnNOUS IIT pue jo yy8ue71) uO SaTROs yysueT] yU9A-}NOUS sa1oeds
yysuag] ped —_—_—_—_——_—_—— II SoTeos TIT pue I [eqsipqns
SSTPOS TROL = yysueTUsIq = [eNSIpqns sosuejeyd [RIOL
suey usiq =————— wsue7]
So em. oe EeQIESS[¥os
TH 29 HE yysueT

“SNOISNAWIC TWHONVIVHG GNV GANINVXY SNAWIONdS “[ ATAV I,

249

EVOLUTION OF THE SUBDIGITAL PAD - Peterson

IV0
cv0
Iv0

Or'0

(F0'0) €F'0
(60°0) €F'0
(90°0) €F'0
(F0'0) FF'0
(Z1°0) G0

(L0°0) 0F'0

cv 0
9F'0
€V'0

v0

(620) 170
€V'0
(G0) LV'0
(6Z'0) 9F'0
(8¥'0) €F'0
vS'0

(62'0) SF'0
(0€°0) Lr'0

0c 0

I¢0

LOO

cc'0

910

0c 0

960

1c0

VCO

910

Il
Il
cI

VI

(6) FI
(6) FI

96
96

(83) FI
(G&) FI
(SZ) FI
(LZ) FI

‘unasnyy A1O}sTF] [BANQeNY AJUNOD sojesuy soT V
‘WOT}I9T[OO Ss IOYNe ay} UI SUBUTIOSdS 6
“AYISIOATU() PIBVAIeH] ‘ABO[OOZ, sanereduioD Jo wnesnyy x

TORS 96 UIUL | "OG UIUL GOT 9LEOFT VOW
(Qepruensy jo dnois suLIneseUNIOUI IO [eseq) SIuDjJNUUD sninDsSDUNLOPW

BONO 1b VG CsI SS 9L8LE ZOW
(eeptuensy Jo dnois suLIneseunIOU IO [eseq) 1015 SNINDSDUNLOJW

ESC SE WO Le: Hone GAEL PEOGE ZOW
(sepruensy fo dnois aulost[loseq) sadisuo] snjoupwanT]

UIUL QT T VE ENS (6 BONES (OTETL SVL8G ZOW

(eepruensy] fo dnois oulostploseq) snzoias snjouvUulav'T
‘PPIGE HNIWNOV I Snwapjna reou ‘u ‘ds ‘y :suauttoads s1zjouy [euOnIPpy

(uuu T°) WU P¢ VE Oe NL ts} Wenes 7 S| I8L¢Z HNWOV'!
(e}9q ‘UOQIpuod sdoION ‘ped MoLeU) snzvIND syouy

Wann eM (7 WU §°6T WON Ag) LOGVE HNWOV'!
(eJeq ‘UONIpUOD sdo1ON ‘ped MOLIeU) 1Way1DG S1JOUY

(WILL ¢°Z) WU OF VE UU OIL DORUE YL ccses ZOW
(e}0q ‘ped MoxteU) szdiysojids snzyouopido.y syouy

(WIL g*T ) WU gg ce HEROD te) WEe Or) Ly 86L8E ZOW
(ej9q ‘ped Moxreu) snjouop1do.4 snjouopido.s4 syouy

(WILL Q°g) LIU FJ vS Cael Wo! (Oy teyZ, 6L69F1T ZOW
(eydye ‘uoytpuoo sdo1oy ‘ped Moxeu) s1jp140jDNbap syjouy

CE ONS eV COOK WERE) ISIS PST9GI ZOW

(eJEq pezi[e1sues) uo1dpjUad syOoUy
“WU 9 G8PPL _ HNINOV' ‘vouttoeds [euontppy

(uLUL [°G) WU TP I€ manne OIL y AV, STIIGST ZOW
(WILL T°) WIL OF O€ CEERI OU OIGS VITGST ZOW
(ejJeq pezije1ouas ‘ped MoxteU) srdajoshiyo syouy

eS, eV ONO ey ete SHS 66E d

eens) te) cV ett) de, Paver) EL) 69¢ d
1UUA1IUA]DA SIOUY

(WILL g°G) ULL ZZ ve wonuor (yg) Coe ORS O0€ d
(WILL F°g) WILL Q°¢ LE wud € OL UIUL ()'OL LOS6ET ZOW
(UIUL Q" J.) LIU QB" 0€ eet te) WONUON Ag) OLS6ET ZOW
(ULL 9'g) WU GE GE eonver TES} Maemo HCC ST) 99S6ET ZOW

250

occur throughout the phalanx ii and iii
regions. In Chamaeleolis and some
Anolis specimens there are not one, but
several series of subdigital scales in the
phalanx iv region. Many anole specimens
also have double, side-by-side scales in
the positions just distal to the pad. In
these instances only one of the series is
counted so that the scale count reported
refers to “scale places” rather than the
total number of subdigital scales.

Gross measurements were taken with
dial calipers (reading to 0.02 mm) or an
ocular micrometer. Phalangeal lengths
were taken with the ocular micrometer to
the nearest 0.1 mm. The accuracy of the
measurement partially depends on the
condition of the specimen and error
estimates vary. The standard deviation of
six replicate measurements of the length
of phalanges ii and iii was 0.1 mm (45%
of the mean value) in five error trials
(specimens of Phenacosaurus, A. occul-
tus, A. sheplani and Chamaelinorops)
and 0.2 mm (11% of the mean value) in
one trial (specimen of Chamaelinorops).
Phalangeal length measurements were
taken along the subdigital scale row.
Actual lengths were defined by the scale
borders rather than the joint axes. For
example, the length of phalanx iv was
defined as the distance from the base of
the most proximal phalanx iv scale to the
distal margin of the most distal phalanx iv
scale. The length of phalanges ii and iii is
in most cases equivalent to the length of
the pad region (Table 1). A separate
measurement of the length of the pad (=
the length of the series of lamellar scales)
is included for species in which this is
not the case.

The digits were prepared for scanning
electron microscopy as described in
Peterson and Williams (1981). Terminol-
ogy for the fine structure follows that of
Peterson and Williams (1981; summa-
rized in Fig. 2), Hiller (1968), Maderson
(1970) and Ruibal and Emst (1965).
Measurements of fine structure were
taken from photographs and are reported

Advances in Herpetology and Evolutionary Biology

as the mean of 12 to 30 individual
measurements with the 95% confidence
interval (for some spinule measurements
n=6). Magnification was adjusted so that
measurement with a dial caliper to the
nearest 0.1 mm resulted in measurement
to the nearest 0.01 um for stalk diameter,
seta tip dimensions and spinule dimen-
sions or to the nearest 0.1 um for stalk
height. The measurement techniques for
fine structure are judged to be adequate
based on F-tests of the variance among
individual setae compared to the vari-
ance among replicate measures (F value
in each test significant at the 0.001 level).

The major difficulty in characterizing
and comparing species in terms of setal
dimensions is to obtain a realistic sample
of measurements. Measurements taken
within about a 30 um radius on one speci-
men are quite consistent and confidence
intervals are small. Measurements from
one or even two such areas underesti-
mate regional variation, particularly in
setal stalk height which increases and
then decreases again along a proximodis-
tal transect on each lamella (e.g., in A.
cuvieri [MCZ 61875] mean stalk height
for the central and proximal regions of a
mid-pad lamella are 16.6 + 1.8 wm and
24.4 + 0.4 pum, respectively). Because
sampling from a large number of areas is
prohibitive in a comparative survey, the
interspecific and intraspecific compari-
sons are made for setae from homologous
regions of mid-pad lamellae. The mean
value for an individual specimen is based
on measurements taken from two to four
of these regions.

The measurements of stalk dimensions
are taken from the center of a mid-pad
lamella on setae adjacent to the junction
between the lamellar core and the free
margin (see Ruibal and Emst, 1965, or
Lillywhite and Maderson, 1968, for
further definition of this region). These
are usually the tallest setae on the
lamella (Ruibal and Emst, 1965). Varia-
tion in stalk diameter is smaller and ap-
pears to be less well correlated with

251

EVOLUTION OF THE SUBDIGITAL PAD - Peterson

‘(LE6L ‘SWEIIII/\\ PU UOSJA}8q WO) SOU JO Salas (\yBu) eJaS-suIds pue (yJ9|) Sulds-ajnuIds SIdAjoydiow ay) “Zz ainbi4

OpiM wit Gy ~
Buoy; wnio7~ uur ‘bs
diy apjnbuotiay di} aytj-yooy url ‘bs7|~ Ajisuap /sanuids O| ~
Ja,BWDIP JUIOd-piwW LUT! G~ JO yybin4sjs sayawoip wuTG’~ dij aytj-yooy sajyawoip wt G|'~
|}, WOS-Ol oywMOZ-G ||O;}wWMIO?-¢ JO} wit G|-G 4 WITT G o} dn JO, jwTt{~ jj} WT O"|-S"

diy a
suapIM pasado, Ajybus Ajisuap \
diy pauasso|4 diy pauayfoi} ‘yun|q O} Vv J9,9WOIP div
TUN, TN, Uy UN,
<— || <—
ajnuids
d;OIPaw aul uaydindysaqoO
Hbuoid/ojes Bbuoid ayids aulds auids |jOwWs ayoinuids
SJY3SH SNIDSE
S3lY4S
O15 TVLISIGENS
SIONV

Q3ZI1VYSAN39

=

252

lamellar region so that the thickness of the
setal stalk at the lamellar/free margin
junction bears no consistent relationship
to the range of dimensions (Peterson and
Williams, in preparation). Stalk height is
defined as the distance between the stalk
base on the lamellar surface and the
spatulate seta tip. Stalk diameter is
measured at a point between one third
and one half of the distance between the
stalk base and the spatulate tip. The stalk
tapers little in this region, and some flex-
ibility in the position of the diameter
Measurement is necessary because of the
presence of debris and the crossing of
stalks.

Measurements of the tip dimensions
are subject to a) regional variation (e.g.,
in A. cuvieri [MCZ 61875] tip area at the
distal margin is 0.165 + 0.023 sq pm
while that midway between the free
margin and the base of the lamella is
0.248 + 0.039 sq mm) and b) variation
apparently related to wear or age of the
Oberhautchen (= functional epidermal
layer; Maderson, 1970). In an A. occultus
specimen (MCZ 156968) the spatulate
tips over the central portion of the
lamellae are extremely frayed and small
(tip area measures 0.175 + 0.014 sq um)
while those near the lateral margins of
the lamella are much less ragged and
larger (tip area measures 0.237 + 0.020 sq
ym). Similar medio-lateral variation in
tip area does not occur in Anolis speci-
mens in which the central setae appear
relatively “unworn.” Wherever possible
the setal tip measurements have been
taken from the central portion of the pad,
but in several cases the area was shifted
laterally to minimize what appears to be
individual variation related to wear or
age of the Oberhautchen. In all cases the
tip dimensions are taken in regions mid-
way along the proximodistal axis of the
lamella. Tip area is calculated from
measurements of the width of the seta
margin and the height (= distance be-
tween the midpoint of the margin and the
junction between the flattened tip and
the oval distal region of the stalk). In

Advances in Herpetology and Evolutionary Biology

most anoles the tip is triangular, and area
can be estimated as 0.5 (width x height).
In Chamaelinorops the tip is square or
rectangular, and area is estimated as
width x height.

The spacing or density of setae is
determined by counting the setae over
250 to 750 sq wm. The entire lamella is
usually folded along its proximo-distal
axis, and in most cases the measurement
is based on a series of 120 to 250 sq wm
photographs taken midway up a gentle
slope where the surface is normal to the
viewing plane. This corresponds to a
position midway between the free
margin of the lamella and its base.

The dimensions and density of
spinules were measured on phalanx i
scales and/or on a lateral digital scale
immediately adjacent to the subdigital
scales, both areas were examined.
Spinule height is total elevation above
the scale surface. Spinules which are less
than 0.5 um tall are nubbinlike, and di-
ameter is measured at their midpoint.
Density is determined by counting the
spinules over 64 to 100 sq um.

RESULTS OF COMPARISONS

COMPARISONS BETWEEN CHAMAELEOLIS
AND ANOLIS CUVIERI

Gross Structure and Pad Proportions

The phalangeal dimensions and scale
distribution of Chamaeleolis are not
different from those in Anolis. The fourth
toe is shorter in Chamaeleolis than in A.
cuvieri (Table 1), but a number of Anolis
species, including A. valencienni, A.
occultus, and A. sheplani are not different
from Chamaeleolis in this respect. The
phalanges that underlie the pad (ii and
iii) are absolutely shorter in Chamaeleo-
lis, but the proportion of pad length/toe
length is similar in Chamaeleolis and A.
cuvieri (Table 1). Chamaeleolis has a
shorter pad in relation to its body length
than A. cuvieri, but several other anoles

EVOLUTION OF THE SUBDIGITAL PAD - Peterson

(e.g., A. sheplani and A. valencienni)
exhibit proportions similar to those of
Chamaeleolis. The total number of sub-
digital scales and the number of pad
scales are similar in Chamaeleolis and A.
cuvieri (Table 1).

The subdigital scales in both forms
exhibit three different morphologies cor-
responding to a) the pad region (phalan-
ges ii and iii); b) the phalanx i region; and
c) the phalanx iv region. In Chamaeleolis
and A. cuvieri all but one (see below) of
the scales in the pad region are lamellae,
and lamellae are absent in the phalanx i
and iv regions. The morphology of the
lamellar scales is indistinguishable in the
two forms (Chamaeleolis lamellae are
shown in Fig. 3A). In most specimens
one of the small scales that is inserted
between the phalanx i and phalanx ii
series to “raise” the pad is exposed on
the subdigital surface and forms the most
distal pad scale. This scale does not have
the characteristic free margin of a lamella
(it also bears spikes rather than setae).
Generalized subdigital scales occur in
the phalanx i and iv regions. Those in the
phalanx i region wrap around the phalanx
more tightly giving them a more convex
cross section. The phalanx iv scales in
Chamaeleolis are very irregular in shape,
and a distinct subdigital scale row is not
present (Fig. 3B).

Fine Structure

Setae cover the lamellae in both A.
cuvieri and Chamaeleolis (Table 2; Figs.
3C, 3D, 4A, B). The setae stalks are
slightly shorter in Chamaeleolis, but
stalk diameter, tip shape and the density
of setae are not different. With the excep-
tion of Chamaeleolis MCZ 59331, tip
area is also comparable in the two forms.
Chamaeleolis MCZ 59331 has a much
larger tip (Table 2). Measurements taken
proximally and distally on the pad and
over the width of four of the mid-pad
lamellae indicate that the enlarged tip is
not confined to a small region. It appears
to be an individual variation.

253

In all the specimens the complete
spine-seta subdigital series shown in
Figure 2 is present in the proximal por-
tion of the lamella.

While the morphology of the pad
region is very similar in Chamaeleolis
and A. cuvieri, the phalanx i and iv
regions exhibit some differences. The
phalanx i scales are grossly similar in the
two forms, but their fine structure is dif-
ferent. In Chamaeleolis the regions ad-
jacent to the distal margin of the scale are
covered by densely packed, small spines
1-2 wm tall and 0.3-0.4 um in diameter).
Spine height decreases toward the base
of the scale. The proximal one-half to
two-thirds of the scale is covered by
spinules (Figs. 3E, F). This area closely
resembles the spinulate Oberhautchen
(Fig. 2 far left). There are two slightly
different populations of spinules (Fig.
3F). The taller ones are 0.79 + 0.08 wm
tall and 0.22 + 0.02 um in diameter.
The shorter, nubbinlike spinules are
0.34 + 0.02 um tall and 0.22 + 0.02 um in
diameter. Together they occur in a den-
sity of 7.0 spinules/sq wm. This morphol-
ogy is present on all the phalanx i scales;
there is no apparent regional differentia-
tion in fine structure. The spinulate
phalanx i region is much more conserva-
tive in terms of the morphotypic series
(Fig. 2) than the lateral digital scales ad-
jacent to the pad. The lateral scales are
covered with spikes, prongs, spines, and
the seta-prong intermediate morphotype
(Figs. 2, 4F). The dimensions and density
of these forms are comparable to those of
the lamellar series in Chamaeleolis and
to those described for the generalized
anole subdigital series in Figure 2.

In A. cuvieri the phalanx i scales are
largely covered with spikes and prongs
(10.4 + 1.04 um tall and 0.53 + 0.07 um
in diameter; Fig. 4E). Small spines (Fig.
2), similar to those in Chamaeleolis, do
occur in A. cuvieri, but only in extremely
proximal areas of the scales. Spinate
areas are most extensive on the scales just
distal to the pad over the phalanx i
region. Regions of small spines also occur

254 Advances in Herpetology and Evolutionary Biology

Figure 3. The subdigital morphology of Chamaelinorops. \n this and all subsequent photographs the small arrow
adjacent to the figure scale points in the direction of the claw. A. Lamellae numbers 3, 4 and 5 in MCZ 57933. The
depth of the frays along the distal margin coincides with the limit of the free margin. B. The phalanx iv subdigital
scales in MCZ 59331. C. Oblique view of the setae tips in a mid-pad region of MCZ 57933. D. Setae stalks on the
free margin of a lamella in MCZ 57933. E. The surface of phalanx i scale 11 in MCZ 8459. The borders of the
epidermal cells are raised, spine-covered ridges. This is the only specimen of an anoline in which this morphology
rather than a bare trackway (Ruibal, 1968) has been found. The morphology does occur in a number of non-
anoline iguanids. F. Spinules on the proximal portion of phalanx i scale 8. Note the presence of spinelike and
nubbinlike spinules.

TABLE 2. COMPARISON OF SETA DIMENSIONS.

EVOLUTION OF THE SUBDIGITAL PAD - Peterson 255

Seta stalk Seta tip
Species/Specimen diameter height area density
Chamaeleolis chameleontides
MCZ 57933 0.56 + 0.03 wm 18.4 + 0.8 um 0.206 + 0.021 sq wm 1.0/sq wm
MCZ 59331 0.58 + 0.03 wm 18.6 + 0.3 um 0.377 + 0.039 sq wm 0.7/sq wm
Chamaeleolis porcus
MCZ 8459 0.53 + 0.02 pm 22.3 + 0.7 um 0.264 + 0.019 sq um 1.0/sq um
Anolis cuvieri
MCZ 61875 0.51 + 0.04 um 22.4 + 0.5 um 0.229 + 0.025 sq um 1.0/sq wm
MCZ 35975 0.60 + 0.02 wm 23.9 + 0.5 um 0.232 + 0.023 sq um 0.9/sq wm
MCZ 35971 0.65 + 0.04 wm 27.2 + 1.4 um 0.253 + 0.017 sq wm 0.9/sq wm
MCZ 35977 0.51 + 0.06 um n.a.* 0.184 + 0.020 sq wm 1.4/sq wm
Chamaelinorops barbouri
MCZ 155851 0.52 + 0.05 wm 6.9 + 0.7 um 1.014 + 0.089 sq um 0.5/sq wm
MCZ 68691 0.55 + 0.05 wm n.a. 0.649 + 0.060 sq um 0.5/sq um
MCZ 156919 0.54 + 0.05 wm 8.4 + 0.4 um 1.472 + 0.149 sq wm 0.6/sq wm
MCZ 126708 0.56 + 0.06 wm, 7.0 + 0.5 wm 0.759 + 0.072 sq wm 0.5/sq wm
MCZ 74594 0.47 + 0.03 wm 5.0 + 0.3 um 0.600 + 0.047 sq wm 0.5/sq wm
Anolis sheplani
MCZ 140021 0.39 + 0.02 um 12.6 + 0.3 um 0.220 + 0.020 sq wm 1.1/sq wm
MCZ 125641 0.41 + 0.03 um 11.8 + 0.5 wm 0.279 + 0.023 sq wm 1.2/sq wm
Anolis occultus
MCZ 156968 0.49 = 0.04 um 11.0 + 0.7 um 0.237 = 0.020 sq wm 1.4/sq wm
[or 0.175 + 0.024 sq um in “wom” regions]
Anolis sp. n. near eulaemus
LACMNH 72741 0.57 + 0.06 um 20.4 + 0.3 wm 0.593 + 0.045 sq wm 1.0/sq wm
Phenacosaurus heterodermus
MCZ 35970 0.41 + 0.02 um 14.6 + 0.2 um 0.226 + 0.028 sq 1.2/sq wm
MCZ 139566 0.46 + 0.02 um 14.8 + 0.3 um 0.214 + 0.021 sq pu 1./2sq wm
P 300 0.45 = 0.09 um 13.7 + 0.4 um 0.308 + 0.025 sq uw 1.2/sq wm
Anolis valencienni
P 329 0.47 + 0.03 um 17.2 + 0.6 um 0.209 + 0.016 sq um 1.1/sq um
P 269 0.40 + 0.02 um 15.3 + 0.4 um 0.171 + 0.018 sq um 1.4/sq wm

* Measurement not available.

A Measurements taken on the third digit pad.

on the lateral digital scales where they
form a landscapelike mosaic with small
bare areas (Fig. 4C, D). The dimensions
of the small spines (0.71 + 0.08 um tall
and 0.35 + 0.05 wm in diameter) are
similar to those of the small spines or
larger spinules in Chamaeleolis.

The major difference between the
spinate areas in A. cuvieri and the spinu-
late regions in Chamaeleolis is that in
the spinate areas the understory of nub-
bin-like spinules is very reduced and the
density of spines-spinules is about half

that in Chamaeleolis (3.6 spinules/sq wm
compared to 7.0 spinules/sq pm). In
terms of the morphotypic series, the most
conservative subdigital scale surface in
A. cuwvieri is less conservative than the
surface of the phalanx i scales in Chamae-
leolis.

Although the phalanx iv scales in
Chamaeleolis are grossly very different
from lamellae, they are covered with
setae. The seta stalk height is less than
that of lamellar setae (11.4 + 0.3 um tall;
cf. Table 2), but the diameter (0.58 + 0.02

256 Advances in Herpetology and Evolutionary Biology

ce
Ve efe.% eo Fee
et te ge

eee g

Figure 4. The subdigital morphology of A. cuvieri (A-E) and Chamaeleolis (F). A. Lateral view of the setae on a
mid-pad lamella in MCZ 35971. B. The setae tips in MCZ 35971. C. The surface of a lateral digital scale in MCZ
35975. Smooth bare area similar to that on the crest of a keel occurs to the lower right (distal portion of the scale).
The remaining surface is covered with fields of spines. The fine irregular lines separating the fields appear to be
the borders of the epidermal cells. D. Detail of the surface of a lateral digital scale showing portions of two small
bare areas and a variety of spine morphologies (MCZ 35975). E. Seta-prong intermediate tips in the phalanx iv
region of MCZ 35971. F. Seta-prong intermediate tips on the lateral digital scale of Chamaeleolis MCZ 8459. Tips
which are turned (lower center and upper right) show compression of the tip (cf. setae Fig. 3C).

EVOLUTION OF THE SUBDIGITAL PAD - Peterson

pm and tip size shape are similar to those
of lamellar setae.

The phalanx iv region in A. cuvieri
consists of well-defined series of general-
ized subdigital scales. The distal scales
bear setae, but within the distal third of
the phalanx (between scales 46 and 49),
the triangular seta tip is replaced by the
flattened, rectangular tip of the seta-
prong intermediate morphotype (Figs.
AE, and 2). The width of the tip is about
0.61 4m compared to 0.79 um for lamellar
setae. At the metatarsal-phalangeal joint,
the seta-prong intermediates are 14.5 +
1.0 um tall and .52 + 14 um in diameter.

COMPARISONS AMONG CHAMAELINOROPS,
A. SHEPLANI AND A. OCCULTUS,
AND COMPARISONS AMONG
CHAMAELINOROPS, A. CHRYSOLEPIS,
A. AURATUS, AND A. SP. N. NEAR EULAEMUS

Gross Morphology of the Digit
and Scales

The gross morphology of the digit in
Chamaelinorops is significantly different
from that in A. sheplani and A. occultus,
although it resembles that of A. chryso-
lepis and A. auratus in most respects.
The comparison with A. sheplani and A.
occultus is described first.

A. sheplani and A. occultus are very
similar. A pad is formed by lamellae
under phalanges ii and iii (scales 7
through 22 in Fig. 5B; scale 6 is a non-
lamellar pad scale). The pad is broad and
occupies at least half the length of the toe
(Table 1). More than half the subdigital
scales are lamellae (Table 1), and gener-
alized subdigital scales occur in the pha-
lanx i (scales 1-5) and phalanx iv (scales
23-30) regions.

In Chamaelinorops the pad has a dif-
ferent position relative to the phalanges:
it is reduced in size, and less than a
quarter of the subdigital scales are pad
scales (Table 1; Fig. 5A). a) In Chamae-
linorops the pad lies under the first and
second phalanges rather than the second
and third phalanges. The distal 2 to 4 pad

257

scales lie under phalanx i (scales 8-10 in
Fig. 5A), and the proximal 3 to 4 scales lie
under phalanx ii (scales 11-13 in Fig.
5A). All the phalanx ii scales are lamellae,
but none of the phalanx iii scales have
the shape or marginal features of a
lamella (scales 14-17 in Fig. 5A corre-
spond to the phalanx iii region). b) The
pad is smaller in Chamaelinorops. Pad
area estimated from the dimensions of
the individual lamellae is 0.41 sq mm in
Chamaelinorops (MCZ 156919; SVL 31
mm) and 0.69 sq mm in A. sheplani (MCZ
125641; SVL 40.8 mm). A scaling equa-
tion relating pad area and body size
(measured as snout-vent length [trunk
height]?) which was developed for six
Puerto Rican anole ecomorphs (including
A. occultus and A. cuvieri) would predict
a pad area of 0.82 sq mm in the Chamae-
linorops specimen (200% of the observed
value) and 0.90 sq mm in the A. sheplani
specimen (130% of the observed value)
(the equation is: pad area = constant
[body size]5° + 963 [y=0.963 for 36
specimens in 6 species] [Peterson et al.,
1982, and Peterson, unpublished data]).
While both specimens have small pads
for their size compared to the Puerto
Rican ecomorphs, the discrepancy is
much’ greater for Chamaelinorops.
c) Differences in phalangeal length and
lamellar morphology also contribute to
the smaller pad in Chamaelinorops. The
“pad” phalanges ii and iii are absolutely
shorter in Chamaelinorops, although the
whole digit is almost twice as long as that
of A. occultus and A. sheplani (Table 1).
Pad width is also lower in Chamaelino-
rops because the lamellae are much
narrower (see below). d) The smaller
number of pad scales (Table 1) is partial-
ly correlated with the shorter pad, but the
shape of the lamellae also contributes to
the difference. In A. sheplani and A.
occultus, and to a greater extent in A.
cuvieri and A. valencienni (Fig. 9B), the
distal phalanx ii lamellae are very short
(proximodistal dimension) so that many
lamellae are packed into a short phalan-
geal length (Figs. 5B, 9). It is the shape of

258

Figure 5. The subdigital surface of the fourth digit in A) Chamaelinorops (MCZ 156919) and B) A. occultus (MCZ ~
156968). In A only the distal 23 of 25 scales are shown. The pad corresponds to scales 8-13. The phalangeal
scale series are: 1-10 phalanx i, 11-13 phalanx ii, 14-17 phalanx iii, and 18-25 phalanx iv. In A. occultus (B) the
regional scale series are: 1-5 phalanx i, 6-22 phalanges ii and iii, and 23-30 phalanx iv. Scale 6 is a non-lamellar
pad scale. The morphology of scale 22 is transitional between that of lamellae and generalized scales.

EVOLUTION OF THE SUBDIGITAL PAD - Peterson

the scales, rather than particularly long
pad phalanges, that account for the high
proportion of pad to total subdigital
scales in most Anolis (cf. phalangeal and
scale ratios for non-anole iguanids and A.
cuvieri in Table 1). In Chamaelinorops
the lamellae are not much shorter than
the generalized subdigital scales and
only three or four pad scales occur in the
phalanx ii region. The width/length pro-
portion of Chamaelinorops lamellae is
lower than that of A. sheplani and A.
occultus lamellae (mean width/length for
the Chamaelinorops [MCZ _ 156919]
lamellae is 1.9 while that for A. sheplani
[MCZ 125641] is 5.0; the maximum for an
individual lamella in Chamaelinorops,
2.4, is much less than the minimum for A.
sheplani, 4.0). e) Lamellae are also dis-
tinguished from generalized subdigital
scales by the presence of a free margin
(=that portion of the scale which is distal
to the dermal core and consists largely of
the outer and inner Oberhautchen layers;
Lillywhite and Maderson, 1968). The
free margin is poorly developed in
Chamaelinorops. In A. sheplani and A.
occultus the free margin on phalanx ii
lamellae is 20-40 wm long and one-
quarter or more of the exposed length of
the scale is actually free margin (the limit
of the free margin is often indicated by
the deeper frays along the lamellar edge;
Figs. 4A, 5, 8, 9A, 10A). In Chamaelinor-
ops a free margin is present only over
the central portion of the scale. It extends
for 10 um proximal to the edge and
accounts for less than 5% of the lamellar
length. Scales 8 and 13 barely have a free
margin; the two layers of Oberhautchen
meet at an angle between 45 and 90° and
are not adpressed to form the typical
bilayer free margin. In both shape and
marginal characteristics, Chamaelino-
rops lamellae are convergent on gener-
alized subdigital scales.

The poor differentiation of lamellae
from generalized subdigital scales is just
one of the factors that contributes to poor
differentiation of regional scale series in
Chamaelinorops. In A. sheplani and A.

259

occultus the distal border of the pad is
clearly defined by an abrupt change in
scale shape and by a series of small scales
set deep between the phalanx i and ii
scales which “raise” the pad. In Chamae-
linorops these deep scales are absent and
the pad is coplanar with the distal pha-
lanx i scales (=the “Norops”’ condition).
Similarly there is no abrupt change in
scale morphology. The distal pad scale
(Number 8 in Fig. 5A) is intermediate in
shape between the adjacent lamella and
the phalanx i scale 7. The scales distal to
the pad have a “velvety” texture over
their central area similar to that of lamel-
lae, and scales 3 to 5 have very small free
margin (Fig. 7A). The central portion of
the phalanx i scales is convergent on the
lamellar scales. The lateral aspect of the
distal phalanx i scales is covered with
hillocks (Peterson and Williams, 1981)
and low keels. The scales under the third
phalanx (numbers 14-17 in Fig. 5A) and
those under the proximal portion of the
fourth phalanx (numbers 21-25) are very
similar. They are so narrow that the
lateral digital scales encroach on the sub-
digital surface. The distal phalanx iv
scales (numbers 18-21) are wider and
resemble the generalized subdigital
scales in the phalanx iv region of A.
sheplani and A. occultus.

While there are few points of similarity
between Chamaelinorops and_ A.
sheplani or A. occultus, the gross mor-
phology of the Chamaelinorops pad is
not entirely unusual among anoles. a)
The “unraised” pad occurs in a variety of
species groups (e.g., A. aequatorialis, A.
barkeri, and A. auratus in Table 1). A.
chrysolepis exhibits an intermediate
condition. b) Many anole and non-anole
species share the phalangeal proportions
of Chamaelinorops (e.g., A. chrysolepis,
A. cuvieri and Morunasaurus; Table 1).
Indeed A. sheplani and A. occultus have
unusually short toes with relatively long
second and third phalanges compared to
A. cuvieri and A. chrysolepis; it is
Chamaelinorops that appears to exhibit
the more generalized phalangeal and

260 Advances in Herpetology and Evolutionary Biology

Figure 6. The morphology of lamellae and setae in Chamaelinorops. A. Lamellae MCZ 156919. Note that the
lamellar margin appears thick and has very shallow frays (cf. Fig. 7A). B. Typical lamellar setae in MCZ 155851
(viewed from the sole of the foot looking toward the claw). Note the large rectangular tip and the absence of an
inflection in the stalk. C. Setae on generalized subdigital scale number 25 in MCZ 155851. The setae (also shown
in E and F) are unusual because they display a consistent tip flexion. The increase in stalk diameter which occurs
about midway along the stalk length marks the beginning of the oval stalk region. A few setae which are turned
(e.g., lower center) show the abrupt compression which forms the tip. D. Lamellar setae in MCZ 155851. These
setae exhibit considerable wear of the tip, but also show the irregular grooves in the stalk and relatively thicker tip
of Chamaelinorops. E. Lateral view of the setae on or scale 25 in MCZ 155851. Setae in profile show the irregular
stalk, thicker tip, and the abrupt compression of the oval region which forms the tip. Note that the inflection occurs
within the flattened tip region and that the distal margin of the tip is often flexed ventrally. F. Normal view of the
setae tips on scale 25 in MCZ 155851. Note the retangular tip shape, reduced density and thick, “rolled” distal
margin.

EVOLUTION OF THE SUBDIGITAL PAD - Peterson 261

Figure 7. Subdigital morphology in Chamaelinorops. A. Phalanx i scales 2 through 4 in MCZ 155851. The small
free margin on scale 3 is evident toward the lower portion of the figure. B. The distal margin of scale 4 at the
transition between the subdigital and lateral portions of the scale. The “seta” tips are small and the margin itself is
studded with large flattened spines. Note how adjacent spines are fused (MCZ 155851). C. The lateral portion of a
lamella and the adjacent lateral digital scale in MCZ 98691. The lateral digital scale is contoured by hillocks and
several small keels. The distal portion has many large irregular spines. D. Detail of the lateral digital scale showing
moderately large spines (lower left) and large diameter spines (top) on adjacent hillocks. Note the irregular,
apparently fused spine at the top right (MCZ 98691). E. Proximal and lateral portion of a lamella in MCZ 155851.
Spines to the right have a 0.5/sq mm spacing and appear to form the base of the Chamaelinorops spine-seta
series. Spines to the top left and lower center have a similar spacing, but a much larger diameter. They may form
the base of the large spine series. Spines similar to those shown in the lower left have a 1/sq mm density and
resemble Anolis spines; they may form the “top” of the spinule-spine series in Chamaelinorops. F. Adjacent
surfaces of a phalanx i and lateral digital scale in MCZ 155851. The cleft between the scales and on a hillock just
above the figure scale are covered with spinules and small spines. The adjacent epidermal cells exhibit a variety
of large spines. Note in the upper left the large, irregular, apparently fused spines.

262

digital proportions. c) The Chamaelino-
rops pattern of lamellar distribution is
approached by a number of anole
species. In A. chrysolepis and A. tro-
pidonotus lamellae are absent in the pha-
lanx iii region. In A. auratus only the dis-
tal two phalanx iii scales are lamellae (the
West Indian grass anole, A. pulchellus,
exhibits a less extreme reduction in
lamellae). d) Pad area in proportion to
body size is similar in A. auratus
and Chamaelinorops. For A. auratus
(LACMNH 72741; SVL 54 mm) pad area
is 1.2 sq mm. The scaling equation de-
scribed above would predict 2.0 mm from
the body dimensions (167% of the
observed value). A. chrysolepis and A.
tropidonotus have only slightly larger
pads in proportion to their size. e) A
similar set of conditions contribute to the
small pad of A. auratus and Chamaelino-
rops. Pad length is reduced because of
the restricted distribution of lamellae,
and the lamellae themselves are longer
and narrower than in A. occultus and A.
sheplani. The mean width/length propor-
tion of the A. auratus lamellae is 2.5 with
a maximum value of 3.3 (cf. A. sheplani
above). Pad length relative to digital
length or body length is similar among
Chamaelinorops, A. auratus, A. chryso-
lepis, and A. tropidonotus (Table 1). f)
The proportion of lamellar to total sub-
digital scales is similar among A. auratus,
A. chrysolepis, A. tropidonotus, and
Chamaelinorops (Table 1). The restrict-
ed distribution of lamellae as well as the
greater length of the individual lamellar
scales contributes to the condition in
each case. g) A. auratus has relatively
poor regional differentiation of the scale
series (Peterson and Williams, 1981).
Lamellae grade into generalized subdigi-
tal scales in the phalanx iii and iv regions.
The distal phalanx i scales are remark-
ably similar to those of Chamaelinorops.
Low keels occur over the lateral portion
of the scale while the central portion has
a “velvet-like” texture (Peterson and
Williams, 1981).

Advances in Herpetology and Evolutionary Biology

Fine Structure

Seta Shape and Dimensions

The “setae” in Chamaelinorops are not
setae as defined by A. sheplani, A. occul-
tus (Fig. 8B, C) or any other anole
examined to date. The seta tip is rectan-
gular (Fig. 6), as is the anole seta-prong
morphotype (Figs. 2, 4E, F), but the
shape and dimensions of the Chamae-
linorops setae are quite distinct from
those of the anole seta-prong or seta.

a) The subdigital spine-seta morpho-
typic series in Anolis appears to be based
on a fairly constant stalk density—
slightly greater than 1/sq wm (values for
individual specimens in this study range
from 0.9-1.4 stalks/sq wm; Table 2;
Peterson et al., 1982 report that there is no
significant difference in stalk density
among 6 Puerto Rican ecomorphs includ-
ing A. occultus and A. cuvieri). Setal stalk
density in Chamaelinorops also appears
to be relatively constant, but at 0.5 stalks/
sq pm, it is roughly half the value for the
anole species or Chamaeleolis (Table 2).

b) The diameter of the Chamaelinor-
ops seta stalks is not different from that
of A. cuvieri or A. occultus, but stalk
height is less than that of any anole setae
examined to date (Table 2; Peterson and
Williams, 1981). The stalks also have a
noticeably irregular surface (Fig. 6B, D,
18),

c) A. sheplani and A. occultus are
typical of anoles in the shape and size of
the tip and in the presence of well-
defined stalk and tip regions (Fig. 8B, C
and Table 2). In Chamaelinorops the
stalk and tip regions are less well defined
by shape change or by flexion of the stalk,
and the tip itself forms in a different
fashion.

In Chamaelinorops and in anoles such
as A. occultus, A. sheplani, and A.
cuvieri, the flattened tip lies just distal to

a portion of the stalk that is oval in cross”

section. In Chamaelinorops this oval
section is 2.0 to 2.8 um long (roughly one-

EVOLUTION OF THE SUBDIGITAL PAD - Peterson 263

~
+.

Figure 8. Subdigital morphology in A. sheplani and A. occultus. A. The surface of midpad lamellae (numbers 5
and 6) in A. sheplani (MCZ 125641). B. Lateral view of the setae on the free margin of lamella number 5 in A.
sheplani (MCZ 125641). C. Setae tips at the proximal portion of the lamellar free margin in A. sheplani (MCZ
125641). D. Setae-prongs on the most proximal digital scale (number 30) in A. occultus (MCZ 156968). E.
Oblique view of the spines adjacent to the distal margin of phalanx i scale 4 in A. occultus (MCZ 156968). F.
Spinules over the proximal portion of phalanx i scale 2 in A. occultus (MCZ 156968). Note the presence of
nubbinlike and spinelike spinules.

264

third of the stalk length); in the anoles it
is 1.0 to 1.5 wm long (or 5—-10% of the stalk
length). The major axis of the oval cross
section corresponds to the proximo-distal
axis of the digit, and this orientation
appears to establish the preferred axis for
flexion of the tip on the stalk. In the
anoles the junction between the round
and oval stalk sections is frequently the
site of a small inflection which tilts the
tip toward the metatarsal—phalangeal
joint. The distalmost section of the oval
stalk forms the apex of the triangular tip
and is the site of a second more frequent
and larger inflection that tilts the seta tip
30 to 70° and points the distal margin
proximally toward the metatarsal-phalan-
geal joint (stalks in the upper right of
Figure 8B are relatively undisturbed and
show this. orientation clearly). In
Chamaelinorops the stalk, including the
oval region, is usually straight and
normal to the scale surface (Fig. 6B, D).
Most of the Chamaelinorops setae have
no inflection of the tip on the stalk; the
flat tip of most lamellar setae is oriented
normal to the scale surface and locomotor
substrate. In a few regions where there is
a consistent tip flexion, the point of
flexion lies within the tip region itself,
just distal to where the tip differentiates
from the stalk (Fig. 6). The inflection tilts
the tip toward the metatarsal-phalangeal
joint as in Anolis or Chamaeleolis. There
is some taper or reduction in the cross
sectional area of the stalk between the
round and oval regions and over the
length of the oval region in A. cuvieri. In
Chamaelinorops the stalk shape changes
slightly, but it is not tapered proximal to
the tip.

In the anole species the triangular tip
appears to be formed as the oval stalk
gradually flattens over the 0.6 wm length
of the tip. Relatively gradual compres-
sion of the oval cross section produces a
triangular surface 0.6—0.8 wm wide (2-3X
wider than the distalmost section of the
oval). Often it appears that compression
of the tip begins in the center and
spreads laterally so that in the proximal

Advances in Herpetology and Evolutionary Biology

portion of the tip the lateral margin is
thicker.

In Chamaelinorops a tip forms by a
more abrupt decrease in the minor axis of
the oval stalk (from 0.5-0.6 um to 0.2-0.3
fm) coupled with a variable increase in
the major axis (from 0.6-0.7 um to 1.0-0.7
pm) (Fig. 6D, E). The amount of material
in a cross section through the proximal
portion of the tip is very close to that
present in any cross section of the stalk.
More distally the tip tapers to 0.15 to 0.25
ym thickness (the minor axis of the oval),
and in three of the five specimens the tip
narrows to 0.5 to 0.6 um (or even slightly
less at the very distal margin; see the
paragraphs at the end of this section for
discussion of the individual variation in
Chamaelinorops specimens). The length
of the tip is subject to obvious wear varia-
tion, but mean length for the five speci-
mens ranges from 1.0 to 1.5 wm. The
Chamaelinorops seta tip is approximate-
ly twice as long and 2 to 3 times thicker
than that in the anole species examined.
The average tip width is greater than that
in anoles, but the width at the distal
margin is less. The distal margin of the
seta in Chamaelinorops is formed by a
chipped, uneven edge (Fig. 6B, D) or bya
ventrally curled “lip” (Fig. 6C, E, F).

There are a number of very clear dif-
ferences between the seta tip in general-
ized anoles and Chamaelinorops. In
Chamaelinorops, 1) there is no taper or
reduction in the cross sectional area of
the stalk proximal to the tip; 2) the tip
appears to form by a more abrupt com-
pression of the oval stalk; 3) this pro-
duces a rectangular or a trapezoidal tip
surface (rather than a triangular one),
4) which is 3 to 6 times larger than that of
the anoles; 5) the tip is substantially
wider, longer and thicker, but the distal
margin is narrower; 6) the tip is usually
oriented parallel to the seta stalk, but
may have an orientation similar to that of

anoles; 7) if an inflection is present, it

occurs within the tip length not in the
stalk region; and 8) lastly, the tip and oval
stalk section make up about 50% of the

EVOLUTION OF THE SUBDIGITAL PAD - Peterson

seta height in Chamaelinorops (e.g., 4
wm in MCZ 156919) and no more than
10% of the height of lamellar setae in
anoles.

d) The distinction between Chamae-
linorops setae and anole setae (as defined
by A. sheplani, A. occultus, and A.
cuvieri) is not significantly altered if
Chamaelinorops is compared to A.
auratus and A. tropidonotus. A. chryso-
lepis has a somewhat wider seta tip and a
larger stalk diameter than other anoles
examined, but it does not resemble
Chamaelinorops (e.g., the larger tip is
distinctly triangular; Peterson and
Williams, 1981).

The only Anolis species examined to
date which has setae that remotely
resemble those in Chamaelinorops is an
undescribed South American form in the
alpha section of the genus (A. sp. n. near
eulaemus LACMNH 72741). The animal
is reported to be riparian and semi-
aquatic (Williams, personal communica-
tion). The gross morphology of the pad,
the fine structure of the setae stalks, and
the position of the tip inflection are simi-
lar to those in other anole species (Table
2). However, the setae tips are rectangular
or oval in outline and have areas almost as
large as those in Chamaelinorops (Table
2; Fig. 10C, D illustrate the rectangular
tip). In both A. sp. n. near eulaemus and
Chamaelinorops the specialized size and
shape of the setae are correlated with in-
creased tip length (e.g., 0.88 + 0.04 um in
A. sp. n. near eulaemus), and with the
morphology of the tip/stalk junction. In
both forms the distalmost stalk section
(apex) is much wider than in the anoles
(0.60 + 0.03 wm wide in A. sp. n. near
eulaemus and 0.6-0.7 wm wide in
Chamaelinorops compared to 0.24 wm
in most anoles). Subsequent flattening of
the apex to form the tip produces a thick
tip (0.1-0.2 wm) which is only slightly
wider (0.75 um) than the distal stalk (0.6
fm). Compression of a wider, less tap-
ered stalk appears to give the tip its
rectangular shape and greater average
_ width. Setae on the pad in A. sp. n. near

265

eulaemus exhibit a gradual flattening to
form the tip as in other anoles, but the
seta (or seta-prong) tips on the phalanx iv
scales are formed from the stalk more
abruptly and exhibit the rectangular
shape of Chamaelinorops setae (Fig.
10C, D). These setae or setae-prongs dif-
fer from Chamaelinorops setae in tip
length (0.72 + 0.06 um compared to 1.1-
1.5 mm in Chamaelinorops), tip area
(0.488 + 0.055 sq wm compared to 0.6—1.4
sq wm in Chamaelinorops), the position of
the tip flexion, stalk dimensions, and stalk
spacing (the latter features are not differ-
ent from those of the lamellar setae).

e) The anole seta and seta-prong mor-
photypes generally differ in some of the
same features as Chamaelinorops setae
and anole setae, and it is possible that the
Chamaelinorops setae should be com-
pared with anole setae-prongs. The major
points of similarity include: 1) the seta-
prong tip surface usually appears to be
thicker and formed by a more abrupt flat-
tening of the stalk, and 2) the apex of the
stalk is wider than that of the seta. These
differences appear to be responsible for
the rectangular tip shape. Further paral-
lels include reduced stalk height and a
less well-defined point of flexion of the
tip on the stalk (a tip flexion can also be
absent; Fig. 8D). However, the tip di-
mensions and stalk spacing of setae-
prongs in any of the anole species
examined are clearly different from
those of Chamaelinorops “setae.” For
example, in A. occultus the poorly differ-
entiated setae (or setae-prongs, the dis-
tinction is not clear in this area) at the
metatarsal-phalangeal joint (Fig. 8D) are
4.4 + 0.3 wm tall and 0.44 + 0.03 um in
diameter with 1.2 stalks/sq wm, and the
tips have an area of 0.285 + 0.025 sq um.
The tip dimensions are 0.63 + 0.06 wm
long by 0.37 + 0.03 wm wide proximally
and 0.45 + 0.02 wm wide distally. In A.
cuvieri the seta-prong tips in the pha-
lanx iv region (Fig. 4E) have somewhat
larger dimensions than those in A. occul-
tus, but the proportions are the same
(0.77 + 0.04 wm long by 0.60 + 0.03 wm

266

wide distally and 0.48 + 0.03 um wide
proximally). In Chamaeleolis the seta-
prongs shown in Figure 4F on the lateral
digital scales adjacent to the pad are
almost perfectly rectangular (0.60 + 0.06
pum long by 0.44 + 0.03 um wide distally
and 0.40 + 0.03 wm wide proximally.)

Chamaelinorops appears to have a
unique setal morphology. Comparisons
beyond those with A. occultus, A.
sheplani, and A. cuvieri reveal no paral-
lels for the reduced stalk spacing in
Chamaelinorops nor for the combination
of shape and size features. Setae prongs
and poorly differentiated setae in a num-
ber of anolines approach the tip shape of
Chamaelinorops setae, but most of the
tip dimensions are significantly smaller
than those of Chamaelinorops. The setae
of A. sp. n. near eulaemus approach the
tip dimensions and some of the shape
features of Chamaelinorops tips, but the
resemblance is far from compelling. The
setae of A. sp. n. near eulaemus have
more in common with other anole setae
than with Chamaelinorops setae.

f) There is significant individual varia-
tion in tip area among the five Chamae-
linorops specimens (Table 2; Ftip area =
82.1; p < 0.005). The variation among the
specimens is much greater than the level
of apparent “wear” variation in A. occul-
tus (Table 2).

The specimen with the greatest tip
area (MCZ 156919) does have what
appear to be “unworn” setae, but the tips
are also about twice as wide as the tips in
three of the other specimens (0.97 + 0.05
zm in MCZ 156919 compared to 0.58 +
0.03 um MCZ 126708, 0.55 + 0.03 wm
MCZ 74594, and 0.52 + 0.03 um MCZ
68691). The fourth specimen (MCZ
155851) exhibits tips of an intermediate
width (0.71 + 0.04 um). The wider tip in
MCZ 155851 and MCZ 156919 occurs all
over the lamellar surface; there is no sug-
gestion that it is a regional or local
variant. Similarly, there is no indication
that the narrower tips in the other three
specimens are “worn” versions of the
larger ones. There are very few scars,

Advances in Herpetology and Evolutionary Biology

gouges or broken surfaces along the
lateral edge of the tip in any specimen.
MCZ 126708 exhibits a few of the wide
tips scattered among the narrow ones.
The width of the seta tip appears to be
subject to significant individual variation
(F=149.8; p < 0.005).

Variation among the Chamaelinorops
specimens in tip length is also significant
(F=10.5; p < 0.005), but the range is
comparable to that of probable wear
variation in A. occultus (Table 2).
Chamaelinorops tip length varies from
1.09 + 0.04 wm in MCZ 74594 to 1.52 +
0.03 wm in MCZ 156919 with interme-
diate values of 1.44 + 0.13 wm in MCZ
156851, 1.31 + 0.10 um in MCZ 126708
and 1.25 + 0.08 um in MCZ 68691. The
specimen with the shortest setae tips
(MCZ 74594) exhibits considerable
debris and variation in the outline of the
distal edge, probably due to chipping and
breakage. Some tips in all the areas are
obviously broken off the stalks (these
were not measured, but they suggest that
the Oberhautchen is old or highly worn).
MCZ 68691 with the second shortest tips
also has more tips with ragged distal
edges than the remaining three speci-
mens. There is no clear evidence of a dif-
ference in wear among the remaining
specimens, and variation in tip length
among them is marginally significant
(F=4.0; 0.05 > p > 0.025).

The Distribution of Setae

In A. sheplani setae cover all the lamel-
lar scales and extend as far proximally as
the metatarsal-phalangeal joint. In A.
occultus setae cover the lamellae and
some of the phalanx iv scales. The pha-
lanx iv scales exhibit sequential steps in
the morphotypic series (Fig. 2). Setae
occur on the distal scales (numbers 22
through 26). The seta tip becomes pro-
gressively narrower on each more prox-
imal scale, and a progressively smaller
central area of the scale bears setae. The
surrounding areas are covered by seta-
prong and spike morphotypes. Setae are

EVOLUTION OF THE SUBDIGITAL PAD - Peterson

absent on scales 27 through 30, and setae-
prongs occur over the central region of
the scales (Fig. 8D). Prongs and spikes
occur more laterally on these scales.

In Chamaelinorops setae occur on all
the subdigital scales. The setae which
cover the generalized subdigital scales
throughout the phalanx iii and iv regions
are somewhat shorter, but otherwise not
different from lamellar setae. For ex-
ample, the setae on the most proximal
subdigital scale (number 25) in MCZ
155851 are 5.6 + 0.3 um tall and the
stalks are 0.58 + 0.06 wm in diameter (cf.
Table 2; Fig. 6C, E, F). The tips are
flexed within the flattened tip region so
that the entire tip area can not be
measured accurately, but the tips are
about 1.5 um long and 0.7 um wide.

Shorter setae similar to those on the
generalized subdigital scales also occur
over the central portion of the distal pha-
lanx i scales (e.g., on scale 4 in MCZ
155851 the stalks are 5.6 + 0.7 um tall).
Laterally the setae grade into other mem-
bers of the morphotypic series (see pha-
lanx i description following).

Although setal distribution in Chamae-
linorops differs from that of A. sheplani
or A. occultus, A. auratus exhibits the
Chamaelinorops pattern. Setae occur on
all the subdigital scales, and those on the
phalanx i scales occur only in the central
portion of the scale (Peterson and
Williams, 1981).

The Phalanx i Region

In A. occultus and A. sheplani the five
small phalanx i scales are covered with
spines and spinules. Distally the surface
is covered with spines which have sharp,
curved tips (Fig. 8E). The proximal half
of the scale is spinulate (Fig. 8F). Scale 5
is entirely covered with spinules. There
are two populations of spinules: short,
nubbinlike spinules 0.19 + 0.02 um tall
and 0.16 + 0.03 wm in diameter, and
taller, spinelike spinules 0.52 + 0.14 um
tall and 0.12 + 0.03 wm in diameter
(actual dimensions for A. sheplani MCZ

267

125641). Together they occur in a density
of 7.4 spinules/sq wm. A. sheplani exhib-
its similar spinules in a density of 8.4
spinules/sq pm.

In Chamaelinorops the distal phalanx i
scales (numbers 1-7 in Fig. 5A) exhibit a
wide range of fine structure. The central
region of the scale is covered with setae.
Setae extend distally onto a very small
free margin on several of the scales (e.g.,
scales 2 and 3 in Fig. 7A). The seta-
bearing area corresponds to the func-
tionally subdigital portion of the scale.
The lateral portions of the scales face
laterally, not ventrally, and the morphol-
ogy of the surface is very similar to that of
the adjacent lateral digital scales. In the
transition from the subdigital to the
lateral surfaces, the distal margin of the
subdigital scale becomes studded with
large diameter spines (Fig. 7B), and the
seta tips become smaller. More laterally
the scale surface becomes contoured into
“hillocks’—the central portion of each
epidermal cell is elevated while the
intercellular junction is depressed
(Peterson and Williams, 1981; Fig. 7C, D,
F). The hillocks create a surface rough-
ness of intermediate magnitude. The
surface of each hillock is covered with
spines which are fairly similar in size, but
adjacent hillocks have spines of very dif-
ferent dimensions (Fig. 7D and text fol-
lowing). True keels occur along the
lateral portion of several of the subdigital
scales (e.g., scales 2, 5, 6, and 7 of MCZ
156919) and on the lateral digital scales
(Fig. 7C).

Chamaelinorops exhibits extraordinary
diversity in the morphology of spines on
the phalanx i subdigital scales and the
lateral digital scales. The diversity in
spines is based on variation in diameter
and density which, in the extreme,
results in the fusion of spines from ad-
jacent epidermal cells. Figure 7 illus-
trates some of this diversity. Small spines
with an understory of spinules occur
toward the sides of the scales and in the
deep folds between the scales (Fig. 7F
lower left and center). The small spines

268

are 0.88 + 0.09 um tall and 0.21 + 0.08 um
in diameter near their base; the spinules
are 0.30 + 0.09 um tall and 0.17 + 0.05
um in diameter. Together they occur in a
density of 34 spines and spinules/sq wm.
The medium-sized spines shown in
Figure 7D bottom are similar to those in
Anolis (1.51 + 0.31 pm tall and 0.48 +
0.03 wm in diameter with about 1.0-1.5
spines/sq wm). Large pyramidal spines
are the most common (Fig. 7D top left
and Fig. 7F center; those in Fig. 7F are
2.12 + 0.20 um tall and 1.66 + 0.20 um in
diameter and occur in a density of about
0.2 spines/sq wm). Even larger, irregular
spines (upper right Fig. 7B, C and upper
left of Fig. 7F) appear to represent a
fusion of several single pyramidal spines.
Some fused or conglomerate spines
assume incredible proportions and cover
more than 100 sq um of the scale surface.
Partially fused large spines form serrate
ridges stretching for more than 50 u par-
allel to the keels on the lateral scales.
The spacing of the conglomerate spines
is highly irregular and extensive aspi-
nate, bare areas occur around an indivi-
dual spine (density less than 0.01
conglomerate spines/sq wm). The very
large conglomerate spines occur on the
distolateral corners of the subdigital
scales in the phalanx i region and on the
lateral scale surfaces distally and ad-
jacent to the keels (Fig. 7C).

A. sheplani and A. occultus exhibit no
counterpart for the single and conglom-
erate spines in Chamaelinorops. Spines
in the anoles have a maximum diameter
of about 0.5 um, one-third or less of the
maximum spine diameter in Chamae-
linorops, and the anole spines occur in a
density of about 1/sq wm. The Chamae-
linorops spine series begins and ends at
lower densities (3-4/sq wm to 0.01/sq
ym) than the spinule-to-spine series in
the anoles (e.g., 7.4/sq wm to 1/sq wm in
A. sheplani). No other anoline examined
to date has fused, conglomerate spines.
But the mosaic landscape formed by dif-
ferent spine types and the variation in
single spine diameter and density which

Advances in Herpetology and Evolutionary Biology

is exhibited by Chamaelinorops does
have some parallel in a number of Anolis
species, including A. cuvieri and A.
auratus. These parallel morphologies
occur only on the lateral digital scales. In
A. auratus there are scattered single
epidermal cells covered with large
pyramidal spines (about 4 um tall; 2.4 wm
in diameter near the base and 0.1 spine/
sq um; cf. 1.9 um tall, 1.8 wm in diameter
and 0.2 spines/sq wm in Chamaelinor-
ops). The lateral digital scales in A.
cuvieri exhibit a mosaic arrangement of
spine types (Fig. 4C, D). The extremes in
diameter and spacing are: spines 1.53 +
0.15 um tall, 0.75 + 0.04 wm in diameter in
a density of 0.23 spines/sq wm and spines
0.72 + 0.04 um tall, 0.32 + 0.02 um in di-
ameter in a density of 3.6—5.0 spines/sq
wm. The variation in diameter is more
conservative than that in Chamaelinor-
ops, and there is no contouring of the

surface by hillocks and keels.

The Morphotypic Series in Fine Struc-
ture

The varieties of fine structure in A.
occultus and A. sheplani are adequately
described by Figure 2. Stalk diameter
may be somewhat smaller in A. sheplani,
and the stalk height of spines, prongs,
setae-prongs and setae is toward the low
end of the ranges given in Figure 2 for
both species. All the morphotypes shown
in Figure 2, except small spines and
spinules which occur on the phalanx i
scales, occur in the proximal area of a
lamella, and there are no novel mor-
photypes.

Chamaelinorops does not exhibit the
anole spine-seta morphotypic series.
There is, however, a parallel, alternative
series. Along a transect from distal to
proximal on a lamella the Chamaelinor-
ops setae become shorter and the tip
becomes less compressed leaving a
pronglike form. The prong is generally
succeeded by a tall spine (2.5-3 pm tall)
with a tapered, hook tip (Fig. 7E right).
The series is similar to the spine-seta

EVOLUTION OF THE SUBDIGITAL PAD - Peterson

series in Anolis in that diameter and
density remain constant as tip shape and
stalk height change. The two series differ
in stalk density (about 1/sq wm in Anolis
and 0.5/sq wm in Chamaelinorops). The
absence of an Anolis-like seta-prong and
spike may be only a formalism. The
Chamaelinorops seta itself is similar to
an anole seta-prong, and variation in stalk
height is much reduced in Chamaelinor-
ops so that the transition to a hooklike tip
simply occurs below the 5 um limit of the
spike. The Chamaelinorops spine-seta
series exhibits continuous variation in
stalk height and tip shape. It is a simpler
series than that of Anolis, but there are no
“gaps or “missing” morphotypes. Even
more proximally on the lamella, density
increases and a small band or scattered
patches of anole-like spines can be dis-
tinguished (0.4—0.5 um diameter and 0.9
spines/sq wm; Fig. 7E lower left). These
grade into regions of small spines about |
pm tall) with an understory of very small,
nubbinlike spinules (together in a den-
sity of 2.4/sq wm). The fields of small
spines and spinules are linked to the
Anolis-type spine by continuous varia-
tion in diameter and density. Fields with
larger “small spines’—those approach-
ing 1 um height and 0.5 wm diameter—
have a very reduced understory of small
spinules. The 1 spine/sq wm density and
the Anolis-like spine appear to be
achieved as the last spinules disappear
from the understory (e.g., the area cited
above exhibits 1.2 small spines/sq wm;
the spinules effectively double the den-
sity in this region). The fields of small
spines and spinules in Chamaelinorops
occur in slightly lower densities than
those of most anole species (e.g., the
maximum density in Chamaelinorops is
4/sq wm, while that in A. sheplani is 7.4/
sq pm), but the shape and size of the
spines and spinules are similar. The
morphotypic series formed between the
spinule and Anolis-like spine is very
similar to that present in the Anolis
species.

A third, larger spine variety occurs in

269

patches along the proximodistal borders
of the lamella (Fig. 7E left). The dimen-
sions (2.62 + 0.42 um tall and 1.46 + 0.09
jm diameter) are similar to those of the
large single spines on the phalanx i and
lateral digital scales, but the lamellar
spines are much more closely packed
(0.5-0.6 spines/sq wm) with almost no
bare surface between the spines. These
large spines resemble the Chamaelinor-
ops spine that forms the base of the
spine-seta series in density and in height,
but they differ in diameter.

The three different spine types—a) the
Anolis-like spine which is continuous
with the spinulate series; b) the large
diameter spine which may be linked to
the large spines in the phalanx i region
and on the lateral digital scales; and c)
the 0.5 um diameter-0.5/sq wm spine
which forms the base of the Chamae-
linorops spine-seta morphotypic series—
are topographically contiguous (Fig. 7E).
There are no epidermal cells bearing
spines with intermediate diameters or
densities.

COMPARISONS BETWEEN
PHENACOSAURUS HETERODERMUS AND
ANOLIS VALENCIENNI

Pad and Scale Morphology

The major differences between the two
forms are in the position of the pad and
the shape of the lamellae.

In A. valencienni lamellae occur under
phalanges ii and iii. The phalanx iv scales
have a lamellar shape in some cases, but
a free margin is absent. The 26 lamellae
(58-60% of the total scales) are packed
into 3.4 to 3.8 wm of the digit (49-45% of
the digital length; Table 1).

In Phenacosaurus phalanges ii and iii
are shorter relative to the length of the
body or digit, and lamellae in this region
account for only about 42% of the sub-
digital scales (Table 1). The distribution
of lamellae is expanded, however, to in-
clude the entire phalanx iv region (scales
27-34 in Fig. 9A) and the proximal pha-

270

lanx i region (scales 6-12 in Fig. 9A;
Lazell, 1969, describes phalanx i lamel-
lae in all Phenacosaurus species). Figure
9C shows the “free” lamellar margin of
the scales just distal to the metatarsal-
phalangeal joint. This expansion of
lamellar scales more than doubles the
length of the pad and doubles the num-
ber of lamellae (better than 80% of the
subdigital scales are lamellae; Table 1).
Although the pad is roughly twice as long
as that in A. valencienni, the number of
lamellar scales and pad area are not
markedly greater in Phenacosaurus.
There are 25 actual lamellae in A. valen-
cienni and 24 to 31 in Phenacosaurus (the
first pad scale is non-lamellar). Estimated
pad area is 4.6 sq mm in A. valencienni
(P 329; 58.5 mm SV length) and 4.3 sq
mm in Phenacosaurus (P 300; 56 mm SV
length). The similarity in these values in
spite of the differences in pad length and
lamellar distribution is related to the dif-
ference in lamellar shape. In A. valen-
cienni the phalanx ii lamellae are very
wide and short (mean width/length
=11.3). Approximately two-thirds of the
pad area (2.9 sq mm) and lamellar scales
(17 lamellae) are concentrated under the
second phalanx. In Phenacosaurus there
is less variation in lamellar proportions,
and the lamellae are narrower and longer
than those of A. valencienni (the mean
width/length ratio for all lamellae is 5.5
in Phenacosaurus and 9.0 in A. valen-
cienni; the standard deviations are 2.0
and 4.7, respectively). As a consequence
of the difference in lamellar shape and
distribution, pad area in Phenacosaurus
is more evenly distributed among the
phalanges (approximately 15% under i,
30% under ii, 20% under iii, and 35%
under iv).

Expansion of lamellae into the phalanx
iv region is unusual among West Indian
anoles, but does occur in Anolis. A.
aequatorialis, a South American alpha
anole with a narrow pad, has lamellae
extending through the proximal half or
two-thirds of the phalanx iv region. This
is the most proximal distribution of

Advances in Herpetology and Evolutionary Biology

lamellae among the anole _ species
examined to date, and Phenacosaurus
may be unique in the presence of lamel-
lae throughout the phalanx iv region.

The expansion of lamellae into the
phalanx i region is unique among anole
species with a raised pad, but unlike the
proximal expansion of the pad, the pha-
lanx i lamellae do not increase the area of
the pad. In A. valencienni, as in many
anoles with a well developed, wide pad,
the distal end of the pad is a flap or shelf
which is attached to phalanx ii and the i/ii
interphalangeal joint, but projects far dis-
tally to lie under phalanx i. In A. valen-
cienni the distal 7 to 9 lamellae and an
estimated 0.65 sq mm of pad area (14% of
the total area) lie on this shelf (measure-
ments for P 329). In figure 9B the i/ii
interphalangeal joint is flexed, and the
portion of the pad which lies under pha-
lanx i, scales 9 to 16, is tilted toward the
viewer. The phalanx i pad area in Phena-
cosaurus (0.64 sq mm; 15% of the total
area; measurements for P 300) is no
greater than that of the pad “shelf” in A.
valencienni. The pad is simply attached
more distally in Phenacosaurus.

Fine Structure

There are no major differences in the
fine structures of Phenacosaurus and A.
valencienni.

The shape and dimensions of the setae
are not different (Table 2; Fig. 10A, B). In
A. valencienni, setae occur on the lamel-
lae under the ii and iii phalanges and
cover the generalized subdigital scales in
the phalanx iv region. In Phenacosaurus
setae occur on all the lamellae (through-
out the ii, iii, iv and proximal i phalangeal
regions).

The phalanx i scales are covered with
spinules and small spines in A. valen-
cienni. In Phenacosaurus the distal
quarter of the scale surface is covered by
prongs and spikes which grade into
spines and finally, small spines (1.27 +
0.12 wm tall, 0.28 + 0.03 wm diameter)
and spinules (0.33 + 0.10 um tall, .21 +

EVOLUTION OF THE SUBDIGITAL PAD - Peterson alt

Figure 9. The subdigital surface of the fourth digit in A) Phenacosaurus (P 300) and B) A. valencienni (P 369). In
Phenacosaurus scales 7 through 12 are lamellae attached to phalanx i. Lamellae 13-26 occur under the second
and third phalanges, and lamellae 27-34 occur under the fourth phalanx. The free margin of the most proximal
subdigital scale (number 34) is shown in C). In A. valencienni scales 17-34 are lamellae attached to phalanges ii
and iii. Scales 9 (or 10) through 16 are lamellae which are attached to the second phalanx or the i/ii inter-
phalangeal joint region, but lie under phalanx i. The i/ii interphalangeal joint (between scales 16 and 17) is flexed
toward the viewer. The ii/iii interphalangeal joint (under scales 27 and 26) is hyperextended.

272 Advances in Herpetology and Evolutionary Biology

*

a ‘ ~~ 4

'

Figure 10. The subdigital fine structures of Phenacosaurus and A. sp. n. near eulaemus A. Lateral view of the
setae on the lamella free margin in Phenacosaurus (P 300). B. Setae tips of Phenacosaurus (MCZ 139566). C.
The setae-prongs of A. sp. n. near eulaemus (LACMNH 42144) on phalanx iv scale 41. D. Detail of the seta-prong
tips in A. sp. n. near eulaemus.

0.05 wm diameter; density of small spines to those which occur proximally on
and spinules = 2.9/sq wm) over the prox- lamellae in A. valencienni and A. cuvieri.
imal three-quarters of the scale surface. The spine-seta morphotypic series in
The prongs, spikes and spines are similar Phenacosaurus is in no way different

EVOLUTION OF THE SUBDIGITAL PAD - Peterson

from that of A. valencienni, A. cuvieri, or
most of the generalized anole species
examined to date.

DISCUSSION

SUMMARY AND PRELIMINARY
EVALUATION OF THE COMPARISONS

Chamaeleolis and A. cuvieri

Chamaeleolis and A. cuvieri share all
the major features of the pad complex.
The morphology of the setae and that of
other fine structural features of the scales
are very similar in the two forms, and the
gross morphology of the pad is virtually
identical.

The features that distinguish Chamae-
leolis from A. cuvieri are: a) the phalanx i
scales have spinules and small spines in
Chamaeleolis and typical subdigital
spines in A. cuvieri; b) setae occur on all
the phalanx iv scales in Chamaeleolis;
c) the seta stalk is taller in C. chamae-
leontides, but not in C. porcus; d) a dis-
tinct subdigital scale row is not well
established in the phalanx iv region; and
e) the toe and pad phalanges are abso-
lutely shorter in Chamaeleolis.

The difference in stalk height between
C. chamaeleontides and A. cuvieri may
be functionally significant, but a number
of anole species (e.g., A. stratulus and A.
evermanni among the Puerto Rican
ecomorphs; Peterson and Williams, un-
published data) exhibit seta stalk propor-
tions similar to those of C. chamae-
leontides. The difference does not
distinguish Chamaeleolis from Anolis,
and it probably is not indicative of a
phylogenetically distinct seta type.

Similarly, the presence of setae on all
the phalanx iv scales is a common differ-
ence among anole species. Even
ecomorph pairs may differ in this feature
(e.g., A. occultus and A. sheplani, and A.
pulchellus and A. auratus; Peterson and
Williams, 1981 and unpublished data).
The expanded distribution of setae in

273

Chamaeleolis may be correlated with the
short digit and pad phalanges. As one
might expect, the number of setae is
highly correlated with body size (essen-
tially with the load presented to the
setae; Peterson et al., 1982). One would
further expect Chamaeleolis and A.
cuvieri to have comparable numbers of
setae. Given that seta density and pad
width are similar in the two forms, the
shorter pad in Chamaeleolis would result
in fewer setae than in A. cuvieri, but for
the expansion of setae over the proximal
phalanx iv scales. The seta-bearing area
is not precisely equal in the two forms (cf.
body size in Table 1), but both are within
10% of the area predicted for their body
size by a scaling relationship developed
for the Puerto Rican ecomorphs
(Peterson et al., 1982).

The presence of extensive spinulate
areas on the subdigital surface and the
absence of a distinct subdigital scale row
in the phalanx iv region do not distin-
guish Chamaeleolis from Anolis, but
these features may be truly primitive
characters. Multiple small scales on the
proximal subdigital surface occur in
many of the preanoline iguanids (e.g.,
Diplolaemus bibroni and Pristodactylus
achalensis; Peterson, unpublished data)
and in several Anolis species that are
considered relatively primitive by other
measures (e.g., A. pentaprion and some
specimens of A. sheplani and A. occultus;
Schwartz, 1974; Williams, personal com-
munication). This condition has not been
observed in more derived anole species,
such as A. cristatellus or A. tropidonotus.
Similarly, small, high density spines or
spinules occur on the subdigital scales of
preanoline iguanids such as Diplolaemus
bibroni and Pristodactylus achalensis,
and on the phalanx i scales of some primi-
tive anole species, such as A. occultus
and A. sheplani. More derived anole
species, such as A. cristatellus and A.
tropidonotus, exhibit spines, spikes or
some combination of morphotypes with-
in the spine-seta portion of the mor-
photypic series. The spinule morphology

274

in Chamaeleolis is not significantly dif-
ferent from that in A. sheplani or A.
occultus. There is no basis for suggesting
it represents an alternative or more prim-
itive condition than those of the relative-
ly primitive anole species.

Chamaeleolis can not be distinguished
from Anolis in the structure of the pad
complex. Setae, lamellae, and other basic
features of the pad appear to be homol-
ogous in the two genera.

Chamaelinorops and A. sheplani, A.
occultus, A. chrysolepis, A. auratus,
and A. sp. n. near eulaemus

Chamaelinorops differs from A.
sheplani and A. occultus in every aspect
of the gross and fine structure of the sub-
digital pad. In spite of their similarity in
body size and habitus, Chamaelinorops
shares little beyond the presence of a pad
with these two anole species. The
Chamaelinorops pad is very small and
different in shape and position on the
digit. The Chamaelinorops setae are dis-
tinct in shape, dimensions and distribu-
tion.

In its gross morphology, Chamaelinor-
ops resembles a variety of the more
terrestrial, derived anole species, partic-
ularly A. auratus. Common features in-
clude: the Norops condition, a narrow
pad, the absence of lamellae under the
third phalanx, expansion of lamellae into
the phalanx i region, and the relatively
conservative width/length proportions of
the lamellae. Chamaelinorops differs
from A. auratus in the extent to which
these features are developed. In almost
every case the trend is more extreme in
Chamaelinorops—pad area is lower rela-
tive to body size, the lamellar proportions
are closer to those of generalized sub-
digital scales, the lamellar free margin is
severely reduced, and lamellae have
expanded further into the phalanx i
region. The Chamaelinorops pad is even
less well developed than that of A.
annectens, a highly derived anole related
to the padless A. onca (Peterson and
Williams, 1981).

Advances in Herpetology and Evolutionary Biology

In fine structure, Chamaelinorops dif-
fers from all other anolines examined to
date in the reduced density and large size
of the seta tip. An undescribed South
American species, A. sp. n. near
eulaemus, has setae whose tips are some-
what larger and approach the rectangular
shape of Chamaelinorops setae. ~A.
auratus exhibits seta distribution similar
to that of Chamaelinorops, but even
these highly derived anole species exhib-
it setae which are more similar to those of
generalized anoles than to those of
Chamaelinorops.

The distinction between the anole
species and Chamaelinorops in fine
structure is not confined to the seta. The
entire Chamaelinorops spine-seta series
is based on a lower stalk density (0.5/sq
m) and reduced variation in stalk height.
Chamaelinorops also exhibits a parallel
series of spines which is based on vary-
ing diameter and stalk density. Ultimate-
ly, increased diameter and decreased
density produce huge keel-like spines.
Comparable spine morphology does not
occur on the subdigital scales of anoles,
although large, pyramidal, single spines
and a more conservative spine series are
found on the lateral digital scales of some
anoles (e.g., A. auratus). Possibly the
most primitive portion of the fine struc-
tural suite, the spinule-spine series, is
comparable to that found in the Anolis

species.
In fine structure, as in gross pad mor-
phology, Chamaelinorops resembles

derived anole species more than the rela-
tively primitive West Indian forms, A.
occultus and A. sheplani. But in fine
structure, the similarity between
Chamaelinorops and any anole is ex-
tremely weak, and it is not at all clear that
the Chamaelinorops spine-seta series is
derived from an Anolis-like series.

Phenacosaurus and A. valencienni

The fine structure and most aspects of
the gross morphology of the pad are very
similar in Phenacosaurus and A. valen-
cienni. The only differences are in the

EVOLUTION OF THE SUBDIGITAL PAD - Peterson

gross structure of the pad. a) The Phena-
cosaurus lamellae are less variable in
shape and have more conservative pro-
portions (the scale width/length ratios are
5.0 compared to 9.0 in A. valencienni).
This feature does not, however, dis-
tinguish Phenacosaurus from the dwarf
twig ecomorphs, A. occultus and A.
sheplani (lamellar width/length ratio is
5.0). b) The lamellae also have an ex-
panded distribution in Phenacosaurus.
They extended throughout the phalanx iv
region and over the proximal phalanx i
region. Both phalanx i and phalanx iv
lamellae occur in Anolis; the condition in
Phenacosaurus is an extreme, but not a
novel pad position.

The case for an independent origin of
the pad complex in Phenacosaurus rests
solely on the difference in pad position.
Among gecko lineages differences in pad
position do indicate alternative pathways
in pad evolution (Russell, 1975, 1976,
1979), but these cases involve topo-
graphic alternatives (e.g., a terminal
versus a mid-digit pad) and often major
differences in the internal structure of
the pad, not simply relative degrees of
expansion (see Discussion). Evidence
from gross pad morphology alone is in-
sufficient to rule out an independent pad
origin, but the fine structural compari-
sons would argue against it, and other
interpretations of the difference in pad
position seem at least as plausible. A.
valencienni and Phenacosaurus have
similar structural habitats, but extreme
pad expansion in Phenacosaurus might
be correlated with subtle differences in
the structural habitat and in foot place-
ment. The pad area in Phenacosaurus is
distributed so that if the interphalangeal
joint mobility is comparable to that in A.
valencienni, the pad can be made to con-
form to a smaller radius of perch curva-
ture. Alternatively the difference might
be correlated with the adaptive history of
the two lineages. The two twig giants
might arrive at the same body size and
pad area from different ancestral sizes
and pad shapes. Phenacosaurus could be
a “scaled up” twig dwarf with negative

275

allometry of the ii and iii phalanges
forcing expansion of the pad. There is
sufficient variation in pad position among
anoles to suggest that the Phenacosaurus
morphology represents an extreme with-
in the context of an Anolis-like pad com-
plex rather than an alterative, non-
homologous pad.

THE EVOLUTION OF THE SUBDIGITAL
PAD COMPLEX AMONG ANOLINES: A
PRELIMINARY VIEW FROM INSIDE THE
RADIATION

Anolis itself poses the greatest chal-
lenge in comprehending the evolution-
ary diversity of the pad complex. To date
only about 15% of the 200 to 250 species
have been examined, but this sample in-
cludes a variety of West Indian, Central
and South American species groups, an
arboreal ecomorph series, and terrestrial
and semi-aquatic species. The sample is
sufficiently large and diverse to make
possible a very tentative assessment of
some aspects of pad evolution within the
genus.

Anolis species exhibit two basic simi-
larities in the pad complex—lamellar
scales forming the pad and the spine-seta
morphotypic series. Lamellae are wide,
flexible, flaplike scales. Anole lamellae
vary in shape, size and position on the
phalanges. The varieties of anole lamel-
lae probably do not represent independ-
ent acquisitions of the anole pad, but at
least the external morphology of the pad
scales does not provide enough informa-
tion to examine the issue carefully.
Scales that are much wider than long and
have a free margin, occur in some
geckos (Ruibal and Emst, 1965) and in
the arboreal skink, Prasinohaema virens
(Williams and Peterson, 1982). These
lamellae are certainly derived independ-
ently of those in Anolis and yet the
external morphology provides no clear
indication of this. The internal connec-
tive tissue and musculoskeletal frame-
work of the digit provide much more
information and probably would be a bet-
ter tool for investigating the macroevolu-

276

tion of the gross pad (see Russell, 1975,
1976, and 1979 for an analysis of these
features in the Gekkonidae). While
lamellar scales are a major feature of the
pad complex in Anolis and appear to be
homologous among anoles, their external
morphology is not a very satisfactory
basis for studying macroevolution among
anoles or anolines.

The spine-seta morphotypic series is
characterized by a 0.5 um stalk diameter,
variable stalk height, 1 stalk/sq um
spacing, and two basic tip shapes—a
hook and a spatula (a flattened triangle
joined to the stalk at the apex; Fig. 2).
The two tip shapes are linked by inter-
mediate shapes. This series occurs in all
the Anolis species examined to date. The
extent to which it discriminates anoles
from other iguanids is under investiga-
tion.

There is relatively little novelty in the
morphotypes within Anolis. None of the
West Indian species examined thus far
(and these include such unusual forms as
A. eugenegrahami) exhibit significant dif-
ferences in the morphotypes in spite of
very significant differences in body size,
habitat, species group, and gross pad
morphology. Instead, adaptation in fine
structure appears to be correlated with
differences in the absolute numbers and
distribution of the morphotypes. Extreme
specialization in the complex is, in some
cases, associated with a “retrograde”
shift in the series and the absence of
terminal morphotypes, such as the seta.
But it is still possible to identify the
series by stalk density, stalk diameter and
usually by the remaining tip shapes. For
example, A. oxylophus and A. macro-
lepis, two Central and South American
semi-aquatic species, have a _ well-
developed pad, but setae are replaced by
typical spike and prong morphotypes
(Peterson and Williams, unpublished
data). A. onca has neither pad nor setae,
but the subdigital spines have the shape
and density of anole spines (Peterson and
Williams, 1981). There is only one case in
which setae are replaced by a novel mor-

Advances in Herpetology and Evolutionary Biology

photype. In A. barkeri the lamellae are
covered by a branched prong, but the
diameter of the stalk, the density of stalks
and the presence of typically anole
spikes, prongs, and spines clearly label
the branched prong as a morphotype
derived from the anole series. There are
only three to four species in which the
dimensions of any of the morphotypes
are unusual or novel. For example, A.
chrysolepis exhibits larger setae tips and
a larger stalk diameter, but the seta den-
sity and tip shape clearly place the mor-
phology within the morphotypic series.
A. sp. n. near eulaemus discussed
previously is a second example.

The morphotypic series also discrimi-
nates the Anolis pad from that of skinks
and perhaps from that of geckos. The
arboreal skink, Prasinohaema virens, has
setae which are remarkably similar to
those of Anolis in shape, but the stalk di-
ameter and tip dimensions are an order of
magnitude greater, and the stalk density
is two orders of magnitude lower
(Williams and Peterson, 1982). The
spine—seta-prong portion of the mor-
photypic series is also absent (Williams
and Peterson, 1982). The setae of some
coleopteran species (Stork, 1980) also
have a shape that is remarkably similar to
that of Anolis setae, but the dimensions
appear to be quite different. Gecko pads
are too diverse and their epidermal fine
structure too poorly known to be certain
that the Anolis morphotypic series is
truly unique. However, the setae of a
variety of geckos appear to represent an
entirely different design for adhesion.
The setae exhibit large diameter
branched stalks in low density and many
triangular terminal spatulae (Ruibal and
Ernst, 1965; Hiller, 1968), but it remains
possible that the anole morphotypic
series is to some extent paralleled in
geckos. Russell (1976) describes a mor-
photypic series of Cyrtodactylus spe-
cies which involves increasing spine
height (with suggested variation in diam-
eter) that leads through an Anolis-like
seta tip to the branched seta with multi-

EVOLUTION OF THE SUBDIGITAL PAD - Peterson

ple spatulae. Stalk density and the
dimensions of the morphotypes, apart
from stalk height are not included and
forms which are intermediate between
the spine and the anole-like seta are not
described. These features might discrim-
inate the Anolis and Cyrtodactylus
series.

The morphotypic series in Anolis is a
heuristic comparative device based on
the existing subdigital morphology in
Anolis. The morphotypes (e.g., seta or
spike) summarize the variety in sub-
digital fine structure, and the order or ar-
rangement of morphotypes reflects the
pattern of topographic variation (e.g.,
fields of spikes grade into spines or
prongs, and are rarely adjacent to
spinules). But the arrangement of the
morphotypes is also consistent with the
patterns of evolutionary change observed
in the A. annectens-onca lineage (Peter-
son and Williams, 1981) and in the “‘retro-
grade” shift and substitution of spikes
or prongs for setae in Central and South
American semiaquatic anoles (Peterson
and Williams, unpublished data). The
morphotypic series may prove to describe
a number of phylogenetic sequences, but
it might become even more useful in inte-
grating developmental mechanisms and
functional change with phylogenetic se-
quences. In some instances the differ-
ences between adjacent morphotypes
moving from spine to seta correspond to
sequentially later stages in the develop-
mental sequence (e.g., density is
determined before stalk height; sug-
gested by Maderson [1970] and Peterson,
unpublished data), and the spine-seta
series which occurs over the proximal
portion of the lamella may be a relic
representation of setal development
(Peterson and Williams, 1981, see Fig.
23). Most of the differences between
morphotypes and thus between species
with different morphotypes in hom-
ologous areas can be interpreted in terms
of the ontogenetic/phylogenetic frame-
work of Alberch, et al. (1979). Develop-
mental interpretations of the series are

20h

still very speculative, although consider-
able information on the shedding cycle
and late stages in seta development are
available through the studies of Maderson
(1970, for example).

Some aspects of the series can already
be interpreted functionally. Three of the
dimensions of the spine-seta series—
stalk diameter, tip size, and stalk den-
sity—are relatively constant among
Anolis species (Peterson, et al., 1982; and
unpublished data). The tip, stalk, and
epidermal scale surface represent se-
quential links in the transmission of
tensile force to the dermal core of the
scale and ultimately to the musculoskel-
etal system of the digit. The consistency
in these dimensions implies that stress is
constant over a range of body sizes.
Scaling for different body sizes occurs in
the absolute number of setae and thus in
pad area (Peterson, et al., 1982). In con-
trast, variation in stalk height has no
effect on tensile stress, and stalk height
may vary independently of the other
dimensions over the surface of a given
lamella or between species.

The application and disengagement of
setae, spines, etc. form a second set of
functional constraints on the system.
Height and diameter determine the stalk
stiffness and the ability of the tip to
“find” the substrate. The size and den-
sity of tips could also influence how the
tips behave. Increased density, essential-
ly crowding of the tips, could reduce the
ability of individual tips to contact the
substrate unless diameter and tip size
are appropriately reduced. Conversely,
extreme reduction in density would give
the tips freedom to contact the substrate,
but could result in too few tips to support
the body mass, unless tip size and stalk
diameter were appropriately increased
(this would in effect reduce the distance
between setae and re-introduce the
problems associated with crowding).

The consistency among anoles in stalk
density, stalk diameter and seta tip size
appear to reflect functional constraints,
but there is no reason to expect the

278

particular dimensions of the anole
morphotypic series to be universal.
Indeed, there appear to be a variety of
setal dimensions and densities (e.g.,
anoles, Prasinohaema_ virens, geckos
and even Coleoptera). The dimensions of
Anolis morphotypic series may reflect
only one of the solutions which might
arise in the evolutionary sequence from
small spines and spinules to the spine-
seta series. Any differences in the char-
acter of the series, particularly in the
“constants, among the anolines and
non-anoline iguanids may indicate an
independent acquisition of the pad fine
structure.

The existing comparative data suggest
that the spine-seta morphotypic series is
homologous at least among Anolis,
Phenacosaurus and Chamaeleolis. The
morphology of the entire pad complex in
Phenacosaurus is quite consistent with
the view that it diverged from an Anolis
stock in which setae and a gross pad were
already well developed.

Chamaeleolis and Chamaelinorops are
thought to have diverged from the
anoline lineage much earlier, perhaps
prior to the differentiation of Anolis (Fig.
1). It is not clear whether the Chamae-
leolis lineage diverged earlier or later
than Chamaelinorops (cf. Fig. 1A, B).
The pad complex in Chamaeleolis
appears to be homologous and virtually
identical with that in relatively gener-
alized Anolis species. The Chamaelinor-
ops morphology is not easily inter-
preted. |

The gross pad morphology of Chamae-

linorops resembles that of highly
derived, more terrestrial anoles, e.g., A.
auratus. The resemblance in _ gross

morphology involves a number of char-
acters that at least within Anolis vary
independently (e.g., the Norops condi-
tion is not always associated with the
absence of phalanx iii lamellae; Table 1).
The similarity between A. auratus and
Chamaelinorops is not readily dismissed
in favor of the dissimilarity in fine struc-
ture, but neither is it easy to accept the

Advances in Herpetology and Evolutionary Biology

verdict that the Chamaelinorops mor-
phology is derived from an Anolis-like
pad. The resemblance between A.
auratus and Chamaelinorops is based on
features which could characterize an
early, nascent pad (e.g., the Norops con-
dition and a very few lamellae that are
not well differentiated from generalized
subdigital scales). The fine structure of A.
auratus clearly labels it a derived anole
(Peterson and Williams, 1981), but the
gross morphology might be convergent
on the early stages in pad evolution. The
Chamaelinorops pad morphology might
be homologous with that of Anolis and
represent a) a highly derived pad condi-
tion or b) a relic, very early stage in pad
evolution, substantially before the well-
defined pads of Chamaeleolis and A.
sheplani. The gross morphology also
permits a third interpretation because
none of the gross characters—lamellar
morphology, pad shape and position, or
phalangeal proportions—are very inci-
sive criteria for pad homology at this
level. The Chamaelinorops pad might
also represent an independent, parallel
acquisition of the pad within the
anolines.

The fine structure of the subdigital
surface is unique among the anolines
examined. It is unlikely that the morphol-
ogy represents a relic, primitive condi-
tion in the sequence leading to Chamae-
leolis and Anolis. a) The spine-seta
morphotypic series appears to arise from
spinules through a decrease in spinule
density and an increase in stalk height
and diameter. The trend would have to
be reversed to shift from the 0.5 stalks/sq
fem series spacing in Chamaelinorops to
the 1 stalk/sq mm series spacing in
Anolis. b) The fine structure in Chamae-
leolis and primitive Anolis species
exhibits regional differentiation. In these
forms the phalanx i scales are covered by
spinules or spines which are more con-
servative and presumably more primitive
members of the morphotypic series. If
the Chamaelinorops morphology repre-
sented an intermediate between the

EVOLUTION OF THE SUBDIGITAL PAD - Peterson

spinate/spinulate scales of preanoline
iguanids and those of Anolis, it seems
unlikely that the phalanx i scales would
be covered with setae. c) If the Chamae-
linorops morphology represented an
intermediate in the sequence of anole
pad evolution, one would expect to find
at least some evidence of the large spine
morphotypic series in more primitive
Anolis species and in Chamaeleolis, not
simply in derived Anolis species. The
fine structure suggests that the Chamae-
leolis pad is derived, but it does not
unequivocably suggest whether it is
derived independently or from an Anolis-
like pad morphology.

There are three morphotypic series in
Chamaelinorops: a) a “spinule to Anolis-
like spine” series which is characterized
by variation in stalk density, diameter
and height; b) a large, pyramidal spine
series that is characterized by varying
spine diameter (minimum 1.5 um), vary-
ing spine density (minimum 0.5/sq “m)
and relatively little variation in spine
height; and c) a spine-seta series which is
characterized by relatively constant stalk
diameter (0.5 wm) and stalk density (0.5/
sq wm) and by varying stalk height and tip
shape. The three series appear to be
linked at the spine morphotype. The
development of individual spines has not
yet been described, but extensions of the
processes implicated in the morphotypic
series provide tentative hypotheses for
their evolutionary relationships. The
spine that forms the base of the
Chamaelinorops spine-seta series (0.5
fm diameter; 0.5 stalks/sq pm) has
the same spacing and height as the smal-
lest and simplest member of the large
spine series. A discrete change in the
developmental program, perhaps an
acceleration (sensu Alberch et al., 1979)
in the growth in diameter at a stage after
the position and density of the spine
stalks were set, would convert the 0.5 um
diameter spine into the 1.5 wm diameter
spine. Local variation in the develop-
mental program for density and diameter
could yield the rest of the large spine se-

2719

ries. The link between the Anolis-
type spine and the Chamaelinorops
spine involves merely a change in

density or spacing from about 1/sq um to
0.5/sq wm. This change might also be
termed an acceleration (Alberch et al.,
1979) in the reduction of spinule density
relative to the changes in the spine di-
ameter and height.

These two shifts in the developmental
program—an acceleration in spine den-
sity and in spine diameter—which could
yield the two distinctive Chamaelinorops
series are more likely to have occurred
early than late in the evolution of the pad
complex. Ifa) the Anolis spine-seta series
was already present, and b) a shift in the
developmental program affected the
entire subdigital Oberhautchen, all mem-
bers of the morphotypic series, including
the spinules and small spines, would be
shifted to a lower density. While the
density of the spinules in Chamaelinor-
ops is lower than that in Anolis, the
Anolis-type spine with a 0.5 wm diameter
and a 1.0/sq wm density would not occur
given this scenario. An alternative pos-
sibility, that the shift in density occurred
only over the fully exposed portions of
the scale, would affect the density of the
seta morphotype, but not that of spines
and the intermediate morphotypes at the
base of the scale. Neither scenario will
produce the Chamaelinorops morphol-
ogy. Among anoles which exhibit any
variation in the series, it is only the seta
which is different. If the Chamaelinorops
fine structures were derived from that of
Anolis, the comparative data would sug-
gest that only the seta should be unique
and not the prong and the spine. The
most plausible interpretation of the char-
acter of the three series and their linkage
through the spine morphotype is that the
shift to a 0.5/sq wm density took place
before setae differentiated, when the
scale surface was spinate. The Chamae-
leolis-Phenacosaurus spine-seta_ series
and the Chamaelinorops spine-seta
series would then have evolved in paral-
lel (see Fig. 1). If this is the case, the

280

pre-anoline genera, Enyalius, and the
para-anoline genera might exhibit com-
parable variation in spine and/or series
density (Fig. 1B). The 0.5 stalks/sq um
spacing of Chamaelinorops does occur in
Enyalius, but the issue needs to be
examined in light of comparative data
from a number of related iguanid genera
(Fig. 1B). A variety of other features,
particularly the unique shape and
dimensions of the seta tip, would support
the argument that the Chamaelinorops
spine-seta series evolved in parallel with
that of Anolis.

The opposite argument—that the fine
structure of Chamaelinorops is derived
from that of Anolis—can be made based
on the similarity between the Chamae-
linorops and Anolis series in stalk diam-
eter and on the parallels between
derived anoles and Chamaelinorops in
seta distribution, in the presence of large,
pyramidal spines, and on the weak par-
allels between Chamaelinorops setae
and those of A. sp. n. near eulaemus. The
unique setal shape and dimensions of
Chamaelinorops might arise through an
early arrest in seta development or
paedomorphosis. Early cessation of
growth in stalk height could leave a
larger, untapered “end” of the stalk, and
a partial differentiation of this “end” into
a tip might produce a thick, square tip.
The lower density of the spine-seta series
is almost certainly an independent
change in the developmental sequence
and presumably represents the modifica-
tion of a much earlier stage in the pro-
gram. It is more difficult to interpret the
series density as a derivative of the
Anolis condition (see above). The argu-
ment that the fine structure is derived
from the Anolis condition is supported by
albumin immunological distance data
which suggest that Chamaelinorops is
close to a relatively generalized His-
paniolan anole, A. cybotes (Wyles and
Gorman, 1980).

The relative consistency of spine and
seta shape and density among Anolis,
Chamaeleolis, and Phenacosaurus sug-

Advances in Herpetology and Evolutionary Biology

gests broad similarity among the forms in
the prehensile and adhesive functions of
the fine structure (Hiller, 1968; Peterson
and Williams, 1981). Behavioral and
ecological interpretations of the Chamae-
linorops morphology are not yet pos-
sible; it is not even known if Chamae-
linorops can maintain grip without its
claws. Two specimens kept in captivity
for two and four weeks were never
observed to voluntarily adhere to glass or
leaf surfaces and could not be encour-
aged to do so. The physical correlates of
the leaf litter habitat are also poorly
known. The presence of moist films and
debris could profoundly affect the
options for an adhesive or a prehensile
grip based on the fine structure. There
also appear to be significant differences
in the morphological design of the sys-
tem. The seta tip in Chamaelinorops has
an area 3 to 6 times larger than that in
Anolis, but a similar stalk cross-sectional
area. If the setae work by adhesion and
the entire flattened area of the tip is in-
volved in an adhesive contact, the stress
on the stalk would be 3 to 6 times greater
in Chamaelinorops, and the stress
created in the epidermal surface sur-
rounding the stalk would be approx-
imately 1.5 to 3 times greater. The
Chamaelinorops setae stalks are shorter
and therefore likely to be much stiffer
than anole seta stalks. The oval stalk and
tip are also thicker and presumably less
flexible than those in Anolis. Both the
increased stiffness of the seta and the
absence of any reasonably consistent tip
flexion which would orient the tip paral-
lel to the substrate suggest that an indivi-
dual Chamaelinorops seta would be
substantially less effective in “finding”
and making contact with the locomotor
substrate than the anole seta. However, it
is possible that the morphological “strat-
egy for adhesion is simply an altemmative
to that in Anolis, Phenacosaurus, and
Chamaeleolis. The Chamaelinorops
“strategy is a lower density of setae that
are distributed over the entire subdigital
surface; each seta having a lower prob-

EVOLUTION OF THE SUBDIGITAL PAD - Peterson

ability of establishing contact, but able to
support 3 to 6 times more load if it does.
This might be as effective a solution as
that of the other anolines, but only if the
functional tip area is very great, and the
non-lamellar setae are as effective as
those on the pad. Estimates of the setae
numbers and tip areas illustrate this
point. Comparing A. sheplani (MCZ
125641; 40.8 um SV length) and Chamae-
linorops (MCZ 156919; 31 mm SV
length), pad area is much greater in A.
sheplani (0.69 sq mm compared to 0.41 sq
mm), but because of the longer toe and
the presence of phalanx i setae, the actual
seta bearing area is much greater in
Chamaelinorops (1.24 sq mm compared
to 0.98 sq mm). Because of the difference
in density the total number of setae is
twice as great in A. sheplani (1.2 x 106
setae compared to 0.62 x 108 setae; the
number of setae is estimated from seta
bearing area and seta density), but the
total surface area of the setae tips is
greater in Chamaelinorops (90.0 x 104 sq
fxm compared to 25.8 x 104 sq um; total
tip area is estimated as mean tip area
[from Table 1] x the number of setae).
Chamaelinorops MCZ 156919 has un-
usually large seta tips (Table 1). If the tip
area of MCZ 126708 is substituted for the
actual value the sense of the comparison
is unchanged, but the difference is
reduced to the point where cumulative
error in the estimate makes inference less
certain (47.0 x 104 sq wm for Chamae-
linorops and 25.8 x 104 sq mm for A.
sheplani; an Anolis species with the
body dimensions of Chamaelinorops
would have an estimated tip area of 18.1
x 104 sq um based on the scaling rela-
tionships developed for Puerto Rican
anoles; Peterson et al., 1982, and Peter-
son, unpublished data). For more typical
Chamaelinorops tip areas the advantage
in total tip area is less than a factor of 3
while the number of setae or possible
points of contact is reduced by half. The
Chamaelinorops and Anolis morpholog-
ical adaptations are equal, alternative
designs for adhesion only if a) all of the

281

flattened tip area in Chamaelinorops can
make effective contact with the substrate,
and b) the short and presumably stiffer
Chamaelinorops setae are equally likely
to “find” the substrate.

Some of the same arguments apply if
the Chamaelinorops setae work like
Velcro in micro-prehension. The greater
stiffness and lower seta density would
presumably reduce the number of poten-
tial contacts and the ability of the tips to
“find” surface irregularities. The tip
shape may also be less effective in pre-
hension than that of Anolis. The anole
seta retains a hook shape and has a wide,
thin distal border which might catch
microscopic irregularities as the seta is
brushed along an environmental surface
at the beginning of the propulsive phase
(Peterson and Williams, 1981). In
Chamaelinorops, the tip narrows distally,
and it may have a rolled, thick distal
edge. These could reduce the probability
that the margin will engage an irregular-
ity. Also, without some initial tip flexion
toward the metatarsal-phalangeal joint,
the distal margin of the tip may not form
the leading edge as the digit makes con-
tact with the substrate. Without this ori-
entation the seta can not act like a hook
and engage surface irregularities.

It is quite possible that neither of the
seta mechanisms presumed in Anolis are
relevant to Chamaelinorops. The distinc-
tive features of the Chamaelinorops
seta—the absence of a consistent tip
flexion, reduced stalk density and in-
creased stiffness—could be advantageous
if the setae functioned to increase the
frictional coefficient of the surface. If the
leaf litter substrate is “slippery” because
of moist films or debris, traction might
become more important than adhesion or
prehension. While it is far from clear how
and in what context the setae of Chamae-
linorops function, there are a number of
reasons to suspect that their behavior and
adaptive role is different from that of
Anolis setae.

The large spine series of Chamaelinor-
ops differs from most anole spine series

282

in that a hook tip is consistently absent. It
is difficult to imagine how the large
spines could act like hooks and form a
prehensile, gripping scale surface. The
pyramidal spines probably function like
minature keels and increase the rough-
ness or frictional coefficient of the scale
surface. The association of large spines
with hillocks (which increase surface
roughness in a larger dimensional range)
and scale regions which have the most
relief (e.g., the distolateral comers of the
phalanx i scales) supports this interpreta-
tion. The series of fused spines con-
verge on the dimensions and appearance
of the keels.

Consideration of the possible func-
tional significane of the differences
between the Chamaelinorops and Anolis
fine structure is highly speculative, but
even interpretations based on alternative
assumptions about the seta mechanism
corroborate the purely morphological
arguments that the Chamaelinorops
morphology is significantly different
from that in Anolis, Chamaeleolis and
Phenacosaurus. Functional interpreta-
tions can not at this point resolve
whether the Chamaelinorops morphol-
ogy represents secondary terrestrial
adaptation of an Anolis-like fine structure
or an independent, perhaps even a less
successful “experiment” in seta evolu-
tion.

CONCLUSIONS

The gross and fine structures of the
Chamaeleolis and Phenacosaurus pads
are similar and appear to be homologous
with those of Anolis. The pad complex,
including the spine-seta morphotypic
series and lamellae, probably appeared
relatively early in the anoline radiation
aes the differentiation of Chamaeleo-
is.

With the exception of the expanded
pad position in Phenacosaurus, the dif-
ferences between the genera and indivi-
dual Anolis species are no greater than

Advances in Herpetology and Evolutionary Biology

the differences between even closely
related Anolis species. If one views
Phenacosaurus, Chamaeleolis, and such
Anolis species as cuvieri, sheplani,
auratus, chrysolepis, and valencienni as
representatives of the taxonomic radia-
tion of anolines, the pad complex exhibits
radical differences in the number, distri-
bution and shape of the lamellae, in
phalangeal proportions, and in the distri-
bution of setae and spines. The complex
appears to be basically conservative in
the spacing and diameter of spines and
setae, the setae tip shape and dimensions,
and the presence of a morphotypic series
linking the spine and the seta.
Chamaelinorops has a unique pad fine
structure, but resembles derived, more
terrestrial anole species in gross mor- —
phology. The Chamaelinorops morphol-
ogy may represent a highly specialized
version of the Anolis-Chamaeleolis-
Phenacosaurus subdigital pad or an
independently derived pad complex.

ACKNOWLEDGMENTS

I am very grateful to Emest E.
Williams for discussions of this problem
and for permission to examine material in
his care and to E. E. Williams, D. B. Wake,
A. P. Russell, and D. G. Buth for com-
ments on the manuscript.

I thank J. Berliner and S. Beydler of the
Department of Pathology, UCLA, for
providing access to an excellent SEM
facility. L. Meszoly, M. Kowalczyk and
H. Kabe prepared the plates. The study
was supported by a UCLA University
Research Grant.

LITERATURE CITED

ALBERCH, P., S. J. GOULD, G. F. OSTER, AND D. B.
WakeE. 1979. Size and shape in ontogeny and
phylogeny. Paleobiology, 5: 296-318.

BOULENGER, G. 1885. Catalogue of the lizards in the
British Museum (Natural History), 2nd edition,
2. London, Taylor and Francis.

EVOLUTION OF THE SUBDIGITAL PAD - Peterson

ETHERIDGE, R. F. 1960. The Relationships of the
Anoles (Reptilia: Sauria: Iguanidae): An Inter-
pretation based on Skeletal Morphology. PhD.
Thesis, Univ. Michigan, University Micro-
films, Inc., Ann Arbor, Michigan, xiv + 236 pp.

HILLER, U. 1968. Untersuchungen zum Feinbau
and zur Funktion der Haftborsten von Reptil-
ien. Z. Morphol. Tiere, 62: 307-362.

LAZELL, J. D. 1969. The Genus Phenacosaurus
(Sauria: Iguanidae). Breviora Mus. Comp.
Zool. No. 325, pp. 1-24.

Liem, K. F. 1973. Evolutionary Strategies and
Morphological Innovations: Cichlid Pharyn-
geal Jaws. Syst. Zool., 22: 425-441.

LILLYWHITE, H. B., AND P. F. A. MADERSON. 1968.
Historical Changes in the Epidermis of the
Subdigital Lamellae of Anolis carolinensis
during the Shedding Cycle. J. Morphol., 125:
379-402.

MADERSON, P. F. A. 1970. Lizard glands and lizard
hands: Models for evolutionary study. Forma
et Functio, 3: 179-204.

OLSON, E. C. 1975. Permo-carboniferous paleo-
ecology and morphotypic series. Am. Zool., 15:
371-391.

PAULL, D., E. E. WILLIAMS, AND W. P. HALL. 1975.
Lizard karyotypes from the Galapagos Islands:
Chromosomes in phylogeny and evolution.
Breviora Mus. Comp. Zool., No. 441, pp. 1-31.

PETERSON, J., S. BENSON, J. MORIN, M. NGAI, AND C.
Ow. 1982. Scaling in tensile “skeletons.”: I.
Structures with scale independent length
dimensions. Science, 217: 1267-1270.

PETERSON, J. A., AND E. E. WILLIAMS. 1981. A Case
Study in Retrograde Evolution; The Onca
Lineage in Anoline Lizards. II., Subdigital
Fine Structure. Bull. Mus. Comp. Zool., 149:
215-268.

RUIBAL, R. 1968. The ultrastructure of the surface of
lizard scales. Copeia, 1968: 698-704.

283

RUIBAL, R., AND V. ERNST. 1965. The structure of the
digital setae of lizards. J. Morphol. 117: 271-
294,

RUSSELL, A. 1975. A contribution to the functional
analysis of the foot of the tokay, Gekko gecko
(Reptilia: Gekkonidae). J. Zool. Lond., 176:
437-76.

1976. Some comments conceming inter-
relationships among gekkonine geckos, pp.
217-244. In A. d’A. Bellairs and C. B. Cox
(eds.), Morphology and Biology of Reptiles.
Linnean Society Symposium, Series 3, Lon-
don, Academic Press.

___.. 1979. Parallelism and integrated design in the
foot structure of gekkonine and diplodactyline
geckoes. Copeia, 1979: 1-22.

SCHWARTZ, A. 1974. A new species of primitive
Anolis (Sauria, Iguanidae) from the Sierra de
Baoruco, Hispaniola. Breviora Mus. Comp.
Zool. No. 423, pp. 1-19.

Stork, N. E. 1980. A scanning electron microscope
study of tarsal adhesive setae in the Coleop-
tera. Zool. J. Linn. Soc., 68: 173-306.

WILLIAMS, E. E. 1976. West Indian anoles: A tax-
onomic and evolutionary summary. 1. Intro-
duction and a species list. Breviora Mus.
Comp. Zool. No. 440, pp. 1-21.

____. 1977. 3. The macrosystematics of the anoles,
pp. 122-131. In E. E. Williams (ed.), the third
Anolis Newsletter. Cambridge, Mass., Mu-
seum of Comparative Zoology.

WILLIAMS, E.. E., AND J. A. PETERSON. 1982. Con-
vergent and alternative designs in the adhe-
sive pad of scincid lizards. Science, 215: 1509-
1511.

WYLES, J. S., AND G. C. GORMAN. 1980. The Classifi-
cation of Anolis: Conflict between Genetic and
Osteological Interpretation as Exemplified by
Anolis cybotes. J. Herpetol. 14: 149-153.

The Axial Skeleton of Chamaelinorops

KARL FORSGAARD !

ABSTRACT. The axial skeletons of the beta-anoles
Chamaelinorops barbouri and Anolis garmani are
described in detail. The caudal transverse process-
es of Chamaelinorops are found to be completely
unlike those of other beta-anoles, and thus the close
relationship that Etheridge (1960) proposed seems
dubious. The zygapophysial plates and other ver-
tebral specializations of Chamaelinorops are
hypothesized to have evolved to allow an excep-
tionally strong contraction of the long epaxial
musculature, possibly to aid in jumping or in the
motions of the head during display behavior.

INTRODUCTION

Etheridge (1960) divided the species of
anoline lizards into two groups, alpha
and beta. This entire assemblage of small
New World iguanids then comprised
eleven genera, with the genus Anolis
containing over 200 species and the other
genera being monotypic or containing
only a few species. Etheridge (1960) syn-
onymized six of these genera with
Anolis: Audantia (2 species), Mariguana
(1 sp.), Diaphoranolis (1 sp.), Xiphocer-
cus (2 sp)., Deiroptyx (2 sp.), and Norops
(4 sp.). Williams (1974) additionally
synonymized Tropidodactylus (1 sp.)
with Anolis. Currently, only four genera
of anoles are recognized: Anolis,
Chamaeleolis, Chamaelinorops, and
Phenacosaurus. Etheridge (1960) classi-
fied Chamaeleolis, Phenacosaurus, and
most Anolis as the alpha section of an-
olines; Chamaelinorops and the rest of
Anolis as the beta section.

1University of Washington School of Law,
Seattle, Washington 98105, U.S.A. (Present address:
1111 Third Ave., Suite 1500, Seattle, Washington
98101, U.S.A.)

The alpha-anoles are characterized by
the absence of transverse processes on
caudal vertebrae posterior to the point
where autotomy arises; beta-anoles have
such processes. Etheridge (1960) de-
fended the validity of this scheme by not-
ing that caudovertebral structure is
useful in distinguishing other lizard fami-
lies as well as non-anoline iguanids. His
phyletic hypotheses, however, have an
element of uncertainty, and he admitted
the possibility that caudovertebral trans-
verse processes could have been inde-
pendently lost in several lines of Anolis.

Perhaps the most questionable assign-
ment in his phylogeny is the position of
Chamaelinorops as a beta-anole. Its ver-
tebrae are quite unlike those of other
anoles, and, by Etheridge’s own admis-
sion, are even significantly different from
those of all other lizards. He noted
(1960) that “each of the vertebrae of the
neck and body has a pair of broad, flat,
laterally projecting winglike processes
arising just below the base of the neural
arch and extending between the pre-
and _ postzygapophyses.” Im the tail
region he observed (1960) “the presence
of heavily ossified structures” and
added, “I am unable to determine just
what these structures are from radio-
graphs. They do not appear to be proces-
ses of the same sort as are found in the
[trunk] vertebrae, yet they seem to be
closely associated with, and possibly
processes of, the caudal vertebrae.” That
Etheridge did not directly examine the
axial skeleton of Chamaelinorops creates
considerable doubt about the validity of
the position of the genus within his an-

AXIAL SKELETON OF CHAMAELINOROPS + Forsgaard

oline phylogeny, particularly in light of
the singular importance of caudoverte-
_ bral structure to that phylogeny.
Relevant skeletal data on Chamaelin-
orops are not available in the literature.
Previous works on the external features
and ecology of the genus are to be found
in Schmidt (1919), Cochran (1928, 1941),

Etheridge (1960), Thomas _ (1966),
Schwartz and Thomas (1975), and
Schwartz and _ Inchaustegui (1980).

Hoffstetter and Gasc (1969) noted the
lack of caudal autotomy, but did not elu-
cidate any other skeletal characteristics.
Chamaelinorops barbouri is the sole ex-
tant representative of the genus, and is
restricted to the mountainous areas of
Hispaniola. In this paper I provide a de-
tailed osteological description of the axial
skeleton of Chamaelinorops barbouri
(Fig. 1) and compare it to that of the beta-
anole Anolis garmani (Fig. 2).

DESCRIPTION OF THE AXIAL
SKELETON OF CHAMAELINOROPS
BARBOURI

The vertebral column of Chamaelinor-
Ops consists of approximately 68 procoel-
ous vertebrae: 8 cervical, 11 thoracic, 5
lumbar, and 2 sacral vertebrae comprise
the trunk vertebral series, and approx-
imately 42 comprise the caudal series.
The centra are constricted midway be-
tween their anterior and posterior ends.

Cervical Vertebrae. There are eight
cervical vertebrae in Chamaelinorops,
the posterior four of which articulate with
ribs. The anterior four have hypapophys-
es, as do the anterior four of all anoles
examined.

The atlas has its anterior face ringlike
and tilted so that the dorsal border is
posterior to the ventral border. It is typ-
ical of the “Category B” atlas described
by Hoffstetter and Gasc (1969). The ring
is formed ventrally by the intercentrum,
which projects ventrally as a hypapophy-
sis, and dorsolaterally by the neural arch,
which is composed of two flat, oblong

285

plates, compressed posteriorly, which are
separated in the dorsal midline and are
overlapped by a forward projecting pro-
cess on the neural spine of the axis. The
atlas of Anolis garmani is longer and
more massive, and the lateral processes
of its neural arch do not extend as far lat-
erally, nor does its hypapophysis extend
as far ventrally.

The axis is comprised of neural arch
and centrum, intercentra nos. 2 and 3,
and the centrum of the atlas. Sutures be-
tween these elements are clearly seen in
A. garmani, but were not evident in any
of the Chamaelinorops specimens exam-
ined. The neural arch of the axis in
Chamaelinorops has a large, laterally
compressed neural spine which projects
anteriorly over the neural arch of the atlas
as well as posteriorly, like the spines of
the remaining trunk vertebrae. The dor-
sal crest of this spine is a rounded ridge
which slants so that its posterior end is
dorsal to its anterior end. Also arising
from the neural arch dorsolaterally are
the zygapophyses. The prezygapophyses
project forward and underlie the post-
zygapophysial area formed by the dor-
somedial surface of the atlas neural-arch
plates. The postzygapophyses of the axis
project laterally as flat plates so that their
rounded tips are at a distance from each
other about four times the diameter of the
axis condyle. The axis of A. garmani is
similar to that of Chamaelinorops, except
that its hypapophysis is ventrally flatten-
ed into a hypapophysial plate, its prezyg-
apophyses extend farther anteriorly, its
postzygapophyses extend laterally rather
than dorsolaterally, and its postzygapo-
physial width is not as great. The hypa-
pophysial plate, similar to the neural-
spine splate of Chamaelinorops, was also
noted in the alpha-anole A. baleatus and
in the beta-anoles A. biporcatus, A.
fuscoauratus, A. chrysolepis, A. nebu-
losus, and A. onca.

In Chamaelinorops, the axis postzyga-
pophyses articulate with the prezygapo-
physes of the third cervical vertebra, the
postzygapophyses of which extend even

286 Advances in Herpetology and Evolutionary Biology

Figure 1. Representative elements of axial skeleton of Chamaelinorops (A-B: MCZ 146960, C-F: MCZ 146958).
A. Atlas. B. Axis. C. Thoracic vertebra no. 17. D. Sacrum. E. Caudal vertebra no. 5, minus the caudal chevron and
one of the transverse processes. F. Caudal vertebra no. 11. For each element, the five views represent lateral,
dorsal, ventral, anterior, and posterior views from top to bottom, left to right.

Abbreviations: atc, atlanto-centrum process of the axis; atp, atlantal plate; as, autotomy spetum; bar, base of the
atlas ring; cch, caudal chevron; cd, condyle; f, foramen; g, groove; gid, groove-like depression; hp, hypapophysis;
hppl, hypapophysial plate; m, medial edge of the zygapophysial plate; ns, neural spine; nsp, neural-spine plate;
pp, parapophysis; prz, prezygapophysis; ptz, postzygapophysis; pzp, prezygapophysial plate; s, spine; tp, trans-
verse process; tpp, transverse-process plate; zp, zygapophysial plate.

AXIAL SKELETON OF CHAMAELINOROPS « Forsgaard 287

Figure 2. Representative elements of axial skeleton of Anolis garmani (MCZ 140048). A. Atlas. B. Axis (compo-
site view including MCZ 7630). C. Thoracic vertebra no. 16. D. Sacrum. E. Caudal vertebra no. 5. F. Caudal
vertebra no. 11 (note lack of terminal bifurcation of transverse process). Views identical to Figure 1.

Abbreviations: see Figure 1.

further apart laterally. This increase in
lateral extension of the zygapophyses
continues in Chamaelinorops so that at
cervical vertebra no. 8, the vertebral
width is about five times the diameter of
the condyle of the eighth vertebra (a
diameter that is considerably greater than
the condyle diameter of the axis) (see
Fig. 3). In A. garmani, the reverse occurs:
the ratio of the zygapophysial width to
the condyle diameter decreases from the
axis to the middle thoracic vertebrae. In
all species examined other than
Chamaelinorops, the zygapophysial
width of virtually all of the presacral
vertebrae was approximately that of the
postzygapophyses of the axis, and none
attained the zygapophysial width of
Chamaelinorops, and the width of the
vertebral body was greatest in the cervi-
cal region and just anterior to the sacrum,
with the thoracic vertebrae having slight-
ly thinner bodies.

On vertebrae nos. 2 to 6 in Chamae-
linorops, the pre- and postzygapophyses
extend laterally so that they remain dis-
tinct from one another in a dorsal view.
But beginning with vertebra no. 7, a
zygapophysial plate extends laterally be-
tween the pre- and postzygapophyses
and includes them, so that the prezy-
gapophyses comprise the anterolateral
comers of the plate, while the post-
zygapophyses comprise the posterolater-
al comers of the plate. No such zyga-
pophysial plate occurs in the vertebrae of
A. garmani or in those of any other mem-
ber of the genera Anolis, Chamaeleolis,
and Phenacosaurus that were examined
in this study.

The neural spines of Chamaelinorops
vertebrae nos. 3 to 5 are smaller than that
of the axis. There is relatively little varia-
tion in neural-spine height throughout
the rest of the vertebral column. Among
other anoles there is considerable varia-
tion in the relative height of the neural
spines. These spines are low ridges on
the midline of the neural arch in
Chamaeleolis, while in A. bimaculatus
the neural spines constitute two-thirds of

Advances in Herpetology and Evolutionary Biology

the vertebral height. In A. valencienni,
the neural spines are highest in the cer-
vical region; in A. ophiolepis, A. sagrei,
A. equestris, and A. coelestinus the neu-
ral spines are highest in the anterior
caudals; in the other anoles examined, as
in Chamaelinorops and A. garmani, the
height of the neural spines is fairly con-
stant throughout the column, except for
diminution in the posterior caudals.

On Chamaelinorops vertebra no. 6
there is a flattening of the posterodorsal
area of the neural spine, and on vertebra
no. 7 this flattened area widens to form a
plate; this neural-spine plate is found on
all vertebrae posterior to no. 7, except for
the posterior caudals. The anterior region
of each neural-spine plate is usually, but
not always, narrower than the posterior —
region of each plate. The dorsal surface of
each plate is usually rugose. Neural-
spine plates were found in most of the
anoline species examined, but only A.
cybotes had plates nearly as large as
those of Chamaelinorops. Small neural-
spine plates occur in Phenacosaurus, A.
latifrons, A. equestris, A. coelestinus, A.
carolinensis, A. baleatus, A. bimacu-
latus, A. fuscoauratus, A. chrysolepis, A.
nebulosus, and A. sagrei. Knobs occur on
the crests of the neural spines of A. an-
gusticeps, A. onca, A. garmani, A. valen-
cienni, and A. ophiolepis, and there is no
buildup of the neural spine at all in
Chamaeleolis. It is perhaps significant
that A. angusticeps and A. onca are at the
end points of two of the four main line-
ages in the Etheridge (1960) phylogeny,
and that A. garmani, A. valencienni, and
A. ophiolepis are in advanced positions
on a third lineage; retention of the
neural-spine plate appears from this lim-
ited data to be a sign of primitiveness.

All the cervical vertebrae posterior to
the axis in Chamaelinorops bear para-
pophyses projecting ventrolaterally from
the base of the neural arch; the para-
pophyses of vertebrae nos. 5 to 7, but not
no. 8, are considerably larger than those
of any other presacral vertebrae (see Fig.
1B). Unlike Chamaelinorops, the other

AXIAL SKELETON OF CHAMAELINOROPS : Forsgaard

289

Figure 3. Trunk vertebrae of Chamaelinorops (MCZ 154450). A. Dorsal view of the presacral vertebrae, minus
the atlas. B. Ventral view of the presacral vertebrae, minus the atlas. C. Dorsal view of the sacral and anterior

caudal area.

anoles examined have the parapophyses
of cervical vertebra no. 8 enlarged like
those of nos. 5 to 7. In the parapophysial
region of the axis, in both Chamaelinor-
ops and A. garmani, are broad, round,
anterolaterally projecting processes
which make contact with the short,
caudolaterally projecting parapophyses
of the atlas; above and between these
processes the floor of the neural canal of
the axis projects forward as a process
which is composed of the centrum of the
atlas. The base of the atlas ring surrounds
this process, much like the mammalian
odontoid.

Thoracic and Lumbar Vertebrae.
There are 16 thoracic and lumbar verte-
brae in Chamaelinorops (nos. 9-24). The
lumbars are the posterior five of these,
lacking ribs and parapophyses on or near

the centrum (they all have small, round-
ed projections from the portion of the
neural arch below the base of the zyga-
pophysial plate). The thoracic vertebrae
are the anterior 11; the first ten of these
articulate with movable ribs, while the
ribs of the last are fused with the para-
pophyses (see Fig. 5). In Chamaelino-
rops these thoracic and lumbar vertebrae
are essentially similar in non-para-
pophysial structure.

The zygapophysial plates of the thor-
acic and lumbar vertebrae have centrally
located foramina which seem to indicate
the point where the pre- and postzyga-
pophyses fuse to form a solid plate. The
postzygapophysis comprises the ventral
surface of the posterior border of each
plate in Chamaelinorops, which overlaps
the dorsal surface of the anterior border

290

of the plate of the adjacent posterior ver-
tebra. In addition, the postzygapophysis
is composed of the thickened lateral mar-
gin of each plate, the posterior portion of
which extends, as a spinelike process,
underneath the anterolateral border of
the plate of the adjacent posterior verte-
bra. This spine articulates along its
length with a groovelike depression in
the ventral surface of the plate of the
posterior vertebra. Thus the zygapo-
physial articulation of thoracic and lum-
bar vertebrae is an interlocking one, but
not at all similar to the zygosphene-
zygantrum articulation found in some
lepidosaurs and noted by Romer (1956).
In Chamaelinorops this spine-groove
mode of articulation occurs on the post-
zygapophyses of vertebrae nos. 8 to 22,
but not in all individuals examined. No
other anole examined has this mode of
articulation.

Additional articulation between the
thoracic vertebrae is provided by grooves
alongside the anterior face of the neural
arch. These are dorsally directed grooves
extending for a short distance posteriorly
from the anterior face of the vertebra,
along the juncture of the arch and the
zygapophysial plate. Each groove articu-
lates with the ventrally directed medial
edge of the zygapophysial plate of the ad-
jacent anterior vertebra; this latter plate
extends posteriorly from the posterior
face of the neural arch, so that a right
angle is formed with the posterior edge of
the plate. This plate-groove mode of ar-
ticulation occurs on the anterior face of
vertebra no. 3 through the posterior
caudals. Romer (1956) does not mention
the plate-groove mode of intervertebral
articulation, but it is found in all the
anoles examined, including A. garmani.
However, in the anoles other than
Chamaelinorops, the articulation seems
to be the result only of the close proxim-
ity of the postzygapophyses to the neural
arch of the adjacent posterior vertebra.
Only in Chamaelinorops is the groove
geeP and the articulation well develop-
ed.

Of the five lumbar vertebrae, the two

Advances in Herpetology and Evolutionary Biology

that are immediately anterior to the
sacrum are shorter than the anterior lum-
bar vertebrae, and hence have shorter
zygapophysial plates. These plates also
do not extend laterally as far as those of
the anterior vertebrae, and the pre- and
postzygapophyses of each vertebra are
distinct from one another, as they are in
the anterior cervical vertebrae (see Fig.
3C). Although five lumbar vertebrae have
been recognized in Chamaelinorops,
Etheridge (1960) notes that the number
of lumbar vertebrae in anoles is usually
three or four but may be as high as seven.
He also cites intraspecific variation in
this number, and concludes that “the
number of lumbar vertebrae is not a par-
ticularly useful taxonomic character.”
The exact number of true lumbar verte-
brae is determined by counting the
number of presacral vertebrae without
rib articulations. This is usually only pos-
sible in alcoholic specimens. Only one
specimen of Chamaelinorops and none
of the other anoles in this study were ex-
amined for this feature. However, the
two or three vertebrae just anterior to the
sacrum were examined for all species in
this study and the other anoles were
found to differ from Chamaelinorops in
that the width of the vertebral body just
anterior to the sacrum is greater, rather
than less, than the width of the body of
the thoracic vertebrae.

Sacrum. The sacrum in Chamaelinor-
ops is comprised of two vertebrae. The
posterior centrum is slightly longer than
the anterior centrum. Their neural spines
are of the same height as the adjacent
anterior and posterior vertebrae. How-
ever, the neural spine of the posterior
sacral vertebra is twice as long as that of
the anterior vertebra, and thus the
neural-spine plate of the posterior verte-
bra has a greater surface area than that of
the anterior vertebra. In A. garmani, the
neural spine of the posterior sacral verte-
bra is somewhat higher than that of the
anterior one. The zygapophyses of the
sacral vertebrae in Chamaelinorops do
not extend as far laterally as those of the
lumbar vertebrae, and those of the pos-

AXIAL SKELETON OF CHAMAELINOROPS « Forsgaard

terior sacral vertebra are narrower than
those of the anterior one. Their para-
pophyses are massive and extend laterally
so that their tips are about as far from
each other as are the tips of the widest
zygapophysial plates of the thoracic ver-
tebrae. The ends of these parapophyses
appear to fuse distally, providing a broad
articulating surface for the pelvic girdle.
The sacral parapophyses of A. garmani, in
addition to the other anoles examined,
are less massive than those of Chamae-
linorops and are unfused.

Caudal Vertebrae. There are approx-
imately 42 caudal vertebrae in Chamae-
linorops. They are not autotomic. In A.
garmani autotomy occurs and autotomy
septa are plainly visible. In Chamae-
linorops and in A. garmani, chevrons oc-
cur on both anterior and posterior caudal
vertebrae. In Chamaelinorops these
chevrons articulate with the ventrolateral
surface of the centrum just anterior to the
condyle, and extend caudoventrally so
that their tips are ventral to the anterior
portion of the centrum of the posterior
vertebra. The anterior caudal chevrons in
A. garmani are much longer than in
Chamaelinorops, and consequently their
tips are ventral to the posterior portion of
the centrum of the posterior vertebra.
Caudal chevrons of posterior caudal ver-
tebrae in A. garmani are considerably
shorter.

The centrum length of caudal verte-
brae in Chamaelinorops increases pro-
gressively posteriorly from the sacrum, so
that the length of the centrum of caudal
vertebra no. 11 is about three times the
length of the centrum of the anterior
sacral vertebra. The centrum lengths of
caudal vertebra no. 11 and subsequent
posterior vertebrae are similar, so that the
posterior caudal centra, while having
diameters substantially less than that of
caudal no. 5, are only slightly shorter in
length (see Fig. 4). In A. garmani, the
centrum length of caudal vertebrae in-
creases posteriorly to only about twice
the length of the anterior sacral vertebra.

The parapophyses (transverse proc-
esses) of caudal vertebra no. 1 in

291

Chamaelinorops extend laterally, slight-
ly posteriorly, and slightly ventrally.
They are as thick as those of the sacral
vertebrae, and slightly greater in length.
Those of caudal vertebra no. 2 are dor-
soventrally compressed, and their lateral
ends are thickened and extend both
posteriorly and anteriorly (see Fig. 4).
Posterior to caudal vertebra no. 5, these
thickened ends of the transverse proc-
esses become flattened into transverse-
process plates in a plane perpendicular to
that of the neural-spine plates and paral-
lel to the dorsal midline. The lateral
extension of the transverse processes
decreases posteriorly. At about caudal
vertebra no. 12 they are approximately
three-quarters the length of the centrum.
At caudal no. 18 they are half the length
of the centrum and considerably less
massive, and at caudal no. 20 they are

Figure 4. Dorsal view of caudal vertebrae nos. 1-21 of
Chamaelinorops (MCZ 146957). Caudal vertebrae
posterior to no. 21 are essentially similar to it. Prezy-
gapophysial plates are not well developed in this
specimen.

Abbreviations: see Figure 1.

292

merely longitudinal ridges on _ the
ventrolateral surface of the centrum (see
Fig. 4). In contrast, none of the other
anoles examined had any expansions on
their caudal transverse processes. The
transverse processes on the autotomic
caudal vertebrae of A. garmani are all
thin and anterolaterally directed, unlike
the large laterally directed processes in
Chamaelinorops. Furthermore, while
Etheridge (1967) asserted that the anter-
olaterally directed caudal transverse
processes of beta-anoles like A. garmani
are “unusual in being terminally bifur-
cate in the vertical plane,” no such bifur-
cation was found in the two specimens of
A. garmani examined.

The neural spines of the caudal verte-
brae in Chamaelinorops continue the
thoracic and lumbar trend of bearing flat
plates. Those neural-spine plates in-
crease in size between caudals no. 1 and
no. 3, in accordance with the increase in
centrum length. This large size is main-
tained further posteriorly than is the size
of the transverse processes. The neural
spines of the caudal vertebrae of A. gar-
mani do not bear neural-spine plates and
are much higher, in relation to the cen-
trum diameter, than those of Chamae-
linorops, and are also much more poster-
iorly directed.

The caudal zygapophyses in Chamae-
linorops remain large but, with the in-
crease in centrum length, lose their
platelike structural contiguity as the pre-
and postzygapophyses become distinct
from one another. The prezygapophyses
of caudal vertebra no. 2 extend further
laterally than do the postzygapophyses
with which they articulate, and the later-
al portion of the prezygapophysis that is
not overlapped is upturmed around the
lateral margin of the postzygapophysis.
Posteriorly from caudal no. 2, it is up-
turned more, so that by caudal no. 4, the
prezygapophyses are lengthened and
their lateral edges, which are uptumed,
support long flat plates in a plane parallel
to and below that of the neural-spine
plate. These prezygapophysial plates are

Advances in Herpetology and Evolutionary Biology

abruptly lost posterior to caudal vertebra
no. 8 in specimen MCZ 146958. This
trait, however, is subject to intraspecific
variation, as the lateral edges of the pre-
zygapophyses of the anterior caudal ver-
tebrae in specimen MCZ 146957 (see
Fig. 4) are thickened and upturned but
do not bear such plates. A similar prezyg-
apophysial plate does not occur in the
anterior caudal vertebrae of A. garmani
nor the other anoles examined.

In his analysis of the caudovertebral
structure of all lizards, Etheridge (1967)
makes almost no mention of neural
spines or chevrons. However, his evalua-
tion of transverse processes is taxonomi-
cally useful and thorough, except that he
lumps Chamaelinorops in a group with
beta-anoles like A. garmani, characteriz-
ed by long transverse processes whose
orientation changes sequentially from
posterolateral “to lateral then acutely
anterolateral.” Such a description of the
caudovertebral structure of Chamaelin-
orops ignores its major _ structural
uniqueness as compared with other
anoles and thus is not adequate to allow
acceptance of his association of Chamae-
linorops with typical beta-anoles.

Ribs. There are 15 pairs of ribs in
Chamaelinorops. Each pair of ribs, ex-
cept the most posterior pair, consists of
two parts: the curved dorsal bony part
and the ventral cartilaginous section,
which in the ribs of vertebrae nos. 9 to 18
forms a “‘parastemum.” The fused ribs of
vertebra no. 19 have no cartilaginous
parts associated with them. There are
four pairs of cervical ribs, the first pair
being about three-quarters the length of
the second pair, which are about half the
length of the third pair. The fourth pair is
about the same length as the third, and
the distal ends of each of these pairs ap-
pears to be in close association with the
scapulocoracoid. The anterior three pairs
of cervical ribs articulate with the enlarg-
ed parapophyses of vertebrae nos. 5 to 7,
and the capitulum of these ribs is corres-
pondingly enlarged, as is each rib’s shaft
diameter. The posterior pair of cervical

AXIAL SKELETON OF CHAMAELINOROPS - Forsgaard

ribs articulate with the parapophyses of
vertebra no. 8, which are similar to those
of the thoracic vertebrae, and conse-
quently the capitulum size and average
shaft diameter of each of the posterior
pair of cervical ribs are much closer to
that of the thoracic ribs than to that of the
anterior cervical ribs. Each of the anterior
two pairs of cervical ribs bears a small,
flat, triangular process on its lateral sur-
face, with the point of the triangle direct-
ed posterolaterally (see Fig. 5).

The cartilaginous tips of the anterior
two pairs of thoracic ribs in Chamaelin-
orops articulate directly with the ster-
num. The third, fourth, and fifth pairs of
thoracic ribs articulate indirectly with the
sternum via a cartilaginous xiphistemum.
This sternal: xiphisternal formula of 2:3
in the one specimen examined is counter
to the claim of Etheridge (1960) that all
anolines except Phenacosaurus (2:2)
have a formula of 3:2. The sixth, seventh,
eighth, and ninth pairs of thoracic ribs are

KES pooo

«

loooes

Figure 5. Ribs and inscriptional ribs of Chamaelinor-
ops (schematic) (MCZ 146960).

293

attached inscriptional ribs, with each pair
meeting at the midventral line to form an
anteriorly directed cartilaginous chevron.
Posterior to the cartilaginous chevron of
the ninth thoracic rib pair is one free
inscriptional rib chevron, which does not
extend to the ventral ends of the cartilagi-
nous tips of the bony elements of the
tenth pair of thoracic ribs, to which it cor-
responds. This fixed-chevron to floating-
chevron ratio of 4:1 in the one specimen
examined differs from the ratios of 4:2
and 3:3 reported for Chamaelinorops by
Etheridge (1965), indicating that intra-
specific variation in that ratio may not be
as conservative as he asserted. The ribs of
the eleventh thoracic pair are the shortest
ribs in the Chamaelinorops column, be-
ing about the same length as the para-
pophyses to which they are fused (see
Fig. 5).

DISCUSSION

The axial skeleton of Chamaelinorops
is radically different from that of Anolis
garmani and the other anoles examined.
The caudovertebral transverse processes
of beta-anoles like A. garmani are slen-
der and are not expanded at their distal
ends. Those of Chamaelinorops, in con-
trast, are thick, and those of the anterior
caudals posterior to caudal no. 5 bear dis-
tal expansions comprising transverse-
process plates. These differences indi-
cate oversimplification in the morpho-
logical and taxonomic analysis of
Etheridge (1960, 1967) insofar as he de-
scribes a single type of caudovertebral
structure purportedly shared by Chamae-
linorops and beta-anoles. The caudal
vertebrae of Chamaelinorops also differ
from those of all other anoles in the large
size of their neural-spine plates and in
the presence of prezygapophysial plates
in some specimens of Chamaelinorops.

The precaudal vertebrae of Chamae-
linorops are also unique in several re-
spects. Specializations of the precaudal
vertebrae of Chamaelinorops include: 1)

294

presence of “wing-like’” zygapophysial
plates; 2) greater zygapophysial width; 3)
presence of spine-groove intervertebral
articulation; 4) more highly developed
plate-groove intervertebral articulation;
5) larger neural-spine plates; and 6) ap-
parent fusion of the distal ends of the
sacral parapophyses.

These specializations raise questions
of function which remain unanswered.
The authors review of other reptilian
vertebrae has been far from exhaustive,
but zygapophysial plates similar to those
of Chamaelinorops have previously been
noted in heavy-bodied arboreal snakes
such as Synophis bicolor (Johnson, 1955).
He speculated that such plates strength-
en the long epaxial musculature through
greater area for muscular insertion and
assist such snakes in “‘the habit of extend-
ing the unsupported body considerable
distances in climbing from hold to hold.”
Therefore, it is possible that Chamae-

Advances in Herpetology and Evolutionary Biologi
S Y SY

linorops might use its vertebral special-
izations to assist in supporting and lift-
ing the forward portions of its body in
such activities as jumping. However, this
has not been functionally studied.
Moermond (1974) found that certain
morphological parameters (ratio of tail
length to snout-vent length and ratio of
humerus length to femur length) cor-
related with the percentage of jumps in
total movements of seven sympatric
species of Anolis. Measurements of these
parameters in preserved specimens of
Chamaelinorops (Table 1) predict that
Chamaelinorops might be a “jumper.”

Limited field and laboratory observa-
tions of Chamaelinorops do not support
the hypothesis that it might be a special-
ized “jumper.” However, adequate data
are still lacking.

Of interest are observations by E. E.
Williams and S. Case (personal com-
munication) regarding aggressive male-

TABLE 1. MEASUREMENTS OF CHAMAELINOROPS.

snout-vent tail

Species length length
Chamaelinorops MCZ # (mm) (mm)
146951 41.28 78.41

56143 39.60 85.75

146956 37.91 83.59

146953 36.44 80.44

146952 32.66 56.71

57728 31.79 67.70

126708 26.59 61.85

146955 26.70 53.68

146954 24.74 47.75

74594 22.36 43.12

38253 22.00 46.38

Chamaelinorops
(mean values)

. semilineatus

. koopmani

. monticola

. hendersoni (adult)
. hendersoni (juv.)

. distichus (in trees)
coelestinus

. distichus (on rocks)
. cybotes

x * X RR RR HX
bP PSP > LS >

*Data from Moermond (1974).

Length Length % jumps
femur humerus Ratio Ratio of total
length length tail: | humerus: movements
(mm) (mm) sn-vent femur in field
13.34 7.30 1.899 0.547 —
11.48 6.35 2.165 0.553 —
11.87 6.34 2.205 0.534 —
10.94 6.46 2.209 0.590 —

8.74 4,42 1.736 0.506 —
9.42 6.60. 2.130 0.701 —
7.81 4.75 2.326 0.608 —
7.56 4.14 2.049 0.548 —
7.18 4.44 1.930 0.618 —
5.98 3.31 1.928 0.554 —
6.49 3.23 2.108 0.498 —
— — 2.062 0.569 ?

_— — 2.4 0.6 60
— — 2.05 0.57 59
— — 2.3 0.56 57
— — 2.55 0.6 46
— — 2.55 0.6 43
— — 1.35 0.8 Si
— = 2.05 0.8 26
— — 1.35 0.8 16
— — 1.8 0.7 9

AXIAL SKELETON OF CHAMAELINOROPS : Forsgaard

male interactions of Chamaelinorops.
These occurred in restrictive habitats in
captivity. Williams noted that the “head
is carried far backward past the vertical to
display the small bicolor dewlap. The
head is then waved slowly several times.
No push-ups or head bobs” were noted,
and “the dewlap is held extended
throughout.” Case noted that the head
was raised at least 90° from the horizon-
tal axis” and that very slow head-bobs
accompanied dewlap extension. This ex-
treme displacement of the head during
dewlapping behavior, possibly the great-
est vertical displacement of the head
among anoles, may require a firmly
anchored epaxial musculature and thus
may help explain the function of the ver-
tebral specializations of Chamaelinor-
ops, but tests of this hypothesis have yet
to be made.

ACKNOWLEDGMENTS
I am indebted primarily to my under-
graduate thesis advisor, Emest E.

Williams, for loan of specimens and
library material and for financial assist-
ance, instruction, and __ inspiration
throughout the course of this study. The
author is also grateful to Benjamin
Shreve, Catherine McGeary, Patricia
Haneline, José Rosado, Susan Rhodin,
Anders Rhodin, A. W. Crompton, William
Fink, James Lazell, Christopher Kelley,
Mark Plotkin, Ross Johnston, Charles
Gougeon, William Amaral, Russell Mit-
termeier, Anthony Russell, Franklin Ross,
Susan M. Case, George Lauder, Lisa
Moeller, and Timothy Moermond. IIlus-
trations were prepared by Laszlo Meszoly
and the author.

APPENDIX: SPECIMENS EXAMINED

Four complete adult skeletons of
Chamaelinorops barbouri were
examined for this study. All had snout-
vent lengths of approximately 36 mm.

Alpha-anoles examined: Phenacosaurus heter-
odermus: MCZ 145323; Chamaeleolis porcus:
MCZ 131918; Anolis latifrons: MCZ 77410; A.
equestris: MCZ 131609; A. coelestinus: MCZ

295

144795; A. carolinensis: MCZ 153516; A. angusti-
ceps: MCZ 59245; A. baleatus: MCZ 7831; A.

bimaculatus: MCZ 28717; A. cybotes: MCZ
134020.

Beta-anoles examined: Chamaelinorops
barbouri: MCZ 38253, 56143, 57728, 74594,

126708, 146951-60, 154450; Anolis biporcatus:
MCZ 24396; A. fuscoauratus: MCZ 140041; A.
chrysolepis planiceps: MCZ 43859; A. nebulosus:
MCZ 133994; A. onca: MCZ 139352; A. garmani:
MCZ 7630, 140048; A. valencienni: MCZ 145321;
A. ophiolepis: MCZ 11181; A. sagrei: MCZ 10498.

LITERATURE CITED

CocHRAN, D. M. 1928. A new species of Chamae-
linorops from Haiti. Proc. Biol. Soc. Wash., 41:
45-47.

___. 1941. The herpetology of Hispaniola. Bull.
U.S. Nat. Mus., (177): vii + 398 pp.

ETHERIDGE, R. E. 1960. The relationships of the
anoles (reptilia: sauria: _Iguanidae): an inter-
pretation based on skeletal morphology. Ann
Arbor, Michigan, University Microfilms, Inc.,
236 pp.

____. 1965. The abdominal skeleton of lizards in the
family Iguanidae. Herpetologica, 21: 161-168.

____. 1967. Lizard caudal vertebrae. Copeia, 1967:
699-721.

HOFFSTETTER, R., AND J. P. Gasc. 1969. Vertebrae
and ribs of moder reptiles, pp. 201-310. In C.
Gans, A. d’A. Bellairs, and T. S. Parsons (eds.),
Biology of the Reptilia, Vol. I. London,
Academic Press.

JOHNSON, R. G. 1955. The adaptive and phylogenet-
ic significance of vertebral form in snakes.
Evolution, 9: 367-388.

MOERMOND, T. C. 1974. Patterns of habitat utiliza-
tion in Anolis lizards. PhD thesis, Harvard
University. Cambridge, Massachusetts, 150

pp.

Romer, A. S. 1956. Osteology of the Reptiles.
Chicago, University of Chicago Press, 772 pp.

SCHMIDT, K. P. 1919. Descriptions of new amphib-
ians and reptiles. Bull. Am. Mus. Nat. Hist., 41:
523-524.

SCHWARTZ, A., AND S. J. INCHAUSTEGUI. 1980. The
endemic Hispaniolan lizard genus Chamae-
linorops Schmidt. J. Herpetol., 14: 51-56.

SCHWARTZ, A., AND R. THOMAS. 1975. A Check-list of
West Indian Amphibians and Reptiles. Pitts-
burgh, Carnegie Museum of Natural History,
216 pp.

Tuomas, R. 1966. A reassessment of the herpeto-
fauna of Navassa Island. J. Ohio Herpetol.
Soc., 5: 73-89.

WILLIAMS, E.. E. 1974. A case history in retrograde
evolution: the onca lineage in anoline lizards.
I. Anolis annectens new species, intermediate
between the genera Anolis and Tropidodac-
tylus. Breviora Mus. Comp. Zool. No. 421 pp.
1-21.

Variation in the Left Lung and Bronchus
Of Thamnophis sirtalis parietalis

THOMAS S. PARSONS!
LUDMELA DJATSCHENKO?

ABSTRACT. Ina sample of 55 Thamnophis sirtalis,
the tracheal attachment of the left lung varied
greatly. A distinct bronchus was present in 12 and
absent in 13. Another 21 showed bronchial struc-
ture ventrally but not dorsally. Nine were damaged
and unclassified. We were unable always to recog-
nize cartilaginous bronchial rings until the material
was stained with toluidine blue. The size of the left
lung varied significantly with the sex of the snake,
and the relative size of the left lung is not a useful
measure. Taxonomic conclusions should not be
based on the examination of respiratory structures
in unstained material or in only one or a few speci-
mens.

INTRODUCTION

The usefulness of the structure of the
respiratory system in the classification of
snakes was suggested by Cope (1900) and
more recently has been championed by
Underwood (1967). However, there are
few detailed studies on the lungs of
snakes, and many workers simply note
whether a left lung is present or absent
and, if present, its length. Beddard (1906
and other papers), Brongersma (1957),
Butler (1895), and Cope (1894 and 1900)
generally provided more information and
they studied many species; in most cases,
however, only one or two specimens of
each were studied. Bergman (1952, 1958,
and other papers) usually studied a large
number of specimens of each species, but
noted almost nothing about the lungs. We

12 Department of Zoology, University of Toronto,
Toronto, Ontario, Canada M5S IAI.

decided to investigate variation in the
left lung of the garter snake, Thamnophis
sirtalis, to test the validity of conclusions
based on observation of a single speci-
men.

Earlier reports on the presence or
absence of a separate left bronchus are
contradictory, and therefore we concen-
trated on this feature. Cope (1900: 695—
696) stated that in the Colubroidea
(roughly equivalent to the Colubroidea of
most authors or to the Caenophidia, ex-
cluding the Viperidae, of Underwood,
1967), the left lung is usually present,
though very small, and is connected to
the right by a “foramen, which perforates
the tracheal cartilage . . . . It is sometimes
connected to the dorsal [= right] lung by
a short tube, in which cartilaginous half
rings are seen in but two of the genera
examined, viz Heterodon and Conophis.”
Thus he concluded that a left bronchus
was absent in most forms, including
Thamnophis (= Eutaenia of Cope) sir-
talis parietalis. Brongersma (1957: 303),
on the other hand, says of the Colubridae
(equivalent to the Dipsadidae, Homa-
lopsidae, Natricidae, and Colubridae of
Underwood, 1967) that “although the left
bronchus is very short, it is usually dis-
tinct.” He also studied Thamnophis
sirtalis parietalis and did not note it as
exceptional in this regard.

Part of the problem here may be in the
definition of a bronchus, usually con-
sidered to be the single tube connecting
a lung to the trachea. In snakes the divi-

}
;

LEFT LUNG AND BRONCHUS OF THAMNOPHIS « Parsons and Djatschenko

sion between the trachea, the right
bronchus, and the lung is obscured by
the frequent occurrence of a tracheal
lung, i.e., pulmonary tissue in the
trachea. We have adopted a fairly simple
definition: a left bronchus is any more or
less tubular connection between the left
lung and the trachea, right lung, or right
bronchus that contains cartilaginous
rings and no alveolar tissue. Admittedly
this is not a perfectly clear-cut distinc-
tion, but the presence or absence of carti-
laginous rings and of alveolar tissue are
determinable features and ones in which
we found considerable variation.

We found it very difficult to determine
the structure simply by inspection with a
dissecting microscope. The cartilaginous
rings are small and could not always be
recognized without staining. Therefore,
we stained all our specimens except for a
few that were badly damaged during the
initial dissection.

MATERIALS AND METHODS

This study is based on 55 specimens of
the red-sided garter snake, Thamnophis
sirtalis parietalis, 37 males and 18 fe-
males, ranging from 300 to 500 mm in
snout-vent length. All were obtained
from Nasco Educational Materials,
Guelph, Ontario, and preserved in
Nasco-Guard; some were later transferred
to 70 percent ethanol. Specimens were
dissected and studied under a stereo-
scopic microscope.

To determine the distribution of carti-
laginous rings in the bronchi, we dis-
sected free and removed the section of
the respiratory tract for approximately 15
mm on either side of the attachment of
the left lung. This section was then
stained with toluidine blue, following
the method given by Hildebrand (1968).
The washing was done in tap water, and
the specimens remained in the stain for
six days. Destaining, as outlined by
Hildebrand, was accomplished in two to
three days. This process left the cartilage

297

a bright dark blue and all other tissues
pale gray. Destaining continued very
slowly in a mixture of absolute alcohol
and benzene or in the 70% alcohol in
which the specimens were stored.

OBSERVATIONS

All of the snakes studied had two
lungs, a functional right and a vestigial
left one (Fig. 1). The right lung begins
dorsal and just anterior to the posterior
tip of the heart, with the right lateral por-
tion of the lung as the most anterior point.
The anterior surface of the right lung
slopes posteriorly for several millimeters
to the anterior end of the left lateral
border. The trachea runs along the ven-
tral aspect of the right lung, midway
between its right and left lateral borders,
for approximately 10 mm, making the
anterior portion of the lung appear
somewhat bilobed. The trachea then
disappears into the substance of the lung,
but in some cases its dorsal wall contains
alveolar tissue (Fig. 2). The functional
respiratory tissue of the lung gradually
becomes more attenuated posteriorly,
leaving a delicate membranous air sac
which ends in the vicinity of the right
gonad.

The vestigial left lung is located
maximally 3 mm anterior to or at the level
of entrance of the trachea! into the sub-
stance of the right lung. It can be over-
looked upon initial examination because
of its small size and close adherence to
the left side of the posterior vena cava.
The shape of the left lung varies con-
siderably. Triangular, oval, and circular
lungs predominate and are encountered
with about equal frequency; vermiform,
pouchlike, and irregularly shaped lungs
occur less frequently. The locations and
lengths of both right and left lungs,

!The trachea, as used here and in most descrip-
tions of snakes, includes, posterior to the entrance
of the left lung, the right bronchus.

298 Advances in Herpetology and Evolutionary Biology

expressed as percentages of snout-vent
length, are given in Table 1.

The greatest variation in the anatomy
of the respiratory organs among speci-
mens involves the attachment of the left
lung to the trachea or right lung. Due to
the delicate consistency of the respira-
tory tissue and its rather colorless trans-
lucent appearance, in addition to the
presence of mesenteries and blood ves-
sels in the area, observations on indivi-
dual snakes are somewhat subjective.
However, in unstained specimens, there
appeared to be four distinct patterns of
attachment of the left lung. As shown

In 23 of the 54 snakes the left lung
seemed to open directly off the trachea;
the alveolar tissue of the lung appeared
to begin immediately adjacent to the
trachea (Fig. 3).

In eight snakes there appeared to be an
intervening tube of membrane between
the actual opening into the trachea and
the start of the alveolar tissue of the lung
(Fig. 4). The membrane ranged from
delicate and mesenterylike to stout and
suggestive of the tissue found between
the cartilaginous tracheal rings. The
length of this membrane varied from 0.1
to almost 1.0 mm.

below, these appearances are often In 12 snakes the attachment appeared
deceiving. different in the ventral and dorsal aspects
TABLE 1. SIZES AND POSITIONS OF LUNGS.*
d and P values for
differences between
Males Females Total males and females

Number of specimens iy 18 55

Snout-vent length (in mm) min. 300 362 300
mean 394 + 30.4 447 + 36.6 411+ 42.8 5.01 (.001)
max. 476 500 500

Position of anterior end min 15.1 14.3 14.3

of right lung mean 17.0£0.81 16.2+£1.02 16.8+0.78 2.85 (.01)
max. 18.9 17.6 18.9

Position of posterior end min 57.1 54.9 54.9

of right lung (including mean Wei as I reese DO) Gil, se SDI 5.39 (.001)

air sac) max. 66.2 63.5 66.2

Relative length of right lung min 39.5 38.6 38.6

(including air sac) as a % mean 45.3+244 42.7+361 444+2.88 2.76 (.01)

of snout-vent length max. 49.8 47.2 49.8

Position of anterior end of min 16.1 15.3 15.3

left lung mean 18.3+0.81 17.3+0.95 18.0 + 0.86 3.30 (.001)
max. 20.0 18.9 20.0

Position of posterior end of min 16.7 15.9 15.9

left lung mean 19.0+0.79 17.88+0.79 186+ 0.78 5.00 (.001)
max. 20.5 19.4 20.5

Relative length of left lung min 0.4 0.3 0.3

as a % of snout-vent length mean 0.7 + 0.22 0.5 + 0.21 0.7 + 0.22 3.64 (.001)
max. 1.3 0.9 1.3

Absolute length of min 0.2 0.1 0.1 1.55 (.10)

left lung (in mm) mean 0.28 + 0.09 025+0.07 0.27 + 0.08
max. 0.5 0.4 0.5

* The positions of lungs are given as percentages of the snout-vent length, with 0 the snout and 100 the vent.

LEFT LUNG AND BRONCHUS OF THAMNOPHIS - Parsons and Djatschenko 299

Figures 1-4. Dissections showing the left lung of Thamnophis before staining with toluidine blue. Anterior is
toward top of page. 1. General view to show the relationships between the two lungs and the trachea. (X 2.8). 2.
Specimen showing alveolar tissue in the trachea (tracheal lung). (X 7). 3. Specimen in which the pulmonary tissue
of the left lung appears immediately adjacent to the trachea; there is no left bronchus. (X 15). 4. Specimen in
which the left lung appears to have a membranous tubular attachment to the trachea; subsequent staining with
toluidine blue showed this attachment to actually be a left bronchus with cartilaginous rings. (X 17).

Abbreviations: A, Alveolar tissue of tracheal lung; L, left lung; R, right lung; T, trachea.

300

of the left lung. In ventral view, nine of
these 12 snakes seemed to have a bron-
chus with cartilaginous rings joining the
left lung to the trachea. The remaining
three appeared to have a heavy mem-
branous tube joining the two. The length
of this bronchus or membrane ranged
from about 0.3 to 1.0 mm. However,
when the left lung was reflected so that
the attachment of its dorsal aspect to the
trachea was visible, the distance between
the trachea and the start of the alveolar
tissue of the lung appeared noticeably
smaller due to the decrease in the length
of the bronchus or membrane as it swung
around to the dorsal aspect of the lung. In
four cases the intervening bronchus or
membrane disappeared completely, leav-
ing the trachea and the alveolar tissue of
the lung apposed.

Eleven of the snakes appeared to have
a left bronchus (under our definition of a
bronchus) joining the left lung to the
trachea. The bronchi ranged in length
from 0.3 to 1.0 mm. In most cases the
separation between the functional respi-
ratory tissue of the left lung and the bron-
chus was abrupt. In a few, the line of
demarcation between the two structures
was less obvious. The left lung of the
fifty-fifth snake was destroyed during
dissection, and is not considered here.

Due to the difficulty of detecting car-
tilaginous rings in unstained material,
the left bronchi of 47 specimens (in eight
this area was destroyed during the initial
investigation) were stained and again
studied without reference to the earlier
observations. The two sets of observa-
tions were then compared. The most
common pattern of attachment of the left
lung was now shown to be the one that
displayed a difference in the structure on
the ventral and dorsal aspects of the left
lung (Table 2; Figs. 5, 6). In all the
snakes the tracheal cartilages were
approximately 0.2 mm wide and about
0.5 to 0.8 mm apart when measured from
the center of one to the center of the ad-
jacent one. In the 21 snakes (11 males
and 10 females) which displayed this pat-

Advances in Herpetology and Evolutionary Biology

tern of attachment, the distance between
tracheal rings decreased slightly as the
left lung was approached and the rings
became less regular in their arrangement.
There appeared to be a distinct left bron-
chus branching off the trachea and run-
ning parallel to and in close contact with
the underlying trachea. However, when
the left lung was viewed from its dorsal
aspect, the amount of cartilage visible in
the area of attachment was greatly re-
duced, and often none could be seen.
The actual structure in this area was,
therefore, an elevation of a portion of the
trachea, which resembled the top of a
tube sliced lengthwise and which joined
the anterior border of the opening in the
trachea to the ventral surface of the left
lung. The dorsal surface of the left lung
was in more direct contact with the
posterior edge of the opening into the
trachea. This type of left tracheopul-
monary junction was found in eight of
the snakes which did not, before staining,
appear to have any bronchus, two which
seemed to have a membranous attach-
ment, and two which appeared to have a
definite bronchus. Nine of the twelve
snakes originally assigned to this cate-
gory were confirmed really to have this
pattem after staining.

The remaining 25 snakes fell into two
categories—those in which the alveolar
tissue of the lamella started immediately
adjacent to the trachea and those in
which there was a definite bronchus. In
13 snakes (11 males and 2 females) the
left lung opened directly into a hole in
the ventral aspect of the trachea, so that
there was only minor disarrangement of
the cartilaginous tracheal rings imme-
diately adjacent to the opening (Fig. 7).
The alveolar tissue of the left lung ex-
tended all the way to the trachea. In 12
snakes (8 males and 4 females) a distinct
left bronchus joined the left lung to the
trachea (Fig. 8). Again there was a
minimal effect of this structure on the
rather uniform arrangement of the cartil-
aginous tracheal rings. The cartilages of
the bronchus did not appear to be con-

LEFT LUNG AND BRONCHUS OF THAMNOPHIS + Parsons and Djatschenko

301

TABLE 2. COMPARISONS OF DIFFERENT MODES OF ATTACHMENT OF THE LEFT LUNG.

Mode of attachment
(for more detailed
descriptions see text)

Estimation from
inspection
of dissection

Observed in specimens
stained with
toluidine blue

males females total males females total

Attached directly to trachea 16 7 23 1 9 13
Membranous, duct like 8 0 8 0) 0 0
attachment

Mode of attachment different 3 9 12 11 10 ot
dorsally and ventrally

Left bronchus present 9 2 iil 8 12
Damaged and omitted from 1 0 1 7 2 9

sample

tinuous with those of the trachea and
usually showed a more irregular pattern.
If there were rings, they were always
thinner and much more closely spaced
than those in the trachea. Often they
were connected by longitudinal bars of
cartilage. In several instances they
formed complete rings around the cir-
cumference of the bronchus, unlike those
of the trachea. In other cases there were
no clear rings, but rather a networklike
arrangement of cartilage. The length of
the bronchus was also less than originally
estimated, 0.8 mm being the maximum.

In no snake was there a purely mem-
branous tube between the trachea and
the left lung as was originally described
in unstained specimens (Table 2).

Two further variations were noted in
the structure of the respiratory tract in
this area. In addition to its attachment to
the trachea, the left lung appeared, in ten
of the snakes (8 males and 2 females), to
share common alveolar tissue with the
right lung. This characteristic is corre-
lated with the type of attachment of the
left lung to the trachea. In half of the
cases (4 males and | female) the left lung
opened directly from the trachea, and in
four instances (3 male and | female) there
was a difference in the method of attach-
ment dorsally and ventrally. In the tenth
specimen (male) the attachment was

destroyed before staining so that definite
confirmation of the structure was impos-
sible. Before staining, however, the
snake appeared to possess a very short
bronchus.

The common alveolar tissue was
always located posterior to, and usually
to the left of, the tracheal attachment of
the left lung and always involved the
dorsal surface of the left lung. Such com-
mon tissue usually formed only a minor
portion of the total alveolar tissue of the
left lung, but was enough to make dissec-
tion of the left lung from the right lung
difficult. When the two lungs were
separated, the area where they had been
fused was always evident on both lungs.
In one case the left lung, on its ventral
aspect, opened directly into the trachea
via a tubelike elevation resembling half a
bronchus cut lengthwise. Posterior to this
on the dorsal aspect of the left lung, the
left lung opened directly into the sub-
stance of the right lung rather than into
the trachea, which by this point had also
entered the substance of the right lung.
Thus, posteriorly the left lung appeared
to be a direct extension or lobe of the
right lung.

Well-developed tracheal lungs were
noted in only two specimens, one male
and one female (Fig. 2). In both cases the
respiratory tissue of the tracheal lung was

302 Advances in Herpetology and Evolutionary Biology

Figures 5-8. Dissections of the left lung of Thamnophis after staining with toluidine blue. Anterior is toward top of
page. 5. Specimen in which dorsal and ventral sides of attachment vary. This ventral view shows a short,
bronchus-like connection. (X 12). 6. Same specimen as in Fig. 5, with the left lung folded anteriorly to show the
dorsal side of its attachment to the trachea. The attachment appears to be direct with no intervening bronchus. (X
12). 7. Specimen in which pulmonary tissue of left lung is directly adjacent to the trachea; there is no left bronchus.
(X 7). 8. Specimen with a distinct left bronchus. (X 12).

Abbreviations: B, Left bronchus; L, left lung; R, right lung; T, trachea.

LEFT LUNG AND BRONCHUS OF THAMNOPHIS - Parsons and Djatschenko

an anterior extension of the alveolar tis-
sue of the right lung. This tissue was
located in the area normally occupied by
the membrane separating the ends of the
discontinuous tracheal rings, i.e., on the
dorsal aspect of the trachea. The tracheal
lung tissue tapered out 7 mm (female) or
8 mm (male) anterior to the point at
which the trachea entered the substance
of the right lung. In the male the left lung
opened directly into the ventral surface
of the trachea; in the female there was a
hint of bronchial structure on the ventral
surface of the attachment of the left lung,
but none dorsally.

DISCUSSION

Contrary to the studies of Cope (1874,
1900) on the same snake, Thamnophis
sirtalis parietalis, we found cartilagi-
nous rings in bronchus-like structures in
most specimens. Cope, however, does
not mention any staining procedure,
and apparently simply inspected un-
stained dissections. The same is true of
the observations by Beddard (1906),
Brongersma (1957), and Butler (1985).
We question the accuracy of the earlier
reports on the presence or absence of car-
tilaginous bronchial rings in snakes, at
least in small ones such as Thamnophis,
since we were often unable to tell if such
rings were present without toluidine
blue staining (Table 2). Our material was
almost certainly at least as well preserved
for our purposes as was that used by
earlier workers who studied museum
specimens fixed in alcohol. Certainly, if
the presence or absence of a bronchus is
to be used as a taxonomic character, then
staining for cartilages appears important
and probably essential.

In Thamnophis there is considerable
variation in the mode of attachment of
the left lung. Whether such variation is
common in other species is not known.
Clearly, however, broad surveys like
Brongersma’s (1957), which are based on
only one or two specimens of most

303

species, are open to considerable ques-
tion. Series of specimens of each taxon
considered must thus be studied before
reliable conclusions can be reached.

Table 1 compares various measure-
ments of the size and position of the
lungs in male and female specimens. In
every case the difference between the
sexes is significant, in all but one case at
the 0.01 level or better (all based on the
d test; Bailey, 1959). The differences in
the lengths and positions of the lungs be-
tween the sexes are not simply reflec-
tions of the difference in lengths of the
specimens, as the left lungs of females
are both relatively and absolutely smaller
than those of males. Thus the sex of the
specimens must be checked in any study
comparing the size and position of the
lungs of different species of snakes.

Another problem is that most previous
workers have used some _ relative
measure of the size of the left lung such
as a percent of the total body length
(Brongersma, 1957; Thorpe, 1975). Un-
fortunately, our data reveal no correla-
tion between snount-vent length and
length of the left lung (correlation coef-
ficent for males = 0.108 and for females
= 0.29; the coefficents for relative
lengths of the left lung are no better—
0.09 for males and 0.10 for females). Thus
in our sample, the absolute, not the rela-
tive, size should be the more meaningful
figure, but obviously it cannot be used in
comparing the lungs of snakes of very dif-
ferent sizes.

The problems noted above do not
mean that the structure of the lungs can-
not be used as a taxonomic character.
Some of the cases in which it has been
used involve such major differences be-
tween taxa that intraspecific variation is
probably not critical, e.g., the example of
Gonyosoma discussed by Underwood
(1967). In most cases, however, detailed
studies of many specimens of each form
are required. For example, Cope (1900)
and Underwood (1967) reported the
general or at least usual absence of a left
lung in the Viperinae, while its presence

304

was noted by Butler (1895) in Cerastes
cornutus and by Brongersma (cited by
Underwood, 1967) in Causus. The char-
acter may well be variable in many mem-
bers of the subfamily.

Our results do not mean that the lungs
are highly variable in all snakes. Thorpe
(1975) found no significant variation
according to size, sex, or geographic
origin in the length of the left lung of
Natrix natrix. He used a different system
of measurement, using scales instead of
absolute measurement, but this should
not change his results greatly. Different
species of snakes, even of natricids, must
show considerably different degrees of

variability.

ACKNOWLEDGMENTS

We wish to thank M. A. Adair, M. N.
Loucks, I. H. McCorquodale, and S. T.
Richter for technical help and typing.
The work was begun by a former student,
J. E. Wilkinson, whose report was very
useful. The photographs were taken by
N. R. Hatton. C. S. Churcher and M. C.
Parsons critically read the manuscript
and made helpful suggestions as did two
anonymous reviewers. The work was
supported by Grant A1724 from the
Natural Sciences and Engineering
Research Council Canada.

Advances in Herpetology and Evolutionary Biology

LITERATURE CITED

BAILEY, N. T. J. 1959. Statistical methods in biology.
New York, John Wiley and Sons, x + 200 pp.

BEDDARD, F. E. 1906. Contributions to the knowl-
edge of the vascular and respiratory systems in
the Ophidia, and to the anatomy of the genera
Boa and Corallus. Proc. Zool. Soc. London,
1906: 499-532.

BERGMAN, R. A. M. 1952. L’anatomie du genre Ptyas
a Java, Riv. Biol. Coloniale, 12: 5-42.

__. 1958. The anatomy of the Acrochordinae.
Proc. Koninklijke Ned. Akad. Wetensch., C,
61: 145-184.

BRONGERSMA, L. D. 1957. Notes upon the trachea,
the lungs, and the pulmonary artery in snakes.
I-II. Proc. Koninklijke Ned. Akad. Wetensch.,
C, 60: 299-313.

BUTLER, G. W. 1895. On the complete or partial
suppression of the right lung in the Amphis-
baenidae and of the left lung in snakes and
snake-like lizards and amphibians. Proc. Zool.
Soc. London, 1895: 691-712.

Corr, E. D. 1894. On the lungs of the Ophidia.
Proc. Amer. Phil. Soc., 33: 217-224.

____. 1900. The crocodilians, lizards, and snakes of
North America. Rep. U.S. Nat. Mus., 1898: 153—
1270.

HILDEBRAND, M. 1968. Anatomical preparations.
Berkeley and Los Angeles, Univ. California
Press, viii + 100 pp.

THORPE, R. S. 1975. Quantitative handling of char-
acters useful in snake systematics with partic-
ular reference to intraspecific variation in the
ringed snake Natrix natrix L. Biol. J. Linn.
Soc., London, 7: 27-43.

UNDERWOOD, G. 1967. A contribution to the classifi-
cation of snakes. Brit. Mus. (Nat. Hist.),
London, x + 179 pp.

On the Dorsal Armor of the Crocodilia

FRANKLIN D. ROSS!
GREGORY C. MAYER?

ABSTRACT. A new method for counting the scales
of the dorsal armor of crocodilians is introduced.
Based on the one-to-one correspondence between
vertebrae and transverse scale rows, it allows com-
parisons between species on the basis of homol-
ogous scale rows. Scale rows are counted away from
the evolutionarily conservative sacro-caudal junc-
ture in precaudal and caudal series, approaching
the more variable anterior and posterior ends of the
dorsal armor from a fixed central point. Primitively,
crocodilian dorsal armor consists of 27 transverse
precaudal rows of two scutes each, as is found in
protosuchians. Changes occurring in the evolution
of the modem crocodilians are 1) broadening of the
trunk armor; 2) breakup of the ancestral scute pair
into many elements per transverse row; 3) loss of
the anteriormost transverse rows; and 4) narrowing
the armor at the cervico-thoracic juncture. Descrip-
tions of the dorsal armor and its variation in all
living species are given. The most frequent kinds of
variation in the living crocodilians, both among and
within species, are the deletion of transverse rows
and the compounding of transverse rows in the
cervical region. Gavialis is the most primitive living
species, departing least from the mesosuchian type
of armor, while Crocodylus acutus and C. porosus
are the most derived, showing great reduction from
the primitive eusuchian pattern in both trunk and
cervical regions.

INTRODUCTION

Nothing is more fugitive than the forms of croco-
diles. —Geoffroy, in Cuvier, 1831.

Our methodological and conceptual
approach to the dorsal armor of the living
Crocodilia differs from that of most
previous authors (Cuvier, 1807; Dumeéril
and Bibron, 1836; Boulenger, 1889;

12Museum of Comparative Zoology, Harvard
University, Cambridge, Massachusetts 02138,
U.S.A.

Deraniyagala, 1939; Wermuth, 1953;
Brazaitis, 1973), but follows directly from
insights available in the morphological
and paleontological literature (Huxley,
1859; Colbert and Mook, 1951; Nash,
1975). Since the latter part of the eight-
eenth century, crocodile taxonomists
have described dorsal armor from the
anterior end first. The traditional meth-
odology involves counting scales (scutes)
from occiput to tail tip in terms of
postoccipitals, nuchals, dorsals and
caudals. The difficulty with counting in
this way is that the greatest and most
interesting variation occurs in the form of
missing, enlarged and fused transverse
scale rows in the cervical region. This
being the case, the most outstanding
variation is encountered initially, result-
ing in unreliable and oftentimes false
homologies. By not establishing the rela-
tionships among the sacrum, hind legs,
and dorsal armor, various authors have
counted the dorsals and caudals begin-
ning or ending with any of several scale
rows. We find it interesting and hearten-
ing, though, that our conclusions general-
ly agree with those of the traditional
method, and that we have little evidence
to suggest that the current classification
of the living Crocodilia is incorrect.
Our method considers the transverse
scale rows on the body and tail in relation
to the underlying vertebrae, with which
there is primitively a one-to-one relation-
ship. There is always an unarmored
space immediately behind the head in
living species, and there may be one or
two more such spaces near the anterior

306

end of the series; but, invariably, there is
a continuous series of transverse rows
from the region of the shoulders, over the
sacrum, and on to the tip of the tail.
These transverse rows of bone and scale
correspond one-to-one with immediately
underlying vertebral elements. Huxley
(1859) first observed the one-to-one cor-
respondence in Caiman crocodilus, and
explained that, “the anterior part of the
inner surface of each of the two middle
scutes is connected by a ligament with
the extremity of the spinous process of a
vertebra; at least this is the case in the
dorsal, lumbar, sacral, and anterior
caudal regions.” Our counts are made
from the evolutionarily conservative
sacro-caudal juncture. We approach the
more variable anterior and posterior ends
of the dorsal armor from a fixed central
point, which is the same for all living and
fossil species.

Each transverse row is intervertebral
in position, the median elements of a row
being attached to the neural spine of the
anterior vertebra, and extending poste-
riorly to overlap the one behind. The
dorsal armor thus exhibits the primary
segmentation which exists prior to
resegmentation in the development of
the vertebral column (Williams, 1959).
Seidel (1979) observed that the blood
supply to the transverse rows of dorsal
armor on the body of Alligator missis-
sippiensis is by means of a series of un-
paired arteries, the dorsal median
arteries, which emerge through the deep
fascia between successive neural spines,
and which also supply the neural spine of
the anterior vertebra. All osteoderms in a
transverse row and the neural spine of
the vertebra anterior to the row receive
blood from a common source. Transverse
dorsal osteoderm rows are thus asso-
ciated with the anterior of the two verte-
brae they overlap.

Huxley (1859) restricted his observa-
tions to the scutes on the body and
anterior part of the tail, and Seidel (1979)
noted that an interruption in the series of
dorsal median arteries occurred in the

Advances in Herpetology and Evolutionary Biology

shoulder region, although it continued
anterior to this. Gadow (1901) also noted
the difference between trunk and cervi-
cal armor in the Crocodilia, stating that,
“the armor of the recent forms consists,
so far as the large scutes are concerned, of
a considerable number of scutes, which
are arranged in transverse rows, each row
corresponding with one skeletal segment
of the trunk proper. Mostly there is a
detached cluster of scutes on the neck.”
Indeed, there is sometimes a detached
cluster of scutes on the neck, particularly
in Crocodylus, but other forms (e.g.,
Gavialis, Tomistoma, Melanosuchus)
possess continuous thoracic and cervical
armor.

As illustrated in Figures 1 and 2,
radiographs and dissection show the one-
to-one relation between vertebrae and
dorsal armor, and that in certain cases
some scutes on the neck overlap two
vertebrae. Dissection confirms Huxley’s
observation of association between
neural spines and osteoderms. There are
normally 26 precaudal vertebrae (24
presacral + 2 sacral) plus a proatlas in all
living and fossil forms examined. For
further discussion of vertebral morphol-
ogy, see Higgins (1923), Hoffstetter and
Gasc (1969), and Romer (1956).

MATERIALS AND METHODS

Our taxonomy of the living Crocodilia
follows Wermuth and Mertens (1961) at
the generic and specific level. We do not
consider subspecies. We have followed
Boulenger (1889) in ordering the genera
as follows: Gavialis, Tomistoma, Croc-
odylus, Osteolaemus, and finally Alliga-
tor and the caimans.

Determinations were made on the
basis of cranial characters, coloration, and
geographic provenance. We examined
skins, mounted, and fluid preserved
specimens of all living species. The
Appendix is a list of specimens examined
and reported on in this paper. In addi-
tion, we incorporate some data from the

DORSAL ARMOR OF CROCODILIA : Ross and Mayer 307

Figure 1. Tracings of radiographs showing endo-and exoskeletal systems in: A. Crocodylus siamensis, MCZ
3716, lateral view of the neck and trunk region. B. Osteolaemus tetraspis, MCZ 3589, dorsal view of the lumbar,
sacral, and anteriormost caudal regions. C. Osteo/aemus tetraspis, MCZ 2017, lateral view of the anterior thoracic
and all of the cervical series.

308

literature; and, in the case of Crocodylus
intermedius, we have relied in part on
photographs and a personal communica-
tion regarding the type. The type of
Protosuchus richardsoni (AMNH 3024)
was also examined.

Specimens of Osteolaemus tetraspis,
Tomistoma _ schlegelii, Melanosuchus
niger, Paleosuchus trigonatus, P. pal-
pebrosus, Caiman crocodilus, C. latirost-
ris, Alligator sinensis, A. mississippien-
sis, and nine species of Crocodylus
were radiographed to determine the rela-
tionship between the dorsal armor and
the skeleton, and for vertebral counts.

Additional vertebral counts were
performed on skeletons.
METHODOLOGY

Although the tradition has been to
count both dorsal armor and vertebrae
from the cranium, we count away from
the sacro-caudal juncture in both pre-
caudal (PC) and caudal (C) series (Figs. 2,
4). PC 1, the posteriormost transverse
precaudal scale row, hangs posteriorly
from the tip of the neural spine of the
posterior sacral vertebra. PC 1 can be
identified unambiguously in fluid pre-
served specimens by a simple dissec-
tion. An incision is made along the
medial, posterior, and lateral borders of
three transverse rows on one side of the
body: the two rows nearest the posterior
edge of the iliac crest and the one an-
terior to them. The skin and osteoderms
are then reflected anteriorly, and the
musculature dorsal to the level of the
sacral ribs removed, thus revealing the
relationship between the vertebrae and
the osteoderm rows. PC | is the trans-
verse row extending posteriorly from the
neural spine of the posterior sacral, and
dorsal to the postzygapophysis of the
same vertebra. The reflected skin and
osteoderms may be folded back, leaving
the specimen externally unchanged, and
the musculature of the other side of the
body intact for anatomical studies.
Radiographs from dorsal view also allow

Advances in Herpetology and Evolutionary Biology

unambiguous identification of PC 1,
provided the osteoderms are sufficiently
opaque to X-rays (Fig. 1B).

Short of dissection or X-ray, the follow-
ing criteria have been found useful in the
identification of PC 1. First, PC 1 lies
between the posterior blades of the ilia.
The posterior edge of the iliac blades
are overlain by PC 1 or the anterior por-
tion of the first caudal row (C 1). The
location of the blades can be determined
by palpation. Secondly, when held
perpendicular to the body, the long axis
of the femurs passes through PC 2 or the
border between PC 2 and PC 1. Third, in
many species, PC 1 is the first transverse
row to become broader after mono-
tonically decreasing in breadth from mid-
body towards the sacrum (see Figs. 2B,
3C). Finally, the posterior edge of PC 1 is
often at the posterior edge of the hind
limb (the traditional point of demarcation
between dorsals and caudals), but this is
not always the case. When these criteria
are applied simultaneously, PC 1 can
usually be identified unambiguously.
Reference to radiographs and dissected
specimens is always helpful. PC 1 has
been identified by these criteria in most
specimens examined in this study.

Skins and stuffed specimens present
special problems. Determination of the
relationship of the transverse rows to the
variously granulated scales around the
base of the hind limb in intact specimens,
and criterion three above are the best
guides to the identification of PC 1 in
such specimens. Comparison with intact
specimens is helpful.

Posterior to PC 1, the caudal series
consists first of double crest caudals and
then single crest caudals. Double crest
caudals are associated with vertebrae
with transverse processes, and have two
to four keels across. Single crest caudals
are associated with vertebrae having re-
duced or absent transverse processes,
and have a single median keel. The keel
of the single crest series and the outer
keels of the posterior double crest series
form a Y-shaped swimming keel.

DORSAL ARMOR OF CROCODILIA - Ross and Mayer 309

Figure 2. Semidiagramatic illustration of the one-to-one relationship between vertebrae and transverse rows of
dorsal osteoderms as revealed by dissection of Melanosuchus niger, MCZ 17726. Odd numbered elements
stippled in the precaudal (PC) series, and even numbered elements stippled in the caudal (C) series. A. PC 26-C
5 in lateral view. B. PC 26-C 2 in dorsal view. C. C 3 in cross section. D. C 4-C 15 in dorsal view. E. C 16-C 28 in
lateral view. F. C 29-C 38 in lateral view.

310 Advances in Herpetology and Evolutionary Biology

Figure 3. Dorsal armor of three selected crocodilians. A. Protosuchus richardsoni, AMNH 3024, type, after
Colbert and Mook, 1951 (PC 27-C 39 shown). B. Gavialis gangeticus, BMS no #, from the head to the sacro-
caudal juncture; C: Caiman crocodilus, MCZ 3394, full body.

DORSAL ARMOR OF CROCODILIA - Ross and Mayer

On the lateral and ventral surfaces of
the base of the tail, scales may occur in
various configurations which break up
the regular (in the sense of showing
plane symmetry; see Breder, 1947) dis-
position of scales in this area. These are
collectively termed basicaudal irregulari-
ties. Four kinds are recognized: 1) a thin
horizontal strip of granular skin separat-
ing the dorsal from the lateral scalation;
2) one or more thick strips of granular
skin on the dorsolateral surface (Fig. 11);
3) lateral intrusions, in the form of small
vertical arrays of scales, two-thirds or
more of a normal scale in height, on the
postcloacal lateral surface, sometimes
present on one side only (Fig. 4; Ross and
Ross, 1974: Fig. la); and 4) large ventro-
lateral intrusions composed of vertical ar-
rays of scales on the postcloacal ventral
and ventrolateral surfaces (Fig. 4; Ross
and Ross, 1974: Fig. lb). This last in-
creases the number of midventral trans-
verse rows relative to the dorsals.

Anterior to and including PC | is the
continuous and contiguous precaudal
armor (Fig. 4). The continuous armor is
the series of transverse rows begin-
ning with PC | and extending to the row
which is bordered in front by a missing
transverse row. Soft skin may intrude
between two rows without a row being
absent. This makes the armor appear dis-
continuous, although all rows are pres-
ent. The Crocodylus moreletii in Fig. 4
has 16 continuous precaudal rows, PC 17
(and others) being absent. Contiguous
scutes are scutes in a series crossing (or at
least touching) the midline in which ad-
jacent members of the series are in con-
tact. Beginning with PC 1, the number of
contiguous scutes per transverse row in
Figure 4 is 4, 4, 4, 4, 6, 4, 4, 4, 6, 3, 3, 6, 5,
24 2.

Additional characteristics of the trunk
armor are the presence of detached
lateral scutes, sometimes forming a longi-
tudinal row (Fig. 4), the alignment and
extent of development of keels on the
scutes, the uniformity of scute size, and
the presence of left/right asymmetries.

311

Note that even development and align-
ment of keels, uniformity of scute size,
and absence of asymmetry impart the
quality of “regularity” to the armor.
The cervical shield is the cluster of
scutes on the neck. The term does not
refer to particular precaudal rows, as the
rows comprising it and its continuity with
the thoracic armor vary among species.
The nape region is the area between the
occiput and the cervical shield. Vertebral
correspondence in the cervical region is
determined in two ways. In species with
a continuous thoracic and cervical armor,
correspondence can be _ determined
directly by counting, with observation of
individual variation revealing compound
rows. In species with discontinuous
cervical and thoracic armor, comparison
with related species is necessary. For
example, in Crocodylus all species have
two enlarged rows in the cervical shield.
In C. cataphractus (Fig. 5D), and fre-
quently in C. johnstoni (Fig. 6A), the
cervical and thoracic armor is continuous,
the last row before the two enlarged rows
being PC 19. Thus the enlarged rows
begin with PC 20. Individual variation in
C. niloticus (Fig. 6B-D) shows that these
enlarged rows are each compound. As-
suming that these rows are homologous
among members of the genus allows ex-
trapolation to other species. Thus the C.
moreletii in Fig. 4 has two scutes in PC
20 + 21 and four in PC 22 + 23. Uncer-
tain homologies might be resolved by
careful dissection of arteries and nerves.
Analysis of the cervical region requires
comparison of many specimens simul-
taneously. We found photographs useful
in comparing specimens from different
institutions. A marker with specimen
number was placed on PC 12, and both
marker and PC 13 forward photographed
from directly above. Properly marked
photographs proved to be as good as
actual specimens for comparative pur-
poses. We would encourage the collec-
tion of photographs of this kind, as it
allows obtaining sizable samples without
damaging populations of the world’s

3412 Advances in Herpetology and Evolutionary Biology

AEPE region ———§__ Gio
Sons
CY

cervical shield === ©

=

|

continuous and
COMNGUOUS Plese~%
caudal armor

detached lateral \
scutes forming

j

ae

lateral intrusion

ventrolateral

Intrusions
ee

a>
ea

Be
Sess
Me

Sy
F

Figure 4. Some terms used in this paper indicated on Crocodylus moreletii, USNM 71955, topotype. From an
original drawing by Aleta Karstad.

DorSAL ARMOR OF CROCODILIA : Ross and Mayer

dwindling supply of crocodilians. The
photographs can be stored as a record of
the specimens examined in natural pop-
ulations without removing individuals
from those populations.

Meristic characters in reptiles are
generally believed not to change in
postembryonic life (Hecht, 1952; Kerfoot,
1970; Fox, 1975). Our observations sup-
port this. Although we have not studied
large series, examination of embryos of
various ages from a variety of species in-
dicates that there is little or no ontogene-
tic change in embryonic life as well. The
scutes of unpigmented embryos are
somewhat more closely spaced, and thus
May appear contiguous or continuous,
whereas in hatchlings and adults they
would not be in contact. Thus available
ontogenetic data does not help to eluci-
date homology of transverse rows.
Medem (1958a) noted that Crocodylus
intermedius embryos did not differ from
adults in scale counts.

There are two exceptions to the gen-
eral rule of no ontogenetic change. First,
the number of single crest caudals de-
creases due to injury and wear on the tail.
Counts of this character are thus possible
only on undamaged individuals. Second,
and more interesting, the number of
scutes per transverse row in Gavialis
decreases from six to four as an animal
ages. This is discussed in the species
accounts. Deraniyagala (1939) also noted
some ontogenetic changes in Crocodylus
porosus embryos, involving a decrease in
the number of small scales interspersed
among the scutes of the trunk armor.

SPECIES ACCOUNTS

The following accounts of the species
provide brief descriptions of the dorsal
armor of each species, with more detailed
treatment of the anterior region, where
the greatest variation within and among
species occurs. Table 1 gives counts of
the contiguous scutes per transverse row

for PC 1 to PC 18 for all species. In Table

313

2, the number of double and single crest
caudals is given for each species. All liv-
ing crocodilians have a Y-shaped caudal
swimming keel, consisting of the single
crest caudals and the posterior part of the
double crest series. The ways in which
the Y-shaped keel is formed from the keel
rows of the trunk armor as it passes over
the pelvis in different taxa are given in
Table 3. Table 4 records the occurrence
of basicaudal irregularities.

Gavialis gangeticus

In Gavialis (Fig. 3B), the continuous
and contiguous precaudal armor is re-
markably uniform in scute size and
shape, and keel row alignment and
development. There are six scutes per
transverse row in PC 1 to PC 18 in hatch-
lings and young. In older specimens,
there are only four scutes across, appar-
ently because the lateral D-shaped scales
of the very young fail to ossify with the
four more medial scutes. This is unique
in the Crocodilia. There are no detached
lateral scutes on the flanks, and the tho-
racic and cervical armor are continuous.
There are 22 continuous rows of pre-
caudal armor, narrowing anteriorly to two
scutes in PC 22. The nape is naked ex-
cept for two prominent scutes and occa-
sional traces of others.

Tomistoma schlegelii

In Tomistoma (Fig. 5A-C), the contin-
uous and contiguous precaudal armor is
quite regular, with five to eight, usually
six, scutes across at midbody (PC 9-PC
12). Detached lateral elements form an
additional row on each flank. The thor-
acic and cervical armor are continuous
forward to PC 23, and narrowest at PC 19
or PC 20. There are two scutes in PC 18,
19, 20, and 21. Typically, the terminal
row is PC 22+23 compound with two
elements. In some specimens PC 23 is
present as a distinct row or asymmetrical-
ly. Fig. 5A-C shows the range of variation
in this region. Scutes on the nape are ir-

314

Advances in Herpetology and Evolutionary Biology

Figure 5. Anterior end of the dorsal armor, PC 13 forward, with selected scale rows numbered. A. Tomistoma
schlegelii FMNH 206755. B. T. schlegelii, USNM 145587. C. T. schlegelii, USNM 58136; D. Crocodylus cata-
phractus, MCZ 54650.

regularly placed, although sometimes
appearing as three transverse rows,
presumably PC 24 to PC 26. This latter
condition is clearly seen in Figure 5A.

Crocodylus cataphractus

In Crocodylus cataphractus (Fig. 5D),
the continuous and contiguous precaudal
armor is regular in scute size and keel
row alignment. There are four to eight,
usually six, scutes across at midbody. The
median scute pair of each row tends to be
broader than the laterals and have lower
keels. Detached lateral elements are
present and sometimes form an addi-
tional keeled row on each side. PC 19 is
bordered front and back by short spaces
of soft skin. PC 20 + 21 and PC 22 + 23 are
compound rows of two elements each.
The compound nature of these rows is
demonstrated in other species of
Crocodylus. The nape rows are variably
developed, but for the most part are

vestigial. The largest elements appear to
be from PC 26.

Crocodylus johnstoni

In Crocodylus johnstoni (Fig. 6A), the
continuous and contiguous precaudal
armor is regular and has six to seven,
usually six, scutes across at midbody.
Detached lateral scutes form one or two
additional well-developed keeled rows
on each flank. The thoracic and cervical
armor are continuous, but small spaces of
soft skin occur between the transverse
rows at the cervico-thoracic juncture.
Sometimes there is but a single scute in
PC 18 or PC 19 (Fig. 6A). PC 20+21 has
two large scutes, and PC 22+23 has four,
for a total of six fairly closely juxtaposed
scutes forming an ovate shield. The two
lateral elements of PC 22+23 are often
displaced posteriorly, their medial edge
bordering the juncture between PC
20+21 and PC 22+23. The nape has

DORSAL ARMOR OF CROCODILIA - Ross and Mayer

315

Figure 6. Anterior end of the dorsal armor, PC 13 forward, with selected precaudal scale rows numbered on: A.
Crocodylus johnstoni, MCZ 35006. B. C. niloticus, USNM 195783. C. C. niloticus, MCZ 12552. D. C. niloticus,
USNM 63592.

three to six scutes in PC 26. PC 25
appears lost or perhaps coalesced with
PC 26. PC 24 has zero to four minor
elements.

Crocodylus niloticus

In Crocodylus niloticus (Fig. 6B-D)
the continuous and contiguous precaudal
armor is moderately regular, particularly
in the sacral and lumbar regions and
along the midline. There are 16 or 17
continuous rows of precaudal armor, with
four to nine, usually six, scutes per trans-
verse row at midbody. The median scute
pair of the anteriormost thoracic row is
often enlarged. Laterally, detached ele-
ments may form up to two additional
keeled rows. In some _ individuals,
especially from Madagascar, a row of two
scutes intervenes between the thoracic
and cervical armor. We interpret this as
PC 19 on the basis of it being closer to the
cervical than the thoracic armor, but this
identification is not certain. PC 20+21
has two scutes, and PC 22+23 has four.
These two transverse compounds tend to
be closely adjacent, forming an ovate
cervical shield. The compound nature of

these rows is demonstrated by occasional
variants exhibiting one or the other row
in an uncompound state (Fig. 6B, D). The
nape has three to six scutes in PC 26, and
zero to four small elements in PC 24. PC
25 is absent, or perhaps fused with PC 26.

Crocodylus palustris

In Crocodylus palustris (Fig. 7D), the
continuous and contiguous precaudal
armor is fairly regular. At the anterior end
of the thoracic armor, small scales some-
times occur between transverse rows and
between the scutes within a transverse
row, leading to a loss of contiguity. There
are four to eight contiguous scutes per
transverse row at midbody, with no out-
standing modal number. Detached
lateral scutes form one or two additional
keeled rows, the more medial particular-
ly well developed. The thoracic and
cervical armor are distinctly separated by
a space of granular skin. PC 17 has zero to
four scutes, and vestiges of PC 18 are
rarely present. PC 19 is absent. PC
20+21 has two scutes, and PC 22+23 has
four. Sometimes an additional pair of
scutes is in contact laterally, giving a total

316

Advances in Herpetology and Evolutionary Biology

Figure 7. Anterior end of the dorsal armor, PC 13 forward, with selected precaudal scale rows numbered on: A.
Crocodylus porosus, USNM 72732. B. C. novaeguineae, from New Guinea, CAS 119186. C. C. novaeguineae,
from the Philippines, FMNH 52362. D. C. palustris, FMNH 63739.

of eight scutes in the cervical shield.
Traces of PC 24 are sometimes evident.
PC 25 is absent or perhaps combined
with the normally four scutes in PC 26.

Crocodylus siamensis

In Crocodylus siamensis (Fig. 8A), the
continuous and contiguous precaudal
armor is regular, though often with small
triangular scales interspersed between
the scutes along the posterior border of a
transverse row, as is also found in C.
porosus. There are four to seven, usually
six, scutes across at midbody. Detached
scutes form one or two additional rows on
the flanks. PC 18 is usually absent, and
PC 19 always is. There is a cervical
shield of two scutes in PC 20+21 and
four scutes in PC 22+23, and sometimes
remnants of PC 24. PC 25 is absent or
perhaps combined with the two to usual-
ly four scutes in PC 26. C. siamensis is
reported to hybridize with C. porosus in
captivity, producing phenotypically
intermediate and allegedly heterotic off-

spring (Youngprapakorn, 1976). The fer-
tility of the hybrids is not reported.

Crocodylus porosus

In Crocodylus porosus (Fig. 7A), the
continuous and contiguous precaudal ar-
mor is regular in keel row alignment and
with six to eight, usually six, narrow ele-
ments per transverse row at midbody.
Small triangular scales occur interspersed
between scutes along the posterior edges
of transverse rows. Anteriorly, these small
elements also occur along the anterior
borders of scute rows, and may contact
the posterior triangles, breaking con-
tiguity. Lateral detached scutes are quite
small and indistinct. The thoracic and
cervical armor are completely discon-
tinuous. PC 17 is usually absent. PC 18 is
sometimes represented by vestiges. PC
19 is absent entirely. PC 20 + 21 has two
scutes, and PC 22 + 23 has four. The
nape is essentially naked. In specimens
examined by us PC 24 to PC 26 are usu-
ally absent, though rarely two distinct

DorSAL ARMOR OF CROCODILIA : Ross and Mayer

CO)
Het
Ch) 9

317

Figure 8. Anterior end of the dorsal armor, PC 13 forward, with selected precaudal scale rows numbered on: A.
Crocodylus siamensis, MCZ 3716. B. C. rhombifer, MCZ 12081. C. C. intermedius, FMNH 75658. D. C. acutus,

FMNH 23147. E. C. acutus, MCZ 10920.

scutes may occur in PC 26. Deraniyagala
(1939) reports specimens with up to four.

Crocodylus novaeguineae

In Crocodylus novaeguineae (Fig. 7B,
C), the continuous and contiguous pre-
caudal armor is regular in scute dimen-
sions and keel row alignment. As in C.
porosus and C. siamensis, interscute tri-
angles occur. There are 16 or 17 contin-
uous precaudal rows, with seven to
twelve, usually eight, contiguous scutes
at midbody. Detached flank scutes,
though reduced, sometimes form an addi-
tional keeled row on each side. The
thoracic and cervical armors are separated
by spaces of skin. PC 18 is represented by
vestiges or elements separated on the
midline. PC 19 is absent, or, in a single
specimen, rudimentary. PC 20+21 has
two scutes, and PC 22+23 has two large
median elements, and may have one or
two smaller elements on either side. In
many individuals, smooth or fine granular
skin separates the two cervical rows (PC
20+21 and PC 22+23) or the left and right
halves of one or both (Fig. 7B). Some-
times PC 22 and PC 23 are not com-
pound. All or any combination of cervical
scutes may be noncontiguous. There are

four to six scutes in PC 26. PC 24 and PC
25 are sometimes not evident, but are
usually present as bluntly keeled scales.

Previous authors have differed on
whether the freshwater crocodiles of the
Philippines should be recognized as a
species (mindorensis) distinct from C.
novaeguineae, or merely as a subspecies
of the latter (Wermuth, 1953; Schmidt,
1956; Wermuth and Mertens, 1961).
Wermuth and Mertens, in their most
recent checklist (1977), have elevated
mindorensis to full species status. Tables
1 and 2 include with C. novaeguineae
data for a few mindorensis. The preced-
ing account of the cervical armor, how-
ever, applies only to C. novaeguineae
proper. C. mindorensis (Fig. 7C) differs
from C. novaeguineae in that PC 19 is
better developed, being present as one or
two elements in three of five individuals;
PC 22+23 has four elements; the scutes
of the cervical shield are little or not
at all separated by skin; and the nape
scutes between the cervical shield and
the prominent nape row (PC 26) are bet-
ter developed, sometimes in two distinct
rows. The extremal count of twelve
scutes at midbody is from an individual
from the Philippines. In general,
mindorensis has a more well-developed

armor than -novaeguineae, though this
conclusion is based on a small sample
and should be considered tentative.

Crocodylus rhombifer

In Crocodylus rhombifer (Fig. 8B), the
continuous and contiguous precaudal
armor is often remarkably even and
regular. There are six or seven, usually
six, scutes across at midbody, and
detached lateral elements form one or
two additional keeled rows. Some indi-
viduals have continuous cervical and
thoracic armor, others do not. PC 18 has
one or two scutes. Sometimes there is a
space of soft skin between PC 18 and PC
19, and invariably between PC 19 and PC
20+21. PC 19 is variously absent or pres-
ent as two scutes. PC 20+21 has two
scutes and PC 22+23 has four. Anterior to
the cervical shield, scutes are poorly
developed, except for a row of four
prominent scutes, presumably PC 26. C.
rhombifer is easily distinguished from C.
acutus by having six or more scutes in all
four transverse rows at midbody. The two
species are reported to hybridize in the
wild and in semi-captivity (Varona, 1966;
Anonymous, 1969). A presumptive wild
hybrid, AMNH 82943, has the black and
white flecked coloration of C. rhombifer,
and regularity of the dorsal armor like
rhombifer, but has fewer contiguous
scutes at midbody than rhombifer, is
lacking PC 17 entirely, and has a head
with C. acutus proportions and snout
character.

Crocodylus moreletii

In Crocodylus moreletii (Fig. 4), the
continuous and contiguous precaudal
armor is fairly irregular and asymmetric,
with the contiguous scutes at midbody of-
ten reduced to three or four. There are two
to six, usually four, contiguous scutes a-
cross at midbody, and detached elements
on the flank are always present. PC 17 and
PC 18 are absent, and PC 19 has zero to

Advances in Herpetology and Evolutionary Biology

two scutes. PC 20+21 has two scutes, and
PC 22+23 usually has four. PC 24 is vari-
ably developed, sometimes absent. PC
25 is absent or perhaps compounded with
the four to six scutes in PC 26.

Natural hybrids have been reported
between Paleosuchus trigonatus and P.
palpebrosus (Medem, 1970) and Croco-
dylus rhombifer and C. acutus. In both
cases, the presumed hybrids closely
resembled one of the parent species,
though possessing certain minor, yet
characteristic, traits of the other parent. It
is probable that hybridization has also
occurred between C. acutus and C.
moreletii. Ross and Ross (1974) used a
combination of several cranial characters
and body color to identify C. acutus from
throughout its range, and found traces of
a C. moreletii character (lateral intrusions
on the basicaudal surface; see Fig. 4) on
C. acutus only where they are sympatric
with C. moreletii (Belize through
Chiapas), or where the population could
have been influenced by hybrid stock
(west coast of Mexico; it is interesting
that fossil C. moreletii have been
reported from Baja [Miller, 1980]). The
dorsal armor of presumptive hybrids
resembles both parent species in scale
counts, but in this case both parent
species have remarkably similar dorsal
armor.

Crocodylus acutus

In Crocodylus acutus (Fig. 8D, E), the
continuous and contiguous precaudal
armor is highly irregular in scute size,
scute shape, and disposition. The median
scute pairs of each row are fairly even in
length, except anteriorly, where midline
irregularities and gross asymmetry may
occur. There are two to six, usually four,
scutes across at midbody, always with
asymmetry in at least some rows. There
are always detached scutes on the flank.
There are 14 to 17, usually 16, continuous
precaudal rows. There is a tendency for
the anteriormost median scute pair of the
continuous armor to be large and flanked

DORSAL ARMOR OF CROCODILIA - Ross and Mayer

by noncontiguous elements. PC 18 is
always absent. PC 19 is almost always
absent. PC 20 — 21 has two elements, and
PC 22 + 23 has four, forming a cervical
shield of six elements, though variations
by deletion are common. PC 24 is ves-
tigial, or more often absent, and PC 25 is
absent or perhaps fused with the two to
four elements of PC 26.

Crocodylus intermedius

In Crocodylus intermedius (Fig. 8C),
the continuous and contiguous precaudal
armor is more or less regular in scute
dimensions and keel row alignment.
There are five or six contiguous scutes
per transverse row at midbody, with a
variable development of detached flank
scutes sometimes forming an additional
keeled row. There are 16 continuous pre-
caudal rows. One of our specimens has a
“normal” Crocodylus cervical shield and
nape region: PC 20+21 with two ele-
ments, PC 22+23 with four elements,
and PC 26 with four. Our other specimen
has three transverse rows in the cervical
shield, with two scutes in each of the two
posteriormost, and four in the anterior-
most. The nape is “normal” with five
scutes. Our sample of C. intermedius is
small, consisting of a single museum
specimen, and photographs of a living
individual in the collection of F. Medem.
C. A. Ross (personal communication)
describes the type (MNHN Paris 7512) as
having heavy thoracic armor of six, some-
times even seven contiguous scutes
across. There are no detached lateral
scutes. The neck has six scutes in a
shield, and there are five scutes in the
nape row.

Osteolaemus tetraspis

In Osteolaemus tetraspis (Fig. 9A), the
continuous and contiguous precaudal
armor is more or less regular, the median
scute pairs the broadest, and with keels
poorly developed or lacking. There are

319

four to eight, usually six, scutes per trans-
verse row at midbody, and detached
scutes on the flanks. PC 18 usually con-
sists of two elements contiguous on the
midline, often with detached lateral ele-
ments with very small, vestigial scales
between them. Anterior to PC 18 there is
a fold of skin, a row of two small ele-
ments, another fold of skin, and then two
rows of two large elements each. The
anteriormost of these large rows is usual-
ly the longest of the two. In a single
specimen (Fig. 9A), only one row of two
small elements occurs between PC 17
and the two large rows of the cervical
shield. There is usually one well-de-
veloped row of four scutes on the nape.
Some specimens show an additional pair
of scutes. Radiographs (Fig. 1C) show
that the large neck scutes overlie verte-
brae PC 21 and PC 22, and PC 23 and PC
24. However, vertebral correspondences
of all cervical scutes and the nape are
uncertain.

Alligator sinensis

In Alligator sinensis (Fig. 9C), the con-
tinuous and contiguous precaudal armor
has six to eight, usually six, scutes at mid-
body, and few or no lateral detached
scutes on the flanks. The thoracic and
cervical armors are separated by at least a
small space of skin. PC 17 is sometimes
absent, but usually present with up to
five elements. PC 18 is usually absent.
Anterior to this there is a strip of skin fol-
lowed by three transverse rows of two
scutes each. The size of these scutes in-
creases anteriorly. The nape has a row of
four to six prominent scutes, and some-
times an additional row of two between
the prominent row and the cervical
shield. The vertebral correspondences of
the cervical shield and the nape are
uncertain due to the discontinuity at the
cervico-thoracic juncture and the rela-
tively slight variation among individuals.
The prominent nape row is likely PC 25,
as it is in A. mississippiensis, leaving PC

320

Advances in Herpetology and Evolutionary Biology

D E

Figure 9. Anterior end of the dorsal armor, with selected precaudal scale rows numbered, PC 13 forward on: A.
Osteolaemus tetraspis, USNM 193289. B. Alligator mississippiensis, MCZ 17723. C. A. sinensis, FMNH 38234.
PC 17 forward on: D. Caiman crocodilus, MCZ 15334. PC 13 forward on: E. C. /atirostris, MCZ 3588.

19 to PC 24 to be accounted for by the
three rows of the cervical shield and the
smaller nape row. The possibility of
compounding is indicated by the large
size of the anteriormost of the three rows
of the cervical shield.

Alligator mississippiensis

In Alligator mississippiensis (Fig. 9B),
the continuous and contiguous precaudal
armor is five to eight, usually eight,
scutes across at midbody, and there are
no detached scutes on the flanks. The
thoracic and cervical armors are barely
continuous, having reduced elements in
PC 18 to PC 19, and possibly PC 20, and
two folds of skin in the cervico-thoracic
juncture. PC 18 is ordinarily present as
two small scutes. There are ordinarily
lateral scutes which may be from PC 18
or traces of PC 19 (we don’t know which).
If PC 19 is absent on the midline, there
are two scutes in PC 20, two very long
elements in PC 21+22, two in PC 23, and
then the nape scales. If the posteriormost
neck scute is PC 19, PC 20+21 has two
very long scutes, and PC 22+23 has two
shorter scutes, and then there are the

nape scutes in two or three rows separat-
ed on the midline. The most prominent
nape row is PC 25, ordinarily with two,
but with up to six ossified elements. PC
25 is the postoccipital of Ross and
Roberts (1979); in Crocodylus this term
refers to PC 26.

Caiman lJatirostris

In Caiman latirostris (Fig. 9E) the con-
tinuous and contiguous precaudal armor
varies from regular to somewhat irregular
in the alignment of the keels. There are
six to nine, usually six, scutes across at
midbody, with detached lateral scutes
tending to form additional rows on the
flanks. The thoracic and cervical armor
are usually discontinuous. PC 18 has two
to four elements. PC 19 is normally
absent, but may be present in rudimen-
tary fashion (Fig. 9E). There are three
transverse rows anterior to PC 19 (or PC
18 when PC 19 is absent) in the cervical
shield, with two scutes in the two pos-
teriormost, and four in the anteriormost
row. The vertebral correspondences of
these rows are uncertain, but are probab-
ly PG 20; PC 2, and) PG 225-23) com

DORSAL ARMOR OF CROCODILIA : Ross and Mayer

pound. In a single specimen (MCZ
17716), PC 24 is well developed and con-
tinuous with the cervical shield. The nape
consists usually of PC 24 to PC 26, with
PC 26 most strongly developed.

Caiman crocodilus

In Caiman crocodilus (Figs. 3C, 9D)
the continuous and contiguous precaudal
armor tends to have low keels in longi-
tudinal alignment, and the detached
scutes on the flanks tend to be aligned in
additional rows. There are six to twelve,
usually eight, scutes across at midbody.
The thoracic and cervical armors are
sometimes not truly continuous, though
they appear so with only narrow strips of
skin where PC 18 or PC 19 are lost. There
are usually four scutes across somewhere
in the PC 21 to PC 23 region, often in PC
93. PC 24 is variously absent, vestigial, or
present and rarely (Fig. 3C) continuous
with the cervical shield and contiguous
across the midline. The nape appears to
be PC 24 to PC 26, sometimes PC 25 to PC
26 only. Some specimens exhibit 26
transverse precaudal rows without loss or
compounding, but about half of the
specimens examined show left-right
asymmetry and fewer than 26 transverse
precaudal rows. In these specimens it is
difficult to determine which rows have
been lost or fused. Figure 9D is a probable
interpretation.

Melanosuchus niger

In Melanosuchus niger (Fig. 2) the con-
tinuous and contiguous precaudal armor
is regular, with nine to twelve, usually ten
scutes across at midbody, and there are no
detached lateral scutes. There are four to
six scutes in PC 18, two to four in PC 19,
two to four in PC 20, two to four in PC 21,
and four in PC 22 and PC 23. Remnants
of PC 24 to PC 26 in approximately three
transverse rows give the nape an overall
granular and studded appearance. There
are invariably 23 continuous precaudal
rows in our sample.

321

Paleosuchus palpebrosus

In Paleosuchus palpebrosus (Fig. 10D)
the thoracic armor is broad with more or
less regular keel row alignment and scute
dimensions. There are four scutes in PC
2 to PC 5, and six to nine, usually eight,
contiguous scutes at midbody. The pre-
caudal armor is continuous to PC 18. PC
18 has three to six elements. PC 19 is
often bordered anteriorly and posteriorly
by soft skin, and is reduced in size. There
are two contiguous elements in PC 19,
PC 20, PC 23, and PC 24. PC 21 and PC
22 may have four or, usually, three ele-
ments. PC 24 may be continuous with or
separated from the cervical shield. PC 25,
as here interpreted, consists of lateral
elements and, in one specimen (Fig.
10D), a single element adjacent to the
midline. The most prominent row on the
nape is PC 26 with four to six elements.
P. palpebrosus is much more ossified
than its congener. This is seen in the
greater breadth of its dorsal armor, the
continuity between the thoracic and
cervical armor, and the greater ossifica-
tion of the skull and the ventral armor
(Medem, 1958b).

Paleosuchus trigonatus

In Paleosuchus trigonatus (Fig. 10A-C)
the dorsal armor is four to eight, usually
six, scutes wide at midbody and often
highly irregular in keel row alignment
and scute size and shape. These irregular
elements form a broad carapace with
some detached scutes on the flanks.
Some P. trigonatus lose the median pair
of scutes in PC 2 to PC 5 in various com-
binations. This character is unique
among the living Crocodilia. Our inter-
pretation of P. trigonatus is based on the
fact that it is less ossified than P. palpe-
brosus. In P. palpebrosus PC 18 usually
has four elements, while in P. trigonatus
there are usually two. PC 19 is reduced
and has two elements in P. palpebrosus.
In P. trigonatus a single transverse row of
the cervical shield is lacking; and, on the

Q99

Advances in Herpetology and Evolutionary Biology

Figure 10. Anterior end of the dorsal armor, PC 13 forward, with selected precaudal scale rows numbered on: A.
Paleosuchus trigonatus, FMNH 69878. B. P. trigonatus, FMNH 69882. C. P. trigonatus, FMNH 69877. D. P.

palpebrosus, FMNH 134989.

basis of its small size in P. palpebrosus,
we believe the missing row to be PC 19.
PC 20 to PC 23 normally have two ele-
ments. In some individuals PC 21 and PC
22 appear to be compound. The nape has
one well-developed row, PC 26, and
small scales lie between PC 26 and PC
23, arranged in one or more additional
rows. The larger of these latter small
rows is probably PC 24, because this row
is better developed in P. palpebrosus.
Some P. trigonatus and P. palpebrosus
combine median scute pairs in the neck
to form median unpaired single or double
keeled elements. This again is unique
among the living Crocodilia.

DISCUSSION

THE FOSSIL RECORD

The presence of middorsal armor is
widespread in the early archosaurs and
their descendants, although it is lacking
or poorly developed in the most primi-
tive thecodonts, the proterosuchians
(Reig, 1970; Romer 1972). It is most high-
ly developed in parasuchians (phyto-

saurs), aetosaurs, certain ornithischians,
and, of course, crocodilians. In some of
these, such as parasuchians (Chatterjee,
1978) and Stagonolepis (Walker, 1961)
there is also a one-to-one correspondence
between vertebrae and transverse scute
rows. The wide occurrence of dorsal
armor suggests that it is primitive for
archosaurs; its general absence in
proterosuchians, however, urges caution
in accepting this hypothesis.

Fossil crocodilians with dorsal armor
in place are rare, but several good ex-
amples exist. The earliest known true
crocodiles, Protosuchus (Colbert and
Mook, 1951), from the Upper Triassic or
earliest Jurassic, and Orthosuchus (Nash,
1975), from the Upper Triassic, have
essentially identical dorsal armor. In the
neck and trunk region paired rectangular
scutes occur in one-to-one correspond-
ence with the vertebrae. At their lateral
edges the scutes are sharply downturned
to form a lateral face. Behind the occiput
an extra small scute or scute pair (corre-
sponding to the proatlas?) occurs. In the
more completely known Protosuchus this
armor extends onto the tail to its tip. The
armor forms a long, essentially parallel

DorRSAL ARMOR OF CROCODILIA - Ross and Mayer

sided, rigid bony adjunct to the vertebral
‘column (Fig. 3A). The armor of another
early crocodilian, Stegomosuchus, seems
to be much the same (Walker, 1968).
Protosuchian dorsal armor did not form a
carapace, though the scutes did enclose
the epaxial musculature, and as such
were protective.

Romer (1974) derived crocodilians
from creatures of amphibious habits
because strong hind limbs and tails func-
tion well in aquatic environments. Nash
(1975) followed Romer, supposing that
Orthosuchus spent much of its time in
the water. However, the completeness of
the dorsal armor of protosuchian croco-
diles would have prohibited tossing the
head upwards and backwards for swal-
lowing prey while keeping the gullet
above water in modem crocodilian
fashion. Protosuchians could not feed in
deep water, and also lacked a com-
pressed tail with a median keel (see
Manter, 1940, on the importance of the
tail in swimming). We suggest that the
group was terrestrial.

The transition from protosuchian to
mesosuchian crocodiles involved reduc-
tion or loss of the anteriormost transverse
scute rows and lateral compression of the
posterior half of the tail to a vertical keel.
The development of a caudal keel and
loss of the anteriormost transverse rows
(thus gaining the ability to toss the head
backwards to swallow) indicate aquatic
habits.

Some mesosuchians retained a parallel
sided armor much like that of proto-
suchians (atoposaurs, Wellnhoffer, 1971),
while others broadened the scutes in the
trunk region (e.g., Teleosaurus, Kalin,
1955). Many mesosuchians had peg and
socket articulations on the lateral edges
of their scutes. This is an elaboration of
peglike processes seen in Protosuchus
and Orthosuchus. Most mesosuchians
still had only a single pair of scutes per
transverse row (Romer, 1956). The
pelagic marine thalattosuchians lacked
armor altogether (Troxell, 1925). (It is
interesting that the most reduced dorsal

323

armor in modem crocodilians occurs in
the two most marine species, Crocodylus
porosus and C. acutus.)

The transition from mesosuchian to
eusuchian crocodiles involved further
caudal compression, and, in some cases,
further loss of anterior transverse scute
rows. Eusuchians broadened and frac-
tured the ancestral scute pairs into paired
sets of two to six elements each (Fig. 3B,
C), lost the lateral peg and socket articu-
lations and consequent rigidity, and
covered the trunk and base of the tail
with a pavement of bony plates forming a
defensive carapace. Interruption or nar-
rowing of the armor in the shoulder
region is also typical of eusuchians (Fig.
3C, 4). In general, eusuchian armor is
broader and more flexible than that of
their mesosuchian predecessors, with a
tendency towards interruption of the
dorsal armor in the shoulder region.
Procaimanoidea (Hassiacosuchus) (Mook,
1941), Leidyosuchus (Erickson, 1976),
and Bernissartia (Kalin, 1955)! are
fossil eusuchians whose armor is known,
and all are of essentially modern form.
Among the living Eusuchia, the dorsal
armor of Gavialis is most reminiscent of
mesosuchian armor.

VARIATION AND EVOLUTION

Primitively, crocodilian dorsal armor
consists of 27 transverse precaudal rows
of two scutes each, as is found in proto-
suchians. The preceding review of his-
torical changes in the primitive armor,
through mesosuchian and _ eusuchian
stages, allows us to make some general
statements on the direction of evolution
of some characters. Advanced character-
istics include loss of rows, a greater num-
ber of scutes per row, broadening of the
trunk armor, and narrowing of the armor

'Buffetaut (1975) has questioned whether
Bernissartia is a eusuchian on the basis of palatal
and vertebral structure; its armor, however, is un-
questionably advanced in character.

324

at the cervico-thoracic juncture. With this
in mind, we can look at evolutionary
steps and directions in the living taxa.
The possibility of reversal cannot be
ignored; for example, the low number of
scutes per row in Crocodylus acutus is
probably a secondary loss.

The variation in dorsal armor among
the living crocodilians shows that a
variety of evolutionary changes have
occurred within the group. The two kinds
of changes which account for most of this
variation are deletion of rows, and the
compounding of rows in the cervical
region.

Deletion of at least the anteriormost
precaudal row (PC 27) occurs in all
modern taxa. In most taxa more are miss-
ing. In the genus Crocodylus a series of
stages (not necessarily a phyletic series)
in the reduction of armor at the cervico-
thoracic juncture can be seen. C. cata-
phractus has continuous precaudal armor
anon JAG Oetos} (aig, SID), lin C,
johnstoni, PC 18 and PC 19 may be re-
duced to a single scute (Fig. 6A). In C.
niloticus reduction continues, with one
or both of PC 18 and PC 19 entirely
absent, leaving the precaudal armor con-
tinuous to PC 17 and the cervical and
thoracic armor discontinuous (Fig. 6B-
D). Further rows may be deleted, so that
in C. acutus, for example, some indivi-
duals have continuous armor only to PC
14. Deletion may also occur in the cervi-
cal shield, and the presence of rows
usually absent also occasionally occurs as
an individual variation.

The compounding of rows, which en-
tails a loss of one-to-one correspondence
of vertebrae and transverse rows, is clear-
ly derived with respect to separate rows
being present. Compounding is indi-
cated by the occurrence of individuals
showing asymmetrical fusion. This is
clearly seen in the series of Tomistoma in
Fig. 5: 5A has the enlarged compound,
5B is compound only on the left, and 5C
has both PC 22 and PC 23 entirely
separate. Though greatly enlarged
scutes, and scutes which overlie two
vertebrae suggest compounding, the

Advances in Herpetology and Evolutionary Biology

recognition of compounding depends on
individual variation. Some species may
be phenotypically fixed for compounds,
which would make the compound row
difficult to detect.

Overall development of the dorsal
armor is reflected in the number of rows
present and the number of scutes per
row. There is considerable variation
among species in the extent to which the
armor is developed. In the genus Paleo-
suchus, for example, palpebrosus, with
more well-developed armor, has 24 con-
tinuous precaudal rows and a mode of
eight scutes per row at midbody, while
the less armored trigonatus lacks PC 19,
and has modally six scutes per row.
Similarly, Crocodylus rhombifer has at
least 17 continuous precaudal rows and
six scutes per row, while C. acutus usual-
ly has 16 or fewer continuous precaudal
rows, and four scutes per row. Overall
reduction of armor is sometimes accom-
panied by marked left/right asymmetries,
as in C. acutus and C. moreletii.

The variation within species is of the
same sort as that which distinguishes
species. Thus, species differ in number
of transverse rows, number of scutes per
transverse row, and the presence of com-
pounds, and all of these characters vary
within species. Many individual varia-
tions are reversions to a more primitive
condition, such as the breakup of com-
pounds in Crocodylus niloticus, or the
presence of PC 17 in some C. acutus.
Other variations are derived and char-
acteristic of more advanced species, such
as the occasional reduction of PC 18 and
PC 19 in the relatively primitive C.
johnstoni. Some individuals show pheno-
types more derived than those character-
istic of even the most advanced species,
such as a C. niloticus with a very reduced
cervical shield, or a C. acutus with only 14
continuous precaudal rows.

ACKNOWLEDGMENTS

About fifteen years ago, one of us
(FDR) brought a young crocodilian to the
Museum of Comparative Zoology for

DorSAL ARMOR OF CROCODILIA - Ross and Mayer

PG

325

ea aaa ee
Sr?

: en a
AN Sa =

Sapa laeash L _.

Figure 11. Lateral views of the anterior ends of the tails of: A. Tomistoma schlegelii, MCZ 9327, showing absence
of basicaudal irregularity. B. Osteolaemus tetraspis, MCZ 2107, showing two thick granular strips separating the

dorsal from the ventrolateral scalation.

identification. The beast had been sold as
Crocodylus acutus, but appeared to dif-
fer from that taxon by the number of
Osseous scutes on the neck. Ernest E.
Williams ducked the question, suggest-
ing that crocodiles were not yet under-
stood, and that work should be done on
the problem. We thank him for inspiring
and facilitating the bulk of this work.
We report primarily on specimens in
the American Museum of Natural History
(AMNH), the Field Museum of Natural
History (FMNH), the Museum of
Comparative Zoology (MCZ), and the
National Museum of Natural History
(USNM). We have also examined speci-
mens in the Boston Museum of Science
(BMS), the Califomia Academy of
Sciences (CAS), the Museo Nacional
de Historia Natural Santo Domingo,
(MNHNSD), the Instituto de Historia
Natural (Tuxtla Gutierrez, Chiapas), the
National Museum of Canada, and the
Yale Peabody Museum. We are indebted
to the curators and staffs of these institu-

tions, especially Emest E. Williams,
Hymen Marx, Miguel Alvarez del Toro,
George Zug, Charles A. Ross, and the late
James A. Peters. We are also indebted to
Aleta Karstad for Figure 4, Jim Lovesek
for photographs of Crocodylus inter-
medius, and E. E. Williams and Don
McAllister for radiographs. In addition,
we thank A. S. and S. B. Avery, J. M.
Clark, F. R. Cook, P. Elias, K. Miyata,
A. Rhodin, J. P. Rosado, J. P. Ross, F. W.
and A. K. Schueler, and R., R., and F.
Whiston. F.D.R. was supported in part by
a grant from the Ella Lyman Cabot Trust,
and G.C.M. was supported by an NSF
graduate fellowship and NRM.

APPENDIX: SPECIMENS EXAMINED

Museum abbreviations are in the Ac-
knowledgments. Numbers in parentheses
are the number of individuals in lots
designated by a single museum number.

326 Advances in Herpetology and Evolutionary Biology

' TABLE 1. NUMBER OF CONTIGUOUS SCUTES PER TRANSVERSE ROW IN PC 1-PC 18.
Number given is mode (or modes) with range undemeath in parentheses. Species are identified by the first
four letters of their generic and specific names. N = sample size.

Species N PC 1 PC2 PC 3 PC4 PC5 PC6 InG ING) IRC)

Gavi gang 8 4 4 4 4 4 4 4 4 4
(4) (4) (4) (4) (4) (4) (4) (4) (4)
Tomi schl 12 4 4 4 4 4 4 6 6 6
(4) (4) (4 (4) (4) (4-6) (6) (5-6) (6-7)
Croc cata 11 4 4 4 4 6 6 6
(4) (4) (4) (4-5) (4-6) (4-6) (4-6) (48) (47)
Croc john ll 4 4 4 4 46 6 6 6
(4) (4) (4) (3-5) (4-6) (6) (6) (6) (6)
Croc nilo 19 4 4 4 6 6 6 6 6 6
(4) (4) (4-6) (4-6) (4-6) (4-6) (4-8) (68) (638)
Croc palu 12 4 4 4 4 6 6 6 6 4,5
(4) (4) (4-5) (4-6) (4-6) (5-6) (6-7) (4=7) (458)
Croc siam 6 4 4 4 4 6 6 6 5 6
(4) (4) (4) (4-6) (4-7) (6-7) (5-8) (48) (5-6)
Croc poro 12 4 4 4 6 6 6 6 6
(4 (4) (4-6) (4-6) (4-6) (6-8) (6-8) (68) (658)
Croc nova 14 4 41 4k 6 6 6 8! 8 8i
(4) (4) (4-5) (4-6) (6) (6-8) (6-8) (8) (611)
Croc rhom 6 4 4 4 4 4 6 6 6 6
(4) (4) (4) (4) (4) (5-6) (6) (C) Gam)
Croc more 8 4 4 4 4 4 4 3 6 6
(3-5) (4) (3-4) (3-5) (3-5) (2-6) (3-6) (2-6) (46)
Croc acut 37 4 4 4 4 4 3,4 4 3 4
: (3-5) (2-4) (2-4) (2-4) (2-5) (2-6) (2-6) (2-6) (2-6)
Croc inte* 2 4.5 4 4 4 5,6 5) 5,6 6
Oste tetr 18 4 4 4 4 4 4 4 6) 6
(4) (4) (4) (4) (4) (4-6) (4-6) (46) (47)
Alli sine 14 4 4 4 4 4 4 4 6 6
(4) (4) (4) (4) (4) (4) (4-6) (46) (68)
Alli miss 17 4 4 4 4 4 6 6 6) 6
(4) 4 @ 6) el @s) G69 (es es)
Caim lati 12 4 4} 4} 4 6 6 6! 8 8
(4-6) (4) (4) (4) (5-6) (6) (6-8) (4-9) (5-9)
Caim croc 24 4° 4° 4° 6° 6° 8° 8 8
(4-6) (4-6) (4-6) (4-8) (6-10) (6-11) (7-12) (8-12) (8-12)
Mela nige 15 6 6 6 6 8 8 8 10 10
(6) (5-6) (6) (6) (6-8) (810) (8-10) (8-12) (9-11)
Pale palp 5 4 4c 4 4 4 4,5 6¢ i gb
(4) (4) (4) (4) (A) (4-6) (6-8) (6-7) (6-9)
Pale trig 25 4 40 2, 2 4 4 40 5 6

Species

_ Gavi gang

' Tomi schl

' Croc cata

Croc john
Croc nilo
Croc palu
Croc siam
Croc poro
Croc nova
Croc rhom
Croc more
Croc acut

Croc inte*
Oste tetr

Alli sine
Alli miss
Caim lati
Caim croc
Mela nige
Pale palp

Pale trig

21

(6-12)
105
(9-12)
TAS
(6-8)
6
(4-7)

DORSAL ARMOR OF CROCODILIA °

TABLE 1. CONTINUED.

PC 12

4

(4-8)

PC 13

4d
(4)
6g
(5-6)
6!
(5-7)
6
(6-7)

PC 14

4d
(4)

10
(8-10)
6
(6)
AS
(4-6)

Ross and Mayer

PC 16

(5-7)
4°
(2-6)

327

Qo
(2-4)

4Range not given since N=2. »N=6°N=94N=10°N=11 !N=122N=13"N=14'N=15JN=16‘N=18!N=19
EN 2 1 N23 °N=25

(oy)
bo
(ee)

Advances in Herpetology and Evolutionary Biology

TABLE 2. NUMBER OF DOUBLE AND SINGLE CREST CAUDALS IN UNDAMAGED SPECIMENS OF LIVING CROCODILIANS.
Species are identified by the first four letters of their generic and specific names.

Number of double crest caudals Number of single crest caudals
Species 9 10 11 12 13 14 15 16 17 18 19 20 21 16 17 18 19 20 21 22 23 24 25
Gavi gang 3 2 1 1
Tomi schl lL ¢ 8 9 @ ®
Croc cata 5 7 2001 2a eS
Croc john OM? my 1 8
Croc nilo 1 GaO eS 1 1 6 6
Croc palu % § 2 Il IL & ® il
Croc siam 1 3 1 I) 2
Croc poro 3 7 @ mie 8 B
Croc nova 7 Ni @ 6 6 3
Croc rhom 3.4 ae! ae]
Croc more 3.4 | I 22) eal
Croc acut lL 7 S15 & IL 20 iy ®
Croc inte* ge 2 8 4
Oste tetr tl 4 @ ®il 2 2 ®
Alli sine % o& & 4 3 4
Alli miss 6 ll 1 G 2 g
Caim lati 3 © 8 1 Tea Ages
Caim croc I 3 QilB 4 87 & §
Mela nige li @ @ Mesh 2
Pale palp lL 6 8 1
Pale trig LB QP Aa:

a Data from Medem, 1958a, Table 1. The single specimen of this species available to us was also examined
by Medem. Medem’s last dorsal corresponds to PC 1 in this specimen; thus, his counts correspond to the
method used in this paper.

TABLE 3. FORMATION OF Y-SHAPED CAUDAL KEEL FROM PELVIC KEEL ROWS.

Median pelvic keel rows merge with

lateral pelvic keel row to become Median pelvic keel rows do not

Y-shaped caudal keel contribute to the Y-shaped caudal keel
Median pelvic keel rows Median pelvic keel rows
become an unpaired maintain their paired
median keel row diminishing nature until diminishing
posteriorly posteriorly

Paleosuchus, Alligator sinensis Caiman, Melanosuchus Gavialis, Tomistoma,

Crocodylus, Osteolaemus,
Alligator mississippiensis

DORSAL ARMOR OF CROCODILIA : Ross and Mayer 329
TABLE 4. OCURRENCE OF FOUR KINDS OF BASICAUDAL IRREGULARITIES IN EXAMINED SPECIMENS OF LIVING
CROCODILIANS.
The irregularities are further described in the Methods. + = present; — = absent; * = thinner than in other
species. Species are identified by the first four letters of their generic and specific names.
One thin One or more Lateral Ventrolateral
Species granular strip thick strips intrusions intrusions
Gavi gang = = al
Tomi schl = is ae
Croc cata t= = =
Croc john +/— — =
Croc nilo f= — —
Croc palu P= at a = =
Croc siam <P /= i= = =
Croc poro ef stills i= =
Croc nova Ei Ja us
Croc rhom — ass on
Croc more B= +/— +
Croc acut P= he = =
Croc inte — +/—4 —
Oste tetr = +/— +/— —
Alli sine = +/- —
Alli miss = +/— +/—
Caim lati = = +/- —
Mela nige il + —
Pale palp — +/— -
Pale trig = +/— — —

2C. A. Ross, personal comminication.

Gavialis gangeticus. INDIA: AMNH_ 81802,
MCZ 5263, 161013, USNM 136615; NO LOCAL-
ITY: AMNH 48500, BMS no #, FMNH 82681, no
#(2), USNM 211524.

Tomistoma schlegelii. MALAYSIA: USNM
84247, 85106; SUMATRA: FMNH 11084—-5, MCZ
9327, USNM 58136, 65509-10, 145586-8; NO
LOCALITY: AMNH_ 66380, 101422, FMNH
134997, 206755, USNM no #.

Crocodylus cataphractus. ANGOLA: MCZ
29360; TANZANIA: MCZ 54650, 54751; ZAIRE:
AMNH 10076, 45863, MCZ 16732; NO LOCAL-
ITY: AMNH 2206, 6584, 7293, 8877, FMNH 37213,
MCZ 5456, USNM 60578.

Crocodylus johnstoni. AUSTRALIA: AMNH
82535, 86536-7, 86539, MCZ 35001-4, 35006,
111978, USNM 64090-1; NO LOCALITY: AMNH
93743.

Crocodylus niloticus. BOTSWANA: USNM
195448, 195782-3; CAMEROON: FMNH 190827-
9; CHAD: MCZ 27070-1; ETHIOPIA: USNM
42860, 136621; GHANA: MCZ 55153; KENYA:
MCZ 40001, USNM 63592; MADAGASCAR: MCZ
2114, 5461, 12552, 16729-31, 16868; MALAWI:
MCZ 30000; MALI: FMNH 20799; SIERRA
LEONE: MCZ 54120; SOUTH AFRICA: FMNH
17155; TANZANIA: MCZ 48000.

Crocodylus palustris. INDIA: AMNH 2413-8,
FMNH 31538, 63739; SRI LANKA: MCZ 42176,
42180; NO LOCALITY: AMNH 45099, FMNH
95995, MCZ 6753, 17595-7.

Crocodylus siamensis. THAILAND: MCZ 3716,
20301, USNM 79476; NO LOCALITY: AMNH
75035, USNM 3906, 8893.

Crocodylus porosus. AUSTRALIA: AMNH
86541; INDONESIA (IRIAN JAYA): AMNH
62041-2; INDONESIA (SUMATRA): MCZ 43742;
PAPUA NEW GUINEA: MCZ 110450-1, 142950,
USNM 211310; PHILIPPINES: FMNH 14346,
5236344, USNM 38081; SANTA CRUZ IDS::
AMNH 42162; SOLOMON IDS.: MCZ 72936;
THAILAND: USNM 72730-6; NO LOCALITY:
AMNH 66388.

Crocodylus novaeguineae. PAPUA NEW
GUINEA: CAS 119186-7, 121997-8; MCZ 652845,
101338, 120335, 140805—-7, USNM 211297, 211305—
6, 082826-7 (field numbers); PHILIPPINES:
FMNH 52357-60, 52362; NO LOCALITY: USNM
083141(field number).

Crocodylus rhombifer. CUBA: FMNH 34677,
MCZ 1249, 1397, 12081, 33394, USNM 13576; NO
LOCALITY: AMNH 77595.

Crocodylus moreletii. BELIZE: FMNH 4430,

330

49369, MCZ 18856; GUATEMALA: USNM 71954,
71957-8; MEXICO: MCZ 8047, 53860, USNM
115357; NO LOCALITY: MCZ 17583.

Crocodylus acutus. COLOMBIA: MCZ 17610;
CUBA: MCZ 1250, 12278, 17718; DOMINICAN
REPUBLIC: MNHNSD no #(8); HAITI: MCZ
1339, 3596(3), USNM 80892; JAMAICA: MCZ
6093(7), 42981, 44853; MEXICO: FMNH 23147;
NICARAGUA: MCZ 5454, USNM 15230;
PANAMA: MCZ 1010, 1265, 2003, 2720, 7318,
18927-9, 28051; UNITED STATES: MCZ 13428-
30, 13432.

Crocodylus intermedius. COLOMBIA: FMNH
75658.

Osteolaemus_ tetraspis. CAMEROON: MCZ
293489, 33594; CONGO: USNM 62226; GHANA:
USNM 207060; LIBERIA: AMNH_ 82166-7,
116354, MCZ 22913, USNM 58273, 193289;
NIGERIA: MCZ 2017; ZAIRE: AMNH 29889,
MCZ 12458, 42855-6; NO LOCALITY: AMNH
7766, 24740, MCZ no #.

Alligator sinensis. CHINA: AMNH_ 28689,
28691-3, FMNH 38234, 38236, MCZ 16706-8,
17545-6, 17548—50, 42982, 161014, USNM 72846.

Alligator mississippiensis. UNITED STATES:
MCZ 969, 3767, 134334, 16715, 17723, 26949,
159136, 159146-7; NO LOCALITY: MCZ 5458,
17592-3, 159149-52.

Caiman latirostris. ARGENTINA: FMNH 9707-—
8; BRAZIL: FMNH 5777, MCZ 1151, 1455, 2854,
3588-9, 3769, 17600, 17716, 29099, 159154, USNM
98780; PARAGUAY: MCZ 34226.

Caiman crocodilus. BOLIVIA: AMNH 97310,
97315-23, FMNH 81564-5: BRAZIL: AMNH
87939, MCZ 1241, 2606, 3394, 3972, 3975; COSTA
RICA: MCZ 1533344, 54997; GUYANA: FMNH
26682; MEXICO: MCZ 5040, 17581; PANAMA:
FMNH 13402, MCZ 2035 2721, 20566, 22332-3,
29760; SURINAM: MCZ 154922: NO LOCALITY:
AMNH 66393.

Melanosuchus niger. BOLIVIA: AMNH 97324;
BRAZIL: MCZ 1011, 2547-9, 3590, 3592, 15134,
17598, 17726, 51851, 160240, no #; NO LOCAL-
ITY: AMNH no #(2).

Paleosuchus palpebrosus. BOLIVIA: AMNH
97328; BRAZIL: AMNH 16574, 97326-7, MCZ
2928, USNM 149134; COLOMBIA: FMNH 69867,
69869, 69875, 134989; NO LOCALITY: MCZ
160000.

Paleosuchus trigonatus. BRAZIL: AMNH
101056, MCZ 2957, 3785; COLOMBIA: FMNH
69877-8, 69881-2; ECUADOR: AMNH 61548-51,
72470, 113635, MCZ 84029; GUYANA: AMNH
16048, 61513, 61516-8, 64823-5; NO LOCALITY:
AMNH 46281-2, 159384.

HYBRID: presumed Crocodylus acutus x C.
rhombifer CUBA: AMNH 82943. Counts from this

animal or any other presumed hybrids mentioned in
the text are not included in Tables 1-4.

Advances in Herpetology and Evolutionary Biology

LITERATURE CITED

ANONYMOUS. 1969. Cuban crocodile. Animal King-
dom, 72(5): 29.

BOULENGER, G. A. 1889. Catalogue of the chelon-
ians, rhynchocephalians, and crocodiles in the
British Museum (Natural History). London,
Taylor and Francis.

BRAZAITIS, P. 1973 (1974). The identification of liv-
ing crocodilians. Zoologica, 58: 59-101.
BREDER, C. M., JR. 1947. An analysis of the ge-
ometry of symmetry with especial reference to
the squamation of fishes. Bull. Amer. Mus. Nat.

Hist., 88: 321-412.

BUFFETAUT, E. 1975. Sur l’anatomie et la position
systematique de Bernissartia fagesii Dollo, L.,
1883, crocodilien du Wealdien de Bemissart,
Belgique. Bull. Inst. r. Sci. Nat. Belg., 51: 1-20.

CHATTERJEE, S. 1978. A primitive parasuchid
(phytosaur) from the Upper Triassic Maleri
Formation of India. Palaeontology, 21: 83-127.

COLBERT, E. H., AND C. C. Mook. 1951. The ances-
tral crocodilian Protosuchus. Bull. Amer. Mus.
Nat. Hist., 97: 143-182.

CuviER, G. 1807. Sur les différentes especes de
crocodiles vivans et sur leurs caracteres dis-
tinctifs. Ann. Mus. Hist. Nat. Paris, 10: 8-66.

__. 1831. The animal kingdom arranged in con-
formity with its organization. Vol. 9. The class
Reptilia. London, Whittaker, Treacher.

DERANIYAGALA, P. E. P. 1939. The tetrapod reptiles
of Ceylon. Vol. 1. Testudinates and crocodil-
ians. Ceylon J. Sci. Colombo Mus. Nat. Hist.
Ser.

DUMERIL, A. M. C., AND G. BIBRON. 1836. Erpétol-
ogie génerdle ou histoire naturelle complete
des reptiles. Vol. 3. Paris.

ERICKSON, B. R. 1976. Osteology of the early
eusuchian crocodile Leidyosuchus formida-
bilis, sp. nov. Monogr. Sci. Mus. Minnesota,
Paleontology, 2: 1-61.

Fox, S. F. 1975. Natural selection on morphological
phenotypes of the lizard Uta stansburiana.
Evolution, 29: 95-107.

Gapow, H. F. 1901. Amphibia and reptiles. The
Cambridge Natural History Vol. 8. London,
Macmillan.

HECHT, M. K. 1952. Natural selection in the lizard
genus Aristelliger. Evolution, 6: 112-124.
Hiccins, G. M. 1923. Development of the primitive
reptilian vertebral column, as shown by a
study of Alligator mississippiensis. Amer. J.

Anat., 31: 373-407.

HOFFSTETTER, R., AND J.-P. Gasc. 1969. Vertebrae
and ribs of modern reptiles, pp. 201-310. In C.
Gans, A. d’A. Bellairs and T. S. Parsons (eds.),
Biology of the Reptilia, Vol. 1. London,
Academic Press.

Huxuey, T. H. 1859. On the dermal armour of
Jacare and Caiman, with notes on the specific
and generic characters of recent Crocodilia. J.
Proc. Linn. Soc. Zool. Lond., 4: 1-28.

DORSAL ARMOR OF CROCODILIA : Ross and Mayer

KALIN, J. 1955. Crocodilia, pp. 695-784, In J.
Piveteau (ed.), Traité de Paléontologie, Vol. 5.
Paris, Masson.

KerFroot, W. C. 1970. The effect of functional
changes upon the variability of lizard and
snake body scale numbers. Copeia, 1970: 252-
260.

MANTER, J. T. 1940. The mechanics of swimming in
the alligator. J. Exp. Zool., 83: 345-358.
MEDEM, F. J. 1958a. Informe sobre reptiles

Colombianos, III. Investigaciones sobre la
anatomia craneal, distribucion geografica y
ecologia de Crocodylus intermedius (Graves)

en Colombia. Caldasia, 8: 175-215.

___. 1958b. The crocodilian genus Paleosuchus.
Fieldiana, Zoology, 39: 227-247.

___.. 1970. Sobre un hibrido inter-especifico del
genero Paloosuchus (Crocodylia. Alligatori-
dae). Rev. Acad. Colombiana de Cien, Exactas
Fis. y Nat., 13: 467-479.

MILLER, W. E. 1980. The late Pliocene Las Tunas
local fauna from southernmost Baja California,
Mexico. J. Paleontology, 54: 762-805.

Mook, C. C. 1941. A new crocodilian, Hassiacosu-
chus kayi, from the Bridger Eocene Beds of
Wyoming. Ann. Carnegie Mus., 28: 207-221.

NasH, D. S. 1975. The morphology and relation-
ships of a crocodilian, Orthosuchus storm-
bergi, from the Upper Triassic of Lesotho.
Ann. S. Afr. Mus., 67: 227-329.

REIG, O. A. 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. 1956. Osteology of the reptiles. Chi-
cago, Univ. Chicago Press.

1972. The Chanares (Argentina) Triassic
reptile fauna. XII. The postcranial skeleton of
the thecodont Chanaresuchus. Breviora Mus.
Comp. Zool. No. 385, pp. 1-21.

___. 1974. Aquatic adaptation in reptiles—primary
or secondary? Ann. S. Afr. Mus., 64: 221-230.

Ross, C. A., AND C. D. Roberts. 1979. Scalation of

331

the American alligator. U.S. Dept. Int. Fish
and Wildl. Serv. Spec. Sci. Rep.—Wildl. No.
22

Ross, C. A., AND F. D. Ross. 1974. Caudal scalation
of Central American Crocodylus. Proc. Bio.
Soc. Wash., 87: 231-233.

SCHMIDT, K. P. 1956. On the status and relations of
Crocodylus mindorensis. Fieldiana, Zoology,
33: 535-539.

SEIDEL, M. R. 1979. The osteoderms of the Ameri-
can alligator and their functional significance.
Herpetologica, 35: 375-380.

TROXELL, E. L. 1925. Hyposaurus, a marine croco-
dilian. Amer. J. Sci., 9: 489-514.

VARONA, L. S. 1966. Notas sobre los crocodilidos de
Cuba y descripcion de una nueva especie del
Pleistoceno. Poeyana Ser. A, 16: 1-34.

WALKER, A. D. 1961. Triassic reptiles from the Elgin
area: Stagonolepis, Dasygnathus and _ their
allies. Phil. Trans. Roy. Soc. Lond. Ser. B, 244:
103-204.

___. 1968. Protosuchus, Proterochampsa, and the
origin of phytosaurs and crocodiles. Geol.
Mag., 105: 1-14.

WELLNHOFFER, P. 1971. Die Atoposauridae (Croc-
odylia, Mesosuchia) der Oberjura—platten-
kalke Bayerns. Palaeontographica Abt. A, 138:
133-165.

WERMUTH, H. 1953. Systematik der rezenten
krokodile. Mitteil. Zool. Mus. Berlin, 29: 375—
514.

WERMUTH, H., AND R. MERTENS. 1961. Schildkro-
ten, krokodile, bruckenechsen. Gustav Fischer
Verlag, Jena.

SAND . 1977. Liste der rezenten amphibien
und reptilien. Testudines, Crocodylia,
Rhynchocephalia. Das Tierreich 100.

WILLIAMS, E. E. 1959. Gadow’s arcualia and the
development of tetrapod vertebrae. Quart.
Rev. Bio., 34: 1-32.

YOUNGPRAPAKORN, U. 1976. Samutprakan, Thailand.
Borpit, Thailand.

The Tarsus and Metatarsus of Protosuchus
And Its Phyletic Implications

MAX K. HECHT '

SAMUEL F. TARSITANO2

ABSTRACT. The tarsus and metatarsus of the
Triassic crocodilian Protosuchus is compared with
that of modem Alligator. The two types of tarsi are
very similar, indicating that the present crocodilian
tarsus is a primitive feature of the Order. Important
differences occur only in the tibial-astragalar artic-
ulation, the position and development of the distal
roller of the astragalus, and degree of development
of the fifth metatarsal.

INTRODUCTION

The Order Crocodilia is of great inter-
est to paleobiology because of its longev-
ity as an amniote taxon and because it is
one of the two survivors of the archosau-
rian adaptive radiation. The earliest
known crocodilian genera are of Triassic
age and generally placed in the suborder
Protosuchia (Bonaparte, 1978; Colbert
and Mook, 1951; Crompton and Smith,
1980; Nash, 1975; Steel, 1973). It is not a
prerequisite axiom, but it is generally ex-
pected that the primitive state of a mor-
phological transformation series is to be
found among the earliest fossil represen-
tatives of the lineage. One of the features
which has been used in determining
lineages among early reptiles has been
the tarsus and metatarsus. The tarsal
region has undergone rapid change
among the early Archosauria but unfor-

!2 Department of Biology, Queens College of the
City University of New York, Flushing, New York
11367, U.S.A.

tunately has not been well preserved in
most fossils. It is for this reason that a
description of the preserved tarsal and
metatarsal elements of Protosuchus is re-
quired.

The description of the type material of
Protosuchus indicated the presence of an
almost complete tarsus and metatarsus
(Colbert and Mook, 1951). Further pre-
paration of the material has now made it
possible to describe in detail the earliest
known crocodilian tarsus and to make
further anatomical comparisons. In order
to properly evaluate these new data a
detailed osteological study of a living
eusuchian crocodilian is required. Until
now only general descriptions were
available for the various crocodilian
tarsal elements (Schaeffer, 1941;
Cruickshank, 1979; Brinkman, 1980b).
Schaeffer (1941) produced the pioneer
study on the amphibian and reptilian
tarsus that illustrated the functional and
phylogenetic problems within the Reptil-
ia. Krebs (1963) discussed the tarsus as
an archosaurian character. Cruickshank
(1979) discussed the tarsus as an impor-
tant character and indicator of lineages
and functional trends within the early
Archosauria. He also described the basic
morphology of the archosaurian tarsus
and clearly demonstrated the complex
interrelationships among the component
parts. Cruickshank stated that early in the
history of the Archosauria two types of
tarsal morphology and function appeared

TARSUS AND METATARSUS OF PROTOSUCHUS : Hecht and Tarsitano

in the groups above the proterosuchian
level. Chatterjee (1978) and Cruickshank
(1979) called these two types, croc-
odile-normal and_ crocodile-reversed.
Cruickshank postulated that these two
types were derived from a primitive
archosaurian condition, the protero-
suchian condition, which had a relatively
mobile relationship between the astraga-
lus and calcaneum. The crocodile-normal
condition was represented by the astraga-
lus and its peg articulating with a
calcaneum and its socket. In this study
Cruickshank (1979) developed a morph-
ological terminology to aid in the de-
scription of these complex structures and
articulations. He used the genus Croco-
dylus as the typical eusuchian crocodile
and the basis of his comparison with
other Archosauria. The variation within
the living Eusuchia was not considered
nor that of other crocodilians. Thulborn
(1980) has also discussed the tarsus of
early archosaurians but primarily at the
proterosuchian level. Brinkman (1980b)
described the mechanics of the eusu-
chian tarsus and metatarsus using
Caiman as a model. The preservation
of tarsal elements among fossil croco-
dilians is not frequent and is unknown in
Hemiprotosuchus (Bonaparte, 1971).
The tarsus, metatarsus, and phalanges
of Protosuchus are almost completely
preserved in the type material of P.
richardsoni (AMNH 3024). It is the in-
tent of this study to compare the tarsus
and metatarsus of Protosuchus with the
typical crocodilian tarsus as represented
by members of Eusuchia. In order to ac-
complish this comparison it is necessary
to describe in detail those components of
the eusuchian tarsus which are preserved
in Protosuchus. Schaeffer (1941) and
Cruickshank (1979) have provided us
with a brief description and terminology
based primarily on Alligator and Croc-
odilus. Brinkman (1980b), not knowing of
Cruickshank’s work, described the tarsus
using a different terminology. All of the
preceding authors considered the func-
tional aspects of the tarsal elements.

333

THE MORPHOLOGY OF THE EUSUCH-
IAN TARSUS AND METATARSUS

As an example of a typical crocodilian
we are utilizing the tarsus of the genus
Alligator. The illustrations are of a large
Alligator mississippiensis in our own
collection (R1200) which is believed to
have had a total length of eleven feet.
The tarsus and metatarsus was well
ossified and preserved and all facets and
condyles well defined. The morphologi-
cal terminology used shall be that of
Cruickshank (1979). Certain modifica-
tions and additions will be required in
this terminology. Morphological varia-
tion in the Eusuchia will be noted.

ASTRAGALUS

The astragalus of eusuchian crocodiles
is firmly attached although not fused to
the tibia (Brinkman, 1980b). The prox-
imal surface of the astragalus (Figs. 1, 2)
presents a U-shaped tibial facet. The
arms of the U are directed laterally while
the belly of the U faces medially. The
anterior arm of the U is much longer than
the posterior arm and culminates in a
ridge. This ridge marks the end of the
articular surface of the tibial facet and the
beginning of the fibular facet lying per-
pendicular to it. Between the arms of the
U-shaped tibial facet lies a moderately
large fossa, the astragalar fossa. It is filled
in life with connective tissue which con-
ceals many nutritive foramina to the
astragalus and helps bind it to the tibia.
The fibular facet is a slightly concave
uneven rectangular surface which artic-
ulates with the medio-distal surface of
the fibula through an interposed men-
iscal cartilage.

The distal surface of the astragalus
(Figs. 3, 4) is slightly complex in that it
consists of two separate articular surfaces
adjacent to one another but separated by
a groove. The larger of the two articular
surfaces is the anteriorly directed distal
roller (Cruickshank, 1979) which articu-
lates with metatarsals 1 and 2. The distal
roller extends from the tibial facet prox-

334

imally and anterior hollow laterally and
continues around the anterior (distal) sur-
face of the astragalus to its posterior
surface. The second articular surface is a
complex low bent scroll-shaped mound
henceforth referred to as the astragalar
trochlea. The lateral surface of the
trochlea forms a prominent feature, the
astragalar peg. The surface of the trochlea
articulates with a concave shelf of the
calcaneum (Cruickshank, 1979) or part of
the secondary articulation of Schaeffer
(1941). The astragalar peg forms the
primary articulation between the socket
of the calcaneum and astragalus (ball and
socket of Schaeffer, 1941). This peg is
formed by the lateral most edge of the
astragalar trochlea and articulates with
the socket of the calcaneum (concave
facet of the calcaneum, Cruickshank,
1979).

The posterior surface (Fig. 4) of the
astragalus contains the trochlea de-
scribed above, part of the distal roller and
a large groove. This groove is hereby
designated the articulating channel. It is
bounded dorsally by the fibular facet,
ventrally by the trochlea and peg, and
extends medioventrally to become con-
tinuous with the lesser groove lying be-
tween the distal roller and the medial
eminence of the trochlea. The articulat-
ing channel also extends to the lateral
surface of the astragalus and receives the
leading edge of the calcaneal tongue as
we calcaneum rotates about the astrag-
alus.

It should be noted that we are not in
agreement with some of Cruickshank’s
terminology because of a contradiction
between his labels and orientation of the
astragalus (Cruickshank, 1979: Fig. 12).
His labeled “groove” in Figures 12b and
12c appear to be different structures and
are therefore confusing.

CALCANEUM

Proximally the calcaneum has a fibular
condyle, which produces a single move-
able articulating surface with the fibula.
The fibular condyle (the fibular facet of

Advances in Herpetology and Evolutionary Biology

Cruickshank, 1979) is rounded proxi-
mally but becomes abruptly flattened
distally for its articulation with distal
tarsal 4, indicating that the movement at
this latter joint is restricted.

Medially the calcaneum (Fig. 5) bears
two articulating surfaces for the astrag-
alus. The most anterior of the two is a
socket (Schaeffer, 1941; or the concave
facet of Cruickshank, 1979), which re-
ceives the peg of the astragalus. Posterior
to this socket is a medially directed
tongue of bone (Cruickshank, 1979)
which is concave anteriorly and _ artic-
ulates with the rounded trochlea of the
astragalus. The dorsal lip of the tongue
articulates with the astragalar groove.

Posterolaterally the calcaneum bears a
large tuber (Figs. 5, 6) which is distinct
from the fibular condyle anteriorly and
the tongue anteromedially. On its pos-
terior surface the calcaneal tuber bears a
deep groove for the tendons of the long
pedal flexors (Brinkman, 1980b), hence-
forth referred to as the deep groove (Fig.
6). It should be noted that most of the M.
gastrocnemius passes over the groove or
medial to the calcaneal tuber. Most of the
long pedal flexor muscles insert on the
fifth metatarsal, the tuber changing the
application of the force of M. ambiens, M.
flexor tibialis externus, and the fibular
head of the M. gastrocnemius. One
muscle, M. fibulo-calcaneum, does insert
onto the dorsal surface of the tuber (Fig.
5). Below this point lies a centrally
directed channel (Fig. 5) which termi-
nates at the bridge. On the lateral side of
the bridge there lies a lateral channel
(Fig. 2).

FOURTH DISTAL TARSAL

Distal tarsal 4 (Fig. 12) is a triangular
bone oriented so that its length is longer
than its width. Although the bone is
small, it has many articular surfaces.
Proximally there is a dorsoventrally
broadened flat area which articulates
through a meniscus with the distal tarsal
4 facet of the calcaneum. Ventral and dis-
tal to the above facet is a nonarticular

TARSUS AND METATARSUS OF PROTOSUCHUS : Hecht and Tarsitano

concavity perforated with many nutritive
foramina. Anterior to this concavity a
slightly concave facet extends in a
postero-dorsoventral direction to articu-
late with the fifth metatarsal. The distal
surface of the bone is imperfectly con-
cave with an indentation in the middle of
its surface. Above the indentation a large
eminence bears the articular surfaces for
metatarsal 3 (medial surface) and meta-
tarsal 4 (lateral surface). Viewed medially
the bone is heart shaped, and its entire
surface articulates with distal tarsal 3.

FIFTH METATARSAL

The fifth metatarsal of all crocodilians
is reduced and bears no phalanges. This
bone viewed dorsally is roughly triangu-
lar in shape and is thickened proximally,
where there is an elongate convex area
for the articulation with distal tarsal 4.
The anterolateral surface bears a conical
shaped projection which comprises one-
third of the total length of the bone.
Dorsoproximally there is a prominence
which tapers posterolaterally and bears a
scarred surface for the insertion of part of
M. gastrocnemius, other pedal flexor
muscles and the cartilage which lies
between the calcaneum and fourth distal
tarsal (Schaeffer, 1941; Brinkman,
1980b).

OTHER METATARSALS

The metatarsals of crocodilians are
modified for a sprawling gait (Brinkman,
1980b) but differently from the sprawl-
ing gait of lizards (Brinkman, 1980a). In
the Crocodilia this adaptation manifests
itself in the torsion and compression of
the proximal portion of all of the meta-
tarsals except 5. The first metatarsal is the
most robust and the least modified of the
four. It is dorsoventrally flattened, and its
lateroproximal surface lies on top of
the dorsomedial surface of metatarsal
2. Metatarsal 2 is slightly proximally
twisted and compressed so that its broad
surface faces dorsomedially and under-
lies the proximal surface of metatarsal 1.
Metatarsal 3 is similar in form to meta-

335

tarsal 2, and its proximal portion lies
under metatarsal 2. Metatarsal 4 is the
most slender of the five metatarsals
similar to the preceding metatarsals in
form. The proximolateral surface of meta-
tarsal 4 is covered by metatarsal 3. Meta-
tarsal 1 articulates proximally with the
distal roller of the astragalus. Metatarsal 2
articulates proximomedially with the dis-
tal roller and proximolaterally with distal
tarsal 3. Metatarsal 3 articulates proxi-
momedially with distal tarsal 4. Meta-
tarsal 4 articulates proximally with the
dorsodistal surface of tarsal 4, whereas
metatarsal 5 articulates proximally with
the ventrolateral surface of tarsal 4.

VARIATION AMONG LIVING EUSUCHIA

The preceding description was based
on the genus Alligator, of a particularly
large individual. Some of the osteological
features are not as prominent in smaller
specimens or in smaller species. There-
fore there appears to be an allometric
component, which is primarily reflected
in slight changes in proportion and
degree of ossification. Comparisons with
other genera indicate a similar degree of
variation as exemplified by the figures of
Brinkman (1980b) of Caiman, and
Cruickshank (1979) and Thulborn (1980)
of different species of Crocodylus. An
examination of Gavialis, the most di-
vergent of the living forms, demonstrates
similar minor alterations but the basic
crocodilian tarsus and metatarsus is still
present. Compared to Alligator, Gavialis
has a reduced deep groove of the cal-
caneal tuber, slightly less spheroidal
distal roller and slightly more robust
metatarsals.

THE MORPHOLOGY OF THE
PROTOSUCHIAN TARSUS AND
METATARSUS

ASTRAGALUS

The astragalus of Protosuchus (Figs. 7—
9) differs from that of modern croc-
odilians in some proportional and minor

336

morphological features. The proximal
surface of the astragalus differs from
eusuchian crocodilians by the presence
of a deep channel which receives the end
of the tibia, thus locking it into place. The
tibial facet is not U-shaped as in the
Eusuchia, though the fibular facet is
almost perpendicular to the tibial facet as
it is in modem crocodilians.

The distal roller is prominent, as in
modern crocodilians, but the orientation
is more medial and vertical in relation to
the remainder of the astragalus. This dif-
ferent orientation causes the tibia be held
at a more acute angle in order to enable
metatarsal 1 and 2 to contact the ground
(compare Figs. 8 and 12).

The anterior hollow adjacent to the dis-
tal roller exists in Protosuchus but is not
as depressed (Fig. 8) as in the modem
Crocodilia.

The medial surface of the astragalus
bears the partially destroyed peg and a
shallow articulating channel (Fig. 9).
This channel intersects the lesser groove
which is not as well developed as in the
eusuchians. There is no astragalar fossa
in Protosuchus because of the absence of
the posterior arm of the tibial facet. As in
modern forms the astragalar trochlea is
directed posterio-laterally to receive the
calcaneal tongue.

CALCANEUM

The calcaneum of Protosuchus bears a
striking resemblance to the typical eusu-
chian despite the latero-medial compres-
sion in the preserved specimen. This
severely compressed portion of the cal-
caneum includes the fibular condyle, dis-
tal tarsal 4 facet, and part of the socket. In
the lateral view the proximal surface
bears a rounded fibular condyle. As in all
Eusuchia, there is a distal flattened
surface below the fibular condyle for the
articulation with distal tarsal 4 (Fig. 10).

The tuber of the calcaneum is a promi-
nent structure distinct from the re-
mainder of the calcaneum. Although the
posterior surface of the tuber is distorted

Advances in Herpetology and Evolutionary Biology

and damaged in the specimen there are
clear indications of the deep groove of
the tuber (Fig. 11).

On the medial surface there is a
medially directed tongue bearing an
anteriorly facing concave surface which
articulates with the astragalar trochlea.
The medial surface of the calcaneum has
the identical structures as the Eusuchia,
the most prominent of which is the
socket.

FOURTH DISTAL TARSAL

Distal tarsal 4 is preserved in the natural
position but is clearly medio-laterally
compressed. As a result the bone is dis-
torted into a half moon shaped disc.
Laterally the articulating surface for
metatarsal 5 has been exposed by the
rotation of metatarsal 5. The posterior
surface is flattened for the articulation
with the distal tarsal 4 facet of the calca-
neum. The anterior or distal surface of
this bone is rounded for articulation with
metatarsal 4 and 3. The medial surface is
flattened, but this may be an artifact of
preservation.

FIFTH METATARSAL

The fifth metatarsal is reduced and as
in modern Crocodilia bears no pha-
langes. Proximally the fifth metatarsal is
bulbous and tapers distally into a long
fingerlike projection. The entire struc-
ture is one third the length of metatarsal
De

The position of metatarsal 5 in the
specimen appears to have been altered
from its natural position. The bone has
been rotated approximately 90° exposing
the articular surface of distal tarsal 4.
Thus the lateral view as exposed in the
fossil represents the dorsal bone surface
and the bottom view, the lateral surface.
It is possible that the entire fifth meta-
tarsal may be medio-laterally com-
pressed. The true dorsal surface bears a
fossa for probable muscle insertions, pos-
sibly M. peroneus anterior and posterior.
The true lateral surface bears a deep

TARSUS AND METATARSUS OF PROTOSUCHUS : Hecht and Tarsitano

groove or elongate pocket. If this is a
natural feature, it would be for the pas-
sage and insertion of part of the gastrocne-
mius muscle.

OTHER METATARSALS AND PHALANGES

The metatarsals of Protosuchus have
the same relationships to each other in
morphology, proportion, lengths, and
articulations as those of modem eusu-
chians (Colbert and Mook, 1951).

The number of phalanges is similar to
most modern Eusuchia having a _ pha-
langeal formula of 2-3-44-0 (Colbert
and Mook, 1951).

CONCLUSIONS

THE PROTOSUCHIAN AND EUSUCHIAN
TARSUS AND METATARSUS COMPARED

1) The crocodilian “normal” tarsus
and its metatarsus were completely
developed by the earliest known Croc-
odilia.

2) Between the protosuchian level and
the eusuchian level the tarsus developed
greater internal mobility by the reduction
of the channel between the astragalus
and the tibia and the development of the
ligamentous support system described by
Brinkman (1980b).

3) The position of the distal roller in
Protosuchus indicates a more primitive
sprawling limb position (Brinkman
1980a, b; Rewcastle, 1980) than is found
in modern Crocodilia and the probable
inability of the protosuchian crocodile to
develop some of the advanced gaits de-
scribed by Brinkman (1980b).

4) The longer and perhaps more com-
plex fifth metatarsal of Protosuchus sug-
gests a more primitive condition but only
slightly different from the eusuchian
condition. It is evident from the muscle
attachments that the vestige of the fifth
digit, the fifth metatarsal, in crocodilians
cannot be reduced for reasons outlined
by Russell and Rewcastle (1979) for the
lizard Sitana.

337

The functions of the hooked fifth meta-
tarsal and its distribution among reptilian
taxa have been described by Robinson
(1975). The hooking of the fifth metatar-
sal is apparently primitive as indicated
by its presence in some early eosuchians
(Gow, 1975; Carroll, 1975). The croc-
odilian condition must be reduced as far
as the loss of the distinct hooking pattern.

THE CROCODILIAN AND PRIMITIVE
ARCHOSAURIAN TARSUS

Thulborn’s analysis (1980) of the
ankles of archosaurian reptiles demon-
strates that previous investigators had
oversimplified the primitive proterosu-
chian level of tarsal development
(Charig, 1972; Cruickshank, 1979;
Hughes, 1968; Krebs, 1963; Reig, 1970).
It is obvious that the complexity of the
morphological transformation series has
been underestimated and the polarities
of the tarsal complex have not yet been
determined. Thulborn’s analysis implies
that he has removed reversals and con-
vergences from the tarsal morphocline,
but instead he has created paraphyletic
and polyphyletic taxa. From our view-
point he has not removed the contradic-
tions in the systematics of the group
(Charig et al, 1976; Krebs, 1974; Reig,
1970: Romer, 1972: Sill, 1967, 1974;
Steel, 1973; Tarsitano and Hecht, 1980;
Walker, 1972; Whetstone and Martin,
1979), but he has pointed out a serious
conflict in the morphological transforma-
tion series (Carroll, 1976; Cruickshank,
1975; Ewer, 1965; Krebs, 1963). At the
present time the most important question
is whether the “crocodile-normal” tarsus
is an indicator of a monophyletic origin of
taxa or whether it has evolved independ-
ently several times (Thulbom, 1980). An
examination of the figures of Chatterjee
(1978) indicates such similarity in the
basic structure of the phytosaurian tarsus
as to indicate that such groups must share
a common ancestor with crocodilians.
The crocodile-normal and_ crocodile-
reversed tarsi are such complex struc-

tures as to-require monophyletic origins
for the lineages they characterize.

ACKNOWLEDGMENTS

The authors wish to express their
thanks to E. Gaffney of the American
Museum of Natural History for permis-
sion to prepare the material and for his
encouragement in making the study; the
PSC/BHE faculty research grants (No.
13317) of the City University of New
York for the financial support for this
study; Matthew Hecht for aid in prepara-
tion of the photographs; Otto Simonis for
his skill in preparation; and Robin Eisner
for one diagram.

The first author would like to express
his appreciation to Emest E. Williams for
the many hours of discussion, begun
years ago, that has made the study of
Evolutionary Biology such a great game.

APPENDIX

The following terminology and their
abbreviations have been used in the
figures for structures on the astragalus
and calcaneum:

ASTRAGALUS
Ac, articulating channel; Ah, anterior hollow; Af,
astragalar fossa; At, astragalar trochlea; Atc, astraga-
lus-tibial contact; Dr, distal roller; Ff, fibular facet;
Lg, lesser groove; P, peg; Tf, tibial facet.

CALCANEUM
B, bridge of tuber caleaneum; Dg, deep groove;
Dt4f, facet of distal tarsal 4; Fc, fibular condyle; Lc,
lateral channel; Mc, medial channel; S, socket; T,
tongue of the calcaneum; Tc, tuber calcaneum.

LITERATURE CITED

BONAPARTE, J. F. 1971. Los tetrapodos del sector
superior de ls Formacion Los Colorados, La
Rioja, Argentina (Triasico sup.) Opera lilloana,
Pipep

___. 1978. E] Mesozoico de America del Sur y sus
Tetrapodos. Opera lilloana, 26: 1-307.

BRINKMAN, D. 1980a. Structural correlates of tarsal
and metatarsal functioning in Iguana (Lacer-
tilia; Iguanidae) and other lizards. Can. J.
Zool., (2)58: 277-289.

Advances in Herpetology and Evolutionary Biology

___. 1980b. The hind limb step cycle of Caiman
sclerops and the mechanics of the crocodile
tarsus and metatarsus. Can. J. Zool., (12)58:
2187-2200.

CARROLL, R. L. 1975. Permo-triassic “lizards” from
the Karroo. Palaeont. Africana, 18: 133-135.

1976. Noteosuchus-The Oldest Known
Rhynchosaur. Ann. S. Afr. Mus., 72: 37-57.

CuHanrIG, A. J. 1972. The evolution of the archosaur
pelvis and hindlimb: an explanation in func-
tional terms, pp. 121-155. In K. A. Joysey and
T. S. Kemp, (eds.), Studies in Vertebrate
Evolution. Oliver and Boyd.

CHARIG, A. J., B. KREBS, H. D. SUESS, AND

F. WESTPHAL. 1976. Thecodontia. Handbuch der
Palaoherpetologie G. Fischer Verlag,
Stuttgartt.

CHATTERJEE, S. 1978. A primitive parasuchid
(Phytosaur) reptile from the Upper Triassic
Maleri formation of India. Palaeontology,
(1)21: 83-127.

COLBERT, E.. H., AND C. C. Mook. 1951. The ances-
tral crocodilian Protosuchus. Bull. Amer. Mus.
Nat. Hist., 97: 147-192.

CROMPTON, A. W., AND K. K. SmiTH. 1980. A new
genus and species of crocodilian from the
Kayenta formation (Late Triassic) of Northem
Arizona, pp. 193-217. In L. L. Jacob, (ed.),
Aspects of Vertebrate History. Museum of
Northern Arizona Press.

CRUICKSHANK, A. R. I. 1975. The affinities of Pro-
terochampsa_ barrionuevoi Reig. Palaeont.
Africana, 18: 133-135.

___. 1979. The Ankle Joint in Some Early Archo-
saurs. South Afric. J. Sci., 75: 168-178.

Ewer, R. F. 1965. The anatomy of the Thecodont
Reptile Euparkeria capensis Broom Phil
Trans. Roy. Soc. Lond. (B), 248: 379-435.

Gow, C. E. The morphology and relationships of
Youngina capemsis Broom and Prolacerta
broomi Parrington. Palaeont. Africana, 18: 89-
131.

HUuGHES, B. 1968. The tarsus of rhynchocephalian
reptiles. J. Zool. Lond., 156: 457-481.

Kress, B. 1963. Bau und Function des Tarsus eines
Pseudosuchiers aus der Trias des Monte San
Giorgio (Kanton Tessin, Schweiz). Palaont.
Zeit., 57: 88-95.

____. 1974. Die Archosaurier. Naturwiss., 61: 17-24.

NasH, D. S. 1975. The morphology and relation-
ships of a crocodilian, Orthosuchus storm-
bergi, from the Upper Triassic of Lestho. Ann.
S. Afr. Mus., 67(7): 227-329.

REIG, O. A. 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.

REWCASTLE, S. C. 1980. Form and function in lacer-
tilian knee and mesotaral joints; a contribution
to the analysis of sprawling locomotion. J.
Zool., 191: 171-192.

RoBINSON, P. L. 1975. The functions of the hooked

TARSUS AND METATARSUS OF PROTOSUCHUS « Hecht and Tarsitano

metatarsal in lepidosaurian reptiles. Colloque
in “Problemes actuels de Paleontologie-Evo-
lution des Vertebres.”” Colloque Intemational
C.N.R.S. No. 218, pp. 464483.

Romer, A. S. 1972. The Chanares (Argentina)
Triassic Reptile Fauna. XVI Thecodont Clas-
sification. Breviora Mus. Com. Zool. No. 395,
pp. 1-24.

RUSSELL, A. P., AND S. C. NEWCASTLE. 1979. Digital
reduction in Sitana (Reptilia: Agamidae) and
the dual roles of the fifth metatarsal in lizards.
Can. J. Zool., 57: 1129-1135.

SCHAEFFER, B. 1941. The Morphological and Func-
tional Evolution of the Tarsus in Amphibians
and Reptiles. Bull. Amer. Mus. Nat. Hist.,
(VI)LXXVIII: 395-472.

SILL, W. D. 1967. Proterochampsa barrionuevoi
and the early evolution of the Crocodilia. Bull.
Mus. Comp. Zool., 135: 415-446.

339

____. 1974. The anatomy of Saurosuchus galilei and
the relationships of the rauisuchid thecodonts.
Bull. Mus. Comp. Zool., 146: 317-362.

STEEL, R. 1973. Crocodylia, pp. 1-116. In O. Kuhn
(ed.), Handbuch der Palaeoherptologie: Part
16. G. Fischer Verlag.

TARSITANO, S., AND M. K. HEcuT. 1980. A recon-
sideration of the reptilian relationships of
Archaeopteryx. Zool. J. Linn. Soc., 69: 149-
182.

THULBORN, R. A. 1980. The ankle joints of archo-
saurs. Alcheringia, 4: 241-261.

WALKER, A. D. 1972. New Light on the origin of
birds and crocodiles. Nature, London, 237:
257-263.

WHETSTONE, K. N., AND L. D. MARTIN. 1979. New
look at the origin of birds and crocodiles.
Nature, London, 279: 234-236.

Z vances } 20,
in Herpetology and Evolutio
nary Biologu
SY

Ff

. |

TARSUS AND METATARSUS OF PROTOSUCHUS : Hecht and Tarsitano 341

ie

K =|

Figure 2. Proximal surface of articulated calcaneum and astragalus of the left side of Alligator and explanatory
diagram. For symbols see text.

Dt 4f

342 Advances in Herpetology and Evolutionary Biology

Tf

Dr

Figure 3. Distal view of the left astragalus of Alligator and explanatory diagram. For symbols see text.

TARSUS AND METATARSUS OF PROTOSUCHUS - Hecht and Tarsitano 343

Ai

Ff

Figure 4. Posterior view of the right astragalus of Alligator and explanatory diagram. For symbols see text.

344 Advances in Herpetology and Evolutionary Biology

Ee

lic

Dt4t

Figure 5. Medial view of the right calcaneum of Alligator and explanatory diagram. For symbols see text.

TARSUS AND METATARSUS OF PROTOSUCHUS : Hecht and Tarsitano 345

Figure 6. Plantar view of right calcaneum of Alligator and explanatory diagram. For symbols see text.

346 Advances in Herpetology and Evolutionary Biology

Figure 7. Proximal view of right astragalus of Protosuchus and explanatory diagram. For symbols see text.

TARSUS AND METATARSUS OF PROTOSUCHUS : Hecht and Tarsitano 347

Ff

Sp

Figure 8. Anterior view of the tibia articulating with astragalus of the left side of Protosuchus and explanatory
diagram. For symbols see text.

348 Advances in Herpetology and Evolutionary Biology

Ff
AG ci

P

Figure 9. Poster view of the tibia articulating with astragalus of the left side of Protosuchus and explanatory
diagram. For symbols see text.

TARSUS AND METATARSUS OF PROTOSUCHUS : Hecht and Tarsitano 349

Dt4f

Figure 10. Medial view of left calcaneum of Proto-
suchus and explanatory diagram. See text for symbols.

F Figure 12. Anterior view of the articulated tarsal skele-
ton of Alligator illustrating its relationships to tibia, fib-
ula, and metatarsals.

fe
Dt4f

Figure 11. Lateral view of left calcaneum of Proto-
suchus and explanatory diagram. See text for symbols.

About the Herpetofauna of Central African

Montane Forest
R. F. LAURENT !

ABSTRACT. The central African montane forest
exists as a series of discontinuous habitats where
incipient speciation may be occurring. The herpeto-
fauna of the area is examined in regard to micro-
habitat preference and geographic differentiation.

INTRODUCTION

Exploration of the central African
region of the Great Lakes of the Wester
Graben has focused on Mt. Ruwenzori
and the Kivu and Rwanda volcanoes.
Other parts of the Nile-Zaire dividing
montane range as well as the mountains
west of Lake Kivu, of the Ruzizi and of
Lake Tanganyika have been poorly
explored. The collections made by the
author in these little known regions dur-
ing the 1950's are interesting from the
point of view of the differentiation of
isolated faunas. Indeed, the montane
forest is a highly discontinuous area in
central Africa. We can actually count no
less than eleven isolated areas: 1) Mt.
Ruwenzori; 2) the highlands west of Lake
Edward with Mt. Tshiaberimu and scat-
tered mountain forest islands (Butembo,
Lubero) in a largely deforested zone; 3)
the lakes Mokoto region; 4) the Virunga
volcanoes; 5) the highlands west of Lake
Kivu with Mt. Kahusi and Mt. Biega; 6)
the Idjwi island forests; 7) the Itombwe
highlands where the ordinary montane
forest surrounds a large grassy central
area, the fauna of which is largely

1Fundacion Lillo,
Tucumdn, Argentina.

Miguel Lillo 205, 4.000

endemic (Laurent, 1964); 8) the high-
lands between Fizi and Kalemie with
Lubitshako and Mt. Kabobo; 9) the
Lutsiro highlands of western Rwanda
between Kisenyi and Kibuye; 10) the
Rugege forest between Bukavu and
Butare; 11) the Bururi forest (see Fig. 1).
Other practically unknown isolated areas
are in the Ituri (Nioka and Blukwa
region) and in western Tanzania
(Kungwe Mountains).

These forests were interconnected and
occupied a larger united area during the
last pluvial period, 18 to 25 thousand
years ago (Moreau, 1966). Therefore,
their flora and fauna is overall rather uni-
form. It was, however, an attractive
problem to investigate if some differen-
tiation had taken place between the iso-
lated parts of this peculiar “milieu”
during this relatively short span of time.
It must be noted that in some cases the
isolation must have been more recent
than in others. For instance, the Ituri
highlands, the Ruwenzori, the Kungwe
Mountains and, to a lesser extent, the
Kabobo area have been isolated longer?
than the Kahusi and Itombwe range or
the forest of the westerm Rwanda and
Burundi where human factors have been
at work during a very recent past, as sug-
gested by the high density of population
in these countries.

2 Evidence for this comes from a) a comparison of
the fauna (and flora) on the regions under study; b)
the distance; and, c) the depth of the depression
between them.

CENTRAL AFRICAN MONTANE FOREST - Laurent

Mt Tshiaberrimu
highlands of Butembo
and Lubero

351

Lakes Mokoto

Mt Ruwenzori

region
Virunga
volcanoes
Mt Kahusi
and
Mt Biega
Idjwi Id Lutsiro
highlands
ltombwe
highlands
Rugege
forest
Lubitshako 5
ururl
forest
Mt Kabobo

Ee a ay!

26°

pO tee

Figure 1. Map of central Africa (eastern Congo, southwestern Uganda, Rwanda, Burundi, and northwestern
Tanzania). Striped areas show highlands mostly covered with montane forest. Some include other formations
generally at altitudes above 3,000 m (Ruwenzori, Virunga Volcanoes, Itombwe Highlands).

Another feature must be emphasized.
On the slopes of the higher mountains
which were first explored (Ruwenzori
and Kivu volcanoes), the flora is rather
distinctly stratified: regular montane
forest with Podocarpus, Bamboos,
Hagenias and alpine meadows. Else-

where, in larger areas between 2,000 m
and 3,000 m, the first three facies are
more or less interwoven and mixed.
Another consequence of the larger
surface of the pertinent regions outside
the Ruwenzori range and the Kivu vol-
cano is that larger swamps can generally

So> Advances in Herpetology and Evolutionary Biology

be found, for instance, in the vicinity of
Mt. Biega and Kahusi near Bukavu. Table
1 shows the habitats of the species en-
countered in the region. This scheme is
not a rigid one. Many species, namely the
commonest which are also generally the
most adaptable, can be found in several
kinds of biotopes besides their preferred
habitat.

SPECIES OF THE REGION

In the large swamps, the commonest
amphibian species of the region are
found in great numbers. This is indeed
a very striking feature of the amphibian
fauna in the montane forest; specimens
can easily be collected during the
day, while at lower altitudes they hide
by day and begin to be active at twilight.

The swamp species are Xenopus laevis
victorianus Ahl and some polyploids
such as the recently described X. vestitus
Laurent, X. ruwenzoriensis Tymowska
and Fischberg, X. wittei Tinsley and
others still undescribed, Bufo kisoloen-
sis Loveridge, Phrynobatrachus versi-
color Ahl, Phrynobatrachus bequaerti
(Barbour and Loveridge), Leptopelis
kivuensis Ahl, and Hyperolius castaneus
Ahl. All of them also inhabit the small
marshes which are scattered in the forest
and overshadowed by trees. They can
also be found in the wells dug by mining
prospectors. Except for Xenopus laevis,
they are often found rather far away from
water, but Phrynobatrachus versicolor is
generally in closer contact with water
than the other forms. Some species live in
the shadowy puddles of the thick transi-
tion forest: Phrynobatrachus pedropede-

TABLE 1. ECOLOGICAL DISTRIBUTION OF THE HERPETOFAUNA OF CENTRAL AFRICA MONTANE FOREST.

Small shaded ponds

Xenopus sp.

Bufo kisoloensis

Phrynobatrachus
petropedetoides

Ph. bequaerti

Hyperolius castaneus

H. frontalis

H. chrysogaster

H. leucotaenius

Large swamps

Xenopus sp.

Bufo kisoloensis

Phrynobatrachus
versicolor

Ph. bequaerti

Rana angolensis
chapini

Hyperolius castaneus

Bushes

Leptopelis kivuensis
Afrixalus laevis
Hyperolius castaneus
Cnemaspis
quattuorseriatus
Chamaeleo rudis
Ch. adolfifriederici
Ch. johnstoni
Philothamnus
heterodermus
Dipsadoboa unicolor
Atheris nitschei

Streams

Rana angolensis

Rana ruwenzorica

Phrynobatrachus
acutirostris

Ph. asper

Hyperolius frontalis

H. discodactylus

Bamboos

Callixalus pictus
Afrixalus orophilus

Ground
(above and below)

Arthroleptis adolfifriederici

Cardioglossa cyaneospila

Schoutedenella
hematogaster

Lacerta jacksoni

Algyroides vauereselli

Leptosiaphos
blochmanni

L. meleagris

L. graueri

Typhlops angolensis

Grassy clearings

Phrynobatrachus
graueri
Mabuya megalura
Duberria lutrix
Psammophylax
variabilis

CENTRAL AFRICAN MONTANE FOREST - Laurent

toides Ahl, whose digital dilations allow it
to climb on leaves, the golden tree frog

_ Afrixalus laevis (Ahl), and the white-

snouted Hyperolius frontalis Laurent.
However, in the forest of Rwanda and
Burundi the second of these tree frogs is
rare, while the third is apparently absent.
Hyperolius discodactylus Ahl, a green
montane tree frog, which was once con-
sidered rare, is actually common
along streams but rarely encountered in
large swamps. Hyperolius leucotaenius
Laurent inhabits the same habitats but its
range is restricted to the southern
Itombwe highlands and northern Kabobo
range, while its northern vicariant H.
chrysogaster Laurent lives in the transi-
tion forest between the latitude of north-
erm Itombwe and Ruwenzori Mt.

Two other montane tree frogs have a
limited distribution. Afrixalus orophilus
(Laurent), described from the Kivu vol-
cano region, is common in Rwanda and
Burundi but is unknown in the Ruwenzori
and the southern Kivu. The species is
often found in rather open places, in con-
trast with A. laevis which likes shade, but
in Muhokole (southern Rugege forest) it is
common in the broken Bamboos.

The Itombwe Bambu tree frog, Callix-
alus pictus Laurent, has a very peculiar
range. West of the Graben it exists in the
Itombwe and Kabobo mountains but has
never been collected in the Kahusi range
nor in the north Kivu, despite careful
searching in the bamboo forests. East of
the Graben it has been found in the
Rugege forest and the Lutsiro highlands
south of Kisenyi, but is apparently absent
from the volcanoes. This distribution is
similar to the range of Gorilla beringei
(sensu Kalin, personal communication),
which is limited to the eastern side of a
hypothetical ancient Nile coming from
upper Katanga, which was captured by
the Zaire. Callixalus pictus is also limited
to the upper levels; rare at 2,300 m, it
becomes more and more frequent at
higher altitudes and is rather common at
2,600 to 2,700 m. An interesting ecolog-
ical shift has been observed in the Mt.

353

Kabobo region where its customary bio-
topes—the bamboos—is lacking. Here
they take refuge beneath the large layers
of moss which cover the trunks of big
trees.

Besides the green tree frogs like
Hyperolius discodactylus and H. fron-
talis, some ranids dwell in the streams:
giant Phrynobatrachus species like acu-
tirostris Nieden and P. asper Laurent, an
Itombwe endemic which is sympatric
with acutirostris around 2,500 m and the
Ranas. According to recent investigations
(Laurent, 1972), two sibling species exist
in the montane forest: a larger form with
more extensive webbing, which appears
to be Rana angolensis chapini Noble
(replacing the nominate form in the Great
Lakes region) and R. ruwenzorica
Laurent, a smaller form with reduced
webbing. At least one undescribed
species seems to live in the Itombwe
highlands, increasing further the already
high rate of endemism in this interesting
region.

In the Itombwe grasslands (see
Laurent, 1964), we noticed that Phry-
nobatrachus bequaerti, which belongs to
the group of small-sized species previ-
ously classified in the genus Arthroleptis
because of the poorly developed web-
bing, is more terrestrial than other
species. The same species behave simi-
larly in the montane forest, but there is
also a closely related species which is
even more terrestrial: P. graueri Nieden.
Like all the species of the genus, it is also
aquatic but, whereas P. bequaerti prefers
still waters, P. graueri is more often seen
in running waters, i.e., very small brooks.
The two species are morphologically
very much alike but differ in their colora-
tion: P. bequaerti has the thighs and the
rear part of the belly orange-colored,
while P. graueri has the belly and the
thighs pinkish.

Completely terrestrial frogs are also
found in the montane forest: Arthroleptis
adolfifriederici Nieden, the most com-
mon, exists everywhere, while the more
secretive and_ strangely blue-colored

354

Cardioglossa cyaneospila Laurent, is
mainly found near tree ferns at the lower
levels. Both have a largely disjunctive
distribution because the forest is inter-
rupted in many places, namely in south-
ern Kivu, Rwanda, and Burundi, but they
are also frequent in the transition forest
which is the main niche of the blue
Cardioglossa. The dwarfed arthro-
leptines of the genus Schoutedenella are
not known in the northern montane
forests except on the Ruwenzori where S.
schubotzi (Nieden) is not uncommon. In
this species, the remarkable sexual
dimorphism associated with the arthro-
leptines, the hypertrophy of the third
finger, is very striking, almost as much as
in S. globosa Witte, since it can be almost
as long as the body. In the western
Itombwe and the Kabobo massif is S.
hematogaster Laurent, remarkable for its
bright red belly, and in the southern
Itombwe and northern Kabobo range the
relatively big S. pyrrhoscelis (Laurent)
with a red network on the belly.

Reptiles are not diverse in these for-
ested highlands. The arboreal species
include the secretive gekko, Cnemaspis
quattuorseriatus (Sternfeld) and three
chamaeleons are present everywhere.
The robust Chamaeleo johnstoni Boulen-
ger is well known by its three-homed
males and their spectacular fights. The
most common species, however, is C.
rudis Boulenger, which Stanley Rand
(1963) has demonstrated to be a species
distinct from C. bitaeniatus ellioti
Gunther, living at lower levels but in the
savanna highlands. The strange long-
tailed C. adolfifriederici Sternfeld is a
rare species, but its disjunct range ex-
tends from Ituri to the Kahusi Mountains
and the Rugege forest. Two still stranger
leaf-nosed chamaeleons, C. xenorhinus
Boulenger and C. carpenteri Parker, are
Ruwenzori endemics; as they are very
rare, Loveridge (1942) supposes that they
are canopy dwellers.

Two lacertids very common through-
out the montane forests bordering the
western Graben are Lacerta jacksoni

Advances in Herpetology and Evolutionary Biology

Boulenger and Algyroides vauereselli
(Tornier).

The semifossorial skinks of the genus
Leptosiaphos! are also common and live
among and under forest floor debris as
well as under and within logs: they
are L. graueri (Stemfeld), L. meleagris
(Boulenger), L. hackarsi (Witte) and L.
blochmanni (Tornier), which have re-
spectively five fingers and five toes, four
fingers and four toes (two species), and
three fingers and three toes.

Another skink characteristic of the
highland savannas, Mabuya megalura
(Peters), lives in the Rugege forest
(Rwanda and Burundi) in the clearings
and Hagenia patches which are not dark
and where an often grassy soil provides it
with its customary niche.

Snakes are infrequent in the montane
forests. As always in the forest regions,
Typhlops angolensis is represented by
light-bellied populations: T. angolensis
polylepis Laurent, in the Itombwe slopes
and T. angolenis irsaci Laurent, near the
Kahusi and the Rugege forest. The only
colubrids still common at this level are
Philothamnus heterodermus ruandae
Loveridge, an active frog eater, and
Dasypeltis ater Sternfeld. Psammo-
phylax variabilis Gunther is ecologically
a highland savanna form like Mabuya
megalura, but it has been collected in
clearings within the montane forest. The
beautiful green tree-viper, Atheris n.
nitschei Tomier is very common in the
montane forest. As is generally the case
with venomous snakes, its diet is not
specialized and comprises frogs as well
as lizards and rodents.

DISCUSSION

Following this survey, the main ques-
tion which arises is whether there is any
proof or any hint of a differentiation
between the montane forest populations

!This genus has been recently merged into
Panaspis.

CENTRAL AFRICAN MONTANE FOREST - Laurent

isolated by distributional gaps created by
a strip of highland savanna or of lowland

- savanna or forest.

An exhaustive analysis of the pertinent
material has at present only been made
for snakes (Laurent 1956b), but we also
have some indications concerning the
tree frogs and the chamaeleons.

Let us consider first the case of the
snake Dipsadoboa unicolor viridiventris
Laurent, which is more an inhabitant of
the transition forest than of the montane
forest, although three specimens have
been collected at 2,100 m in the Virunga
National Park. There is an apparently
homogeneous group of populations liv-
ing on the Ruwenzori slopes, the forests
west of Lake Edward and around Lake
Kivu, on the eastem as well as the west-
ern side. Four specimens are quite out-
side of the range of variation of this group
and might belong to taxonomically recog-
nizable populations (Fig. 2).! Apparently
one such distinct population occurs on
the Blue Mountains of the Ituri, which
have been isolated from other montane
forests for a very long time; another is
on the Kabobo range, west of Lake
Tanganyika and south of the Itombwe,
and still another in the Bururi forest
(Burundi), which is severed from the
Rugege forest by a large deforested area.
The fourth specimen comes from the
Rwindi plain, which is an unexpected
environment for the species.

The viper Atheris nitschei nitschei
(Tornier) also demonstrates a regional
variation, as shown on the diagram (Fig.
3). It can be seen that two samples are
especially differentiated: the one on
Idjwi Island which is the most complete-
ly isolated montane forest of all, and the
East Itombwe sample. The eastern
slopes of Itombwe are very steep and the
inhabited zone consequently narrowed;
moreover, the forest appears to be dis-
continuous, so that some populations can

1They have not been named because these are
single specimens.

355

at the same time be small and isolated,
which suggests a Sewall Wright effect
(Wright, 1931). Other striking differences
appear between Kahusi and Itombwe
ranges as well as between the Itombwe
range and Upper Lubitshako. In each
case, although the two regions are not far
away from each other, a ow and uninhab-
ited zone intervenes and the gene flow is
consequently interrupted. Such a con-
trast has not been seen for the Bururi
sample, which can be explained by the
fact that Atheris nitschei is well able to
withstand the destruction of the forest
provided there remain some patches of
forestlike undergrowth not too distant
from each other. It appears that the
Bururi forest is still or has recently been
connected to the Rugege forest by a chain
of uncultivated areas where Lobelias and
other montane forest plants form an envi-
ronment where Atheris nitschei can live.
If, in the same conditions, Dipsadoboa
unicolor underwent more differentiation,
it is probably because this species is
more decidedly a forest dweller and lives
at lower altitudes where cultivation has
more completely destroyed the suitable
habitats. Thus it appears that greater
degrees of isolation can be created by
human factors at low altitudes than by
solely climatic changes at higher alti-
tudes. The apparent reason for this is that
the human population is denser and
therefore more destructive at moderate
heights than on mountain tops.
Chamaeleo rudis schoutedeni Laurent,

Cardioglossa cyaneospila __inornata
Laurent, Hyperolius castaneus sub-
marginatus Laurent, and Hyperolius

leucotaenitus allogynus Laurent demon-
strate that the populations of the Kabobo
range do show an appreciable degree of
differentiation. This is a feature which
was to be expected when one considers
the lowness of the Fizi “bridge,” which
connects the Kabobo range with the
southern Itombwe.

Hyperolius discodactylus appears
somewhat differentiated in the Bururi
forest. In the Ruwenzori range the

356 Advances in Herpetology and Evolutionary Biology

Range of Subcavdals

ae Om Beh os
Gini ae oe. oe
199° a mai
of Poy
cee : |
acm ao
us SS, aaa
nee Ses oss a
EMG = ee Sey oye a ee es ee ee perenne kee AG tie
Katanda Pe roo me noacnsuoDOe Sh °
N.Kivu ? 0
: 2 ge ee!
60 ae me Nilo ee i
Nene ! T Ubitcheko vane
ITURI 9° 'S-Kivu !
Range oF ventrals
= Bururi
BURUNDI

190 200 210 230

Figure 2. Correlation between the numbers of subcaudals (ordinate) and ventrals (abscissa) in Dipsadoboa
unicolor viridiventris. The specimens from Katanda (Rwindi region), Nioka (Ituri), Lubitshako (southern Kivu), and
Bururi (Burundi) are outside of the range of variation of the subspecies.

357

CENTRAL AFRICAN MONTANE FOREST - Laurent

"yso10} uNung E}) +}Sa10}
ebebny @ ‘abuei ogogey @) ‘uoibes oyeysyigny 6) ‘@MQWO}| yinos (8) ‘@MQUWO}| se5Z) ‘@MqQUIO}| YON Q) ‘a6ues isnyey G) ‘pue|s} IMIPI (¥)
‘uoiba1 SAaOUBdD|OA eBunii, (€) ‘uoiba1 saye7 OJOHOW () ‘uoibe1 o19qn7 (1) SaJ0] aUeJUOW ay} JO Sued JUa/ayIp 0] PUuOdsauO0D sainBy pa/osiD

‘aul| peyop e Aq peyoeuuOo aie SUBS ay] “siSeujUaJed
useMmjeg ae ajdwes yoeA UI SUBWIDAdS Jo SYequUNU ay, eYOS}U “U SUAYIY Ul Sjepneoqns puke sje4jUBA JO SJaquuNU Bu} JO UOHWEWeA jed1ydesBoey “¢ aunbi4

SS a (9) @) = ar a (2) @
ase (S) @) ee (s) G)
ee ae ©) @) a ee 1) O
= ae Ne eel (9) ©) oS aa ae (9) G)
a (t) (8) SS ceo () 4%
ees (i) @ abe gaara Ore
eG BQ) =e SSS Se eee SS OLS
re eae (8) ©) > <= e n (4) (Q)
See soca aaa s (9) ® Sa ae Se (9) &
=e @) © SSS (ic) (€)
Soe (9) S a (1) 2

(3) > ae ans (s)

bo 46

ae (2) isa DE
Tease! (Q > a ee w) ()
SS (Z) ©) > (1) @
tease a = >) (6) ——$<——<—$<$ (01) (6)
ee ay p (ce) ©) Sete (Cc)
Se a (ht) © S22 == ee Oe
SSS SS (4) (9) ea ee a) (3)
7 (8) © [= a a ae (8) G)
BES Share cer (+) &) en ()@)
2 Set eee COG =3 me

See (6) (@) ‘ (al
(y) DAP ‘ OO

09 Os Ob S¢ OZ} 09} OS} Ob
S|PPAVIqGNS S{D4 jue

square-snouted Hyperolius xenorhinus
has recéntly been described (Laurent,
1972). Both are extreme areas where such
differentiation effects or even endemism
(Ruwenzori) were to be expected. Noth-
ing conspicuous has so far been demon-
strated elsewhere, but a detailed analysis
is likely to prove some incipient diverg-
ences wherever the montane forest is
discontinuous.

ACKNOWLEDGMENTS

It is a pleasure to dedicate this small
essay to Ernest E. Williams, who several
years ago revised the text and provided
some valuable suggestions to the author.
The figures have been executed by Otilia
Brizuela.

LITERATURE CITED

BARBOUR, T., AND A. LOVERIDGE. 1930. Reptiles and
Amphibians from the Central African Lakes
Region, pp. 764-785. In Strong’s African
Republic of Liberia and the Belgian Congo.
Cambridge, Mass.

Gans, C. 1964. Further comments on the forms of
the African snake genus Dasypeltis (Reptilia:
Serpentes). Rev. Zool. Bot. Afr., 69: 279-295.

LAURENT, R. F. 1950. Diagnoses préliminaires de

treize Batraciens nouveaux d'Afrique Centrale.

Rev. Zool. Bot. Afr., 44: 210-212.

1951. Deux Reptiles et onze Batraciens
nouveaux d Afrique Centrale. Rev. Zool. Bot.
Afr., 44: 360-381.

__. 1952a. Reptiles et Batraciens nouveaux du
massif du mont Kabobo et du plateau des
Marungu. Rev. Zool. Bot. Afr., 46: 18-34.

___. 1952b. Quelques données nouvelles sur la
Systématique et l’Ecologie du genre Hyperol-
ius. Rapp. Ann. Soc. roy. Zool. Belg., 82: 324—
334.

___.. 1952c. Reptiles et Batraciens nouveaux de la
région des grands lacs africains. Rev. Zool. Bot.
Afr., 46: 270-279.

___.. 1954a. Remarques sur le genre Schoutedenel-

la Witte. Ann. Mus. Congo. Nov. ser., in 4°, 1,

Miscellanea Zoologica H. Schouteden: 3440.

1954b. Apercu de la Biogéographie des

Batraciens et des Reptiles de la région des

Grands Lacs. Bull. Soc. Zool. France, 79: 290-

310.

Advances in Herpetology and Evolutionary Biology

___. 1956a. Esquisse d’une faune herpétologique
du Ruanda-Urundi. Les Naturalistes Belges,
37: 280-287.

___. 1956b. Contribution a l’Herpétologie de la

Région des Grands Lacs de |’ Afrique Centrale.

I. Généralités II. Chéloniens III. Ophidiens.

Ann. Mus. Roy. Congo Belge, sér. in 8°, Sci.

Zool., 48: 1-190, pls. I-XXI.

1960a. Notes complémentaires sur les

Chéloniens et les Ophidiens du Congo

Oriental. Ann. Mus. Roy. Congo Belge, sér. in

8°, Sci, Zool., 94: 1-88.

____. 1960b. Les Serpents du Kivu. Les Naturalistes
Belges, 41: 437-452, figs. 1-4.

___. 1964. Adaptive modifications in frogs of an
isolated highland fauna in central Africa.
Evolution, 18: 458—467.

___. 1971. Caracteres biomeétriques de trois genres
monotypiques de la famille des Hyperoliidae.
Bull. Séances Acad. Sc. Outremer. Bruxelles.
1971-2: 256-275, 18 figs.

___. 1972. Amphibiens. Explor. Parc Nat. Virunga,
(2)22: 1-124, figs. 1-29, cartes 1-3, pls. I-XI.

____. 1973. Le genre Leptopelis Gunther (Salientia)

au Zaire. Ann. Mus. roy. Afr. Centr. sér. in 8°,

Zool., 202: 1-62.

. in press. Le genre Afrixalus en Afrique Cen-

trale. Ann. Mus. roy. Afr. Centr., sér. in 8°,

Zool.

LOVERIDGE, A. 1942. Scientific results of a fourth
expedition to forested areas in East and
Central Africa IV. Reptiles. Bull. Mus. Comp.
Zool., 91: 237-373, pls. I-VI.

Moreau, R. E. 1966. The Bird Faunas of Africa and
its islands. New York, Academic Press, 424 pp.

NIEDEN F. 1912. Amphibia. Wiss. Ergebn. Deutsch.
Zentral Afrika Exped., 4: 165-195, 1 taf.

RAND, A. S. 1963. Notes on the Chamaeleo bitaenia-

tus complex. Bull. Mus. Comp. Zool., 130: 3-29.

Roux-EstTEvE, R. 1974. Revision systématique des
Typhlopidae d’Afrique. (Reptilia: Serpentes).
Mem. Mus. Nat. Hist. Nat., (n.s.), A. Zool., 87:
1-313.

STERNFELD, R. 1912. Reptilia. Wiss. Ergebn.
Deutsch. Zentral Afrika Exped., 4: 197-279,4
Mate

TINSLEY, R. C., H. R. KOBEL, AND M. FISCHBERG.
1979. The biology and systematics of a new
species of Xenopus (Anura: Pipidae) from the
highlands of Central Africa. J. Zool. London,
188: 69-102.

WITTE, H. G. DE. 1933. Batraciens et Reptiles re-
cueillis par M. L. Burgeon au Ruwenzori, au
Kivu et au Tanganyika. Rev. Zool. Bot. Afr., 24:
97-103.

____. 1941. Batraciens et Reptiles. Explor. Pare Nat.
Albert, 33: I-XVII, 1-621, figs. 1-54, pls. I-
LXXVI.

WriGurT, S. 1931. Evolution in Mendelian popula-
tions. Genetics, 16: 97-159.

Zoogeography of the Skinks (Sauria: Scincidae)
Of Arno Atoll, Marshall Islands

A. ROSS KIESTER!

ABSTRACT. The distribution of four species of
scincid lizards (Emoia cyanura, E. arnoensis, E.
boettgeri, and Lamprolepis smaragdinum) on Arno
Atoll, Marshall Islands, reveals the following pat-
terns: 1) very small islands are inhabited only by E.
cyanura, the smallest species; 2) somewhat larger
and vegetationally more complex islands usually
have E. cyanura and Lamprolepis, the largest spe-
cies; 3) as reported by Brown and Marshall (1953),
most larger islands have these two plus either E.
arnoensis or E. boettgeri, which are of similar size
and intermediate between the first two; 4) two
islands were found to have all four species, but
these were not the largest or most complex islands.
These pattems are discussed in terms of Williams’s
(1969) generalizations about the ecology of coloni-
zation in the Anolis of the West Indies.

INTRODUCTION

With the publication of the work of
MacArthur and Wilson (1963, 1967), a
new methodology, the equilibrium an-
alysis of colonization and _ extinction
rates, revitalized one of the central topics
of evolutionary biology. As always, issues
involving methodology have correspond-
ing theoretical issues. In biogeography
the key issue is often seen as the relative
importance of the role of history as op-
posed to any of those processes which
may produce repeated patterns in com-
munity structure. This is a general prob-
lem in evolutionary ecology (an introduc-
tion to it is given in Kiester, 1980). In
island biogeography the contrast is nicely
exemplified by two books which take

1 Department of Biology, Tulane University, New
Orleans, Louisiana 70118, U.S.A.

opposite viewpoints for the most part.
MacArthur (1972) attempts to develop a
generally nonhistorical approach to bio-
geography, while Lack (1976) argues that
the role of history is always paramount.

Some case studies in island bioge-
ography may help to frame the outlines of
this problem. In this paper I discuss the
pattern of distribution of four species of
skinks which occur on Arno Atoll in the
Marshall Islands of Micronesia in the
hope that analysis of these patterns may
contribute to a further understanding of
the problem of the role of history.

BACKGROUND AND RESULTS

Arno Atoll in the Marshall Islands,
Micronesia, is in most respects a typical
coral atoll. Located at about 7° 7’ N and
171° 41’ E, it consists of 133 islands of
various sizes comprising a total of about
five sq. miles of dry land. Many of the
islets are no larger than emergent coral
boulders while some are several miles
long, and, in the case of Amo Island
nearly one half mile wide. Arno has been
described by Agassiz (1903) and by
several articles in the Atoll Research
Bulletin (Marshall, 1951; Stone, 1951;
Wells, 1951; and Mason,’ 1952).

Six species of skinks occur on Arno
Atoll. Two of them will not be considered
further here for lack of data. One of these,
Lipinia noctua, is nocturnal and the
other, Eugongylus albofasciolatus, is
cryptic, living under piles of debris and

360

in houses. During my visit I saw no
Lipinia and only one Eugongylus.

The other four species of skinks found
on Arno are diurnal and are Emoia
cyanura, E. arnoensis, E. boettgeri
orientalis, and Lamprolepis smarag-
dinum (placed in the genus Dasia until
recently). These four species form three
distinct size classes with E. cyanura the
smallest (40 mm _ snout-vent length
[SVL]), E. arnoensis and boettgeri simi-
lar and of intermediate size (70-80 mm
SVL), and Lamprolepis the largest (110
mm SVL).

As a result of a visit made to Arno in
1950 by J. T. Marshall and I. LaRivers,
Brown and Marshall (1953) detailed an
unusual distributional pattern for these
four species. They found E. cyanura
present on all islands visited. Lam-
prolepis and the two larger Emoia were
only found on those islands large enough
to support a substantial number of coco-
nut palms. But of the 12 islands which
contained either E. arnoensis or E.
boettgeri, none had both species. This
remarkable lack of overlap formed no
recognizable geographical pattern within
the atoll. Further, islets which were
inhabited by different species were
sometimes only several hundred meters
apart. Brown and Marshall (1953) con-
cluded that competition was the probable
mechanism which kept these two sibling
species allopatric.

In the summer of 1968 I visited Arno
for the express purpose of obtaining more
distributional data about the skinks, es-
pecially the two allopatric species. I cen-
sused skink species on 33 islands (about
25% of all of the islands) including 11 of
the islands reported on by Brown and
Marshall. Each island was searched for at
least one rain-free hour. Because of the
activity and conspicuousness of these
diurnal species, it was usually possible to
determine which species were present in
a short time. Specimens were collected
on seven of these islands and placed in
the Museum of Comparative Zoology,
Harvard University.

Advances in Herpetology and Evolutionary Biology

Figure 1 shows the distribution of E.
arnoensis and E. boettgeri throughout
the atoll. Table 1 gives the summary data
for the four common diurnal species. Of
23 islands which had either E. arnoensis
or E. boettgeri, 13 had E. arnoensis, 8 had
E. boettgeri, and 2 had both species.
Thus the basic pattern found by Brown
and Marshall still holds, but the allopatry
is not perfect; there are two exceptions.
All of the islands which had been previ-
ously investigated were found to have
the same species as before. Thus the
pattern was stable for at least 18 years in
these cases.

Table 1 gives the distribution of the
four species over the 33 islands and the
area in hectares of each island. Areas
were estimated from the U.S. Navy Chart
H.O. 6004 and are approximate. Overall,
larger islands have more species. The
Spearman rank correlation coefficient for
area against number of species is 0.54,
and the probability that this correlation
does not differ from zero is p < 0.002.

Consideration of islands possessing
one, two, or three species as detailed in
Table 1 shows another pattern: islands
which have only one species have E.
cyanura, the smallest species. Four of the
islands which have two species have E.
cyanura and Lamprolepis (the smallest
and the largest species) while a single
two-species island had E. cyanura and E.
arnoensis. There are then 20 islands with
three species and two islands with four
species. This pattern (with the one ex-
ception) of the smallest species in the
simplest communities followed by the
largest and then the medium-sized spe-
cies in the successively more complex
communities is reminiscent of some of
the “assembly rules” of Diamond (1975).

Casual observations on the ecology of
the four species revealed that E. cyanura
was commonly active on the ground or on
piles of dead vegetation. Lamprolepis
was usually observed on the trunks of

coconut palms, but would move onto the

ground to forage. The other species of
Emoia were also usually associated with

ARNO ATOLL SKINK ZOOGEOGRAPHY : Kiester

361

Figure 1.

coconuts, but seemed to spend somewhat
more time on the ground.

Two islands, Lojor Ngak and Bokale-
Laj, which had only E. cyanura, were
devoid of coconut palms. Their dominant
vegetation consisted of two shrub spe-
cies, Scaevola and Messerschmidtia.
These two genera are usually the earliest
shrubs to invade a bare island (Stone,
1951).

Outside of Arno Atoll these four spe-
cies are widely distributed throughout
the Pacific (Brown, 1956). E. cyanura is
found throughout the Caroline and
Marshall Islands in Micronesia, and in
much of Melanesia and almost all of

Distribution of Emoia arnoensis (A) and E. boettgeri (B) on Arno Atoll, Marshall Islands.

Polynesia, even reaching Clipperton
Island in the western Pacific. Lampro-
lepis smaragdinum is found over all of
Micronesia and Melanesia, and also New
Guinea, Indonesia, and the Philippines.
E. boettgeri is apparently found through-
out the Marshall Islands and the eastern
Caroline Islands. The distribution of E.
arnoensis is less well known. Specimens
in the United States National Museum
are listed from Kusaie and one specimen
in the California Academy of Science is
reported from Truk. On either of these
islands arnoensis would be sympatric
with boettgeri. I have seen boettgeri
common on several of the larger islands

Q

362 Advances in Herpetology and Evolutionary Biology

TABLE |. DISTRIBUTION OF THE FOUR SPECIES OF COMMON DIURNAL SKINKS ON THE ISLANDS OF ARNO ATOLL,

MARSHALL ISLANDS.

Islands are listed in counterclockwise order from the northwest horn of Amo Atoll. Place names are from
Mason (1952): those given in brackets are the names on U.S. Navy Chart H.O. 6004.* = an island also
censused by Brown and Marshall (1953).

ISLAND

NAMWI [NAMI]

LOJOR NGAK

MWET-DRIK [METORIKKU]

BIKAREIJ [PIKAAREF JI]

BIKONELE

LIBEIJ-EN’

BOKALE-LAJ

RAKIJEDR

BAR-BAR-EN’

ENEAITOK

BOKANBIT

KIDRENEONG

BOKANKORA

JAR-KWIL [CHARUKURU]

ENELIK, ENEAR [Now fused
into one]

TUTU [DODO]

KIJ-BWE [KEJIBOI]

JIL’ANG [CHIRAN]

ENIDRIK [ENIRIKKU]

TAK-LIB [TAGELIB]

TAK-LIB-EJ [EAST
TAGELIB]

L’ANGAR [RAKAARU]

KILANGE [KIRAGE]

MALEL [MARERI]

4th from East in L’OM’ARIN
MALEL group

IJOEN’ [IJOEN]

ALJALTUEN’ MATOL-EN’
[CHITTAKINMATOROEN]

AOT-LE [AUTORE]

INE [INE]

ARNO [ARNO]

WWRrWNNNEWNWwWHw | Number of Species

WWWWrE WW ORWARWE

Qe

Www

Previously Censused

Te FP ORR

of the Truk group and on Ponape, but saw

no arnoensis at either

locality. E.

arnoensis also apparently does not occur
on Majuro Atoll in the Marshalls, the

Area (hectares)
Emoia boettgeri

>> > |Emoia arnoensis

> >
fe et eet et tet || canny arroliens

_
Co
(<e)
(ee)
AAA AS GOGOGAG GBAGGOG GOAOAARODAAAOS |amae aren
> SP >> Pe >
locflosslociocles) lesiles)
ists! fle eis

Dm wh
Fea EST i

closest island to Arno. I did not see it dur-
ing my visit of three days. It is known
from other atolls in the Marshalls, how-
ever.

)

ARNO ATOLL SKINK ZOOGEOGRAPHY «- Kiester

DISCUSSION

Williams (1969) presented a list of ten
preliminary generalizations on _ the
ecology of colonization of small islands
by lizards of the genus Anolis. Several of
these claims are open to both historical
and equilibrium interpretations and
highlight the contrast between the dif-
ferent views of the role of history men-
tioned above. Although the data are not
extensive, the case of the skinks of Amo
Atoll may add to an understanding of the
generality of some of the claims.

With regard to island size, Williams
(1969) claims that “there appears to be
some critical island size below which a
colonizing species may exclude every
other congener. On tiny islands E.
cyanura does occur alone, but whether it
actively excludes the other Emoia spe-
cies and Lamprolepis is not known. E.
cydanura may usually succeed in coloniz-
ing small islands first and then be able to
prevent the successful colonization of
other species. An alternative possibility
is that other species are not found on the
smallest islets because of a lack of suit-
able vegetation for perching and perhaps
other ecological factors such as food
supply and predation. On this view, his-
tory plays a dominant role in the sense
that there may not have been enough
time for suitable vegetation to appear.
Were this the case, a species other than
E. cyanura arriving at such an island
would not succeed in colonizing whether
E. cyanura was there or not.

However, colonization may proceed to
the point where Williams (1969) claims
that “even on small islands the phe-
nomena of exclusion and coexistence
appear to be relatively stable, once
achieved.” Data from Arno Atoll indicate
stability over a period of at least 18 years
for those islands which have either E.
arnoensis or E. boettgeri, but not both.
The two islands which possessed both
species were not larger or more complex
than those islands which had only one of
the two. I do not know whether this co-

363

existence represents an equilibrium for
these islands. Comparison with the other
islands indicates that it is not. This
generalization would predict that once
one species excluded the other, then
stability would occur.

Direct data on the colonization and ex-
tinction rates for the lizards of Micronesia
is needed to definitely assess the validity
of these two claims. This data has not
been collected, but some aspects of the
natural history of the area suggest that
these processes may be both episodic
and time dependent between episodes.
The destructive typhoons which strike
most atolls an average of two or three
times a century (Wiens, 1962; see also
Brown and Marshall, 1953, for discussion
of the case of Arno Atoll) often remove
almost all plant and animal life from an
atoll island. Obviously this creates very
episodic extinction. It may also tend to
produce episodic colonization since
typhoons may be the primary mechanism
by which individual lizards are removed
from islands to become rafters and pos-
sibly colonists. This pattern of extinction,
far from contributing to the creation of an
equilibrium species number, moves an
island away from any equilibrium.
Successful colonization by skinks cannot
occur until the islands have been revege-
tated to some degree. The evidence
presented here indicates that E. cyanura
can move onto islands which have only
the earliest plant colonizers (the shrubs
Scaevola and Messerschmidtia) whereas
the other species may not be able to in-
vade until the coconuts retum. Thus the
ontogeny of the vegetation of an island
affects which species may invade at a
given time. The question is: is there
enough other colonization and extinction
to create an equilibrium between epi-
sodes of typhoon-induced extinction and
colonization? If the exclusion of E.
arnoensis and E. boettgeri is due to some
form of competition, then the possibility
of an equilibrium can be considered.

Williams (1969) made two generaliza-
tions about coexistence which are rele-

364

vant to the zoogeography of these Pacific
skinks. First, “Coexistence of colonists is
possible if the colonizing species have
been preadapted in sympatry on more
complex source islands...” Islands on
which coexistence occurs are not neces-
sarily the source areas for islands with
fewer species. The two small islands
within Arno Atoll are not necessarily
source islands even for the rest of the
atoll, just as large islands such as Kusaie
and Ponape are not necessarily source
islands for the rest of the Marshall and
Caroline Islands. For Micronesian
skinks, it is not at all clear that there is
any source area in the conventional
sense. These skinks form a group which
is more or less endemic to the Pacific
islands (with the exception of Lampro-
lepis), and it is likely that almost all
colonization events took place from one
island to another rather than from an out-
side source area to the islands.

Second, “Coexistence may be evolved
in situ in the special circumstances of
complex geography...” This is the
likely pattern for the coexistence (if the
reports are accurate) of E. arnoensis and
E. boettgeri on Kusaie and Truk which
are large, complex volcanic islands.
These islands certainly have enough
ecological diversity to support a much
larger lizard fauna than Amo Atoll.

In summary, Williams (1969) observes
that “exclusion is an extremely common
phenomenon but one of low visibility.
Coexistence is a less frequent but highly
visible phenomenon.” Only in special
circumstances where the kind of geo-
graphical arrangement found in an atoll
which essentially constitutes multiple
replications of a natural experiment can
one expect to see the low visibility
phenomenon of exclusion highlighted so
well. It is the combination of the replica-
tion in space and continued censuses
through time of the lizards of Arno Atoll
and other islands of Micronesia which
will enable us to determine the relative
importance of equilibrium and historical
factors in island lizard zoogeography.

Advances in Herpetology and Evolutionary Biology

ACKNOWLEDGMENTS

I wish to thank the Evolutionary
Biology Committee of Harvard Univer-
sity, the Society of Sigma Xi, and Ermest
E.. Williams for support of the field work.
I am grateful to A. E. Greer, E. E. Wil-
liams, and the late T. P. Webster for dis-
cussion on the zoogeography of Pacific
lizards. I thank R. A. Lillehei for com-
ments on the manuscript.

LITERATURE CITED

AGassiz, L. 1903. Reports of the scientific results of
the expedition to the tropical Pacific... IV,
The coral reefs of the tropical Pacific. Mem.
Mus. Comp. Zool., 28, xxxiii + 410.

Brown, W. 1956. The distribution of terrestrial rep-

tiles in the islands of the Pacific basin. Proc.

Eighth Pac. Sci. Cong., 3A: 1479-1491.

BROWN, W., AND J. T. MARSHALL, JR. 1953. New
scincoid lizards from the Marshall Islands,
with notes on their distribution. Copeia, 1953:
201-207.

DIAMOND, J. 1975. Assembly of species communi-
ties, pp. 342-444. In M. L. Cody and J.
Diamond (eds.), Ecology and the Evolution of
Communities. Cambridge, Harvard Univ.
Press.

KiesTeER, A. R. 1980. Natural kinds, natural history
and ecology. Synthese, 43: 331-342.

Lack, D. 1976. Island Biology. Berkeley, Univ. of
Calif. Press, xvi + 445 pp.

MACARTHUR, R. 1972. Geographical Ecology. New
York, Harper & Row, Publ., xviii + 269 pp.

MACARTHUR, R. AND E. O. WILSON. 1963. An equi-
librium theory of insular zoogeography. Evolu-
tion, 17: 373-387.

AND) . 1967. The Theory of Island Bio-
geography. Princeton, N.J., Princeton Univ.
Press, 203 pp.

MARSHALL, J. T., JR. 1951. Vertebrate ecology of
Amo Atoll, Marshall Islands. Atoll Res. Bull.
No. 3, pp. 1-38.

Mason, L. 1952. Anthropology-geography study of
Arno Atoll, Marshall Islands, Atoll Res. Bull.
No. 10, pp. 1-21.

STONE, E. L., Jr. 1951. The soils, of Arno Atoll,
Marshall Islands. Atoll Res. Bull. No. 5, pp.
1-56.

WELLS, J. W. 1951. The coral reefs of Amo Atoll,
Marshall Islands. Atoll Res. Bull. No. 9, pp.
1-14.

WIENS, H. J. 1962. Atoll Environment and Ecology.

New Haven, Yale Univ. Press, xxii + 532 pp.

WILLIAMS, E. E. 1969. The ecology of colonization
as seen in the zoogeography of anoline lizards
on small islands. Quart. Rev. Biol., 44: 345—
389.

The Fossil Record and

Early Distribution of Lizards

RICHARD ESTES!

ABSTRACT. The fossil record of lizards is scanty,
but some hypotheses about the early evolution and
distribution of the group can be made. Four phases
in lizard history may be identified. The first or eo-
lacertilian phase is essentially Triassic, with the
group progressively declining during the later
Mesozoic. The second phase, during the Jurassic,
involved the initial radiation of true lizards, or
lacertilians, which probably underwent a primary
dichotomy in the latter part of the Early Jurassic
when Laurasia and Gondwana began to separate.
This vicariance produced a Gondwanan iguanian
group and a Laurasian group ancestral to all other
lizards; each major group further diversified, both
through plate separation and the influence of
epicontinental seas or other intracontinental bar-
riers. Some primitive living families may have
evolved during this phase although as yet no fossil
record documents their presence. In the third, or
Cretaceous phase, the remaining Holocene lacer-
tilian families appeared. Plate separation continued
to be important in lizard diversification, and was
instrumental in partitioning iguanids and agamids
into subgroups, perhaps also formed gekkonid and
scincid subgroups, and caused the agamid-
chamaeleonid dichotomy. During the fourth or
Cenozoic phase, expansion of family ranges took
place during Paleogene peak tropicality. Cooler and
drier climates in the Neogene caused restriction of
lizard groups to their present distributions.

An evaluation of Renous’s (1979) study on lizard
biogeography shows that her cladogram, based
essentially on limb nerve character states, is dis-
cordant with those based on more diverse data.
Owen’s (1976) continental reconstruction provides
a better background for biogeographic movements
of agamids, gekkonids, scincids, and varanids than
does the conventional arrangement that includes a
widely open eastern Tethys Sea and separation of
southeast Asia from Gondwana.

1Department of Zoology, San Diego State Uni-
versity, San Diego, Califormia 92182, U.S.A.

INTRODUCTION

Lizards, as a group, have always pro-
vided outstanding opportunities for many
kinds of studies. Many anatomical, physi-
ological, ecological, and population
studies utilizing these animals have been
significant in formulation of new ideas
regarding structure and function in rep-
tiles. Lizard taxonomy has always been a
popular field, since the diversity of liz-
ards and the extensive modifications of
their skeleton and external surface offer a
natural laboratory for development of
hypotheses of interrelationships and
evolution of taxa, as well as providing
data for use in biogeographical analysis.

One of the most outstanding examples
of such integrated studies in lizards is to
be found in the studies of Emest E. Wil-
liams and his colleagues on the genus
Anolis. Although I have not participated
in this project directly myself, Professor
Williams facilitated my own work on the
lizard fossil record through his vivid in-
terest in anatomy and the history of rep-
tiles, and he has provided an important
source of stimulation, information, and
constructive criticism directly applicable
to my work. It is a pleasure to acknowl-
edge his help and to have the opportunity
to contribute to this volume.

In this paper I discuss some aspects of
my recent compilation of the lizard fossil
record that appears in the Handbuch der
Paldoherpetologie (Estes, 1982), a
summary that I have often informally and
somewhat ruefully called “The Tele-
phone Book of Fossil Lizards.” While

366

this work has provided both the tedious
aspects of compilation as well as the
stimulating advantages of a comprehen-
sive view, it is the latter that I hope to
provide here. In the Handbuch, each
described fossil species is discussed,
diagnoses are given for each genus and
species if possible, and a provisional list
of character states for each family and
other higher categories is provided. The
length of the Handbuch makes it desir-
able to give a summary of some aspects of
it that will help paleontologists and
herpetologists who are interested in what
is known of lizard evolution and the fossil
record, and who may then use the Hand-
buch itself for documentation.

The format I have adopted here is a
narrative one; there is no opportunity to
include more than a few of the more im-
portant data and bibliographic citations,
which are given more fully in the Hand-
buch. Eldredge (1979) has argued that a
scenario, to be classified as “good sci-
ence’, should be explicitly based on a
well-established genealogy, and Rosen
(1978) demonstrated that such a gene-
alogy is essential for biogeographic
studies. The cladogram of lizard relation-
ships offered here (Fig. 1) is based on
data given in the Handbuch, and for the
most part is in accord with that generally
accepted by herpetologists, at least as far
as the primitive-derived branching se-
quence Iguania—Gekkota—Scincomor-
pha—Anguimorpha is concerned. How-
ever, there are still many differences of
opinion regarding the branching se-
quence of families (or within families),
and while this paper demonstrates some
agreement between genealogy and vi-
cariant events, it fails in other areas be-
cause a robust cladogram is not yet avail-
able for lizards. Estes and de Queiroz (in
preparation) are providing this informa-
tion based on more than 100 characters
from both the skeletal system and soft
anatomy.

Advances in Herpetology and Evolutionary Biology

THE FOSSIL RECORD OF LIZARD
GROUPS

The most important features of the liz-
ard fossil record are summarized below,
preliminary to discussion of paleobio-
geography of the groups. Marine vara-
noids have been omitted, as have
Ophidia and Amphisbaenia. The latter
two are for me separate Orders; the
former may actually be convergent on

JURASSIC |CRETACEOUS| CENOZOIC

—

mM || SS |e m ie

© 3
iS 3

a

os)
S€i
(001)
so

Iguanidae
{Euposauridae
Agamidae

Chamaeleonidae

Ardeosauridae

{ Bavarisauridae

Gekkonidae

Pygopodidae

Teiidae

Gymnophthalmidae

Lacertidae

Scincidae

| +Paramacellodidae

Xantusiidae

Cordylidae

PUR | Xenosauridae

a arene ee ee ae {Dorsetisauridae

Anguidae

I---------- | +marine varanoids?
{+ Necrosauridae

Helodermatidae

Varanidae

al

Figure 1. Suggested interrelationships of lizards and
times of divergence. Thick black lines indicate extent of
the known fossil record of each group. Dotted lines

indicate extinction of the group. Cumulative m.y.b.p. —

given in column at left, which is in proportion to actual
time involved.

3)

varanoids (J. Gauthier, personal com-
munication 1982). The lizards as a whole

may be placed in a monophyletic Order

Sauria of the Superorder Squamata and
divided into two major groups, the prob-
ably paraphyletic Suborder Eolacertilia
(eolacertilians) and the monophyletic
Suborder Lacertilia (lacertilians or true
lizards).

EARLY SAURIANS

The Eolacertilia first emerge in the
Upper Permian and Lower Triassic of
South Africa. These include small insec-
tivorous diapsids that have been referred
to the family Paliguanidae by Carroll
(1975, 1977), including the genera Pali-
guana, Palaeagama, and Saurosternon.
Paliguana itself appears to be an authen-
tic saurian in having an external conch on
the quadrate; the other two generally
appear more primitive although they are
probably lepidosaurian. In Paliguana, the
first stages of temporal region modifica-
tions that are associated in living forms
with a streptostylic quadrate appear. In
the Upper Triassic, early saurians are
found only in England and North Amer-
ica; these later forms, the Kuehneo-
sauridae, are gliding animals (Kuehneo-
saurus, Kuehneosuchus, Icarosaurus)
although other lizards in the English
localities were apparently of more nor-
mal build (P. Robinson, personal com-
munication). The kuehneosaurs are
authentically saurian and had a more
progressive streptostylic temporal region
than the paliguanids (although the loose
quadrate-pterygoid contact necessary for
functional streptostyly was not yet pres-
ent). Another Upper Triassic (possibly
Lower Jurassic) form that may be related
to the eolacertilians is Fulengia (Fulen-
gidae) from southern Asia. It has leaf-
shaped, polycuspate teeth like those of
Iguana, and if actually a member of the
Sauria, suggests the development of

FOSSIL LIZARD DISTRIBUTION - Estes 367

herbivory in the squamates even at this
early date.

Evans (1980) has denied the saurian
relationship of eolacertilians, proposing
them as merely another group of eo-
suchians that lost the lower temporal bar,
paralleling lizards, some sphenodontids,
and prolacertilians. Parsimony suggests,
however, that this single character cannot
be used to deny a relationship, in view of
a number of other derived character
states that kuehneosaurs, and at least
Paliguana, share with squamates.

TRUE LIZARDS

The fossil record of the true lizards
begins in the Late Jurassic and is thus
separated from that of the Triassic eo-
lacertilians by more than 50 million
years, approximately the length of the
entire Cenozoic. Extrapolation across the
gap is thus not possible, and the lack of a
fossil record of the primitive precursors
of true lizards is unfortunate.

Iguania

The most primitive true lizards appear
to be the Iguanidae, a group whose fossil
record does not extend prior to the Late
Cretaceous, presumably owing to the
paucity of Gondwana localities. Pristi-
guana, from the late Cretaceous of Brazil,
seems more primitive than the most
primitive living iguanids (moruna-
saurines), although some similarities to
teiids were also suggested (Estes and
Price, 1973). The Late Paleocene of
Brazil has produced a number of interes-
ting undescribed fossil iguanids (Estes,
in progress), which include both primi-
tive and derived taxa, one of the latter
reminiscent of the tropidurines. This
suggests that all major extant iguanid
groups were diversified by the beginning
of the Cenozoic at least. North American
records include several Eocene genera,
among which Parasauromalus is a primi-

368

tive form with possible iguanine rela-
tionships, but there is no special similar-
ity to Sauromalus as the name implies.
Aciprion, from the Oligocene of North
America, is a primitive iguanid that
seems to be allied to the basiliscines. An
undescribed, related North American
Oligocene genus has a basilisklike oc-
ciptial projection on the skull, helping to
corroborate the basiliscine relationship
of the group. In Mexico the Paleogene
(Oligocene) Paradipsosaurus is another
misnomer, since this primitive moruna-
saurine-like iguanid is unrelated to the
living Dipsosaurus, although resembling
it somewhat in having an inflated and
shortened snout. Other Oligocene
iguanids are known from Canada and
represent sceloporines and_ possibly
tropidurines.

North American fossil iguanids from
Miocene through Pleistocene time for
the most part represent living genera and
occur generally in or near existing
ranges. An interesting exception is the
presence in the Miocene of Florida and
Nebraska of the living genus Lei-
ocephalus, now restricted to the West
Indies.

Only a single taxon suggests the pres-
ence of iguanids in Europe. Geiseltaliel-
lus, from the Eocene of Central Europe,
seems to have been a small agile iguanid
perhaps related to Aciprion, but the spe-
cimens need further study. Presence of
iguanid-like forms in Asia is indicated by
the Eocene Arretosauridae from Central
Asia. Arretosaurus was a large lizard with
a sculptured skull and mandibular char-
acter states like those of iguanids, but
again further study of these specimens is
necessary.

Turning to other iguanian families, the
extinct Euposauridae from the Upper
Jurassic of France seems to be iguanian
but its position is still in question. There
is a strong derived agamid similarity but
euposaurs share derived states lacking in
the most primitive living agamids.

Many agamid fossils occur within the

Advances in Herpetology and Evolutionary Biology

general limits of the family today: a
primitive Late Cretaceous agamid
Mimeosaurus was present in Central
Asia; Eocene records referred to the
primitive agamid Tinosaurus occur in
southern Asia; Agama (or Stellio) is
recorded from the Paleogene on into the
Holocene in Europe. Other fossils indi-
cate a wider Early Cenozoic dispersal of
some agamids; Tinosaurus has also been
recorded in the Eocene of North America
and Europe, while Uromastyx appears in
the Paleogene of Europe.

The Chamaeleonidae have a _ very
limited fossil record, but Chamaeleo is
present in Central Europe during the
Miocene. An unusual Miocene record of
Chamaeleo from East Africa is known
from a cast, in calcite, of the head region.

Gekkota

Some of the earliest true lizard fossils
appear to be gekkotans. These animals
are represented by complete skeletons
and have been placed in the families
Ardeosauridae and Bavarisauridae by
Hoffstetter (1964). Ardeosaurs occur in
the Upper Jurassic lithographic lime-
stones of Germany (Ardeosaurus, Eich-
staettisaurus) as well as in the Upper
Jurassic of eastern Asia (Yabeinosaurus).
Ardeosaurus is known from complete
specimens and is quite gekkonid in gen-
eral appearance. It and its close relative
Eichstaettisaurus (represented by an
exquisitely preserved skeleton) have
procoelous vertebrae and complete
supratemporal arches, the latter showing
a derived, “hockey-stick” squamosal that
lacks the dorsal process found in Iguania
and the majority of the Teiidae. The
bavarisaurs Bavarisaurus and _ Palae-
olacerta are less well preserved, but also
appear to be gekkotan; they have derived
amphicoelous vertebrae. Other gekkotan
features of these families appear in many ©
details of the skull, the fine, numerous
teeth, and the configuration of the trans-
verse processes of the first six or seven

caudal vertebrae. The two families ap-
pear to be early members of the Gekkota,
with the same heterogeneity of vertebral
type that occurs in living Gekkonidae.
The skull roof of Ardeosaurus shows,
however, a distinctive pattern of regular,
enlarged epidermal scutes. This pattern
may be primitive for lacertilians and lost
in iguanians and Gekkonidae, or could be
interpreted as a derived state of all liz-
ards above the iguanian level but lost in
gekkonids. Since pygopodids have such
enlarged scales, and they may appear in
some iguanids, as well as being present
in amphisbaenians and snakes, I believe
the first alternative is more parsimonious,
indicating that granular scales are de-
rived for both Iguanidae and Gekkoni-
dae. Robinson (1967) suggested that pres-
ence of an “hockey-stick” squamosal in
ardeosaurs and bavarisaurs is evidence of
scincomorphan rather than gekkotan rela-
tionships, but the few living gekkonids
that have retained remnants of both
squamosal and supratemporal show the
former to lie closely pressed against the
latter and the parietal and appearing to
be of “hockey-stick” type. Thus gekko-
tans may have lost the dorsal process
independently of other lizards, if they
arose before the scincomorphs as I sup-
pose, since the dorsal process is still
present in the vast majority of teiids. Al-
ternatively, if gekkotans are scincomorph
derivatives, their possession of an
“hockey-stick”” squamosal is to be ex-
pected.

Described fossils of the family Gek-
konidae are all of Cenozoic age. An un-
described gekkonid is known in the Late
Cretaceous of Central Asia. Fragmentary
New World records occur in the Upper
Paleocene of Brazil and the Upper Eo-
cene of California. Cenozoic records are
known from Europe (Eocene-Pliocene)
and Morocco (Miocene), but these fossils
are fragmentary and not distinctive.
Cenozoic gekkonids indicate that the
family was more or less cosmopolitan at
the beginning of the Cenozoic and that

FOSSIL LIZARD DISTRIBUTION - Estes 369

geckos occurred north of their present
latitudinal limits in Europe.

Scincomorpha

The primitive scincomorphs (Lacer-
toidea), have a rather good fossil record
that for the teiids goes back to the Late
Cretaceous of North America and Central
Asia, and Paleocene of South America;
lacertids first appear in the Paleocene of
Europe. Teiids are already represented
by both teiine (Meniscognathus, Pen-
eteius) and tupinambine (Chamops)
tribes in the Late Cretaceous of North
America. Polyglyphanodon is also pres-
ent there at that time, and is the most
highly derived member of the most
primitive subfamily of teiids, the Poly-
glyphanodontinae (see Estes, 1982). The
polyglyphanodontines are better repre-
sented in the Upper Cretaceous of Cen-
tral Asia, where they had a brief radiation
into a number of rather closely related
forms, becoming extinct at the end of the
Cretaceous. Rather large (Iguana-like) in
general, most of them had enlarged mo-
lariform or polycuspate-spatulate teeth,
indicating that both omnivorous and her-
bivorous adaptations had evolved. No
Cretaceous teiids are known in South
America as yet (few localities are known),
but they must have been there since the
Late Paleocene Itaborar deposits in
Brazil have produced well differentiated
members of both teiine and tupinambine
lines that were similar to modern genera
(Estes, in progress). Teiids disappear
from northern latitudes in North America
at the end of the Cretaceous with appear-
ance of Cnemidophorus there in the
Miocene. In South America, post-
Paleocene records of teiids are essen-
tially limited to the tupinambines Tupi-
nambis and Callopistes. No fossils are
yet known of the Gymnophthalmidae
(“microteiids’, see Estes, 1982).

The record of lacertids in the Late
Paleocene of Europe is of a genus closely
allied to Lacerta (Plesiolacerta), and in

370

the Middle Eocene Eolacerta appears;
the latter is a large primitive lacertid with
an open supratemporal fenestra. Other
records of lacertids are primarily Euro-
pean and referable to Lacerta itself or to
closely related forms. Lacertids may have
occurred in the Late Jurassic of Europe
(see below).

The skinks (Scincoidea) have a very
poor fossil record, somewhat surprising
in view of their present cosmopolitan dis-
tribution and great diversity. Some of the
Late Jurassic cordyloid lizards that I have
included in the family Paramacellodidae
(Estes, 1982, Saurillodon) show hints of
scincoid relationship and one apparently
had reduced limbs. There is, however, no
compelling evidence at present to in-
clude them in the Scincoidea. Two Late
Cretaceous genera from North America
(Sauriscus and Contogenys) have tenta-
tively been included in the Scincidae.
The former is possibly a cordyloid, but
Contogenys is more definitively scincoid
and has a Eumeces-like aspect, as well as
showing some derived dental and man-
dibular similarities to primitive xantu-
siids (Palaeoxantusia). In the North
American Oligocene, Eumeces itself ap-
pears. South America has produced an
undescribed Paleocene form resembling
Contogenys (Estes, in progress). No
skink fossils have been found in Europe
although a Miocene record of Eumeces
from Morocco is known. No fossil skinks
have been found in Asia.

The Cordyloidea as constituted here
include a primitive Late Jurassic family
Paramacellodidae and the living families
Cordylidae and Xantusiidae. The para-
macellodids occurred in England, Por-
tugal and North America and include
Paramacellodus and_ several related
genera (Hoffstetter, 1967; Estes, 1982).
These taxa show a number of cordyloid
features of the pelvis, caudal vertebrae
and mandible, but according to Hoff-
stetter (1967) Paramacellodus had a
primitive open supratemporal fenestra
rather than the closed one of the living
families. In my opinion, the supratempo-

Advances in Herpetology and Evolutionary Biology

ral region is not preserved in the speci-
mens of this genus and the fenestral
condition cannot be demonstrated.
Paramacellodus had, however, a derived
cordylidlike expanded coracoid region
and well developed compound osteo-
scutes of cordyloid type. It is possible
that Sauriscus, a Late Cretaceous form
mentioned above under the Scincoidea,
is in fact a cordyloid, but definitive
character states are lacking.

True cordylid fossils are few, and are
limited to the Late Eocene or Early
Oligocene of France (Pseudolacerta) and
the Miocene of Africa (Gerrhosaurus).
The Xantusiidae, contrary to what is
usually stated, has no demonstrable rela-

tionship with gekkotans (see Estes, 1982

for further discussion). There are a num-
ber of derived character states of the skull
and mandible that suggest cordyloid rela-
tionships, but a scincoid relationship is
not falsifiable at present. Xantusiids are
represented by fossils now referred to
Palaeoxantusia from the Eocene of Cali-
fornia, Wyoming, and Montana, and the
Oligocene of Canada. Some of these fos-
sils appear referable to Xantusia, and the
only known skull shows some similarity
to that of the Central American Lepid-
ophyma and perhaps was near the an-
cestry of the latter.

Anguimorpha

The fossil record of anguimorphs is the
best for any group of lizards, primarily
because of their extensive, old, Northern
Hemisphere distribution. The Xen-
osauridae has a North American fossil
record extending from the Late Cre-
taceous into the Oligocene. If the Holo-
cene Asian Shinisaurus is actually a xen-
osaur, the group originally had a wider
distribution. The Dorsetisauridae is an
Upper Jurassic family found in North
America and Europe. It seems more
derived than the xenosaurs in its pre-
daceous dentition and anguid-like brain-
case, while its skull roof scutellation is
derivable from the primitive scinco-

morph type. If Dorsetisaurus is authen-
tically anguimorphan it suggests a split
between the lines leading to Anguidae
and Xenosauridae prior to the uppermost
Jurassic. The Anguidae first appears in
the Late Cretaceous of North America, as
the common genus Odaxosaurus. This
genus appears to have been near the
ancestry of the glyptosaurs, which like
the polyglyphanodontine teiids had en-
larged teeth and were probably omnivo-
rous. The group had an extensive Early
Cenozoic North American radiation,
forming two tribes, the primitive
Melanosaurini (Melanosaurus, Xestops,
Peltosaurus) and more derived Glypto-
saurini (Glyptosaurus, Paraglypto-
saurus, Eoglyptosaurus, Heloder-
moides). The glyptosaurinid Placosaurus
occurs in the Eocene and Oligocene of
Europe; the melanosaurinid Xestops also
occurs there in the Eocene. Fragmentary
far northern records of both Odaxo-
saurus-like and glyptosaur-like taxa oc-
cur in the Eocene of the Canadian Arctic
Archipelago (Estes and Hutchison, 1980).
Anguids of modem aspect are also pres-
ent in the Late Cretaceous and Early
Cenozoic of North America, where prob-
able gerrhonotine fossils have been
found, and the Diploglossinae, now West
Indian, Mexican, Central and South
American, appears in the Early Eocene of
North America (Gauthier, 1982). The
Anguinae is best represented by fossils in
Europe, where Ophisaurus occurs from
the Eocene on and Anguis appears in the
Oligocene; Ophisaurus does not appear
in North America until the Oligocene.
The Anniellinae is now considered a sub-
family of Anguidae (Bellairs, 1970) and is
represented by the North American Eo-
cene Apodosauriscus and Miocene-
Holocene Anniella (Gauthier, 1980,
1982).

The Necrosauridae is an extinct family
of primitive varanoid lizards. This group
has a number of derived varanoid charac-
ter states (most completely known for the
Early Cenozoic European genus Necro-
saurus), while retaining a more primitive

FOossIL LIZARD DISTRIBUTION : Estes 371

skull roof. A number of undescribed
forms that may belong here occur in the
Late Cretaceous of central Asia (Borsuk-
Bialynicka, 1982). Eosaniwa, from the
Eocene of central Europe, may also be-
long to this group; it is a large lizard with
an elongated fish-catching snout unlike
that of other necrosaurids. North Ameri-
can necrosaurids include Parasaniwa
(Cretaceous-Eocene), and _— perhaps
Provaranosaurus (Paleocene) and
Colpodontosaurus (Late Cretaceous).
Necrosaurs all had a predaceous denti-
tion and so far as known lacked retraction
of the naris; they appear, however, to
have had at least a partially developed
intramandibular jaw hinge like that of
Heloderma.

The Helodermatidae may be present in
the North American Cretaceous (Para-
derma) and without doubt appears in
North America by the Oligocene and
Miocene (Heloderma); an helodermatid
has also been found in the Late Eocene
or Early Oligocene of France (Eur-
heloderma).

The Varanidae had a wide distribution
in the Late Cretaceous and Early
Cenozoic. Varanids are known from the
Late Cretaceous of Central Asia (Borsuk-
Bialynicka, 1982) and the North Ameri-
can varanoid Palaeosaniwa may be a
varanid; Varanus-like forms occur in the
Eocene of North America and Europe
(Saniwa). A definitive Early Eocene
record of varanids occurs in the Canadian
Arctic Archipelago (Estes and Hutchison,
1980). Saniwa extends into the Oligo-
cene of North America, and other vara-
nids persist until the Pleistocene in
Europe (Iberovaranus and Varanus). A
Late Cretaceous relative of the extant
Lanthonotus occurs in central Asia
(Borsuk-Bialynicka, 1982). The relation-
ship of Lanthanotus to the marine vara-
noids suggested by McDowell and
Bogert (1954) has not been generally ac-
cepted (Rieppel, 1980a), and both Estes
(1982) and Gauthier (personal communi-
cation 1981) include Lanthanotus in the
Varanidae. Rieppel (1980b) maintained a

372 Advances in Herpetology and Evolutionary Biology

separate family for the genus, but his
comparisons included only Lanthanotus,
Heloderma, and Varanus. Since Rieppel
lacked a more extensive generic series,
and in view of the many derived charac-
ters shared by Lanthanotus and Varanus,
I believe that it is reasonable to place
Lanthanotus in the Varanidae as a sub-
family.

LIZARD PALEOBIOGEOGRAPHY

The summary of the lizard fossil record
presented above and derived from Estes
(1982) now makes it possible to offer
some preliminary suggestions on the
biogeographic history of the different
families of lizards. Lillegraven et al.
(1979) have emphasized the tenuous data
on which mammalian Mesozoic bioge-
ography is based, and have observed that
this lack of data has not hampered the
flow of supposition; indeed, as A. S.
Romer often observed, “increase in data
leads to triumphant loss of clarity.” I am
well aware that the present paper is
based on data equally tenuous as those
for mammals so far as the Mesozoic is
concerned, and much more so for the
Cenozoic. Nevertheless, since the Hand-
buch (Estes, 1982) represents the first
codification of data regarding all fossil
lizards, some discussion of what it sug-
gests is desirable. No previous detailed
discussion of this sort has been based on
the fossil record, although a recent study
of lizard biogeography by Renous (1979)
has made use of continental plate recon-
structions. The zoogeographic models
discussed here are consistent with the
generally accepted movements of con-
tinental plates and also provide some cor-
roboration for the position of the South-
east Asian region suggested by Owen
(1976, 1981, 1982).

The diagram of continental plates used
here in Figures 3-13 is adapted from the
Early Cretaceous map in Smith and
Bryden (1977). Continental masses are
highly diagrammatic, and I have modi-
fied the map to reduce the polar distor-

tion of continents, but the relative dis-
tance between points of continental plate
contact or proximity are kept approxi-
mately the same as in Smith and Bryden.
The map has utility, however, principally
as a base map to be used in the context of
this paper; the maps of Smith and Bryden
(1977), Owen (1976), Lillegraven et al.
(1979) and Howarth (1981) should also be
consulted. I have chosen the Early
Cretaceous map as a base because conti-
nental separation first becomes marked at
that time, yet the original Pangaean con-
figuration is still apparent. The Early
Cretaceous was in addition an important
time in lizard evolution and distribution.
The known Jurassic record consists en-
tirely of extinct families, although they
adumbrate extant groups to one degree or
another. Since most extant families were
in existence by at least the Late Creta-
ceous, the importance of Early Creta-
ceous time is clear. It should be empha-
sized, however, that data from times
other than the early Cretaceous appear
on the maps, and the figure legends and
text should be consulted for explanation.
The account of lizard fossils above
demonstrates how scanty our record is, as
well as how few fossils we have from
Gondwana. It is not possible to develop
an adequate interpretation of lizard bio-
geography without this southem record,
and a variety of equally plausible hypo-
theses can be constructed with the exist-
ing data. I have chosen the interpretation
that seems most parsimonious to me
when the cladogram of lizard relation-
ships, the fossil record, and the present
distribution of lizards is confronted with
data on plate movements, but my account
will have to be modified as more informa-
tion is obtained from the South American
and African Jurassic and Cretaceous, the
Cretaceous of Europe, and of course any
Mesozoic and Cenozoic data from Asia.

PERMIAN- TRIASSIC

The Permian and Triassic record
shows only that up to the Early Triassic,

the Squamata was Gondwanan in distri-
bution. The group is not yet recorded
from South America but was presumably
present there as well. By the Late Trias-
sic the more derived kuehneosaurids
appear in Laurasia, and an herbivore,
Fulengia, occurs in eastern Asia (Fig. 2).
Although only a few localities are known,
the variety of lizards in the English
Triassic fissures (mostly still unde-
scribed) suggests that by the end of the
Triassic there was some diversity in
Laurasia as well. Cox (1974) has shown
that a “world fauna” existed during the
Triassic since there were few barriers to
animal movement during that time.
These data, poor as they are, suggest that
lacertilians were Pangaean in_ the
Triassic, and that the various groups
originated as vicariants following the
breakup of Pangaea, either by plate
separation or through separation by
intracontinental barriers.

JURASSIC

The single Lower Jurassic fossil that
has been referred to the lizards is the
poorly preserved Protolacerta, from
Argentina. It is not demonstrably a lizard;
part of the specimen is from a fish and
part from an indeterminate reptile (J.
Bonaparte, personal communication
1981).

In the Late Jurassic, the presence of
Cteniogenys in North America and
Europe indicates a significant extension
of the geological range of the eolacerti-
lians, already known from Laurasia in the
Late Triassic. The most significant aspect
of Jurassic fossil lizards, however, is that
all the major groups of extant true lizards
(Superfamilies of Romer, 1956; Infra-
orders of Estes, 1982) are represented in
the Late Jurassic of a relatively small
area—western E,urope—and that the
groups show very little mixture in locali-
ties (Table 1, Fig. 3). Greatest similarity
occurs among the English, Wyoming, and
Portuguese localities, all of which con-
tain Paramacellodidae and _ Dorset-

FossIL LIZARD DISTRIBUTION : Estes SIS

isauridae. Cteniogenys, the eolacertilian,
is lacking in the English localities, but
occurs in Portugal as well as in Wyoming.
Primitive scincomorphs (lacertoids?)
occur at Lerida, Spain, and the presumed
lacertoid Durotrigia overlaps with the
paramacellodids in England. This series
of localities has thus yielded Cteni-
ogenys, scincomorphs, and_ primitive
anguimorphs. In the Cerin lithographic
limestone in France, only the iguanian
euposaurs occur, and in the Solnhofen
lithographic limestone in Germany only
the ardeosaurs and bavarisaurs are found;
an ardeosaur also occurs at the same time
in Eastern Asia. The Chinese and Ger-
man ardeosaurs suggest that the gekkotan
diversification was Eurasian. The param-
acellodid—dorsetisaur—eolacertilian as-
semblage suggests that some _ scinco-
morphan and anguimorphan groups were
originally Euro-North American. Al-
though ecological and/or sedimentary
environment differences may be _ in-
volved, some barriers to movement seem
to have been present (see Howarth,
1981), and there is some geological evi-
dence for this hypothesis. The major ex-
tent of European epicontinental seas at
the approximate time of these Late Juras-
sic localities was in Central Europe, ex-
tending south through France (Ager,
1975; Hallam and Sellwood, 1976). There
was an emergent area in northem
Europe, north of the Solnhofen portion of
this seaway, where the ardeosaurs and
bavarisaurs must have been living, their
remains being incorporated into the
back-reef lagoons that later became the
lithographic limestone (Hallam, 1975). A
similar situation occurred at the Cerin
locality in France, with an emergent re-
gion to the southeast (Delfaud, 1980)
being a source for the euposaurs. The
Purbeck and Portuguese localities, on the
other hand, were associated with an
emergent region to the west, more
closely linked with North America and
Greenland, as indicated both by the plate
tectonic evidence as well as that of the
fossils. Details of the oscillations of these

374

Advances in Herpetology and Evolutionary Biology

NA

Figure 2. Known distribution of Triassic Sauria. Base map for Pangaea in the Late Triassic (200 m.y.b.p.) from

Smith and Bryden (1977).

Abbreviations: AF = Africa; AN = Antarctica; AU = Australia; EA = Eurasia; IN = India; NA = North America; NZ =
New Zealand; SEA = Southeast Asia; Greenland and Madagascar not marked.

epicontinental seas are not entirely clear,
but the archipelagic area formed in
western Europe could have permitted
vicariance of these and perhaps other liz-
ard groups. For the most part these seas
disappeared in the latest part of the
Jurassic, when most of Europe (and many
other parts of the world) became emerg-
ent and provided at least a brief connec-
tion with North America before Creta-
ceous epicontinental seas isolated the
two areas by inundating most of Europe
once again (Prothero and Estes, 1980;
Hallam, 1981).

With the Triassic and Jurassic record in
mind, we may now proceed to discussion
of the biogeography of Cretaceous and

later lizards and their relationship to the
Jurassic forms.

IGUANIAN DISTRIBUTION AND THE
MODEL OF RENOUS

A recent review of lizard biogeography
by Renous (1979) was based mainly on
structures and distributions of living
groups. Although she did consider past
continental movements, her discussion of
lizard fossils ignored many recent contri-
butions in this area. Renous emphasized
the Iguania in her discussion, and formu-
lated hypotheses of iguanian (and other
lizard) relationships and distribution,
based principally on her extensive

FossiIL LIZARD DISTRIBUTION - Estes 375

Figure 3. Known distribution of Jurassic Lacertilia. Base map for Early Cretaceous (120 m.y.b.p.) from Smith and
Bryden (1977). & = paramacellodids, dorsetisaurs, and possible lacertoid Durotrigia (England); @ = scinco-
morphs (possible lacertoids) Meyasaurus and /laerdesaurus (Spain); * = paramacellodids, dorsetisaurs and
possible scincoid Saurillodon (Portugal); A = Paramacellodus, Dorsetisaurus (Wyoming); 0 = Euposaurus
(France); © = ardeosaurs Ardeosaurus and Ejchstaettisaurus, bavarisaurs Bavarisaurus and Palaeolacerta
(Germany); © = ardeosaur Yabeinosaurus (People’s Republic of China). Continents identified in Figure 1. Note
large eastern Tethys Sea; see text and compare continental reconstruction in Figure 14.

studies of the limb innervation (Renous,
1978, 1979, 1981; the latter contains little
new information). In the lizard forelimb,
the course of the ulnar nerve is superfi-
cial to the muscles as in most tetrapods
(primitive or lacertide condition of
Renous = L). In some lizards, however,
the nerve takes a deep course (derived or
varanide condition = V). In the hind
limb, the dorsal muscles are generally
innervated in tetrapods by the peroneal
nerve (primitive condition = A), but
uniquely in some lizards the peroneal
nerve is absent and its function taken
over by the interosseous nerve (derived
condition = B). The polarity of these

character states seems well established.
Derived conditions are as follows: B oc-
curs in most iguanids, all agamids, some
cordylids, and Heloderma; V occurs in
varanids and Afroasian lacertids; V and B
together are found in Malagasy iguanids,
chamaeleonids, teiids and gymnophthal-
mids. All other groups have the primitive
states L and A. From the developmental
standpoint the difference between the
lacertide and varanide conditions of the
forelimb does not appear to be funda-
mental. Since nerves grow, in develop-
ment, toward the muscle _ blastemata,
slight differences in timing or conditions
in the mesenchyme could lead to adop-

376

TABLE 1. DISTRIBUTION OF KNOWN LATE JURASSIC
LIZARDS BY LOCALITY, SHOWING THAT VERY LITTLE
OVERLAP IN INFRAORDINAL GROUPINGS OCCURS.

~”
aS 5

a & eS
fee Se 8 2 BE
© 6 (Sy) Mo ss > 3 =} a
20 eI c Ss y fos)
Ses ef a
= = 3 S 2)
Seae® Ss 8 ® e
S = = Ss & & Ss
~ os Oo =~ s oS =) se)
Oeane ASS © gs

Purbeck,

England xX X

Guimarota,

Portugal

Quarry 9,

Wyoming X X X

Lérida,

Spain xX X

Cerin,

France XxX

Solnhofen,

Germany xX X

People’s Republic

of China

tion of a different pathway; such charac-
ters are readily subject to both parallelism
and reversal. The difference between the
A and B states is less ambiguous since a
loss is involved; here again, however, the
phylogenetic significance is doubtful
since multiple loss can occur. Reversal is
possible if the “loss” were to be the result
of epigenetic processes not accompanied
by loss of potentiality (see Alberch et al.,
1979).

Renous (1979) also discussed character
states of the dentition, forearm and
shoulder musculature, scapulocoracoid
fenestrations, humeral foramina, form of
the carpus, cervical vertebrae, and fe-
moral-preanal pores. Most of these char-
acter states show extensive parallelism;
some are correlated, and they will not be
discussed further here owing to space
restrictions. In any case, Renous’ cladis-
tic models for iguanian relationships
were based only on the nerve pathways
and the presence of an acrodont denti-
tion. She then used these results in bio-
geographic analysis, an analysis that is

Advances in Herpetology and Evolutionary Biology

generally unsuccessful because homo-
plasy (which occurs in most characters)
has a greater effect on a cladogram when
only a few characters are used; her cladis-
tic models are discordant, in general,
with those based on more extensive data.
Her view (1979:395) is stated unambigu-
ously in the original as well as in my
translation:

Within the classification of lizards, all these char-
acters [i.e., the nerve pathways] do not appear to
us to be distributed by chance, but to obey a cer-
tain logic, a certain rationality that cannot be other
than the phylogenetic history of the group. The
distribution of these characters reflects phylo-
genetic relationships.

It is possible to accept that in some cases
the distribution of the nerve pathways
indicates phylogenetic relationships, but
if all possible cladistic arrangements of
these characters are diagrammed it is
clear that homoplasy must have occurred,
since all possible combinations of the
two character states occur in living liz-
ards (i.e., LA, LB, VA, VB). Since B is a
state defined by loss, it seems less likely
that A has been obtained by reversal, al-
though as we shall see, this possibility
has to be considered in some situations.

Several separate origins of V are
demonstrated by its occurrence in some
iguanians, Afroasian lacertids, teiids and
gymnophthalmids, and the varanids,
occurrences discordant with the exten-
sive data that indicates lack of relation-
ship between the iguanians, the varanids,
and the other groups. The situation with
B is more complicated. It is widespread
in iguanids, but A is present in iguanines,
while the sceloporines and tropidurines
may be either A or B. Since these three
groups are quite homogenous in most
other derived character states, it is pos-
sible that the absence of the peroneal
nerve in iguanids is the result of genetic
suppression without loss of potentiality.
Alternatively one must argue for numer-
ous independent losses of the nerve
within iguanids, and this is possible yet
much less parsimonious. Jullien and

Renous (1972) gave data for morunasaurs
and Crotaphytus, and these primitive
iguanids also have both L and B (con-
firmed by K. de Queiroz, personal com-
munication 1980), suggesting that rever-
sal in iguanines, some of the tropidu-
rines, and some of the sceloporines is the
most parsimonious alternative. Separate
developments of B have occurred in
Heloderma, and in some cordylids as
well (in the latter group Renous noted,
[1979: 394] that the conditions of B are
not perfectly attained’’), and in both the
teiids and gymnophthalmids. In the latter
two groups the presence of both derived
conditions V and B may be phylogeneti-
cally significant.

Renous (1978, 1979) was particularly
concerned with the relationship between
the Malagasy iguanids, the agamids, and
the chamaeleonids, and emphasized in
this regard the common presence of de-
rived conditions V and B in the oplurines
and chamaeleonids, but as noted above
this combination occurs also in the teiids
and gymnophthalmids. She suggested
that the agamids, in retaining the primi-
tive condition L with the derived condi-
tion B, are more closely related to the
New World iguanids, while oplurines
gave rise to the chamaeleonids, the latter
two groups later invading Madagascar. In
her view, the agamids underwent vi-
cariance on the Indian block, only
secondarily invading Africa. This hy-
pothesis required her to postulate a paral-
lel origin of acrodonty in both agamids
and chamaeleonids. On the other hand,
Moody (1980), who has reevaluated the
relationships and biogeography of
agamids, indicated that determination of
relationships among the _ iguanids,
chamaeleonids, and agamids requires
further study owing to the fact that
chamaeleonids share approximately the
same number of derived states with both
agamids and Malagasy iguanids (oplu-
rines), and he hypothesized that various
agamid groups had centers of origin on
Asian, Southeast Asian, and Australian
plates, following the continental recon-

FOSSIL LIZARD DISTRIBUTION : Estes 377

struction of Owen (1976). Moody’s de-
tailed analysis is in agreement with
Renous’ conclusion that Africa secon-
darily received its agamids, although he
brings them from Asia rather than from
India. Moody also suggested that Aus-
tralia and Southeast Asia received
agamids simultaneously; he emphasized
the closer relationship between South-
east Asia and Australia required for this
interpretation that can be found in
Owen's (1976) reconstruction but not in
the more usual continental arrangements
(e.g., Smith and Bryden, 1977, and see
Fig. 14). Owen’s interpretation will be
discussed further below.

My own view of early lacertilian evolu-
tion is that a Jurassic ancestral moru-
nasaur-like population of iguanoids (es-
sentially at iguanid level) occupied at
least the southern portions of the South
American-African region of Gondwana
(Fig. 4). This iguanoid group was the re-
sult of vicariance following separation of
Pangaea into two more or less distinct
areas, Gondwana and Laurasia; this basal
lacertilian dichotomy left the Laurasian
lizards to develop into the more derived
groups that will be treated in the follow-
ing section. The iguanid group would
have undergone further vicariance when
South America and Africa separated (a
process initiated in the very latest Juras-
sic, primarily accomplished in the Early
Cretaceous and completed by about
Middle Cretaceous, see Tarling, 1980),
separating the South American and Afri-
can forms (Fig. 5). Within the African re-
gion, some of the ancestral morunasaur-
like forms (which were surely in the
Madagascar region before separation of
the latter from Africa) became oplurines,
and developed condition V in parallel
with chamaeleonids (recall that this is
simply a different, deeper path for the
ulnar nerve, and that V has arisen inde-
pendently in other lizard groups).

In the South American region, it is pos-
sible that Pristiguana from the Late
Cretaceous of Brazil was a representative
of the early morunasaur group that di-

Advances in Herpetology and Evolutionary Biology

Figure 4. Suggested areas of origin (stippled) in primitive Iguania; see Figure 3 for map data and see text for
further explanation. Only initial separation into groups shown; subgroup differentiation shown in Figure 5.

versified into the non-oplurine iguanids.
North American iguanids seem to have
been the product of rather limited dis-
persals from South America (Savage,
1966), probably beginning in the Late
Cretaceous and extending on into the
Early Cenozoic. These movements did
not involve vicariance but resulted from
chance dispersal across a series of vol-
canic islands that arose during the Late
Cretaceous between North and South
America owing to collision of a Pacific
aseismic ridge with this island arc system
(Schmidt-Effing, 1979), or perhaps tran-
sient upwarpings produced by heat of
plate movements (H. Owen personal
communication 1980). Rage (1978, 1981)
has suggested that a number of snake
genera dispersed across this region,
teiids also took this path (see below), as
did some _ hadrosaurine dinosaurs

(Casamiquela, 1980) and mammals
(Ferrusquia-Villafranca, 1978; Lillegra-
ven et al., 1979; Hallam, 1981). The
North American Late Cretaceous record,
however, shows no definitive record of
iguanids, which are first recorded by
Paradipsosaurus in the Paleogene of
Mexico and Parasauromalus in the Early
Eocene of Wyoming; both genera appear
to be relatively primitive morunasaur-
like forms. The Oligocene Aciprion and
an undescribed crested relative (see
above) apparently represent more
northern representatives of basiliscines.
The sceloporines probably came from
limited dispersal of members of a com-
mon ancestor with tropidurines. While
the latter diversified in South America
and the West Indies, sceloporines ex-
ploited the more open, arid habitats in
southwestern North America, and only a

single tropidurine, the presently West
Indian genus Leiocephalus, ever reached
North America (Miocene). Iguanines may
have been derived from _ primitive
iguanids that also reached Central
America from the south. Finally, a single
iguanid appears to have reached Europe;
the Middle Eocene Geiseltaliellus dis-
persed across the North Atlantic route at
the same time as the agamid Tinosaurus,
glyptosaurine anguids, the varanid
Saniwa, and a number of mammals and
other lower vertebrates during a period
of peak tropicality in the Eocene.

Part of the ancestral morunasaurlike
iguanid group in the African region may
have been separated by some ecological
or physical barrier in the northem or
northeastern portion, developed acro-
donty, and thus became an ancestral
group for the agamids and the chamae-
leonids. This group would probably have
had nerve conditions L and B as do
iguanids and agamids today. Following
Moody (1980), agamids then originated
from portions of this population that
underwent vicariance on the Asian,
Southeast Asian, Indian, and Australian
plates following separation of Africa from
eastern Gondwana in the Early Creta-
ceous (Fig. 5). Southeast Asia and eastern
Asia have localities that yield a Triassic
Glossopteris—Lystrosaurus biota (Col-
bert, 1973), suggesting an association of
these areas with Gondwana, an associa-
tion corroborated by the continental
reconstruction of Owen (1976, 1981,
1982, see below). The primitive agamid
stock is represented in the Late Creta-
ceous of Central Asia by Mimeosaurus,
and later, in the Eocene, by the primitive
agamid Tinosaurus, which has been
recorded from the Paleocene and Eocene
of eastern Asia and the Eocene of Europe
and North America as well. All records of
Tinosaurus are fragmentary, and while
the Asian forms are primitive agamids the
generic reference is in doubt. The North
American and European specimens are
very similar, however, and appear to be
congeneric; corroboration for this is the

FOSSIL LIZARD DISTRIBUTION : Estes 379

presence of Eocene lizard, turtle,
champsosaur, crocodilian, and mammal
genera common to both these areas,
suggesting dispersal across the North
Atlantic corridor during Eocene peak
tropicality. While the Asian fossil record
is very scanty, no data suggest that this
dispersal was across the Bering Route,
although this is a possibility (Fig. 5).

Portions of the ancestral L-B-acrodont
population of iguanoids remaining in
Africa after the vicariance of agamids
developed nerve condition V and gave
rise to the chamaeleonids, which by the
Late Cretaceous invaded Madagascar (or
perhaps had been there prior to its sepa-
ration) and in the Cenozoic reached
Eurasia.

The Late Jurassic Euposauridae in
France may have been a more highly
derived branch of the common acrodont
ancestor of agamids and chamaeleonids,
a group that moved northward, briefly,
into Laurasia when agamid ancestors
spread to Asia and Southeast Asia. This
would explain their general similarity to
agamids, their retention of certain primi-
tive conditions such as paired bones of
the skull roof, and their development of
some derived conditions not found in the
most primitive living agamids.

In brief summary, then, a southwestern
Gondwanan (South America—Africa—
Madagascar) primitive iguanian group
diversified during Middle-Late Jurassic
time after separation from Laurasian liz-
ard groups. Neotropical iguanid groups
and oplurines developed by vicariance as
South America and Africa diverged (a
process completed by the Middle Creta-
ceous). A more northern Gondwanan
acrodont iguanian group underwent vi-
cariance as the Asia-Southeast Asia-
Australia-India blocks separated from
Africa. The agamids and chamaeleonids
resulted from this plate separation. Much
later, following their Mesozoic subgroup
differentiation, primitive agamids spread
widely in the Eocene, occurring in North
America, Europe and Asia; sympatry
with iguanids in North America, at least,

380 Advances in Herpetology and Evolutionary Biology
= _-— -_—
= ~ -_
Figure 5. Suggested areas of origin (stippled) and later dispersals (arrows) in the families of iguanians. Iguanid

subgroups formed after vicariant separation of South America and Africa; oplurines relict in Madagascar; Late
Cretaceous or Early Cenozoic dispersal of iguanids to North America and in the Eocene, apparently, to Europe.
Agamid subgroups formed by vicariance following separation of Southeast Asia, Australia and India; dispersal of
agamids into Europe and North America in the Eocene (North American route chosen here; possible Bering route
shown as dotted line). Chamaeleonids disperse to Europe in Miocene; to India after collision with Asia (not shown).
Map data as in Figure 3. Dates in or at end of arrows represent earliest known pre-Pleistocene fossils of the group in

that area. See text for further explanation.

confirms this as dispersal. Some Asian
agamids also invaded Africa and India in
the Cenozoic in a limited way (Agama,
Uromastyx, Leiolepis). Agama and
Uromastyx reached Europe no later than
the Late Eocene or Early Oligocene,
either from Asia Minor or Africa, with
Agama persisting in Europe until the
Early Pleistocene (it still occurs there
peripherally). It is likely that the chamae-
leonids reached India as a result of range
extension in the Early or Middle Ceno-
zoic, after collision of India with Asia.
While there is no present record of such
an extension during Eocene tropicality,
the group had reached Central Europe by

the Miocene returm of tropical conditions
(see e.g., Cracraft, 1973a). Cracraft
(1973b, 1975) suggested that agamids and
chamaeleonids diverged after separation
of South America and Africa in the
Middle Cretaceous; it is likely to have
been earlier than that time.

A major unanswered question in
iguanid distribution is the absence of
iguanids in Africa today. If a few have
survived in Madagascar and the major
evolution of the group was South
American, what happened to the inter-
vening African populations? There is no
obvious answer, but it is possible that
iguanids may have been reduced by

)

shifting climatic belts (as discussed by
Robinson, 1973), perhaps in combination
with other, as yet unknown factors.

POST-IGUANIAN LIZARD DISTRIBUTIONS

Gekkota. Renous (1979) suggested
Europe as a center of origin for gekko-
tans. While the apparently gekkotan
ardeosaurs and bavarisaurs occur in the
Upper Jurassic of northern Europe, the
presence of an ardeosaur in eastern Asia
indicates a wider distribution (Fig. 3).
Since the localities with ardeosaurs con-
tain no trace of the contemporary radia-
tion of paramacellodids, lacertoids and
euposaurs, gekkotans may have origi-
nated on a portion of Europe and Asia
that was at that time not connected with
southwestern Europe. During the Late
Jurassic, Eurasia began to be bisected by
the so-called Obik Sea, which separated
the European region and isolated Siberia
and eastern Asia from the rest of the
world (Cox, 1980). Although the Obik Sea
(or Turgai Straits) separated Europe from
Asia completely by Cretaceous time, the
initiation of this seaway from the north
during the Late Jurassic would have ef-
fectively separated the northern Euro-
pean ardeosaurs and bavarisaurs from the
Asian ones that are more likely to have
been the actual gekkonid ancestors from
the geographic standpoint; an Asian ori-
gin of gekkonids is supported by their
present distribution. Kluge (1967) sug-
gested that eublepharine gekkonids gave
rise to the now wholly Australian diplo-
dactylines and the gekkonines in Asia,
with the latter spreading westward to
India, Africa, and Madagascar. South
American gekkonines are low in diver-
sity; they and the sphaerodactylines ap-
pear to be derivatives of highly dispers-
able cosmopolitan genera (see Bons and
Pasteur, 1977). Whether they reached the
New World by Cretaceous predrift dis-
tribution or postdrift rafting is unknown,
but their low present diversity suggests
that rafting is the more likely possibility.
Tarling (1980) suggested that South

FOSSIL LIZARD DISTRIBUTION - Estes 381

Atlantic oceanic currents would have
fostered dispersal from Africa to South
America but not the reverse, and further
pointed out that a series of large oceanic
islands may have been available along
the Ceara-Sierra Leone Rises up to the
end of the Eocene. The only South
American fossil gekkos are from the
Paleocene of Brazil; they have not yet
been identified to subgroup, but give a
minimum date for emplacement of gek-
konids on that continent. Eublepharines
are likely to have reached North America
from either Asia or Europe in the Early
Cenozoic, rather than in the Late Creta-
ceous when the Bering connection first
became available, since there are no
gekko fossils in the relatively extensive
Late Cretaceous deposits of North
America and they were present in Cen-
tral Asia at the time (unpublished ma-
terial being studied in Warsaw). An Eo-
cene gecko occurred in California but
cannot be identified to subgroup; this
animal as well as the Eocene-Pliocene
European geckos probably dispersed
during Eocene and Miocene peaks of
tropicality (see e.g., Cracraft, 1973a).
The distribution pattern for geckos is
thus very similar to that for agamids as
described by Moody (1980), with primi-
tive Southeast Asian forms, a_ well-
defined Australian radiation, and several
endemic genera in India, dispersing later
into Africa and Madagascar (Fig. 6). Both
geckos and agamids are found in the Late
Cretaceous of Central Asia and the Eo-
cene of North America and Europe.
Again, however, lack of fossils from many
areas of present gecko distribution makes
analysis of past movements difficult. The
pygopodids lack a fossil record, but if
derived from diplodactylines in Australia
as Kluge supposes (1981, personal com-
munication), must have lost the type C
double visual cells present in all gekkonid
subfamilies. Alternatively as G. Under-
wood believes (personal communication
1981) they may have branched off prior to
gekkonid acquisition of these cells; this
would fit with their retention of the large,

382

a

Paleoc.

Advances in Herpetology and Evolutionary Biology

e
(so.

Figure 6. Suggested area of origin (stippled) and later dispersals (arrows) in geckos. Map data as in Figure 3.
Dates at end of arrows represent earliest known pre-Pleistocene fossils of the group in that area. Possible Bering
route to North America for Eocene geckos shown in dotted line, see text. Note vicariance of gekkonid subgroups in
Southeast Asia, Australia, and India. Only major paths shown; island or rafting dispersal routes not included. See

text for further explanation.

regular dorsal head scales. Dibamids (in-
cluding anelytropsids) seem to be related
to the gekkotans (Underwood, 1971) and
while they also lack a fossil record, again
point to an originally Laurasian deploy-
ment of the Gekkota.

Scincomorpha. While the iguanoids
were radiating in western Gondwana,
scincomorphs were diversifying in
Laurasia, as shown by the presence of
cordyloids and perhaps scincoids in the
Late Jurassic of Europe. Definitive
lacertoids (the most primitive of scinc-
omorphs) are not known at this time, al-
though the presence of the more ad-
vanced cordyloid group indicates that
lacertoids must have already evolved,
and ancestral lacertoids may have split
off from a common ancestor with iguan-

ians in Eurasia (Fig. 7). Meyasaurus and
Ilaerdesaurus from the Late Jurassic of
Spain, and Durotrigia from the Late
Jurassic of England, may represent early
lacertoids (Fig. 3).

The unity of living lacertoids seems to
be demonstrated by a number of derived
character states (Hoffstetter, 1962; Riep-
pel, 1980c; Estes, 1982; Gauthier, in
preparation), although Presch (1981,
personal communication) believes lacer-
tids and teiids to be widely separated.
Presch also allies the gymnophthalmids
with the lacertids, but teiids and gymno-
phthalmids share derived character states
of the hemipenial musculature (N.
Arnold, 1982, personal communication),
and the presence of derived nerve condi-
tions V and B; I thus continue to ally

teiids and gymnophthalmids while await-
ing the results of the studies of Presch
and Gauthier.

Cordyloids (and perhaps scincoids) are
present in the Late Jurassic, and as noted
above, lacertoids were present in west-
ern lLaurasia at least; one possible
hypothesis is that the teiid-lacertid
dichotomy arose by separation of a part of
the ancestral population on one of the
large continental islands in the European
region during the late Jurassic. This vi-
cariance could then have produced the
lacertids, while the teiids developed in
North America, where they are exten-
sively represented by the Late Creta-
ceous, including teiine, tupinambine,
and polyglyphanodontine groups. A rela-
tively restricted radiation of poly-
glyphanodontines was also present in
eastemm Asia at this time. The poly-
glyphanodontines were a group that
capitalized on relatively large size and
herbivorous-omnivorous habits, and
while they are a primitive group of teiids
their divergent dietary specializations
preclude them from ancestry to teiids as a
whole. This hypothesis presumes that
teiids reached South America from the
north. While this is against the ocean cur-
rent pattern that Tarling (1980) supposes
would have fostered only south to north
movement, such groups as hadrosaurian
dinosaurs clearly must also have come
from the north.

So far as lacertids are concerned, much
of Europe was submerged for the major
part of Cretaceous time, so that extensive
habitat for lizard groups was not present.
Eurasian lacertids appear to be the most
primitive group, and the more derived
African forms (with some later reentrants
into Eurasia) may have developed from
isolation on northwest Africa in the Late
Cretaceous, when that region was sepa-
rate from both Europe and the rest of
Africa.

An alternative hypothesis for the lacer-
tid-teiid dichotomy would be to postulate
the presence in western Gondwanaland
of an ancestral lacertoid population (of

FOSSIL LIZARD DISTRIBUTION : Estes 383

which Durotrigia would then, like the
euposaurs, be a Laurasian extension) that
underwent vicariance during the Early-
Middle Cretaceous separation of Africa
and South America, producing lacertids
in the former, teiids in the latter. Since
teiines and tupinambines are known
from the Late Paleocene of Brazil, they
must have thus had a long history in
South America, although since the
Cretaceous record in South America is
poor we do not know if they were present
there at that time. Estes (1970) favored
this hypothesis.

In this second hypothesis, the lacertids
would be primarily African, but separa-
tion of the more primitive, now Eurasian
forms from the more derived African
forms need not be much different than in
the first hypothesis.

Neither of the above alternatives pro-
hibits the gymnophthalmids from having
been endemic to the South American
region as postulated by Presch (1980). In
the first alternative the gymnophthalmid
stock could have been the result of exten-
sion of a teiid stock into South America
from the north during the (Late?) Creta-
ceous, with divergence possibly from
intracontinental factors; in the second
hypothesis the source would have been
from Africa. When the affinities of the
gymnophthalmids are established more
clearly perhaps a choice of these alterna-
tives will be easier to make.

In Figure 7, I have chosen the first of
these hypotheses, principally on the
basis that the known fossil record is
earlier in North America and Asia and
that Durotrigia, the possible ancestral
lacertoid, occurs in Laurasia. Once again,
however, the record is poor, and I do not
think the second hypothesis can be falsi-
fied at present.

Beyond the lacertoid level of organiza-
tion lie the more derived scincoid-
cordyloid groups, together forming the
sister group of the lacertoids. Origin of
these groups must have been no later
than the Upper Jurassic, since primitive
lacertoids seem already to have been

io)
(o')

Paleoc.

4 Advances in Herpetology and Evolutionary Biologi
&l Y Sy

Paleoc.

Figure 7. Suggested area of origin (stippled) and later dispersals in lacertoids. Map data as in Figure 3. Dates at
end of arrows represent earliest known pre-Pleistocene fossils of the group in that area. Alternative hypothesis
discussed in text. + = Late Jurassic ?lacertoids; © = Late Cretaceous polyglyphanodontine teiids.

evolved, and primitive cordyloids occur
in the Late Jurassic of England, Portugal,
and Wyoming. Here in particular the lack
of a better African and Asian record
makes interpretation difficult. At this
stage of knowledge the scincoid radiation
seems to have taken place in a different
area from that of cordyloids. Perhaps the
simplest explanation is to presume that a
primitive scincomorph stock had de-
veloped in Laurasia before the latest
Jurassic. By then, the Tethys Sea was
relatively open from the Pacific to at least
as far as southern Asia, effectively isolat-
ing much of Laurasia and Gondwana.
The fluctuating epicontinental seas of the
European region could have first caused
the lacertid-teiid dichotomy, and then
fostered the divergence of western
Laurasian populations that became the

cordyloids, and eastern populations that
evolved into scincoids.

The scincoids (Scincidae) are now so
diverse and cosmopolitan a group that it
is difficult to determine their bio-
geographic history. Greer (1970, 1974,
1977) has given some discussion of the
matter, but analysis of the primitive
eumecoid skinks is still in progress.
Moreover, many skinks are well known
for their ability to cross water barriers
(see e.g., Darlington, 1957) and they have
colonized islands extensively. The most
primitive scincines are placed in the sub-
family Scincinae, which includes
Eumeces, probably the most primitive
living scincid. This group is most rele-
vant to the present discussion of early
diversification of lizard families; the
more derived lygosomine and _ other

scincid groups will not be considered
here (Fig. 8). The scincines seem to have
spread widely in southern Asia, and other
scincid groups reached Australia, Africa,
and Madagascar. Since Neotropical
scincids are very limited in diversity,
being represented (like the geckos) by
probable raftings of cosmopolitan genera,
the group probably did not arrive in
Africa before the Late Cretaceous, after
separation of Africa from South America.
At that time Africa approached Laurasia
more closely as a result of northeastward
movement and rotation engendered by
separation from South America; this
would have facilitated entrance of a
number of groups (e.g., geckos, agamids)
into Africa. Another factor that may have
operated to keep scincids (and perhaps

Paleoc.

FOSSIL LIZARD DISTRIBUTION - Estes 385

geckos and cordylids as well) from reach-
ing South America (except by rafting) was
the fact that in the Late Cretaceous, much
of northwestern Africa was separated into
several continental islands by epiconti-
nental seas (see Howarth 1981: Fig.
13.18).

A primitive, probable scincine genus
Contogenys occurred in the Late Creta-
ceous and Paleocene of North America,
together with the more problematic and
perhaps cordyloid genus Sauriscus. In
the Paleocene of Brazil an apparent rela-
tive of Contogenys is found. At present
there are no fossil scincids known in
Europe (Cretaceous localities are essen-
tially unknown, owing to extensive sub-
mergence, and no scincids occur in the
relatively rich Cenozoic localities there),

primitive

Figure 8. Suggested area of origin (stippled) and later dispersals (arrows) in primitive Scincoidea (Scincinae).
Entrance of Contogenys-like scincine into South America can have been either Late Cretaceous or earliest
Paleocene. Map data as in Figure 3. Dates at end of arrows represent earliest known pre-Pleistocene fossils of the

group in that area. See text for further explanation.

386

or in the large and diverse samples of fos-
sil lizards in the Late Cretaceous of Cen-
tral Asia; their direction of entrance into
North America is thus difficult to predict.
As noted above, entrance through South
America is unlikely. It is possible that
scincids were actually present in Central
Asia during the Cretaceous, but in areas
or environments different from those that
have provided the known fossils. If so,
they could have reached North America
over the Bering connection, which was
utilized by other lizards and mammals in
the Late Cretaceous, finally reaching
South America briefly in the Paleocene.
This may be the simplest explanation at
present, since the living South American
scincids include only a few African-
derived (and probably rafted) lygos-
omines. Eumeces itself occurs in North
America in the Oligocene and is very
likely to have been a later Bering migrant
during the Early Cenozoic. Greer (1974)
suggested this route for Scincella, an
Asian lygosomine that also occurs in the
southeastemm United States. Origin of the
endemic North American Neoseps was
perhaps through a similar migration path.

The cordyloids may have been initially
a Laurasian group as indicated by their
past distribution in Wyoming, England
and Portugal in the Upper Jurassic (Fig.
9, paramacellodids). This wide distribu-
tion of a primitive cordyloid group pro-
vides a means by which xantusiids can
have been North American vicariant
derivatives of a primitive cordyloid stock,
while the Cordylidae itself followed a
southern direction, eventually becoming
limited to subsaharan Africa and Mada-
gascar after the development of the
Sahara Desert. Cordylids were present in
the Early Cenozoic of Europe, and it is
possible that they dispersed northward
from an original African center, perhaps
during peak Eocene tropicality when
other groups also extended their ranges
(iguanids, agamids, gekkonids), and
varanids). This is perhaps the best choice
given that the European cordylid record
at present is limited to a single species

Advances in Herpetology and Evolutionary Biology

and because much of Europe was sub- |

merged during the Cretaceous (Hallam,
1981). Given present data, I suggest that
an as yet unknown Cretacous paramacel-
lodid derivative produced widely distri-
buted populations that were eventually
limited to North and Central America,
becoming the Xantusiidae, and to Africa
and Madagascar, becoming the Cord-
ylidae.

Anguimorpha. As Renous_ (1979)
stated, origin and distribution of an-
guimorphs was not simple. Anguimorpha
and Scincomorpha appear to be sister
groups (see e.g., Camp, 1923) and some
similarities between some of the Jurassic
scincomorphs (paramacellodids) and the
earliest anguimorphs (dorsetisaurs) sup-
port this conclusion. Again we are faced
with the great hiatus in the record be-
tween the Late Jurassic and the Late
Cretaceous, but by the end of Mesozoic
time the anguimorphs had diversified
widely. In North America all known
families (including most marine forms, if
actually related) were present by the
Late Cretaceous, and at the same time in
Central Asia necrosaurs and varanids are
known, suggesting a long Late Mesozoic
radiation in Laurasia (Cracraft, 1973b
also described anguimorphs as_ Laur-
asian). Both the Late Jurassic dorseti-
saurs and the Cretaceous-Holocene
xenosaurs appear to represent a primitive
level of anguimorph organization, and
the only evidence that more derived
levels were present in the Jurassic is the
presence of the presumed aigialosaur
Proaigialosaurus from the Solnhofen
limestones in Germany (Hoffstetter,
1964), if that group is in fact varanoid.
Both from the standpoint of the fossil
record and the distribution of Xeno-
saurus today, the xenosaurs seem to have
originated in North America (Fig. 10),
although if the Holocene Shinisaurus
from eastern Asia is actually related to
Xenosaurus, a possible wider area of
origin could be considered (Asiamerica
of Cox, 1974 and Rage, 1981). Xenosaurs
had a limited radiation in the Late

Jurassic
paramacellodids

xantusiids
Paleoc.’

FoOssIL LIZARD DISTRIBUTION - Estes 387

Figure 9. Suggested area of origin (stippled) and later dispersals in cordyloids. Movement of cordylids into Europe
would have been during Eocene tropicality, unless they had had a previous wider distribution there dating from the
Mesozoic. Map data as in Figure 3. Dates at end of arrows represent earliest known pre-Pleistocene fossils of the

group in that area. See text for further explanation.

Cretaceous and Cenozoic in North
America (and perhaps Asia) and were
eventually limited to Mexico and Central
America (and perhaps eastern Asia).
They were not present in the Cenozoic of
Europe, corroborating a North American
or Asiamerican origin; their distribution
is thus similar to that for xantusiids, ex-
cept for the possible Asian connection.
The anguids themselves also seem to
have had a North American origin. They
Occur there in some diversity from the
Late Cretaceous on into the present, and
the Early Cenozoic anguids from Europe
seem to have been a more restricted
group derived from Early Cenozoic
North American forms, rather than being
part of a broad Laurasian distribution
(Fig. 11). Lack of a European Cretaceous
record again makes this difficult to cor-

roborate, but in the excellent sample of
fossil lizards from the Late Cretaceous of
Central Asia no anguids are present, sug-
gesting that they were either limited to
North America at that point, or to the lat-
ter and westem Europe. Primitive
(melanosaurinid) anguids in the Early
Cenozoic of North America include
Xestops and Melanosaurus; at least the
first of these genera also occurs in the
Eocene of Europe. The more derived
(glyptosaurinid) genera in North America
seem to be more closely related to each
other than to Placosaurus, an Eocene
glyptosaur from Europe and Asia. The
occurrence of Placosaurus in Asia indi-
cates that the Obik Sea, which separated
Europe and Asia at that time, was not a
barrier to at least some lizards. For the
above reasons, I have chosen the North

Jurassic

ancestral

Advances in Herpetology and Evolutionary Biology

Figure 10. Suggested area of origin (stippled) and later dispersals (arrows) in primitive anguioids. Map data as in
Figure 3. Dashed line indicates possible dispersal of xenosaurs to Central Asia. Dates at end of arrows represent
earliest known pre-Pleistocene fossils of the group in that area. See text for further explanation.

Atlantic route for glyptosaur entrance
into Europe (Fig. 11), rather than the
Bering connection. An apparent glypto-
saurine from the Eocene of the Canadian
Arctic Archipelago (Estes and Hutchin-
son, 1980) indicates that dispersal could
have been through the North Atlantic
connection to Europe, which was main-
tained at least until the end of the Early
Eocene.

In North America the gerrhonotine and
diploglossine groups appeared early, the
former perhaps as early as the Late
Cretaceous, the latter in the Eocene
(Gauthier, 1982). Diploglossine em-
placement on the West Indies was
probably from North America; how and
when they reached South America is less
easily determined.

While gerrhonotines and diploglos-
sines appear to have originated in North

America, the case of the anguines is less
clear. Ophisaurs appear earlier in Europe
(Eocene) than in North America (Oligo-
cene), suggesting that they were Eura-
sian derivatives from an anguioid stock.
The fossil record for these times, from
both areas, is good enough to allow some
confidence in this hypothesis. Several
possibilities for origin of anguines exist
that can be tested by future work. One is
that the predominantly “grass-swim-
ming’ ophisaurs arose in the early Ceno-
zoic in association with the distribution of
grassland and savannah environments,
finally achieving their present extensive
Laurasian distribution. Another possi-
bility, which I consider less likely, is that
the Old and New World ophisaurs are
parallel developments from primitive
anguid ancestors (see e.g., Klembara,
1979). In either case, ophisaurs appear to

FOSSIL LIZARD DISTRIBUTION : Estes 389

Figure 11.

Suggested area of origin (stippled) and later dispersals (arrows) in Anguidae. Glyptosaur movement

was in the Eocene. Date of diploglossine dispersal to South America not known. Ophisaurs not shown. Map data as
in Figure 3. Dates at end of arrows represent earliest known pre-Pleistocene fossils of the group in the area. See text

for further explanation.

have originated subsequent to the early
diversification of anguids in the Creta-
ceous. The anniellines are North Ameri-
can endemics, first appearing in the Early
Eocene, and again suggest a North Ameri-
can origin for Anguidae.

The extinct varanoid necrosaurs oc-
curred with true varanids in the Late
Cretaceous of North America and Central
Asia (different genera appear to be in-
volved and the Asian forms have just re-
cently been described by Borsuk-
Bialynicka, 1982). In the Early Cenozoic
there are two derived forms in Europe
(Fig. 12). Necrosaurs may have origi-
nated in Asia, from varanoid ancestors,
and dispersed across the Bering connec-
tion together with the true varanids dur-
ing the Late Cretaceous, but it is not pos-
sible to falsify an origin in North America

or Asiamerica, perhaps from a common
stem with the helodermatids. The Eo-
cene European necrosaurs are derived
with respect to the North American Late
Cretaceous genus Parasaniwa, and prob-
ably reached Europe after its emergence
in the very early Cenozoic.

The helodermatids (Fig. 12), now iso-
lated in southwestern North America,
were once more widely distributed; they
occur in the Late Paleogene of Nebraska
and Colorado (Heloderma), and had un-
doubted members in the Paleogene of
France (Eurheloderma). Possible helo-
dermatids occur in the Late Cretaceous
of Wyoming (Paraderma), and if this
record is correctly referred, the group
had a long North American history.
Renous (1979) supposed that heloderma-
tids had a Gondwanan origin, since they

390

Advances in Herpetology and Evolutionary Biology

Figure 12. Suggested area of origin (stippled) and later dispersals (arrows) of primitive varanoids. Helodermatid
and necrosaur dispersal to Europe was in the Early Cenozoic. Map data as in Figure 3. Dates at end of arrows
indicate earliest pre-Pleistocene fossils of the group in that area. See text for further explanation.

have the derived nerve condition B that
she believes appeared in the southem
hemisphere. There is, however, no rea-
son to suggest an origin outside of Laur-
asia, and if the early evolution of
anguimorphs was also Laurasian, helo-
dermatids could have been derived from
a primitive varanoid stock, perhaps in
North America, during the Late Creta-
ceous. Presence of Eurheloderma in
France was probably a later dispersal
during peak tropicality in the Eocene, as
for varanids, gekkonids and agamids.
Among the anguimorphs, only the
varanids achieved a wide southern con-
tinent distribution so far as known. They
may have formed subgroups in Asia,
Southeast Asia, India and Australia, as
suggested for agamids, geckos, and
skinks (Fig. 13). Varanids have a high
level of activity compared with most

other lizards, and I believe them to have |
evolved originally in Laurasia as preda- —
tors on small mammals of the Late»

snakes, which was probably a comple-
mentary Gondwanan radiation and (at!
least so far as the colubroid and more de- ©

rived groups are concerned) primarily a

Cenozoic phenomenon. There would!

have been no problem for varanids to
reach Africa, which maintained connec-
tions of various sorts with Laurasia. Vara-
nids also spread beyond the Arctic Circle

in the Eocene, as did other lizard groups |

discussed above. Like agamids, they are °
found
America and Europe, and lingered in the
former region until the Oligocene and
until the Late Pleistocene in Europe. I

have chosen the Bering route for Early |
Cenozoic entrance of Saniwa into North)

|
|
|
|
|
|
|
|
|
.
|
)
|
|
|
|

in the Eocene of both North?!

Eoc.

FOssIL LIZARD DISTRIBUTION - Estes 391

Cret.

“primitive

Figure 13. Suggested area of origin (stippled) and later dispersals (arrows) of varanids. X = position of Lan-
thanotus. Varanid dispersal from Asia into North America and Europe was in the Early Cenozoic. Map data as in
Figure 3. Dates at end of arrows indicate earliest pre-Pleistocene fossils of the group in that area. See text for further

explanation.

America (Fig. 13), but the paucity of the
Cretaceous record in Europe and Paleo-
gene in Asia makes a clear-cut decision
difficult to make.

OWEN’S HYPOTHESIS AND THE POSITION
OF SOUTHEAST ASIA

The continental reconstructions of
Owen (1976, 1981, 1982) are of interest
here since they provide a much more
extensive and intimate connection be-
tween eastern Laurasia and Gondwana
during the Mesozoic, in contrast to the
reconstructions commonly used (e.g.,
Smith and Bryden, 1977) in which a large
Eastern Tethys seaway separates these
two regions widely. In Figure 14, I have
used Owen’s reconstruction for the Early
Cretaceous as a base map. In brief,

Owen's view is that this eastern Tethys is
a geometric artifact produced by assum-
ing an Earth of constant dimensions. He
suggested that the Early Mesozoic earth
was about 20% smaller in diameter than
at present, and that it has expanded ex-
ponentially throughout Phanerozoic
time. The assumption of an equal amount
of subduction for each area of oceanic
crust production by sea floor spreading is
unwarranted in this view. Owen has
given additional support for his hypothe-
sis in later papers (1981, 1982), and the
reviews by Sunderman and Brosche
(1978) and Scrutton (1978) are also of in-
terest in this regard. Beyond distortions
that may be the result of an expanding
Earth, the choice of a map base may fur-
ther distort this region. A case in point is
the Lambert equal-area projection used

392

ae

telids, xenosaurs

necrosaurs bavarisaurs
Vian xantusiids paramacellodids tid
anguids dorsetisaurs acertids

euposaurs

helodermatids

gymnophthalmids

5

ardeosaurs

Advances in Herpetology and Evolutionary Biology

4
Chamaeleonids 6 agamids eee
cordylids geckos
skinks
varanids

anids
ee 2

Figure 14. Suggested areas of origin of lizard families and suggested dispersals between areas, given on
continental reconstruction of Owen (1976) for Early Cretaceous time (120 m.y.b.p.). Family names occur approxi-
mately on presumed area of origin. Where names span more than one continent, origin is presumed to have been on
the conjoined area. Continents identified in Figure 2. Number of lines connecting regions indicates numbers of
families presumed to be involved in dispersals (as opposed to vicariances) during Jurassic, Cretaceous, and
Cenozoic time. The number code identifies the families involved as follows: 1) iguanids, teiids, scincids, anguids; 2)
iguanids, glyptosaurs, helodermatids, necrosaurs; 3) agamids, scincids, varanids; 4) agamids, gekkonids, scincids,
varanids; 5) gekkonids, scincids; 6) chamaeleonids, lacertids.

by Lillegraven et al. (1979). In this out-
standing summary of paleogeographic
data, the map distortion makes the actual
close faunal relationships of Southeast
Asia and Gondwana _ biogeographically
inexplicable. In Owen’s reconstruction
the enormous Tethys seaway in the
eastern regions—at least as an entity
floored by simatic (oceanic) crust—
disappears. Owen stated, however, that
this does not rule out a narrower epicon-
tinental Tethys in this region that could
have been an effective barrier to animal
movement in some cases. The absence of
a large Tethys ocean also reduces the
northward movement of India to about
1,600 km, thus reducing also the high rate

of movement usually predicted for this
region to a figure more like that of rates of

plate movement in other areas. In
Owen’s reconstruction greater India
remained close to Southeast Asia

throughout Mesozoic time, although its
principal movements were the same as in
other reconstructions: a counterclock-
wise rotation that eventually oriented its
eastern margin toward the north, and
continued northward movement that
finally brought its present-day northern
margin up against Asia in the Late
Cretaceous, when the first phase of
Himalayan orogeny began. Based on the
conventional reconstruction of India as a
free-floating plate in the developing

ES ee eee

)

Indian Ocean, Lillegraven et al. (1979:
293) stated that “significant barriers to
the exchange of land vertebrates from
any continent with India existed from the
Early Cretaceous through the remainder
of the Mesozoic.” The presence of
endemic genera of agamids, geckos,
lacertids, and skinks in India, however,
attests to an ancient connection, almost
certainly of Late Mesozoic age, with
Southeast Asia. Colbert (1975) demon-
strated that India had a fauna of cosmo-
politan dinosaurs during the Cretaceous,
which corroborates this view; see also
Hallam (1981). Based on the double
thickness of sialic (continental) crust
under the Himalayas and Tibetan Pla-
teau, Owen (1976) suggested that there
was originally a more extensive, conti-
nental portion of the Indo-Tibetan plate
forming a connection to Laurasia, permit-
ting at least a limited amount of exchange
to take place. The implications of this
connection for the Agamidae have been
examined by Moody (1980, see above)
and they probably apply to the geckos,
skinks, and varanids as well.

Movements northward of India and
Australia during the Cenozoic, associated
with extensive production of new sea
floor in the Wharton Basin west of Aus-
tralia, deformed Southeast Asia and
pressed it to the east, twisting it and
bringing Indonesia and the Celebes
against Australia-New Guinea (Owen,
1976; Crostella, 1977). This juncture is
Webers Line (Darlington, 1957), an
imaginary line of “faunal balance”
separating an essentially Oriental fauna
to the north from the Australian-New
Guinea fauna. Owen’s reconstruction
thus gives Weber’s Line a tectonic ra-
tionale.

The relevance of Owen’s work here
lies in the explanatory power it provides
for the distribution of agamids, geckos,
skinks, and varanids. In all these cases
the most primitive forms of these families
are not those found in the Australian re-
gion, but are in general Asian (Mertens,
1942; Kluge, 1967; Greer, 1970; Moody,

FOSSIL LIZARD DISTRIBUTION - Estes 393

1980), a situation suggesting that sub-
groups were formed by vicariance of
these areas and India. Cracraft (1975)
suggested that Antarctic routes might
account for the Australian and New Zea-
land distribution of some amphibian and
reptile groups, although lizards were not
among the ones he cited. The West
Antarctic connection may have been
available to (e.g.) sphenodontid reptiles
and leiopelmatid frogs to New Zealand
(Estes and Reig, 1973; Cracraft, 1975),
and the East Antarctic route to Australia
appears to have been open through the
Mesozoic (Owen, 1976; Tarling, 1980).
Relationships of the lizard groups men-
tioned above, evidence from fossils, and
inferences from recent distributions do
not indicate, however, that Antarctica
was ever occupied by lizards or used as a
migration route, although climatically (at
worst cool temperate) and physically it
seems to have been available. Perhaps
this simply indicates once again that
agamids, geckos, skinks, and varanids
arose more to the north, or towards
Laurasia, and did not have access to this
route; alternatively the temperatures may
simply have been marginal or prohibitive
for lizards.

The above discussion strongly suggests
that Owen’s continental reconstruction,
or some other arrangement that permits
closer contact of Southeast Asia and
Australia, is preferable zoogeographi-
cally to those of Smith and Bryden (1977)
or Lillegraven et al. (1979), which sepa-
rate these areas widely. Owen’s ar-
rangement in addition helps to explain
the Gondwanan biota in some Triassic
localities in Indonesia and eastern Asia,
provides a tectonic rationale for Weber’s
Line of faunal balance in the Holocene
fauna, and reduces the high relative
speed of the Indian plate required by
conventional reconstructions to one more
in accord with other estimates of plate
movement. The additional support for
this hypothesis (Owen, 1981, 1982)
makes it very difficult to falsify although
his expanding Earth model has not yet

394

been widely accepted; in any case any
alternative arrangement should explain
adequately the above factors in order to
be zoogeographically as well as geologi-
cally acceptable.

SUMMARY AND CONCLUSIONS

It seems clear from the above that
much of early lizard evolution took place
on an essentially Pangaean continental
arrangement, and that most of the crucial
early stages of lizard development oc-
curred during the Jurassic and Creta-
ceous. Some eolacertilian group, as yet
unknown, gave rise to generalized lacer-
tilians, perhaps as early as the Late
Triassic or Early Jurassic. This general-
ized group became widely distributed in
Pangaea, and a primary dichotomy took
place as Laurasia separated from Gond-
wana in the Middle Jurassic, the Gond-
wana region retaining the primitive
iguanians, the lLaurasian populations
forming a gekkotan—scincomorph—
anguimorph ancestral group, the sister
group of iguanians. This latter group
appears to have been fragmented by
epicontinental seas in the European
Jurassic as well as by separation of Africa
from North America in Western Gond-
wana. This led to the development of
Late Jurassic groups that already showed
the “basic patents” of some living fami-
lies. Aside from the known Late Jurassic
groups, it is probable that some of the
more primitive (particularly iguanian)
living familes were also present in
Gondwana, although not yet docu-
mented. In this view, the primitive
iguanids (at a morunasaurine level) al-
ready occupied southem parts, at least, of
both Africa and South America (western
Gondwana). The origin of the Iguanidae
seems to have been the result of plate
separation of Laurasia and Gondwana,
while the development of an ancestral
acrodont agamid-chamaeleonid ancestor
in northwestern Gondwana would have
been through intracontinental factors. In

Advances in Herpetology and Evolutionary Biology

the earliest Cretaceous, the initiation of
separation of South America and Africa
led to the divergence, in isolation, of the
South American morunasaurines and
other Neotropical iguanids. Loss of Afri-
can iguanids resulted in the present relict
distribution of oplurines on Madagascar.
Movement of iguanids from South
America into North America and Europe
was, as discussed above, essentially a
dispersal phenomenon across preexisting
barriers and probably did not occur
earlier than the latest Cretaceous or the
beginning of the Cenozoic.

Late Jurassic and Cretaceous seas
extending south along the east coast of
Africa, and the separation of Africa from
eastern Gondwana (India—Madagas-
car—Antarctica—Australia) that began in
the Early Cretaceous, caused division of
the ancestral acrodont population of
iguanians. This vicariance resulted in the
formation of the Agamidae in the South-
east Asian region and the Chamaeleoni-
dae from the remaining African popula-
tion.

In Laurasia, intracontinental epicon-
tinental seas fragmented the ancestral
group of non-iguanian lizards. The gek-
kotans became restricted to Asia as the
Obik Sea spread southward, perhaps as
early as the beginning of the Cretaceous
or the end of the Jurassic, resulting in the
evolution of gekkonids in Asia (or South-
east Asia), India, and Australia. Diversifi-
cation within the Gekkonidae probably
followed from the separation of the Aus-
tralia-Southeast Asia-India plates during
the Cretaceous. Thus gekkonids vicari-
ated by the action of intracontinental bar-
riers (Obik Sea) ‘but diversified through
plate separation, as in the scenario given
for agamids by Moody (1980). Since Aus-
tralia was bisected during the Early
Cretaceous by an enormous epicontinen-
tal sea, this may have played a part in the
evolution of pygopodids from the early
gekkonid stock.

Primitive scincomorphs must have
arisen from marginal southem Laurasian
populations of the ancestral stock of non-

iguanian lizards. Such primitive lacer-
toids could thus have deployed in North
America as teiids, while lacertids de-
veloped from the Old World populations,
separated by the Jurassic and Cretaceous
epicontinental seas that covered, vari-
ably, much of what is now Europe.
Origin of the gymnophthalmids from
teiids in South America was the result of
intracontinental factors as yet unknown.

Cordyloids and scincoids must have
diverged as a result of fragmentation of
the remaining primitive scincomorphan
groups in Laurasia by Late Jurassic epi-
continental seas. Cordyloids appear to
have evolved on the North America—
Greenland—Great Britain—Iberian Pe-
ninsula plate association while scincoids
may have been evolving on the other side
of the European Jurassic seas, in Asia.
Scincoids subsequently dispersed
through Laurasia and into Australia,
South America, and Africa. Scincid evo-
lution into subgroups may have followed
that of agamids and geckos by vicariance
in the Southeast Asian—Australian—In-
dian regions. Xantusiids and cordylids
resulted from restriction of widespread
primitive cordyloids to areas in North
America and Africa, probably through
separation by Cretaceous epicontinental
seas.

This fragmentation of the Laurasian
ancestral group of non-iguanian lizards
must have taken place within a relatively
short time (Middle to Late Jurassic). In
simplified form, I suggest that gekkon-
oids diverged to the north, lacertoids to
the east and west, followed by an east-
west dichotomy of cordyloids and scinc-
oids, following the cladistic relationships
given here (Fig. 1).

The origin of anguimorphs brings fur-
ther complexity. These are the most de-
rived of the major lizard groups, and
probably diverged from a scincomorph or
pre-scincomorph stem group after the lat-
ter had separated from a common ances-
tor with gekkotans. It is likely that the
divergence of xenosaurs, dorsetisaurs
and anguids was in the European-North

FOSSIL LIZARD DISTRIBUTION - Estes 395

American region, with intracontinental
factors rather than plate separation in-
volved. Necrosaurs and varanids evolved
in Asia, being separated from Europe by
the Late Jurassic European epicontinen-
tal seas and the Cretaceous Obik Sea.
Limited dispersal of primitive varanoids
into North America in the Late Creta-
ceous, over the Bering connection, re-
sulted in the North American evolution
of helodermatids; a dispersal path for
necrosaurids was thus also available.

It is interesting that in the present
model, there is a decrease in Gondwanan
influence in lizard evolution as more
derived groups are considered. The
Iguania thus appears wholly Gond-
wanan, gekkotans seem to have been
originally Laurasian, and among scinco-
morphs only the gymnophthalmids, the
pygopodids, and possibly the cordylids
may have had a Gondwanan origin.
There is no evidence for Gondwanan
origin of any anguimorph group.

Several phases in saurian evolution
may thus be proposed. The first is a
Triassic phase, in which a number of
different (perhaps unrelated) pre-lacer-
tilian groups exploited the insectivorous
habitus, with one group assuming herbi-
vorous habits. Only a few aspects of this
phase are known, the primitive pal-
iguanids in Africa, with normal body
type, followed by the gliding kuehne-
osaurs. After the Triassic, the eolacertil-
ians dwindled as the more progressive
lacertilians evolved.

After the climatic extremes of the
Triassic, a Jurassic phase of lacertilian
evolution resulted in the presence of all
living suborders of modern lacertilians
by the end of the Jurassic, indicating a
long previous history of which we have
no record. Since the continents up to the
Middle Jurassic were still closely asso-
ciated in a Pangaean arrangement, Early
Jurassic diversification must have been
primarily the result of vicariant separa-
tion of populations within a widespread
primitive lacertilian group by epiconti-
nental seas or other intracontinental fac-

396

tors. By the Middle Jurassic, however,
separation of Laurasia and Gondwana
seems to have been the catalyzing factor
in separating the Gondwana iguanians
from the more derived Laurasian lizard
groups, and if these iguanians were then
essentially at iguanid level it could be
said that that family arose by this vicari-
ance. At the end of the Jurassic, wide-
spread emergence must have permitted
dispersal of the groups earlier formed by
isolation in various regions of Laurasia.
Very likely this dispersal heightened the
competition among the groups already
formed, and led to the canalization of
various adaptive types within lizards.

In the third or Cretaceous phase of liz-
ard evolution the present family groups
evolved. Some of the more primitive liv-
ing families must have already been in
existence in the Late Jurassic although
no record of them has yet been found.
The Early Cretaceous was a time in
which vicariance by plate separation was
very important, and resulted in the for-
mation of some iguanid and agamid sub-
groups and the Chamaeleonidae, as
South America and Africa separated and
Africa itself diverged from eastern
Gondwana. Other than these relatively
clear plate movements, epicontinental
seas were extensive in the Cretaceous
and must have separated populations of
various lizard groups that resulted in the
evolution of other living families.

The Cenozoic may be thought of as a
fourth phase of lizard evolution, in which
expansion of ranges in the Paleogene oc-
curred, during periods of peak tropical-
ity, and progressive restriction in the
cooler and drier Neogene gradually for-
med the boundaries of the present distri-
butions. Perhaps as Darlington (1957) has
suggested, the rise of colubroid snakes,
which seems to have been mainly a
Cenozoic phenomenon, played a part in
the progressive reduction of diversity
and restriction of range seen in lizards.

As shown in Figure 14, there is no
major center of lizard evolution in South-
east Asia or even the Old World Tropics,

Advances in Herpetology and Evolutionary Biology

as was suggested by Darlington (1957) for
many vertebrate groups, although this
region was important in the evolution of
agamids, gekkotans, scincids, and vara-
nids. Such tropical areas were wide-
spread in the past (Lillegraven et al.,
1979), and origin of lizard families took

place in all parts of the world, except, so >

far as known, Antarctica. The widespread
tropical and temperate conditions of the
past permitted lizard dispersal during the
later Mesozoic and (particularly) Ceno-
zoic, where connections existed.

ACKNOWLEDGMENTS

The data that permitted this paper to

be written were gathered with the help of
many colleagues and friends too numer-
ous to mention here; a partial list is given
by Estes (1982). For help in specific ways
related to this paper I thank E. N. Amold,
Jose Bonaparte, Richard Etheridge,
Jacques Gauthier, Max Hecht, Scott
Moody, Michael Novacek, Hugh Owen,
William Presch, Kevin de Queiroz, and
Garth Underwood, although they do not
necessarily accept all of my conclusions.

LITERATURE CITED

AGER, D. 1975. The Jurassic world ocean. In K.
Finstad and R. Selley (eds.), Jurassic Northem
Sea Symposium Proceedings, Norwegian
Petrol. Soc., Stavanger.

ALBERCH, P., S. GOULD, G. OSTER, AND D. WAKE.
1979. Size and shape in ontogeny and phy-
logeny. Paleobiology, 5: 296-317.

BELLAIRS, A. D’A. 1970. The life of reptiles. London,
Weidenfeld and Nicholson, 2 vols., xi + 589

pp.
Bons, J., AND G. PasTEuR. 1977. Solution histo-
logique 4 un problem de taxonomie herpéto-
logique intérresant les rapports paléobio-
logiques de l’Amerique du Sud et de l'Afrique.
C. R. Acad. Sci. Paris, 284: 2547-2550.
BoRSUK-BIALYNICKA, M. 1982. Anguimorphan and
related lizards from the late Cretaceous of the
Gobi Desert, Central Asia. Paleont. Polonica,
in press.
Camp, C. 1923. Classification of the lizards. Bull.
Amer. Mus. Nat. Hist., 48: 289-481.
CARROLL, R. 1975. Permo-Triassic “lizards” from
the Karroo. Paleont. Afr., 18: 17-87.

___. 1977. The origin of lizards, pp. 359-396. In S.
Andrews, R. Miles and A. Walker (eds.), Prob-
lems in Vertebrate Evolution. Linn. Soc.
London, Symp. Ser. 4.

CASAMIQUELA, R. 1980. Considerations ecologiques
et zoogeographiques sur les Vertébrés de la
zone littorale de la mer du Maestrichtien dans
le Nord de la Patagonie. Mem. Soc. géol.
France, 139: 53-55.

COLBERT, N. 1973. Continental drift and the distri-
bution of fossil reptiles, pp. 395-412. In D.
Tarling and S. Runcorn (eds.), Implications of
continental drift to the earth sciences. New
York, Academic Press, NATO Adv. Study Inst.,
Vol. 1.

___. 1975. Mesozoic tetrapods and the northward
migration of India. J. Paleont. Soc. India (Orlov
Memorial Volume), 20: 138-145.

Cox, B. 1974. Vertebrate palaeodistributional pat-
terms and continental drift. J. Biogeog., 1: 75—
94.

___. 1980. An outline of the biogeography of the
Mesozoic world. Mem. Soc. geol. France, 139:
15-79.

CRACRAFT, J. 1973a. Continental drift, paleoclima-
tology and the evolution and biogeography of
birds. J. Zool., 169: 455-545.

___.. 1973b. Vertebrate evolution and biogeography
in the Old World tropics: Implications of con-
tinental drift and paleoclimatology, pp. 373-
393. In D. Tarling and S. Runcom (eds.), Im-
plications of continental drift for the earth
sciences. New York, Academic Press, NATO
Adv. Study Inst., Vol. 1.

—__. 1975. Mesozoic dispersal of terrestrial faunas
around the southem end of the world. 17th Int.
Zool. Congr. 1972, Monaco. Mem. Mus. Nat.
Hist. Nat. Paris, 88: 29-54.

CROSTELLA, A. 1977. Geosynclines and plate tec-
tonics in Banda Arcs, Eastern Indonesia. Bull.
Amer. Assoc. Petrol. Geol., 61: 2063-2081.

DARLINGTON, P. 1957. Zoogeography: the geo-
graphical distribution of animals. New York,
John Wiley and Sons, xi + 675 pp.

DELFAUD, J. 1980. Les formations sedimentaires
des bassins et plates-formes mesozoiques et
cenozoiques. In: A. Autran and J. Dercourt,
eds., Evolutions Geologiques de la France. 26°
Congr. geol. int. collog. C7, Mem. B.R.G.M.
France, 107: 260-265.

ELDREDGE, N. 1979. Cladism and common sense,
pp. 165-198. In: J. Cracraft and N. Eldredge,
eds., Phylogenetic analysis and paleontology.
New York, Columbia University Press.

EsTES, R. 1970. Origin of the Recent North Ameri-
can lower vertebrate fauna: an inquiry into the
fossil record. Forma et Functio, 3: 139-163.

——. 1982. Handbuch der Palaoherpetologie, vol.
10, Sauria terrestres. Gustav Fischer Verlag,
Stuttgart.

ESTES, R., AND J. HuTcHISON. 1980. Eocene lower
vertebrates from Ellesmere Island, Canadian

FOssIL LIZARD DISTRIBUTION : Estes 397

Arctic Archipelago. Palaeogeog., Palaeocli-
matol., Palaeocol., 30: 325-347.

EsTES, R., AND L. PRICE. 1973. Iguanid lizard from
the Upper Cretaceous of Brazil. Science, 180:
748-751.

Estes, R., AND O. REIG. 1973. The early fossil
record of frogs: a review of the evidence, pp.
11-63. In J. Vial (ed.), Evolutionary Biology of
the Anurans. Columbia, Univ. Missouri Press.

Evans, S. 1980. The skull of a new eosuchian rep-
tile from the Lower Jurassic of South Wales.
Zool. Journ. Linn. Soc., 70: 203-264.

FERRUSQUIA-VILLAFRANCA, I. 1978. Distribution of
Cenozoic vertebrate faunas in Middle America
and problems of migration between North and
South America. Univ. Nac. Auton. Mexico,
Inst. Geol. Bol., 101: 182-321.

GAUTHIER, J. 1980. Anniella (Sauria; Anguidae)
from the Miocene of California. PaleoBios, 31:
1=7.

___.. 1982. Fossil Xenosauridae and Anguidae from
the Lower Eocene Wasatch Formation, south
central Wyoming, and a revision of the An-
guidae. Univ. Wyoming Contr. Geol., in press.

GREER, A. 1970. A subfamilial classification of
scincid lizards. Bull. Mus. Comp. Zool., 139:
151-184.

___. 1974. The generic relationships of the scincid
lizard genus Leiolopisma and its relatives.
Austr. J. Sci. (Suppl. Ser.), 31: 1-67.

____. 1977. The systematics and evolutionary rela-
tionships of the scincid lizard genus Ly-
gosoma. J. nat. Hist., 11: 515-540.

HALLAM, A. 1975. Jurassic environments. Cam-
bridge Earth Sci. Ser.; Cambridge Univ. Press,
ix + 269 pp.

____. 1981. Biogeographic relations between the
northem and southem continents during the
Mesozoic and Cenozoic. Geol. Rundschau, 70:
583-595.

HALLAM, A., AND B. SELLWoopD. 1976. Middle
Mesozoic sedimentation in relation to tec-
tonics in the British area. J. Geol., 84: 301-321.

HOFFSTETTER, R. 1962. Revue des récentes aquisi-
tions concemant l’histoire et la systématique
des squamates. Collog. Intern. Prob. Pal.
C.N.R.S., 104: 243-279.

____. 1964. Les Sauria du Jurassique supérieur et
specialement les Gekkota de Baviere et de
Mandchourie. Senck. Biol., 45: 281-324.

____. 1967. Coup d’oeil sur les sauriens (lacertili-
ens) des couches de Purbeck (Jurassique
supérieur d Angleterre: résumé dun
Mémoire). Colloq. Intern. Prob. Pal. C.N.R.S.,
163: 349-371.

HowartTH, M. 1981. Palaeogeography of the
Mesozoic, pp. 197-220. In Cocks (ed.), The
evolving Earth. British Mus. (Nat. Hist.) and
Cambridge University Press.

JULLIEN, R. AND S. RENOuS. 1972. Variations du
trajet du nerf ulnaire et de l’innervation des
muscles dorsaux de la jambe chez les lacertil-

398

iens. Bull. Mus. Nat. Hist. Nat. Paris, 29: 247—
252.

KLEMBARA, J. 1979. Neue funde der Gattungen
Ophisaurus und Anguis (Squamata, Reptilia)
aus dem Untermiozan Westbohmens (CSSR).
Vest. Ustred. Ust. geol., 54: 163-169.

KiuGE, A. 1967. Higher taxonomic categories of
gekkonid lizards and their evolution. Bull.
Amer. Mus. Nat. Hist., 135: 1-59.

LILLEGRAVEN, J., M. Kraus, AND T. Bown. 1979.
Paleogeography of the world of the Mesozoic.
In J. Lillegraven, Z. Kielan-Jaworowska and
W. Clemens (eds.), Mesozoic mammals: the
first two-thirds of mammalian history.
Berkeley, Univ. Calif. Press, 321 pp.

MCDOWELL, S. AND C. BOGERT. 1954. The system-
atic position of Lanthanotus and the affinities
of the anguinomorphan lizards. Bull. Amer.
Mus. Nat. Hist., 105: 1-142.

MERTENS, R. 1942. Die Familie der Warane (Var-
anidae). Abh. Senck. Nat. Ges., 462: 1-391;
465: 117-234; 466: 235-391.

Moopy, S. 1980. Phylogenetic and historical bio-
geographical relationships of the genera in the
Agamidae (Reptilia: Lacertilia). Ph.D. thesis,
Univ. Michigan, xv + 373 pp.

OweEN, H. 1976. Continental displacement and ex-
pansion of the earth during the Mesozoic and
Cenozoic. Phil. Trans. Roy. Soc. London, ser.
A, 281: 223-291.

___. 1981. Constant dimensions or an expanding
Earth?, pp. 179-192. In Cocks (ed.), The evolv-
ing Earth. British Mus. (Nat. Hist.) and Cam-
bridge University Press.

1982. Ocean floor spreading evidence of
global expansion. Tectonophysics, in press.

PRESCH, W. 1980. Evolutionary history of the South
American microteiid lizards (Teiidae: Gymno-
phthalminae). Copeia, 1980(1): 36-56.

PROTHERO, D., AND R. Estes. 1980. Late Jurassic
lizards from Como Bluff, Wyoming, and their
palaeobiogeographic significance. Nature,
286: 484-486.

RaGE, J.-C. 1978. Une connexion continentale entre

Amérique du Nord et Amérique du Sud au

Crétacé supérieur? L’ exemple des vertébrés

continentaux. C. R. Somm. Géol. France, 1978:

281-285.

1981. Les continents peri-atlantiques au
Cretace supérieur: migrations des faunes
continentales et problems _ paléogeo-
graphiques. Cret. Res., 2: 65-84.

RENOuS, S. 1978. Confrontation entre certaines
données morphologiques et la repartition
geographique des formes actuelles des
Sauriens: interpretation phylogénétique et
hypothése paléobiogeographique. Bull. Soc.
Zool., 103: 219-228.

___. 1979. Application des principes cladistiques a
la phylogénése et la biogéographie des Lacer-
tiliens. Gegenbaurs Morph. Jahrb., 125: 376-
432.

Advances in Herpetology and Evolutionary Biology

___. 1981. Development de l’aspect historique de
la biogéographie par la superposition de deux
theses: proposition d’une hypothése phylo-
génétique batie selon les principes Hen-
nigiens et theorie de la dérive des continents.
C. R. Soc. Biogeogr., 57: 81-102.

RIEPPEL, O. 1980a. The phylogeny of anguimorph
lizards. Denkschr. Schweiz. Naturf. Ges., 94:

1-86. .
___. 1980b. The postcranial skeleton of Lan-
thanotus borneensis (Reptilia, Lacertilia).

Amphibia-Reptilia, 1: 95-112.

___. 1980c. The trigeminal jaw adductor muscula-
ture of Tupinambis with comments on the
phylogenetic relationships of the Teiidae
(Reptilia, Lacertilia). Zool. Linnean Soc., 69:
1-29.

ROBINSON, P. 1967. The evolution of the Lacertilia.
Colloq. Intern. Prob. Pal., C.N.R.S., 163: 395—
407.

____. 1973. Paleoclimatology and continental drift,
pp. 451-476. In D. Tarling and S. Runcom
(eds.), Implications of continental drift to the
earth sciences. New York, Academic Press,
NATO Adv. Study Inst., Vol. 1.

RoMER, A. 1956. Osteology of the reptiles. Univ.
Chicago Press, xxi + 772 pp.

ROSEN, D. 1978. Vicariant patterns and historical
explanation in biogeography. Syst. Zool., 27:
159-188.

SAVAGE, J. 1968. The origins and history of the
Central American herpetofauna. Copeia, 1966:
719-766.

SCHMIDT-EFFING, R. 1980. Geodynamic history of
oceanic crust in southem central America.
Intem. Alfred-Wegener-Symp., Berliner
Geowiss. Abh., A19: 201-202.

SCRUTTON, C. 1978. Periodic growth features in
fossil organisms and the length of the day and
the month, 154-196. In P. Brosche and J.
Sundermann (eds.). Tidal friction and the
earth’s rotation. Berlin, Springer Verlag.

SMITH, A., AND J. BRYDEN. 1977. Mesozoic and
Cenozoic paleocontinental maps. Cambridge
Univ. Press, 70 pp.

SUNDERMANN, J., AND P. BROSCHE. 1978. Numerical
computation of tidal friction for present and
ancient oceans, pp. 125-144. In P. Brosche and
J. Sundermann (eds.), Tidal friction and the
earth’s rotation. Berlin, Springer Verlag.

TARLING, D. 1980. The geologic evolution of South
America with special reference to the last 200
million years, pp. 1-41. In R. Chiochon and A.
Chiarelli (eds.), Evolutionary Biology of New
World Monkeys and Continental Drift. New
York and London, Plenum Press.

UNDERWOOD, G. 1971. A modem appreciation of
Camp’s “Classification of the Lizards.” Intro-
duction to facsimile reprint, Soc. Stud.
Amphib. Rept., pp. vii-xvii.

Commentary on a Frog and Lizard Newly
Recorded from Central Durango, Mexico

ROGER CONANT!

ABSTRACT. Herpetological and hydrographic fea-
tures of the Rio Nazas, a stream of internal drainage
in Durango, Mexico, are discussed. The cliff frog,
Syrrhophus guttilatus, is newly reported from the
Nazas valley, and the earless lizard, Callisaurus
(Cophosaurus) texanus scitulus, apparently in-
vaded a new habitat created near the source of the
river as a result of a major flood that occurred in

1968.

INTRODUCTION

The Rio Nazas, one of the streams of
interior drainage of the altiplano of
northern Mexico, has long been of in-
terest to zoogeographers, largely because
its fauna includes a number of aquatic
and semiaquatic organisms that show
close affinities with those of the Rio
Grande which forms the southerm
boundary of Texas. Meek, as long ago as
1904, called attention to the similarities
exhibited by the fish faunas of the two
rivers, even though they are now sepa-
rated by broad areas of arid terrain. A
former connection must be assumed,
probably during pluvial periods of the
Pleistocene. I briefly reviewed the sub-
ject (Conant, 1963) and published a map
of the general area in a paper that was
chiefly concerned with natricine snakes
of the Rio Nazas and certain other still-
existing but disjunct streams that obvi-
ously were part of the old Nazas system.

While I was assembling data for that
study and a later one (Conant, 1969), I

1Department of Biology, University of New
Mexico, Albuquerque, New Mexico 87131, U.S.A.

made a number of visits (during 1960,
1961, 1962, and 1965) to the area at the
source of the Nazas near El Palmito,
Durango, during which a number of
amphibians and reptiles were obtained,
most of which were deposited in the
collection of the American Museum of
Natural History.

The Rio Nazas is formed by the union
of the Rio de Ramos and the Rio del Oro,
but its water is impounded at its source
by a large dam, the Presa Cardenas, and
is controlled by the opening or closing of
tunnels at the base of the structure. The
Nazas, which flows chiefly eastward
across the Chihuahuan Desert, formerly
emptied into a large interior basin in
southwestern Coahuila. Goldman (1951:
P]. 44) portrayed a sump in the now arid
basin where some of the runoff from the
river accumulated during earlier years.
The water of the Nazas is currently used
for agriculture and other purposes in the
Laguna District that includes the cities of
Torreon, Gomez Palacio, and Ciudad
Lerdo. (For information on the hy-
drography of the Nazas and other streams
of the region see Tamayo and West,
1964).

I was able to retum to the locality near
El Palmito on 20 and 21 July 1976 in
company with the late Dr. James D.
Anderson, Keith A. Hawthome, Salome
Litwin, and Anderson’s daughter, Susan,
as part of a field trip through the
Chihuahuan Desert as far south as central
Zacatecas. Many changes were noted in
the relatively broad floodplain of the
Nazas immediately below the Presa

400

Cardenas, which formerly had been oc-
cupied in large part by arid mesquite
scrub. That vegetation had largely disap-
peared but was being replaced by clumps
of miscellaneous growth. An exception
was the intermittent gallery forest of very
large Mexican cypresses (Taxodium)
along the edge of the river.

A concrete spillway, which had been
constructed near the village of E] Palmito
to serve as a relief for the main dam dur-
ing periods of high water, had vanished,
and the ravine below it had been steeply
eroded to a depth of 10 m or more as was
clearly indicated by a comparison with
photographs taken at the site about 15
years earlier. Water from the spillway
had evidently poured across the Nazas
floodplain, sweeping everything before it
except the larger rocks, the giant cy-
presses, and a few much smaller trees
along the perimeter or at other places
outside the main path of the torrent.
Presumably little else remained except
rocks, sand, and gravel after the water
subsided.

The cause of the flood was Hurricane
“Naomi that struck the Mazatlan area of
Sinaloa in September, 1968 with a deluge
of rainfall that spilled eastward over the
Sierra Madre Occidental and blanketed
the watershed feeding the Nazas. The
rainy cyclone period lasted from 8
through 13 September during which
period 213 mm of rain fell at El Salto and
99 mm at El Palmito-Presa Cardenas (Dr.
Robert H. Schmidt, Jr., in litt., through
Dr. Robert G. Webb). Schmidt (1976)
reported that the same storm resulted in
320 mm of rain at Siqueros, Sinaloa, in a
single 24-hour period.

During the almost eight years that had
passed between the time of the deluge
and our 1976 visit, part of the herpeto-
logical fauna had returned, although we
saw no snakes. Turtles (Kinosternon hir-
tipes murrayi and Chrysemys scripta
ssp.) were abundant in the river and were
also present in several small ponds in
rocky areas below the former spillway
that had been scoured out by the rush of

Advances in Herpetology and Evolutionary Biology

water and its burden of debris. A large
female of the latter species measured 308
mm in carapace length. Two species of
spiny lizards (Sceloporus) were observed
and two S. undulatus consobrinus were
collected. Hyla arenicolor was heard
calling and specimens of Bufo punctatus
and members of the Rana pipiens com-
plex were seen. Two taxa of unusual
interest, Syrrhophus guttilatus and
Callisaurus (Cophosaurus) texanus
scitulus, both of which are discussed in
detail below, also were found.

Syrrhophus guttilatus (Cope)

Five of these frogs (AMNH 107551-5)
were obtained, three during the evening
of 20 July 1976 and two the following
evening. Anderson heard a large number
calling from a small cliff face near the
northeastern edge of the Nazas flood-
plain and from talus slides and rock piles
at the base of the cliff. He described
three distinct calls in his field notes as
follows: “A short chirp of low intensity; a
louder, more distinct and more metallic
chirp; and a soft trill very much like the
encounter call of Hyla crucifer.” A tape
was made of the calls of three different
frogs, but because their brief, high-
pitched sounds are difficult to record
well (and to graph), even under the best
conditions, only one (from AMNH
107551) was suitable for the production
of an audio-spectrogram (Fig. 1).

Some of the frogs responded to crude
imitations of their notes, but the majority
fell silent when approached. Most
seemed to be calling from crevices in the
cliff face or from under the piles of rocks.
Six were seen, one of which was wedged,
head upward, in a vertical opening,
between two adjacent rocks each about a
half meter high; the frog was about 3 cm
inward from the exposed face of the rocks
and about 6 cm above their base. One
was calling from a small burrow beneath
a rock, and another was heard among
small rocks at the base of a dead bush.

FROG AND LIZARD FROM DURANGO - Conant

2

4
TIME

Figure 1.

401

6 8 1.0

IN SECONDS

Audio-spectrogram of Syrrhophus guttilatus; same note displayed with 50 Hz (left) and 300 Hz (right)

filters. Based on a tape recording made 20 July 1976 (from AMNH 107551) at the edge of the Rio Nazas
floodplain near El Palmito, Durango, Mexico, temperature 17.8° C.

On 21 July there was a light rain off and
on during the late afternoon and early
evening, and many more Syrrhophus
were calling than on the previous night.
The cliff apparently was high enough to
escape inundation during the flood; at
least there were no high water marks on
it such as were visible on the cypresses
and some of the rocks along the edge of
the floodplain. I had not examined it dur-
ing my earlier visits to the area, nor had I
heard any chirps, chiefly because my
evenings were fully occupied searching
for natricine snakes along the river and in
the pools near it (Conant, 1969).

In Anderson’s field notes the frogs
were described in life as being, “Pale
greenish yellow with dark markings. The
eye seemed dark although up close it
appeared to have some light markings.”
My field notes were slightly different,
viz.: “Pale dorsal areas olive green. Dark
dorsal markings black or dark gray. Dor-
sal and concealed surfaces of hind legs
yellowish olive; dark markings match
back.”

Two of the specimens were sent to Dr.
John D. Lynch, an authority on lepto-
dactylid frogs, for a determination as to
species. His response is quoted, virtually

402

in full, as follows: “Both specimens are
adult males agreeing with S. guttilatus in
the morphology of the hands and feet,
size, proportions, and skin texture, but
differing in having a more acuminate
snout (compare S. guttilatus and S.
marnockii in my 1970 paper [Figure 9];
these resemble marnockii in snout
shape), more intense marbling of the
dorsum (rather than flecks and _ thin
vermiculations), and in having less dis-
tinct tympana. The tympana are large
enough for guttilatus (59.1% eye length
in the 26.7 mm specimen [AMNH
107554] and 55.2% in the 25.0 mm
specimen [AMNH 107553]). The colora-
tion in life as recorded on the specimen
tag [Conant notes] also is at variance with
what is known of S. guttilatus.

“The frogs either represent somewhat
variable S. guttilatus or another weakly
differentiated allopatric population. My
gut reaction is to call them S. guttilatus
and attribute the differences to geo-
graphic variation. This decision is based
on the lack of difference in hand and foot
morphology, the knowledge that tech-
niques of preservation and fluid concen-
trations affect distinctness of tym-
pana, ...and a sense that pattern differ-
ence is not significant enough to warrant
species description. The most vexing fea-
ture is the difference in the snout shape,
but on the whole it does not compel me
to argue that the beast is new.”

The three specimens not seen by
Lynch agree in having more acuminate
snouts and a more intense dorsal mar-
bling by dark pigment (the latter particu-
larly noticeable in AMNH 107551) than
in the guttilatus depicted by Lynch (loc.
cit.). All five specimens have an inter-
orbital bar (slightly interrupted in
AMNH_ 107552) that is both much
broader and darker than shown in
Lynch’s figure.

The discovery of Syrrhophus gut-
tilatus at the source of the Rio Nazas in
central Durango consitutes a major range
extension both for the species and the
genus. According to Lynch, in his taxo-

Advances in Herpetology and Evolutionary Biology

nomic revision of Syrrhophus (1970: Fig.
10), the nearest localities for guttilatus
were in the Big Bend National Park,
Texas, extreme southeastern Coahuila,
extreme southwestern San Luis Potosi,
and Guanajuato. He has since com-
mented (in litt.) that every other speci-
men of guttilatus he has seen since 1969,
when his manuscript was submitted for
publication, has been from the area he
mapped for this frog. Lynch’s postulated
range for the genus Syrrhophus as a
whole (1970: Fig. 21), shows a large
blank space surrounding the Nazas River
locality.

It is likely that guttilatus will eventu-
ally be found at additional stations along
the Nazas or its principal tributaries. The
upper portions of the river valley abound
with rocky cliffs, some of them at the
edges of the stream where it passes
through gorges. Because of the constant
flow of water on a year round basis, the
humidity should be high enough in many
places to support such small anurans as
Syrrhophus.

Callisaurus (Cophosaurus) texanus scitulus
(Peters)

Four lizards of this taxon (AMNH
119225-8) were collected 21 July 1976 on
the floodplain of the Rio Nazas where
sand and gravel had replaced the many
dense clumps of mesquite and other
vegetation that were present prior to the
1968 immersion and scouring. Several
other scitulus were also seen actively
prowling during late morning of the same
day. Most permitted us to approach
closely, but they were watchful and wary.
When we moved too near they raced for
cover but seldom entered it, stopping
short instead and curling their tails and
waving them from side to side in
characteristic fashion. When we retreated
they promptly resumed prowling. Ander-
son recorded the temperatures of three
immediately after their capture at 36.8°,
39.2°, and 40.4° C.

FROG AND LIZARD FROM DURANGO - Conant

This lizard was not found during my

) previous trips to the area even though I

had walked repeatedly through virtually
all open parts of the floodplain during
favorable weather at various times in the
months of June through September, col-
lectively. Presumably scitulus entered
the newly created habitat after the 1968
flood, perhaps as soon as the spotty vege-
tation had grown large enough to provide
temporary shelter from the sun and such
predators as men, dogs, and birds. It was
almost certainly an invader, and in that
capacity it would be of interest to Dr.
Ernest E. Williams, whose studies on
such phenomena are exemplified by his
scholarly “The Ecology of Colonization
as Seen in the Zoogeography of Anoline
Lizards on Small Islands” (1969). It is
likely, however, that barring additional
severe inundations, the Nazas floodplain
eventually will revert, at least in part, toa
habitat less favorable for this species.

Among the specimens obtained are two
adult males (snout-vent lengths 54 and 59
mm) and two adult females (snout-vent
lengths 56 and 58 mm), all of which were
found on gravelly or rocky substrata.
These lizards agree meristically with
Peter's (1951) description of the south-
western race of texanus as follows: a)
ventral scales in midline, 78 to 82 (mean
81.0); b) dorsal scales along midline,
counted for a distance equal to the length
of the head measured from tip of snout to
posterior edge of interparietal, 36 to 39
(mean 37.8); and c) femoral pores (the
sum of the counts for both legs), 27 to 31
(mean 29.5).

No color notes were made in the field,
but the color patterns of the four Nazas
lizards were compared with six speci-
mens of scitulus from Chihuahua and six
from Grant County, New Mexico, in the
collection of the Museum of South-
western Biology of the Department of
Biology of the University of New Mexico.
The patterns were all basically similar.
The paired dark spots, one on each side
of the midline of the back, were less con-
spicuous on the Nazas specimens than on

403

those from the more northern localities,
but the dark, wavy crossbands on the tail
and the longitudinal, light-bordered dark
stripe on the rear of the thigh were about
equally intense in the entire sample from
the three areas. The dark ventrolateral
stripes did not penetrate quite so far
anterodorsally in the two Nazas males as
they did in the males from Chihuahua
and New Mexico. The venters in all were
comparable, both in regard to the pale
ground color and the intensity and posi-
tion of the dark pigmentation of the
males. Local demes of scitulus vary in
coloration and pattern and, with this fact
in mind, I could detect no significant dif-
ferences among the several population
samples.

The Nazas lizards represent a range
extension west of the localities plotted by
Peters (1951: Map 1) who listed geo-
graphical data for the 1,254 specimens of
the texanus complex then present in
museums in the United States. His
nearest localities were approximately 190
km downstream in the Nazas valley be-
tween Lerdo and La Goma and “14 miles
north of Pedricana” (sic. = Pedricena).

There are widely conflicting opinions
about the proper generic name for this
lizard. The principal ones, arranged in
chronological order, are summarized as
follows: 1) Axtell (1958) considered
texana to belong to the genus Cal-
lisaurus, and he therefore did not include
it in his proposed monographic revision
of the genus Holbrookia. 2) Earle (1961),
based on a study of the middle ear,
thought that texana represented an
“earless” Callisaurus. 3) Clarke (1965),
as a result of observations on its push-up
pattern, assigned texana to monotypic
status, resurrecting Troschel’s (1852)
name of Cophosaurus for it. 4) Guttman
(1970), using electrophoretic evidence
from hemoglobin, showed that all the
members of the group have a close rela-
tionship, and he doubted the validity of
Cophosaurus as a separate entity. Gutt-
man also cited papers by Carpenter
(1962, 1963) on the display action of two

404

species of Urosaurus and three of Uma
that indicated sufficient intrageneric
variation to suggest that the splitting off
of Cophosaurus on the basis of its be-
havior was untenable. 5) Cox and Tanner
(1977), in a study on the osteology and
myology of the head and neck regions,
advocated the retention of Callisaurus,
Holbrookia, and Uma as separate genera
and the return of Cophosaurus to the
synonymy of Holbrookia. 6) Adest (1978),
in his electrophoretic study, proposed
uniting all under the single genus Cal-
lisaurus.

Cox and Tanner (1977) may be con-
sulted for a detailed review of the tax-
onomic peregrinations of the group, the
so-called sand lizards, up to the time they
submitted their manuscript. Obviously
there is wide disagreement about the
generic classification, and the group is in
need of thorough study, preferably by an
investigator who can approach the prob-
lem objectively, consider all its facets,
and not be overly influenced by his own
personal methods of research, however
important they may be. In the meantime,
the approach I personally favor is to con-
sider Callisaurus, Cophosaurus, Hol-
brookia, and Uma as subgenera of the
genus Callisaurus Blainville (1835).
Such an arrangement would reflect most
of the diverse opinions as well as to pro-
vide zoologists in general, many of whom
are not especially familiar with these
iguanids, with a quick and easily envi-
sioned concept of their relationships.
Herpetologists who, with a few notable
exceptions, have resisted using the sub-
genus, might do well to emulate practi-
tioners of other disciplines who have
found the subgenus to be a useful and
convenient category. As Simpson re-
marked long ago (1961), the aversion to
subgenera is mostly psychological.

ACKNOWLEDGMENTS

The several persons who accompanied
me on the field trip through the

Advances in Herpetology and Evolutionary Biology

Chihuahuan Desert to Durango during
1976 were helpful in many ways, es-
pecially the late James D. Anderson
whose contagious enthusiasm and broad
knowledge of the amphibians and rep-
tiles of the region inspired us all. I am
grateful to John D. Lynch who provided
his opinion on the taxonomic status of the
Syrrhophus reported in this paper, and I
am indebted to Richard G. Zweifel for
preparing the audio-spectrogram of the
call of one of these frogs. John S. Apple-
garth, Ralph W. Axtell, Charles M.
Bogert, Charles J. Cole, William G.
Degenhardt, John D. Lynch, Samuel B.
McDowell, Jr., Hobart M. Smith, Robert
G. Webb, and Daniel C. Wilhoft under-
took a critical reading of the manuscript
or were helpful in a variety of other ways
or both. The Directors General de la
Fauna Silvestre, Departamento de Con-
servacion, in Mexico City very kindly
issued collecting permits to me over a
period of many years. My field work in
Mexico was supported in large part by
the National Science Foundation Grants
G-9040, G-22657, and GB-2177. Also I
must mention my deep obligation to my
late wife, Isabelle Hunt Conant, who
accompanied me and was my staunch
companion in the field during all of my
trips to Mexico from 1949 through 1967,
but who was too ill to participate with us
during 1976.

LITERATURE CITED

ADEST, G. A. 1978. The relations of the sand lizards
Uma, Callisaurus and Holbrookia (Sauria:
Iguanidae): an electrophoretic study. Ph.D.
dissertation, Univ. Califormia, Los Angeles,
131 pp. Diss. Abstr. Int. B 39: 36.

AXTELL, R. W. 1958. A monographic revision of the
iguanid genus Holbrookia. Diss. Abstr., 19:
1476-1477.

BLAINVILLE, H. M. D. DE. 1835. Description de
quelques especes de reptiles de la Californie
précédée de l’analyse d’un systeme général
d’herpétologie et d’amphibiologie. Nouv. Ann.
Mus. d’Hist. Nat. (Paris), 4:(3) 232-296.

CARPENTER, C. C. 1962. A comparison of the pat-
terns of display of Urosaurus, Uta, and
Streptosaurus. Herpetologica, 18: 145-152.

FROG AND LIZARD FROM DURANGO - Conant

___. 1963. Patterns of behavior in three forms of the
fringe-toed lizards (Uma-Iguanidae). Copeia,
1963: 406-412.

CLARKE, R. F. 1965. An ethological study of the
iguanid lizard genera Callisaurus, Coph-
osaurus, and Holbrookia. Emporia State Res.
Stud., 13: 1-66.

CONANT, R. 1963. Semiaquatic snakes of the genus

Thamnophis from the isolated drainage system

of the Rio Nazas and adjacent areas in Mexico.

Copeia, 1963: 473-499.

1969. A review of the water snakes of the
genus Natrix in Mexico. Bull. Amer. Mus. Nat.
Hist., 142: 1-140.

Cox, D. C., AND W. W. TANNER. 1977. Osteology and
myology of the head and neck regions of Cal-
lisaurus, Cophosaurus, Holbrookia, and Uma
(Reptilia: Iguanidae). Great Basin Nat., 37: 35-
56.

EARLE, A. M. 1961. The middle ear of Holbrookia
and Callisaurus. Copeia, 1961: 405-410.
GOLDMAN, E. A. 1951. Biological investigations in
México. Smithsonian Misc. Coll., 115: xiii +

A476 pp.

GuTTMaAN, S. I. 1970. An electrophoretic study of
the hemoglobins of the sand lizards, Cal-
lisaurus, Cophosaurus, Holbrookia, and Uma.
Comp. Biochem. Physiol., 34: 569-574.

405

LyNncH, J. D. 1970. A taxonomic revision of the
leptodactylid frog genus Syrrhophus Cope.
Univ. Kansas Publ., Mus. Nat. Hist., 20: 1-45.

MEEK, S. E. 1904. The fresh-water fishes of Mexico
north of the Isthmus of Tehuantepec. Field
Columbian Mus., Zool. Ser., 5: Ixiii + 252 pp.

PETERS, J. A. 1951. Studies on the lizard Holbrookia
texana (Troschel) with descriptions of two new
subspecies. Occ. Papers Mus. Zool., Univ.
Michigan, 537: 1-20.

SCHMIDT, R. H., JR. 1976. A geographical survey of
Sinaloa. Southwestern Stud., Texas Westem
Press, Mono. 50: 1-77.

Simpson, G. G. 1961. Principles of animal tax-
onomy. New York, Columbia Univ. Press, xiv
+ 247 pp.

Tamayo, J. L., AND R. C. West. 1964. The Hy-
drography of Middle America, pp. 84-121. R.
Wauchope and R. C. West (eds.), Handbook
Middle American Indians, Vol. 1. Austin,
Univ. Texas Press. 84-121.

TROSCHEL, F. H. 1852. Cophosaurus texanus, neue
eidechsengattung aus Texas. Archiv. Naturg.
(Berlin), 16: 388-394.

WILLIAMS, E. E. 1969. The ecology of colonization
as seen in the zoogeography of anoline lizards
on small islands. Quart. Rev. Biol., 44:
345-389.

List of Peruvian Anolis with
Distributional Data (Sauria: Iguanidae)

NELLY CARRILLO DE ESPINOZA!

ABSTRACT. Species of the iguanid lizard Anolis
known to occur in Peru are: A. boettgeri, A. bom-
biceps, A. chrysolepis scypheus, A. dissimilis, A. f.
fuscoauratus, A. laevis, A. ortoni, A. punctatus
boulengeri, A. p. punctatus (new record for Peru),
A. trachyderma, and A. transversalis. Distribu-
tional maps for these species in Peru are provided.

INTRODUCTION

This checklist of Peruvian species of
the iguanid lizard Anolis that I have
prepared is based primarily on data con-
tained in the works published by Wil-
liams, Vanzolini, Dixon and Soini (see
Literature Cited), as well as on the col-
lections of the Section of Herpetology of
the Museo de Historia Natural “Javier
Prado” of the Universidad Nacional
Mayor de San Marcos in Lima, Peru.
Maps have been compiled from the
available distributional data (Figs. 1, 2,
3);

It is my desire that this contribution
will prove useful to all students inter-
ested in the Peruvian herpetofauna. I
dedicate the paper to the distinguished
herpetologist, Dr. Ernest E. Williams, in
honor of his recent retirement as Curator
of Herpetology at the Museum of Com-
parative Zoology.

1Section Herpetology, Museo de Historia Natural
“Javier Prado’, Universidad Nacional Mayor de
San Marcos, Lima, Peru.

Abbreviations. MCZ, Museum of
Comparative Zoology; TCWC, Texas
Cooperative Wildlife Collection, Texas
A&M University; MHNJP, Museo de
Historia Natural “Javier Prado”; FMNH,
Field Museum of Natural History, Chi-
cago; ANSP, Academy of Natural Sci-
ence, Philadelphia; AMNH, American
Museum of Natural History.

CHECKLIST

Anolis boettgeri Boulenger 1911

Type Locality. Huancabamba, Peru,
1,000 m, restricted to Oxapampa, Peru by
Barbour (1934).

Material. AMNH: Huancabamba,
Piura; MCZ: Oxapampa, Pasco.

Distribution. Known only from Huan-
cabamba and Oxapampa of the Cis-
Andean region.

Anolis bombiceps Cope 1876

Type Locality. Nauta, Peru.

Material. MHNJP: Saur.0131, San
Regis, Rio Maranon, Pv. Loreto, Loreto;
Saur.0208, Santa Maria de Nieva, Pv.
Bagua, Amazonas; TCWC: Centro Union:
Mishana; Moropon (Dixon and Soini,
1975). Additional records: Estiron; Rio
Ampiyacu; Rio Itaya; Rio Nanay; San
Regis (Vanzolini and Williams, 1970).

Distribution. Amazonian region of
Peru, Ecuador, and Brazil plus the
Orinoco basin and the Guianas.

Anolis chrysolepis scypheus Dumeril and
Bibron 1837

Type Locality. French Guiana and
Surinam.

Material. TCWC: Centro Union (Dixon
and Soini, 1975); Pebas (Burt and Myers,
1942). Additional records: Barranca;
mouth of Rio Santiago; Rio Itaya; Iquitos;
Yarinacocha; Pedrera; Pampa Hermosa;
Rio Huallaga, Hudnuco (Vanzolini and
Williams, 1970).

Distribution. Amazonian region of
Colombia, Ecuador, Peru, and Brazil plus
the Orinoco basin, the Guianas and cen-
tral Brazil with adjacent Paraguay.

Anolis dissimilis Williams 1965

Type Locality. Itahuania, Upper Rio
Madre de Dios, Peru.

Material. FMNH: Itahuania, Upper Rio
Madre de Dios.

Distribution. Known only from the type
locality.

Anolis fuscoauratus fuscoauratus D’Orbigny
1837

Type Locality. “Chile,” corrected by
D’Orbigny (1847) to Rio Mamore be-
tween Loreto and Rio Sara, Bolivia, re-
stricted to Provincia Moxas, Bolivia, by
Bocourt (1873).

Material. MHNJP: Saur.0009, Chan-
chamayo, Pv. Tarma, Junin; Saur.0013,
Pucallpa, Pv. Cmel. Portillo, Ucayali;
Saur.0127, La Merced, Pv. Tarma, Junin;
Saur.0354, Bellavista, Pv. Jaén, Cajamar-
ca; Saur.0440, Yurinaqui Alto, Pv. Tarma,
Junin; Saur.0841—0870, Rio La Torre, Rio
Tambopata, Pv. Tambopata, Madre de
Dios; Saur.0899, Tingo Maria, Pv.
Huanuco, Hudnuco; TCWC: Centro
Union; 5 km NNE Iquitos; Mishana;
Moropon; Yanamono. Additional records:
Napo and upper Rio Maranon (Cope,
1868); Pebas (Cope, 1869); Nauta (Cope,
1876).

Distribution. Amazonian regions of
Ecuador, Bolivia, Peru, and Brazil, the

PERUVIAN ANOLIS - Espinoza 407

northern tier of NE Colombia and NW
Venezuela, the Orinoco and the Guianas,
and the Brazilian coastal forest.

Anolis laevis (Cope 1876)

Type Locality. Between Moyobamba
and Puerto Balsa, Rio Huallaga, eastern
Peru.

Material. ANSP: Between Moyobamba
and Puerto Balsa, Rio Huallaga, eastern
Peru.

Distribution. Known only from the type
locality.

Anolis ortoni Cope 1868

Type Locality. Rio Napo or Upper Rio
Maranon, Ecuador or Peru.

Material. MHNJP: Saur.0426, Pucall-
pa, Pv. Cmel. Portillo, Ucayali; TCWC:
Iquitos; Moropon; Centro Union; Mi-
shana. Additional records: Napo or Upper
Rio Maranon (Cope, 1868); Nauta (Cope,
1876); Pebas (Boulenger, 1885).

Distribution. Amazon and Orinoco
basins and the Guianas and the Brazilian
coastal forest.

Anolis punctatus punctatus Daudin 1802

Type Locality. “South America.”

Material. MHNJP: Saur.0201, Pucall-
pa, Pv. Cmel. Portillo, Ucayali; Saur.
0898, Tingo Maria, Pv. Hudnuco, Hudan-
uco (first records from Peru).

Distribution. Amazonian Brazil and the
Orinoco basin plus the Pucallpa and
Tingo Maria regions of Amazonian Peru.

Anolis punctatus boulengeri O’Shaugh-
nessy 1881

Type Locality. Canelos, Ecuador.

Material. CWC: Centro’ Union;
Mishana; Moropon. Additional record:
“Pebas” (Cope, 1871).

Distribution. Amazonian regions of
Peru, Ecuador, Colombia, and Brazil, the
Orinoco and the Guianas and the Bra-
zilian coastal forest.

408

Anolis trachyderma Cope 1876

Type Locality. Nauta, Peru.

Material. TCWC: Centro Union; 5 km
NNE Iquitos; Indiana; Mishana; Moro-
pon; Rio Momon. Additional records:
Pebas (Cope, 1885); Iquitos (Williams and
Vanzolini, 1966).

Distribution. Known from Amazonian
Peru between Nauta and Pebas plus the
Orinoco basin and the Guianas.

Anolis transversalis Dumeril 1851

Type Locality. “South America,” re-
stricted to Sarayacu, Peru, by Williams
and Vanzolini (1966).

Material. TCWC: Centro Union;
Mishana; Moropon. Additional records:
Nauta (Cope, 1876); Pebas (Cope, 1885);
Achinamisa; Rio Huallaga; Bombo, Rio
Tapiche; Cashiboya; Rio Ucayali; Cerro
Azul; Iquitos; Pampa Hermosa, near
mouth of Rio Cushabatay; Rio Ucayali
valley; Sarayacu (Williams and Vanzolini,
1966).

Distribution. Western Amazonian re-
gion, from Loreto in Peru to the Rio Soli-
moes in Brazil, including Ecuador and
Colombia. In Ecuador it extends up the
Andean foothills for an indeterminate
distance.

Comment. The generalized statements
on distribution of these species outside
Peru are based on works by Peters and
Donoso-Barros (1970) and Williams
(1965, 1974, 1976) and are not necessarily
supported by documented specimens.

LITERATURE CITED

BarBour, T. 1934. The anoles II. The mainland
species from Mexico southward. Bull. Mus.
Comp. Zool., 77: 121-155.

BocourTt, M. F. 1873. Mission scientifique au
Mexique. Rept. Livr. 1.

BOULENGER, G. A. 1885. Catalogue of the Lizards in

Advances in Herpetology and Evolutionary Biology

the British Museum (Natural History), Lon-
don, 2: 11-95.

Burt, C. E., AND G. S. MYERS. 1942. Neotropical
lizards in the collection of the Natural History
Museum of Stanford University Stanford Univ.
Publ., Univ., Ser., Biol. Sci., 8(2): 1-52.

Core, E. D. 1868. An examination of the Reptilia
and Batrachia obtained by the Orton Expedi-
tion to Ecuador and the Upper Amazon, with
notes on other species. Proc. Acad. Nat. Sci.
Philadelphia, pp. 96-119.

___. 1869. Seventh contribution to the herpetology
of Tropical America. Proc. Amer. Phil. Soc., 11:
147-169.

___. 1871. Ninth contribution to the herpetology of
Tropical America. Proc. Acad. Nat. Sci. Phila-
delphia, pp. 200-224.

___.. 1876. Report on the reptiles brought by Pro-
fessor James Orton from the middle and upper
Amazon, and westem Peru. J. Acad. Nat. Sci.
Philadelphia, 8(2): 159-188.

____. 1885. Catalogue of the species of batrachians
and reptiles contained in a collection made at
Pebas, Upper Amazon, by John Hauxwell.
Proc. Amer. Phil. Soc., 23: 94-103.

Dixon, J., AND P. SoInt. 1975. The Reptiles of the
Upper Amazon Basin, Iquitos Region, Peru. I.
Lizards and Amphisbaenians. Milwaukee
Public Museum, No. 4, pp. 1-58.

D‘OrsiGNy, A. 1847. Voyage dans l’Amerique
meridionale ... pendant les annees 1826...
1833. Tome cinquieme l.ere partie: Reptiles.
12 pp. 7, pls. Paris: P. Bertrand Strasbourg: V.
Levrault (1835-1847).

PETERS, J., AND R. DONOSO-BarROS. 1970. Catalogue
of the Neotropical Squamata. Part II. Lizards
and Amphisbaenians U.S. Nat. Mus., Bull. No.
297, pp: 43-69.

VANZOLINI, P., AND E. WILLIAMS. 1970. South
American Anoles: The Geographic. Differenti-
ation and Evolution of the Anolis chrysolepis
Species Group (Sauria: Iguanidae). Arq. Zool.
S. Paulo, 19: 1-298.

WILLIAMS, E. 1965. South American Anolis (Sauria,

Iguanidae): two new species of the Punctatus

Group. Breviora, Mus. Comp. Zool. No. 233,

pp. 1-15.

1974. South American Anolis: Three new
Species related to Anolis nigrolineatus and A.
dissimilis. Breviora Mus. Comp. Zool. No. 422,
pp. 1-15.

___. 1976. South American Anoles: The species
Groups. Pap. Avuls. Dept. Zool. Sao Paulo,
29(26): 259-268.

WILLIAMS, E., AND P. VANZOLINI. 1966. Studies on
South American Anoles. Anolis transversalis
A. Dumeril. Pap. Avuls., Dept. Zool. Sao Paulo
19, Art. 17: 197-204.

PERUVIAN ANOLIS - Espinoza 409

| oO
? }

A>
\ a cl ea 3
ie) i b , 9
4° TUMBES \ iE ctorero g
oh
PIURA Wat N\ a
er a
O 4 ivy :
aN ( ;AMAZONAS, A
H Sead : :
}LameayEQue\ 4 4°
‘CAJAMARCA.7
O pa, y “SAN
4,
“Fog A \MaRTN
* oe : ‘.
ile H
LA \
LIBERFAD\
i NS
ANCKSH
O é
4 id : ;
Re ene PASCO\ |
MADRE
DE DIOS s
12°
F
: )
PE " 2
é
PUNO
, LS
© A.boettger! AREQUIPA 4g
16°%

O) A. bombiceps

\|+ A.chrysolepis scypheus

Elk A.dissimilis ;
ESCALA 12000000 STACNA

Capital de Dpto. o 2

BS
One \

i paurcbe
of -

.
i cree 7 7 -

Figure 1. Distribution of Anolis in Peru.

410 Advances in Herpetology and Evolutionary Biology

2° fl

4°

|
2 iG
) oo
6° : sat \ AS : ee _S :
j * \ 0 72 {

i \, © 3
» CAJVAMARCA

RLIBERFAD\,

102)
MADRE
DE DIOS
12°
i
|
@ A.f. fuscoauratus Gis F
A A. laevis
&] A-ortonil ene '
Co = i
srs f
ESCALA: #2009000 an
°
|} Capital de Dpto. o \|

Figure 2. Distribution of Anolis in Peru.

PERUVIAN ANOLIS - Espinoza All

>
fe)

4 ee
: : Es a
i cue) AMAzgNAS :
6° # bo ( \ Sag" :
| L AMBAY E QQ "4Q
| | : £ \ a » B i j
| GAGA |
ee Ges a
ss rn 3
B° aL s

SS aS; RES ST aaa

ro)
°

‘,

oF ee !pASCO

ove enone

A. punctatus boulengert
A: punctatus punctatus
A- trachyderma
A. transversalis
ESCALA 1:2000000

Capital de Dpto. o

Figure 3. Distribution of Anolis in Peru.

Two General Problems Involved in
Systematics and Zoogeography of Bats

KARL F. KOOPMAN!

ABSTRACT. It is difficult to determine true sym-
patry (at time of mating) in species which are mi-
gratory and at the same time have an extended ges-
tation period. The basis for the ecological peculi-
arity of species which are confined to mountainous
regions, but not to high altitudes, is not readily ap-
parent.

INTRODUCTION

My first contact with bat taxonomy and
biogeography resulted from an identifica-
tion problem brought to my attention by
Ernest E. Williams, which eventually
resulted in a joint paper (Koopman and
Williams, 1951). My first field experience
with bats was on the same collecting trip
to Jamaica, in 1950, that was Emest’s first
field experience with Anolis. Since then,
I have studied the systematics and distri-
bution of bats in many parts of the world.
In the course of these studies, two gen-
eral problems have emerged which, to
my knowledge, have not been generally
recognized. I would like to discuss these
on a more or less theoretical level since I
think they have some general implica-
tions for evolutionary studies.

HOW DOES ONE DETERMINE SYM-
PATRY IN MIGRATORY BATS?

Bats, having the ability to fly, are able
to cope with the problem of annual fluc-

! Department of Mammalogy, American Museum
of Natural History, New York, New York 10024,
U.S.A.

tuations in climate (winter vs. summer,
wet vs. dry seasons) by changing their
ranges. In this they resemble migratory
birds, though as far as we know, no bats
migrate as far as some birds do. However,
unlike birds, bats are viviparous mam-
mals with a fairly long gestation period.
Moreover, a number of bats have various
mechanisms for sperm storage, delayed
implantation, or delayed development
(Orr, 1970; Wilson, 1979). This more or
less extended gestation period may mean
that some months will elapse between
mating and the birth of young. During
this period, the males, females, or both,
may fly hundreds of miles.

In birds, it is common to find two or
more subspecies occurring sympatrically
in the winter range. (Many examples may
be found in the American Omithologists’
Union’s Checklist of North American
Birds [1957]. A good example is found
among the subspecies of Sturnella
magna.) However, what is important is
that these subspecies are allopatric in the
breeding season. In birds, the range dur-
ing the breeding season is quite simple to
determine once nests are found. It can
then be safely inferred that mating took
place nearby since birds are oviparous
and sperm storage is unknown (W. E.
Lanyon, personal communication).
However, as indicated above, no such
inference may be drawn in the case of
bats if there is any significant migration.
Two taxa may be sympatric during part of
the year, allopatric during another part;
however, without direct evidence of

mating, it is impossible to determine
whether or not these two taxa have an
opportunity to interbreed. Obviously, the
presence of reproductive isolation, the
crux of the biological species concept,
can only be determined if there is an
unrealized opportunity to interbreed.

A few specific cases may point up the
problems. Lasiurus borealis borealis is a
common migratory bat in eastern North
America. In the southeastern United
States there is another form, seminolus,
differing chiefly in color. Originally
described as a subspecies of L. borealis,
seminolus was considered a distinct
species by Barkalow (1948) based on the
fact that pregnant females and females
with newborn young of both forms could
be found sympatrically in Alabama. Also,
they apparently both occur throughout
the year in Louisiana (see Lowery, 1974:
114-121). The taxonomic status of these
two forms is far from clear (see Koopman
et al., 1957), and it is quite possible that
they are separate species, but the point to
be made here is that sympatry at time of
parturition is not evidence of sympatry
during mating in migratory bats and
therefore is not relevant to the question
of whether or not separate species are
involved.

Tadarida brasiliensis is a widespread
New World bat some of whose northern-
most populations in the southem half of
the United States migrate and others do
not (see Cockrum, 1969; Spenrath and
LaVal, 1974, and references therein).
Some of the authorities cited have con-
sidered that more than one species is
involved, but the discussion is clouded
by uncertainty as to where mating actu-
ally takes place. To my knowledge, no
one has demonstrated two different
populations of the Tadarida brasiliensis
group mating in the same place at the
same time, and it is only this kind of evi-
dence that could establish that more than
one species is involved.

In the genus Leptonycteris, two spe-
cies are currently recognized in North
America (Davis and Carter, 1962; Baker

ZOOGEOGRAPHY OF BATS - Koopman 413

and Cockrum, 1966), nivalis, and a taxon
variously known as sanborni or yer-
babuenae. These two forms: were ori-
ginally considered as subspecies, and
they are allopatric at the northem ends of
their ranges (southwestern Texas and
southern Arizona plus southwestern New
Mexico respectively). Each occurs at its
northern end in summer (where the
young may be born) but is confined to
more southern regions, chiefly in Mexi-
co, during the winter. In Mexico, their
winter ranges certainly overlap, at least
macrogeographically, though judging by
published records, the two species have
not been collected at the same place at
the same time. The authors of both the
papers mentioned above concluded that
the macrogeographical sympatry, which
they documented, was evidence that
species rather than subspecies were in-
volved. I am inclined to agree with their
taxonomic conclusions, but would point
out that no one, to my knowledge, has
demonstrated that nivalis and sanborni
(or yerbabuenae) mate at the same time
and place and that therefore no true bio-
logical sympatry has been demonstrated.

The demonstration of true biological
sympatry between closely related taxa in
many bats will not be easy. Numerous
species living in areas with marked sea-
sonal fluctuations are known or sus-
pected to be migratory to a greater or
lesser degree. However, in most cases we
have no clear picture conceming the dis-
tribution of a species at different times of
the year. The best documentation we
have for any species of bat with an exten-
sive migration is that of Findley and
Jones (1964) for Lasiurus cinereus in
both North and South America. Cockrum
(1969) has summarized what is known
about migration (or its absence) in vari-
ous populations of one subspecies of
Tadarida brasiliensis. Tuttle (1976) has
summarized the movement patterns of a
species (Myotis grisescens) whose mi-
grations are limited to movements be-
tween summer roosts (including ma-
ternity colonies) and winter hibernacula.

414

Hoffmeister (1970) has recognized the
problem of summer vs. winter ranges of
temperate North American bats and
collected all data available to him for
these seasonal ranges in the state of Ari-
zona, There have been other studies as
well, but these are the principal ones.
Assuming that the ranges of all mi-
gratory species can be determined for
various times of the year, there is still the
problem of when and where mating
occurs. Most bats are small, nocturnal,
and difficult to observe under natural
conditions, so direct observation of mat-
ing is rarely possible, though this has
been done in some cases (see Thomas et
al., 1979). Where a particular species is
physiologically capable of mating only
during a limited part of the year, indirect
evidence such as presence of sperm in
the epididymes of males and of active
ovulation and fresh vaginal plugs in
females, may fix the time and place of
mating quite accurately. This kind of
evidence is unfortunately available for
very few species of migratory bats. The
determination of sympatry in a true bio-
logical sense will probably entail work
by many investigators operating in dif-
ferent places for a long period of time.

THE PROBLEM OF “PARA-MONTANE”
DISTRIBUTIONS

During the course of determining geo-
graphical and altitudinal ranges of Peru-
vian bats (Koopman, 1978), I found a
rather peculiar distributional pattern
emerging in several cases. Though some
75 species were widespread in the Ama-
zon basin, ten had essentially Andean
distributions, ranging considerable dis-
tances along this mountain chain but not
evidently extending far to the east of it.
Five of these (Sturnira erythromos, S.
bidens, Vampyrops vittatus, Vampyressa
melissa, Myotis keaysi) are not known
below 1,000 m (the semiarbitrary upper
limit for the lowlands) in Peru, but five
others (Lonchophylla robusta, Sturnira

Advances in Herpetology and Evolutionary Biology

bogotensis (=“ludovici’), Vampyrops
dorsalis, Artibeus harti, Myotis oxyotus),
although in most cases occurring in the
highlands, are also known below 1,000 m.
Even among the species apparently con-
fined to the highlands in Peru, two spe-
cies (V. vittatus and M. keaysi) are known
from lower elevations in Venezuela
(Handley, 1976). What keeps these five
(or perhaps seven) Andean lowland
species from extending their ranges
across the Amazon basin as so many other
eastern Peruvian lowland species have
done? One possibility might be reduced
rainfall farther east as the climatic influ-
ence of the Andes lessens. However,
judging by the rainfall map that Haffer
(1974: 17) gives, such is not the case. The
only other explanation I can think of is
that those species that are restricted to the
vicinity of the Andes require caves or rock
crevices for roosting and are not able to
use trees for this purpose as the species
that spread across the Amazon basin ob-
viously can. Unfortunately, most collec-
tions of these Andean species have been
made by netting and little or no data on
roosts are available. It is, of course, pos-
sible that some presently unknown factor
affecting forest composition and therefore
food or roosting sites might explain this
anomalous. distribution pattern. How-
ever, this seems unlikely since species
which feed on nectar, on fruit, and on
insects and belong to two quite distinct
families fall into this pattern. Another
point to be emphasized is that bats
actually live in two quite different
environments, that of the roost and that of
the foraging area. These may be separated
by many miles and require quite different
adaptations. Thus, different species may
have similar roosting habits but forage
under quite different conditions and vice
versa. Where a number of species occur in
the same general area, as in tropical
America, it is difficult to determine which
of these two general factors is more im-
portant in restricting numbers and dis-
tribution of a given species.

Unlike the problem of determining

true biological sympatry, that of para-
montane distributions is not, of course,
peculiar to bats. For any group of animals
for which reasonably good distributional
data are available, it would be interesting
to look for species with fairly extensive
distributions along a range (or ranges) of
mountains that occur in adjacent low-
lands but do not occur far from the moun-
tains involved. Explanations for this kind
of distribution could well be different in
different cases. However, it would cer-
tainly be interesting to see how wide-
spread this phenomenon is.

ACKNOWLEDGMENTS

I would like to thank Wesley E.
Lanyon of the Department of Or-
nithology, American Museum of Natural
History, for valuable information on
birds, and Guy Musser, my colleague in
the Department of Mammalogy, for criti-
cal reading of the manuscript.

LITERATURE CITED

AMERICAN ORNITHOLOGISTS’ UNION. 1957. Checklist
of North American Birds, Edition 5. Baltimore,
Maryland.

BAKER, R. J., AND E. L. CockruM. 1966. Geographic
and Ecological range of the long-nosed bats,
Leptonycteris. J. Mammal., 47: 329-331.

BARKALOW, F. S., Jr. 1948. The status of the Semi-
nole bat, Lasiurus seminolus (Rhoads). J.
Mammal., 29: 415-416.

CockruM, E. L. 1969. Migration of the guano bat,
Tadarida brasiliensis. Univ. Kansas Mus. Nat.
Hist. Misc. Publ., 51: 303-336.

Davis, W. B., AND D. C. CarTerR. 1962. Review of

ZOOGEOGRAPHY OF BATS - Koopman 415

the genus Leptonycteris. Proc. Biol. Soc.
Washington, 75: 193-198.

FINDLEY, J. S., AND C. JONES. 1964. Seasonal Dis-
tribution of the Hoary Bat. J. Mammal., 45: 461-
470.

HAFFER, J. 1974. Avian speciation in tropical South
America. Publ. Nuttall Omith. Club, 14: 1-390.

HANDLEY, C. O., Jr. 1976. Mammals of the Smith-
sonian Venezuelan Project. Brigham Young
Univ. Sci. Bull., Biol. Ser., 20(5): 1-91.

HOFFMEISTER, D. F. 1970. The Seasonal Distribu-
tion of Bats in Arizona: a case for improving
Mammalian Range Maps. Southwestern Nat.,
His JUL.

KoopMaN, K. F. 1978. Zoogeography of Peruvian
Bats with Special Emphasis on the Role of the
Andes. Amer. Mus. Novitates, 2651: 1-33.

KOOPMAN, K. F., M. K. HECHT, AND E. LEDECKy-
JANECEK. 1957. Notes on the Mammals of the
Bahamas with Special Reference to the bats. J.
Mammal., 38: 164-174.

KOOPMAN, K. F., AND E. E. WILLIAMS. 1951. Fossil
Chiroptera collected by H. E. Anthony in
Jamaica, 1919-1920. Amer. Mus. Novitates,
1519: 1-29.

Lowery, G. H. 1974. The Mammals of Louisiana
and its Adjacent Waters. Baton Rouge, Loui-
siana State University Press, xxiii + 565 pp.

Orr, R. T. 1970. Development: Prenatal and Post-
natal. Biology of Bats, Vol. 1, 217-231. New
York, Academic Press.

SPENRATH, C. A., AND R. K. Lava. 1974. An Eco-
logical Study of a Resident Population of
Tadarida brasiliensis in Eastern Texas. Occ.
Pap. Mus. Texas Tech. Univ., 21: 1-14.

THOMAS, D. W., M. B. FENTON, AND R. M. R. BARC-
LAY. 1979. Social Behavior of the Little Brown
Bat, Myotis lucifugus. I. Mating Behavior.
Behav. Ecol. Sociobiol., 6: 129-136.

TuTTLE, M. D. 1976. Population Ecology of the
Gray Bat (Myotis grisescens): Philopatry,
Timing and Patterns of Movement, Weight
Loss during Migration, and Seasonal Adaptive
Strategies. Occ. Pap. Mus. Nat. Hist. Univ.
Kans., 54: 1-38.

WILSON, D. E. 1979. Reproductive Patterns. Spec.
Publ. Mus. Texas Tech. Univ., 16: 317-378.

Modes of Reproduction of Philippine Anurans

WALTER C. BROWN!
ANGEL C. ALCALA?

ABSTRACT. Sufficiently complete data on charac-
ters of eggs, larvae, and egg deposition sites are
available for 28 of the 65 species of anurans known
from the Philippines to assign to them a specific
mode of reproduction. Less complete but adequate
data are available for 28 more species which enable
us to assign the General Mode of Reproduction
(either “aquatic” or “direct development”) with a
high degree of probability. Data on reproduction for
the remaining nine species are so limited that even
the General Mode must be considered as unknown.
The diversity of modes of reproduction exhibited
by Philippine anurans is discussed in terms of sys-
tematics. This diversity is compared to that of the
tropical anuran community of 81 species in the area
of Santa Cecilia, Ecuador as reported by Crump
(1974).

INTRODUCTION

As currently understood, the Philip-
pine anuran fauna is assigned to 65 spe-
cies distributed in six families and 19
genera. Until relatively recently little
was known about their life histories.
Taylor (1920) and Inger (1954) published
brief notes on eggs, larvae, or both for 14
species. Detailed life histories, however,
were known for only two of these,
Polypedates leucomystax (Villadolid and
Rosario, 1930) and Kaloula picta (Cen-
dana and Fermin, 1940).

Since 1954, field and laboratory studies
by the present authors have provided
detailed life history data on 25% of the

11070 Fremont Street, Menlo Park, California
94025, U.S.A.

2Silliman University, Dumaquete City, Philip-
pines.

species and data on eggs, egg deposition,
larvae, or both for most of the remaining
species. Only about half of this data has
been published (Alcala, 1955, 1962;
Alcala and Brown, 1956, 1957). It there-
fore seems appropriate at this time to
summarize the available information on
modes of reproduction of the Philippine
anurans, and to compare the diversity
with that exhibited by an anuran fauna of
about the same size in a similar, tropical
forest habitat. The anuran fauna of Santa
Cecilia, Ecuador, reported by Crump
(1974) best meets these criteria. The
evolution and adaptive value of amphi-
bian modes of reproduction are not con-
sidered in this paper. These topics have
been treated in some detail in a number
of early and recent papers (Lutz, 1947,
1948; Noble, 1927; Jameson, 1957;
Orton, 1949, 1957; Salthe and Duellman,
1973; Salthe and Mecham, 1974; Crump,
1974; Wassersug, 1975; Lamotte and
Lescure, 1977; Wake, 1978, 1980).

CLASSIFICATION OF REPRODUCTIVE

MODES

As Salthe and Mecham (1974: 380)
point out, the modes of reproduction of
anurans are so varied and their evolution
so poorly understood that no formal clas-
sification has been proposed, although
they regarded that of Jameson (1957) to
be the most complete up to that time.
Examination of the general papers cited
in the introduction illustrates this lack of
any consensus as to a system of classifica-

=e eee aeena eae eeeeteteeeteeneteeteetenataaaeeenee tas teatnaeenenaeeeeetzeseeneeeeeeeteteeete ote etot eee tt ee

MODES OF REPRODUCTION OF PHILIPPINE ANURANS - Brown and Alcala

tion. Salthe and Mecham themselves
(1974: 380) recognize two primary cate-
gories, “Aquatic” and “Terrestrial,” with
four subcategories and seven _ sub-
subcategories under “Aquatic” and two
subcategories and six sub-subcategories
under “Terrestrial.” Salthe and Duell-
man (1973: 243) list nine anuran modes
without subdivisions. Two of these in-
volve “direct” development. Crump
(1974) lists three primary categories
based on egg deposition and larval habi-
tat. She lists 10 specific modes exhibited
by the Santa Cecilian frogs with three or
four of the modes in each primary cate-
gory. Six of the modes have aquatic lar-
vae. Lamotte and Lescure (1977) classify
anuran reproductive modes in six pri-
mary categories and 18 modes. The pri-
mary categories are largely based on egg
deposition sites, larval habitats, and
degree of parental care.

In the following scheme we use the
two primary categories (General Modes)
of Salthe and Mecham but recognize a
greater number of Specific Modes (18).
The Specific Modes are based on size
and pigmentation of eggs, egg deposi-
tion, reduction or modification of larval
stages and time of retention in egg cap-
sules, larval nourishment, and type and
degree of care by parents.

GENERAL MODE I—HAVE AN AQUATIC
LARVAL STAGE

SPECIFIC MODE

1) Large to moderate number of
small, pigmented eggs deposited
in clusters in water. Larvae feed.

2) Moderate number of moderate-
sized, pigmented eggs deposited
in small groups attached to rocks
on stream bottom or sides. Larvae
feed.

3) Small to medium-sized eggs
deposited in foam nests in water.
Larvae feed.

4) Moderate number of small to
medium-sized, pigmented eggs

417

deposited on rocks, sand, moss,
leaves in close proximity to a
body of water to which larvae find
their way. Larvae feed.

5) Moderate to large number of
small to medium-sized, nonpig-
mented eggs deposited in foam
nests on objects overhanging
water to which advanced tadpoles
find their way. Larvae feed.

6) Small number of relatively large,
nonpigmented eggs attached to
wood above water in tree holes.
Larvae fall into water where they
develop and feed.

7) Small to moderate number of eggs
deposited in water-filled leaf
axils, tree holes, or other small
volumes of water. Larvae feed.

8) Small number of relatively large
eggs deposited in waterfilled leaf

axils, pools or other small
volumes of water. Larvae non-
feeding.

9) Eggs deposited in small water-
filled basins constructed in sand
or gravel. Larvae feed.

Eggs deposited on land. Larvae
carried to water. Larvae feed.
Small to moderate number of eggs
and larvae carried in some
manner by one of the parents
until larvae enter water. Larvae
feed.

Small to moderate number of eggs
and/or larvae carried in some
manner by one of the parents
until larvae enter water. Larvae
nonfeeding.

12)

GENERAL MODE II—LACK AN AQUATIC
LARVAL STAGE AND HAVE DIRECT

DEVELOPMENT
13) Small to moderate number of
large unpigmented eggs de-

posited in nest in terrestrial site.
Tadpoles hatch in nest but never
have an aquatic stage.

Small number of large unpig-
mented eggs deposited on land or

14)

418

in leaf axils of trees, shrubs,
epiphytic ferns. Froglets hatch
directly from eggs.

Small number of large eggs
carried externally in same way by
one or the other of the parents
until the froglets are released
fully metamorphosed.

Gastric brooding through meta-
morphosis of small number of
large eggs.

Small number of large eggs re-
tained within the oviducts of the
females, with ovoviparous de-
velopment.

Small number of small to moder-
ate-sized eggs retained within the
oviducts of the female, with
viviparous development.

15)

16)

17)

18)

RESULTS

REPRODUCTIVE MODES OF
PHILIPPINE ANURANS

We have fairly complete life history
data on 16 species of Philippine anurans,
part published (Alcala and Brown, 1955;
Alcala, 1962) and part in manuscript. We
also have sufficiently extensive data on
eggs and larvae of 12 additional species,
including three for which the information
is limited to populations from northem
Borneo (Inger, 1956, 1966), to assign all
28 to a specific Reproductive Mode as
recognized in the above classification.
We know that another 10 species (includ-
ing two Bornean populations reported by
Inger) have aquatic larvae but we do not
have data on the site of egg deposition
and therefore cannot determine the
Specific Mode. The information for these
38 species is summarized in Table 1.

Based on the habitat preference, clutch
number, size, and pigmentation of ap-
proximately mature eggs dissected from
gravid females as well as the general
mode of reproduction of close relatives,
we can reasonably assume the general
mode for 18 more species of Philippine

Advances in Herpetology and Evolutionary Biology

frogs. Thus we assume direct develop-

ment within the egg capsule for 11 addi-

tional species of Philautus, Oreophryne,
and Platymantis (Table 2). The eggs of

these species are very large, nonpig- |

mented, and few in number, very similar
to those of the species for which direct
development is well established. Also
the adults of these 11 species occupy
similar terrestrial
humid, forest habitats. Similarly, in terms

of characters of the eggs and habitats of ©
adults, we assume aquatic development |

for seven species of Pelophryne,

Rhacophorus, Rana, and Ooeidozyga. —
Thus, of the 65 species of Philippine ©
(Barbourula bu- |
Platy-

anurans only nine
suangensis, Rana _ micrixalus,
mantis cornutus, P. subterrestris, P.
polilloensis, Micrixalus mariae,
autus alticola, P. williamsi,

and do not include gravid females.

However, based on the uniformity of |

“mode” known or deduced for other
Philippine members of the genera Phil-
autus, Rana, Platymantis and Rhac-
ophorus, as well as the habitats of the
adults, we predict that seven of the nine

species listed above will eventually be |
shown to exhibit a mode which is con- |
sistent with that of known Philippine —

species of the same genera. We make no
prediction conceming B. busuangensis
or M. mariae.

When we consider only the 38 species
for which we have fairly complete evi-
dence on the mode of reproduction, 31
(81.6%) have aquatic larvae and seven

(18.4%) have direct development. When |

we consider the 65 known species, 39
(60.0%) very probably have aquatic
larvae, 17 (26.0%) direct development
and nine (14%) are unknown. The dis-
crepancy in the percentages in the two
comparisons is attributable to the much
greater difficulty in obtaining reasonably

(usually arboreal), |

Phil- ¥
and —
Rhacophorus hecticus) are completely —
lacking in evidence as to their general —
reproductive mode. With the exception ~
of B. busuangensis, these species are ©
known from only one or two specimens —

RE Aa at oe Sag Ye FS eS RED wy

oe

)

MODES OF REPRODUCTION OF PHILIPPINE ANURANS - Brown and Alcala

complete evidence of direct develop-
ment. It is much easier to find aquatic
larvae than it is to find eggs which are
deposited in hidden sites not in close
proximity to some body of water.

In terms of the two general modes,
comparable data is available for the new
world tropical anurans in the Santa
Cecilia area, Ecuador (Crump, 1974). Of
the 81 anurans recorded for that area,
Crump reported 61 (75%) as having
aquatic larvae, 17 (21%) as being direct
developers, and 3 (4%) as unknown.
These percentages for aquatic and direct
development are close to those we found
for the Philippine anurans.

At the level of the specific modes, I(1),
(2), etc., we find that we have sufficient
data to enable us to so classify 28 of the
65 species (Table 1). Six specific modes
are exhibited under aquatic development
and one under direct development. Of
the 28 species, 10 (35.7%) are type I(1);
two (7.1%) I(2); two (7.1%) 1(4); three
(10.7%) 1(5); one (3.5%), 1(6); one (3.5%)
(7); and two (7.1%) 1(8). Seven species
(25%) are known to have direct develop-
ment, all of type II(14).

Crump (1974) recognized 10 specific
modes for the Santa Cecilia frogs as
compared to seven for the Philippine
anurans. Based on Crump’s_ system,
modes three (eggs and larvae in a con-
structed basin), six (eggs carried by
parents, larvae aquatic), seven (eggs and
larvae in terrestrial nest), and nine and 10
(eggs and young carried in various ways
on dorsum of adults) are not represented
among the known Philippine anurans.
The correspondence of the modes of
Crump to specific modes in our system
and the proportions of species exhibiting
each of the specific modes are shown in
Table 3.

Even though the proportion of the
Philippine anurans for which the specific
mode is known is much less than that for
the Santa Cecilian frogs, the evidence
suggests similar trends in the diversity of
reproductive modes exhibited by these
two tropical anuran communities.

419

SYSTEMATICS AND MODE

Data on egg deposition sites, larval
habitat, and general reproductive mode
are summarized by family in Table 4.

Discoglossidae. Adults of Barbourula
busuangensis are completely aquatic.
However, the low number of large,
unpigmented eggs in the ovaries of
gravid females (Table 2) suggests a
specialized mode of reproduction, pos-
sibly lacking an aquatic larval stage.

Pelobatidae. Only a single species in
each of two genera, Leptobrachium and
Megophrys, occurs in the Philippines.
Both species have aquatic, feeding
larvae. These are found in ponds or slow
streams or in pools adjacent to them.
Based on the fact that both species have
eggs pigmented on one hemisphere, it is
probable that they are deposited in the
water or possibly in nearby locations
from which the larvae can reach the
water on hatching. Thus these two may
be any of three specific modes, I(1), I(2),
or I(3).

Bufonidae. Definitely five and prob-
ably all seven Philippine species have an
aquatic larval stage. Bufo marinus and B.
biporcatus use rain filled pools or slow
moving streams for egg deposition. Lar-
vae have also been collected from such
pools (Inger, 1966: 63). The specific
mode is I(1).

The larvae of Ansonia have been found
in rapid mountain streams (Inger, 1954:
240). The mature eggs removed from
gravid females are nonpigmented, but
the egg deposition site is unknown. The
specific mode is probably I(2) or I(4).

Details of the life history of the
diminutive bufonid Pelophryne brevipes
are in manuscript and verify Inger’s ob-
servation (1960: 416) that the larvae are
nonfeeding. The large, nonpigmented
eggs and the small clutch size (8-12) of
the other two Philippine species (P.
albotaeniata and P. lighti) suggest that
they may well have similar specific
reproductive modes. This is mode I(8).

Ranidae. Twelve and most probably all

420

Advances in Herpetology and Evolutionary Biology

TABLE |. PHILIPPINE ANURANS FOR WHICH GENERAL OR SPECIFIC REPRODUCTIVE MODE IS KNOWN.

Species

Megophrys
monticola

Leptobrachium
hasselti

Bufo biporcatus

Bufo marinus

Pelophryne
brevipes

Ansonia mulleri
Ansonia megregori
Ooeidozyga laevis
Rana cancrivora
Rana erythraea
Rana everetti

Rana leytensis

Rana limnocharis
Rana magna

Rana microdisca
parva

Rana nicobarensis

Rana sanguinea

Rana signata

Staurois natator
Platymantis
dorsalis

Platymantis
guentheri
Platymantis

hazelae

Polypedates
leucomystax

Polypedates
macrotis

Site of

egg
deposition

In water
In water
In water
in leaf

axils

In water
In water
In water
On rocks
in water
On leaves,
rocks etc.,
near water
In water
On moss,
rocks sand
near water

On rocks
in water

Soil crev-
ices, epi-
phytic
ferns
Epiphytic
ferns
Epiphytic
ferns, leaf

axils
Foam nest

Foam nest

Number of eggs

400-500

400-500
1000-2000
thousands

5-20
100-200
400-500
100-150

2000-2500
1200-1900

800-100
40-60
1000-2000
140-200
10-20
200-400

180+

about 300
200-300

5-9
150-225

300-400

Eggs with dark pigment

light
Xx
xX
Xx

mr

nx

nA AMM

~ Eggs without dark pigment

Peete

Direct development

Aquatic
larvae
©
70}
)
=
©
ea
g
ae
3 2
(0) Q,
o o
a
»«
XxX
xX
xX
XxX
XxX
XxX
xX
»«
XxX
xX
»«
XxX
xX
XxX
XxX
XxX
xX
x

Non-Feeding

Habitat
of adults

semi-terrestrial
semi-terrestrial

semi-terrestrial
semi-terrestrial

terrestrial-arboreal
semi-terrestrial
semi-terrestrial
aquatic

aquatic

aquatic
semi-aquatic
semi-aquatic
aquatic
semi-aquatic
semi-aquatic
semi-aquatic

semi-aquatic

semi-aquatic
semi-aquatic

terrestrial

arboreal

arboreal
semi-arboreal

semi-arboreal

MODES OF REPRODUCTION OF PHILIPPINE ANURANS

- Brown and Alcala 42]

TABLE 1. CONTINUED

Edwardtayloria On wall of
picta tree hole 8
above
water (one clutch)
Edwardtayloria Probably
spinosa as for E.
picta (?) 20+
Philautus Epiphytic
schmackeri ferns 6-15
Philautus Epiphytic
lissobrachius ferns and
leaf axils
of
Pandanus 7-30
Philautus sp.* Leaf
axils of
Pandanus 5-13
Rhacophorus
appendiculatus 300-400
Rhacophorus
pardalis Foam nest 44-50
Chaparina fusca In water
(small
pools) 50-100
Kalophrynus In water
pleurostigma (small
pools) 500-1000
Kaloula baleata In water several hundred
Kaloula conjuncta In water about 1000
Kaloula picta In water 500-1000
Kaloula rigida In water 500-1000
Oreophryne Under
annulata moss on
trees 3-9

xX
(gray) X 1(6) arboreal
x X I arboreal
(brown)
XX 1I(14) arboreal
xX X II(14) arboreal
xX X II(14) arboreal
I arboreal
1(5) arboreal
x XxX 1(7) semi-aquatic
xX X I(8) terrestrial
X XxX I(1) terrestrial-arboreal
X XxX I(1) terrestrial-arboreal
x XxX I(1) terrestrial
X Xx I(1) terrestrial
xX X 1I(14) arboreal

*This unidentified species also appears in Table 4 as one of the species for which larvae have not been identified.

16 species of the genus Rana have an
aquatic larval stage. The known Philip-
pine species exhibit three specific modes
of aquatic development. Rana leytensis
and R. magna exhibit mode I(3), R.
everetti and R. signata mode I(2), and R.
cancrivora and R. limnocharis mode I(1)
(Table 3).

Ooeidozyga laevis larvae have been
collected from temporary ponds and
mountain streams, and males and females
congregate around such ponds during the
Tainy season. Alcala (1962: 686) records
observing a pair in amplexus in a shallow
ditch. Although eggs have not been seen
in the turbid ponds (possibly because
they are deposited at the bottom), dis-

section of gravid females reveals a clutch
size of about 100 to 150 pigmented eggs.
The specific mode is I(1) or I(2).

Inger (1966: 244) assigned three
mountain stream larvae, two from Borneo
and one from Mindanao, to Staurois
natator. The deposition site of eggs has
not been observed, but eggs from two
gravid females are brownish pigmented
and number 200-300. The specific mode
is not known.

Definitely three (very probably four)
and probably all 11 species of the genus
Platymantis have direct development.
The fourth species indicated as very
probable is P. corrugatus. During field
work on Negros in 1970, egg capsules

422, Advances in Herpetology and Evolutionary Biology

TABLE 2. PHILIPPINE ANURANS FOR WHICH GENERAL REPRODUCTIVE MODE IS UNKNWON OR ONLY LISTED AS

PROBABLE.

Species

Barbourula busuangensis
Pelophryne albotaeniata
Pelophryne lighti
Platymantis cornutus
Platymantis corrugatus
Platymantis ingeri
Platymantis insulatus
Platymantis lawtoni
Platymantis laevigatus
Platymantis polilloensis
Platymantis subterrestris
Ooeidozyga diminutiva
Micrixalus mariae

Rana diwata

Rana melanomenta
Rana micrixalus

Rana woodworthi
Philautus acutirostris
Philautus alticola
Philautus aurifasciatus
Philautus emembranatus
Philautus leitensis
Philautus surdus
Philautus williamsi
Rhacophorus everetti
Rhacophorus hecticus

Rhacophorus zamboangensis

Oreophryne nana

Number of eggs*

probably! hundreds
unknown
200-300
16-20
unknown
12-15
25+
10=
10+

Eggs with dark pigment

ww KK

~ xxx = MX | Eggs without dark pigment

KA KM KRM OM

General reproductive mode (probable)

unknown

II
unknown
unknown

I
unknown

I

I
unknown
unknown

unknown
I
II

*Based on counts or approximations (over 100) of enlarged or mature ovarian or ovaducal eggs.

Habitat
of adults

aquatic
terrestrial-arboreal
terrestrial-arboreal
arboreal?
terrestrial
terrestrial-arboreal
terrestrial

arboreal
semi-terrestrial
arboreal

arboreal?

aquatic

unknown
semi-aquatic
semi-aquatic
semi-aquatic
semi-aquatic
terrestrial-arboreal
unknown
terrestrial-arboreal
arboreal
terrestrial-arboreal
arboreal

unknown

arboreal

unknown

arboreal

arboreal

tThis is a rough estimate of eggs in the ovaries of the types of these species. The ovaries were not removed from the types.

The eggs in the only female available are in an early stage of maturation. We are assuming that they will all mature at one time.

There were 15 large eggs in the ovaries and about 30 small to moderate. Whether or not all the latter will mature for the same clutch as the

large ones is not known.

with early embryos were collected from a
soil crevice. Though these were tenta-
tively identified as P. corrugatus, the
material was lost before this could be

verified. Eggs
documented species,
guentheri, and P. hazelae,) are deposited
in such sites as leaf axils of Pandanus,

of the
(P. dorsalis, FB

three well-

MODES OF REPRODUCTION OF PHILIPPINE ANURANS : Brown and Alcala 423
TABLE 3. COMPARISON OF SPECIFIC MODES OF PHILIPPINE AND SANTA CECILIAN FROGS.
a : e
oe o Ge
iS a rs)
~ 9 ~
: 2 z
= 2 :
= & 5 a
() 1S) (3)
5 2 S B
ie es & <=
= I fa]
5 of 3 5
& = = es
3 z : 5
= z=) g z)
(5) ie Lol
= 3 a 3
5 = H =
a = S) 5
n Z, ©) Z,
Small, pigmented eggs in water. Larvae
feed. I(1) 10 (36%) (1) 34 (44%)
Moderate sized pigmented eggs attached
to objects in water. Larvae feed. I(2) 2 (7%) (4) 14 (18%)
Moderate-sized, pigmented eggs on
leaves, mosses, etc. near water. Larvae
feed. 1(4) 3 (11%) (5) 6 (8%)
Relatively large eggs above water in tree
holes. Larvae feed. 1(6) 1 (3+%) not included 0 +
Moderate to large eggs in water in leaf
axils, tree holes, etc. Larvae feed. 1(7) 2 (7%) (2) 1 (14+%)
Relatively large eggs in water in leaf
axils, tree holes, etc. Larvae nonfeeding. 1(8) 1 (3+%) not included +
Eggs in water-filled basins in sand or
gravel. Larva feed. 1(9) 0 (3) 1 (1+%)
Eggs on land. Larvae carried to water by
parents. Larva feed. 1(10) 0 (6) 5 (6%)
Eggs and larvae carried by parents till
larvae enter water. Larvae feed. I(11) 0 (9) 1 (1+%)
Large, unpigmented eggs in terrestrial
nest. Tadpoles in nest but not aquatic. 11(13) 0 (7) 1 (1+%)
Large, unpigmented eggs in soil, leaf
axils, ferns, etc. Direct development. 11(14) 7 (25%) (8) 14 (18%)
Large eggs carried externally in same
way by one of the parents. Direct
development. 11(15) 0 (10 1 (1+%)

epiphytic ferns and soil crevices. Four of
the remaining seven species are known
to have a small clutch of large, unpig-
mented eggs and the adults of all 11 spe-

cies occupy similar terrestrial habitats.
An interesting morphological feature of
the directly developing larvae of the
known species is the use of expanded

Advances in Herpetology and Evolutionary Biology

424

Arerodurez JO S[Ixe fea, = yoo _ Assour : (eTqeqoid Gc) (eTqeqord gc)

ul soroods F ul soroeds % uo so1oeds & G GI € 63 oepiuey
snuepueg
jo s[ixe
jeg] Ul Ioyem
ul soloeds | snuepueg
sureo.s JO S[Ixe JeoyT
UuTe}yUNOUL peTTyaerem
ul soloeds % ul so1eds | BUIpsoj}
spuod spuod -uou ,(e,qeqoid
ul soroeds % ul soloeds Z G+) [ ‘Sulpse; F 0 L oepruojng
sjood
pue sureoas (t9yeM
uIe}UNOUL ur Ajqeqoid)
ul SsoIoeds & uMOUAU/) G 0 Gs sepyneqoledg
UMOUAUL) T T depIsso[soosiq
a ————————————————EE
Quowdojeasp (yuouldojaAsp (queuIdoj[eAsp (@PAIET IO oevAIeT onenby yuouldolsaoaq sa1oeds ATIUIe
onenbe) yelp) onenbe) $839 JO WIM Joquinyy polq surosiepun jo ‘ON
syeqIqey S839 OF $339 10f QOUSPIAV YO") Joaquin
jeAre’y SOUS SOzIS uMoUyU ()
uoyIsodeq uoyIsodeq

Seen ee — eee
_———— el

*SOOUA ANIddITIHG JO SLIGVH TVAUVI GNV ALIS NOILISOdAG SOY “P ATAV.L

425

- Brown and Alcala

MODES OF REPRODUCTION OF PHILIPPINE ANURANS

“syeyIqey
TE[IWIS BUIAdNDIO SaAQe[aI asojo Lay Aq payqryxe spour oy} se [[aM Se s}]Npe jo syeIIQey pure ‘sofeuloy plAvis 10; Joquinu 33a oy} UO ATLeEWLId paseq suoneodenxa ore santiqeqoid esoyy,

(sotqeqord tim 6¢) — (Setqeqord YAM gT)

LG 6 T€ L G9 sTRIOL,
so[ppnd
UIvI syung
Jo spuod 909 UO Ssoul IayeM (a[qeqoid T)
ul sotoeds g Ul sotoeds [ ul sotoeds 9 9 T 8 sBpI[AYOIOI
gjoy 90Q
UI IoyeM
aaoqe poom syueq
sjood 0} peyory ASSeIB IO
wieoys -ye soisods [ ‘10}eM I9A0
ul sotoads ¢ snuepueg seaq jo
sejoy 90Q JO SUI9q JO SOAvaT] UO
UI IayeM S[EX® feol] Ur SySOU UIBO} (2]qeqo1d Z) (e1qeqoid F)
ul so1oads Z = soloads F 10 ¢ ul soloeds ¢ € 9 € SI sepuioydooeyy
[IOS }StouL
IO S[Ixe Feo]
sureoys ul sotoeds Z Ioyem
ul sotoeds Q syuejd 19y}0 UI SoToads G
spuod juou pue snuep Io}eM Ieou

-euliod 10 -ueg JO SUIoy ‘syueq 10

426

abdominal sacs in respiration (Alcala,
1962). The specific mode is II(14) and is
probably exhibited by all 11 species.

Rhacophoridae. Seven, probably eight
species in three genera have aquatic lar-
vae. Polypedates leucomystax, P. macro-
tis, and Rhacophorus pardalis are known
to lay nonpigmented eggs in foam nests
on leaves of trees or shrubs or grasses at
the edge of forest streams or ponds. The
hatching larvae find their way to the
water where they complete develop-
ment. These are examples of Mode I(5).
Aquatic larvae, but not the site of egg
deposition, are also known for Rhaco-
phorus appendiculatus and R. zam-
boangensis. The lack of data on larvae
and the lower number of relatively large
unpigmented eggs in a gravid female of
Rhacophorus everetti leaves the general
mode of reproduction for this species in
some doubt, although we list it as aquatic
with a question mark (Table 2).

Edwardtayloria picta eggs have been
found attached singly to the sides of
water filled tree holes just above the
level of the water (Inger, 1966: 352).
Since larvae of the closely related E.
spinosa in southeastern Philippines have
been found in_ identical tree-hole
microhabitats, it is very probable that the
similar eggs (Table 3) are also deposited
on the walls. This is mode I(6).

Three species in the genus Philautus
have direct development, specific mode
II(14), and probably at least four other
species have the same mode based on
characters of eggs and habitats of adults
(Table 2). Detailed life histories of the
three species noted above are in manu-
script. Only P. alticola and P. williamsi,
based on single specimens, are without
even the indirect evidence of the type of
egg or clutch size. The eggs of the three
species reported at this time (Table 1) are
deposited in leaf axils of epiphytic ferns
or Pandanus. Like Platymantis in the
Ranidae, the larvae do not show any evi-
dence of external gills, oral disks, or
coiled guts. They differ from Platyman-
tis, however, in possessing highly vas-

Advances in Herpetology and Evolutionary Biology

cularized, broad tail fins which presum-
ably function in respiration (Alcala and
Brown, in manuscript).

Microhylidae. Six of the Philippine
species have aquatic larvae. We have
found eggs and larvae of Chaparina fusca
and Kalophrynus pleurostigma in small
temporary bodies of water in the forest
(see also Inger, 1956: 397, 1966: 135).
The larvae of Chaparina feed, mode I(7),
but those of Kalophrynus do not, mode
1(8).

Apparently all the Philippine species
of Kaloula deposit eggs in ponds or
similar bodies of water such as ditches.
Life histories of K. conjuncta and K. picta
have been well studied (Alcala, 1962:
715; Cendana and Fermin, 1940).
Kaloula rigida and K. baleata use small
temporary pools for breeding (Taylor,
1922: 177; Inger, 1966: 125). Alcala
(1962: 719) reported that larvae of K. con-
juncta feed on algae. This is mode I(1).

In the genus Oreophryne, one species
(O. annulata) is known to undergo direct
development (Inger, 1954: 447). The
highly vascularized tail is the respiratory
organ. Most probably the second Philip-
pine species O. nana has a similar repro-
ductive mode, since the habitat of the
adults and the mature eggs from a gravid
female (Table 2) are so similar. The
specific mode is II(14).

DISCUSSION

The evidence so far available for the
Philippine anurans indicates there is lit-
tle or no intrageneric diversity of mode of
reproduction except for the genus Rana.
The latter exhibits three specific modes.
Otherwise, evidence suggests that a
specific mode of reproduction is prob-
ably generically uniform and therefore
useful in classification. For example, all
known species of Rhacophorus and the
two species of Polypedates exhibit mode
1(5); all Philautus (Rhacophoridae) mode
11(14), with the larvae using a vascular-
ized tail for respiration; all Platymantis

MODES OF REPRODUCTION OF PHILIPPINE ANURANS : Brown and Alcala

(Ranidae) mode II(14), with the larvae
using abdominal sacs for respiration.
Also, in view of habitats of the adults and
characteristics of the eggs and larvae
(where the latter are known), it is very
probable that Oreophryne and Ed-
wardtayloria have uniform modes.

As shown in Table 4, aquatic larvae are
known for 31 species and are assumed to
be most probable for eight more species.
These are distributed in five of the six
families of Philippine anurans. Direct
development is known for seven species
(3 in Ranidae, 3 in Rhacophoridae, and 1
in Microhylidae) and is most probable for
11 more (5 each in Ranidae and
Rhacophoridae and 1 more in Micro-
hylidae), see Tables 1 and 2. The family
with unknown mode is the Discoglos-
sidae represented by the unique species
Barbourula busuangensis.

The Ranidae with 29 species exhibits
four specific modes. The Rhacophoridae
(18 species), Bufonidae (7 species), and
Microhylidae (8 species) each exhibits
three specific modes. The Discoglos-
sidae (1 species) and Pelobatidae (2
species) probably have only one specific
mode.

In the Santa Cecilian fauna 61 species
in seven of the eight families of anurans
are known to have an aquatic larval stage.
The tadpoles of one species in the
Leptodactylidae develop through
metamorphosis within a foam nest. Six-
teen species (14 Leptodactylidae, 1
Pipidae, and 1 Hylidae) undergo direct
development. The Hylidae with 37
species exhibits five specific modes, and
the Leptodactylidae (25 species) four

specific modes. The other families
(Pipidae, 1 species; Bufonidae, 4;
Dendrobatidae, 5; Centrolenidae, 3;

Microhylidae, 5; Ranidae, 1) each exhibit
only one specific mode.

In both areas the two families with the
greatest diversity of species exhibit the
greatest diversity of specific modes of
reproduction. Also, with the exception of
the Pipidae, direct development is ex-
hibited only by families with relatively

427

high species diversity. These families,
however, are different for the old and the
new worlds. It should also be noted that
direct development is limited to one
specific mode in the Philippine fauna but
three specific modes for the Santa Ce-
cilian fauna.

ACKNOWLEDGMENTS

We would like to thank Lawton Alcala,
Teodulo Luchavez, Crescencio Lumbod,
and Braulio Gargar for assistance in the
field. We also thank Marvalee H. Wake
and William Duellman for critical sug-
gestions on the manuscript and Robert F.
Inger for data on Bornean populations of
some of the Philippine species.

Our studies of Philippine amphibians
and reptiles were supported by Grants
from the National Science Foundation
(the most recent, GB 41947).

LITERATURE CITED

ALCALA, A. C. 1955. Observations on the life history
and ecology of Rana erythrae Schlegel on
Negros Island, Philippines. Silliman J., 2: 175-
192.

___. 1962. Breeding behavior and early develop-
ment of frogs of Negros, Philippine Islands.
Copeia, 1962: 679-726.

ALCALA, A. C., AND W. C. Brown. 1956. Early life
history of two Philippine frogs with notes on
deposition of eggs. Herpetologica, 12: 241-
246.

_____ AND . 1957. Discovery of the frog Cornufer
guentheri on Negros Island, Philippines, with
observations on its life history. Herpetologica,
13: 182-184.

CENDANA, S. M., AND D. V. FERMIN. 1940. Biology of
Kaloula picta (Dumeril and Bibron) with
special reference to its development and
breeding habits. Philip. Agriculturist, 28: 626—
655.

Crump, M. L. 1974. Reproductive strategies in a
tropical anuran community. Univ. of Kansas
Mus. Nat. Hist. Misc. Publ. 61: 1-68.

INGER, R. F. 1954. Systematica and zoogeography of
Philippines amphibia. Fieldiana, Zool., 33:
183-531.

____. 1956. Some amphibians from the lowlands of
north Bomeo. Fieldiana, Zool., 34: 389-424.

____. 1960a. Notes on Toads of the genus Pel-
ophryne, Fieldiana, Zool, 39: 415418.

428

____. 1960b. A review of the oriental toads of the
genus Ansonia Stoliczka. Fieldiana, Zool., 39:
473-503.

____. 1966. The systematics and zoogeography of
the amphibia of Bormeo. Fieldiana, Zool., 52:
14402.

JAMESON, D. L. 1957. Life history and Phylogeny.
Syst. Zool., 6: 75-80.

LAMOTTE, M., AND J. LESCURE. 1977. Tendances
adaptives a Jl’affranchissement du milieu
aquatique chez les amphibiens anoures. Terre
et la Vie, 2: 225-312.

Lutz, B. 1947. Trends toward non-aquatic and
direct development in frogs. Copeia, 1947:
242-252.

____. 1948. Ontogenetic evolution in frogs. Evolu-
tion, 2: 29-39.

NOBLE, G. K. 1927. The value of life history data in
the study of the evolution of amphibians. Ann.
N.Y. Acad. Sci., 39: 31-128.

OrTON, G. L. 1949. Larval development Necto-
phrynoides tornieri (Roux) with comments on
direct development in frogs. Ann. Carnegie
Mus., 31: 257-268.

____. 1957. The bearing of Larval evolution in some
problems in frog classification. Syst. Zool., 6:
79-86. .

SALTHE, S. N., AND W. E. DUELLMAN, 1973. Quanti-
tative constraints associated with reproductive
mode in anurans, pp. 229-249. In J. L. Vial
(ed.), Evolutionary Tiology of the Anurans.
Columbia, Missouri, Univ. Miss. Press.

Advances in Herpetology and Evolutionary Biology

SALTHE, S. N. AND J. S. MECHAM. 1974. Reproduc-
tive and courtship pattems in Physiology of the
Amphibia, B. Lofts, (ed.), New York, Academic
Press, 2: 309-521.

TAYLOR, A. C., AND J. J. KOLLROS. 1946. Stages in the
normal development of Rana pipiens larvae.
Anat. Rec., 94: 7-23.

TAYLOR, E. H. 1920. Philippine Amphibia. Philip. J.
Sci., 16: 213-359.

____. 1922. Additions to the herpetological fauna of

the Philippine Islands, I. Philip. J. Sci., 21:
161-206.

VILLADOLID, D. V., AND N. DEL Rosario. 1930.
Studies on the development and feeding
habits of Polypedates leucomyatax (Graven-
horst), with a consideration of the ecology of
the more common frogs of Los Banos and
vicinity. Philip. Agriculturist, 18: 475-503.

WakE, M. H. 1978. The reproductive biology of
Eleutherodactylus jasperi (Amphibia, Anura,

Leptodactylidae), with comments on the
evolution of live-bearing systems. J. Herpetol.,
12: 121-133.

1980. The reproductive biology of Necto-

phrynoids malcolmi (Amphibia: Bufonidae),
with comments on the evolution of reproduc-
tive modes in the genus Nectophrynoides.
Copeia, 1980: 193-209.
WASSERSUG, R. J. 1975. The adaptive significance of
the tadpole stage with comments on the main-
tenance of complex life cycles in anurans.
Amer. Zool., 15: 405-417.

Reproductive Data on Platemys platycephala
(Testudines: Chelidae) in Colombia

FEDERICO MEDEM !

ABSTRACT. Mating in Platemys platycephala
occurs mainly towards the end of the wet season.
Egg deposition is primarily in the early dry season.
A single enormous egg is laid. Incubation lasts
approximately 150 days. Hatchlings are large, and
subsequent growth appears to be slow. Maximum
adult size may be around 160 mm.

INTRODUCTION

Platemys platycephala (Schneider
1792) is a flat-shelled, small chelid turtle,
whose habitat comprises marshes, ponds,
and creeks within the tropical rain forest
of northern South America. Its range in
Colombia includes the Colombian
Amazon from Leticia to the Rio Atacuari,
the upper Putumayo region (Rio
Guamués), the upper Caquetd area
(Florencia, Rio Orteguaza), and the
middle course of the Rio Apaporis which
belongs to the Caquetd system. It has not
yet been recorded from the Rio Vaupés
drainage.

Almost nothing is known about the
reproductive parameters of P. platy-
cephala. According to information given
by natives, Medem (1960) stated that it
lays between one and six eggs. Dixon and
Soini (1977) reported that two females
killed for food by natives each contained
one well-developed egg. They also
reported that on July 31 copulation was
observed in shallow water of a small for-
est stream at Centro Union, near Iquitos,

1Estacion de Biologia Tropical “Roberto Franco,”
Apartado aereo 22-61, Villavicencio (Meta), Co-
lombia.

Peru. Espinoza (1970) and Moll (1979)
reported the clutch size to be four to six
and five to seven eggs respectively,
though without documentation. Ewert
(1979) stated that an average egg mea-
sures 51 X 28 mm and that the hatchling
carapace length averages 56.5 mm. No
other reproductive information is avail-
able on this species.

This paper now provides observations
on courtship and reproductive behavior,
as well as new data on clutch size, egg
size, incubation period, hatchling size,
and subsequent growth rate. All observa-
tions were carried out in outdoor en-
closures at the Instituto “Roberto
Franco” in Villavicencio, Colombia.

COURTSHIP AND COPULATION

Courtship takes place during the rainy
season (late March/April to November/
December), day or night, principally in
shallow water (5 cm) but also observed in
water up to 50 cm in depth. Copulation
occurs infrequently on the bank close to
the edge of the water. Courtship is stimu-
lated in captivity by running water, and
the animals react immediately if the tank
is cleaned by means of a hose. Reproduc-
tive behavior was observed on a total of
56 occasions. These occurred in the fol-
lowing frequency: in March, April, and
May once each month; June twice; July
seven times; August none; September
twice; October eight times; November
twenty-two times; and December twelve
times.

430

The male initiates courtship by per-
secuting the female until she is cornered,
at which time he touches her face with
the tip of his snout. He then climbs upon
her carapace from behind, folding his tail
under her shell. His forelegs continue to
scratch the female’s carapace dorso-
laterally, until he can secure a grip on the
sides of the carapace at the level of mar-
ginals 3 to 4. The male then hits and
pushes the female’s head until she is
forced to seek relief by withdrawing her
head within her shell, remaining motion-
less. In this position the male begins
trembling motions of the head and neck.
Copulation appears to take place at this
time, but the exact duration is difficult to
ascertain. The total time spent for court-
ship, mounting and copulation can last
from four or five minutes to four and a
half hours. The sequence can be inter-
rupted briefly by both sexes in order to
lift their heads out of the water to
breathe. Successful intromission was
specifically noted to occur on October 5
and 8 between the hours of 08:30-08:34
and 09:24—09:29.

EGGS AND HATCHLINGS

During egg deposition the female does
not excavate a nest, but rather makes a
shallow groove or lays her egg directly on
the ground. The nesting site is always to
be found under rotten leaves, and the egg
is partially covered by sand or earth or
not at all.

Between 1968 and 1975 ten episodes of
egg deposition were observed during the
months of August through February
(Table 1). On each occasion only a single,
hard-shelled, white oblong egg was laid.
In comparison with the size of the fe-
male, the egg is enormous, and one
wonders how it can pass the orifice of the
cloaca. A female with a carapace length of
148 mm was observed to lay an egg mea-
suring 57 X 28 mm, and an average egg
measures 54.8 X 27.6 mm and weighs
27.0 gm. It is possible that more than one

Advances in Herpetology and Evolutionary Biology

egg may not have sufficient space within
the reproductive tract. In addition, the
body cavity volume of this flat-shelled
turtle species is quite small, and a second
egg of the same proportions could prob-
ably not be accomodated within the
space available.

The ten eggs obtained were each incu-
bated in the laboratory under rotten
leaves at a temperature of 28 to 30°C,
which corresponds to the ambient
temperature. For unknown reasons, only
four of the eggs hatched. The six others
were found to be decomposed; two with
small dead embryos, and four without
embryos, possibly unfertilized. The in-
cubation period lasts from 124 to 177
days (average 150 days). The four viable
hatchlings had an average carapace
length of 47.8 mm and a weight of 16.2
gm (Table 1).

An egg caruncle was present in two
hatchlings only, and fell off after four and
35 days, respectively. The hatchlings
were first kept isolated in the laboratory
for about six months, and later put in a
large enclosure with hatchlings of other
species (Phrynops, | Rhinoclemmys,
Kinosternon spp.). One has survived for
five years, the other three died after one
or two years. Death was probably due to
environmental stress, which was also
observed in other chelid turtles (Phryn-
ops gibbus, P. nasutus) when maintained
under crowded conditions.

OTHER COMMENTS

Platemys platycephala apparently
grows very slowly (Fig. 1), and evidently
to a considerable age. A male (IRF 43—
carapace length 146 mm) and a female
(IRF 45—carapace length 148 mm) were
kept in captivity for 22 and 15 years re-
spectively. Bowler (1977) has recorded a
specimen still alive in captivity at the age
of nearly 20 years, and Mertens (1954)
discussed a 29 year old individual.

The maximum size attainable by P.
platycephala is unclear. Goeldi (1906)

REPRODUCTION OF PLATEMYS PLATYCEPHALA

- Medem 431

Figure 1.

20 30

Growth in captivity of 12 specimens of Platemys platycephala followed for one year or greater

(average length 5.8 years, longest 22 years). Positions of individual curves interpolated on the graph to mimic a
continuous growth curve. Carapace length (mm) plotted against elapsed time (years). Numbers on the graphs

represent catalogue numbers from the IRF collection.

stated that the largest specimens were
“never greater than 250 mm,” and Ewert
(1979) gave 200 mm as the size of a “large
adult.”’ The largest measured specimens,
however, are a 163 mm male (Niceforo
Maria, 1952) and a 161 mm male (Duell-
man, 1978). The largest individual ex-
amined in this series of 22 animals was a
male measuring 161 mm (Table 2).
Reproductively active females usually
measure from 140 to 155 mm carapace
length.

This species, known in Leticia as
“Matamata” (as are Chelus fimbriatus
and Phrynops nasutus), can be con-
sidered amphibious, since it is frequently
found walking on the ground far from
water. In captivity it can remain for

several days under rotten leaves, and
during the dry season it undergoes a
prolonged period of estivation even if
water is present. Both sexes are vora-
cious, eating daily, and aggressively bite
the necks, hind legs, and tails of cage
mates.

It should be noted that whereas each
clutch deposited by P. platycephala con-
tains but a single egg, the possibility of
multiple clutches being laid per season
cannot be excluded, especially in view of
the broad range of egg-laying dates.

ACKNOWLEDGMENTS

It gives me great pleasure to make a
contribution to the collection of essays in

Advances in Herpetology and Evolutionary Biology

432

‘ejeydaohje/d shwajeyqd jo Buijyoyey pue 665 “ge oinbi4

a as = rae G86 86 xX 9S OSE StI cL sny FI Or
cor LV vol GL AP FZ 0:96 8c x IG = =F GL url 0Z 6
OCI 1314 el pL Ady 61 TSG LG X GS = oS, €L AON L6G 8

a = = = Cle 9¢x0S Zé PST oh Ue GT L

= = = = 9 LG 86 x 9S = = GL AON 6G 9
881 €S LLY OL unf 8g 8°8C LG X 19 =a =n ig) PENG FIL S)

= a =e = £96 86 xX LS O00& Srl 69 99H eT v

= = aa Ser OLG LG X SS a a Gi} tat Ik €
e LT SP 9ST 69 idy 0Z LOE 6G X LS a a SOR ONG ean G

a aa = a C86 86 x €S — a= Soot I
(urs) (uiut) yBuUaTT (skep) BuTyoye Hy (UI) (uu) (us) (wut) YQSuUsT uontsodeaq ON 33Y

JYBIOM aordviey poised jo ayeq JUSIOM 9ZIS FUSION aordeiey jo ayeq
BuITyoyeY BUTT yo}eY uoljyeqnouy 334 834 a[eula J a][euls J

“WIVHddOXKLV Id SKNALVId NO VLVCO HAILLONGOYWdAY “T ATAVL

REPRODUCTION OF PLATEMYS PLATYCEPHALA « Medem

433

TABLE 2. SIZE OF PLATEMYS PLATYCEPHALA*,

Carapace Plastron
IRF No. Sex Length (mm) Length (mm)
653 m 161 144
238 m 161 —
769 f 158 143
44 ie 155 142
D000 m 152 131
239 it 150 =
764 m 149 131
45 f 148 134
690 m 148 128
43 m 146 130
350 i 141 =
42 m 138 =
809 f 134 124
810 m 125 LS}
366 f 115 101
43 m 110 —
814 fi 105 95
770 j 98 86
860 j 85 tal
815 j 79 2,
870 j 1 70
2 j 61 52
2) h 53 45
870 j 52 45
1 h 48 40
860 h AT 39
815 h 43 36

Width (mm) Height (mm) Weight (gm)
99 46 277
98 40 —_

106 46 392

103 43 350
97 43 325
96 39 —
93 4] 263

103 45 350
92 43 258
95 42 350
93 42 —
93 38 =
91 4] 263
83 36 222
83 33 150
75 35 —
74 30 129
63 26 82
63 24 68
61 22 76
57 21 46
45 17 2,
36 17 18.8
38 15 IB)
32 17 17.3
39 18 16.5
29 16 12.0

*Carapace and plastron length measured in midline. Width is maximum value. Height is maximum value along midline.

honor of Ernest E. Williams, as I consider
him one of the last of the great classical
herpetologists and anatomists, as well as
a good friend and mentor. Our associ-
ation, which began at the MCZ in 1962,
has always been intellectually stimulat-
ing and rewarding for me, and I owe
much of my knowledge of reptile
anatomy to him, especially as regards
turtles. I owe him a special debt of grati-
tude for the constant generosity he has
shown me in providing classic and recent
bibliographies without which I could
never have written some of my papers on
turtles and crocodilians.

I also wish to express my gratitude to
the following persons: Olga Castano
Mora, Biologist at the Instituto “Roberto
Franco,’ Alan Lieberman, Herpetologist

and Peace Corps Volunteer, for the prep-
aration of the Tables and the correction of
the text respectively, and to Anders and
Susan Rhodin for the editing of the
manuscript.

APPENDIX: SPECIMENS EXAMINED

IRF = Instituto “Roberto Franco,” Villavicencio,
Colombia. IRF 42-3, 653, 764, 769, 809-10, 814,
870, Leticia, Amazonas, Colombia; IRF 238,
Larandia, upper Rio Orteguaza, Caqueta, Colom-
bia; IRF 44-5, Santa Rosa de los Kofanes, Rio
Guamués, Putumayo, Colombia; IRF 555, Centro
Union, Ancaya, nr. Iquitos, Loreto, Peru; IRF 239,
350, Puerto Narino, Amazonas, Colombia; IRF 690,
Rio Caillaru, Loreto, Peru, ca. 40 km West Leticia,
Colombia; IRF 366, Xeruint, Rio Negro, Brazil; IRF
770, Uru-mutti creek, NE of Leticia, Amazonas,
Colombia; IRF 1-2, 815, 860, bom in captivity.

434

LITERATURE CITED

BowLeR, J. K. 1977. Longevity of reptiles and am-
phibians in North American collections. Soc.
Stud. Amph. Rept. Herp. Circ., 6: 1-32.

Dixon, J. R., AND P. SoINnI. 1977. The reptiles of the
upper Amazon basin, Iquitos region, Peru. II.
Crocodilians, turtles and snakes. Milwaukee
Publ. Mus. Contr. Biol. Geol., 12: 1-91.

DUELLMAN, W. E. 1978. The biology of an equa-
torial herpetofauna in Amazonian Ecuador.
Univ. Kansas Mus. Nat. Hist. Misc. Publ., 65:
1-352.

EspINozA, N. C. DE 1970. Contribucion al
conocimiento de los reptiles del Peru
(Squamata, Crocodylia, Testudinata: Reptilia).
Publ. Mus. Hist. Nat. “Javier Prado” Zool., 22:
1-64.

EWERT, M. A. 1979. The embryo and its egg: de-
velopment and natural history. In M. Harless,

Advances in Herpetology and Evolutionary Biology

and H. Morlock (eds.), Turtles: perspectives
and research. New York, Wiley, pp. 333-413.

GoELDI, E. A. 1906. Chelonios do Brazil (Jabotys—
Kagados—Tartarugas). Bol. Mus. Goeldi, 1906:
699-756.

MEDEM, F. 1960. Datos zoo-geograficos y eco-
logicos sobre los Crocodylia y Testudinata de
los rios Amazonas, Putumayo y Caqueta.
Caldasia 8(38): 341-351.

MERTENS, R. 1954. Bemerkenswerte Schildkroten
aus Sud- und Zentralamerika. Aquar. Terr.
Zeitsch., 7: 239-2492.

MOLL, E. O. 1979. Reproductive cycles and adapta-
tions. In M. Harless and H. Morlock (eds.),
Turtles: perspectives and research. New York,
Wiley, pp. 305-331.

NICEFORO MariA, H. 1952. Testudineos del sub-
orden Pleurodira en el Museo de La Salle. Bol.
Inst. La Salle, 39: 14-21.

The Prehensile-Tailed Skink
(Corucia zebrata) on Bougainville Island,

Papua New Guinea
FRED PARKER !

ABSTRACT. Little information has been recorded
on Corucia zebrata. Four years of part-time field
work and collecting on Bougainville Island has
yielded data on habitat, live color, behavior, feed-
ing, reproduction, predators and vernacular names.
Reports indicate that C. zebrata may undergo
periods of communal torpidity and coprophagy, but
this has yet to be verified.

INTRODUCTION

From 1960 to 1963 and for a part of
1966, the author collected reptiles and
amphibians on Bougainville Island,
which, together with Buka Island and
some offshore islets, comprises the North
Solomons Province of Papua New
Guinea. During that period, and through
the following years while the author con-
tinued collecting in other parts of Papua
New Guinea, Emest E. Williams provid-
ed all the assistance, information, and
copies of published material that a part-
time field worker could ask for. I am
grateful for his patience and persever-
ance, without which the value of my field
collections would have been considera-
bly less. The major part of my herpetolog-
ical collections from Papua New Guinea
and Australia are deposited in the
Museum of Comparative Zoology, in-
cluding the specimens referred to in this
paper.

1717 Ross River Road, Kirwan Qld 4814, Aus-
tralia.

Corucia zebrata is a large and unusual
skink (Fig. 1), but since its description by
Gray in 1885, very little has been re-
corded about the species in the wild.
Guppy (1887) reports that the species was
said to live in the foliage of higher trees;
Kinghorn (1928) repeats information from
the village people about feeding and be-
havior; Schmidt (1932) comments on the
suitability of the tail and claws for ar-
boreal life; and Loveridge (1946) notes
that a specimen was taken from the
stomach of a crocodile.

Some specimens have reached zoos,
and Honegger (1975) provides informa-
tion on captive breeding, food, and infec-
tions.

DISTRIBUTION AND HABITAT

On Bougainville, C. zebrata is fairly
common in the lowlands and foothills
below 300 m elevation. In the Guava
area, west of Kieta, it is reported by vil-
lagers to occur on the plateau at about
900 m. It is found in primary forest,
including swamp and _ littoral habitat
characterized by Casuarina spp. Pre-
ferred trees are those with dense foliage
and extensive epiphytic growth. Since
these are generally the oldest trees, C.
zebrata does not occur in secondary
forest where large trees have been re-
moved as part of traditional gardening
practices, or in areas selectively logged
for timber.

436

Advances in Herpetology and Evolutionary Biology

Figure 1. Corucia zebrata.

Villagers from Buka Island report that
Corucia is also to be found there, but it is
restricted to the undisturbed forests on
the low mountain ranges along the south-
west coast.

Corucia zebrata is also found on all the
larger islands in the Solomon Islands,
from Choiseul to San Cristobal (McCoy,
1980).

DESCRIPTION

Corucia zebrata is a large skink, with a
large and distinct head, small eyes, well-
developed limbs, and digits with long
curved claws. The tail is longer than the
body, cylindrical and prehensile, and
with a blunt tip.

Two gravid females in my series
measured 31.5 cm snout-vent length plus

35.5 cm tail length (MCZ 78292 [part])
and 28.5 and 29.7 cm (MCZ 78393).
Kinghom (1928) records a larger speci-
men 70.5 cm total length and Schmidt
(1932) one of 31.7 and 39.8 cm. The
snout-vent length comprises 47 and 49%
of the total length of my specimens, and
44.3% of Schmidt’s. New bom specimens
measuring 17.8 and 17.5 cm (50.4%)
(MCZ 78292 [part]) and 15.5 and 18.2 cm
(46%) (Honegger, 1975) extend the range
of relative body length, and there may be
geographical variation or sexual di-
morphism in this feature.

The normal color pattern for C. zebrata
on Bougainville is pale to medium olive
green on head and face with the dorsum
olive-brown to brown with narrow ir-
regular pale green to yellow-green bars,
usually discontinuous at the vertebral
line. The tail lacks pale bars but has

CORUCIA ZEBRATA ON BOUGAINVILLE -° Parker

scattered paler scales. The dorsal pattern
continues onto the flanks, where the pale
bars widen and merge into the uniform
pale olive or green ventral coloration.
The limbs may have pale bars dorsally
and laterally, or on the lateral surfaces
only.

A number of specimens have scattered
wholly or partly black scales on the post-
cephalic dorsal and lateral body surfaces
(see illustration in Parker, 1970).
Loveridge (1946), Burt and Burt (1932),
and Kinghom (1928) also refer to this
apparently common variation.

The iris is bright yellow. The pupil is
black and, when contracted, wider above
than below, with a small inferior ‘‘tail]’’
which points postero-ventrally. This
small “tail” is also present when the
pupil dilates.

Hatchlings (e.g., MCZ 78292 [part] and
72919) are distinct in having the dorsum
pale brown, almost translucent, with the
head and venter paler, longitudinal
stripes on the nape and transverse bars
on the dorsum. There is no green pig-
ment present at birth. I have not ob-
served the ontogenetic change to adult
color.

A mechanism for color change exists in
this species. On 11 July 1962, at about
midday, village people found two sub-
adult skinks (MCZ 72916-17) in dense
foliage at the extremity of a branch 15 m
above the ground in a large tree in pri-
mary swamp forest. The branch was cut
off and lowered by vines to the ground
where I noted: “. . . the color of the head
and the pale dorsal bars as well as ventral
surfaces is dull pale blue... the rest of
dorsum and tail are normal brown.” By
the following morning: “...the bluish
color is still present but blotches of
normal green are appearing on the heads
of both (skinks)—the blotches are irreg-
ular in shape and appear to to be expand-
ing from the center of each scale.” By the
evening of July 13, the heads, ventral
surfaces and dorsal bars of both lizards
were pale dull green. On the 15th, when
both lizards were feeding, the same areas

437

were ©... normal dull green but some of
the dorsal head scales are darker green in
the center, paler at the edges.”

This method of color change is similar
to that observed in captive sub-adult
green tree pythons, Chondropython viri-
dis, where the bright yellow, orange, or
red juveniles change to green in 2 to 3
days, through a mechanism where a
green speck first appears in the center of
each scale, then slowly expands to cover
the scale.

The pale transverse bars on the dorsum
of C. zebrata may also become indistinct.
An adult female in captivity (MCZ 78292
[part]) became uniformly dull brown
dorsally over a period of two months.
Honegger (1975) illustrates a strongly
marked specimen newly arrived at
Zurich Zoo and states the pale dorsal bars
faded in captivity.

BEHAVIOR

This skink is arboreal, nocturnal, and
according to village informants from a
number of different parts of Bougainville,
wholly vegetarian. By day it rests in an
upright position in dense foliage, grasp-
ing twigs and branches with limbs and
tail. It avoids sun and light and may shift
its position if sunlight does reach it.
McCoy (1980) states that it shelters in
hollows in the larger forest trees.

When active, its movements are slow
and deliberate, with the claws gripping
climbing surfaces firmly. The tail is
braced against trunks and larger
branches, and coiled around smaller
branches. Skinks sometimes hang sus-
pended by the tail while feeding or mov-
ing. The tongue is used constantly to
investigate the surroundings, and more
frequently if the skink is disturbed. Its
defensive behavior is very similar to that
of the larger species of Australian and
New Guinean Tiliqua—mouth wide
open, tongue extended, body com-
pressed to present maximum area to the
“intruder, and rapid inhaling and exhal-

438

ing causing loud hissing. The bite is very
powerful and prolonged. Hatchlings
display this behavior more readily than
adults.

By day these skinks are usually found
alone or in twos. Some villagers report
that they occasionally find three to five
skinks of different sizes and sexes to-
gether. Groups of two taken like this
were MCZ 72916-17, 93727-28, and
93729-30.

Village informants were consistent in
asserting that Corucia will not willingly
descend to the ground. During my many
nocturnal collecting trips, not once did I
find one of these skinks on the ground.
McCoy (1980), however, states that
“_..it sometimes forages on the ground
at night.” On Bougainville, villagers say
it will use aerial vines and branches in
contact for movement between trees.
They associate this skink with the pos-
sum, often finding the two animals in
close proximity in foliage and in hollows
in trees, by day and night. Kinghom
(1928) comments on this association.

Villagers report that this species eats
the leaves, flowers, and fruits of various
trees and epiphytes. Food is neatly
nipped off, fruits in portions and leaves
in small circular pieces, and each item
swallowed. Kinghorn (1928) notes the
stomach contents of one specimen as a
mass of circular leaf particles.

The preferred food items on Bougain-
ville are the leaves and fruit of an epi-
phytic Scindapsus vine with elongated
leaves, known as bakuna in the Nasioi
language in the Kieta area. The vine is
illustrated in Parker (1970). The ripe
fruits are red and sweet. The vine itself,
however, has a white milky juice “with
numerous needle-like crystals of calcium
oxalate which should irritate mucous
membranes’ (Womersley, in litt.). A
second common food plant is a wild pep-
per vine, Piper sp., known as konangko-
nang in the Nasioi language. Again,
leaves and fruit are eaten. Paradoxically,

Advances in Herpetology and Evolutionary Biology

the presence of the alkaloid piperine,
especially in the fruits, should make it
unpalatable (Womersley, in litt.). The
arboreal agamid lizard also found on
Bougainville, Gonocephalus godeffroyi,
feeds readily on this Piper.

In captivity on Bougainville, these
skinks remained vegetarian, rejecting
meat and a variety of insects even if left
without food for up to two weeks.
Domestic fruits—tomatoes, bananas (skin
and pulp) pawpaw, and pineapple—were
taken, but all were neglected if either of
the preferred food plants were placed in
the cage. Villagers report that Corucia is
not known to enter gardens in search of
food. This may be because subsistence
gardening as practiced on the island re- —
quires felling large forest trees and clear-
ing undergrowth. Dietary preference
may be another reason the skinks do not
enter gardens. Honegger (1975) reports
that the skinks at Zurich Zoo accepted
apple, shoots of Monstera sp., and
Scindapsus aureus (the preferred food of
all those offered), but rejected bananas,
peaches, vegetables, Tradescantia, Phil-
odendron scandens, and Codiaeum.
Honegger’s specimens are also said to
have taken cockroaches, maggots, meal-
worms and insects, but to have rejected
fish. Specimens in the San Diego Zoo
took insects and small mice and birds
(Anonymous, 1978).

The captive skinks on Bougainville
showed a preference for leaves covered
in water droplets, and would not take
water from containers in the cages. How-
ever, Honegger (1975) states that some of
his specimens took water from containers

at night.
Corucia defecates frequently and the
feces have a characteristic odor.

Honegger (1975) confirms this. Leaf fiber
and portions of undigested leaves are
visible in feces of newly captured speci- __
mens. Honegger (1975) observed speci-
mens eating feces and portions of skin
slough. He also noted that the feces of all

CORUCIA ZEBRATA ON BOUGAINVILLE : Parker

specimens had a high Salmonella con-
tent, and that dead skinks had high
Salmonella and amoeba infestations.

COMMUNAL TORPIDITY

Corucia zebrata may exhibit com-
munal torpidity at times. Kinghorn (1928)
reports field information obtained by N.
S. Heffernan on Ysabel Island:

Westem natives assert that if there is more
than one of these lizards in the same hollow tree,
the topmost one only comes out to feed, the other
undemeath remaining and feeding on the
excreta of the top one, which is always fat and
healthy, the second one being thinner, and the
third very thin.

People of six major language groups in
areas where I collected on Bougainville
also report this behavior with more de-
tails and embellishments. The reports
seem to indicate that four to ten of these
lizards, of all sizes and sexes, are occa-
sionally found together in a hollow high
in a forest tree. Sometimes there are also
possums in the same hollow. The lizards
at the top are well fed and healthy, while
those below are progressively thinner.

As elsewhere in Papua New Guinea,
the people traditionally attach anthro-
pomorphic significance to animal be-
havior, and this apparent communal
torpidity is explained by the people in
the following fashion. When the lizards
cannot find enough food on adjacent
trees, and they cannot reach trees farther
away to feed, they enter a tree hollow to
sleep and wait for fresh growth of food
plants. The lizards nearest the entrance
to the hole commiserate with the others,
saying how unfortunate it is that they are
all without food. They offer to go out at
night to see if there is any food available.
They may find small quantities of food,
and eat this, but when they return to the
others they say they found nothing at all.
Because they have had something to eat,
they defecate, and remain fairly healthy.

439

The lizards below them in the hollow eat
the feces, as there is nothing else for
them.

Nothing in these stories indicates that
there is a seasonal or even regular pattern
to this behavior. During nearly four years
of field work on Bougainville, such a
group of lizards was never found, nor did
village hunters report finding one.

Thus the actual occurrence of com-
munal torpidity remains unconfirmed.
From my experience on Bougainville and
elsewhere in Papua New Guinea, tradi-
tional interpretations of animal behavior
are almost invariably based on actual
observations. Should such reports be
obtained from different language groups
within the range of the species con-
cerned, it can be accepted that they are
based on fact.

The reports on Corucia from six of the
more than twenty different language
groups of Bougainville and Buka Islands,
and from Ysabel Island to the southeast
in the Solomon Islands would have a
high probability of being based on ob-
served, but apparently infrequent, be-
havior. The coprophagy part of the
reports has now been confirmed by
Honegger (1975).

REPRODUCTION

One, or occasionally, two live young
are born at a time, covered in a mem-
branous sac from which they soon
emerge and are immediately active and
able to feed and fend for themselves.
Breeding is reported by villagers to occur
at any time of the year. One specimen
born on 11 March 1961 measured a total
of 33 cm, while another bom on 17 April
1963 was 17.8 cm snout-vent, the tail 17.5
cm. Honegger (1975) reports a newly-
born specimen of 15.5 and 18.2 cm
length. The birth of twins is also re-
corded (Anonymous, 1978).

440

PREDATORS

On Bougainville, man appears to be
the major predator of Corucia. Villagers
do not specifically hunt these skinks, but
frequently find them while searching
trees for possums, a favorite subsistence
food. When a hunter finds one of these
skinks, he grasps it by the tail, and, avoid-
ing its claws, swings its head against a
tree to kill it. The body is cooked whole
in the ashes of a fire. The skin is removed
after cooking. If the lizard is to be kept
alive, the hunter ties a length of vine to
the end of a stick, makes a noose at the
end of the vine, and places the noose
around the skink’s neck. He then lifts the
lizard off the ground until it grasps the
stick, and then binds the body, limbs, and
tail to the stick with another vine.

Loveridge (1946) records an 18 inch
Corucia taken from the stomach of an 11
foot crocodile (Crocodylus porosus) in
the Shortland group, south of Bougain-
ville.

VERNACULAR NAMES

In the areas where data collection was
carried out, this was one of the better
known species of reptile. This may be
because of its large size, vegetarian diet,
association with possums, use as food, or
because it is the only viviparous lizard on
the island. No information indicates that
this species is a clan totem. Clan mem-
bership is inherited matrilineally on
Bougainville, and members may not eat
or touch their clan totem. The totem
name is often used as the clan name.
Other large reptiles, including the croc-
odile (Crocodylus porosus) and some of
the marine turtles (Dermochelys coriacea
and Eretmochelys imbricata) are clan
totems in parts of Bougainville. Corucia
appears to be absent from creation and

Advances in Herpetology and Evolutionary Biology

spirit mythology on Bougainville, where-
as the crocodile and the ground boa
(Candoia carinata) figure prominently.

TABLE 1. NAMES USED FOR CORUCIA ZEBRATA IN SOME
OF THE LANGUAGES OF BOUGAINVILLE.

Kieta District Nasioi Language Katau
Guava Katau
Eivo Katakau
Buin Buin Katau, Kama
Banoni Takorai
Kunua Kunua Bokapuo
Tinputz Tinputz Kokorot
ACKNOWLEDGMENTS

I am grateful to J. Womersley, then
Chief of the Division of Botany in Lae,
for identification of food plants.

LITERATURE CITED

ANONYMOUS. 1978. First time at the zoo. ZoonooZ
(San Diego Zoo) February, 1978.

Burt, C. E., AND M. D. Burr. 1932. Herpetological
results of the Whitney South Sea Expedition.
VI. Bull. Amer. Mus. Nat. Hist., 63: 461-597.

Guppy, H. B. 1887. The Solomon Islands and their
Natives. London, Swan Sonnenschein, Lowrey
& Co., i-xvi + 384 pp.
KINGHORN, J. R. 1928. Herpetology of the Solomon
Islands. Rec. Aust. Mus., 16(3): 123-178.
HOoNEGGER, R. E. 1975. Beitrag zur Kenntnis des
Wickelskinkes Corucia zebrata. Salamandra,
11: 27-32.

LOVERIDGE, A. 1946. Reptiles of the Pacific World.
New York, MacMillan, i-xii + 259 pp.

McCoy, M. 1980. Reptiles of the Solomon Islands.
Wau Ecology Institute, i-vi + 80 pp.

ParRKER, F. 1970. Collecting reptiles and amphib-
ians in New Guinea. Aust. Nat. Hist. (Austral-
ian Museum) March 1970, pp. 309-314.

SCHMIDT, K. P. 1932. Reptiles and amphibians from
the Solomon Islands. Field Mus. Nat. Hist.
Zool. Ser., 18: 175-190.

Seasonal and Spatial Variation in the
Annual Cycle of a Tropical Lizard

ROBIN M. ANDREWS!
A. STANLEY RAND?
STELLA GUERRERO?

ABSTRACT. The population ecology of the small
tropical lizard, Anolis limifrons, was studied over a
two year period on two sites at Barro Colorado Is-
land in Panama. Population density was the highest
late in the wet season following recruitment of
juveniles and the lowest early in the wet season
prior to the hatching of eggs. Seasonal fluctuation in
density is the result of low egg production during
the dry season and annual population turnover.
Seasonal changes in egg production appear to be
the result of food or water stress, or both. Individual
growth rates, physical condition of all individuals,
and food intake by juveniles and adult females were
lower in the dry than the wet season, paralleling
seasonal lows in arthropod abundance. Variation in
annual recruitment on Lutz site over four years was
not related to food intake by adult females during
the period when egg production is maximal. Popu-
lation density varied spatially with lizards about
one-half as abundant on Lutz as on AVA site. The
high population density on AVA site was associated
with relatively low female reproductivity, relatively
poor physical condition, and low food intake. Egg
mortality was higher on Lutz than on AVA site re-
sulting in a lower net reproductive rate on Lutz than
AVA. The discordance between population density
and measures of lizard fitness on the two sites was
the result of independence of egg and lizard fitness
components.

1Department of Biology, Virginia Polytechnic
Institute and State University, Blacksburg, Virginia
24061, U.S.A.

2Smithsonian Tropical Research institute, P.O.
Box 2072, Balboa, Panama.

3Department of Entomology, University of Cali-
fornia, Berkeley, California 94720, U.S.A.

INTRODUCTION

Population density of Anolis limifrons
on Barro Colorado Island (BCI) in
Panama fluctuates seasonally and from
year to year. Low egg production during
the dry season and annual population
turnover produce the seasonal fluctua-
tions (Sexton et al., 1963, 1971; Andrews
and Rand, 1974, 1982). As a result, popula-
tion density is the highest late in the wet
season following the major influx of
juveniles into the population and lowest
early in the wet season prior to the hatch-
ing of eggs.

Population density also fluctuates as
much as sixfold from year to year.
Andrews and Rand (1982) found a nega-
tive correlation between density late in
the wet season and the length of the
preceding dry season for their ten year
study. Thus, both seasonal and year to
year fluctuations have been associated
with variation in rainfall. Two explana-
tions have been proposed for this associa-
tion.

The first explanation is that there is in-
sufficient soil moisture for eggs to com-
plete development successfully during
the dry season, or desiccation is a prob-
lem for hatchlings, or both (Sexton, 1967;
Sexton et al., 1971). Thus, egg production
is reduced during the dry season and
recruitment is inversely related to the
length and severity of the dry season.

442

Evidence to support this hypothesis is in-
direct, specifically that populations of A.
limifrons are not food limited during the
dry season (Sexton et al., 1972). Although
food is less abundant during dry season
months, Sexton et al. (1972) found that
the proportion of adult lizards with
empty stomachs, the mean number of
prey per lizard stomach, and prey size
were similar in the dry and the wet
season. Thus, Sexton et al. (1972) argued
that because individuals get as much to
eat in the dry season when food is scarce
as in the wet season when it is abundant
that food is not limiting.

A second explanation argued by Wright
(1979) is that A. limifrons populations are
food limited during the dry season when
food is least abundant. Moreover, since
variation in insect abundance is associ-
ated with variation in rainfall (Wolda,
1978a), the magnitude of annual recruit-
ment should also be associated with food
availability. Wright (1979) further
pointed out that the food habit data col-
lected by Sexton et al. (1972) was from a
dry season that was unusually wet and
suggested that even if food were not
limiting when Sexton et al. (1972)
sampled, it might be in most years when
the dry season is more severe.

In this paper we describe seasonal
changes in A. limifrons populations at
two sites on BCI, one with a more dense
lizard population than the other. The site
of relatively low population density is the
one on which density has been moni-
tored for ten years (Andrews and Rand,
1982). Comparative data on population
density, survival, recruitment, individual
growth rate, reproductive condition of
females, body size, and food intake were
collected over a period of two years.
Arthropod abundance was not measured.
Since the plasticity of reptilian growth
and egg production rates in response to
changes in arthropod abundance is well
documented (Andrews, 1976; Ballinger,
1977; Dunham, 1978; Smith, 1981), we
assume that food intake and growth rates
provide direct assessments of food avail-

ability.

Advances in Herpetology and Evolutionary Biology

We propose to evaluate the role of food
as a factor regulating population density
of A. limifrons. We will do so by deter-
mining the relationship between food
availability and seasonal changes in egg
production, year to year changes in an-
nual recruitment, and spatial variation in
density. The data analyzed here was col-
lected as part of a study of the effects of
malaria on A. limifrons (Rand, Guerrero,
and Andrews, this volume.)

BARRO COLORADO ISLAND

The vegetation on BCI is classified as
tropical moist forest (Holdridge and
Budowski, 1956). Seasonality in rainfall
is marked. The dry season (months with
less than 100 mm of rain, Janzen, 1967)
generally includes January through April
(Rand and Rand, 1979). The dry season
months are characterized by sharply de-
creased rainfall, decreased relative
humidity, increased wind speed, and in-
creased light in the understory as a result
of partial leaf drop by canopy trees
(Leigh et al., 1982). The wet season
includes May through December. We
have arbitrarily divided the wet season
into an early wet season (May-August)
and a late wet season (September-
December). The wet season often begins
abruptly, and its onset is associated with
such seasonal phenomena as leaf flush-
ing of forest trees, reproductive activity
of plants and animals, and peak insect
abundance (Leigh, 1975; Wolda, 1978a,
b; Robinson and _ Robinson, 1970;
Rubinoff, 1974; Toft and Levings, 1979).

One of our study sites, Lutz, is an 890
m? area located in the Lutz watershed
(Rubinoff, 1974). The other site, AVA, is a
990 m? area located at the 2,500 meter
marker of Armour trail. These sites were
established in November 1971 and May
1976, respectively, and they were located
about 3 km distant from one another.
AVA site was structurally similar to Lutz
site. Although in most places the under-
story was sparse enough to walk through
easily, both sites included several

ANNUAL CYCLE OF A LIZARD : Andrews, Rand, and Guerrero

clumps of dense and tangled vegetation
resulting from tree falls.

METHODS

Thirteen censuses were conducted at
each site (Fig. 1). Four minor censuses
consisted of one or two consecutive days
of collecting. Nine major censuses con-
sisted of two consecutive days of collec-
ting followed by two additional days of
collecting six to seven days later. Lizards
were collected in the moming (0930-
1200 h). During each census, the sites
were searched by two or three people
and about 95% of all lizards seen were
captured. We moved through the sites
slowly, systematically examining all pos-
sible perches from tiny stems to large
tree trunks from ground level to 2 to 2.5 m.
Lizards were captured by hand and indi-
vidually placed in plastic bags and trans-
ported to a laboratory where size, sex,
and individual identification number
were recorded. Snout-vent length (SVL)
was measured to the nearest mm and
weight (W) to one-hundreth of a gram.
Individuals were identified by clipping
the terminal phalanges of toes in unique
combinations that required only one toe
per foot. In addition, blood smears were
taken from each female to assess repro-
ductive condition. Color characteristics
of Giemsa stained smears indicates indi-
viduals that are depositing yolk (Rand,
Guerrero, and Andrews, this volume).
Lizards were held in plastic bags for 24 h
and then released at the place of capture.
Scats produced during this period were
dried at 60°C, weighed to one-tenth of a
mg, and used to estimate food intake (see
below).

Lizard classes were based on age, size,
and reproductive attributes. Males and
females on BCI both reach a maximum
size of 49 to 50 mm SVL. Individuals 40
mm SVL or more were considered adult
because most are sexually mature at
these sizes (Sexton et al., 1971). For some
analyses, the adult class was partitioned
into young adults (40-43 mm SVL) and

443

old adults (> 43 mm SVL). These adult
classes were recognized because some
individuals cease linear growth at 44 mm
SVL and can no longer be aged by size.
The juvenile I class consisted of indi-
viduals 29 mm SVL or less, and the
juvenile II class consisted of individuals
30 to 39 mm SVL.

Total population size for the 1976-1977
censuses was estimated by the method of
Jolly (1965). Major censuses provided
two estimates of population size, and
minor censuses provided one estimate.
Bailey's modification of the Lincoln
index (Poole, 1974) was used to estimate
population size for the 1978 censuses.

Survivorship was determined as

S = ll = Gln IN, = Ih Wi)
where S equals the daily survivorship
rate, N, is the number of individuals
present or known to be alive in an initial
census, N, is the number of individuals
included in the initial census present or
known to be alive in a following census,
and t is the interval between censuses in
days. Survivorship was expressed as S?°,
the survivorship over a 28 day period.
Although males and females were initial-
ly distinguished, preliminary analyses of
survivorship did not detect differences
related to sex. Therefore, survivorships
were determined for two classes, adults
and juveniles. Because home ranges of
both juveniles and adults are quite stable
(Andrews and Rand, 1983), we assume
that few individuals disappearing from
the censuses had emigrated. Jolly (1965)
estimates for survivorship were not used
because they do not distinguish age
classes.

Survivorship curves were constructed
for the cohorts hatched in the 1976 wet
season. Because the high attrition of
hatchlings precluded following one set of
individuals through time, the curves
were based on two age groups. Survivor-
ship to 90 days of age was determined for
all individuals 29 mm SVL or less at first
capture. Survivorship from 90 days on-
ward was determined for all juveniles

444

that reached at least 90 days of age.
Juvenile ages were determined from
growth rates of recaptured individuals.

Growth data from census records were
fitted to the logistic growth model using
the interval equation derived by
Schoener and Schoener (1978). This
equation is

iL, =

where the estimated parameters are the
rate constant (r) of the logistic growth
equation and the asymptote (a). The data
provided by the censuses were the size at
first capture, L,, the size at second cap-
ture, L,, and the interval between cap-
tures in days, D. Estimates of r, a, and the
95% confidence intervals for both
parameters were determined by non-
linear regression procedures (Barr et al.,
1976). Snout-vent length was used as an
index of size because weight gave poorer
fits to the logistic model as judged by
relative widths of the 95% confidence
intervals.

The physical condition of adult A.
limifrons was estimated as

JE aA, ae = Ib) See)

IC = (W°?33/SVL) * 100

where IC is the index of condition. For A.
limifrons, the exponent of the power
function relating W to SVL is 3 (Andrews,
unpublished data). Therefore, to obtain
an estimate of IC that was independent of
size, W was raised to the 1/3 power.

Changes in the ratio between body
weight and body length of adult indi-
viduals should reflect changes in their
physical condition. High values of IC
would be expected when females were
producing eggs since eggs and maturing
Ova may comprise as much as 10% of
female body weight (Andrews and Rand,
1974). High values would also be ex-
pected during periods of high food in-
take. Although levels of hydration have
not been measured at different times of
the year, it is possible that some differ-
ences in IC could be attributable to this
source of variation.

Advances in Herpetology and Evolutionary Biology

Food intake was estimated from scat
production as

I = SCTG * 2.94 * 22,773

where I is food intake in J.g-!.d-', SETG
is dry scat weight in g divided by lizard
live weight in g, 2.94 is the correction
factor which converts scat production
into food intake (see below), and 22,773
is the energy value of food items in J.g".
The 2.94 correction factor was deter-
mined from energy budgets of 23 adult
female A. limifrons whose food intake
and scat production were monitored
daily over a 35 day period (Andrews,
1979). During this period, the mean ratio

of dry scat weight to the dry weight of

food eaten was 0.34 + 0.009(K + SE). In
order to determine if individuals cap-
tured during censuses would have ex-
hibited similar ratios, 20 females were
taken from their cages in late morning
after they were fed and placed indi-
vidually in plastic bags. They were then
treated similarly to individuals captured
in the field census. After 24 h scats were
collected and the females returned to
their cages. Each female was used as its
own control in this experiment, using as
the control value the ratio determined for
the day prior to the experiment when the
lizards were not disturbed. “Captured”
females had ratios of 0.33 + 0.042, and
the control females had ratios of 0.32 +
0.024 (p > 0.05, paired comparison t-test).
A conspicuous difference between the
“captured” and the control females was
that seven of 20 “captured” females did
not defecate during the 24 h confine-
ment, but only two of 20 control females
did not defecate during 24 h (p = 0.0637,
Fisher Exact Test). The low frequency of
defecations by “captured” individuals
was apparently the result of stress and
not of lower food intake. “Captured”
individuals that did not defecate had
eaten similar quantities in the previous
24 h as those that did defecate (17.6 vs.
18.7 mg, respectively, p > 0.05, t-test).
For this reason, individuals that did not
defecate were not included in the an-

ANNUAL CYCLE OF A LIZARD - Andrews, Rand, and Guerrero

alyses of the census data. We assume that
their food intake was the same as that of
individuals that produced scats during
their 24 h confinement.

Because A. limifrons eat a wide variety
of small arthropods year round (Sexton,
unpublished data; Andrews, unpub-
lished data), it is probably appropriate to
use a single energy value for their prey.
Because the scat weights collected were
not corrected for ash, the value of 22,773
+ 556 J.g 1 (det. by RMA) for adult
crickets was used to represent prey
items. The crickets had an ash corrected
energy value of 24,577 J.g-! which was
similar to the 24,171 J.g~! that Reichle et
al. (1971) considered representative of 15
species of litter arthropods.

All mathematical nd _ statistical
manipulations of the SCTG data sets
were conducted after natural log trans-
formations. This procedure reduced the
significantly (p < 0.05, Sokal and Rohlf
1969) skewed data sets (lizard class by
census) from 39 to 15%.

RESULTS

POPULATION DENSITIES

During the 1976-1978 study period we
observed seasonal and spatial differences
in population density of A. limifrons. In
1976, population densities reached
maximum levels in the late wet season
following large increases in population
density due to recruitment of juveniles
(Fig. 1). The proportion of small juve-
niles on Lutz was the highest in July to
August (Fig. 2). The low proportion of
this class during the dry and the early wet
season suggest most recruitment occurs
within a relatively few months in the
middle of the wet season. Proportional
changes in class representation were
similar on AVA site.

Recruitment was apparently higher in
1976 than in 1977, judging by the num-
bers of individuals present in the
December 1976 to January 1977 and the
January to February 1978 censuses, re-

445

spectively (Fig. 1). As pointed out by
Andrews and Rand (1982), the high re-
cruitment of 1976 was associated with a
short preceding dry season (January to
mid-April) and the low recruitment of
1977 was associated with a long preced-
ing dry season (mid-December 1976 to
mid-May 1977).

Spatial differences were pronounced.
The density of the A. limifrons popula-
tion on Lutz site was about half that on
AVA site (Fig. 1). Densities reached
respective maximum values of 1,150 and
1,860 individuals/ha in the late wet sea-
son of 1976 and respective minimum
values of 260 and 660 individuals/ha in
the early wet season of 1977. The maxi-
mum density observed on Lutz site dur-
ing the study period was similar to den-
sities observed in the late wet seasons of
1971 and 1978; 1,150 individuals/ha thus
represents the maximal density attained
on this site over ten years of censusing
(Andrews and Rand, 1982).

SURVIVORSHIP

Survivorship of lizards on Lutz and on
AVA site was similar overall (p > 0.05,
Mann-Whitney U-test, Table 1). Adult
survivorship on Lutz site averaged 0.725
per 28 days and did not vary appreciably
through time (p > 0.05, 2 x 2 chi-squared
tests for all pairwise comparisons of the
1976 to 1977 data). Adult survivorship on
AVA site averaged 0.756 per 28 days, but
in contrast to Lutz site, survivorship did
vary temporally. Survivorship during the
last two intervals of 1977 on AVA site was
significantly lower than during the pre-
ceding four intervals (p > 0.05, 2 x 2 chi-
squared tests for all pairwise compari-
sons). This change in survivorship had no
obvious cause.

The similarity of adult and juvenile
survivorship during each census interval
(Table 1) was an artifact caused by lump-
ing all juvenile survivorship data. The
analysis of survivorship using cohort data
indicated that the survivorship of indi-
viduals of 90 days of age or less was lower
than the survivorship of individuals more
than 90 days of age on both sites (Fig. 3,

446 Advances in Herpetology and Evolutionary Biology

150
ep)

Y 100
<x
=)

= 50
>
a)
z
ins
O
og
Lu
(0a)
=
D
z

20097 RAINFALL (mm)

Figure 1.

The total size of Anolis limifrons populations on Lutz and AVA sites is shown in the top two graphs.

Estimates are presented plus and minus their standard errors (see text for estimation procedures). Weekly rainfall
(mm) is shown in the bottom graph. Incomplete censuses on AVA precluded population estimates for June 1976

and June 1978.

p < 0.01, 2 x 2 chi-squared tests). The
low survivorship of the smallest juveniles
was indicated by a significant (r = 0.74, p
< 0.05) increase in survivorship with
mean size (Table 1).

Survivorship curves of the 1976 cohort
on Lutz and AVA sites were similar in
their general appearance (Fig. 3). Sur-
vivorship to 90 days of age was 0.636 and
0.539 per 28 days, respectively. The
largest juveniles and adults had survivor-
ships on Lutz and AVA sites of 0.797 and
0.845 per 28 days, respectively. The
probability of death was constant until
about 210 days of age and then dropped

to 0.616 and 0.658 per 28 days, respec-
tively. Although the cohort analysis was
based on relatively few individuals, it
confirmed the general results of the sur-

vivorship analyses based on census inter-
vals (Table 1).

FEMALE REPRODUCTION

Old and young adult females had very ~
different patterns of reproductive ac-
tivity. Old females exhibited continuous
oogenesis on both sites with 88-97%
judged reproductively active (Table 2).
Fewer young females were judged

ANNUAL CYCLE OF A LIZARD - Andrews, Rand, and Guerrero

@oacoosotoscone ~» Adult males
o—— Adult females

60 e@———-e Young adults
s Ey o——> Juvenile |
® 50le % (6-1-1) —e Juvenile II
Ss \ 2
42) y L
2 AO Ne ey . ea
SOMME ef NO ge Rs D
= y
“= 0) |r
(
9S
= 20 \, DB
S \oupeeea
S [aa
ao AOF \-7

laslaealsc
J A& S ON DO V
1976

FMAM J J
1977

Figure 2. Proportional representation of the five lizard
classes on Lutz site during the period of most intensive
censusing. The darkened bar on the time axis indicates
dry season months.

447

)
a
O
2
>
a
=)
Oo
2
—
0 50 150 250
AGE CDAYS)

Figure 3. Survivorship curves for Anolis limifrons
populations on Lutz (@) and on AVA (0) sites. Sample
sizes are shown for the initial cohorts at about 30 and
90 days of age. The ordinate indicates the natural log
(In) of the number of survivors with N=1,000 at t=0.

TABLE 1. SURVIVORSHIP OF ADULT (= 40 MM SVL) AND JUVENILE (< 40 MM SVL) ANOLIS LIMIFRONS ON LUTZ AND AVA

SITES.

Daily survivorship has been converted to survivorship over a 28 day period (S?*). The total number of
individuals counted in the initial census follows survivorship in parentheses. Survivorship was determined
only when the initial number of individuals was 20 or more. Mean juvenile snout-vent length (SVL) shown is

for the initial census.

Census Intervals

LUTZ SITE
(days)
May-July 1976 (82)
July-Oct. 1976 (75)
Oct—Dec. 1976 (49)
Dec.-Feb. 1977 (70)
Feb.-Apr. 1977 (73)
Apr.-June 1977 (70)
June-Jan. 1978 (204)
Jan—May 1978 (119)
AVA SITE
(days)
June-Sept. 1976 (80)
Sept.—Oct. 1976 (56)
Oct.Jan. 1977 (85)
Jan.—Mar. 1977 (71)
Mar.-June 1977 (63)
June-Aug. 1977 (64)
Aug.Feb. 1978 (183)
Feb.June 1978 (140)

I

I

Survivorship

Adults
.798 (27)
.761 (29)
.658 (31)
.743 (44)
.736 (42)
613 (37)
.745 (25)
752 (30)
£725
838 (53)
836 (50)
842 (64)
.809 (116)
.678 (100)
577 (66)

— (14)
723 (30)

757

Juvenile
Mean
Juveniles SVL
a (4) pes
.670 (29) 25.3
671 (24) 30.4
.786 (20) 34.7
a= (14) moss
— (i) a
= (6) —_
.630 (21) 26.8
.689
— (2) _—
655 (51) 29.7
.789 (45) 32.0
858 (25) 33.7
eee (14) —
= (ll) —
— (0) pie
.736 = (23) 35.6
.760

448

Advances in Herpetology and Evolutionary Biology

TABLE 2. PERCENT OF ADULT FEMALE ANOLIS LIMIFRONS DEPOSITING YOLK DURING DRY SEASON (JAN. TO APRIL)
AND WET SEASON (May TO DECEMBER) CENSUSES (SEE TEXT FOR DETAILS).

LUTZ SITE AVA SITE
Dry Wet Dry Wet
Class % N % N % N %
Young adults (40-43 mm SVL) 59 = (27) 78 ~=(18) 36 = (45) 74 = (19)
Old adults (44 mm SVL +) 97 ~=(38) 95 (60) 90 (68) 88 (94)

oogenic than old adults. This difference
was significant in the dry season on both
sites (p < 0.05, 2 x 2 chi-squared tests).
More young adults on both sites were
reproductively active in the wet season
than in the dry season, but seasonal dif-
ferences were significant only on AVA
site (p < 0.01, 2 X 2 chi-squared tests)
where 74% were judged oogenic in the
wet season in comparison with 36% in
the dry season.

The blood staining technique is largely
insensitive to rate of egg production. It
distinguishes reproductive individuals as
those that have high levels of circulating
lipoprotein associated with yolk deposi-
tion (Ayala and Spain, 1975). Old (large)
females thus exhibited high levels of
oogenic activity year round despite more
than a twofold difference in egg laying
rate between the wet and the dry season
(Andrews and Rand, 1974). However,
since young (small) adult females have an
energy commitment to both growth and
reproduction, oogenic events may not be
continuous. Thus, the seasonal and site
differences in the proportion of oogenic
individuals may reflect either differences
in egg laying rates, or age (size) of sexual
maturity, or both.

GROWTH

Individual growth records, partitioned
by sex, site, and season, were fit to a
logistic growth model. Preliminary an-
alyses showed that males and females did
not differ either in their asymptotic size
or their rate constants (p > 0.05, two-
tailed t-tests). Therefore, only the results
of site and season comparisons are pre-

sented (Fig. 4, Table 3). The mean maxi-
mum SVL (asymptote) attained was about
47 mm for all the fitted growth curves. In
contrast, the rate (r) at which asymptotic
size was reached varied considerably.
The rate constant was greater in the wet
season than the dry season on both sites
(p < 0.05, two-tailed t-test). The magni-
tude of the difference was such that an
individual hatched in the wet season
reached sexual maturity (40 mm SVL) 30
days sooner than an individual hatched
in the dry season.

Growth constants (r) were higher on
Lutz site than on AVA site during both
the wet and the dry season (p < 0.05, two-
tailed t-tests). For example, during the
wet season on Lutz site sexual maturity
was reached at 93 days and on AVA site at
125 days. During the dry season, the dif-
ference was also about 30 days.

PHYSICAL CONDITION

Index of condition (IC) varied as a
function of site, sex, and season (Fig. 5).
A three-way analysis of variance using
only the first record for each individual
for a census period (N = 917) was used to
test these main effects and all interaction
terms. Season, sex, and site all had a sig-
nificant effect on IC (p < 0.05). The
interaction term between sex and time
was also significant (p < 0.001). In gen-
eral, the early wet season had the highest
values of IC, the late wet season inter-
mediate values, and the dry season had
the lowest values. Thus, both males and
females were in the poorest condition in
the dry season and the best condition in
the early wet season. Females generally

ANNUAL CYCLE OF A LIZARD : Andrews, Rand, and Guerrero

449

TABLE 3. FITTED PARAMETERS OF THE LOGISTIC GROWTH EQUATION OF ANOLIS LIMIFRONS POPULATIONS ON LUTZ
AND AVASITES.
The rate (r) constants and asymptotes (a) are shown followed by their 95% confidence intervals in paren-

theses.
Season N r a
LUTZ SITE Wet 38 .0228 (.0194—.0262) 46.8 (46.347.6)
Dry 62 .0177 (.0156—.0198) 47.6 (46.7-48.4)
AVA SITE Wet 74 .0176 (.0156—.0196) 47.7 (47.2-48.3)
Dry 144 0135 (.0119-.0150) 47.3 (46.7-48.0)

€
s
<=
ol
(o)
Zz
WW
=)
=
Zz
uw
a
ke
=)
oO
Zz
no
100 200 300 400
TIME (DAYS)

Figure 4. Fitted logistic growth curves for A. limifrons
individuals on Lutz and AVA sites during the 1976 wet
and the 1977 dry season. Parameter estimates are
given in Table 3.

Qa
QQ OL
qq 2
Qoe
%e0

INDEX OF CONDITION (IC)
= s
(oy) (oe)

EW Lw DRY EW LW =DRY EW
76 77 78
Figure 5. Index of condition for Anolis limifrons on

Lutz (@) and AVA (0) sites in different seasons and
years. Seasonal means with the same letter are not
significantly different (p > 0.05, Duncan’s multiple
range test). See text for details.

had higher values of IC than males on
each site during all times of the year.
Finally, values of IC on Lutz site were
generally higher than those of AVA site
for within sex comparisons.

FOOD INTAKE

The scat production method of estimat-
ing food intake was supported by an in-
dependent determination of food intake.
In May and June 1977, adult females
feeding ad libitum in captivity ate 284
J.g_1.d~! (Andrews, 1979). In comparison,
adult females censused on Lutz site in
May and June to July 1977 had an esti-
mated food intake of 265 and 304 J.g-'.d“
(Fig. 6). Since anoles were probably not
food limited on Lutz site during this
period at least (see Discussion), the cor-
respondence between the two sets of
measurements is corroborative.

In order to evaluate the differences in
food intake between the sites, lizard
classes, and seasons, a three-way analysis
of variance was used to test these main
effects and all of the interaction terms.
Censuses were combined into six con-
secutive seasons: early wet 1976, late wet
1976, dry 1977, early wet 1977, dry 1978,
early wet 1978. In this analysis all main
effects were significant, as was the inter-
action between season and class (p <
0.001). Differences between means were
then tested with Duncan’s multiple
range tests. These results are discussed
below.

Food intake was significantly higher
on Lutz than on AVA site with respective

450

5257

315

JUVENILES

105
8@

GigE ede

420

210

FEMALES 17 C)

INTAKE

315

FOOD

105

w

(N/.25m*) 102
= Le)

FOOD
ABUNDANCE

JAN SOUND
1976

Advances in Herpetology and Evolutionary Biology

e 7
of
18
6 e@
19
@)
7 Se
11
e 25
e@
12
7 fo)
8 @)

JFMAMJJA "dO ie
1977 1978

Figure 6. Mean food intake (J.g"!.d~') for juvenile, adult female, and adult male Anolis limifrons individuals on
Lutz (@) and AVA (0) sites. Sample sizes for each census are shown next to the means. Diamonds represent
mean food intake for the previous year (data from Andrews and Asato, 1977). Food abundance was determined
as the number of litter arthropods 2.5 to 23.5 mm in length collected in weekly 9.25 m? samples in Lutz watershed
(S. Levings and D. M. Windsor, 1982, unpublished data). The values presented here are bimonthly means

representing two (rarely three) collections.

means of 263 and 220 J.g4.d' (p <
0.001). Juveniles, adult females, and
adult males differed in their food intake
with respective means across sites of 296,
261, and 194 J.g"'td“! (p < 0.05). The
high food intake of juveniles was prob-
ably related to their weight specific
metabolic rate (Bennett and Dawson,
1976; Andrews and Asato, 1977) and the
energy costs of growth processes (War-
ren, 1971). The wide fluctuation in intake
by juveniles may be the result, in part, of

seasonal changes in mean juvenile size
(Fig. 2). The food intake of adult females
was more similar to that of juveniles than
to that of adult males. This suggests that
the energy cost of egg production is high
relative to the cost of reproduction by
males. A comparison of class means for
each site further illustrates both site and
class differences. Respective means were
330, 293, and 209 J.g"!.d~! for Lutz site
and 272, 243, and 184 J.g-'.d“! for AVA
site. Only the comparisons of adult males

ANNUAL CYCLE OF A LIZARD : Andrews, Rand, and Guerrero

vs. adult females and juveniles vs. adult
males differed significantly (p < 0.05,
Duncan’s multiple range tests).
Although the effect of season on food
intake was highly significant in the com-
plete model, only the early and the late
wet seasons of 1976 could be distin-
guished (p < 0.05, Duncan’s multiple
range test). Inspection of food intake
through time (Fig. 6) shows that the
highest values for food intake for all clas-
ses were observed in that period. The
pattern of temporal change in food intake
differed between the three classes as in-
dicated by a significant (p < 0.001)
interaction between time and class.

DISCUSSION

RESPONSE TO SEASONALITY IN FOOD
AVAILABILITY

Seasonal variation in arthropod abun-
dance on Barro Colorado Island is docu-
mented. In general, abundance is lowest
during the dry season and highest in the
early wet season in association with the
advent of the rainy season (Fig. 6; Robin-
son and Robinson, 1970; Rubinoff, 1974;
Willis, 1976; Wolda, 1978b; Toft and
Levings, 1979; Levings and Windsor,
1982). Depending on the arthropod taxon,
abundance may gradually decline during
the late wet season to dry season levels or
abundance may reach seasonal lows as
early as August (Rubinoff, 1974).

Wright's (1979) hypothesis that food is
limiting during the dry season was sup-
ported by our data. Individual growth
rates were lower during the dry than the
wet season on both sites (Fig. 4). Both
males and females had lower weight to
length ratios during the dry than the wet
season (Fig. 5). Food intake by adult
females and juveniles was lower in the
dry than in the wet season (Fig. 6). This
variation in growth, physical condition,
and food intake does parallel known sea-
sonal changes in arthopod abundance on
BCI. However, the conclusion that popu-
lations are limited by food shortage dur-

451

ing the dry season must be qualified by
two additional considerations.

First, changes in food availability may
have behavioral as well as physiological
consequences. For example, food intake
by adult males was similar year round. If
food were limiting during the dry season,
males should have increased their food
intake during wet season months since
females clearly were able to do so. On the
other hand, dry season food shortage for
males may have increased the time allo-
cated to foraging as has been observed for
Anolis cupreus (Fleming and Hooker,
1975). Food limitation thus might not be
manifested in food intake per se but by
other costs. For example, increased forag-
ing activity may expose individuals to
higher risk of being eaten by predators or
may reduce effectiveness of territorial
defense. Since females spend most of
their activity period searching for prey
year round (Andrews, 1971; Fleming and
Hooker, 1975), seasonal changes in food
intake by females should be more reflec-
tive of changes in prey availability.

Second, food may not be the limiting
nutrient. Stamps and Tanaka (1981) have
demonstrated that water availability
overrides food as the limiting factor for
growth of juvenile Anolis aeneus during
dry season months. Presumably, other
physiological processes such as oogene-
sis could be affected detrimentally by
water shortage as well. Thus, a more ap-
propriate conclusion for our study is that
conditions during the dry season are suf-
ficiently stressful to reduce growth rates
and food intake of young individuals and
to reduce rates of egg production and
food intake of adult females.

RESPONSE TO YEAR TO YEAR VARIATION
IN FOOD AVAILABILITY

In order to evaluate the effect of year to
year variation in arthropod abundance
(Rubinoff, 1974; Wolda, 1978b), on an-
nual recruitment of A. limifrons we can
compare food intake of adult females dur-
ing the early wet season when oviposi-
tion rates are probably maximal (An-

drews and Rand, 1982) with annual
recruitment as measured by population
density at the end of the wet season. For
1975-1978, respective mean values for
food intake by adult females on Lutz site
in the early wet season were 471, 401,
287, and 305 J.g''.d“! (Fig. 6). The cor-
responding values for annual recruitment
were 23, 100, 66, and 112 (1975 and 1978
population density data are from An-
drews and Rand, 1982). Thus, year to
year changes in food intake by females
were not translated into numbers of sur-
viving offspring.

The independence of annual recruit-
ment and food intake by adult females
suggests that annual recruitment may be
more influenced by the survivorship of
eggs (and possibly very small hatchlings)
than by the rate at which they are pro-
duced. This argument is supported by
the observation that survivorship of liz-
ards did not vary appreciably over 10
years (Andrews and Rand, 1982) and
that malaria parasites do not affect sur-
vivorship, growth, or fecundity of lizards
(Rand, Guerrero, and Andrews, this
volume). Moreover, observations on the
water relations of eggs of A. limifrons
suggest that moisture may not be limiting
during the dry season in most years (An-
drews and Sexton, 1981).

SPATIAL DIFFERENCES

Comparisons between our two sites
suggest that the population at AVA site
was food limited relative to the popula-
tion at Lutz site. Individuals on AVA site
grew more slowly, matured later, had
lower food intake, and had a lower index
of condition than individuals on Lutz
site. Reproductive activity of young adult
females was lower during the dry season
and egg production rates of young and
old females were presumably lower at
AVA than at Lutz site.

The comparisons of food intake,
growth, and index of condition suggest
that lizards on AVA site were laying
fewer eggs than lizards on Lutz site. This
creates an apparent paradox. Lizard den-

Advances in Herpetology and Evolutionary Biology

sities were greater on AVA than on Lutz
site despite poorer individual perform-
ance. Given a presumed lower reproduc-
tive rate and similar mortality schedules
(Table 1, Fig. 3), population density on
AVA should have declined relative to
Lutz site.

However, this result may not be as
paradoxical as first appears. For A. limi-
frons, lizard performance is transformed
into population density through survivor-
ship of eggs. Selection pressures in the
egg environment (soil-leaf litter inter-
face) and the lizard environment may be
quite different. Moreover, optimal habi-
tats for eggs and lizards may be found in
different places. This model is supported
by measurements of mortality rates of
eggs (Andrews, 1982). During May and
June 1981, eggs on AVA and Lutz sites
had mortality rates of 56 and 82%, respec-
tively, measured for a 40 day incubation
period. A life table analysis showed that
lower mortality during the egg stage
accounted for the consistently higher
population density on AVA than on Lutz
site despite the apparent low fitness of
lizards in the AVA population. We there-
fore suggest that spatial and temporal
variation in egg mortality may produce
the observed variation in lizard density
from place to place and from year to year.

ACKNOWLEDGMENTS
This study was funded by National
Science Foundation (NSF) Grants

BO19801X and GB37731X to E. E. Wil-
liams and NSF Grant 76—05758 to A. S.
Rand and S. Ayala. We thank N. Fetcher,
S. J. Wright, and two anonymous re-
viewers for their comments on the manu-
script, J. Birch of the Statistical Consul-
ting Service of VPI&SU for statistical

help, and S. Levings for allowing us to use ©
unpublished litter arthropod data. The
Fortran program used to determine the
Jolly estimates of population size was
written by G. G. Lawrence and R. H. Giles
of the Department of Fisheries and Wild-

ANNUAL CYCLE OF A LIZARD - Andrews, Rand, and Guerrero

, life at Virginia Polytechnic Institute and
_ State University.

LITERATURE CITED

ANDREWS, R. M. 1971. Structural habitat and time
budget of a tropical Anolis lizard. Ecology, 52:
262-270.

1976. Growth rate in island and mainland
anoline lizards. Copeia, 1976: 477-482.

_ ___.. 1979. Reproductive effort of female Anolis

limifrons (Sauria: Iguanidae). Copeia, 1979:
620-626.

—__. 1982. Spatial variation in egg mortality of
Anolis limifrons (Iguanidae). Herpetol. 38:
165-171.

ANDREWS, R. M., AND T. ASATO. 1977. Energy util-
ization of a tropical lizard. Comp. Biochem.
Physiol., 58A: 57-62.

ANDREWS, R. M., AND A. S. RAND. 1974. Reproduc-
tive effort in anoline lizards. Ecology, 55:
1317-1327.

= AND . 1982, in press. Long term changes in
population density of the lizard Anolis limi-
frons, pp. 000-000. In E. G. Leigh, Jr., A. S.
Rand, and D. M. Windsor (eds.), The ecology
of a tropical forest: seasonal rhythms and long-
term changes. Smithsonian Institution Press.

= AND . 1983. Limited dispersal of juvenile
Anolis limifrons. Copeia (in press).

ANDREWS, R. M., AND O. J. SEXTON. 1981. Water
relations of eggs of Anolis auratus and Anolis
limifrons. Ecology, 62: 556-662.

AYALA, S. C., AND J. L. SPAIN. 1975. Annual oogene-
sis in the lizard Anolis auratus determined by
a blood smear technique. Coepia, 1975: 138—
141.

BALLINGER, R. E. 1977. Reproductive strategies:
food availability as a source of proximal varia-
tion in a lizard. Ecology, 58: 628-635.

Bark, A. J., J. H. GOODNIGHT, J. P. SAIL, AND J. T.
HELwic. 1976. A user’s guide to SAS.76. SAS
Institute Inc., Sparks Press, Raleigh, North
Carolina, USA.

BENNETT, A. F., AND W. R. Dawson. 1976. Metabo-
lism, pp. 127-224. In C. Gans (ed.), Biology of
the reptilia, volume 5 (Physiology A). New
York, Academic Press.

DunuHaM, A. E. 1978. Food availability as a proxi-
mate factor influencing individual growth rates
in the iguanid lizard Sceloporus merriami.
Ecology, 59: 770-778.

FLEMING, T. H., AND R. S. Hooker. 1975. Anolis
cupreus: the response of a lizard to tropical
seasonality. Ecology, 56: 1243-1261.

HOLDRIDGE, L. R., AND G. BuDOwsKI. 1956. Report
of an ecological survey of the Republic of
Panama. Caribbean Forester, 17: 92-110.

453

JANZEN, D. H. 1967. Synchronization of sexual
reproduction of trees within the dry season in
Central America. Evolution, 21: 620-637.

JOLLy, G. M. 1965. Explicit estimates from capture-
recapture data with both death and dilution—
Stochastic model. Biometrika, 52: 225-247.

LEIGH, E. G., JR. 1975. Structure and climate in
tropical rain forest. Ann. Rev. Ecol. and Sys-
tem. 6: 67-86.

LEIGH, E. G., JR., A. S. RAND, AND D. M. WINDSOR
(EDS.). 1982. The ecology of a tropical forest:
seasonal rhythms and long-term changes.
Smithsonian Institution Press, in press.

LEVINGS, S., AND D. M. WINDsorR. 1982. Seasonal
and annual variation in litter arthropod popula-
tions. In E. G. Leigh, Jr., A. S. Rand, and D. M.
Windsor (eds.), The ecology of a tropical forest:
seasonal and long-term changes. Smithsonian
Institution Press.

POOLE, R. W. 1974. An introduction to quantitative
ecology. McGraw-Hill Series in Population
Biology.

RAND, A. S., S. GUERRERO, AND R. M. ANDREWS.
1983. The ecological effects of malaria on
populations of the lizard Anolis limifrons on
Barro Colorado Island, Panama. This volume.

RAND, W. M., AND A. S. RAND. 1979. The rainfall of
Barro Colorado Island, Panama, pp. 31-49.
Actas del IV Simposium Internacial de
Ecologia Tropical, Panama, Panama.

REICHLE, D. E. 1971. Energy and nutrient metabo-
lism of soil and litter invertebrates, pp.
465477. In P. Duvigneaud (ed.), Productivity
of forest ecosystems. Paris UNESCO.

ROBINSON, M. H., AND B. ROBINSON. 1970. Prey
caught by a sample population of the spider
Argiope argentata (Araneae: Araneidae) in
Panama: a year’s census data. Zool. J. Linn.
Soc., 49: 345-357.

RUBINOFF, R. W. 1974. Environmental monitoring
and baseline data (1973). Smithsonian Institu-
tion Environmental Sciences Program, Tropi-
cal Studies. Washington, D.C., Smithsonian
Institution.

SCHOENER, T. W., AND A. SCHOENER. 1978. Estimat-
ing and interpreting body-size growth in some
Anolis lizards. Copeia, 1978: 390-405.

SEXTON, O. J. 1967. Population changes in a tropical
lizard Anolis limifrons on Barro Colorado
Island, Panama Canal Zone. Copeia, 1967:
219-229.

SEXTON, O. J., H. HEATWOLE, AND E.. MESETH. 1963.
Seasonal population changes in the lizard,
Anolis limifrons, in Panama. Am. Midland
Nat., 69: 482-491.

SEXTON, O. J., E. P. ORTLEB, L. M. HATHAWAY, R. E.
BALLINGER, AND P. LICHT. 1971. Reproductive
cycles of three species of anoline lizards from
the Isthmus of Panama. Ecology, 52: 201-215.

SEXTON, O. J., J. BAUMAN, AND E. ORTLEB. 1972.
Seasonal food habits of Anolis limifrons.
Ecology, 53: 182-186.

454

SmiTH, D. C. 1981. Competitive interactions of the
striped plateau lizard (Sceloporus virgatus)
and the tree lizard (Urosaurus ornatus).
Ecology, 62: 679-687.

SOKAL, R. R., AND F. J. ROHLF. 1969. Biometry: the
principles and practice of statistics in biologi-
cal research. San Francisco, Califomia, W. H.
Freeman and Co.

STAMPS, J. A., AND S. K. TANAKA. 1981. The influ-
ence of food and water on growth rates in a
tropical lizard (Anolis aeneus). Ecology, 62:
33440.

Tort, C. A., AND S. C. LEvINGS. 1979. Tendencias
estacionales relacionadas con las poblaciones
de arthropodos en la hojarasca, pp. 559-576.
Actas del IV Simposium Intemacial de
Ecologia Tropical, Panama, Panama.

Advances in Herpetology and Evolutionary Biology

WaRREN, C. E. 1971. Biology and water pollution
control (Bioenergetics and growth, Chapter
11). Philadelphia, Pennsylvania, W. B.
Saunders Co.

WILLIS, E. O. 1976. Seasonal changes in the inver-
tebrate fauna on Barro Colorado Island,
Panama. Rev. Brasil. de Biologia, 36: 643-657.

WoLDA, H. 1978a. Fluctuations in abundance of
tropical insects. Am. Nat., 112: 1017-1045.

___. 1978b. Seasonal fluctuations in rainfall, food
and abundance of tropical insects. Journal of
Animal Ecology 47: 369-381.

WRIGHT, S. J. 1979. Competition between insecti-
vorous lizards and birds in Central Panama.
Am. Zool., 19: 1145-1156.

The Ecological Effects of Malaria on
Populations of the Lizard Anolis limifrons on
Barro Colorado Island, Panama

A. S. RAND!
STELLA GUERRERO?
ROBIN M. ANDREWS?

ABSTRACT. Prevalence of malaria varied between
two sites, with season, and was different in lizards
of different sizes. High levels of parasitemia pro-
duced anemia, though without measurable effect on
growth, reproduction, and survival of the lizards.
Anolis limifrons and the Plasmodium species that
infect it may exhibit a coevolved host-parasite rela-
tionship.

INTRODUCTION

This project was undertaken as part of a
long term study to investigate the factors
that influence population levels in a
tropical organism. The results demon-
strate an apparently well-adjusted rela-
tionship between a disease and its verte-
brate host. Population fluctuations in the
small, forest-understory, iguanid lizard
Anolis limifrons (Fig. 1) on Barro Colo-
rado Island (BCI), Panama have been
reported and discussed several times
over the past 14 years (Sexton, 1967; Sex-
ton et al., 1972; Wright, 1979; Andrews
and Rand, in press). Attention has fo-
cused largely on the importance of food
and water as controlling or limiting fac-
tors. When Guerrero et al. (1977) dis-

1§mithsonian Tropical Research Institute, Bal-
boa, Panama.

2Department of Entomology and Parasitology,
University of California, Berkeley, California
94720, U.S.A.

3Virginia Polytechnic Institute and State Univer-
sity, Blacksburg, Virginia 24061, U.S.A.

covered that these lizards had a high
prevalence of malaria and that this varied
from place to place on BCI, it seemed
possible that this disease was influenc-
ing, if not controlling, A. limifrons popu-
lation levels.

We set out to investigate this possi-
bility by examining lizards at roughly
monthly intervals at two small study sites
on Barro Colorado Island. One site had a
high lizard density and lower malaria in-
cidence, the other had fewer lizards and
more malaria. Anolis limifrons at the lat-
ter site had been censused for the past

Figure 1.
Panama.

Anolis limifrons from Barro Colorado Island,

456

eight years (Andrews and Rand, in press).
The ecological results of this study are
reported in the preceding paper (An-
drews, Rand and Guerrero, this volume);
those dealing with malaria are described
here.

Few studies have been carried out on
the effect of parasites in wild populations
(Anderson, 1976; Bennett et al., 1976).
Studies on the effect of helminths in
populations of wild moose and deer have
shown the different effects that a parasite
can have in two sympatric host popula-
tions, the deer acting as an unaffected
reservoir host and the moose being more
susceptible and their population density
limited by the parasite (Anderson, 1976).
Other host-specific parasite infections
affect only a single host species. There
are only a few reports on the effect of pro-
tozoans on populations of birds, such as
Wamer’s (1968) study on the effects of
malaria on the Hawaiian honeycreepers.

Malaria in reptiles represents a good
model for studies on the evolution and
epidemiology of natural infections
(Ayala, 1977). Reptiles make good sub-
jects for such studies because they are
sedentary and easy to capture so that the
same animals can be examined several
times. Following the first report that
plasmodia occurred in New World lizards
(Aragao and Neiva, 1909), many papers
have been written on reptilian malaria
(156 contributions from 1909-1975:
Ayala, 1978). Lizards and snakes are the
only reported hosts; of 87 species of New
World lizards reported to be infected
with malaria 34 are anoles (Ayala, 1977).
Most reports have very little information
on the ecology of the host and parasite,
and there are no studies of the impact of
malaria on lizard populations (Ayala,
1977). Ayala and Spain (1976) worked in
western Colombia with Anolis auratus
infected with Plasmodium colombiense,
finding fluctuations in the prevalance of
malaria through a 12-month period.
Scorza (1971) in a detailed study in
Venezuela, on Tropidurus torquatus
infected with Plasmodium tropiduri,

Advances in Herpetology and Evolutionary Biology

reported changes in hematology and in)
behavior; captive animals with high !
infections were less active and main--
tained lower body temperatures. Telford |
(1974, 1977) described the species of!
Plasmodium infecting anoline lizards at i
different localities in Panama. He»
showed fluctuations of natural popula- |
tions of Plasmodium and suggested ai
relation to the ecology of the vector popu- |
lation. Jordan (1964) also reported fluctu-
ations through the year and changes from |)
year to year in Plasmodium floridense }
infecting Anolis carolinensis in Georgia. |

Malaria may be an old disease of liz- |
ards, possibly much older than malaria in |
mammals (Ayala, 1977). Our findings |
suggest that the specific relationships of /
Anolis limifrons with the Plasmodium \
species that infect it are long-standing ;
and the species closely coevolved. As |
parasitologists have predicted in such;
situations the disease seems to have little +
effect on its natural host. In our study, the +
prevalence of malaria varied between the °
two sites, with season, and was different ©
in lizards of different sizes. Levels of ©
parasitemia affected the blood picture ©
and high levels produced anemia. How- -
ever, in the variables which should be.
relevant to the population dynamics,
such as growth, reproduction, and sur- —
vival, we found little measurable effect of
malaria, at least during the chronic phase
of the infection.

THE TYPICAL COURSE OF A MALARIA
INFECTION

To provide a background for the de-
scriptions which follow, we present a _
brief review, largely from Ayala (1977), of —
the typical course of a malaria infection ©
in a lizard. A detailed description of the ©
Plasmodium infecting Anolis limifrons, —
and their development and behavior, |
(Guerrero and Pickering, in preparation) ©
will be published elsewhere. Some in- |
formation on the course of the infection
in captive A. limifrons has been pre- |
sented in Pickering (1980).

)

ANOLE MALARIA : Rand, Guerrero, and Andrews

Malarial infections in reptiles follow
the same general course as in mammals
and birds: a rise, peak, decline, and
chronic period, sometimes followed by a
relapse. The vectors of malaria in reptiles
are largely unknown. Ayala (1971)
showed that the sandfly Lutzomyia vex-
trix was possibly the vector transmitting
Plasmodium mexicanum to the iguanid
lizard Sceloporus occidentalis in Cali-
fornia.

The parasite has two phases, a sexual
phase in the invertebrate vector and an
asexual phase in the vertebrate host.
Plasmodium gametocytes are picked up
by the vector when it takes a blood meal
from an infected host. Following fertil-
ization in the vector the zygote may
produce up to a thousand sporozoites
(Ayala, 1977) which invade the hemocoel
of the insect, some going to the salivary
glands. When the vector bites a lizard the
parasites are transferred when saliva is
injected and the lizard becomes infected.

In the lizard, the sporozoites enter red
blood cells (RBC’s) and become tropho-
zoites which grow, becoming either
sexual forms (gametocytes) or asexual
forms. Sexual forms remain quiescent
until ingested by a vector. Asexual forms
become schizonts and then segmentors,
dividing several times to produce many
merozoites which emerge, destroying the
host cell, and infecting new RBC’s.

During the first stage of the infection in
the lizard the number of parasites in the
blood cells is increasing and asexual
forms are predominant. This phase of the
infection is longer in reptiles than in
mammals and birds and takes from one to
two months to reach the peak of para-
sitemia (about 34 weeks in A. limifrons).
The level of parasitemia at this point
depends upon both the host and the
species of Plasmodium (Ayala, 1977).
(The highest we have seen in A. limi-
frons under natural conditions is 100 to
170 per 1,000 RBC’s.) The peak of the
infection is followed by a crisis period
when the parasitemia can still be high
and the host is responding to it by mas-

457

sive destruction of old RBC’s and re-
placement by new, immature RBC’s.
Sometimes animals wih high parasitemia
die because of the severe anemia (Ayala,
1977). (This apparently happened to two
of our captive A. limifrons.) The post-
crisis period shows a decrease in the
level of the infection with sexual forms
becoming more common than asexual
forms, indicating a low rate of parasite
reproduction. After the crisis the infec-
tion enters a chronic period.

The chronic period is the longest one,
and animals once infected may remain so
for the rest of their life time (Ayala, 1977).
Gametocytes are the common forms
found in the RBC’s and blood pictures
are very similar to those of healthy indi-
viduals. During this period, the infection
may be at so low a level that it is not
detected in the blood smears. Also the
final stages of schizogamy could possibly
occur in the hemopoietic sinuses with
subsequent reinvasion of the blood
(Ayala and Spain, 1976). (Most of the A.
limifrons caught in the wild with malaria
showed chronic infections.)

A lizard with chronic malaria may suf-
fer a relapse during which the parasites
rapidly reproduce asexually and levels of
parasitemia rise. Another crisis may then
occur with high levels of immature
RBC’s and, if the lizard survives, a return
of the infection to the chronic level.

MATERIALS AND METHODS

Intensive censusing, during which
lizards were marked, bled, measured,
and released, of two study areas on Barro
Colorado Island, Panama, continued for
about one year and less intensively for an
additional year. During the study, 735 A.
limifrons and 1,287 blood smears were
examined. Many anoles were recaptured
several times.

The study areas, census periods and
measurement techniques have been
described in detail in the preceding
paper (Andrews, Rand, and Guerrero,
this volume). In this paper we describe in

458

detail only those techniques specific to
the malarial study.

Sites. Two sites were used: Lutz,
characterized by a lower population
density of anoles and higher incidence of
malaria; and AVA, characterized by more
anoles and less malaria. Both were in
tropical forest.

Census. Lizard malaria information
was collected during 13 censuses (Table
1). For several analyses censuses were
grouped into wet and dry seasons. In
some cases, wet and dry season data from
different years were combined.

Class. For most analyses, five classes of
lizards were recognized: juveniles (SVL
< 40 mm), small males, small females (39
mm < SVL < 44 mm), large males, and
large females (43 mm < SVL).

Blood smears. Blood was obtained
from the tip of a toe when the terminal
phalanx was clipped as part of the mark-
ing procedure. No more than one blood
sample was taken from an animal during
a census. It was difficult to bleed young
animals without injuring them, so only in
cases when the blood was easily obtained
were blood smears made from animals
less than 30 mm in SV length. Standard
thin blood smears were fixed in methanol
and stained for 40 minutes with 5%
Giemsa solution pH 7.0-7.2 (Ayala and
Spain, 1976). Slides were searched, at
1,000 times magnification, long enough
to examine at least 20,000 RBC’s; often
the whole slide was searched before
recording it as negative. Once a Plas-
modium-infected cell was encountered,
the microscope field was shifted random-
ly and 1,000 RBC’s were examined, shift-
ing the field systematically. The numbers
of Plasmodium balli and P. “tropiduri”
present in these 1,000 RBC’s were re-
corded. Following Telford (1977), para-
sitemia levels below 10 per 1,000 were
considered low, probably indicating
chronic infections.

Next, 100 RBC’s were examined, and
the number which were immature was
recorded. The cytoplasm of an immature
RBC is darker than that of a mature
RBC’s and the nucleus is bigger. Arbi-

Advances in Herpetology and Evolutionary Biology

trarily, immature RBC counts were con- |

sidered high when they were equal to or
greater than 10 per 100.

The blood smears made from female
Anolis producing eggs show a very dis-
tinctive reddish coloration surrounding
the red blood cells. This, as described by
Ayala and Spain (1975), is due to the high
concentration of lipoprotein that appears
in the blood of reptiles at the end of the
hydration state of follicle growth and is
maintained during yolk deposition.

Species of Plasmodium. Four species
of Plasmodium have been described from
A. limifrons in Panama (Telford, 1974).
Because controversy exists about the
number and nomenclature of the species
present (Ayala, 1976; Ayala and Spain,
1976; Guerrero et al., 1977; Telford,

1977), we distinguished two morphospe- |

cies: Plasmodium balli belonging to the
Sauromoeba group and P. “tropiduri,” of
the Tropiduri group. In this latter cate-
gory we included parasites that might

have been classified as Plasmodium |

floridense, P. tropiduri or P. minasense

following Telford, 1974 and 1977 or more ©

recently as a single undescribed species,
Plasmodium sp. (Telford, 1979).

Plasmodium balli

This parasite is bigger than a red blood
cell nuclei; it does not show malarial
pigment in the cytoplasm and has a
vacuole. It causes marked distortion to
the host RBC or its nucleus (see Telford,
1969, for descriptions, and 1974, for fig-
ure, hosts and geographical distribution).

Plasmodium “tropiduri”

This parasite is approximately the size
of a RBC nucleus. It is oval or elongate in
shape with malarial pigment in the cyto-
plasm, and sometimes a vacuole. These
parasites do not visibly distort the host
cell or its nucleus (see Telford, 1974, for
descriptions, figures, hosts and geo-
graphical distributions of P. floridense, P.
tropiduri, and P. minasense and Telford,
1979 for a description of P. sp.). We be-

ANOLE MALARIA « Rand, Guerrero, and Andrews

lieve that most, if not all, of our “tropi-
duri’ infections involve a single species.

The insect vectors of these parasites
has not yet been positively identified.
The similarities in distribution of the two
parasites seen in this study suggests that
they share vectors, at least on Barro Colo-
rado Island. The most likely candidate
for vector is the phlebotomine fly,
Luzomya trinidadensis, that feeds on liz-
ards and is common in the study areas
(Kimsey, in preparation).

Statistical analysis of the data was
largely done at the Virginia Polytechnic
Institute and State University using the
Statistical Analysis System programs
(Helwig and Council, 1979). Because of
the unbalanced design, SAS general
linear model procedures were used for
the analysis of variance where the effects
of several variables were to be con-
sidered together.

RESULTS AND DISCUSSION

The results are divided into two sec-
tions. First, we consider the prevalenceof
malaria, levels of parasitemia, and the
species of Plasmodium as they varied
with site, times of sampling, and lizard
class as determined by lizard size and
sex. Second, we examine the effects of
malaria on the lizards considering its
prevalence, the species of malaria, and
the level of parasitemia on: 1) blood pic-
ture as measured by the proportion of
immature red blood cells; 2) physical
condition as indicated by the ratio of
body weight to the cube of length; 3) food
intake as estimated by scat weight; 4)
time of activity as measured by hour of
capture; 5) female reproductive activity
as measured by presence of lipoproteins
in the blood indicating oogensis; 6)
growth as measured by increase in snout
vent length in recaptured animals; 7)
survival as measured by recapture; 8)
intensity of predation as estimated by tail
breakage; and, finally, 9) differential
mortality among the size classes as indi-
cated by the size structure of the popula-
tion.

459

THE EFFECTS OF SITE, CENSUS, AND
CLASS ON MALARIA

Overall. The prevalence of malaria in
Anolis limifrons on Barro Colorado
Island was quite high. Plasmodia were
found in 28% of the 1,287 blood smears
examined. Of those with malaria, the
majority (67%) had low parasitemias
(<10/1,000), presumably indicating
chronic malaria; the remainder had
higher parasitemias, presumably indica-
ting acute phase malaria. The prevalence
of malaria is probably higher than these
figures suggest. Cage experiments sug-
gest that malaria is never lost but may be
present at such a very low level that it is
undetected in a blood smear (Guerrero
and Pickering, in preparation). Some
captive A. limifrons tested first negative
and then positive for malaria even though
the screened cages in which they were
kept made the possibility of a new infec-
tion very unlikely. It is possible that the
parasite may be absent from the circu-
lating blood but present in other tissues.
Even at levels below which we would
reliably detect blood parasites, and we
examined at least 20,000 RBC’s before
declaring a slide negative, a lizard could
still be infective to a vector taking a blood
meal. A phlebotomine fly typically takes
between 0.3 and 0.5 mg of blood, a
mosquito 2.0 mg (B. Chaniotis, personal
communication) and a mg of A. limifrons
blood contains about 76,000 RBC’s so
that a phlebotimine probably ingests at
least 22,800 RBC’s.

If malaria is never lost, we can estimate
the number of undetected infections
from the proportion of animals that tested
positive for malaria on one capture that
tested negative the next time they were
examined. At both Lutz and AVA this is
about 50%, suggesting that there are
about as many undetected malaria infec-
tions among the lizards that we examined
as there are infections that we detect.
This estimate is supported in another
way. Half of the 309 blood smears made
from large adult A. limifrons at Lutz were
positive for malaria; 79% of the 75 indi-

460

viduals that were tested two or more times
as large (43 mm < SVL) adults were posi-
tive for malaria at least once; 95% of the 36
tested three or more times; and all of the
14 tested four or more times. Probably
almost all of the large anoles at Lutz and
almost half of those at AVA had malaria at
some time during their lives and probably
continued to harbor the parasite through-
out their lives.

The two morphospecies of malaria, P.
balli and P. “tropiduri,’ were found
alone and in combination. Overall (N =
370), mixed infections were the most
commonly seen (54%), pure P. balli the
next most common (34%), and pure P.
“tropiduri’ the least (13%).

The effects of site, census and class
were considered separately, and three
aspects of malaria were analyzed: the
prevalence of malaria (Table 1), the fre-
quency of high levels of parasitemia
among infected animals (Table 2), and
the relative abundance of the two species
of Plasmodium (Table 3).

Site. The overall prevalence of malaria
was consistently higher at the Lutz site
(43%) than at AVA (19%) (N = 1287, X?, =
87.3, p < 0.0001). The proportion of in-
fections with high parasitemia were simi-
lar at both sites though somewhat higher
at Lutz (36%) than at AVA (27%) (N =
370, X?, = 3.34, p = 0.067). Mixed spe-
cies infections were relatively more
common at AVA (63%) than at Lutz (48%)
(N = 370, X?, = 5.42, p = 0.034). The two
species of Plasmodium occurred in about
the same proportions at the two sites (1 P.
“tropiduri’ to 1.35 P. balli at Lutz, 1 to
1.29 at AVA).

Lutz had about half as many lizards as
did AVA (Andrews et al., this volume).
This lower lizard density might be ex-
pected to reduce the chances of a vector
transmitting the disease from one Anolis
limifrons to another and so reduce the
prevalence of the disease. This did not
happen; at least 43% of the anoles at Lutz
and 19% at AVA had malaria. We initially
expected to find that the lower anole
population levels at Lutz were the result
of higher levels of malaria there. As will

Advances in Herpetology and Evolutionary Biology

be discussed below, we can not support
this idea on the data at hand. Even if it
were true, it would not explain why
malaria was more common at Lutz than at
AVA. If it is not an historical accident,
possibly the result of sampling a multi-
year disease cycle at different points at
the two sites, we can only suggest that it
might be due to differences in the two
host populations in their susceptibility to
the malaria or the result of differences in
vector populations. The Lutz site is on a
slope close to a small stream; AVA is on
level ground much further from water. It
is difficult to evaluate these differences
until we identify the vector.

Census. For analysis by contingency
tables censuses made during the same
wet or dry season were combined (see
Table 1 for details of the times of census-
ing and of seasons).

Prevalence of malaria changed over
time from census to census in the course
of the study (Table 1), with no correlation
between sites (N = 12, r= 0.12). No clear
seasonal pattern is evident, but our late

wet season samples were high at both —

sites. Lutz showed a significant change
among seasons censused (N = 517, X?, =
23.5, p = 0.015) because of a very high
prevalence in a single census; this did
not occur at AVA (N = 769, X?; = 13.6, p.
= 0.33).

The proportion of the infections with
high (> 9/1,000) or very high (> 19/1,000)
parasitemia changed over time. There is
a cross correlation between sites in these
changes (N = 12, r = 0.80). High para-
sitemia was more common during the
early wet season and became progres-
sively less so throughout the rest of the
wet season and into the dry season (Table
1). Looking only at the year of intensive
censusing, the differences between sea-
sons are significant at both sites (Lutz N
= 178, X2, = 37.5, p. < 0.001; AVA N =
128, X?, = 26.0, p. < 0.001). The three
successive early wet seasons sampled
show successively lower proportions of
high parasitemia (Table 1).

The high frequencies of high para-
sitemia presumably reflect high frequen-

ANOLE MALARIA : Rand, Guerrero, and Andrews 461

TABLE |. PREVALENCE OF MALARIA (THE PERCENTAGE OF LIZARDS EXAMINED THAT HAD MALARIA) AND LEVEL OF
PARASITEMIA (THE PERCENTAGE OF INFECTIONS WITH HIGH PARASITEMIA IE. > 9 INFECTIONS PER 1,000 RBC’s)
DURING SUCCESSIVE CENSUSES.

Infections with
High Parasitemia

A limifrons
with Malaria

Lutz AVA Lutz AVA

CENSUS SEASON N %o N % N % N %
6 May-14 Jun 76 Early Wet 36 «41.7 Syl WSs) 15 93.3 8 62.5
30 Jun-6 Aug 76 " 33. 45.5 45 17.8 15 80.0 8 62.5
1 Sep-21 Oct “76 Late Wet 43 46.5 55 = 2.1.8 20 ~=40.0 12 25.0
27 Oct-19 Nov 76 me 89 27.0 24 12.5
1-17 Dec 76 is 59 69.5 35 20.0 AL Blea amlArs
12-27 Jan 77 Dry 33 42.4 18 BD) 14 28.6 30 =. 20.0
923 Feb 77 ie 55 40.0 39 =.20.5 IES 8 37.5
16 Mar-7 Apr 77 rf 10 40.0 98 15.3 A 2510) LS ls3
20 Apr-12 May 77 i 59 4386 42.4 48 8.3 25 = 20.0 4 25.0
25 May-10 Jun 77 Early Wet DAR So. 79 11.4 QS Side 9 55.6

29 Jun4 Aug 77 4 36 =. 38.9 14 21.4 14) 14.3 3 0

18 Jan-9 Feb 78 Dry 65 40.0 59 16.9 26 =. 30.8 10 = 20.0
17 May-29 Jun 78 Early Wet 64 29.7 42 21.4 19 26.3 9 44.4

N = number examined.

cies of new infections or relapses. These
coincide with periods of general insect
abundance (see Andrews et al., this
volume) and may reflect times when the
malaria vector is particularly abundant,
rather than times when lizards are parti-
cularly susceptible to malaria. It is also a
time when food is most abundant and
when reproductive and social activity are
probably at their highest and the anoles
might be the most susceptible to malaria.
As one might expect, the peak in new
infections precedes the peak in total
prevalence of malaria.

The year to year changes make the
reality of the apparent seasonal pattern
hard to assess and are themselves hard to
interpret.

Class. The prevalence of malaria varies
with size and sex class (Table 2). It is
higher in males than in females, though
significantly so only at AVA (N = 629, X?,
= 19.1, p < 0.001; at Lutz N = 410, X?, =
3.49, p = 0.06), and higher in larger
adults than in small ones (Lutz N = 517,
X?, = 89.8, p = 0.001; AVA N = 770, X?,
36.7, p < 0.001). At both sites large males
are more likely to have malaria than is
any other size class. Small adults have

lower incidences of malaria. At both sites
large females are intermediate; at AVA
they are more like large males and at
Lutz more like small adults. At both sites
juveniles are less likely to have malaria
than are adults (Lutz N = 517, X?, = 62.7,
p < 0.001; AVA N = 770, X2, = 30.0).

The differences between size classes
are due, at least in part, to differences in
age because once a lizard has been infec-
ted with malaria, it probably rarely, if
ever, loses the parasite. It is not clear
why males have a higher prevalence of
malaria than do females. Perhaps males,
because they tend to use higher and more
exposed perches than do females (Talbot,
1979), come into more frequent contact
with vectors.

Levels of parasitemia also vary be-
tween classes (Table 2). These differ-
ences are not significant if the sites are
considered separately. However, at both
sites the pattern is the same and if they
are combined the difference between the
small and large adults is marginally signi-
ficant (N = 353, X?, = 5.24, p = 0.02).
Juveniles have the smallest proportion of
high parasitemia, small adults the highest
proportion, and large adults are inter-

462

Advances in Herpetology and Evolutionary Biologi
§ Y SY

TABLE 2. PREVALENCE OF MALARIA (THE PERCENTAGE OF LIZARDS EXAMINED THAT HAD MALARIA) AND LEVELS OF
PARASITEMIA (THE PERCENTAGE OF INFECTIONS WITH HIGH PARASITEMIA IE. > 9 INFECTIONS PER 1,000 RBC’s) AT

DIFFERENT SITES AND IN DIFFERENT LIZARD CLASSES.

A limifrons

Infections with

with Malaria High Parasitemia
N % N %

Lutz
subadults (SVL<40 mm) 107 9 10 20
small females (39<SVL<44) 54 28 15 AT
small males (39<SVL<44) 47 32 15 53
large females (43<SVL) 116 55 64 37
large males (43<SVL) 193 62 119 34
Total 517 43 223 36
AVA
subadults (SVL<40 mm) 141 5 7 14
small females (39<SVL<44) 76 7 5 40
small males (39<SVL<44) 77 17 11 46
large females (43<SVL) 201 7 34 29
large males (43<SVL) 75) 32 88 24
Total 770 19 147 27
Grand Total 1287 29 370 33

N = number examined.

mediate. If one compares the frequency
of high parasitemias among all smears
(not just those with malaria) juveniles
have lower frequencies than do adults
(Lutz N = 517, X2, = 19.4, p < 0.01; AVA
N = 770, X?, = 5.24, p < 0.05). This sug-
gests either that juveniles contract ma-
laria less frequently than do adults or that
the early stages of the infection are more
frequently fatal for juveniles. Males and
females are very similar in the proportion
of cases of malaria in which the level of
parasitemia is high. Though, as would be
expected, where there is a difference
males have more high parasitemias.

Both sexes and all three size classes are
essentially identical in the species of
malaria infecting them.

THE EFFECTS OF MALARIA ON THE
ANOLES

Blood Picture. Malaria, at least when
parasitemia levels are high, is associated
with an abnormally high percentage of
immature RBC’s. This usually occurs
during the crisis phase of an infection

and immediately following it, apparently
due to a large scale destruction and
replacement of RBC’s by the lizard in
response to a high level of parasitemia
(Ayala, 1977). In our study the number of
these immature RBC’s per 100 RBC’s
was routinely counted in all slides with
Plasmodium. Figure 2 shows a positive
correlation between the level of par-
asitemia and the proportion of the RBC’s
that are immature (Lutz N = 220, r =
0.45, p < 0.001, AVA N = 146, r = 0.42, p
< 0.001).

Testing simultaneously for the effects
of the two species of malaria and three
levels of parasitemia, we found (Table 4)
that Plasmodium species either alone, or
in interaction with level of parasitemia,
made no difference in the proportion of
IRBC’s. It is surprising that P. balli
which visibly distorts the nucleus of an
RBC apparently has no more effect than
does a P. “tropiduri” that looks to be
much less harmful to the host cell.

Testing simultaneously the effects of
the three levels of parasitemia, site, cen-
sus, class, and their possible interactions

ANOLE MALARIA -: Rand, Guerrero, and Andrews

oo
un
wy
co
Ee
Lil
oe
J
os
= e @
= ee
5S ®
ww @
® e
om ®
= ®
i © eco
= @ e
aa © e
ee
®@ e
ee8 @ 0 ®
e @ @
@ @@e
e ® )
© eee e
®
e
e @
@

Preise elie) Claire Wes) tals

463

LO L2 L4 LE
aioe eK 7S)

Figure 2. Percentage of red blood cells which are immature, related to number of plasmodia found per 1,000 red

blood cells in Anolis limifrons on Barro Colorado Island.

on the proportion of immature RBC’s, we
found (Table 5) that the level of par-
asitemia has as significant an effect as do
census, class, and the interaction be-
tween census and level of parasitemia.
Acute malaria produces an effect in
lizards which reduces the hemoglobin
levels in blood (Scorza, 1971). This
anemia presumably occurs in A. limi-
frons in cases where we find large per-
centages of IRBC’s. The effect of high
parasitemias on IRBC’s and presumably

on anemia is more severe in smaller
anoles than it is in larger ones. Any
reduction in hemoglobin must mean a
reduction in oxygen carrying capacity of
the blood and therefore, if nothing else,
of the lizard’s ability to repay an oxygen
debt quickly. Iguanid lizards are well
known for their ability to sustain activity
anaerobically (Bennett and Licht, 1972),
sO anemia might not affect a lizards abili-
ty to catch prey, to flee a predator or to
interact socially, but it should reduce its

snout-vent length cubed, was tested
(Table 6) for the effects of malaria species
and three levels of parasitemia, size, and
sex class of lizard and the interactions of
these variables. Class and season and
their interaction did produce effects at
the two sites but at neither site did ma-

464 Advances in Herpetology and Evolutionary Biology
TABLE 3. SPECIES OF MALARIA.
P. balli &
P. “‘trop- P. “trop-
P. balli iduri” iduri”’
N % N % N %
Lutz 83 37 106 48 34 315
AVA AL 92 63 14 #10

ability to recover quickly from such ac-
tivity: As the following pages will show,
we did not find this disability reflected in
such things as growth, survival, or repro-
duction.

Condition. Lizards with malaria were
as heavy as those without malaria. Condi-
tion, the weight of a lizard divided by its

laria show any effect. As we have de-
scribed in the previous paper, males
were generally lighter than females of
comparable length. Part, at least, of the
greater weight of the females was due to
the egg or eggs which they were produc-
ing. That females with malaria were no
lighter than those without is evidence
that they are equally reproductive.
Food intake. Anoles with malaria were

not getting measurably less food than

TABLE 4. STATISTICAL ANALYSIS SYSTEM. GENERAL LINEAR MODELS PROCEDURE—TYPE IV SUMS OF SQUARES.

Degrees of

Source Freedom

Dependent Variable: Immature Red Blood Cells

Species of Plasmodium 2

Level of Parasitemia 1

Species * Parasitemia 2

Model 5

Error 363
* = interaction between variables.

Sum of
Squares F Value Pr. F
106.40 1.29 0.28
1215.06 29.54 0.0001
3.74 0.05 0.96
2383.00 11.59 0.0001
14930.59

TABLE 5. STATISTICAL ANALYSIS SYSTEM. GENERAL LINEAR MODELS PROCEDURE—TYPE IV SUMS OF SQUARES.

Degrees of

Source Freedom

Dependent Variable: Immature Red Blood Cells

Site 1
Census 12
Size & Sex Class 4
Level of Parasitemia 9)
Site * Parasitemia 1
Census * Parasitemia 12
Class * Parasitemia 4
Census * Class * Prsta 50
Model 86
Error 283

Sum of

Squares F Value Pr. F
105.89 3.49 0.06
809.47 222 0.01
779.25 6.42 0.0001
387.45 6.38 0.002

0.50 0.02 0.90

813.71 DIS 0.01
218.52 1.80 0.13

2194.46 1.45 0.03

8791.77 B87 0.0001

8593.17

*%

= interaction between variables.

ANOLE MALARIA - Rand, Guerrero, and Andrews

465

TABLE 6. STATISTICAL ANALYSIS SYSTEM. GENERAL LINEAR MODELS PROCEDURE— TYPE IV SUMS OF SQUARES.

Degrees of Sum of
Source Freedom Squares F Value Pe, |e

Dependent Variable: Condition (Weight/SVL?) at Lutz
Size & Sex Class 4 0.0005 7.03 0.0001
Season 5 0.002 28.10 0.0001
Malariat 6 0.00008 0.74 0.62
Class * Season 19 0.0007 1.99 0.008
Class * Malariat 19 0.0001 0.36 0.99
Season * Malariat 26 0.0003 0.75 0.81
Clss * Ssn * Malariat 21 0.0003 0.90 0.59
Model 100 0.0097 5.48 0.0001
Error 412 0.007
Dependent Variable: Condition (Weight/SVL?) at AVA
Size & Sex class 4 0.0004 6.03 0.0001
Season 5 0.001 15.67 0.0001
Malariat 6 0.00002 0.21 0.97
Class * Malariat 18 0.0007 2.31 0.002
Class * Malariat 13 0.0002 0.86 0.59
Season * Malariat 17 0.0001 0.48 0.96
Clss * Ssn * Malariat 16 0.0003 1.00 0.46
Model 79 0.014 10.39
Error 686 0.012

tSeven categories used: no malaria and low (< 10/1,000 RBC’s) and high (> 9/1, DUD RBC’s) levels of parasitemia for Plasmodium balli, P

tropiduri, and both species of Plasmodium.
* = interaction between variables.

those without. Food intake was estimated
by the weight of scat per gram of lizard
produced by a lizard over the 24 hours
following capture. A log transformation of
the data makes the samples more nearly
normally distributed. A test (Table 7) on
the transformed data shows that only liz-
ard class had a significant effect on food
intake. At Lutz but not at AVA, juveniles
and adult females produced a relatively
greater weight of feces than did adult
males.

Time of activity. Malaria did not seem
to affect the time of day at which the liz-
ards were active. Each census was di-
vided into three periods: before 9 am, be-
tween 9 and 10 am, and after 10 am. Con-
tingency tables were constructed to com-
pare the times at which lizards were
caught. At neither site and in neither the
wet nor the dry season did the presence
of malaria (Wet season: Lutz N = 293, X?,

= 0.33, p = 0.85, AVA N = 381, X?, =
0.05, p = 0.98; Dry season: Lutz N = 221,
X?, = 0.74, p = 0.69, AVA N = 378, X*, =
282, p = 0.24) or the level of parasitemia
(Wet season: Lutz N = 293, X?, = 7.85, p
= 0.097, AVAIN = 381, 2, = 1.90. p =
0.75; Dry season: Lutz N = 221, X*, =
3.72, p = 0.45, AVA N = 378, X2, = 3.84,
p = 0.43) show a significant effect. Liz-
ards with malaria were not compensating
for it by being active earlier nor were
they debilitated by it and becoming ac-
tive later.

Reproduction. Reproduction did not
seem to be depressed in those females
showing malaria. The proportion of fe-
males which tested positive for oogenesis
varied with size (Table 8). Those tested
below a SVL of 35 mm were never posi-
tive; those few above 48 mm always
were; at intermediate sizes the propor-
tion that were oogenic varied, increasing

466

Advances in Herpetology and Evolutionary Biology

TABLE 7. STATISTICAL ANALYSIS SYSTEM. GENERAL LINEAR MODELS PROCEDURE—TYPE IV SUMS OF SQUARES.

Degrees of

Source Freedom

Dependent Variable: Food Intake (Log of Scat Weight) at Lutz

Size & Sex Class 4
Season 5
Malariat 6
Class * Season 19
Class * Malariat 19
Season * Malariat 25
Clss * Ssn * Malariat 13
Model 91
Error 279

Dependent Variable: Food Intake (Log of Scat Weight) at AVA

Size & Sex Class 4
Season 4
Malariat 6
Class * Malariat 14
Class * Malariat 11
Season * Malariat 12
Clss * Ssn * Malariat 7
Model 58
Error 437

Sum of
Squares F Value Pr. F.
8.19 6.21 0.0001
2.78 1.69 0.14
1.26 0.64 0.70
8.28 1.32 0.17
3.61 0.61 0.89
7.84 0.95 0.53
5.36 1.25 0.24
41.40 1.38 0.02
91.98
3.26 1.96 0.10
2.04 1.22 0.30
0.73 0.29 0.94
5.70 0.98 0.47
4.78 1.04 0.41
3.76 0.75 0.70
3.69 1.27 0.26
181.79 1.68 0.0023
222.27

Seven categories used: no malaria and low (< 10/1,000 RBC’s) and high (> 9/1,000 RBC’s) levels of parasitemia for Plasmodium balli, P.

“tropiduri,’ and both species of Plasmodium.
* = interaction between variables.

as SVL increased. Once an Anolis limi-
frons begins to yolk eggs she usually con-
tinues to do so. Only 13% of the 93 fe-
males tested again after once testing posi-
tive tested negative on the second occa-
sion.

A two by two contingency table shows
a significant effect of the presence of
malaria on the frequency of oogenesis in
potentially reproductive (36 mm SVL)
females (N. =) 504) 32) = 2529) p =
0.0001). Surprisingly the effect is posi-
tive, a higher proportion of the females
with malaria are oogenic than of those
without malaria.

A tabulation (Table 4) of oogenic fe-
males with and without malaria shows no
negative effect of malaria on oogenesis
and suggests that the apparent positive
effect is possibly a spurious side effect of
the fact that both the prevalence of ma-

laria and the frequency of oogenesis in-
crease with size. A similar lack of effect of
malaria on oogensis was seen in Anolis
auratus infected with Plasmodium
colombiense in Colombia (Ayala and
Spain, 1976).

Looking at the effect of malaria in
another way, we can consider only those
adult females that had a very high (> 19
per 1,000 RBC’s) parasitemia or a high (>
9 per 100) percentage of IRBC’s. Of these
24 anoles, all but one tested positive for
oogenesis. Therefore even the acute
phase of malaria does not inhibit
oogenesis to a measurable extent. Pos-
sibly the changes in the blood associated
with egg production favors assexual re-
production in the Plasmodium so that
parasites become more common in
oogenic anoles.

Our evidence of reproductive activity

ANOLE MA.ariA : Rand, Guerrero, and Andrews

TABLE 8. THE EFFECTS OF MALARIA ON OOGENESIS.

Female Anolis limifrons
Without Malaria With Malaria

Snout Vent

Length (mm) N %Oogenic N % Oogenic
21 1 0
BP) 1 0
23
24
25 1 0
26
a 1 0
28
29 ] 0
30 2 0
31 2 0
32 6 0 1 0
33 5 0 1 0
34 11 0
35 5 20.0
36 9 0
37 12 16.7 1 0
38 10 20.0 4 50.0
39 10 50.0 1 0)
40 22 45.5 2 50.0
Al 20 30.0 6 66.7
42 28 64.3 2 100.0
43 28 60.7 0 100.0
44 28 89.3 8 100.0
45 49 89.8 20 95.0
46 33 90.6 24 87.5
AT 38 84.2 18 94.4
48 14 85.7 11 100.0
49 7 100.0 3 100.0
30 5 100.0 2, 100.0
al 1 100.0 1 100.0
52 2 100.0

in these anoles in the field is whether
their blood smears stained reddish indi-
cating that they were circulating lipopro-
teins and therefore presumably yolking
eggs. One can be reasonably sure that
females that are not yolking eggs are not
actively reproductive. However, there is
no way of knowing from our field data if
two oogenic females are producing eggs
at the same rate. There is some evidence
from females maintained in captivity.
Seven captive females with malaria laid
as many eggs (Mean = 0.79 per female
per week) of the same size (Mean weight
= 0.14 g, N = 27) as did 12 similar sized
females without malaria (Mean = 0.81

467

eggs per female per week; Mean weight
= (0.14 g, N = 48) (Andrews, unpublished
data). .

Growth. Anoles with malaria seem to
grow as rapidly as those without. A test
(Table 9) of the effects, on growth rate, of
size and the presence of malaria (at the
beginning, or end, or both of the growth
interval) showed a strong effect of size (p
= 0.0001) but none of malaria, alone or in
interaction with size. The lower growth
rates in malarial animals, which one sees
when one groups animals without regard
for size, seems to be due to a higher
prevalence of malaria among larger ani-
mals (as noted above) and that larger
animals grow more slowly than do smal-
ler ones (Andrews et al., this volume).

Survivorship. Malaria did not appear to
decrease the survivorship of infected
anoles. To determine if malaria affected
the likelihood that a lizard would disap-
pear from the population, contingency
tables were prepared for each site com-
paring the numbers of lizards caught in
each census with and without malaria,
with our records as to whether they had
been recaptured during the same census,
recaptured at some subsequent census or
never captured again and had probably
died or emigrated. At neither site was
there a significant difference in the like-
lihood of recapture during the same or
during a subsequent census due to hav-
ing malaria (Lutz N = 517, X?, = 1.80, p
= 0.41, AVA N = 770, X?, = 1.72, p =
0.42), or to level of parasitemia (Lutz N =
517, X?, = 3.12, p = 0.54, AVA N = 770,
X?, = 1.96, p = 0.74). Malaria did not
seem to affect either conspicuousness or
survivorship.

Predator pressure. Anoles with malaria
seemed to be no more at risk from preda-
tors than those without malaria. To esti-
mate intensity of predation, frequencies
of regenerated tails were tabulated for
most captured anoles. Only the condition
of the tail on the last occasion when a
lizard was captured was scored to avoid
counting the same break twice. Com-
bining sites, size, and sex classes shows a

468 Advances in Herpetology and Evolutionary Biology

TABLE 9. STATISTICAL ANALYSIS SYSTEM. GENERAL LINEAR MODELS PROCEDURE— TYPE IV SUMS OF SQUARES.

Sum of
Source DF Squares F Value Pr. F |
Dependent Variable: Growth Rate :
Lutz, Wet Season |
Snout Vent Length 1 0.87 72.41 0.0001
Presence of Malaria 1 0.001 0.89 0.35 |
SVL * Malaria ] 0.001 1.01 0.32
Model 3 0.12 BOT 0.0001 |
Error 34 0.04
Lutz, Dry Season
Snout Vent Length 1 0.31 20.95 0.0001 |
Presence of Malaria 1 0.00002 0.01 0.92
SVL * Malaria 1 0.00001 0.01 0.93
Model 3 0.04 8.71 0.0004
Error 24 0.04
AVA, Wet Season
Snout Vent Length 1 0.26 10.98 0.0017
Presence of Malaria 1 0.0000004 0.00 0.99
SVL * Malaria 1 0.0000002 0.00 0.99
Model 3 0.04 5.53 0.002
Error 49 0.11
AVA, Dry Season
Snout Vent Length 1 0.004 1.71 0.20
Presence of Malaria 1 0.0005 0.22 0.64
SVL * Malaria 1 0.0005 0.24 0.63
Model 3 0.06 9.06 0.0001
Error 63 0.14

DF = Degrees of Freedom
* = interaction between variables

significant effect of malaria on frequency
of tail breaks (overall % broken = 19.2, N
= 640, X?, = 8.35, p = 0.004). This is
probably spurious because, as one might
expect, at both sites, larger animals are
more likely to have their tails broken
than are smaller ones (Lutz N = 272, X?,
= 8.67, p = 0.013); AVA N = 368, X?, =
10.56, p = 0.005). Larger lizards should
more frequently have broken tails if only
because they are generally older than
small ones and the scar left by a broken
tail never disappears.

For all lizard classes where the sample
size is large enough to allow testing
(large males at Lutz (% broken = 26.3, N

= 99, X?, = 1.16, p = 0.28), and large
males (% broken = 22.0, N = 132, X?, =
0.002, p = 0.97) and females at AVA (%
broken = 31.0, N = 100, X?, = 0.06, p =
0.81)), the malarial state when last cap-
tured had no significant effect on the fre-
quency of tail breakage.

Among large adults, males have their
tails broken more often than do females
but the difference is small and not signi-
ficant (Lutz N = 159, X?, = 2.76, p =
0.097; AVA N = 232, X2,
0.12). Because males fight more than do
females, if aggressive encounters be-
tween conspecifics were an important
cause of tail breakage one would expecta

II
bo
aN
us
uo)

|

ANOLE MALARIA « Rand, Guerrero, and Andrews

substantially higher percentage of tail
breaks among males than among females.
We do not find this and so aggressive
encounters are probably not important.
Most tail breaks probably come from
failed predation attempts. It is certainly
not necessary that there be a close cor-
relation between failed and successful
attempts, but it seems likely that this
would be true except where the sample
includes differing proportions of naive
predators or of predators with different
capture techniques. In our situation there
could possibly be a seasonal difference in
predators but there is unlikely to be a site
difference and we can think of no way in
which either of these would bias the pic-
ture either for or against anoles with
malaria.

Differential mortality. Malaria did not
have a detectable differential effect on
the mortality of the different size and sex
classes of anoles. One possible effect of
malaria which we have not yet con-
sidered is that it might kill a significant
number of lizards during the early stages
of the infection, leaving the others to re-
cover essentially unaffected even though
still circulating the parasite. Two of the
several score of naturally infected anoles
which we kept in captivity died during
the crisis stage of their malaria infection.
The malaria-induced anemia may well
have been at least a contributing factor.

We can not test for a general effect of
this sort but we might detect an effect if
the malaria was affecting different lizard
classes differently. One might expect that
if malaria was sometimes fatal it might
kill a higher percentage of young anoles
than older ones, or kill more of the egg
producing females than males. If this is
so one would expect that the classes most
strongly affected should be proportion-
ately rarer at Lutz than at AVA where
malaria is much less common. A compari-
son of size and sex classes at the two sites
shows no siginficant difference between
them during either dry (N = 601, X?,, p =
0.53) or wet (N = 686, X?,, p = 0.40) sea-
sons.

469

There appears to be no statistically
significant difference of the effect of
malaria on the two sexes, or on the size
classes. However, this approach would
not detect an effect in which malaria was
killing proportionally more animals in
the smallest size class. This early mor-
tality would affect all succeeding size
classes as well and so appear as a general
lowering of the population level without
changing the size structure, unless it was
compensated for by reduced mortality at
later stages. Since we do not find a differ-
ence between the two sites except that
the site with more malaria has fewer liz-
ards, we can not exclude the possibility
that malaria has a significant negative
effect on the survival of an infected
anole. We can say that if malaria does
have such an effect, it occurs during the
early stages of the infection, and that if
the effect is different on different age
classes it must be more severe on young
animals.

We found lower frequencies of high
levels of parasitemia (new infections)
among juveniles than among adults. This
is consistent with the hypothesis that
malaria is more serious for younger than
for older animals but it is also consistent
with the hypothesis that small lizards are
less attractive to the vectors and so are
infected less frequently.

SUMMARY AND CONCLUSIONS

Malaria was extremely common among
Anolis limifrons during our study. In one
population 95% of the large adults cap-
tured three or more times had a patent
malaria infection at least once. Two
morphologically different types of
Plasmodium were involved which were
similar in their distribution in time, space
and class of A. limifrons and both were
frequently present together in the same
individual.

The disease was present at all times of
year, and there did not seem to be
marked seasonality in either overall

470

prevalence or frequency of high levels of
parasitemia indicative of new infections,
though the latter were most common in
early rainy season.

Malaria was much more common
among large individuals than among
small ones. Individuals seemed, once
infected, to continue to circulate the
parasite for a long time, perhaps through-
out their lives, usually at a low level,
though relapses with high levels of para-
sitemia were observed in captive ani-
mals. The higher prevalence in large
animals is in part the result of this per-
sistence of the parasite. If the probability
of being infected remained above zero
throughout life then the probability of
ever being infected would increase with
age and older (larger) animals would
have, as we find that they do, a higher
prevalence of infection.

Since the frequency of high levels of
parasitemia was higher among small
adults than large ones, the probability of
contracting the infection is as high, or
higher, for a small adult as it is for a larger
adult. The prevalence of malaria and the
occurrence of high levels of parasitemia
was low in the juveniles. Either the
probability of contracting malaria is low
for these small animals, or malaria when
contracted is more likely to be fatal to a
small animal than it is to a large animal.

Chronic malaria with low levels of
parasitemia produced little effect on the
blood picture. Higher levels of par-
asitemia correlated with high levels of
immature RBC’s and this in turn presum-
ably reflected a high level of anemia.
This effect was stronger for small animals
than for larger ones.

We examined a variety of other aspects
of the anoles from the two study sites in-
cluding: weight, food intake, time of ac-
tivity, female reproductive condition,
growth, survivorship, and tail breakage.
Despite large sample sizes we found no
factor, except IRBC’s, on which malaria
had a statistically significant effect on the
lizards in which we detected it, whether
we look at all malaria infections or only

Advances in Herpetology and Evolutionary Biology
§& 8Y

the relatively small number of cases with
high parasitemias.

Apparently malaria does not have a
marked effect on the anoles that it infects,
at the very least this is true for adults with
chronic malaria. This is what one would
expect of a well adapted parasite in a
closely coevolved host-parasite relation-
ship. It is to the parasite’s advantage, as
well as to the advantage of the host, for
the host to live for a long time and leave
many offspring. However, we can not
conclude, with certainty, that malaria has
no ecologically significant effect on these
populations, because we can not exclude
the possibility that the initial stages of a
malaria infection, which are rarely repre-

sented in our sample, may be fatal, at —

least to young animals. Still, at least in its
chronic phase, malaria in Anolis limi-
frons seems to be behaving as a prudent
parasite.

ACKNOWLEDGMENTS

We thank Steve Ayala for his advice
and counsel on lizard malaria during the
planning stages as well as in the course of
the project. A large number of people
helped in the field, among them: John
Pickering, Bonifacio deLeon, Gordon
Burghardt, Lauri McHargue, Sharon
Strauss, Hal Hertzog, Marina Guerrero;
others assisted in the lab: Humberto and
Fanny Carvajal and Luis Carlos Zamora;
Don Windsor at STRI and Ned Fetcher,
Biology Department and W. R. Price, the
Statistical Consulting Service of VPI&SU
advised on statistical analysis.

We are indebted to STRI for logistic
and financial support, to National
Science Foundation Binational Grant No.
INT 7605758 to Rand and Ayala, National
Institute of Health Grant 2 ROI] AI12511
to Ayala, and a Smithsonian Institution
Research award to Rand for financial
support.

We thank John Pickering, Emest E.
Williams, Joe Wright, and Bob Kimsey
for stimulating discussions and valuable

ANOLE MALARIA : Rand, Guerrero, and Andrews

suggestions on the problems of lizard
malaria and anole ecology, and those who
read and commented on the manuscript:
John Pickering, Joe Wright, Pat Rand,
Bob Kimsey, Mike Ryan, Steve Ayala,
and Joe Schall.

LITERATURE CITED

ANDERSON, R. C. 1976. Helminths, pp. 35-43. In L.
A. Page (ed.), Wildlife Diseases. Section II.
Impact of parasitic diseases on wildlife popula-
tions. New York, Plenum Press.

ANDREWS, R. A., AND A. S. RAND. 1982. Long-term
changes in population density of the lizard
Anolis limifrons. In E. G. Leigh, Jr., A. S.
Rand, and D. M. Windsor (eds.), The Ecology
of a Tropical Forest: Seasonal Rhythms and
Long-term Changes. Washington, D.C.,
Smithsonian Institution Press. (in press).

ARAGAO, H. DE B., AND NEIvA, A. 1909. A contribu-
tion to the study of the intraglobular parasites
of the lizards. Two new species of Plas-
modium, Pl. diploglossi n. sp. and Pl. tropiduri
n. sp. Mem. Inst. Oswaldo Cruz, 1: 44-50.

AYALA, S. C. 1971. Sporogony and experimental
transmission of Plasmodium mexicanum. J.
Parasitol., 57: 598-602.

____.. 1977. Plasmodia of reptiles, pp. 267-309. In J.
P. Kreier (ed.), Parasitic Protozoa. Vol. III.
Gregarines, Haemogregarines, Coccidia,
Plasmodia and Haemoproteids. New York,
Academic Press.

__. 1978. Checklist, host index, and annotated
bibliography of Plasmodium from reptiles. J.
Protozool., 25: 87-100.

AYALA, S. C., AND J. L. SPAIN. 1975. Annual oogene-
sis in the lizard Anolis auratus determined by
a blood smear technique. Copeia, 1975: 138-
141.

aa AND . 1976. A population of Plasmodium
colombiense sp. n. in the iguanid lizard, Anolis
auratus. J. Parasitology., 62: 177-189.

BENNETT, A., AND P. LICHT. 1972. Anaerobic me-
tabolism during activity in lizards. J. Comp.
Physiol., 81: 277-288.

BENNETT, G. F., E. C. GREINER, AND W. THRELFALL.
1976. Protozoans, pp. 25-33. In L. A. Page
(ed.), Wildlife Diseases. Section II. Impact of

A471

parasitic diseases on wildlife populations.
New York, Plenum Press.

HELWwIG, J. T., AND K. A. COUNCIL. 1979. SAS User’s
Guide, 1979 Edition. SAS Institute Inc.,
Raleigh.

JORDAN, H. B. 1964. Lizard malaria in Georgia. J.
Protozool., 11: 566-572.

GUERRERO, S., C. RODRIGUEZ, AND S. C. AYALA.
1977. Prevalencia de hemoparasitos en
lagartijas de la Isla Barro Colorado, Panama.
Biotropica, 9: 118-123.

Markus, M. B. 1974. Arthropod-bome disease as a
possible factor limiting the distribution of
birds. Intern. J. Parasit., 4: 609-612.

PICKERING, J. 1980. Sex ratio, social behavior and
ecology in Polistes (Hymenoptera, Vespidae),
Paschysomoides (Hymenoptera, _Ichneu-
monidae) and Plasmodium (Protozoa,
Haemosporida). Ph.D. Thesis. Harvard Uni-
versity, Cambridge, Mass.

SEXTON, O. J. 1967. Population changes in a tropical
lizard Anolis limifrons on Barro Colorado Is-
land, Panama Canal Zone. Copeia, 1967: 219-
2225

SEXTON, O. J., J. BAUMAN, AND E. ORTLEB. 1972.
Seasonal food habits of Anolis limifrons.
Ecology, 53: 182-186.

SEXTON, O. J., J. HEATWOLE, AND E.. MESETH. 1963.
Seasonal population changes in the lizard,
Anolis limifrons, in Panama. Am. Midland
Natur., 69: 482-491.

SCORZA, J. W. 1971. Anaemia in lizard malaria infec-
tions. Parassitologia, 13: 391-405.

TELFORD, S. 1969. A new saurian malarial parasite
Plasmodium balli from Panama. J. Protozool.,
16: 437-441.

___. 1974. The malarial parasites of Anolis species
(Sauria: Iguanidae) in Panama. Inter. J.
Parasitol., 4: 91-102.

____. 1977. The distribution, incidence and general
ecology of saurian malaria in Middle America.
Intem. J. Parasitol., 7: 299-314.

____. 1979. A taxonomic reconsideration of some
Plasmodium species from iguanid lizards. Ann.
Parasitol. (Paris), 54: 129-144.

WARNER, R. E. 1968. The role of introduced di-
seases in the extinction of the endemic Ha-
waiian avifauna. Condor, 70: 101-120.

WRIGHT, S. J. 1979. Competition between in-
sectivorous lizards and birds in central
Panama. Amer. Zool., 19: 1145-1156.

Eurythermy and Niche Breadth

jal o
=)

West Indian Anolis Lizards:

A Reappraisal

PAUL E. HERTZ!

ABSTRACT. A comparison of the thermal biologies
of the solitary species Anolis roquet on Martinique
and three species from the complex fauna of Puerto
Rico, A. cristatellus, A. evermanni, and A. gund-
lachi, demonstrates no consistent differences in
thermal niche breadth that can be attributed to A.
roquet’s occupation of an island in the absence of
congeneric competitors. Three behavioral-ecologi-
cal “phenotypes” with respect to thermal biology
are described: 1) thermal passivity; 2) obligate
thermoregulation; and 3) thermoregulatory versa-
tility. Anoles from the complex faunas exhibit any of
the three phenotypes, but solitary anoles all appear
to be versatile, at least on thermally complex is-
lands. Eurythermy and thermoregulatory versatility
probably represent an ancestral “colonizing
phenotype” in West Indian anoles. The thermal
biology of solitary anoles may represent a retention
of this ancestral phenotype rather than a product of
ecological release.

INTRODUCTION

The concept of “ecological release”
was developed by ecologists to explain
the apparent increase in the range of re-
sources utilized by island populations
relative to that used by conspecifics on
the mainland (Wilson, 1961). In theory,
“solitary” populations are relatively free
to exploit a wider variety of resources be-
cause of the absence (or near absence) of
closely related competitors in depauper-
ate island faunas.

1 Department of Biological Sciences, Barnard Col-
lege, Columbia University, New York, New York
10027, U.S.A.

During the last decade, researchers
studying the ecology of Anolis lizards in
the West Indies have extrapolated from
the original formulation of the concept to
compare patterns of resource utilization
among the island-dwelling members of
the genus. These researchers have con-
cluded that solitary anoles generally ex-
hibit expanded structural habitat (i.e.,
perch site) and/or thermal niches relative
to those of species from the complex
faunas of the Greater Antilles (e.g.,
Lister, 1976; Rand and Rand, 1967;
Ruibal and Philibosian, 1970). Proof of
ecological release in the strictest sense
(Grant, 1967) is usually lacking, however,
in the absence of ecological data on the
ancestors of the solitary populations
(Huey and Webster, 1976; Ruibal and
Philibosian, 1970).

The presumed expansions of structural
habitat and thermal niches of the solitary
species are thought to be reflections of a
single phenomenon: solitary species
occupy a greater variety of habitats and
microhabitats. In the complex Anolis
faunas of the Greater Antilles, interspe-
cific resource partitioning of structural
and climatic habitats suggests that com-
petitors restrict the range of thermal
environments that some species may
occupy (Rand, 1964; Schoener and
Schoener, 197la; Rand and Williams,
1969). Indeed, Ruibal and Philibosian
(1970) explicitly stated that the relatively
broad thermal niche of the solitary spe-

THERMAL NICHE BREADTH IN ANOLIS : Hertz

cies A. oculatus on Dominica resulted
from its occupation of “an array of habi-
tats which on Cuba or Puerto
Rico...are occupied by different spe-
cies. One could logically argue that the
occupation of a large array of microhabi-
tats would also result in an expansion of
the thermal niche.

The broad thermal niche of a species
could be produced by either of two
phenomena (Huey and Webster, 1975).
On the one hand, each population of the
species might be eurythermal (i.e., hav-
ing a broad thermal niche), exhibiting a
niche breadth nearly as great as that
shown by all populations of the species
considered together. On the other hand, a
broad species thermal niche could be
produced by population differentiation
among the niche breadths of several
stenothermal (i.e., having a narrow
thermal niche) populations living under,
but not behaviorally compensating for,
distinct climatic regimes (Hertz, 1977).
The distinction of “within-population”
and “among-population” components of
a broad species niche is analogous to
Roughgarden’s (1972) “within-pheno-
type’ and “between-phenotype” com-
ponents of a population’s niche breadth.

Although previous researchers have
considered population stenothermy and
eurythermy in their comparisons of
thermal niche breadths among Anolis
species (Huey and Webster, 1975, 1976;
Ruibal and Philibosian, 1970), none has
made and tested explicit predictions
about both within- and among-popula-
tion niche breadths in solitary versus
sympatric species (but see Lister, 1976).
A systematic knowledge of these
phenomena is essential to our under-
standing of the evolution and bioge-
ography of the huge West Indian radia-
tion of this genus (Williams, 1969, 1972).
The purpose of this paper is to make and
test two such predictions, one at the
population level and another at the spe-
cies level.

1) If it is true that populations of soli-
tary anoles are more eurythermal and

473

occupy broader arrays of microhabitats
than do populations in the complex
faunas, then we should observe that the
within-population niche breadths of the
solitary species are at least equal to the
combined population niche breadths of
two or more sympatric species in the
complex fauna.

2) If it is true that the solitary species
occupy broader arrays of habitats than do
species in the complex fauna, then we
should observe that the among-popula-
tion niche breadth (hereafter, the species
niche breadth) of the solitary species is at
least equal to the combined species
niche breadths (among populations and
among species) of those species that
replace each other over an equivalent
range of habitats in the complex fauna.

To test these predictions, I systemati-
cally compared the structural habitats
and field thermal biology of a solitary
Anolis of the Lesser Antilles with those
of other Anolis species that are, at least
superficially, its ecological equivalents
in the complex fauna of Puerto Rico. On
Martinique, roquet is the dominant
terrestrial vertebrate in many habitats
from sea level to approximately 750 m
elevation (Lazell, 1972). On Puerto Rico,
cristatellus is the dominant trunk-
ground anole of the lowlands, and
gundlachi its upland replacement in
montane forests up to approximately
1,000 m elevation. Two other species,
stratulus and evermanni, also commonly
occupy tree trunks in the Puerto Rican
lowlands and uplands, respectively
(Rand, 1964); however, stratulus was
rare at the lowland cristatellus study site,
and I did not sample it there.

I sampled along altitudinal transects on
both islands. Although such a sampling
regime does not ensure the representa-
tion of every habitat type present on an
island, altitude does provide predictable
gradients in ambient temperature, cloud
cover, moisture, vegetation, and, on
Puerto Rico, community composition.

If the predictions made above are cor-
rect, I would expect the following rela-

474

tionships to emerge from the data on
structural habitat and thermal niche
breadths. 1) The niche breadths of indi-
vidual roquet populations should be
greater than those of individual popula-
tions of cristatellus, gundlachi, and
evermanni and at least equal to the com-
bined population niche breadths of
sympatric evermanni and gundlachi
populations. 2) The overall species niche
breadth of roquet (among populations
sampled along the transect) should be at
least equal to the combined species
niche breadths of cristatellus and
gundlachi sampled over an equivalent
altitudinal range.

MATERIALS AND METHODS

FIELD STUDIES

I sampled three populations of roquet
(at 20 m, 350 m, and 650 m elevations) in
northeastern Martinique during late June
and early July 1975. In Puerto Rico, I
studied a lowland population of crist-
atellus on the western outskirts of
Luquillo (5 m) and four populations (at
270 m, 480 m, 675 m, and 850 m) of its
upland replacement, gundlachi, in the
Luquillo Experimental Forest during
July, August, and September 1975. In
addition, I simultaneously sampled
evermanni, sympatric to gundlachi, at the
two intermediate upland sites. The loca-
tions of the study sites, brief descriptions
of the vegetation at each, and the sub-
specific identifications of the roquet
populations are provided elsewhere
(Hertz, 1977, 1981).

I obtained data records from adults and
subadults captured with standard tech-
niques and precautions (Hertz, 1977)
from dawn to dusk at each locality. For
each individual captured, I recorded the
following data: sex, snout-to-vent length,
body (cloacal) and air (shaded bulb, 1 cm
above perch site) temperatures with a
quick reading Schultheis thermometer,
perch height, and time of capture.

Advances in Herpetology and Evolutionary Biology

Samples varied from 132 to 180 records
per population and were uniformly dis-
tributed throughout the animals’ full
periods of activity at each site. Because
there were neither size- nor sex-
dependent trends in any population, I
have lumped all records within each
population in this analysis. A treatment of
other aspects of the field data can be
found elsewhere (Hertz, 1977, 1981).

ESTIMATION OF NICHE BREADTHS

The estimation of niche breadth for the
structural habitat of each population is
straightforward. I follow Lister (1976) in
using perch height diversity as an index
of niche breadth along this resource
dimension, using the formula B = 1/2 pi’
(Levins, 1968), in which pi is the fre-
quency with which lizards perched in
each one-foot height interval.

Calculation of thermal niche breadth is
much more complex, largely because
environmental heterogeneity can
strongly affect the variability of body
temperatures in a population (Soule,
1963; see also Huey and Slatkin, 1976;
Huey, 1982). Hence, one must consider
both environmental and body tempera-
tures to accurately describe thermal spe-
cialization. For the data presented here, I
follow Ruibal and Philibosian (1970) and
Huey and Webster (1975, 1976) in using
the range of hourly mean body tempera-
tures to estimate thermal niche breadth.
Similarly, the range of hourly mean air
temperatures indexes the environmental
heterogeneity (hereafter, the thermal
habitat breadth) experienced by each
population. However, the range of hourly
mean air temperatures is an imperfect
measure of environmental heterogeneity
because air temperatures only approxi-
mate operative environmental tempera-
tures (Bakken, 1976). Furthermore, the
data underestimate the available en-
vironmental heterogeneity because these
air temperatures are not independent of
the animals under study (i.e., I did not
measure air temperature in microen-

THERMAL NICHE BREADTH IN ANOLIS - Hertz

vironments that the lizards did not use).
Nevertheless, these statistics have the
advantage of minimizing the importance
of extreme temperatures in any data set.
They also provide convenient summaries
which are directly comparable among
studies in which lizards were sampled
over a range of time periods.

RESULTS

POPULATION NICHE BREADTHS: TESTS
OF PREDICTION 1

Perch height diversities for popula-
tions of roquet (maximum = 6.62) on
Martinique (Fig. 1) were similar to those
for each of the three species on Puerto
Rico (Fig. 2). These data indicate that
roquet, in the absence of congeneric
competitors, does not use a wider variety
of perch heights than does any of the
common species in the complex fauna.
Indeed, perch height diversities for two
combined samples of sympatric gund-
lachi and evermanni were higher (8.50 at
480 m and 7.12 at 675 m) than those for

20m 350m 650m
3:] x= 89-6+4-3 x=89:04+4-3
=6- B=6-41
a B=6-62
E24
2:1
r
5 1-8
wy.
as
pe 2
oO
OS)
ao
0-6
0-3
FREQUENCY
Figure 1. Sex-weighted perch height frequency dis-

tributions for Anolis roquet on Martinique. Mean perch
heights (cm, X + s.e.) and perch height diversity (B =
1/p?) indicated for each population.

A475

roquet. These data fail to confirm the first
prediction. (The data on perch heights
may be biased insofar as evermanni
perched very high in a tree are less likely
to be seen than are those perched lower
on the trunk (Rand, 1964). However, such
a bias is conservative because it would
have fostered an underestimate of ever-
manni's niche breadth.) Schoener (1975)
has shown that the structural habitat
breadth of some Anolis populations is
dependent upon the availability of perch
sites in various habitats. Although I did
not quantitatively measure perch height
availabilities, I do not believe that inter-
locality differences in the height of the
vegetation influenced these data. Lizards
in all populations failed to use the full
range of perch heights that were avail-
able.

Estimates of the heterogeneity in the
thermal environment (thermal habitat
breadths, Table 1) show that roquet
populations experienced more variable
environments than did most populations
of gundlachi and evermanni, but the dif-
ference is neither consistent nor of great
magnitude. The lowland sample of crist-
atellus inhabited the most heterogeneous
thermal environment of all the popula-
tions surveyed. Moreover, the combined
thermal habitat breadths of sympatric
gundlachi and evermanni (3.4 C at 480 m
and 2.6 C at 675 m) were indistinguish-
able from those of the roquet popula-
tions. Some of these data confirm the first
prediction.

Thermal niche breadths (Table 1) of
roquet populations were consistently
greater than those of gundlachi and
evermanni populations, but the niche
breadth of the cristatellus population
was nearly as large as that of the most
eurythermal roquet population. Thermal
niche breadths of each of the roquet
populations was also greater than the
combined niche breadth of sympatric
gundlachi and evermanni populations
(3.4° C at each site). These data also pro-
vide a partial confirmation of the first
prediction.

476 Advances in Herpetology and Evolutionary Biology
evermanni evermanni
x=184-44+5-6 x=167-444-8
B= 7:90 B= 7:26
cristatellus gundlachi gundlachi gundlachi gundlachi
X=1H2-44+45 %K=17-9440 xK=113-343-:9 xK=118-7436 , x=121-5+4-2
37. B=7:24 B=6:75 B=6-69 B=5-95 B=6-52
/
oo 5m 270m / 480 m 675m 850m
3+ Y
a Y),
= ee : Yj
= Na j Y,
ae
Oe \I Y Y
a 2
ane ye
5 15 S UY,
edhe SS Yj
0-9 a Y)
5 NI
SS h,
03 SS i
=| eA “| 2.23 AT A Gyn eek De ee eS Do ee eS

FREQUENCY

cristatellus [4 gundlachi

V//\_ evermanni

Figure 2. Sex-weighted perch height frequency distributions for three Anolis species on Puerto Rico. Mean
perch heights and perch height diversities as in Figure 1.

SPECIES NICHE BREADTHS: TESTS OF
PREDICTION 2

The environmental heterogeneity ex-
perienced by all roquet populations
combined (6.0° C) was not notably larger
than that experienced by the four gund-
lachi populations (5.9° C), but the former
was larger than that evident among the
two evermanni populations (2.6° C).
(These data may, however, underesti-
mate species breadths for evermanni
because this species occupies a larger
elevational range than is evident in my
samples [Rand, 1964; Schoener and
Schoener, 1971b].) My data on one crist-

atellus population are insufficient as an
estimator for the entire species, but Huey
and Webster (1976), using identical
techniques in summer, found a species
thermal habitat breadth of 8.8° C for 27
hourly samples of cristatellus from di-
verse localities. The species thermal
habitat breadth for roquet (above) was
slightly less than that of the combined
thermal habitat breadth of cristatellus
and gundlachi sampled over an equiva-
lent range of elevations (6.7° C, when the
850 m gundlachi sample is excluded).
The above data, in combination with
Huey and Webster’s (1976) observation
of a species thermal habitat breadth of

THERMAL NICHE BREADTH IN ANOLIS : Hertz

A477

TABLE 1. MINIMUM AND MAXIMUM HOURLY MEAN TEMPERATURE STATISTICS, THERMAL HABITAT BREADTHS (THB),
AND THERMAL NICHE BREADTHS (TNB) (ALL °C) IN 10 SAMPLES OF WEST INDIAN ANOLIS LIZARDS.

Air Temperatures

Body Temperatures

in minimum maximum minimum maximum
Population N X + s.e. X + s.e. THB XGeEuste: EGE, TNB
roquet: 20 m 12 Mss 08 7/7 a= OZ! 2.6 26.4 + 0.2 30.0 + 0.2 3.6
350 m 12 23.7 + 0.1 27.0 + 0.3 3.3 WA EOD Bil jl Of 6.4
650 m 10 *91.7+0.3 24.8 + 0.4 3.1 0) a= (0), 11 26.7 + 0.2 4.5
cristatellus: 5m 12 24.2 + 0.1 28.3 + 0.2 4.1 24.3 + 0.1 30.4 + 0.2 6.1
gundlachi: 270 m 12 23.70.11 *26.5+0.1 2.8 24.0+0.1 *27.1+0.1 3.1
480 m 12 22.1+0.1 24.3 +0.1 2.2 22.3 + 0.1 25.1+0.1 2.8
675 m 12 21.6+0.1 93.9 + 0.2 2.3 21.7+ 0.1 94.5 + 0.2 2.8
850 m 11 *90.6 + 0.1 22.2+0.1 1.6 *90.8 + 0.1 22.6 + 0.2 1.8
evermanni: 480m ll 22.7+0.1 *24.5+ 0.2 1.8 23.2+0.2 *25.7+ 0.2 2.5
675 m 10 *9219+0.1 94.2 + 0.2 2.3 *99.2+0.1 95.1+ 0.2 2.9

+N = number of hourly samples collected at each locality.

*These values used for calculations of species breadths; see text for explanation.

9.5° C for 19 hourly samples of gundlachi,
collectively refute the second prediction.

The species thermal niche breadth
among roquet populations (8.9° C) was
substantially greater than that among
gundlachi and evermanni populations
(6.3° C and 3.5° C, respectively). In addi-
tion, the species thermal niche breadth
for roquet was remarkably similar to that
of the combined species thermal niche
breadths of cristatellus and gundlachi
sampled over an equivalent altitudinal
range (8.7° C, when the 850 m gundlachi
sample is excluded). These data clearly
support the second prediction. However,
Huey and Webster (1976) provide con-
tradictory evidence. They found species
thermal niche breadths of 9.6° C and 8.7°
C for cristatellus and gundlachi, respec-
tively; these values more nearly ap-
proximate those for roquet and therefore
refute the second prediction.

DISCUSSION
EURYTHERMY AND SPECIES NICHE
BREADTHS

These new data and those of Huey and
Webster (1976) only partially confirm the

two predictions made in the Introduc-
tion. Populations of roquet do not exhibit
broader structural habitat or thermal
habitat niches than do species in the
complex Puerto Rican fauna. However,
populations of the former species do
exhibit broader thermal niches than do
some populations in the complex fauna.
The comparisons of species breadths
indicate no consistent differences be-
tween this solitary species and those
from the complex fauna of Puerto Rico.

How general are these results? In the
two decades since Ruibal’s (1961) pio-
neering study of the Cuban anoles, re-
searchers have studied the thermal bi-
ology of numerous West Indian Anolis
species. Unfortunately, many of the data
are not useful for comparison because
they were collected over unspecified
time periods and are not divisible into
temporally restricted samples (e.g.,
Brooks, 1968, Heatwole et al., 1969;
Rand, 1967). However, I have been able
to compile data on the thermal habitat
and thermal niche breadths for 56 popu-
lations of 15 species (Table 2). These data
may have limited generality because no
researchers have examined seasonal
variation in the thermal biology of West
Indian anoles; however, the data set is

Advances in Herpetology and Evolutionary Biology

TABLE 2. SPECIES THERMAL HABITAT AND THERMAL NICHE BREADTHS (THB AND TNB, °C) FOR 15 SPECIES OF
WEST INDIAN ANOLIS LIZARDS.

Species Island pS st THB TNB Reference
Solitary Anoles:
acutus* St. Croix 1 14 6.9 7.0 McManus and Nellis, 1973
marmoratus Guadeloupe 9 29 9.0 9.9 Huey and Webster, 1975
oculatus* Dominica 4 ri od 7.3 Ruibal and Philibosian, 1970
roquet Martinique 3 oo 6.0 8.9 Hertz, 1977, 1981, this paper
sagrei* Abaco 2 al ? 10.6 Lister, 1976
Anoles from Complex Faunas:
allisoni* Cuba 2 4 4.7 2.3 Ruibal, 1961; Ruibal and Philibosian, 1970
allogus* Cuba 2 3 RD, 1.7 Ruibal, 1961; Ruibal and Philibosian, 1970
homolechis* Cuba 3 6 4.1 2.6 Ruibal, 1961; Ruibal and Philibosian, 1970
lucius* Cuba 1 3 1.9 2.1 Ruibal, 1961; Ruibal and Philibosian, 1970
sagrei* Cuba 1 4 3.7 3.0 Ruibal, 1961; Ruibal and Philibosian, 1970
sagrei* Exuma 3. 19 ? 6.8 Lister, 1976
cybotes Hispaniola 3) 29 9.6 8.8 Hertz, 1977; Hertz and Huey, 1981
shrevei Hispaniola 1 5 1.7 3.1 Hertz, 1977; Hertz and Huey, 1981
cooki Puerto Rico 3 £19 BLY 4.8 Huey and Webster, 1976
cristatellus Puerto Rico 8 29 8.8 10.1 Hertz, 1977; Huey and Webster, 1976
evermanni Puerto Rico Bi All 2.6 3.5 Hertz, 1977; this paper
gundlachi Puerto Rico 9 65 9.5 9.6 Hertz, 1977, 1981; Huey and Webster, 1976

Sp = number of populations sampled.

S = number of temporally restricted samples represented. See text for explanation.
*THB calculated from midpoints of air temperatures ranges; mean air temperatures not available.

*Data read from figure.

internally consistent because nearly all
studies were conducted during summer
months.

Three general trends are apparent in
Table 2. 1) All of the solitary species ex-
hibit broad thermal habitat and thermal
niches. 2) Some species from the com-
plex faunas of Hispaniola and Puerto
Rico are as broad-niched as are the soli-
tary species. 3) There is a strong associa-
tion between species thermal habitat
breadth and species thermal niche
breadth (r, = 0.829, P < 0.01). These
conclusions contradict Ruibal and Phili-
bosian’s (1970) hypothesis that solitary
anoles may generally have broader
niches than do species from the complex
faunas (see also Huey and Webster,
1976).

Despite the clarity of the above con-
clusions, the data in Table 2 largely re-
flect differences in the sampling effort

devoted to different species. Among the
15 species, thermal habitat breadth is
significantly rank correlated with num-
ber of populations sampled and with the
number of hourly samples collected (1, =
0.765 and 0.775, respectively, P’s < 0.01).
Similarly, species thermal niche breadth
is also rank correlated with these vari-
ables (r, = 0.606 with number of popula-
tions and r, = 0.835 with number of
hourly samples, P’s < 0.01).

We can limit the effect of this sampling
bias by examining only those species in
which three or more populations were
sampled. (This restriction unfortunately
eliminates all but one Cuban species
from consideration.) Four species in the
complex faunas (cristatellus, cybotes,
gundlachi, and sagrei) exhibit niche
breadths comparable in magnitude to
those of three solitary anoles (marm-
oratus, oculatus, and roquet). Two other

THERMAL NICHE BREADTH IN ANOLIS - Hertz

well-sampled species from complex
faunas (cooki and homolechis) exhibit
narrow thermal niches, but the small
breadth of the latter may be attributable
to the limited number of temporally re-
stricted samples available.

EURYTOPY AND THERMOREGULATORY
VERSATILITY

The interspecific differences in ther-
mal niche breadths listed in Table 2 re-
flect interactions between the diversity
of ambient thermal regimes which popu-
lations occupy and the degree to which
populations can alter the precision with
which they thermoregulate in response
to varying environmental conditions
(Hertz, 1977, 1981; Hertz and Huey,
1981). The precision of thermoregulation
in a population of small arboreal ecto-
therms such as Anolis can be estimated
from the slope of the regression of body
temperatures on air temperatures (Huey,
1981; Huey and Slatkin, 1976). A slope
near zero indicates the independence
of body and air temperatures (perfect
thermoregulation), whereas a slope near
one indicates a strong dependence of
body temperature on air temperature
(thermal passivity, Hertz, 1974). The
thermoregulatory versatility of a species
can therefore be estimated by the range
of regression slopes among populations
living in diverse thermal habitats (Hertz,
1977). Theoretically, this index of ther-
moregulatory versatility can vary be-
tween 0.0 and 1.0. Although this statistic
can describe the range of effects of
thermoregulatory behaviors, behavioral
observations are still necessary to de-
scribe the nature and extent of thermo-
regulation (Hertz and Huey, 1981; Huey,
1982).

The niche breadths of forest-dwelling
Anolis species in complex faunas may be
determined by their stenotopy (i.e., oc-
cupation of a narrow range of habitats)
and lack of thermoregulatory versatility.
The best studied example is gundlachi,
the species that is restricted to low

479

perches in the relatively dark montane
rainforests of Puerto Rico. Populations of
gundlachi are stenothermal, and, be-
cause of their thermal passivity at all sites
(range of regression slopes = 0.05), body
temperatures track air temperatures
throughout its broad altitudinal range
(Hertz, 1981). However, because popula-
tions of gundlachi experience narrow and
distinct ranges of air temperatures at dif-
ferent altitudes, the species thermal habi-
tat and thermal niche breadths are large.
Ruibal (1961) has shown that populations
of allogus and lucius are also steno-
thermal and thermally passive in the
shaded forests of Cuba. However, be-
cause these species were not sampled in
geographically and thermally diverse
localities, one cannot estimate the magni-
tude of either their species breadths or
their thermoregulatory versatility.
Several other stenotopic anoles in
complex faunas also show low thermo-
regulatory versatility. However, because
they are restricted to open habitats where
the cost of thermoregulation is low
(Huey, 1974, Huey and Slatkin, 1976),
they thermoregulate precisely, and, as a
result, their thermal niche breadths are
apparently narrow. For example, cooki
always regulates body temperature with-
in narrow limits (low regression slopes in
all populations, R. B. Huey, personal
communication), probably because the
potential for overheating is high in the
deserts of Puerto Rico (Huey and Web-
ster, 1976). In the montane savanna of
Hispaniola, shrevei always thermo-
regulates carefully under sunny skies,
presumably because of the benefit ac-
crued from having a high body tempera-
ture in a habitat where air temperatures
are usually low (Hertz and Huey, 1981).
In contrast to the low versatility shown
by the thermoconformers and obligate
thermoregulators described above,
eurytopic species in complex faunas (e.g.,
cristatellus, cybotes, and sagrei) exhibit
high thermoregulatory versatility. Popu-
lations of cristatellus living in shaded
versus open habitats exhibit a range of

480

regression slopes (0.80) that virtually
span the entire theoretical range from
thermal passivity to perfect thermo-
regulation (Huey, 1974; Huey and
Slatkin, 1976). Lister (1976) described an
analogous situation for sagrei on Exuma
in which the precision of thermoregula-
tion varied with the amount of direct
solar radiation that penetrated the vege-
tation in different habitats. Similarly, in
cybotes sampled along an altitudinal
gradient of 1,150 m, the precision of
thermoregulation varied (range of slopes
= (0.60) among populations. In each of
these versatile species, animals predict-
ably shift habitat and/or microhabitat,
alter basking frequency, and vary times
of activity with geographical changes in
air temperature and/or the energetic
costs of basking in different habitats.
Thermoregulatory versatility in these
species therefore arises from their ability
to shift regulatory behavior with the oc-
cupation of different habitats.

It is difficult to assess the thermo-
regulatory versatility of the solitary spe-
cies with great accuracy. Several re-
searchers have noted that they show
thermal passivity in many habitats
(Brooks 1968 on oculatus; McMannus
and Nellis 1973 on acutus) but exhibit
opportunistic thermoregulatory behavior
under some circumstances (Lazell, 1972;
Ruibal and Philibosian, 1970 on ocu-
latus; Huey and Webster 1975 on marm-
oratus). However, my data on three pop-
ulations of roquet suggest at least a
moderate degree of versatility (range of
regression slopes = 0.59); precision of
thermoregulation varied in response to
altitudinal changes in air temperatures
and basking opportunities (Hertz, 1981)
in a fashion analogous to that seen in
cybotes on Hispaniola (Hertz and Huey,
1981). Presumably, my estimate of versa-
tility in roquet would be enlarged by the
inclusion of a sample of A. r. zebrilus or
A. r. salinei, the subspecies that occupy
the most arid and xeric habitats on
Martinique (Lazell, 1972).

The above analysis suggests that West
Indian Anolis species have adopted one

Advances in Herpetology and Evolutionary Biology

of three behavioral-ecological “pheno-
types’ with respect to their thermal
biology: 1) thermal passivity under a
broad range of ambient conditions; 2)
careful thermoregulation whenever pos-
sible; and 3) thermoregulatory versatility
as a response to varying conditions. (We
must, of course, use caution in specifying
the phenotype of any species because
estimates of thermoregulatory versatility
will, like estimates of niche breadth, be
sensitive to the number of populations
sampled as well as the seasonal schedule
of sampling.) Anolis species from the
complex faunas exhibit any one of the
three phenotypes; examples of inter-
mediate phenotypes undoubtedly will be
found with additional research. The

solitary anoles, however, all exhibit the —

versatile phenotype.

THERMOREGULATORY VERSATILITY AND
COLONIZING POTENTIAL

Anoles are thought to have evolved in
the moist tropical forest of the New
World as eurythermal and passive spe-
cies with relatively low activity tempera-
tures (Huey and Webster, 1975). How-
ever, because these characteristics might
be grossly maladaptive for the trans-
oceanic dispersal of a small animal,
thermal passivity and low activity
temperatures were probably not ances-
tral within the West Indian members of
this genus. Rather, the colonizing species
from which the West Indian Anolis radi-
ation is derived almost certainly ex-
hibited thermoregulatory versatility as
well as a tolerance of high temperatures
(Heatwole et al., 1969; Ruibal, 1967) and
desiccation stress (Williams, 1969).
These characteristics would enable a
small ectotherm not only to survive trans-
oceanic rafting but also to exploit the
wide array of habitats available on the
newly inhabited island. Thermoregula-
tory versatility and eurythermy can there-
fore be considered an ancestral “coloniz-
ing phenotype” in most, if not all, island
colonizations by members of this genus.
Williams (1969: 374-375) has constructed

THERMAL NICHE BREADTH IN ANOLIS - Hertz

a similar portrait of colonizing anoles in a
more general context.

Thermoregulatory versatility and
eurythermy in solitary anoles may not
represent “ecological release” or “niche
expansion’ at all. Rather, these species
may have simply retained the widely
adaptive colonizing phenotype while
exploiting diverse habitats in the absence
of congeneric competitors. Morphologi-
cal differentiation within the solitary
anoles of the Lesser Antilles appears not
to be related to their thermal ecology
(Hertz, 1977; Ruibal and Philibosian,
1970), and these species exhibit rela-
tively little physiological differentiation
among populations (Hertz, 1977, 1980,
1981; Hillman et al., 1979). Indeed,
Listers (1976) observations on sagrei,
and, to some extent, those on monensis
(but cf. Gorman and Stamm, 1975: 201)
are the only reliable demonstration of
ecological release of thermal biology
among West Indian anoles; the physio-
logical and genetic bases of this phe-
nomenon have not been explored in
these species.

Some species in the complex faunas of
the Greater Antilles (e.g., cristatellus,
cybotes, and sagrei) exhibit phenotypes
that are similar to the hypothetical colo-
nizing phenotype, and their occupation
of disturbed habitats in the Greater
Antilles provides further evidence of
their colonizing abilities (e.g., Turner
and Gist, 1970). Williams (1969) has, in
fact, listed two of these species (crist-
atellus and sagrei) among the six best
colonizers within West Indian anoles. If
Williams’s (1972) phylogeny for the
Puerto Rican radiation is correct, how-
ever, we must assume that, for crist-
atellus at least, thermoregulatory versa-
tility represents convergence to the
colonizing phenotype, rather than reten-
tion of it.

CONCLUSIONS AND SUGGESTIONS FOR
FUTURE RESEARCH

The above analysis does not in any way
support the thesis that solitary anoles ex-

481

perience “ecological release” of their
thermal biology. Nonetheless, most
Anolis ecologists would probably agree
that the field biology of solitary anoles is
markedly different from that of species in
the more complex Greater Antillean
faunas. This statement is supported by
the intuition of numerous field workers
and stems from the belief that competi-
tion will almost inevitably constrain a
species ecological flexibility (see, e.g.,
MacArthur, 1972).

The failure of Anolis biologists to
demonstrate the “ecological release” of
thermal biology is perhaps a methodo-
logical, rather than a conceptual, failure.
The data gathered during the last two
decades apparently provide poor resolu-
tion of the biological realities that Rand
(1964) and Williams (1972) have de-
scribed as the “climatic habitat” of each
species. Researchers should now adopt
more sophisticated methodologies to test
the hypotheses generated by theoretical
ecologists. The simple measurement of
lizard body temperatures and air tem-
peratures, though useful for acquiring
baseline information about the field bi-
ology of many species, is insufficient for
testing general theory about the effects of
competition.

The demonstration of ecological re-
lease in the thermal biology of these
animals will require simultaneous analy-
sis of three factors that interact to pro-
duce a species’ “thermal biology”: 1) the
range of habitats and microhabitats avail-
able to a population; 2) the populations’
use of the available environmental heter-
ogeneity; and 3) the breadth of physio-
logical capacity that has evolved in dif-
ferent populations. The use of biophysi-
cal models and _ micrometeorological
measurements (Porter and Gates, 1969),
in combination with observations of
animals in a diversity of habitats, might
adequately address the first two of these
factors. Indeed, Roughgarden, et al.
(1981) have successfully applied these
techniques in a study of resource parti-
tioning by anoles on Grenada and St.
Kitts. An adequate analysis of the physi-

482

ological breadth of a population will re-
quire detailed studies of the thermal
dependence of whole-animal physiologi-
cal functions (e.g., locomotion, Huey and
Stevenson, 1979; paper by Huey in this
volume) in addition to the now typical
measurements of rates of water loss and
critical and preferred temperatures.

Finally, future research on this topic
should be directed towards those species
that Williams (1969) has identified as the
most successful invaders of isolated is-
lands in the Greater Antillean subregion
of the Caribbean: angusticeps, caro-
linensis, cristatellus, distichus, grahami,
and sagrei. These species, and their close
relatives, will provide the best oppor-
tunities for the demonstration of ecologi-
cal release. Lister’s (1976) study of sagrei
provides a good starting point for addi-
tional work: in the absence of detailed
information on source island (Cuba)
populations, our understanding is far
from complete.

ACKNOWLEDGMENTS

I thank A. Garcia, M. Perez, and D.
Viera for field assistance in Puerto Rico,
and W. Hall and F. Wadsworth for the
use of facilities at their disposal. R. B.
Huey offered valuable criticism and dis-
cussion. W. H. Bossert, K. I. Miyata, A. S.
Rand, R. Ruibal, T. W. Schoener, E. E.
Williams, B. Wu, and E. Zouros all had
something to say about previous drafts.

Financial support of field work was
generously provided by grants from the
Richmond and Anderson Funds of
Harvard University, Sigma Xi, the Ex-
plorers Club, and the National Science
Foundation Grants DEB 75-16334 and
GB 37731X. The manuscript was pre-
pared with the assistance of the Research
and Travel Fund of Barmard College.

This paper is dedicated to the Prin-
ciple of Unsympathetic Magic.

Advances in Herpetology and Evolutionary Biology

LITERATURE CITED

BAKKEN, G. S. 1976. A heat transfer analysis of
animals: unifying concepts and the application
of metabolism chamber data to field ecology. J.
Thermal Biology, 6: 337-384.

Brooks, G. R. 1968. Body temperatures of three
lizards from Dominica, West Indies. Herpetol.,
24: 209-214. ’

Cow .es, R. B., AND C. M. BoOGERT. 1944. A prelimi-
nary study of the thermal requirements of
desert reptiles. Bull. Amer. Mus. Nat. Hist., 83:
263-296.

GorMAN, G. C., AND B. STAMM. 1975. The Anolis
lizards of Mona, Redondo, and La Blanquilla:
Chromosomes, relationships, and natural his-
tory notes. J. Herpetol., 9: 197-205.

GRANT, P. R. 1972. Convergent and divergent
character displacement. Biol. J. Linn. Soc., 4:
39-68.

HEATWOLE, H., T. H. Lin, E. VILLALON, A. MUNIZ,
AND A. MATTA. 1969. Some aspects of the
thermal ecology of Puerto Rican anoline liz-
ards. J. Herpetol., 3: 65-77.

HERTZ, P. E. 1974. Thermal passivity of a tropical
forest lizard, Anolis polylepis. J. Herpetol., 8:
323-327.

____. 1977. Altitudinal variation in thermoregula-
tory strategies, physiological ecology, and
morphology of some West Indian anoles.
Doctoral Dissertation, Harvard University.

___. 1980. Responses to dehydration in Anolis liz-
ards sampled along altitudinal transects.
Copeia, 1980: 440-446.

____. 1981. Adaptation to altitude in two West In-
dian anoles. I. Field thermal biology and
physiological ecology. J. Zool., London, 195:
25-37.

HERTZ, P. E., AND R. B. HurEy. 1981. Compensation
for altitudinal changes in the thermal environ-
ment by some Anolis lizards on Hispaniola.
Ecology, 62: 515-521.

HILLMAN, S., G. C. GORMAN, AND R. THOMAS. 1979.
Water loss in Anolis lizards: evidence for ac-
climation and intraspecific differences along a
habitat gradient. Comp. Biochem. Physiol.,
62A: 491-494.

Huey, R. B. 1974. Behavioral thermoregulation in

lizards: importance of associated costs.

Science, 184: 1001-1003.

1982. Temperature, physiology, and the
ecology of reptiles, pp. 25-91. In C. Gans and F.
H. Pough (eds.), Biology of the Reptilia, Vol. 12,
in press. New York, Academic Press.

Huey, R. B., AND M. S.aTKIN. 1976. Cost and
benefits of lizard thermoregulation. Quart.
Rev. Biol., 51: 363-384.

Huey, R. B., AND R. D. STEVENSON. 1979. Integrat-
ing thermal physiology and ecology of ecto-

——— .

THERMAL NICHE BREADTH IN ANOLIS : Hertz

therms: A discussion of approaches. Amer.
Zool., 19: 357-366.

Huey, R. B., AND T. P. WeBsTER. 1975. Thermal
biology of a solitary lizard: Anolis marmoratus
of Guadeloupe, Lesser Antilles. Ecology, 56:
445-452.

___ AND . 1976. Thermal biology of Anolis liz-
ards in a complex fauna: the cristatellus group
on Puerto Rico. Ecology, 57: 985-994.

LAZELL, J. D., JR. 1972. The anoles (Sauria,
Iguanidae) of the Lesser Antilles. Bull. Mus.
Comp. Zool., 143: 1-115.

LEvINS, R. 1968. Evolution in Changing Environ-
ments. Princeton, N.J., Princeton University
Press.

LISTER, B. C. 1976. The nature of niche expansion
in West Indian Anolis lizards. I. Ecological
consequences of reduced competiton. Evolu-
tion, 30: 659-676.

MacArtTuHur, R. H. 1972. Geographical Ecology.
New York, Harper and Row.

McManus, J. J., AND D. W. NELLIS. 1973. Tempera-
ture and metabolism of a tropical lizard, Anolis
acutus. Comp. Biochem. Physiol., 45A:
403-410.

Porter, W. P., AND D. M. GaTEs. 1969. Thermo-
dynamic equilibria of animals with environ-
ment. Ecol. Monogr., 39: 227-244.

RAND, A. S. 1964. Ecological disribution in anoline
lizards of Puerto Rico. Ecology, 45: 745-752.

—__. 1967. The ecological distribution of the ano-
line lizards around Kingston, Jamaica. Breviora
Mus. Comp. Zool. No. 272, pp. 1-18.

RAND, A. S., AND P. J. RAND. 1967. Field notes on
Anolis lineatus in Curacao. Stud. Fauna
Curacao Other Carib. Islds., 93: 112-117.

RAND, A. S., AND E. E. WILLIAMS. 1969. The anoles
of La Palma: aspects of their ecological rela-
tionships. Breviora Mus. Comp. Zool. No. 327,
pp. 1-18.

ROUGHGARDEN, J. 1972. Evolution of niche width.
Amer. Nat., 106: 683-718.

ROUGHGARDEN, J., W. PORTER, AND D. HECKEL.
1981. Resource partitioning of space and its
relationship to body temperature in Anolis
lizard populations. Oecologia. 50: 256-264.

483

RUIBAL, R. 1961. Thermal relations of five species of
tropical lizards. Evolution, 15: 98-111.

___. 1967. Evolution and behavior in West Indian
anoles, pp. 116-156. In W. Milstead (ed.), Liz-
ard ecology: a symposium. Kansas City, Mis-
souri, University of Missouri Press.

RUIBAL, R., AND R. PHILIBOSIAN. 1970. Eurythermy
and niche expansion in lizards. Copeia, 1970:
645-653.

SCHOENER, T. W. 1975. Presence and absence of
habitat shift in some widespread lizard spe-
cies. Ecol. Monogr., 45: 233-258.

SCHOENER, T. W., AND A. SCHOENER. 1971a. Struc-
tural habitats of West Indian Anolis lizards. I.
Lowland Jamaica. Breviora Mus. Comp. Zool.
No. 368, pp. 1-53.

AND . 1971b. Structural habitats of West
Indian Anolis lizards. II. Puerto Rican up-
lands. Breviora Mus. Comp. Zool. No. 375, pp.
1-39.

SOULE, M. 1963. Aspects of thermoregulation in
nine species of lizards from Baja California.
Copeia, 1963: 107-115.

TURNER, F. B., AND C. S. Gist. 1970. Observations of
lizards and frogs in an irradiated Puerto Rican
forest, Chapter E-2, pp. E25-E49. In H. T.
Odum and R. F. Pigeon (eds.), A tropical rain
forest. A study of irradiation and ecology. Div.
Tech. Inf., U.S. A.E.C. Publ. TID-24270
(PRNC-138).

WILLIAMS, E. E. 1969. The ecology of colonization
as seen in the zoogeography of anoline lizards
on small islands. Quart. Rev. Biol., 44:
345-389.

____. 1972. The origin of faunas. Evolution of lizard
congeners in a complex island fauna: a trial
analysis, Vol. 16, pp. 47-88. In Th.
Dobzhansky, M. K. Hecht, and W. C. Steere
(eds.), Evolutionary Biology. New York,
Appleton-Century-Crofts.

WILSON, E. O. 1961. The nature of the taxon cycle in
the Melanesian ant fauna. Amer. Nat., 95: 169-
193.

Natural Variation in Body Temperature and
Physiological Performance in a
Lizard (Anolis cristatellus)

RAYMOND B. HUEY'

ABSTRACT. Field data on changes in body tempera-
ture (with habitat, time of day, and thermoregula-
tory behavior) of the Puerto Rican lizard Anolis
cristatellus are integrated with laboratory data on
the effect of body temperature on relative sprint
speed. Lizards that thermoregulate in an open
park in the early morming should be able to run
slightly faster than thermoconforming lizards in a
nearby forest. At midday, however, ambient tem-
peratures in the open are high, and the T, of lizards
in the open is elevated above thermal preferences;
lizards in the forest may then have a performance
advantage. The results suggest that the temporal
and spatial variation in physiological performance
of lizards in nature can be dynamic.

INTRODUCTION

Body temperatures of some active liz-
ards vary in time and in space (Avery,
1982). The extent to which such variation
influences physiological and thus eco-
logical performance is a fundamental, but
understudied problem in physiological
ecology (Crowder and Magnuson, 1982;
Huey, 1982). This problem can be ap-
proached by determining how the
physical environment influences body
temperature (Porter et al., 1973), by in-
vestigating the thermal sensitivities of
physiological systems that have direct
ecological relevance (Bartholomew,
1958; Brett, 1971; Huey and Stevenson,
1979; Bennett, 1980), and then synthe-
sizing ecological and physiological data

1Department of Zoology NJ-15, University of
Washington, Seattle, WA 98195, U.S.A.

(Huey and Slatkin, 1976; Waldschmidt,
1978; Christian and Tracy, 1981; Muth,
1980; Huey, 1982).

I first became interested in integrating

physiology and ecology while on a field
trip with Emest E. Williams in the
Caribbean and soon decided to pursue
this topic for my dissertation under his
direction. During this period, I grew to
appreciate the relevance of whole-animal
physiological performance such as
locomotion to ecological analyses, an
approach that Ernest supported. Con-
sequently, it is my pleasure to dedicate
this paper, which is a direct extension of
my thesis research, to Professor Emest E.
Williams for this Festschrift in his honor.

This paper reports a preliminary in-
vestigation of the effects of temporal and
spatial variation in body temperature on
physiological performance in the Puerto
Rican lizard Anolis cristatellus. Labora-
tory data on the thermal sensitivity of
sprint locomotion are integrated with a
previous analysis of body temperatures
in nature (Huey, 1974).

This integration of physiological per-
formance and ecology focuses on sprint
speed as a single, but ecologically impor-
tant, metric of physiological perform-
ance. Sprint speed can strongly affect the
efficiency of predation, escape from
predators, and social interactions (Rand,
1964; Webb, 1976; Elliott et al., 1977;
Bennett, 1980; Christian and Tracy,
1981; Huey and Hertz, 1982). Ultimately,
however, this approach must be gen-

PHYSIOLOGICAL ECOLOGY OF A LIZARD - Huey

eralized to incorporate additional physi-
ological (e.g., digestion, growth, repro-
duction) and ecological (e.g., environ-
mental productivity, risks of predation)
information (Huey and Slatkin, 1976).

BACKGROUND ECOLOGICAL DATA

The body temperature (T,) of the
trunk-ground lizard Anolis cristatellus
depends upon habitat, time of day, and
thermoregulatory behavior. In open habi-
tats where basking sites are frequently
available on lower tree trunks, these liz-
ards bask early and late in the day and
behaviorally control body temperatures
at relatively high and constant levels
(Fig. 1, from Huey, 1974). However, in a
nearby forest where basking sites are few
and distant, these lizards rarely bask, and
thus are thermoconformers. Consequent-
ly, average body temperatures in the
forest are relatively low and variable
(Fig. 1). (Males and females did not differ
significantly in T, in any census.) Com-
parable patterns hase been documented
for other Anolis (Lister, 1976; Lee, 1980;
Hertz and Huey, 1981).

In laboratory thermal gradients, lizards
from both forest and open habitats prefer

30 a mie? Aes oH

26 Vl forest

Anolis cristatellus

Body Temperature (°C)
i)
@

0800

1200
Time of Day

i600

Figure 1. Body temperatures of Anolis cristatellus in
an open and in a forest habitat in lowland Puerto Rico
(redrawn from Huey, 1974). Vertical lines represent
range, horizontal lines indicate means, and boxes en-
close 95% confidence limits of the means.

485

an average body temperature of 29.6°C
(Huey and Webster, 1976). In nature, liz-
ards in the open habitat reach these
temperatures very early in the day by
basking (Fig. 1). However, the high en-
vironmental temperatures in this habitat
at midday apparently force these lizards
to be active at temperatures a few de-
grees above preferred levels (Note: this
forcing should be verified with a bio-
physical analysis.) In the shaded forest,
however, lizards are active at T, below
preferred levels until late TOraRAIME.
Thereafter, “forest” lizards are active at
near_preferred T,, for the rest of the day.

MATERIALS AND METHODS

Six adult male Anolis cristatellus used
in these experiments were collected
along the northern coast of Puerto Rico
near the site of the field study (Pta.
Salinas) in the summer of 1979 (July), and
acclimated for 10 days to a 13:11 L:D
cycle and a temperature of 28°C. Crickets
(dusted with Vionate® vitamin-mineral
supplement) were provided every other
day.

Lizards were stimulated to sprint on a
horizontal 2.4 x 0.2 m racetrack (rubber-
ized substrate). The track contained 12
photocell stations at set distances (2 m
maximum separation). When a lizard
broke the first photocell beam, a multi-
channel timer was activated, and the time
when each subsequent beam was broken
was automatically measured and stored
in memory (Huey et al., 1981). The time
for each 0.25 m interval was determined,
and the fastest time converted to speed
(to the nearest 0.1 m s“').

Each lizard was raced daily (6 trials/
temperature, 1 temperature/day). The
sequence of temperatures (Fig. 2) was
random, except that the trials at 32°C
were run last to minimize the possibility
of injury resulting from exposure
to high T,.

The ds after the sprint trials were
completed, lizards were transferred to
15°C for 1 h and then cooled (~#2°C

486

a 100 Anolis cristatellus a as alin

: 7

Ee

x 80-4

‘ a

2 604

a

. +

= 40-

Q

op)

220m]

2 t ;

S Aa / § Vv
SSS SI ] ] ||

15 25 35

Body Temperature, °C

Figure 2. Normalized sprint speeds of Anolis crist-
atellus (mean + 95% confidence limits) as a function
of body temperature. The arrow indicates the average
preferred body temperature of lizards in a laboratory
thermal gradient (data from Huey and Webster, 1976).

min‘) until they lost the righting re-
sponse (Critical Thermal Minimum).
Two days later lizards were transferred to
30°C for 1 h and then heated until they
reached the Critical Thermal Maximum,
following procedures of Huey and Web-
ster (1976).

Sprint speeds and critical temperatures
for each lizard were fitted to a product-
exponential curve (Huey and Stevenson,
1979). These equations were then solved
to determine the “optimal” T, for sprint
performance (T, where speed maxi-
mized) and the “thermal performance
breadth” range of T, over which sprint
speed is greater than some arbitrary
level, see Huey and Stevenson (1979).
Tolerance ranges (the difference be-
tween the Critical Thermal limits) were
determined directly.

RESULTS

THERMAL SENSITIVITY OF SPRINT SPEED

Normalized sprint speeds of Anolis
cristatellus are diagrammed in Fig. 2,
and several statistical summaries are
presented in Table 1. The lower tem-

Advances in Herpetology and Evolutionary Biology

perature threshold for locomotion is
about 9°C, and the upper threshold is
about 37°C. Sprint speed increases from
the lower threshold T, until 30°C, which
approximates the average optimal tem-
perature for sprint speed. At higher
temperatures, sprint speeds drop precipi-
tously. Interestingly, the optimal T,, for
sprint speed is nearly identical to the
preferred T,, of these lizards in laboratory
thermal gradients (Table 1).

Thermal performance breadth mea-
sures the extent to which physiology is
independent of temperature at eco-
logically meaningful T, (Huey and
Stevenson, 1979). The range of T, over
which these lizards are able to run at 95%
and 80% of maximal speed is 6.5° and
12.7°C, respectively.

INTEGRATING SPRINT DATA WITH FIELD
TEMPERATURES

The average T, of lizards from each
habitat and time (Fig. 1) are used as an
index of temporal and spatial variation in
T, of Anolis cristatellus. Of course, this
measure ignores within-sample vari-
ability in T,, which is sometimes large in
the open habitat (Fig. 1). To summarize
the average effect of T, on sprint velocity,
I used the product-exponential equations
to calculate relative performance of each
individual at each of the T,,’s measured in
the field (Fig. 1) and then averaged
performance among individuals. This
measure ignores within and between-
individual variability (Table 1), but
nevertheless serves as a useful first-
approximation of how the average rela-
tive performance of cristatellus might
change as a function of habitat, time, and
behavior.

Activity begins near sunrise in natural
populations. Body temperatures at this
time are cool (~24.6°C), but lizards
should be able to sprint at about 92% of
maximum speed. In less than an hour,
lizards in the open elevate their tempera-
tures to about 30.6°C, a temperature at
which sprint speed is near maximal (Figs.

PHYSIOLOGICAL ECOLOGY OF A LIZARD - Huey

487

TABLE 1. STATISTICS DESCRIBING THERMAL SENSITIVITY OF SPRINT SPEED FOR ANOLIS CRISTATELLUS FROM LOW-

LAND PUERTO RICO (N = 6).

Character

Critical Thermal Minimum
“Optimal” T,,

Preferred T,,

Critical Thermal Maximum

Tolerance Range

Thermal Performance Breadth
95% of maximum
80% of maximum

Tolerance range = CTMax—CTMin.
Preferred Ty from Huey and Webster (1976).

2, 3). In contrast, body temperatures and
sprint speeds of the “‘forest’’ lizards,
which do not bask, remain at lower levels
until 1000 or 1100 h (Fig. 3). Thus, lizards
in the forest may have slightly submaxi-
mal locomotor abilities in the early morn-
ing.

By midday, ambient temperatures are
hot. Lizards in the forest are finally active
at T,, near preferred levels (Fig. 1), and
thus their sprint speeds are high and
remain high for the remainder of the day
(Fig. 3). In contrast, T,,’s of lizards in the
open are several degrees above preferred
levels (Fig. 1), and sprint speeds appear
slightly reduced (Fig. 3). The apparent,
slight hyperthermia of lizards in the open
should also increase metabolic rate and
evaporative water loss, thereby further
increasing the possible physiological
Beovantage of the open habitat at mid-

ay.

COMPARISONS WITH OTHER SPECIES

Data on thermal sensitivity of sprint
speeds are now available for several liz-
ards. The close correspondence between
the preferred T, and the optimal T,, for
sprint locomotion of A. cristatellus, a
species with a low thermal preferendum
(Table 1), is also found in several
thermophilic species (Cnemidophorus,
Dipsosaurus, Sceloporus; Tracy, 1979;

Temperature °C

X + s.e. range
8.92 + 0.50 7.0-10.2
29.53 + 0.85 27.0-32.1
29.6 + 0.59
37.00 + 0.20 36.6-38.0
28.12 + 0.48 26.8-29.8
6.48 + 0.39 5.4— 7.7
12.73 + 0.73 10.4-15.2

Bennett, 1980; Huey, unpublished data).
Curiously, however, two species with
low thermal preferenda do not show this
correspondence (Eumeces, Gerrhonotus;
Bennett, 1980). Further studies are
required to determine whether there is a
general pattern (or patterns) among liz-
ards.

Data on thermal niche breadth (Table
1) support the proposition that sprint
abilities of lizards are somewhat tem-
perature independent (Bennett, 1980): A.
cristatellus can sprint at 95% of maxi-
mum speed over a 6.5°C range of T,.

forest

1004 e—_ ®—_ ®—_e ®

©
open

te)

OX,
OF Se Oe

j

Speed

Relative
(Co)
ro)

=a |

Predicted

Ze
| Anolis cristate/lus

OF il
0800 1200 1600

Time of Day

Figure 3. Predicted relative sprint speeds of Anolis
cristatellus in open and forest habitats as a function of
time of day. Average values only are plotted.

488

Nevertheless, the actual diurnal variation
in J, in nature (Fig. 1) seems to be suffi-
ciently large to affect sprint abilities of A.
cristatellus. Whether this pattern is valid
for other species and for other physio-
logical activities of A. cristatellus re-
mains to be determined.

Data on thermal niche breadth are also
relevant to an hypothesis (Tracy, 1979)
concerning the relative degree of physio-
logical specialization for temperature in
frogs versus lizards. Because high rates of
evaporative water loss hinder the ability
of many frogs to control T, by basking
(Tracy, 1976), selection might have
favored a_ relatively temperature-
independent physiology in frogs. Indeed,
data on the thermal performance
breadths (80% level) for Rana clamitans
(Huey and Stevenson, 1979) and for
Anolis cristatellus (21.7° and 12.7°C,
respectively), are in accord with Tracy’s
(1979) prediction.

Interestingly, even though the thermal
performance breadths of Rana and Anolis
differ strikingly (1.7X), the tolerance
ranges for these two species are similar
(Rana = 30.0°C, Anolis = 28.8°C). This
observation supports the argument that
selection for tolerance range may be
independent of that for thermal perform-
ance breadth (Huey and Stevenson,
1979; Huey, 1982).

DISCUSSION

If the preferred temperatures of lizards
reflect underlying physiological per-
formance (Dawson, 1975), then A. crist-
atellus in open habitats might have a
physiological advantage over “forest”
lizards in the morning; but “forest” liz-
ards might have an advantage at midday.
On the other hand, if some aspects of
physiology are relatively independent of
T,, near the thermal preferendum (see
Bennett, 1980), then the observed varia-
tions in T, (Fig. 1) of Anolis cristatellus
might not be of sufficient magnitude to
impair physiological performance.

Advances in Herpetology and Evolutionary Biology

Data available to evaluate these alter-
natives are preliminary rather than defi-
nitive: 1) the sample sizes are small (only
6 individuals, only 6 trials/T,), 2) the
temperature intervals are too broad for
detailed analysis (see Fig. 2), 3) the ac-
climation was at a constant temperature
rather than at natural cycles of T,; and 4)

the performance of all “forest” and of —

“open” lizards at a given T, is assumed
identical. Keeping these limitations in
mind, the data suggest that the observed
natural variation in T, (Fig. 1) is of suffi-
cient magnitude to influence one aspect

of the physiological performance of —

Anolis cristatellus in lowland Puerto
Rico. In particular, habitat, time, and
behavior appear to affect sprint perform-
ance. For example, “open” lizards, which
bask frequently, may have a locomotor
advantage for a few hours in the early
morning, but perhaps not at midday.
Interestingly, because temperature
levels in the forest are warm at midday,
“forest” lizards are then active at T,, near
optimal for sprinting even without ex-

Pituophis melanoleucus

8 @ mean acceleration

a4 maximum speed

o

BSS
SS

Relative strike success

T Ar — Tu
4 6 8 ie}

Relative speed /acceleration

(0)

Figure 4. Body temperature affects maximum speed,
mean acceleration, and strike success of gopher
snakes (Pituophis melanoleucus) striking at mice (data
from Greenwald, 1974). However, neither relative
speed or acceleration has a 1:1 relationship with rela-
tive success (slope of each relationship is plotted).

PHYSIOLOGICAL ECOLOGY OF A LIZARD : Huey

hibiting thermoregulatory behaviors
(Huey, 1974).

The dynamics apparent in these pat-
terns encourage further investigation. In
particular, it would be interesting to
measure the effect of T, on digestion,
metabolism, and water balance and to
learn whether populations in the forest
and open differ in food availability or in
average growth rates. Moreover, by
repeating such studies during winter,
when the T, differential between “‘for-
est’ and “open” lizards should be ex-
aggerated (see Lister, 1976), the impor-
tance of season could be elucidated as
well.

Because of the importance of speed to
the ecology of lizards (Bennett, 1980;
Christian and Tracy, 1981; Huey and
Hertz, 1982), the described physiological
patterns should have _ considerable
ecological implications. However, pre-
dicting the magnitude of these ecological
effects is risky from sprint speed alone.
Acceleration and reaction times are
sometimes very important in predator-
prey encounters (Webb, 1976; Elliott et
al., 1977; Huey and Hertz, in prepara-
tion). Moreover, the actual decrement in
rate of food capture that results from a 5%
decrement in sprint performance de-
pends in part upon the biology of the
prey as well as that of the predator (Huey
and Stevenson, 1979). For example, body
temperature affects the relative accelera-
tion, velocity, and percent success of
gopher snakes (Pituophis melanoleucus)
striking at mice (Greenwald, 1974).
Nevertheless, the two laboratory mea-
sures of sprint performance (acceleration,
velocity) are only fair predictors of actual
strike success (Fig. 4). This problem of
scaling physiological to ecological per-
formance is fundamental but presently
elusive (Huey, 1982).

ACKNOWLEDGMENTS

This research was supported by the
National Science Foundation under

489

Grants DEB 78—-12024 and GB 37731X,
the Graduate School Research Fund of
the University of Washington, the Miller
Institute for Basic Research in Science,
and the Museum of Vertebrate Zoology
(University of California, Berkeley). I
thank C. R. Tracy and P. E. Hertz for
comments on the manuscript. Finally, I

thank the editors for the opportunity to
contribute to this Festschrift.

LITERATURE CITED

AVERY, R. A. 1982. Field studies of reptilian thermo-
regulation, pp. 93-166. In C. Gans and F. H.
Pough (eds.), Biology of the Reptilia Vol. 12,
Physiology (B). New York, Academic Press.

BARTHOLOMEW, G. A. 1958. The role of physiology
in the distribution of terrestrial vertebrates, pp.
81-95. C. L. Hubbs (ed.), Zoogeography.
AAAS, Washington, D.C.

BENNETT, A. F. 1980. The thermal dependence of
behavioural performance of small lizards.
Anim. Behav., 28: 752-762.

BRETT, J. R. 1971. Energetic responses of salmon to
temperature. A study of some thermal relations
in the physiology and freshwater ecology of
sockeye salmon (Onchorhynchus _ nerka).
Amer. Zool., 11: 99-113.

CHRISTIAN, K. R., AND C. R. Tracy. 1981. The effect
of the thermal environment on the ability of
hatchling Galapagos land iguanas to avoid
predation during dispersal. Oecologia, 49:
218-223.

CROWDER, L. B., AND J. J. MAGNUSON. 1982. Cost-
benefit analysis of temperature and food re-
source use: a synthesis with examples from the
fishes. In W. P. Aspey and S. I. Lustick (eds.),
Behavioral Energetics: Vertebrate Costs of
Survival. Columbus, Ohio State Univ. Press, in
press.

Dawson, W. R. 1975. On the physiological signifi-
cance of the preferred body temperatures of
reptiles, pp. 443-473. In D. M. Gates and R.
Schmerl (eds.), Perspectives of Biophysical
Ecology New York, Springer-Verlag.

ELLIOTT, J. P., I. MCT. Cowan, AND C. S. HOLLING.
1977. Prey capture by the African lion. Can. J.
Zool., 55: 1811-1828.

GREENWALD, O. E. 1974. Thermal dependence of
striking and prey capture by gopher snakes.
Copeia, 1971: 98-106.

HERTZ, P. E., AND R. B. Huey. 1981. Compensation
for altudinal changes in the thermal environ-
ment by some Anolis lizards on Hispaniola.
Ecology, 62: 515-521.

Huey, R. B. 1974. Behavioral thermoregulation in

490

lizards: importance of associated costs. Sci-
ence, 184: 1001-1003.

1982. Temperature, physiology, and the
ecology of reptiles, pp. 25-91. In C. Gans and
F. H. Pough (eds.), Biology of the Reptilia Vol.
12, Physiology (B). New York, Academic Press,
in press.

Huey, R. B., AND P. E. HERTZ. 1982. Effects of body
size and slope on sprint speed of a lizard
Stellio (Agama) stellio. J. exp. Biol., 97: 401-
409.

Huey, R. B., W. SCHNEIDER, G. L. ERIE, AND R. D.
STEVENSON. 1981. A field-portable racetrack for
measuring acceleration and velocity of small
cursorial animals. Experientia, 37: 1356-1357.

Huey, R. B., AND M. SLATKIN. 1976. Costs and bene-
fits of lizard thermoregulation. Quart. Rev.
Biol., 51: 363-384.

Huey, R. B., AND R. D. STEVENSON. 1979. Inte-
grating thermal physiology and ecology of
ectotherms: a discussion of approaches. Amer.
Zool., 19: 357-366.

Huey, R. B., AND T. P. WEBSTER. 1976. Thermal
biology of Anolis lizards in a complex fauna:
the cristatellus group on Puerto Rico. Ecology,
57: 985-994.

LEE, J. C. 1980. Comparative thermal ecology of
two lizards. Oecologia, 44: 171-176.

LISTER, B. C. 1976. The nature of niche expansion
in West Indian Anolis lizards I: ecological con-
sequences of reduced competition. Evolution,
30: 659-676.

Advances in Herpetology and Evolutionary Biology

MutTH, A. 1980. Physiological ecology of desert
iguana (Dipsosaurus dorsalis) eggs: tempera-
ture and water relations. Ecology, 61: 1335-
1343.

PORTER, W. P., J. W. MITCHELL, W. A. BECKMAN,
AND C. B. DEWITT. 1973. Behavioral implica-
tions of mechanistic ecology. Oecologia, 13: 1—
54. =

RAND, A. S. 1964. Inverse relationship between
temperature and shyness in the lizard Anolis
lineatopus. Ecology, 45: 863-864.

Tracy, C. R. 1976. A model of the dynamic ex-
changes of water and energy between a ter-
restrial amphibian and its environment. Ecol.
Monogr., 46: 293-326.

___. 1979. Further thoughts on anuran thermo-
regulation. In E. H. Burtt (ed.), The Behavioral
Significance of Color. New York, Garland
Publishers.

WALDSCHMIDT, S. 1978. Monthly variation in
thermoregulatory behaviors and space utiliza-
tion in the lizards
Sceloporus undulatus.
Thesis, Colorado State
Collins.

WEBB, P. W. 1976. The effect of size on the fast-start
performance of rainbow trout Salmo gairdneri,
and a consideration of piscivorous predator-
prey interactions. J. exp. Biol., 65: 157-177.

Unpublished M.S.
University, Fort

Uta stansburiana and:

On the Voluntary Departure of Lizards from

Very Small Islands

THOMAS W. SCHOENER!
AMY SCHOENER?

ABSTRACT. Individuals of the iguanid lizard Anolis
sagrei were placed on extremely small fragments of
natural islands and their subsequent behavior
recorded. Between a third and half of the 54 trials
resulted in the lizard leaping into the sea and, when
allowed, swimming and floating to shore. Lizards
that departed showed more exploratory activity
than those not jumping; lizards departed more dur-
ing sunny than cloudy weather. Results support the
hypothesis that lizards will leave islands of their
own volition if those islands are inhospitable
enough.

INTRODUCTION

A significant portion of Ermest E. Wil-
liams’s scientific contribution has been to
the field of island biogeography.
Through an examination of distributional
patterns in the Caribbean, he inferred the
particular characteristics of Anolis lizards
that predisposed them for island coloniz-
ation (Williams, 1969). He concluded that
the very abundant species of open vege-
tation, A. carolinensis and A. sagrei, were
especially apt colonizers. Williams’s
work in this area to a large degree in-
spired our research on the lizards of very
small islands, research begun in 1974 and
still underway.

The first stages of this research in-
volved several expeditions by boat, dur-
ing which we visited approximately 600

1 Department of Zoology, University of California,
Davis, Califomia 95616, U.S.A.

2Department of Oceanography, University of
Washington, Seattle, Washington 98195, U.S.A.

islands spanning most of the Bahamas.
Among many other characteristics, we
recorded the diurnal lizard species
present on these islands, including A.
sagrei and A. carolinensis. Our studies
revealed that an enormous number of
small islands in the Bahamas had at least
one species of lizard: single-species is-
lands range down to 0.1 to 2.0 acres (1
acre = 0.4 ha), depending on the general
locality.

Surprising though these results were to
us, we could still ask the question: why
do lizards not occur on yet smaller is-
lands? What causes extinction on such
islands, assuming the lizards reach them
occasionally, or at least were there before
the islands became islands, that is before
the final post-Pleistocene submergence?

To investigate this problem, in 1977
we initiated a series of experiments,
seeding 30 islands smaller than the
smallest inhabited-island size with
propagules of A. sagrei or the iguanid
Leiocephalus carinatus. A propagule
comprised 5 to 10 adult individuals, in
the ratio 3 females: 2 males. The details
of these introductions will be reported
elsewhere. Here the following facts are
relevant. First, of the 25 islands seeded
with A. sagrei, three lost their entire
complement of lizards in four to five
days; of the five islands seeded with L.
carinatus, one became empty in eight
days. One of the sagrei introductions
even involved 10 individuals—all disap-
peared without trace. These four islands,
measured in terms of vegetated area,

492

were the smallest representatives from
their respective localities: vegetated
areas for sagrei were 6.3 to 39 ft? (total
areas 3,020-4,750 ft?). Second, somewhat
larger islands lost their individuals more
gradually. Even now, however, four
years after the introductions, islands
down to three orders of magnitude
smaller than the smallest islands na-
turally having sagrei still have popula-
tions, many thriving, that descended
from our propagules. As we will report
elsewhere, this second result plus some
others suggests to us that periodic catas-
trophes, in this case hurricanes, set the
naturally observed single-species island
size well above that inhabitable by
lizards most of the time.

But what about the first result? What
happened to the lizard propagules on our
smallest, most miserable islands? They
clearly did not starve to death or desic-
cate, because initially healthy sagrei can
live without food or water for longer than
four to five days, and because the scal-
loped, pock-marked surfaces of the is-
lands provide much shade for small ani-
mals. Weather was good during this time:
lizards could not have been washed away
by waves. Predators are more likely, but
still unlikely. Terrestrial lizard predators,
such as certain hawks, cannot inhabit
islands as tiny as those in question.
Transient marine birds—herons, terns,
gulls, and oystercatchers—are a possi-
bility; all were commonly seen in the
general area and a few were seen on the
islands in question. However, none of
these birds is primarily adapted for catch-
ing quick and clever lizards, and it is hard
to imagine that they would have found it
profitable, much less possible, so totally
to wipe out our propagules.

We are, however, left with another
alternative. Perhaps the lizards, after siz-
ing up their desperate plight, simply
decided to leave the islands of their own
volition. Better islands lie nearby and are
perhaps visible to lizards, and the future
fitness of individuals not leaving would
with certainty be zero. This hypothesis is

Advances in Herpetology and Evolutionary Biology

made less outrageous by an additional set
of observations. Some lizards, and sagrei
in particular, are excellent swimmers and
especially floaters. We discovered this in
the field, and later in the laboratory we
floated lizards in wave-tanks with sea-
water for varying periods of time
(Schoener and Schoener, in preparation).
Results show that sagrei commonly can
survive, floating without aid, for 12
hours, and some can float for at least 24
hours with no visible harm. Hence a liz-
ard, jumping into the relatively quiet
waters of our study sites, could quite
conceivably reach a larger island, were
the currents pulling favorably.

Despite this rather optimistic assess-
ment, we would be more convinced that
lizards departed islands voluntarily were
we actually to witness it. We could repeat

exactly the seeding of the same islands
used in 1977, but the likelihood is that

we would have to spend days on the is- |

lands before seeing a departure. We

therefore decided to perform a series of —
experiments to encourage early depar- —

tures, making certain variables as ex-
treme as possible, much as physiologists
sometimes do. We placed individual liz-
ards on even smaller, more miserable
islands than those of the 1977 introduc-
tions and watched their behavior. Would

they leave voluntarily, and if so, how ©

would they go about it? These experi-—
ments are the subject of the present —

paper.

SUBJECTS AND METHODS

Our experimental procedure was to
place a single lizard on a rock of cl-2' (1
= (0.305 m) diameter and record its sub-
sequent behavior. If the lizard jumped
from the rock immediately upon release,
the trial was not counted. Rocks were
placed on the average 5 to 10’ from shore,
although tidal flux caused considerable
variation in this distance. The rock on
which the lizard was placed was set upon

LEAPING LIZARDS : Schoener and Schoener

one or more additional rocks such that at
low tide the lizard was still on an island.
Lizards were watched until 1) they
departed from the rock, or 2) if they did
not depart, for varying lengths of time in
half-hour segments up to 6 hours. In the
second case, of course, we followed no
stopping rules, since we had no idea how
long a lizard might remain on a rock be-
fore departing, if it departed at all. How-
ever, we discovered empirically that no
lizard departed a rock once it performed
no major movement (body tum or crawl;
see below) for a half-hour. As it turned
out, we stopped observing four subjects
less than a half-hour since their last major
movement, and because these individu-
als might well have departed their rocks,
we excluded them from certain compari-
sons (see below).

For each lizard we recorded details of
its behavior while on the rock, including
head turns, body turns, number of crawls,
and total distance crawled. We also
recorded the fates of those individuals
departing the rocks.

Experiments were performed April 19-
24, 1979, and April 14—25, 1980, at Staniel
Cay, Exuma Islands, Bahamas. To avoid
wind as much as possible, we rotated
trials between three sites with varying
degrees of wind exposure. All sites were
in fairly protected salt-water “creeks”
with little or no current. Experiments
were mostly performed between 930 and
1600 hrs (Fig. 1), during April the most
favorable time of day for lizard activity.
Total observation time for all trials com-
bined was 5,796 minutes.

Lizard subjects were all Anolis sagrei,
collected from Staniel Cay zero to three
days before the experiments. This is the
same source locality that supplied
colonists for our original 1977 introduc-
tion. No individual was used more than
twice; and with two exceptions, we did
not re-use the same individual during the
same day. With one exception, only
adults were used, and we attempted to
standardize size as much as possible by
selecting the largest individuals.

493

RESULTS AND DISCUSSION

GENERAL OUTCOME OF THE
EXPERIMENTS

We ran 54 trials, 29 with males (28
adults, 1 subadult) and 25 with females
(all adult). Experiments resulted in three
possible outcomes: I) the lizard leapt
from the island; II) the lizard remained
on the island until the experiment was
stopped; and III) the island was nearly
flooded by high tide, and then either a)
the lizard swam off or b) the experiment
was stopped. Among males, 12 jumped
from the island and 17 did not; of the
latter, 11 were in Class II and 6 were in
Class III. Among females, 8 jumped from
the island and 17 did not; of the latter, 13
were in Class II and 4 were in Class III.

To test for sexual differences in the
tendency to depart from islands, we need
to modify these numbers slightly. Be-
cause no lizard jumped from an island
after remaining motionless for at least a
half-hour (see above and below), we
deleted the four cases in which the ex-
periment was terminated less than a half-
hour after the last major movement. We
also deleted the one subadult-male trial.
The modified data show that 12 of 26
males jumped, and 8 of 23 females did. In
other words, about half the males jumped
as compared to a third of the females. The
sexual difference, however, is not statis-
tically significant (x? = 0.663).

TEMPORAL AND CLIMATIC CORRELATES
OF DEPARTURES

Although the primary purpose of our
experiments was to determine whether
or not lizards depart from islands a sub-
stantial proportion of the time, we can in
retrospect search for associations be-
tween departure behavior and environ-
mental characteristics.

Most lizards that departed did so close
to midday. However, most of our obser-
vations were centered around midday
also. As Figure 1 shows, males tended to
jump from their rocks soniewhat earlier

494

e)

Percent
observation time
Ol

3 Jumps > OSes
Time of day TO) im l2
2 Jumps

Percent
observation time

lO

Figure 1.

Advances in Herpetology and Evolutionary Biology

Times of day during which males and females jumped from islands. Top histogram is the percent

observation time in various diel intervals for observations of males only. Bottom histogram is the same for

females. O = times of jumps.

than females, but curiously, our observa-
tions also tended to be distributed over
earlier portions of the day for males than
females. To test whether males departed
earlier in the day than would be expected
from our observational bias, we per-
formed a Kolmogorov-Smirmnov 1l-sample
test. In this test the expected distribution
gives the relative amount of time we
spent observing lizards for each diel
interval, and the observed distribution
gives the proportion of jumps during
each of those intervals. Even though
males tended to jump earlier than ex-
pected from our observational bias, the
difference was not statistically significant
(maximum difference in cumulative
frequency, D = 0.205). And, while fe-
males tended to jump later than ex-
pected, this difference was also not sta-
tistically significant (D = 0.139). Hence
in the absence of more data, we must
conclude that departure times for both

sexes were scattered randomly through-
out the day.

Considerable variation in wind speed
occurred over the course of our experi-
ments. We therefore tested whether or
not animals tended to depart more fre-
quently during times of light wind. Wind
speeds were divided into two categories:
high and moderate-to-slight (We did not
measure wind speed, but during the
times of our experiments the categories
were totally discrete and obvious!). Four
departures took place during high winds,
and 16 took place during relatively wind-
less times. During high winds 10 lizards
did not jump and during light winds 24
did not jump. While a slightly greater
fraction of lizards jumped during rela-
tively still periods, the association is not
statistically significant (x? = 0.489).

During 1980 we recorded the fraction
of the observation time that was cloudy
and the fraction that was sunny. Do liz-

LEAPING LIZARDS : Schoener and Schoener

ards jump more during sunny periods?
Because air temperatures ranged be-
tween 24.8 and 28.0°C (in 1979; in 1980,
temperatures were slightly higher), we
might expect this to be so. Most of our
observation periods were fairly sunny:
only 13% had greater than 50% cloud
cover. Therefore, to divide the data, we
had to use a relatively low cutoff for
“percent cloudy.” Of the lizards that
jumped, 15 of 16 did so during weather
that was cloudy less than 15% of the time.
Of the lizards that did not jump, 13 of 22
were watched during weather that was
cloudy less than 15% of the time. This
difference is statistically significant (x? =
5.739). If 5% clouds rather than 15% is
used, a similar result is obtained (x? =
4.968). Thus lizards tended to depart
from islands more on sunny than cloudy
days.

MOVEMENT CHARACTERISTICS OF
DEPARTING VS. NONDEPARTING LIZARDS

Table 1 gives movement statistics,
measured per-unit-time, for four types of
data: number of crawls, total inches (1
inch = 2.54 cm) crawled, number of body
turns and number of head turns. The
table shows that lizards eventually
departing (Class I) had higher movement
rates in nearly all categories than did
lizards that stayed on the island (Class
II). Lizards whose island was nearly
covered by high tide (Class III) had rates
intermediate between those of Classes I
and II. Thus lizards that eventually
jumped were relatively exploratory,
crawling more and farther, reorienting
more, and looking around more. For
comparisons between Classes I and II,
differences are significant for 1) mean
crawls/min (df = 19, t = 2.152, one- or
two-tailed P < 0.05), 2) mean inches
crawled/min (df = 19, t = 2.084, one-
tailed P < 0.05), and 3) mean body turns/
min (df = 25, t = 1.731, one-tailed P <
0.05), but not for mean head turns/min (df
= 16, t = 1.703; all tests corrected for

495

inequality of variances according to
Bailey, 1959). Lizards that were (or pos-
sibly were) disturbed by water splashing
over their rocks moved more than those
not so disturbed, but not as much as liz-
ards that actually jumped. Of Class III
lizards, several were even partially
covered by water before they swam off,
though in no case did water cover the liz-
ard’s nostrils or mouth.

Table I also shows that males tended to
move more than did females. Differ-
ences, however, are not usually as great
as those between classes of outcomes.
Indeed only in Class I were they consis-
tent and high: males crawled about twice
as far as females, reoriented about twice
as often, and looked around about ten
times as much! Of the several compari-
sons possible, only one is statistically
significant, mean head turns/min: Class I
only (df = 8, t = 2.110, one-tailed P <
0.05), all classes combined (df = 21, t =
2.144, one- or two-tailed P < 0.05). The
greater exploratory behavior of males is a
natural correlate of their tendency to leap
more frequently from islands. Moreover,
as shown elsewhere (Schoener and
Schoener, 1980), males of sagrei on large
islands are more likely to be mobile than
are females.

DEPARTURE CHARACTERISTICS

For animals of Class I, the elapsed time
since the start of the experiment before
jumping into the water varied enor-
mously, from 5 seconds to 208 minutes;
the mean is 56.3 minutes. Males jumped
on average slightly sooner than females
(50.3 vs. 65.4 minutes), but the difference
is not statistically significant. For all data
combined, time before jumping (Fig. 2),
and the number of individuals not jump-
ing by a certain time, are approximately
exponential. Similarly, Moermond (1979)
found that times between movements
roughly resembled exponential distribu-
tions in several species of Haitian anoles.
Such distributions would result from a

496

TABLE l.

I

lizard eventually

jumped*

Type of movement Sex N Xx
crawls/min fe} 9 ~.103

Q 7 O71
inches crawled/min re) 9 225

2 U 111
body turns/min fc) 9 .056

2 U .022
head turns/min fo} 9 315

Q U .039

Advances in Herpetology and Evolutionary Biology

MOVEMENT CHARACTERISTICS FOR LIZARDS PLACED ON VERY SMALL ISLANDS.

Ill
II island nearly
lizard stayed covered with

on island water by end

N x s N x S
104 ll 0386 .037 6 044 .039
064 13.042 ~—S 051 4 047 = =.025
250 11 052 #£.064 6 .061 .061
123 13 .069 8 .105 4 058 =.038
056 ll .009 #.008 6 016 .018
030 13 =©.023~—S 053 4 033 .023
389 ll 07 .089 6 101 .086
045 13 =©.0388 =. .08 1 4 .076 §=.063

*Only cases where the lizard stayed on the island for more than five seconds were included.

constant per-unit-time probability of
jumping or movement.!

While Class I outcomes cannot be
distinguished from those of Class II on
the basis of total duration of the experi-
ment, they are distinguishable on the
basis of time since the last major move-
ment (crawl or body tum). All animals
that jumped did so within a half-hour of
their last major movement, and most did
so considerably sooner than this: only
three jumped more than 5 minutes after
their last major movement. Those
animals not jumping were watched up to
a total of 240 minutes after their last
major movement, and 10 were watched
more than an hour from this time (or from
the start of the experiment, in cases of no
movement at all).

All lizards that jumped did so with a
decisive leap into the water. Jumping

1This is well known in probabilists’ circles but is
given here for the interested. Let kAt be the proba-
bility of movement (e.g., jumping) in a small
amount of time At. In G time units, the probability
of not moving Po 7 = (1 — kAt)© = eT, where AtG
= T. The exponential function Por is the same as
ite fraction of individuals not moving by time T.
The fraction moving between times 0 and 1 is then
Pyro = 1- Po =1-—e-, and between times t and
t+ lis Pye = Poy — Pots) =e kt — e kt) = (il =

é*) (ey Pan 0 ieee Thus the fraction moving in
each of a set of unit time intervals is exponential.

distances varied from approximately 0.5

to 3.5’. Jumping was sometimes, but not
always, oriented toward the nearest land.
Some lizards that jumped were allowed
to swim and float in the water until they
reached shore, a process lasting up to five

10

Number of trials

0 1 2 3 4
Time before jumping (hrs.)

Figure 2. Distribution of the total elapsed experimen-
tal time before jumping for those lizards (Class |) that
jumped from islands.

LEAPING LIZARDS : Schoener and Schoener

minutes. Often they actively swam more
than 4’, and three individuals swam more
than 10’ (Fig. 3). Lizards would com-
monly change direction while floating, in
several cases orienting toward the near-
est above-water object, such as a rock or
mangrove shoot. In no case did a lizard
drown, and most individuals were kept
for one to three days after the experiment
with no apparent ill effects (indeed, some
were reused in a second trial).

CONCLUSIONS

The major objective of our experi-
ments, to show that lizards commonly
leave very small islands on their own
volition, has been accomplished: be-
tween a third and half the trials resulted
in this outcome. Our results during sunny
versus partly-to-entirely cloudy periods
suggest that, were weather conditions
sunnier (and perhaps warmer), the frac-
tion departing would have been greater.
Many of the lizards not jumping were
cold to the touch when removed, and a
number had curled up in rock crevices
and become motionless. Thus it ap-

Number of jumps
(o?)

OE 4 On 68) 10> 12
Distance swum (ft.)

Figure 3. Distribution of the distance swum (not in-
cluding floating) for those lizards that jumped from
islands (1 ft = 0.305 m).

497

peared that sunny periods were generally
those in which activity was possible (or
optimal), rather than those in which ac-
tivity was engendered by thermal stress.
That thermal stress was not the primary
motivation for departures is additionally
indicated by the presence of a departure
on at least one entirely cloudy day and by
the availability of some shade on most of
our rocks during most of the day. More-
over, lizards staying in the sun for long
periods of time and never jumping
showed no apparently ill effects. None-
theless, because of uncontrolled climatic
conditions, we do not wish to ascribe any
significance to the actual fraction of indi-
viduals that jumped during our experi-
ments.

Because our experiments were run at
the same time of year and under the same
general climatic conditions as the origi-
nal 1977 introduction experiments, we do
feel that we have adequately simulated
most conditions of the latter. One might
argue that the greater island size in the
1977 introductions makes the two sets of
experiments not very comparable. Our
guess is that exploratory behavior would
have lasted longer on the larger islands,
but that the eventual outcome would
have been much the same. But of course
we have not proven that lizards volun-
tarily depart from the same-sized island
used in the 1977 experiments. Even if we
had shown this with experiments of the
present type, however, we could not
have proven that lizards actually jumped
from the islands in 1977. We have docu-
mented, however, an exploratory be-
havior on the part of lizards placed on
very small islands, and a rather remark-
able tendency for those lizards to leap
into the sea rather than passively await

death.

ACKNOWLEDGMENTS

Chris Bayer performed some of the
experiments during 1980. Research was
supported by National Science Founda-
tion Grant DEB-22798.

498

LITERATURE CITED

BaILey, N. J. T. 1959. Statistical Methods in Bi-
ology. London, English Universities Press.

MOERMOND, T. C. 1979. The influence of habitat
structure on Anolis foraging behavior. Be-
havior, 70: 147-167.

SCHOENER, A., AND T. W. SCHOENER. in preparation.

Advances in Herpetology and Evolutionary Biology

Experimental zoogeography: floatation of insu-
lar Anolis lizards.

SCHOENER, T. W., AND A. SCHOENER. 1980. Densi-
ties, sex ratios and population structure in four
species of Bahamian Anolis lizards. J. Anim.
Ecol., 49: 19-53.

WILLIAMS, E. E. 1969. The ecology of colonization
as seen in the zoogeography of anoline lizards
on small islands. Quart. Rev. Biol., 4: 345-389.

Experimental Evidence of Strong Present-Day
Competition Between the Anolis Populations
Of the Anguilla Bank—A Preliminary Report

JONATHAN ROUGHGARDEN!
JOHN RUMMEL?
STEPHEN PACALA?

ABSTRACT. Preliminary experimental results in-
dicate strong present-day competition between the
Anolis populations native to the Anguilla Bank. A.
wattsi is absent from xeric habitat in which we
show it can successfully survive and reproduce.
Introductions of A. wattsi to such habitat have low
initial survival if A. gingivinus is present and higher
initial survival if A. gingivinus is partially removed
prior to the introduction. Furthermore, if A. wattsi
is introduced in the presence of A. gingivinus then
the introduced animals are forced into marginal
territories. Introduced animals exhibit great site
fidelity. The panting temperature of A. wattsi is
lower than that of A. gingivinus. The apparent com-
petitive exclusion of A. wattsi by A. gingivinus from
xeric habitat seems to be caused by a carrying ca-
pacity disadvantage of A. wattsi relative to A.
gingivinus in xeric sites, together with a high com-
petition coefficient from A. gingivinus to A. wattsi.
In addition, a four-year survey of three sites on St.
Maarten reveals no change in the rank order of
abundance of all the populations even though the
absolute abundance of every population has fluctu-
ated. The abundance of lizards on an off-shore cay
fluctuated less than those on the St. Maarten sites
during the same period. Hurricane Frederick in
August 1979 caused no reduction in the abundance
of the lizard populations at these sites, including
the off-shore cay.

INTRODUCTION

Studies on the Anolis lizard popula-
tions on the island of St. Maarten have

1,2,3 Department of Biological Sciences, Stanford
University, Stanford, California 94305, U.S.A.

from the beginning revealed exceptional
results. Lazell (1972) in his taxonomic
monograph on the anoles of the Lesser
Antilles reported that the smaller species
from St. Maarten, Anolis wattsi, was in
danger of extinction. He reported also
that A. wattsi from nearby Anguilla had
already become extinct, a consequence of
the destruction of its preferred woody
habitat. Lazell felt also that A. wattsi was
destined for extinction on St. Maarten
because he believed the species to be
concentrated in small, mesic wooded
ravines. In contrast, the native popula-
tions of A. wattsi from the other banks in
the Northern Lesser Antilles are found to
be distributed throughout all habitats on
the island, and were thought to be in no
danger of extinction.

In 1972, Williams highlighted the is-
land of St. Maarten as an exception to the
biogeographic rule of body size that per-
tains to islands with two anole species in
the Lesser Antilles. The basis for these
rules comes from Schoener (1969). Only
the island of St. Maarten, among all eight
islands with two anole species, does not
have a species whose body size in adult
males exceeds 100 mm (SVL). In fact, A.
gingivinus has a body size among adult
males of approximately 60 mm (SVL), the
size that is typical of endemic species
who are the solitary residents of other
islands. Since A. wattsi is only slightly

500

smaller than this “solitary size,” there is
great overlap in body size between the
two species on St. Maarten.

Recently, Roughgarden et al. (1982)
have shown that A. wattsi is not re-
stricted to small mesic wooded pockets
on the island of St. Maarten, but is, in
fact, distributed throughout the hills in
the center of the island. The maximum
elevation of these hills 424 m. The woods
on these hills extend down into the xeric
habitat at sea level along the walls of
ravines. Lazell’s observation that A.
wattsi was restricted only to ravines
results from his observing A. wattsi at the
margin of its distribution where it ap-
proaches sea level elevations through
these wooded ravines.

Roughgarden et al. (1982) also pre-
sented four lines of evidence that sug-
gested the existence of strong present-
day competition between the popula-
tions of St. Maarten. Moreover, they sug-
gested that the absence of A. wattsi from
sea level habitat is caused by competitive
exclusion, in the sense of the Lotka-
Volterra equations, by A. gingivinus.
Thus, they suggest not only that there is
bona fide present-day ecological compe-
tition between these two populations,
but also that it is actually strong enough
to result in the competitive exclusion of
one population by the other in a certain
habitat. The experiments we report here
are designed to provide an experimental
test of this hypothesis. We report pre-
liminary results which indicate that
strong present-day competition is pres-
ent; our data are not sufficient to estab-
lish that this competition is strong
enough to explain the actual absence of
A. wattsi from sea level habitat on the
island of St. Maarten. Also, we present
results of censuses of three sites in St.
Maarten which we have been monitoring
since 1977. These sites serve to indicate
whether there is any long term variation
in population sizes on the Anguilla Bank.

Advances in Herpetology and Evolutionary Biology

RESULTS

Almost all the small cays of the An-
guilla Bank contain populations of A.
gingivinus. The vegetation on these cays
is scrub with occasional or no trees at all.
This vegetation is similar in aspect to, but
more xeric, than the habitat at sea level
on St. Maarten, from which A. wattsi is
absent. Most of the cays are formed from
Pleistocene coral. For this experiment,
we selected a cay known as Anguillita
which is due west of the island of An-
guilla. We have measured its maximum
elevation above sea level as 6 m. This cay
is easily accessible by boat from St.
Maarten and Anguilla, but is rarely
visited. Figure 1 presents a map of this
cay. The map also shows the location of
two of the study sites that we have es-
tablished there. The censusing on the
cays was conducted by a team of four
people. Each site was censused for three
days. The lizards were marked with a dif-
ferent color each day. The mark-
recapture data were analyzed by finding
the simplest log-linear statistical model
consistent with the data and then projec-
ting the total population size based on
this statistical model. This census tech-
nique is presented in detail in Heckel
and Roughgarden (1979).

EXPERIMENTAL INTRODUCTIONS TO
ANGUILLITA

The first site (SK) was established as a
pilot study in August 1979. This site is an
isolated grove (clone) of Sea Grape trees
(Coccoloba uvifera L.). This grove has a
profile that has been sculptured by the
prevailing wind and exhibits a classic
Krummholtz shape. The maximum
height of the trees is 4 m towards the
leeward side and grades down to ap-
proximately 0.5 m towards the windward
side of the site. The site is approximately
circular with an area of 558.6 m/?.

In August 1979, we censused the resi-

ANOLIS POPULATIONS OF ANGUILLA BANK : Roughgarden, Rummel, and Pacala

dent population of A. gingivinus and
estimated its size at approximately 196
lizards. Then we collected 103 A. wattsi
from the hills of St. Maarten. Approxi-
mately half were collected from the
Columbier Ravine at the western margin
of the distribution of A. wattsi, near sea
level. The other A. wattsi were collected
from Pic du Paradis and are slightly
greener in body color than the lizards
near the species’ border. Hence, it was
possible that geographical variation ex-
isted in physiological ability to withstand
drought stress (Hertz, 1981). Both collec-
tions had approximately a 50/50 sex ratio.
The lizards were toe-clipped according
to the location from which they were col-
lected and were then released into the

ANGUILLITA (B.W.I.)

100 Meters

Figure 1. Map of Anguillita.

501

center of the SK site. The resident A.
gingivinus were left undisturbed.

Table 1 presents census data for this
site. The population size of the resident
A. gingivinus was essentially unchanged
in March 1980. However, only 15 (ap-
proximately) remained of the 103 A.

wattsi that had been introduced; eight

distinct animals were seen. Furthermore,
these A. wattsi were found solely at the
periphery of this site where the vegeta-
tion was very low. No A. wattsi had as-
sumed territories in the center of the
grove. The survivors were generally
paired, we observed approximately a 50/
50 sex ratio, and individuals from both
collection locales were equally repre-
sented among the survivors.

In October 1980, and March 1981 the
populations of A. gingivinus at SK had
continued to number approximately 200.
The A. wattsi that had survived the first
six months continued to survive in their
territories at the periphery of the site. By
March 1981 their number had dropped to
about 10. This experiment demonstrates
the ability of A. wattsi of both sexes and
from two collection locales to survive in
the environment of the cay for more than
one year.

Our next study site (MK) was estab-
lished in March 1980. It, too, is a
Krummbholtz of trees (Manzanilla or
Manchineel—Hippomane mancinella L..,
and Sea Grape—Coccoloba uvifera L.).
The area of this site is 262.6 m?, ap-
proximately one half as large as our first
site. The resident population of A.
gingivinus was estimated at 116, also
about one half that of SK. We removed 41
residents, leaving a population of 75 A.
gingivinus, and we then introduced 55 A.
wattsi from St. Maarten into the center of
the site.

When we returned in October 1980, we
discovered that the population of A.
gingivinus had returned to its original
size (116 animals). We attributed this

TABLE |. CENSUS DATA— ANGUILLITA.

502 Advances in Herpetology and Evolutionary Biology
number
Site Date A. gingivinus
SK Aug. 79 196.3 (7.2)
area = May 80 204.6 (9.3)
558.6
sq. m. Oct. 80 180.0 (6.6)
Mar. 81 194.1 (14.5)
MK Mar. 80 115.6 (8.9)
area = 41 removed
262.6
sq. m. Oct. 80 116.4 (5.4)
Mar. 81 106.0 (5.7)

number density density
A. wattsi A. gingivinus A. wattsi
103 intro 35.1 (1.3) 18.4 intro
14.5 (7.0) 36.6 (1.6) 2:6) i(1e3)
(8 seen)
15.0 (2.1) SP (2) 2.7. (0.4)
10.3 (2.6) 34.7 (2.6) 1.8 (0.5)
50. intro 44.0 (3.4) 20.9 intro
28.4 (3.4)
after removal
15.5 (1.7) 44.3 (2.0) 5.9 (0.6)
8.2 (0.6) 40.4 (2.2) 3.1 (0.2)

Note: Standard error of estimate given in parentheses. Density is in units of number of animals per 100 sq. m.

increase to recruitment from outside the
site, as indicated by the large body sizes
of the new residents. Of the 55 A. wattsi
that had been introduced, only 16 re-
mained. This initial survivorship is twice
that observed in the first experiment in
which no residents had been removed, as
discussed further below. Furthermore,
some of the territories of A. wattsi at MK
were located in the center of the site
where they had been released and where
residents had been nosed-out. Finally,
we observed one hatchling of A. wattsi at
this time. By March 1981 the number of
A. gingivinus had dropped to about 106
and of A. wattsi to about eight.

We know that the A. wattsi that we
have introduced at each of these sites
have not merely migrated out of the sites
to set up territories elsewhere because
we have searched extensively for this
possibility.

The principal difference between the
introductions at MK, where the resident
A. gingivinus were partially removed,
and SK where the residents were un-
disturbed, is that the initial survivorship
of the introduced A. wattsi is higher in
MK than SK. But the survivorship of A.
wattsi within MK for subsequent periods
is about the same or possibly even lower
than in SK. By the end of the first period
the A. gingivinus density within MK had
retumed to its original value due to re-

cruitment from outside the site. Here in
the subsequent periods the MK site no
longer serves as an experiment to be con-
trasted with SK but instead as a replicate
of SK for the assay of the longevity of A.
wattsi in the environment provided by
the cay.

The length and season of the initial
periods for MK and SK are different. The
monthly survival probability, 1, can be
calculated based on the following for-
mula,

N(At) = [*N,

where N(At) is the number remaining
after At months, N, is the number intro-
duced, At is the number of months be-
tween census and | is the monthly sur-
vival probability. Upon rearranging we
have

1]
INGA aes
N

O

The number of months for the SK intro-
duction is eight and for the MK introduc-
tion is seven yielding

ANOLIS POPULATIONS OF ANGUILLA BANK : Roughgarden, Rummel, and Pacala

These numbers indicate that after a
standard length of time, say six months,
an introduction with the SK monthly
survival would have 23% survivors and
with the MK value would have a 35%
survival, or almost twice as much.

We suspect the mortality differences
between SK and MK mostly occur within
the first month or two of the introduction
and, if so, the values calculated based on
the entire census interval underestimate
the difference between the sites. Also,
the introduction to MK occurred during
the dry season, which Stamps and Tanaka
(1981) have shown is generally the
period of greatest environmental stress
for Anolis lizards. It is possible that the
MK introduction would have succeeded
even better if it had been made in the wet
season, as was the introduction to SK.

We are repeating these experiments,
and further results will be reported as
they become available.

303

LONG TERM CENSUS ON ST. MAARTEN

Table 2 presents census results from
three sites on St. Maarten begun in 1977.
These results provide the beginning of a
longitudinal study of the population sizes
of lizards on a tropical island. One site,
Naked Boy Mountain, on the windward
side of the island, contains members of A.
gingivinus only. This site is xeric and the
vegetation is scrubby. Maximum height
of the trees here is about 5 m.

Another site, called Boundary, is on the
leeward side of the island, near the spe-
cies border of A. wattsi. It is a xeric site,
but not as xeric as Naked Boy Mountain
above. The maximum tree height is ap-
proximately 6 m and cacti are abundant.
From July 1977 through March 1980 it
contained only A. gingivinus, but in
November 1980 some A. wattsi were
found there as well.

The third site is located at the highest

TABLE 2. CENSUS DATA—ST. MAARTEN.

number

Site Date A. gingivinus
Naked ualyan vn CBE, (783)
Boy July 78 107.9 (11.6)
Mt. July 79 69.6 (9.0)
area = Mar. 80 193.6 (14.3)
215.6 Nov. 80 GEO (Ga)
sq. m. Mar. 81 98.8 (7.6)
Bound- |e 100.4 (4.4)
ary July 78 130.3 (6.0)
area = July 79 130.0 (5.7)
200.6 Mar. 80 260.4 (11.1)
sq. m. Nov. 80 191.4 (6.0)
Mar. 81 144.7 (6.6)

Pic du July 77 38) (AUO.1)
Paradis July 78 56.1 (21.7)
area = July 79 29.9 (24.5)
396.0 Mar. 80 57.9 (12.3)
sq. m. Nov. 80 157.4 (18.6)
Mar. 81 52.8 (11.5)

density

number A. gingiv- density
A. wattsi inus A. wattsi
-0- 43.2 (3.4) -0-

-0- 50.1 (5.4) -0-

-0- 32.3 (4.2) -0-

-0- 89.8 (6.6) -0-

-0- 77.5 (3.1) -0-

-0- 45.8 (3.5) -0-

-0- 50.1 (2.2) -0-

-0- 65.0 (3.0) -0-

-0- 64.8 (2.8) -0-

-0- 129.8 (5.5) -0-

8.8 (3.7) 95.4 (3.0) 44 (1.8)
4.0 (2.0) T2323) 2.0 (1.0)
41.4 (2.6) 8.4 (2.6) 10.5 (0.7)
98.7 (12.9) 14.2 (5.5) 24.9 (3.3)
83.8 (8.8) 7.6 (6.2) PAL — (1.22)
174.0 (20.8) 14.6 (3.1) 43.9 (5.2)
225.0 (10.8) 39.8 (4.7) 56.8 (2.7)
156.1 (9.7) 13.3 (2.9) 39.4 (2.5)

Note: Standard error of estimate given in parentheses. Density is in units of number of animals per 100 sq. m.

304

point on St. Maarten, on Pic du Paradis
(424 m). This site contains both species
and is in the center of the distribution of
A. wattsi. Pic du Paradis is the most
mesic of our sites, and the vegetation
consists of woods whose maximum
height is about 8 m.

All three sites showed an increase in
population size in 1980 for both species.
This broad trend appears to be correlated
with a dry season that was unusually wet
in 1979. In 1980 many more juveniles
were observed than in previous years.
One of the most interesting points about
these data is that they reveal greater fluc-
tuations in population sizes on St.
Maarten than on the cay, Anguillita.

DISCUSSION

The results of our study so far establish
that A. wattsi can survive for over a year
and can produce offspring in habitats
equivalent to the xeric, scrubby habitat
that they do not occupy on St. Maarten.
Furthermore, the results show that the
initial survivorship of A. wattsi upon
introduction into this habitat is higher
when resident A. gingivinus are partially
removed prior to the introduction. These
findings support the hypothesis that the
absence of A. wattsi from sea level loca-
tions around St. Maarten is caused by
competitive exclusion by A. gingivinus in
that habitat.

Our observations suggest that there is a
great site fidelity among introduced
animals. T. Schoener (personal com-
munication) has reported a similar high
site fidelity in experimental introduc-
tions of anoles to small cays in the
Bahamas, provided that the cay already
contains a resident anole. The observa-
tion of high site fidelity also agrees with
anecdotal observations on accidental in-
troductions throughout the Caribbean.
For example, a population of A. porcatus
at the site of a former World Trades Fair
in Santa Domingo in the Dominican
Republic has remained restricted to ap-
proximately three city blocks over many

Advances in Herpetology and Evolutionary Biology

years. Similarly, a population of A. crist-
atellus has remained in a very small area
in the town of La Romana in the Domini-
can Republic since the 1930’s. These
propagules simply do not spread in the
presence of a resident. Still other ex-
amples have been compiled by Williams
(personal communication). We term a
propagule that is not spreading in this
sense an “enclave.”

We are fortunate to be able to report on
the effect of a large hurricane on a popu-
lation of A. gingivinus on the Anguillita
cay. Our census in August 1979 preceded
hurricane David and another large hur-
ricane (Frederick) that came two weeks
later. Hurricane David is the hurricane
that demolished much of Dominica.
Hurricane Frederick came very close to
the Anguilla Bank. Surprisingly enough,
this hurricane produced little structural
damage to the cay. The hurricane threw
chunks of coral up onto the cay, but
caused no apparent damage to the trees.
Our census of March 1980 reveals that
there was essentially no change in the
population size of A. gingivinus in our
study site. This observation suggests that,
at least on small islands, there is little
mortality to Anolis lizards caused by
hurricanes, unless there is severe struc-
tural damage to the vegetation in which
they live.

Finally, we conjecture on the mecha-
nisms that cause the apparent competi-
tive exclusion of A. wattsi by A. gingiv-
inus from xeric sea level habitats. The
phenomenon of competitive exclusion
involves two considerations. The first
pertains to the disadvantage in carrying
capacity (K) that one species has relative
to another. The other pertains to the
existence of a competition coefficient (a)
between the two species. These distinct
considerations arise upon inspecting the
condition for one species to invade a
habitat in the presence of another species
according to the Lotka-Volterra competi-
tion equations. Here, we are interested in
whether A. wattsi can invade low-land
habitat in the presence of A. gingivinus.

ANOLIS POPULATIONS OF ANGUILLA BANK - Roughgarden, Rummel, and Pacala

For A. wattsi to be prevented from invad-
ing the habitat, it is necessary that:

or in words, that the competition coef-
ficient of A. gingivinus against A. wattsi
exceed the ratio of the carrying capacity
of the habitat for A. wattsi relative to the
carrying capacity of A. gingivinus.
Obviously, for A. wattsi to be prevented
from invading the lowland habitat in the
presence of A. gingivinus, both the com-
petition coefficient a,, should be of
moderate magnitude and the ratio of the
carrying capacity for A. wattsi relative to
A. gingivinus should be low.

We speculate that the principal
mechanism contributing to a carrying
capacity disadvantage to A. wattsi in a
xeric habitat relative to A. gingivinus is
that A. wattsi is not as physiologically
well adapted to xeric conditions as is A.
gingivinus. Preliminary data obtained at
Stanford in collaboration with Hal
Mooney and at the University of Wis-
consin in collaboration with Warren
Porter show that A. wattsi has a higher
rate of water loss per unit of surface area
than A. gingivinus. Also, in Figure 2 we
show that A. wattsi on St. Maarten has a
lower panting temperature than A.
gingivinus. This result parallels our
earlier finding that A. wattsi on St.
Eustatius has a lower panting tempera-
ture than A. bimaculatus on that island
(Roughgarden et al., 1982). A. bimacu-
latus is closely related to A. gingivinus,
and this lineage generally appears to
have more xeric tolerance than the A.
wattsi lineage. As a consequence, we
suspect that A. wattsi may find less of the
microhabitat suitable as territories and
may have a more narrow window in its
time budget during which it can forage
than A. gingivinus. Both factors would
translate into lowering the carrying
capacity ratio of A. wattsi relative to A.
gingivinus in xeric habitat.

505

We should add that panting tempera-
ture data not withstanding, individuals of
A. wattsi and A. gingivinus typically
have the same body temperature in the
field at sites where they co-occur, and
there is no evidence that A. gingivinus
prefers hotter microsites than A. wattsi
under field conditions.

There are three mechanisms that cause
a competitive effect of individuals of A.
gingivinus against individuals of A.
wattsi (i.e., that lead to a high a, ,). First,
members of A. gingivinus use insect prey
that A. wattsi would otherwise eat.

Second, A. wattsi and A. gingivinus are
interspecifically territorial. The space
used for territories is the most obvious
resource for which these species are
competing. The mechanism of competi-
tion for this space is interspecific terri-
torial behavior, including both aggres-
sion and defense.

Third, interspecific predation may be
important as well. Large male A. gingiv-
inus have been observed capturing and
eating adult as well as juvenile A. wattsi.
Similarly, A. wattsi are capable of captur-
ing and eating at least juvenile A. gingiv-
inus. Analysis of stomach contents and
direct observation establish that there is
both inter- and intra- specific predation
among Anolis lizards, but the magnitude
of the effect of this form of interference
competition on population dynamics has

PANTING TEMPERATURES-St. Maarten

Anolis gingivinus
16

NUMBER OF INDIVIDUALS
O=aNWA

OoOouNW

PANTING TEMPERATURE (°C)

Figure 2. Distribution of panting temperatures for liz-
ards from St. Maarten.

506

not yet been determined. It is possible,
for example, that the readily observed
association of A. wattsi with rock piles is
not an adaption whose function is to
provide a humid and cool microclimate,
but is a trait whose function is to provide
an escape from predation by A. gingiv-
inus as well as other predators.

In xeric habitat the magnitude of these
competitive interactions of A. gingivinus
against A. wattsi probably increases. We
may summarize these speculations by
imagining a transect from the somewhat
mesic habitat of the hills on St. Maarten
to the xeric habitat of sea level. Along the
transect the competition coefficient of A.
gingivinus against A. wattsi rises while
the carrying capacity ratio of A. wattsi to
A. gingivinus falls. The species border of
A. wattsi lies approximately at the point
where the competition coefficient equals
the carrying capacity ratio. Below this
point only A. gingivinus occurs.

Over the years Emest E. Williams has
introduced many to the Anolis system.
He championed the view that this system
might someday prove as important to
evolution, ecology, and biogeography as
Drosophila has in genetics. Although it is
premature to say with certainty, there is a
real possibility that this vision will be
realized. Our preliminary efforts with
field experiments using Anolis popula-
tions have provided us with great opti-
mism that Anolis will play a leading role
as experimental terrestrial community
ecology develops during the coming
years.

ACKNOWLEDGMENTS

We thank James Bryan of Bryansville,
West End, Anguilla for the permission to

Advances in Herpetology and Evolutionary Biology

conduct these studies on Anguillita and
the Fleming family of St. Maarten for
their permission to census the population
on their property. We also thank Carl and
Vikki Leisegang of the VIKKI TOO for
their seamanship and hospitality and
Stephen Adolph for his able assistance
during all phases of the field work. We
thank Paul Hertz and an anonymous
reviewer for exceptionally helpful com-
ments on the manuscript. We gratefully
acknowledge the support of the National
Science Foundation.

LITERATURE CITED

HECKEL, D., AND J. ROUGHGARDEN. 1979. A tech-
nique for estimating the size of lizard popula-
tions. Ecology, 60: 966-975.

HERTZ, P. 1981. Compensation for altitudinal
changes in the thermal environment by some
Anolis lizards on Hispaniola. Ecology, 62: 515—
O21.

LAZELL, J. 1972. The Anoles (Sauria, Iguanidae) of
the Lesser Antilles. Bull. Mus. Comp. Zool.,
143: 1-115.

ROUGHGARDEN, J., D. HECKEL, AND E. FUENTES.
1982. How coevolutionary theory explains the
biogeography and community structure in
Anolis lizard communities in the Lesser
Antilles. To be published in the proceedings of
a symposium on the Ecology of Lizards, held at
the ASZ meeting in Seattle, December 1980;
to be edited by R. Huey, E. Pianka, and T.
Schoener.

SCHOENER, T. 1969. Size patterns in West Indian
Anolis lizards I. Size and species diversity.
Syst. Zool., 18: 386-401.

STAMPS, J., AND S. TANAKA. 1981. The influence of
food and water on growth rates in a tropical
lizard (Anolis aeneus) Ecology, 62: 33-40.

WILLIAMS, E. E. 1972. Origin of faunas. Evolution of
lizard congeners in a complex island fauna—a
trial analysis. Evol. Biol., 6: 47-89.

Competition Between Anolis and Birds:

A Reassessment
TIMOTHY C. MOERMOND !

ABSTRACT. Birds and anoles show complementary
distributions in the West Indies and Central Amer-
ica. The possible effect of these different taxonomic
groups on each other is assessed in terms of compe-
tition and predation. Birds which are likely compe-
titors and those which are Anolis predators are
poorly represented in the West Indies. The princi-
pal likely competitors of Anolis in the West Indies
are migratory parulid warblers. The warblers
primarily use upper levels of vegetation while
anoles predominantly use the lower levels near the
ground. Within the lower levels lizards may exist at
high population levels and scan many surfaces con-
tinuously, possibly depressing arthropod abun-
dance and/or activity. Their hunting patterns may
exclude insectivorous birds from lower portions of
habitats when lizard densities are high. The ability
of lizards to outcompete birds in particular portions
of habitat is based on the mechanistic view of habi-
tat use derived from the concept of the structural
niche. Anolis populations are also sensitive to
predation by birds, with the reduced populations in
the Central American mainland apparently due to
the high level of avian predation, not to competition.

tion.

INTRODUCTION

Anolis lizards are a conspicuous part of
the fauna of the West Indies both
in species richness and abundance
(Williams, 1969, 1976; Schoener, 1969)
and may be the dominant arboreal insec-
tivores. In contrast, on the Central and
South American mainland, Anolis lizards
are much less abundant (Rand and
Humphrey, 1968; Andrews, 1979). Con-
versely, birds are far more diverse on the

1 Department of Zoology, University of Wisconsin,
Madison, Wisconsin 53706, U.S.A.

mainland and notably depauperate in the
West Indies (Terborgh, 1973; Cox and
Ricklefs, 1977; Terborgh and Faaborg,
1980b).

The possibility of competitive interac-
tions between anoles and _foliage-
gleaning birds is compelling and has
been suggested by a number of workers
(Schoener, 1969, 1970, 1972; Schoener
and Schoener, 1978b; Moermond, 1974,
1976; Morse, 1975; Lister, 1976;
Andrews, 1979; Wright, 1979, 1981).
Competition is suggested by the many
parallels between communities of Anolis
lizards and _ of birds, particularly
of the small foliage-gleaners: both in-
clude generalized insectivores (Schoener
and Gorman, 1968; Hespenheide, 1975;
Andrews, 1979; Wright, 1979), both are
highly arboreal, visual, diurnal foragers
and both are equipped with color vi-
sion (Forbes et al., 1964; Sexton, 1964;
Sillman, 1973).

The demonstration of competition re-
quires evidence 1) that both organisms
are using the same resources and 2) that
this resource use results in adverse
effects to one or both organisms (cf.
Schoener, 1972; Pianka, 1978). In the
case of anoles and small insectivorous
birds, we can infer use of similar re-
sources by similarity in size and taxa of
invertebrate prey (e.g., Wright, 1979) or,
less certainly, by similarity of hunting
locations (i.e., prey taken from same sub-
strates) by animals with similar-sized
trophic appendages (i.e., mouth and bill
lengths). The latter criterion is probably
justified since it is likely that most anoles

508

and small avian insectivores are relative-
ly catholic in taxa choice within similar
size ranges (cf. Schoener and Gorman,
1968; Hespenheide, 1975; Andrews,
1979).

Adverse effects of one organism on
another due to competition are difficult
to document. The usual approach is to
look for complementary distribution, i.e.
distributions in which the density of one
organism is low where the other is high

(e.g., Kiester, 1971). Complementarity
between birds and lizards has been
demonstrated within several island

groups. For example, Wright (1979) has
shown higher lizard densities and lower
bird densities on small than on large
islands in Gatun Lake, Panama. Lister
(1976) refers to another example in the
northern Bahamas where Grand Bahama
and Abaco, which are unusually rich in
small insectivorous birds, support only
one species of Anolis in contrast to
equivalently sized islands farther south.
Schoener and Schoener (1978b) carefully
document an inverse relationship in bird
species richness and Anolis density be-
tween Andros and Bimini in the
Bahamas. Wright (1981) has also shown
for a selected series of West Indian
islands that smaller islands have higher
densities of Anolis lizards but lower den-
sities of birds than do larger islands.
The difficulty in using such comple-
mentary patterns is in determining the
causal factors. Are the patterns primarily
results of competition or is predation in-
volved? If the densities of each group are
affected by the presence of the other,
what factors control the pattern? The con-
cept of the structural niche of Anolis
developed by Rand and Williams (Rand,
1964; Rand and Williams, 1969;
Williams, 1972) offers a mechanistic
approach with which to analyze the pat-
terns of habitat use and hence resource
use by Anolis lizards (see Moermond,
1974, 1979a,b). I shall show here that the
same concept can be usefully employed
to analyze the interactions between
lizards and birds and that such an appli-

Advances in Herpetology and Evolutionary Biology

cation leads to a more precise definition
of the problem. This in turn allows one to
distinguish among competing hypothe-
ses.

HYPOTHESIS: COMPETITION OR |
PREDATION?

The following hypotheses have been
offered to account for complementary
distributions of lizards and birds in main-
land and West Indian habitats:

1) Schoener (1972) suggested that,
“,.. competition with endotherms, the
metabolic and therefore home-range
requirements of which are too great to
allow them to exist on small islands,
could reduce the populations of lizards in
tropical mainlands.”

2) Wright (1979, 1981) argued that
lizards opportunistically harvest re-
sources which are not harvested by birds.
Higher extinction rates for birds on
islands may leave larger portions of
resources available for anoles.

3) Terborgh (quoted in Lister, 1976)
suggested that Anolis lizards, by virtue of
their high abundance, tight packing, and
low extinction rates, out-compete birds
on these islands (i.e., the West Indies).

4) Andrews (1979) suggested that,
‘,.. island anoles, with few predators,
may be food limited relative to heavily
preyed-upon mainland species.“

These statements, although not mutu-
ally exclusive clearly do not agree on the
outcome of competition between lizards
and birds. The foregoing suggestions can
be combined as follows to form two sep-
arate hypotheses:

1) At high species densities birds are
capable of out-competing lizards, but
there are fewer species of birds in the
West Indies due to higher extinction
rates of birds over anoles. This presum-
ably allows higher densities of lizards on
these islands.

2) At high numerical densities, lizards
are capable of out-competing birds, but
lizards are sensitive to predation by
birds. Thus the higher densities of West

«

Indian anoles are presumably due to
much lower numbers of avian predators
compared to the mainland.

The question is not only whether the
birds are responsible for the low densi-
ties of mainland anoles (Andrews, 1979;
Wright, 1979; see also Schoener and
Schoener, 1978b) or whether the anoles
are responsible for the paucity of bird
species in the West Indies (Moermond,
1974; Terborgh and Faaborg, 1980b) but
whether the birds are implicated as com-
petitors or as predators. The answer will
be sought through a series of compari-
sons.

MATERIALS AND METHODS

The observations and data presented
here were gathered on many trips to the
West Indies and Costa Rica during which
I have observed over 50 Anolis species.
Most emphasis here is placed on data col-
lected in Haiti (Nov. 1971 and July-Aug.
1972) and the Dominican Republic
(March 1979). Data on lizards were
derived mainly from detailed chron-
ologies of the behavior of individuals in
the field (described in Moermond, 1974,
1979a). Data on birds were distilled from
daily lists and observations on foraging
behavior taken at the West Indies sites
concurrently with Anolis studies as well
as from extensive observations of under-
story birds at the La Selva Biological Sta-
tion in Costa Rica (Aug.-Dec. 1979, June-
Aug. 1980, and Dec. 1980-Jan. 1981).

AVIAN PREDATORS AND
COMPETITORS

A comparison of mainland and West
Indian habitats shows that, although total
bird densities are not appreciably lower
in the West Indies, there is a striking
decrease in total species diversity and in
the population densities of certain
foraging types—potential lizard preda-
tors, bark-gleaning insectivores and foli-
age-gleaning insectivores (Cox and

ANOLIS—BIRD COMPETITION - Moermond

309

Ricklefs, 1977; Terborgh and Faaborg,
1980b). (Islands such as Trinidad with
geologically recent connections to South
America are included with the main-
land.) If one includes migrants, the totals
for species and foliage-gleaning insecti-
vores on the Greater Antilles increase
substantially but still remain below cor-
responding values for mainland habitat
(Terborgh and Faaborg, 1980a).
Numerous authors have noted that
several families which contribute sub-
stantially to the three foraging types
listed above are entirely absent from the
West Indies (Bond, 1974): Momotidae,
Bucconidae, Dendrocolaptidae, Fuma-
riidae, and Formicariidae. Geographical
distributions for these species suggest
that the members of all five families are
poor over-water colonists (MacArthur, et
al., 1972; Terborgh, 1973); their absence
from the West Indies may be independ-
ent of the presence of high lizard densi-
ties. At a montane site in southern Haiti
(Les Platons) I recorded 38 resident
species of birds and 11 migrants
(Moermond, 1974). Of the resident
species, only six appear as potential
competitors of Anolis: Icterus domini-
censis, Vireo altiloquus, Coereba flav-
eola, Anthracothorax dominicus, Todus
subulatus, and T. angustirostris. Only
the first two are usually considered as
foliage-gleaners. Closer examination sug-
gests that competition between these two
species and the anoles is probably not
intense. The Icterus is much larger than
the anoles, and both I[cterus and Vireo
may eat an appreciable amount of fruit.
The diet of V. altiloquus in Puerto Rico
was reported by Wetmore as 57% fruit
(Wetmore, 1916). Both Coereba and
Anthracothorax are heavily nectarivorus,
and even though both take small insects
around flowers, neither is likely to be a
substantial competitor of Anolis. The two
todies take most of their prey on the wing
by sallying and hovering, but the major-
ity of these are taken from the surface of
foliage at the forest edge (Kepler, 1977;
Moermond, unpublished observation).

510

The todies are thus potential competitors
for these insects, but they are also occa-
sionally predators on small juvenile
Anolis (Wetmore, 1916). Of the six
species, only the nectarivores were seen
daily at the Les Platons site.

In contrast, the 11 migrant parulids I
recorded at Les Platons are all potential
competitors of Anolis. Six species were
seen daily, and three were abundant.
Likewise in the area around Constanza,
Dominican Republic, I recorded 10
species of parulids in two weeks in 1979;
of these, four were abundant.

This predominance of migrant parulid
warblers among the foliage-gleaning
insectivores in the West Indies is well
documented (Lack and Lack, 1972; Lack
1976; Terborgh and Faaborg, 1980a;
Keast, 1980). These migrants may remain
on their wintering grounds in the West
Indies for upwards of seven months
(Keast, 1980) and thus are potentially
important competitors. Pianka (1975) has
stressed the possible influence of mi-
grant birds in Africa in depressing the
Kalahari lizard fauna.

Distinction of birds as anole competi-
tors from birds as anole predators is con-
founded by the dual role of many bird
species. In addition to large birds
of several families, e.g., Momotidae,
Bucconidae, and Cuculidae, that contain
known or potential predators on anoles,
many smaller species of birds are known
to prey occasionally on juvenile anoles.
For example, in the rainforest at La Selva
field station in Costa Rica, at least half of
nearly two dozen understory insectivores
are known or likely opportunistic preda-
tors on juvenile anoles (personal observa-
tion). They include the antbird Thamno-
philus punctatus, which was one of the
principal Anolis competitors considered
by Wright (1979). Clearly many of these
species are likely to have dual roles in
their interactions with Anolis. It is impor-
tant to emphasize that the major sources
of insectivore competitors with Anolis in
the West Indies, the parulid warblers, are
not likely to be significant predators on
anoles due to their small size (Table 1).

Advances in Herpetology and Evolutionary Biology

ARTHROPOD DENSITIES AND FOOD
LIMITATION

Wright (1979) argued that birds, as
competitors, lower the abundance of
arthropod prey used by anoles. Sampling
a series of small islands in Gatun Lake,
Panama, he demonstrated a negative cor-
relation between density of insectivorous
birds (mist net samples) and arthropod
abundance (sweep net samples) and a
negative correlation between bird and
lizard densities. Note that the animal
communities on the small islands Wright
sarnpled are reduced subsets of the ad-
jacent mainland communities and are not
analogous to the West Indian-mainland
comparison. In the latter comparison the ~
correlation between bird and arthropod
abundance is opposite that found by
Wright. Three separate studies of insect
abundance at mainland sites in Costa
Rica and at sites on Puerto Rico and
Dominica have shown a significantly
lower abundance of arthropods on West
Indian islands (Allan et al., 1973; Janzen,
1973; Andrews, 1979) and significantly
fewer large insects in West Indian sam-
ples (Andrews, 1979).

With respect to insectivores Morse
(1975) reports that Odum et al. (1970)
found the biomass of reptiles and am-
phibians in a Puerto Rican forest to be
nearly 30 times that of carnivorous birds
(including insectivores). Despite the
higher metabolic rates of the birds,
Morse suggests from preliminary calcula-
tions of Lavigne [no reference given] that
the birds may account for only 1 to 2% of
the insects consumed by both groups.

Andrews’s (1976, 1979) experimental
studies clearly suggest that island anoles
are food limited but that mainland anoles
are far less so. Andrews has shown that
mainland anoles, when compared with
island species, show faster growth rates,
earlier maturation, lower adult survivor-
ship, smaller adult size and lower popu-
lation densities. Andrews (1979) suggests
that the most likely cause for this differ-
ence is a higher predation rate on the
mainland anoles. Experiments with food

ANOLIS—BIRD COMPETITION - Moermond

dll

TABLE 1. OVERLAP IN SIZE OF TROPHIC STRUCTURE BETWEEN ANOLIS LIZARDS AND FOLIAGE-GLEANING BIRDS AT
TWO SITES IN HISPANIOLA.

Trophic size in birds is given as length of exposed culmen. Data from Cruz, 1980; Lack, 1976; and Wetmore

and Swales, 1931. Trophic size in Anolis is given as head length from the anterior edge of the ear to the tip of

the snout. Data from Schoener, 1969. Scale is nonlinear to accommodate the large number of entries at small

sizes.
LES PLATONS, HAITI CONSTANZA, DOMINICAN REPUBLIC
Bird Lizard | Bird
Bill length (mm) Head Length (mm) Bill length (mm)
22 22 pa
Icterus dominicensis
21 6 Todus subulatus A. cybotes 3 21 A.cybotes 3 Todus subulatus 3 21
2 Todus subulatus Todus subulatus @
Phaenicophilus palmarum
20 A. coelestinus 6 20 A.chlorocyanus 6 20
19 19 19
18 Vireo altiloquus 18 18
17 17 17
16 A.hendersoni 3d 16 A.aliniger 6 16
A. coelestinus 2 A. chlorocyanus
15 15 15
A. cybotes 9 A. cybotes @
14 14 14
A. distichus 3 A. christophei 3
A. distichus 3
13 A. monticola 3 13 13
A. cochranae 3 Seiurus novaeborealis
A. semilineatus 3
12 A.hendersoni 2 12 12
Seiurus aurocapillus A. christophei 2 Seiurus aurocapillus
Mniotilta varia A. distichus 3 Mniotilta varia
A. etheridgei 3
A. distichus 3
11 A. koopmani 3 11
Dendroica petechia A. monticola @
Geothlypis trichis
10 Dendroica coronata A. semilineatus 2 10 A.cochranae 2 Dendroica coronata 10
Parula americana A. etheridgei @ Dendroica tigrina
Dendroica palmarum Dendroica caerulescens
Dendroica tigrina
Dendroica caerulescens
9 Dendroica discolor A. koopmani 2 9 9
Setophaga ruticilla Setophaga ruticilla
8 Coereba flaveola 8 Coereba flaveola 8

supplementation in enclosures showed
no significant increase in growth rates for
A. limifrons on the mainland but a signif-
icant increase in growth for an island
species, bringing it into line with the rate
for limifrons (Andrews, 1976; see also
Licht, 1974). Her data clearly suggest that
island anoles are food limited while
mainland forest anoles are far less so
(see also Stamps, 1977; Schoener and
Schoener, 1978a).

COMPARISON OF EXPLOITATION
PATTERNS

OVERLAP IN SIZES OF TROPHIC
STRUCTURES

We can roughly evaluate the overlap in
prey size of birds and anoles foraging in
similar habitats by examining the relative
sizes of their trophic structures. Bill size
in birds and head size in Anolis have

912

been shown to correlate well with size of
prey items taken (Schoener, 1965, 1967,
1968: Schoener and Gorman, 1968;
Hespenheide, 1975; Howe, et al., in
preparation). The bill lengths for insecti-
vorous gleaners and Todus species for
the two Hispaniolan sites mentioned
above are compared to head lengths of
the Anolis species present at each site in
Table 1. Because the head length
measure of lizards overestimates their
mouth length by a couple of millimeters
(varying from species to species), I would
expect an Anolis with a head length
equal to the bill length of a bird to take
somewhat smaller prey on average than
the bird. Therefore to use Table 1 to
compare average prey size taken by
lizards and birds, the lizard columns
should be shifted downward by a couple
of millimeters relative to the birds. The
sizes presented for the lizards are for
large adults of each sex. All the species
presented hatch out as active insectivores
but with body and head sizes below that
listed for the adults of A. koopmani, the
smallest species in Table 1. Therefore
the ontogenetic series for each species
listed in Table 1 overlaps the trophic
sizes of all smaller species. In contrast
birds fledge and begin foraging with
adult-sized bills. Given these two con-
siderations the parulids clearly fall well
within the range of trophic sizes encom-
passed by the lizards.

FORAGING HEIGHT OVERLAP

Within-habitat comparisons of foraging
locations reveal important differences
between Anolis lizards and small foliage-
gleaning birds in the West Indies. There
is little overlap between the foraging
heights of parulid warblers and those of
Anolis lizards in forest habitats. Data on
within-habitat distribution of Anolis on
Jamaica (Schoener and Schoener, 1971a)
and Puerto Rico (Schoener and Schoener,
1971b) (Table 2) and on Hispaniola (Fig.
1) show that the anoles are active pre-

Advances in Herpetology and Evolutionary Biology

dominantly within 3 m of the ground. In
contrast, parulids and other insectivorous
foliage-gleaners forage predominantly
above 3 m (Lack and Lack, 1972;
Moermond, 1974; Cruz, 1980; Terborgh
and Faaborg, 1980a; Bennett, 1980). The
separation is greater if foraging height
distributions are examined in more de-
tail. Perch height distributions for seven
Haitian Anolis (Fig. 1) show strongly
skewed distributions with the mode
usually within a few centimeters of the
ground. The modes of similar distribu-
tions for the birds are usually above three
meters, and there is a sharp decline in
observations within a meter of the
ground (see Diamond, 1973; Cruz, 1980;
references in Keast and Morton, 1980).

The height overlap distributions sug-
gest different zones of activity for the
two types of organisms. Although it is
tempting to interpret these apparent-
ly reciprocal distributions as niche
shifts, this supposition is difficult to
test. Comparisons of foraging height
distributions in the West Indies, in main-
land wintering grounds, and in breeding
and migrating areas over a range of habi-
tats only emphasize the variability of
such distributions. As expected, the
height of foraging activity is strongly in-
fluenced by vegetation structure and,
obviously, but less easily documented,
by local distribution of available prey
(Parnell, 1969; Morse, 1979; and refer-
ences in Keast and Morton, 1980). Never-
theless, almost all recorded parulid
foraging distributions cited in Keast and
Morton (1980) show modes at middle or
upper ranges of the vegetation, nearly
always above three meters. Although
Todus species may feed below 3 m
(Kepler, 1977), these birds are primarily
taking prey on the wing from leaf tips.

In contrast, at La Selva, a Costa Rican
lowland rainforest, there are at least a
dozen species of ground or foliage-
gleaning insectivorous birds which, I
estimate, spend more than half their
foraging time below three meters. A
similar pattern is suggested by a careful

ANOLIS—BIRD COMPETITION - Moermond

513

TABLE 2. FREQUENCY OF ANOLIS LIZARDS ABOVE TEN FEET (~3 M) ON TWO WEST INDIAN ISLANDS.

No. species No. individuals Percent above
Site in census in census 10 feet (~3 m) Source
Jamaica
Mona 4 2,458 7 Schoener & Schoener, 1971a
Port Antonio open 4 1,097 5 Schoener & Schoener, 1971a
Port Antonio closed 3 749 4 Schoener & Schoener, 1971a
Whitehouse 4 1,656 1 Schoener & Schoener, 1971a
Puerto Rico
Maricao, 20 road edge 3 199 20 Schoener & Schoener, 1971b
Maricao, open forest 3 317 9 Schoener & Schoener, 1971b
Maricao, with cleared 3 70 6 Schoener & Schoener, 1971b
understory
Maricao, forest 3 371 W Schoener & Schoener, 1971b
intermediate
El Verde 3 394 4 Schoener & Schoener, 1971b
7.5!
ra)
2 36 |
=)
3 distichus | coelestinus
ro) = &
45 N=209 N=412
WW
>
(e)
oO
3.0
7)
a cybotes hendersoni monticola koopmani semilineatus
= N=104 N=139 N=100 N=180 N=114
= |
s |.
0 —_ 2 ear scmmgnecams ecemre

PER CENT OF INTERVALS

Figure 1. Foraging height distributions for seven species of Anolis at Les Platons, Haiti (after Moermond, 1979a).
N = the total number of perch heights recorded for each species. © the Ecological Society of America.

survey of data presented by Karr (1971)
and Orians (1969). Pearson (1977) also
observed species of foliage-gleaning ant-
birds feeding predominantly below 3
meters in rainforests in South America.
The primary candidates for Anolis com-
petitors listed in these mainland studies
are suboscines (especially antbirds) and

wrens which are largely absent from the
West Indies.

INFLUENCE OF HABITAT STRUCTURE

The differences in foraging heights
between parulids and Anolis suggest that
the habitat constrains the two groups dif-

ol4

ferently. For our purpose the forest habi-
tat can be divided into two zones: canopy
and understory. Canopy includes the
upper and outer branches, twigs and
leaves of trees. The understory includes
inner limbs and boles of trees as well as
shrubs and ground cover. It is not surpris-
ing to find that foraging in these different
zones requires different adaptations.
Below I will briefly discuss the types of
constraints that apply to each group of
foragers.

FOLIAGE-GLEANING BIRDS

The ability to fly plus the sizes and
hence maneuverability of small foliage-
gleaners would seem to make all height
levels readily exploitable. Although
numerous studies show height separa-
tions among closely-related birds (Lack,
1971; Lack and Lack, 1972; Cruz, 1980),
species often expand their height range
of habitat use on islands where total
species diversity is lower (MacArthur
et al., 1966; MacArthur et al., 1972;
Terborgh et al., 1978; Terborgh and
Faaborg, 1980a). Nevertheless, studies of
niche expansion among colonizing birds
on islands (Keast, 1970; Diamond, 1970;
Ricklefs and Cox, 1972, 1977; Terborgh
and Faaborg, 1973; Terborg et al., 1978)
suggest that not all portions of the habitat
are equally available.

Small differences in habitat use near
ground levels are sometimes associated
with changes in wing shape and especial-
ly foot and leg structure (Dilger, 1956;
Pearson, 1977; Fitzpatrick, 1978;
Moermond, et al., in preparation). The
most specialized morphological adapta-
tions appear to be associated with exten-
sive use of trunks or ground (Bock and
Miller, 1959; Osterhaus, 1962; Diamond,
1970; Feduccia, 1973). Thus, while foli-
age-gleaners may readily use much of the
canopy zone on islands, they do not
appear to readily expand to the trunk or
ground in the West Indies even in the
absence of the usual specialists on these
substrates.

Advances in Herpetology and Evolutionary Biology

One might, however, expect foliage-
gleaners to expand into the understory
shrubs. Nevertheless, those warblers,
such as Dendroica palmarum and D.
caerulescens, which heavily use low
shrubs, primarily do so in the open or at
the edges of forest, not usually under the
canopy. Here again the mainland groups
using this zone, e.g., Formicariidae and
Troglodytidae, are absent or rare in the
West Indies.

ANOLIS LIZARDS

Although anoles are noted for their
arboreal habits and can be found from
ground to canopy, their foraging activity

is heavily weighted towards the ground.

My own detailed observations on the
movements of over thirty species have
shown that anoles are constrained to
move on those surfaces which are either
connected to each other or separated by
no more than half a meter (Moermond,
1979a,b, and unpublished observations).
Anoles use the habitat as if it were a
“jungle-gym” with struts (=perches) of
varying diameters and spacings. A num-
ber of studies have shown the importance
of perch diameter to different species of
anoles (Rand, 1964; Schoener, 1968;
Moermond, 1979a). Any assemblage of
four or more species is likely to include
species with complementary adaptations
such that perch diameters of 2 mm thick
twigs to large trunks are used (see com-
ments on ecomorphs in Rand and
Williams, 1969; Williams, 1972; and
Moermond, 1974).

Different spacings of the habitat “jun-
gle-gyms” also favor different species
types, but here we find that combinations
of small perches and wide spacings
appear difficult for any anole to use effec-
tively (Moermond, 1979a). Such combi-
nations occur predominantly in the
canopy zone. Inter-perch distances in the
canopies of most trees appear to be so
great that most foraging movements are
constrained to connected surfaces. Since
these surfaces are also narrow, the area

ANOLIS—BIRD COMPETITION : Moermond

which can be scanned at one time by a
hunting lizard is small. As a result, there
are frequent changes in scanning loca-
tions and probably low prey capture rates
(Moermond, 1979b). Consequently
anoles cannot easily use much of the can-
opy and observations within the canopy
suggest that the few individuals foraging
there usually penetrate the canopy only
on the outer tips of large branches which
act as “highways” back to other lower
locations (personal observation).
Although there are a few species of
anoles, such as A. valencienni of Jamaica
and the giant anoles (Williams, 1962),
specialized for movement in the crowns
of trees, these species are generally rarer
in any given habitat than other anole
species (see Schoener and Schoener,
197la). The “giant” forms may have
diverged into different trophic niches.
(Schoener [1970] has found that at least
one, A. equestris, is frugivorous.) I have
also observed a large male A. leachi on
Barbuda picking and eating berries.
Other more numerous “canopy” forms
such as the trunk-crown or “green”
anoles (Williams, 1965; Rand and
Williams, 1969) have been observed
foraging in canopy (Andrews, personal
communication; Moermond, unpublished
observation) but still heavily use foraging
sites below three meters (see data in
references cited in Table 2 and Fig. 1).
Clearly birds have a decided advantage
in exploiting the canopy, although I
doubt that the scarcity of anoles there is
due to competitive pressure from birds. It
is more likely due to the inherent diffi-
culties of the nonflying anoles to forage
in that zone. Lack of overlap of the two
groups in the lower zone in the West
Indies is more difficult to explain.

CONSEQUENCES OF HUNTING PATTERNS

In the lower parts of trees, in shrubs
and on other surfaces near the ground,
easy movement from one surface to
another allows anoles (as well as birds)
access to most prey. This fact, coupled

515

with the high lizard densities in these
areas (Odum et al., 1970; Schoener and
Schoener, 1980; Andrews, 1979), sug-
gests that many surfaces may be under
virtually continuous surveillance for
much of the day (e.g., Rand, 1967;
Andrews, 1971; Fleming and Hooker,
1975; Moermond, 1979b and unpub-
lished observations).

Because Anolis hunt by scanning the
surfaces around them for prey, particular-
ly moving prey (Curio, 1970; Moermond,
1981), a dense population of anoles may
depress prey abundance and _ activity
levels in lower habitat zones (cf.
Schoener, 1969; Odum et al., 1970;
Schoener and Schoener, 1978b; see also
Pough, 1980). This may constitute signifi-
cant resource depression for active,
visual foragers such as birds (cf. Charnov
et al., 1976). Schoener (1969) has also
pointed out that, “Anolis, with its several
age-size classes overlapping spatially, is
more likely to uniformly reduce food
than... birds.”

Although both Anolis lizards and foli-
age-gleaning birds find prey principally
by visually scanning branches, leaves or
ground from the same or nearby surfaces,
two differences are particularly notice-
able: 1) the times spent scanning from
each perch, and 2) the rates of prey cap-
ture. By either measure birds appear to
be much more active foragers. Although
few data on rates of movement or prey
capture are available for foliage-gleaning
birds in the West Indies (Moermond,
1974; Moermond and Denslow, in prep-
aration), the data available suggest a rate
of prey capture more than an order of
magnitude greater than that reported for
Anolis in the West Indies (Moermond,
1974; Andrews, 1976).

Several relevant aspects of both birds
and anoles are summarized in Table 3.
The comparison draws from temperate
data on vireos (Williamson, 1971) be-
cause these data are quite complete and
were taken in a similar fashion to those
on parulids and because similar vireos
occur among the resident foliage-

516

Advances in Herpetology and Evolutionary Biology

TABLE 3. A COMPARISON OF METABOLIC RATES, FEEDING RATES, AND RATE OF MOVEMENT OF ANOLIS LIZARDS WITH

SELECTED FOLIAGE-GLEANING BIRDS.

length of
trophic struc. (mm) weight (g)
Vireo spp. 10 -15* 16 -20*
Parulid spp. 7.5 — 14* q =la
Anolis spp. I = 1A 15- 7.5 ||
(40-70 mm) SVL)
RATIOS Vireo/Anolis 2.5 —10
Parulid/Anolis 9 = 45

Values from Roberts, 1955.

Estimates from Lasiewski and Dawson, 1967, Kcal./day = 129 W

*
t
; -Calculated from data in Williamson, 1971.
§

cal./day prey/hr. moves/hr.
6500 — 9000 ¢ 60 —- 120; 810 — 1380 ¢
3600 — 6000 + — 900 — 1800 **
= Sy O8= 62 5 = 25am
30- 100 20 —120 do— 160
= — 70— 180

.724

Data from specimens listed in Appendix II of Moermond, 1974.
|| Estimated from weights for similar-sized lizards given in Jensson, 1973.
{ Estimated from Licht and Jones, 1967, 40-44 cal/gram body wt./day.

# Values from Moermond, 1974.

** Moermond, unpublished data on wintering warblers at Les Platons, Haiti.
++ Values calculated from mean intervals between moves from data in Moermond, 1974.

gleaners in the West Indies (e.g., Cruz,
1980).

The correspondence between move-
ment rate and prey capture rate is con-
sistent for both groups; in general, an
individual foliage-gleaning bird must
find and capture prey at a much greater
pace than an Anolis to survive. The
obvious implication is that a population
of anoles could successfully survive on a
much lower prey base than their avian
competitors (Pough, 1980) and may also
contribute substantially to depressing
that base below the minimum needed for
the birds. If one is to look for significant
competition between birds and lizards it
is here.

The apparent advantage of Anolis in
lower vegetation zones depends on 1)
their manner of hunting which is well
defined and characteristic of the genus,
and 2) their population density. Factors,
such as predation, which consistently
lower population levels, potentially
leave more resources available for birds
to exploit—a contrasting argument to that
proposed by Wright (1979).

DISCUSSION

The hypotheses presented above can
now be reassessed. I find little evidence
that birds are capable of outcompeting
lizards although the possibility cannot be
ruled out. At least lizards are more likely
to be constrained in their resource use by
the habitat rather than by the species of
birds present.

On the mainland, evidence that Anolis
lizards are generally below carrying
capacity is persuasive; predation, not
competition from birds, appears to be the
most likely cause (Andrews, 1979;
Schoener and Schoener, 1978b). In the
West Indies, the high densities of lizards
and paucity of bird species are unlikely
to be due solely to extinction rates
of birds as argued by Wright (1981).
Schoener (1976) has shown that the high
per-species immigration rate for birds
can compensate for their relatively small
population sizes on islands. Lack of par-
ticular species and types of species of
birds in the West Indies is likely due toa
combination of poor colonization ability

ANOLIS—BIRD COMPETITION - Moermond

and diffuse competition from other birds
(Terborgh and Faaborg, 1980b) and pos-
sibly from Anolis (op cit, Moermond,
1974).

Instead, I suggest that where there are
few predators, Anolis may be capable of
excluding insectivorous birds from par-
ticular portions of the habitat via resource
depression by the Anolis (Moermond,
1974, 1976). The paucity of predators on
Anolis in the West Indies would not
seem to be due to the Anolis since they
are ubiquitous. Instead many predators
on Anolis, such as the Saurothera and
Coccyzus cuckoos or Tyrannus fly-
catchers probably depend heavily on
large insects which are relatively scarce
in the West Indies (Andrews, 1979).
These and probably all avian predators
on Anolis take lizards opportunistically
as a minor portion of a typically insectiv-
orous diet. Therefore if anoles are partly
responsible for the depression of insect
densities in the West Indies, it is possible
this “depression” affects their “‘preda-
tors’ also. At least the presence of many
species of “large-headed drop-to-the-
ground birds’ in forests in Costa Rica
does appear to be associated with high
insect densities which include higher
numbers of large insects (Orians, 1969;
Schoener, 1971).

I suggest in summary 1) that Anolis
populations are sensitive to predation by
birds which are capable of depressing the
lizard populations below carrying capa-
city; 2) that anoles may effectively ex-
clude small insectivorous birds from the
lower, more connected portions of forest
habitats if the anoles are near carrying
capacity; and 3) that in forest the small
insectivorous birds only have a competi-
tive advantage in the outer canopy, al-
though the potential carrying capacity of
this portion for lizards may not be high
regardless. The latter two points are
based primarily on a close view of be-
havioral mechanisms of habitat and re-
source use.

The validity of the latter two assess-
ments depends on the accurate identifi-

517

cation of the site of overlap in resource
use between anoles and the relevant bird
species. It is at this same level that accu-
rate identification of site and pattern of
foraging overlap between groups has led
to demonstration of competitive interac-
tion between rodents and ants (Brown
and Davidson, 1977) and between fish
and golden-eye ducks (Ericksson, 1979).
In the present case the structural view
of the habitat vis-a-vis the concept of
the structural niche developed by Rand
and Williams (Rand, 1964; Rand and
Williams, 1969) provided the key for a
mechanistic assessment of habitat use
both within the genus Anolis (Williams,
1972; Schoener, 1975; Moermond,
1979a,b) and among various groups of
arboreal insectivores (Moermond, 1974).

ACKNOWLEDGMENTS

Numerous people have contributed in
many ways to this paper over the past
nine years. T. W. Schoener offered
observations, encouragement, and _ in-
sights throughout. R. M. Andrews, J. S.
Denslow, S. A. Foster, and T. W.
Schoener read the manuscript and
offered numerous useful suggestions.
J. S. Denslow assisted in the Dominican
Republic. Field work was supported by
National Science Foundation (NSF)
grants to E. E. Williams and an NSF grant
to the author. Observations from Costa
Rica were supported by an NSF grant to
the author and J. S. Denslow. The largest
debt is to Ernest E. Williams, whose
understanding of anoles and people pro-
vided the forum for this and other ideas.

LITERATURE CITED

ALLAN, J. D., L. W. BARNTHOUSE, R. A. PRESTBYE,
AND D. R. STRONG. 1973. On foliage arthropod
communities of Puerto Rican second growth
vegetation. Ecology, 54: 628-632.

ANDREWS, R. M. 1971. Structural habitat and time
budget of a tropical Anolis lizard. Ecology, 52:
262-270.

518

1976. Growth rate in island and mainland
anoline lizards. Copeia, 1976: 477-482.

____. 1979. Evolution of life histories: a comparison
of Anolis lizards from matched island and
mainland habitats. Breviora Mus. Comp. Zool.
No. 454, pp. 1-51.

BENNETT, S. E. 1980. Interspecific competition and
the niche of the American redstart (Setophaga
ruticilla) in wintering and breeding communi-
ties, pp. 319-335. In A. Keast and E. S. Morton
(eds.), Migrant Birds in the Neotropics.
Smithsonian, Washington, D.C.

Bock, W. J., AND W. DEW. MILLER. 1959. The
scansorial foot of the woodpeckers, with com-
ments on the evolution of perching and climb-
ing feet in birds. Amer. Mus. Nov. No. 1931, pp.
1-45.

BOND, J. 1974. Birds of the West Indies. 4th Ed.
London, Collins.

BROWN, J. H., AND D. W. DavipDson. 1977. Competi-
tion between seed-eating rodents and ants in
desert ecosystems. Science, 196: 880-882.

CHARNOV, E. L., G. H. ORIANS, AND K. HyaTT. 1976.
Ecological implications of resource depres-
sion. Amer. Nat., 110: 247-259.

Cox, G. W., AND R. E. RICKLEFS. 1977. Species
diversity and ecological release in Caribbean
land bird faunas. Oikos, 28: 113-122.

Cruz, A. 1980. Feeding ecology of the black-
whiskered vireo and associated gleaning birds
in Jamaica. Wilson Bull., 92: 40-52.

Curio, E. 1970. Die Selektion drier Raupenformen
eines Schwarmes (Lepidopt., Sphingidae)
durch einen Anolis (Rept., Iguanidae). Z.
Tierpsychol., 27: 889-914.

DIAMOND, A. W. 1973. Habitats and feeding
assemblages of St. Lucia forest birds. Ibis, 115:
313-329.

DIAMOND, J. M. 1970. Ecological consequences of
island colonization by Southwest Pacific birds.
I. Types of niche shifts. Proc. Nat. Acad. Sci.,
67: 529-536.

DILGER, W. C. 1956. Adaptive modifications and
ecological isolating mechanisms in the thrush
genera Catharus and Hylocichla. Wilson Bull.,
68: 170-199.

ERIKSSON, M. O. G. 1979. Competition between
freshwater fish and goldeneyes, Bucephala
clangula (L.), for common prey. Oecologia, 41:
99-107.

Fepuccia, A. 1973. Evolutionary trends in the
neotropical ovenbirds and woodhewers.
Omith. Monogr. No. 13, pp. 1-69.

FITZPATRICK, J. W. 1978. Foraging behavior and
adaptive radiation in the avian family Tyran-
nidae. Ph.D. thesis. Princeton University.
University Microfilms International Order No.
78-18333. Ann Arbor, Michigan.

FLEMING, T. H., AND R. S. HOOKER. 1975. Anolis
cupreus: the response of a lizard to tropical
seasonality. Ecology, 56: 1243-1261.

ForRBES, A., S. Fox, E. MCCarRTHY, AND E. YAMA-

Advances in Herpetology and Evolutionary Biology

SHITA. 1964. Quantitative response to color shift
in diurnal lizards. Proc. Nat. Acad. Sci., 52: 667—
672.

HESPENHEIDE, H. A. 1975. Prey characteristics and
predator niche width, pp. 158-180. In M. L.
Cody and J. M. Diamond (eds.), Ecology and
Evolution of Communities. Cambridge,
Harvard Univ. Press.

HowE, R. W., J. W. BRUSKEWITZ, T. C. MOERMOND,
AND K. A. RUSTERHOLZ. In preparation. Prey
handling efficiency by captive warblers. II.
Interspecific comparisons.

JANZEN, D. H. 1973. Sweep sample of tropical foli-
age insects: effects of seasons, vegetation
types, elevation, time of day, and insularity.
Ecology, 54: 687-702.

JENSSEN, T. A. 1973. Shift in the structural habitat of
Anolis opalinus due to congeneric competi-
tion. Ecology, 54: 863-869.

Karr, J. R. 1971. Structure of avian communities in
selected Panama and Illinois habitats. Ecol.
Monogr., 41: 207-233.

KEAST, A. 1970. Adaptive evolution and shifts in
niche occupation in island birds, pp. 39-53. In
W. L. Sterm (ed.), Adaptive Aspects of Insular
Evolution. Pullman, Washington State Press.

____. 1980. Spatial relationships between migratory
Parulid warblers and their ecological counter-
parts in the Neotropics, pp. 109-130. In A.
Keast and E. S. Morton (eds.), Migrant Birds in
the Neotropics, Smithsonian, Washington,
D.C.

KeasT, A. AND E. S. MorTON. 1980. Migrant birds in
the neotropics: ecology, behavior, distribution
and conservation. Symp. Nat. Zool. Park,
Smithsonian Inst., Washington, D.C., 576 pp.

KEPLER, A. K. 1977. Comparative study of Todies
(Todidae): with emphasis on the Puerto Rican
tody, Todus mexicanus. Publ. Nuttall Omith.
Club No. 16, pp. 1-190.

KIESTER, A. R. 1971. Species density of North
American amphibians and reptiles. Syst. Zool.,
20: 127-137.

Lack, D. 1971. Ecological Isolation in Birds. Cam-
bridge, Harvard Univ. Press.

____. 1976. Island Biology Illustrated by the Land
Birds of Jamaica. Studies in Ecology, Vol. 3,
Berkeley, Univ. Calif. Press.

LACK, D., AND, P. Lack. 1972. Wintering warblers in
Jamaica. Living Bird, 11: 129-153.

LASIEWSKI, R. C., AND W. R. Dawson. 1967. A re-
examination of the relation between standard
metabolic rate and body weight in birds.
Condor, 69: 13-23.

LIcHT, P. 1974. Response of Anolis lizards to food
supplementation in nature. Copeia, 1974: 215—
Die

LICHT, P., AND R. E. JONES. 1967. Effects of exogen-
ous prolactin on reproduction and growth in
adult males of the lizard Anolis carolinensis.
Gen. Comp. Endocrinol., 8: 228-244.

LISTER, B. C. 1976. The nature of niche expansion

ANOLIS—BIRD COMPETITION - Moermond

in West Indian Anolis lizards. I. Ecological
consequences of reduced competition. Evolu-
tion, 30: 659-676.

MacArTHurR, R. H., J. M. DIAMOND, AND J. R. KARR.
1972. Density compensation in island faunas.
Ecology, 53: 330-342.

MacArTHUuR, R. H., H. RECHER, AND M. Copy. 1966.
On the relation between habitat selection and
species diversity. Amer. Nat., 100: 319-332.

MOERMOND, T. C. 1974. Patterns of habitat utiliza-
tion in Anolis lizards. Ph.D. thesis, Harvard
University.

____. 1976. The basis for resource partitioning in
birds. Abstract. Symp. on Behavioral Allocation
of Resources in Vertebrates. Midwest Anim.
Behav. Conf. Chicago, Illinois.

___. 1979a. Habitat constraints on the behavior,
morphology, and community structure of
Anolis lizards. Ecology, 60: 152-164.

___. 1979b. The influence of habitat structure on
Anolis foraging behavior. Behaviour, 70: 147-
lie

____. 1981. Prey-attack behavior of Anolis lizards. Z.
Tierpsychol. 56: 128-136.

MOERMOND, T. C., AND J. S. DENSLOw. In prepara-
tion. A comparison of the foraging patterns of
migrant Parulid warblers and Méicrolegia
palustris in montane Dominican Republic.

MOERMOND, T. C., R. W. HowE, J. W. BRUSKEWITZ,
AND K. A. RUSTERHOLZ. In preparation. Experi-
ments on perch use in warblers: interspecific
differences due to morphological and be-
havioral attributes.

MorsE, D. H. 1975. Ecological aspects of adaptive
radiation in birds. Biol. Rev., 50: 167-214.

___. 1979. Habitat use by the blackpoll warbler.
Wilson Bull., 91: 234-243.

Opum, H. T., W. ABBoTT, R. K. SELANDER, F. B.
GOLBY, AND R. F. WILSON. 1970. Estimates of
chlorophyll and biomass of the tabonuco forest
of Puerto Rico, pp. 3-19. H. T. Odum (ed.), A
Tropical Rain Forest, Section I. Oak Ridge, U.S.
Atomic Energy Comm.

ORIANS, G. 1969. The numbers of bird species in
some tropical forests. Ecology, 50: 783-801.

OSTERHAUS, M. B. 1962. Adaptive modifications in
the leg structure of some North American
warblers. Amer. Midl. Nat., 68: 475-486.

PARNELL, J. F. 1969. Habitat relations of the
Parulidae during spring migration. Auk, 86:
505-521.

PEARSON, D. L. 1977. Ecological relationships of
small antbirds in Amazonian bird communi-
ties. Auk, 94: 283-292.

PIANKA, E.. R. 1975. Niche relations of desert lizards,
pp. 292-314. In M. L. Cody and J. M. Diamond
(eds.), Ecology and Evolution of Communities.
Cambridge, Harvard Univ. Press.

——.. 1978. Evolutionary Ecology, 2nd Edition.
Harper & Rowe, New York.

Poucu, F. H. 1980. The advantages of ectothermy
for tetrapods. Amer. Nat., 115: 92-112.

519

RAND, A. S. 1964. Ecological distribution in anoline
lizards of Puerto Rico. Ecology, 45: 745-752.

___. 1967. Ecology and social organization in the
Iguanid lizard, Anolis lineatopus. Proc. U.S.
Nat. Mus., 122: 1-79.

RAND, A. S., AND S. S. HUMPHREY. 1968. Interspecif-
ic competition in the tropical rain forest:
ecological distribution among lizards at Belem,
Para. Proc. U.S. Nat. Mus., 125: ]—17.

RAND, A. S., AND E. E. WILLIAMS. 1969. The anoles
of La Palma: aspects of their ecological rela-
tionships. Breviora Mus. Comp. Zool. No. 327,
pp. 1-17.

RICKLEFS, R. E., AND G. W. Cox. 1972. Taxon
cycles in the West Indian avifauna. Amer. Nat.,
106: 195-219.

eee AND) . 1977. Morphological similarity and
ecological overlap among passerine birds on
St. Kitts; British West Indies. Oikos, 29: 20-66.

ROBERTS, T. S. 1955. A manual for the identification
of the birds of Minnesota and neighboring
states. Minneapolis, Univ. Minnesota Press.

SCHOENER, T. W. 1965. The evolution of bill size
differences among sympatric congeneric
species of birds. Evolution, 19: 189-213.

___. 1967. The ecological significance of sexual
dimorphism in size in the lizard Anolis con-
spersus. Science, 155: 474477.

____. 1968. The Anolis lizards of Bimini: resource

partitioning in a complex fauna. Ecology, 49:

704-726.

1969. Size patterns in West Indian Anolis
lizards. I. Size and species diversity. Syst.
Zool., 18: 386-401.

1970. Size pattems in West Indian Anolis
lizards. II. Correlations with the size of partic-
ular sympatric species—displacement and
convergence. Amer. Nat., 104: 155-174.

1971. Large-billed insectivorous birds: a
precipitous diversity gradient. Condor, 73:
154-161.

____. 1972. Competition and the niche, pp. 35-136.
In C. Gans and D. W. Tinkle (eds.), Biology of
the Reptilia, Volume 7, Ecology and Behavior.
New York, Academic Press.

____. 1975. Presence and absence of habitat shift in

some widespread lizard species. Ecol.
Monogr., 45: 233-258.
___. 1976. The species-area relations within

archipelagos: models and evidence from island
land birds. Proc. 16th Internat’]. Omith. Cong.;
Canberra, pp. 629-642.

SCHOENER, T. W., AND G. C. GORMAN. 1968. Some
niche differences in three lesser Antillean
lizards of the genus Anolis. Ecology, 49: 819-
830.

SCHOENER, T. W., AND A. SCHOENER. 1971a. Struc-
tural habitats of West Indian Anolis lizards. I.
Lowland Jamaica. Breviora Mus. Comp. Zool.
No. 368, pp. 1-53.

___. AND ___. 1971b. Structural habitats of West
Indian Anolis lizards. II. Puerto Rican Uplands.

220

Breviora Mus. Comp. Zool. No. 375, pp. 1-39.

SS VAND . 1978a. Estimating and interpreting
body-size growth in some Anolis lizards.
Copeia, 1978: 390-405.

_____. AND . 1978b. Inverse relation of survival of
lizards with island size and avifaunal richness.
Nature, 274: 685-687.

AND . 1980. Densities, sex ratios and popu-
lation structure in four species of Bahamian
Anolis lizards. J. Anim. Ecol., 49: 19-53.

SEXTON, O. J. 1964. Differential predation by the
lizard Anolis carolinensis upon unicoloured
and polycoloured insects after an interval of no
contact. Anim. Behav., 12: 101-110.

SILLMAN, A. J. 1973. Avian vision, pp. 349-387. In
D. S. Farner and J. R. King (eds.), Avian Biol-
ogy, Volume III. New York, Academic Press.

STAMPS, J. 1977. Rainfall, moisture and dry season
growth rates in Anolis aeneus. Copeia, 1977:
415419.

TERBORGH, J. W. 1973. Chance, habitat and dis-
persal in the distribution of birds in the West
Indies. Evolution, 27: 338-349.

TERBORGH, J. W., AND J. R. FAABORG. 1973. Turnover
and ecological release in the avifauna of Mona
Island, Puerto Rico. Auk, 90: 759-779.

___ AND . 1980a. Factors affecting the distribu-
tion and abundance of North American
migrants in the Eastem Caribbean region, pp.
145-155. In A. Keast and E. S. Morton (eds.),
Migrant Birds in the Neotropics. Smithsonian,
Washington, D.C.

____ AND ___. 1980b. Saturation of bird communi-
ties in the West Indies. Amer. Nat., 116: 178—
195.

Advances in Herpetology and Evolutionary Biology

TERBORGH, J. W., J. R. FAABORG, AND H. J. BROCK-
MAN. 1978. Island colonization by lesser An-
tillean birds. Auk, 95: 59-72.

WETMORE, A. 1916. Birds of Puerto Rico. Bull. U.S.
Dept. Agr., 326: 1-140.

WETMORE, A., AND B. SWALES. 1931. The birds of
Haiti and the Dominican Republic. Bull. U.S.
Nat. Mus., 155: 1-483.

WILLIAMS, E. E. 1962. Notes on Hispaniolan herpe-
tology. G. The giant anoles. Breviora Mus.
Comp. Zool. No. 155, pp. 1-15.

____. 1965. The species of Hispaniolan green anoles
(Sauria, Iguanidae). Breviora Mus. Comp.
Zool. No. 227, pp. 1-16.

___.. 1969. The ecology of colonization as seen in
the zoogeography of anoline lizards on small
islands. Quart. Rev. Biol., 44: 345-389.

____.. 1972. The origin of faunas. Evolution of lizard
congeners in a complex island fauna: a trial
analysis. Evol. Bio., 6: 47-89.

____. 1976. West Indian anoles: a taxonomic and
evolutionary summary. I. Introduction and a_
species list. Breviora Mus. Comp. Zool. No.
440, pp. 1-21.

WILLIAMSON, P. 1971. Feeding ecology of the red-
eyed vireo (Vireo olivaceus) and associated
foliage-gleaning birds. Ecol. Monogr., 41: 129-
152.

WRIGHT, J. S. 1979. Competition between insectivo-
rous lizards and birds in central Panama. Amer.
Zool., 19: 1145-1156.

___. 1981. Extinction-mediated competition: the
Anolis lizards and insectivorous birds of the
West Indies. Amer. Nat., 117: 181-192.

A Synecological Study of Surinam Monkeys

RUSSELL A. MITTERMEIER!
MARC G. M. VAN ROOSMALEN?

ABSTRACT. Surinam’s eight monkey species were
studied in the Raleighvallen-Voltzberg Nature
Reserve to determine how these species partition
available habitat and food resources. Cebus apella,
Saguinus midas midas, and Saimiri sciureus were
the most adaptable species and regularly occurred
in all five forest types in the study area. Ateles
paniscus paniscus, Chiropotes satanas chiropotes,
Pithecia pithecia, and Cebus nigrivittatus were far
more restricted in choice of habitat, and Alouatta
seniculus, though it entered all five forest types,
had a strong preferece for one. Seven of the eight
species were found most often in high forest, but
Saimiri was most frequently seen in liane forest and
occupied lower levels of the forest than any other
species. Saguinus was found in edge habitats more
often than the other species, but Saimiri and Cebus
apella regularly used edges as well. Saguinus,
Saimiri, Cebus apella, and Cebus nigrivittatus
were omnivorous, with Saimiri apparently eating
the highest percentage of arthropods. The four
other species were not seen to feed on insects or
other animal life. All eight species ate a high pro-
portion of fruit, with flowers playing only a very
small role. The two largest species (Alouatta and
Ateles) ate the most leaves, whereas the three
smallest (Saguinus, Saimiri, Pithecia) ate no leaves
at all. Chiropotes was found to be an important seed
predator, and Ateles and Alouattu were major seed
dispersers.

INTRODUCTION

Although a number of studies (e.g.,
Gautier and Gautier-Hion, 1969;

1 World Wildlife Fund-U.S. Primate Program, 1601
Connecticut Avenue, N.W., Washington, D.C.
20009, U.S.A. and Department of Anatomical Sci-
ences, SUNY, Stony Brook, New York 11794, U.S.A.

2Rijksinstitut voor Natuurbeheer, Kasteel Broek-
ee 2, P.O. Box 46, 3956ZR Leersum, Nether-
ands.

Charles-Dominique, 1971; Gartlan and
Struhsaker, 1972; Rodman, 1973a, b,
1978; Dunbar and Dunbar, 1974; Quris,
1976; Moreno-Black and Maples, 1977;
Struhsaker and Leland, 1979) have inves-
tigated certain aspects of habitat utiliza-
tion and niche separation in a number of
Old World primate communities, very
little in the way of quantitative syn-
ecological data has been collected on rain
forest-dwelling Neotropical primates.
Hladik and Hladik’s (1969, 1971, 1975)
studies provided quantitative data only
on diet, and other investigations of New
World monkey synecology (e.g., Klein
and Klein, 1975, 1976, 1977; Izawa, 1975,
1976) have been mainly qualitative.

As a graduate student in the Depart-
ment of Anthropology and a long-term
squatter in the Herpetology Department
of the Museum of Comparative Zoology,
R.A.M. was much impressed with and
inspired by the rigorous quantitative
analyses of ecological relationships in
various Caribbean Anolis communities
carried out by Emest E. Williams and
many of his students. Although Neotrop-
ical monkeys are not quite the coopera-
tive subjects that Anolis lizards are, it
seemed possible to apply at least some of
the methods being used to learn more of
monkeys in South America. Consequent-
ly, when we found a suitable study site in
the form of a diverse and undisturbed
primate community in the interior of
Surinam, we decided to make a prelimi-
nary attempt to define quantitatively the
niches occupied by sympatric New

Ps COPPENAME ]
MONDING

[pa

g

- Parimaribo—_

Bs j

ARALEIGHVALLEN- ;

=<4/VOLTZBERG / ABROKOPONDO
}

J £:RESERVOIR
I

ijne

\
Marow!

LP

ZEN : ye

Figure 1. Map of Surinam showing the location of the
Raleighvallen-Voltzberg Nature Reserve and the other
protected areas in the country.

World monkeys and to determine how
they divide up available forest resources.

There are eight primate species in
Surinam, and they include seven mem-
bers of the family Cebidae and one
species of Callitrichidae. The callitrichid
is the golden-handed tamarin (Saguinus
midas midas), which weighs about 500 g
and lives in groups that average six indi-
viduals. The cebids range in size from
the squirrel monkey (Saimiri sciureus),
which weighs about 700 g and travels in
huge groups that can exceed 50 individ-
uals, to the black spider monkey (Ateles
paniscus paniscus), with males and
females averaging about 7.8 kg. The
social structure of spider monkeys is
unique among the Surinam species in
that the animals apparently live in tem-
porary “parties” or “subgroups” within
larger, loosely-organized communities.
Other cebids include the white-faced
saki (Pithecia pithecia) and the bearded
saki (Chiropotes satanas chiropotes),
both of them members of the unusual

Advances in Herpetology and Evolutionary Biology

cebid subfamily, Pitheciinae. The white-
faced saki lives in pairs or small family
groups, weighs about 1.5 to 2.0 kg, and is
the only Surinam species that is strongly
sexually dimorphic in color. The bearded
saki is larger, weighing about 3.0 kg, lives
in group of eight to more than 30 individ-
uals, and is not dimorphic. Two species
of capuchin monkey also occur in
Surinam. The tufted capuchin (Cebus
apella apella) is found throughout the
country and is probably the most adapt-
able monkey. Males average about 3.9 kg
and females 3.0 kg, and the animals live
in groups of six to 20 or more individuals.
The weeper capuchin (Cebus nigrivit-
tatus), in contrast to its relative, is prob-

ably the rarest Surinam monkey, but its.

weight and group size are similar to that
of C. apella. And finally, there is the red
howler monkey (Alouatta seniculus), a
large species that lives in one male
groups averaging four to five animals and
is sexually dimorphic in size, with males
weighing about 8.5 kg and females about
6.0 kg (Mittermeier, 1977).

METHODS

The study discussed here was con-
ducted in the Raleighvallen-Voltzberg
Nature Reserve, a 56,000 ha protected
area established in 1961 and currently
managed by STINASU, the Surinam
Nature Conservation Foundation. In
early 1976, we established a 300 ha study
site at the foot of the 240 m Voltzberg
dome, one of a number of huge granitic
inselbergs found in the reserve.

The study area has four major forest
types and a fifth transition type, all of
which are briefly described in the
Appendix. High forest is the most com-
mon formation, followed by liane forest
and mountain savanna forest, and pina
swamp forest grows along the small
creeks flowing through the reserve. The
Voltzberg area apparently has a greater
diversity of forest types and edges than
would be usual in a 300 ha tract of forest
in the interior of Surinam (J. Lindeman,

SYNECOLOGY OF SURINAM MONKEYS - Mittermeier and Van Roosmalen

personal communication), probably be-
cause of the many granite formations that
do not always provide adequate support
for high forest trees, but do allow the
growth of relatively rare forest types like
mountain savanna forest and liane forest.
As a result of this unusual diversity, it
was possible to study the habitat prefer-
ences of all eight Surinam monkey
species in a much more restricted area
than would have been possible in most
other parts of the country.

The study discussed here was con-
ducted on 147 field days between March,
1976 and February, 1977. Data were
gathered during 12 census walks of the
whole study area, and also while search-
ing for “target species” selected on a
given day. Every time a group was en-
countered, the following kinds of infor-
mation were recorded: time, location in
the study area, forest type, edge or non-
edge situation, activity (resting, travel-
ling, feeding on plant food, foraging for/
feeding on insects), height and stratum in
the forest, how located, path animal and
observer animal distances (for census
purposes), group size and composition (if
possible), and direction of travel.

A sighting was recorded as an edge if
the first animal encountered was within
20 m of a clearing or another forest type.
Feeding on plant food was considered
the activity if the first animal observed
was eating fruits, flowers, leaves, or other
vegetable matter. Foraging for/feeding
on insects was recorded as the activity if
the first animal seen was eating, catching,
or actively searching for insects or other
animal life. For stratification observa-
tions, the forest was divided into six sec-
tions: shrub layer (0-3 m), understory (3—
15 m), lower part canopy (15-20 m),
middle part canopy (20-25 m), upper part
canopy (25-30 m), and emergents (= 30
m). In order to avoid possible bias be-
cause of the presence of the observer,
only first sighting data are used for the
histograms of utilization of forest types,
edges, vertical stratification, and plant vs.
insect eating. Diet histograms, which are

523

unlikely to be affected by observer pres-
ence, are based on all feeding observa-
tions.

RESULTS

HABITAT UTILIZATION

Forest type utilization. High forest
was the most frequently used forest type
for seven of the Surinam species, where-
as the eighth, Saimiri sciureus) was
most often in liane forest (Fig. 4 a—h).
Ateles was found almost exclusively in
high forest; Cebus nigrivittatus was very
dependent on it as well; and Chiropotes
was found only in high forest and moun-
tain savanna forest. Pithecia was similar
to Chiropotes, but also entered liane
forest on occasion. Saguinus, Saimiri,
Cebus apella, and Alouatta spent at least
some time in all five formations, but
Alouatta was more dependent on high
forest than the three smaller species.

Edge utilization. Three of the eight
species studied were regularly seen in
edge situations, with Saimiri and C.
apella being found in edges about half
the time and Saguinus more often than in
non-edges. By contrast, Ateles was almost
never found in edge habitats, and
Pithecia, Chiropotes, C. nigrivittatus
and Alouatta only entered them on
occasion (Fig. 5a-h).

Vertical stratification. Ateles and
Chiropotes were found primarily in the
two highest strata of the forest; Alouatta
was most often in the middle to upper
strata; and Saguinus, Pithecia, and the
two Cebus were mainly in the understory
and lower to middle strata. Saimiri, in
keeping with its preference for liane for-
est, was in the understory stratum more
often than anywhere else. None of the
eight species in the Voltzberg are was
seen on the ground during the present
study, and only three (Saguinus, Saimiri,
C. apella) entered the shrub layer. Two
species (Pithecia, C. apella) were never
seen in emergents. The understory, and

524 Advances in Herpetology and Evolutionary Biology

Figure 2. a. Adult golden-handed tamarin (Saguinus midas midas). b. Young adult male squirrel monkey (Saimiri
sciureus). c. Adult male white-faced saki (Pithecia pithecia). d. Adult female white-faced saki (Pithecia pithecia).
e. Adult bearded saki (Chiropotes satanas chiropotes) (photo courtesty of the Koln Zoological Society).

SYNECOLOGY OF SURINAM MONKEYS -: Mittermeier and Van Roosmalen Byd5)

Figure 3. a. Juvenile tufted capuchin monkey (Cebus apella apella). b. Young adult male weeper capuchin
monkey (Cebus nigrivittatus). c. Adult female (above) and subadult (below) red howler monkey (Alouatta senicu-
lus). d. Juvenile black spider monkey (Ateles paniscus paniscus).

526

SAGUINUS

HIGH RAIN! | LOW RAIN! | MOUNTAIN! 'PiNa LIANE
FOREST FOREST SAVANNA — FOREST FOREST
FOREST

a

PITHECIA

or 00

RAINT U MOUNTAIN! TRINA ~T Uiane
SAVANNA FOREST FOREST
FOREST

Cc

L
HIGH RAIN T1¢
JOREST FOREST

CEBUS APELLA

n=104

HIGH RAIN ' ' LOW RAIN T’ MOUNTAIN! 'PiINA LIANE
FOREST FOREST SAVANNA FOREST FOREST
FOREST
e
ALOUATTA

737 n=74

SIGHTINGS

TOTAL

IAIN” “MOUNTAIN” 'PINA ") UIANE
SAVANNA FOREST FOREST
FOREST

g

% TOTAL SIGHTINGS

S

G

TOTAL SIGHTIN¢

%

% TOTAL SIGHTINGS

SIGHTINGS

TOTAL

Advances in Herpetology and Evolutionary Biology

SAIMIRI

HIGH RAIN! | LOW RAIN| MOUNTAIN! 'PiINA LIANE
FOREST FOREST SAVANNA FOREST FOREST
FOREST

b

CHIROPOTES

n=I8

0.0 ee

HIGH! RAIN
FOREST FOREST

LOW RAIN” MOUNTAIN PINA
SAVANNA FOREST
FOREST

d

CEBUS NIGRIVITTATUS

n=i7
18
5.9
HIGH RAIN’ TLOW.RAINT 'MOUNTAIN | IPINA LIANE
FOREST FOREST SAVANNA FOREST FOREST

FOREST

f

ATELES

n= 118

HIGH RAIN | | MOUNTAIN” 'PiNA
FOREST SAVANNA FOREST
FOREST

h

SYNECOLOGY OF SURINAM MONKEYS - Mittermeier and Van Roosmalen

middle, lower, and upper levels of the
canopy were utilized to varying extent by
all eight species (Figs. 6a-h).

DIET

Herbivory wus. omnivory. Sharp dis-
tinctions can be made in diet since half
the species are partly insectivorous
whereas the other four appear to be en-
tirely herbivorous. Saguinus, Saimiri,
and the two Cebus species regularly in-
clude arthropods in their diet (Fig. 7a-d),
while Pithecia, Chiropotes, Alouatta,
and Ateles were never seen eating
insects or other animals.

Fruits and seeds: seed predation vs.
seed dispersal and seed dropping. A
further distinction can be made between
species that are heavily dependent on
seeds as a food source and those that rare-
ly or never actually digest seeds. Here it
is necessary to distinguish between
ingestion of seeds and digestion of seeds.
All species swallow at least some seeds
along with the exocarp, mesocarp and/or
aril that they eat, but most species simply
swallow the seeds intact without mastica-
tion and pass them through the digestive
system without destroying their powers
of germination. The result is referred to
as endozoochory and is an important
method of seed dispersal for forest trees.
In addition, seeds are often dropped,
especially by the smaller monkey
species, after the softer, edible parts are
removed with the hands or mouth. Seed
dropping may also be a form of seed dis-
persal if the monkey carries the seeds to
another tree before dropping them
(exozoochory). However, in most cases, it
appears that the seeds are simply
dropped out of the tree from which they
are taken and do not travel any further
than if they had fallen by themselves.
This kind of seed dropping is not really a

527

form of seed dispersal, but we have in-
cluded it together with endo- and
exozoochory in Figure 8 because none of
these feeding behaviors destroys the
seeds ability to germinate and also be-
cause it is at times difficult to distinguish
between exozoochory and simple seed
dropping.

Contrasting with these behaviors is
seed predation in which the seeds of
certain families (e.g., Leycthidaceae) are
actually eaten by the monkeys. The seeds
are removed from the fruit, chewed into a
pulp and digested, obviously losing their
powers of germination.

Although all eight monkeys species
sometimes perform endozoochory with
very small seeds of genera like Ficus
(Moraceae), Coussapoa (Moraceae),
Cecropia (Moraceae), Bagassa (Mora-
ceae), Passiflora (Passifloraceae), Duroia
(Rubiaceae), Bellucia (Melastomataceae),
and Hylocereus (Cactaceae), the major
primate seed dispersers are Ateles and
Alouatta. These monkeys ingest many
large and small seeds every day, and
several hundred seeds of up to five
species are at times found in single fecal
samples. Ateles is probably a more im-
portant dispersal agent because of its
much larger range and higher activity
level, but Alouatta also disperses seeds
hundreds of meters from where they
were eaten. All eight species also per-
form exozoochory on occasion, but pri-
mates apparently play only a very minor
role in this form of dispersal in Surinam.

Seed predation appears to be of major
importance for Chiropotes (Fig. 8d), and
quite possibly for Pithecia (Fig. 8c) and
C. nigrivittatus (Fig. 8f) as well, although
the samples for these two species are
quite small. Saguinus, C. apella, and
Ateles sometimes eat seeds, but much
less than Chiropotes (Fig. 8a,e,h), where-

xs

Figure 4. Utilization of the five forest types in the Voltzberg study area. a. Saguinus midas midas. b. Saimiri
sciureus. c. Pithecia pithecia. d. Chiropotes satanas chiropotes. e. Cebus apella apella. f. Cebus nigrivittatus. g.

Alouatta seniculus. h. Ateles paniscus paniscus.

Advances in Herpetology and Evolutionary Biology

y

SIVLIGWH SLVLIGVH
3903 NON 3903

Vb

€el=u
6S6

S4131LV
p

SLVLIBVH SIVLIGVH
3903 NON 3903

668

S4LOdOUIHO

3903 VNid
1S3403 HOIH

SONILHDIS IVLOL %

SONILHOIS TVLOL %

SLVLISVH SLVLIGVH
3903 NON 3903

> 28

VLIVNO1V

to)

SLVLIGVH SIVLIGVH
3903 NON 3903

VIOSHLId

SONILHOIS TVLOL %

SONILHDIS 1WLOL %

4

SLVLIGVH SLVLIGVH
3903 NON 3903

SONILHOIS WLOL %

288 =

SNIVLLIIAIYSIN SNE39

q

S1V1IGVH SLVLIGVH
3903 NON 3903

SONILHOIS 1VLOL %

(e)
oo

62s

ge9 su

IMIWIVS

SLVLIGVH SLVLIGVH
3903 NON 3903

W114adv SNE39

SIVLIGVH SLVLIGVH
3903 NON 3903

€ 29

SANINSVS

(2)
wo

SONILHOIS TWLOL %

‘snosiued snosiued saajy “y ‘sninojuas eyenojy °6 ‘snjeyiaubiu sngaD *} ‘ejjade ejjade sngaD ‘a ‘sajodouiyo
seuejes Ssajodouiy) ‘p ‘ejoeyjid eIDaYjIg ‘9 ‘snasNnoS uiWweS °q ‘sepiw sepiw snuinBeg ‘e ‘eae Apnjs BiaqzyoA 9u} ul sebpe jo uolezinM ~s eunbi4

SONILHDIS TWLOL %

529

SYNECOLOGY OF SURINAM MONKEYS - Mittermeier and Van Roosmalen

y

SONILHOIS TVLOL %
Os (0)

[r- T T T 1 T

00 Yy3AV7

> | § AYOLSHSQNN

AdONVOD
1uVd Y3M071

AdONVI
1uvd JT0GIN

AdONVD

Svs Lud Y3ddN

evisu

4LN3943W3

S43 141V

p

SONILHOIS TVLOL %
OS (0)
r T T T T T

Yy3AV7
8NYHS

AYOLSYSONN

AdDONYD

Gd) Lud Y3M07

AdONVI
1uvd JT00IN

AdONVO
LYVd Y3ddN

dou ;
vil

S43 LOdOUIHO

41N3943W3

‘snosiued snosiued sajajy “y ‘sninojuas eyenojy 6 ‘snjeyiauBblu sngay ‘4 ‘eyjade ejyjade sngayD ‘a ‘sajodoulyo
SPBUBR]ES SA]OAOIIYD “p “BlDaYIIA eIDaYIId ‘9 ‘SNaINIDS WIWIeS ‘gq ‘sepiw sepiw snuinBes “e ‘ease Apnjs 61aqz}O/ 9u} ul eye1}S }SEOJ JO UOHEZINA “9 aunBi4

SONILHOIS IWLOL %

+6 AYOLSY3GNN

AdONVI
1uVd Y3M07

AdONVD

0'0e Luvd 310GIW

AdONVD

Be Lud Y3ddN

cgzxu
O¢!

VLLIVNONV

4IN3943W3

)

SONILHOIS TVLOL %

AYOLSY30NN

AdONVD

8a3 1uVd Y3MO7

AdONVO
Luvd 3T0GIW

AdONVO,
1uVd Y3ddN

Qlsu
4IN3943W3

VIOSHLId

4

SONILHODIS TVLOL %
os (e)
—— T ia T

£ Se

AdONVO
ud ¥3M07

5 AdONVI
OL 4uVd 3T0GIW

AdONVI,
ud Y3ddN

4IN39Y3W3

SNIVLLIAIYSIN SNES9

q

SONILHOIS TVLOL %
OS”

vis AYOLSYZONN

AdONVD
1uVd Y3M07

AdONVID
Lud 3700IN

AdONVI
Luv Y3ddN

eg=u
LN39Y3N3

IYIWIVS

SONILHOIS IWLOL %

Os"

SS lee T

v2

99¢ AYOLSY3ONN

AdONVD
4uVd ¥3M07

AdONVI
Lud J1G0IN

AdONVD
1YVd YaddN

egleu
1N39u3W3

V1 lsdV SN€so

SONILHOIS TWLOl %

AdONVO

982 Luvd Y3MO7

: AdONVO
vee Luvd STOGIN
AdONVD
‘Lud ¥addn

g6 aU
o1f} inz9uaW3

SNNINSVS

SAIMIRI

% TOTAL SIGHTINGS

% TOTAL SIGHTINGS

" FORAGING
FOR INSECTS

FORAGING
FOR INSECTS

CEBUS APELLA

CEBUS NIGRIVITTATUS
T

66.7

% TOTAL SIGHTINGS

% TOTAL SIGHTINGS

FEEDING

FORAGING
FOR INSECTS

FEEDING | FORAGING
FOR INSECTS

c d

Figure 7. Feeding on plant food vs. foraging for/feed-
ing on insects and other animals in the Voltzberg study
area. Only four species were seen feeding on insects
or other animals. a. Saguinus midas midas. b. Saimiri
sciureus. c. Cebus apella apella. d. Cebus nigrivit-
tatus.

as Saimiri and Alouatta were never seen
acting as seed predators (Fig. 8b,¢).
Fruits us. flowers and leaves. Fruits
made up the majority of the plant diet
for all species, with flowers and leaves
usually contributing only a small portion.
Fruits (including seeds) accounted for
over 90% of the plant foods eaten by
Saguinus, Saimiri, Pithecia, Chiropotes,
and C. apella, and over 80% of those
eaten by C. nigrivittatus and Ateles (Fig.
8a-f,h). Alouatta was the only species

Advances in Herpetology and Evolutionaru Biology

that ate a relatively high percentage of
leaves, and even it appears to be highly
dependent on fruit in the wet season
(Fig. 8g). Of the five species eating
leaves, Alouatta was the only one that
included mature leaves in its diet.
Chiropotes, C. apella, C. nigrivittatus,
and Ateles were only seen eating flush
leaves, and the three smallest species
(Saguinus, Saimiri, Pithecia) did not eat
leaves at all (Fig. 8a-h).

Differential utilization of fruit fami-
lies and fruit species. More detailed
analysis of the plant foods eaten by the
eight monkeys indicates that they were
feeding on 151 identified and 30 uniden-
tified species of at least 54 families dur-
ing this study. Only one family, the
Moraceae, was utilized by all eight
monkeys, and no single plant species was
eaten by more than six of them. Seven
monkeys species were observed to eat
members of the Leguminosae Mimosa-
ceae, six fed on various Sapotaceae,
Rubiaceae, and Palmae, and five ate
some Burseraceae, Lecythidaceae,
Myrtaceae, and Passifloraceae. Several
families like the Capparaceae and
Meliaceae were very important seasonal-
ly for some monkeys, but were not eaten
by others. The majority of plant families
were only occasionally used as dietary
supplements for a few monkey species.
Comparison of species rather than fami-
lies reveals even less overlap, since most
plant species are infrequently eaten by
only one or two monkeys species (Mitter-
meier, 1977; van Roosmalen, et al., in
preparation).

Several very interesting examples of
niche separation were evident in utiliza-
tion of the same food species at different
stages of growth. For example, Chiro-
potes broke open unripe Brosimum parin-
arioides (Moraceae) and ate the seeds,
thus acting as a seed predator, whereas C.
apella and Ateles swallowed the meso-
carp and seeds of ripe fruits and later
excreted the seeds, thus acting as dis-
persers. A similar situation was observed
with Ecclinusa guianensis (Sapotaceae).

o31

SYNECOLOGY OF SURINAM MONKEYS - Mittermeier and Van Roosmalen

‘snosiued snosiued sajajy ‘y ‘sninojuas eyenojy 6 ‘snjeyiauBbiu sngay ‘} ‘ejjade ejjade snqayd ‘a ‘sajodosyo seuejes sajodouyo
‘Pp ‘eIDaYjIC BIDAYIIg ‘9 ‘SN@INIOS WIWIeS ‘q ‘sepiw sepiw snuinBes ‘e ‘suoieniasgo BHulpse} jo saquunu |e}0} ay} UO paseg aie sabejusdIa4 ‘ade|d Hulye}
SEM UOI}EPIJd Paas J}! SUILWUS}EP 0} BjQeUN 348M 9M YDIUM JO} Saldeds jue|d asOU} Sepnjou! AYOHayed _¢,, ay, ‘Huiddoip peas ajduuis osje pue (AvOYD00zZ0xa
pue Avoyoo0ozopus) \eiadsip peas jenjoe sapnjou! Aiobayeo ,pasiadsip spaas,, ay, ‘eae Apnjs Biaqz}O/ au} Ul SpOO} jURId JO UOHeZINM ‘8 ainbi4

y 6

S$3AV31 SY¥3M013 SLinus

OSIW S3AV37_, | SY3MO74 suns 6 0
|
4 982 5
50 Lvs xe oe ee
03Su3¢S10 4 4H =| |
4 saaas |] oO iS) S
=o ~ D >
i ie & 4os 6
4 S
|j0S 9 os fe) 9 9
oOo n
oO Q cy m
m m ina} D
al 2 eo Ss 3 I s
5 » ie 5 5
3 ae 8 | 2
BN al ose | i) o
aasuagsia EY
Ss!
‘aasu3dsia| 4
2e1-u 4 Zbeu 5 eu ee-u saa3s_ |
Gee |
4 } 7 =
“001 "eto AIWLLIAINSIN Saad ~°°! v11adv sng€30 °°
S3131V VLLVNO1V SNIVL
p 9 q e
S3Av31
TNE SUAMOTS, | SLINYd | _S3AV37 | 6 _S3AV37_ |_| SY3MO74 Suns | _S3AV37 |
$2 00 00 OO
+ Dal
= 4 fed
| al aT
BY xe BS ae
I pt x a : =
g g ies i g
D> H ~
= =
40S 0 FFE 40S o 40s 9 ‘0S 0
Ge asuadsia @ n | D
m $0335 mm m a)
ay 2D J) elle oes
< Ss BS <
, Bb a Ss 98 5
S2is = = fo) *. =
S g 2 aasuadsio is)
Gaswadsia|] = fa sie |7 @& SEES 5)
sa33s a3asuadsid
$033S
O8 =u 4 Gl=u lZ=u 7 6e5u |
ee ale 4
ec ce| Eee
©'26
~00! J | 4
S3LOdOYIHO VIOSHLId IMIWIVS oe SANINOVS oa

932

Chiropotes fed on the young seeds of this
species, while Ateles swallowed the ripe
fruits and excreted the seeds. Competi-
tion of this kind was seen in the same tree
at the same time, with Chiropotes feed-
ing on the young fruits to such an extent
that few had a chance to ripen to a stage
at which Ateles would find them pala-
table.

Finally, it is necessary to point out that
the sample sizes for diet in Pithecia and
C. nigrivittatus are very small (Fig. 8c,f)
and give only an approximate indication
of the foods eaten by these animals.
However, the data on the other six
species are a good representation, though
by no means complete, of the kinds of
plant foods eaten by these monkeys.
More in-depth coverage of this same
study is provided in Mittermeier (1977)
and Mittermeier and van Roosmalen
(1981), and an additional two years of
data are now in the process of being
analyzed (van Roosmalen, et al., in prep-
aration) and should add greatly to our
knowledge of these monkeys.

DISCUSSION

COMMENTS ON NICHE SEPARATION

It should be clear from the above data
that there is overlap in almost all the
variables considered and that the
monkeys are at least in competition for
many of the fruits of the major edible-
fruit producing families like the Mora-
ceae. However, it should be equally clear
that the eight species are avoiding com-
petition and are dividing up the habitat
in a number of important ways. Some
species (e.g., Ateles, Chiropotes) are
spending most of their time in the upper
parts of high forest, others (e.g., Saimiri)
are in the lower strata of formation like
liane forest; half the species investigated
(Saguinus, Saimiri, Cebus apella, Cebus
nigrivittatus) are eating arthropods and
other animals; others (e.g., Ateles,
Alouatta) are not. Three species

Advances in Herpetology and Evolutionary Biology

(Saguinus, Saimiri, Cebus apella) fre-
quent edges, perhaps to take advantage
of higher insect abundance in such
ecotone areas, whereas others (especially
the highly frugivorous species, e.g.,
Ateles) are rarely found in edges. Some
species act as seed predators (e.g., Chiro-
potes), whereas others (e.g., Saimiri,
Alouatta) rarely or never eat seeds; and
each monkey is feeding on a particular
set of food items that overlaps partly, but
never entirely, with other species.

IMPORTANCE OF THE SYNECOLOGICAL
STUDIES FOR CONSERVATION

Wild populations of nonhuman pri-
mates are diminishing rapidly in many
parts of the world, making the need for
adequate protected areas ever more
urgent. New national parks and reserves
must be created in important areas in
which they do not yet exist, and estab-
lished protected areas must be assessed
to determine if they are of long term con-
servation value. When selecting sites for
new reserves and assessing those already
in existence, synecological data of the
kind presented here can be a very useful
tool. Such data can be used to determine
if habitats in a particular area are likely to
support populations of the primates
found in a given country or region and, if
not, what kinds of forest should be added
for a more complete representation. If
one or two primate species are of major
concern, a reserve can be selected with
their particular ecological requirements
in mind. Data from an undisturbed site
like the Voltzberg area in Surinam can
also be used to assess disturbed areas to
determine the extent of damage and the
potential for recovery as useful primate
habitat.

Synecological data can also be of great
help in determining priorities in primate
conservation. For instance, the study
covered in this paper indicates that
Saguinus midas, Saimiri sciureus, and
Cebus apella are very adaptable species
occurring in a wide variety of habitats,

SYNECOLOGY OF SURINAM MONKEYS - Mittermeier and Van Roosmalen

whereas Ateles and Chiropotes are re-
stricted to undisturbed formations and
are therefore far more vulnerable. Infor-
mation of this kind, used together with
data on hunting and habitat destruction,
makes it possible to formulate an overall
picture of primate status in a particular
country or region, to indicate which
species should be the focus of conserva-
tion attention, and to evaluate potential
reserves with sound ecological princi-
ples. Finally, synecological studies can
be used to monitor primate populations
in established protected areas like the
Raleighvallen-Voltzberg Nature Reserve
in Surinam, and to ensure that these areas
continue to serve their essential conser-
vation functions.

ACKNOWLEDGMENTS

We would like to express our thanks to
J. P. Schulz, H. Reichart, and F. Baal of
STINASU (Surinam Nature Conservation
Foundation) for permission to work in
the Raleighvallen-Voltzberg Nature
Reserve, and to J. G. Fleagle for reading
and commenting on earlier drafts of this
manuscript. R.A.M. would also like to
express special thanks to Emest E.
Williams for continual support, intellec-
tual stimulation, ever-expanding office
space, and a most congenial atmosphere
in which to work during his years at
Harvard.

R.A.M. was partly supported by the
New York Zoological Society, partly by
the World Wildlife Fund-U.S., and partly
by a Graduate Fellowship, a Doctoral
Dissertation Enrichment Grant, and
Grant BNS 7724921 from the National
Science Foundation. M.G.M.v.R. was
supported by WOTRO, the Netherlands
Foundation for Tropical Research.

Figures were prepared by L. Meszoly.

APPENDIX

Forest types in the Voltzberg study
area (modified from Lindeman and

533

Moolenaar, 1959 and P. Teunissen,
personal communication).

HIGH RAIN FOREST

High rain forest is the typical terra firme tropical
forest that is not affected by seasonal fluctuations in
river level. It can usually be divided into three to
four storeys. The upper storey consists of emergents
which reach 40-45 m and, on rare occasions, as high
as 60 m. Below the emergents is the canopy, which
ranges from 15-30 m and can itself be subdivided
into several sub-categories. The understory consists
of slender trees up to about 15 m. Saplings and
undergrowth species make up the bottom 3 m. High
rain forest is very rich in tree species and has far
more edible fruit species than any other formation.
In the Voltzberg area, the understory is dominated
by boegroemaka palms (Astrocaryum sciophilum).

Low RaIN FOREST

We use the term low forest to designate a sub-
category of high rain forest that usually does not
exceed 20 m in height, is far richer in lianes than
neighboring high forest, and has far fewer
boegroemaka palms. It is usually a transition be-
tween high forest rich in boegroemaka and liane
forest, or sometimes between high forest and
mountain savanna forest.

MOUNTAIN SAVANNA FOREST

Mountain savanna forest is a rare formation that
occurs on bauxite hills and low mountains in the
coastal region of Surinam and in parts of the interior
(e.g., the Voltzberg area) where only a thin layer of
soil covers the underlying rock. It is similar to the
better known white and sand savanna forest in
xeromorphy, thin-stemmed aspect, coriaceous
structure of leaves and lack of differentiation into
storeys, but differs in floristic composition. In strik-
ing contrast to high forest, boegroemaka palms are
absent from the understory. The term savanna forest
in general covers several forest types that grow in
areas with very permeable soil that has poor water
retention capacity.

LIANE FOREST

Liane forest is a formation found in areas with
stony lateritic soils that provide bad rooting condi-
tions and poor footholds for trees. It is especially
noteworthy for the absence of storeys. Tall trees do
occur, but they are so widely separated from one
another that no true canopy exists. The space be-
tween the trees is filled with dense tangles of lianes
and vines that grow in abundance because of un-
restricted exposure to sunlight. Although occasional
tall trees may reach 30 m or more, the liane tangle
itself rarely exceeds 10 to 15 m.

+

534

PINA SWAMP FOREST

The term swamp forest is used for a number of
two-storeyed forest types that are inundated to
some extent throughout the year. In the Voltzberg
area, this forest type is represented by what we call
pina swamp forest, a formation dominated by the
pina palm (Euterpe oleracea) and growing in a
narrow belt along the small creeks flowing through
the rain forest.

LITERATURE CITED

CHARLES-DOMINIQUE, P. 1971. Eco-éthologie des
prosimiens du Gabon. Biol. Gabon., 7(2): 121-
228.

DunBakR, R. I. M., AND E. P. DUNBAR. 1974. Ecolog-
ical relations and niche separation between
sympatric terrestrial primates in Ethiopia.
Folia primatol., 21(1): 36-60.

GaARTLAN, J., AND T. T. STRUHSAKER. 1972. Poly-
specific associations and niche separation of
rain-forest anthropoides in Cameroon, West
Africa. J. Zool., 168: 221-266.

GAUTIER, J. P., AND A. GAUTIER-HION. 1969. Les
associations polyspecifiques chez les Cerco-
pithecidae du Gabon. Terre et Vie, 116: 164—
201.

HLADIK, A., AND C. M. HLapixk. 1969. Rapport
trophiques entre vegetation et primates dan la
forét de Barro Colorado (Panama). Terre et Vie,
23: 25-117.

Huapik, C. M. 1975. Ecology, diet, and social pat-
tern in Old and New World primates, pp. 3-35.
In R. Tuttle (ed.), Socioecology and Psychol-
ogy of Primates.

HLADIK, C. M., A. HLADIK, J. BROSSET,

P. VALDEBOUSE, G. VIROBEN, AND J. DELORT-LAVAL.
1971. Le regime alimentaire des primates de
Vile de Barro Colorado (Panama): resultats des
analyses quantitatives. Folia primatol., 16: 85-
2,

Izawa, K. 1975. Foods and feeding behavior of

monkeys in the upper Amazon basin. Primates,

17: 295-316.

1976. Group sizes and compositions of
monkeys in the upper Amazon basin. Primates,
17: 367-399.

KLEIN, L. L., AND D. J. KLEIN. 1975. Social and eco-
logical contrasts between four taxa of Neo-
tropical primates, pp. 59-85. In R. Tuttle (ed.),

Advances in Herpetology and Evolutionary Biology

Socioecology and Psychology of Primates. The
Hague, Mouton.

____. AND’°___. 1976. Neotropical primates: aspects
of habitat usage, population density, and re-
gional distribution in La Macarena, Colombia,
pp. 70-78. In R. W. Thorington, Jr. and P. G.
Heltne (eds.), Neotropical Primates: Field
Studies and Conservation. National Academy
of Sciences, Washington, D.C.

AND . 1977. Feeding behavior of the
Colombian spider monkey, Ateles belzebuth,
pp. 153-181. In T. Clutton-Brock (ed.), Primate
Ecology: Studies of Feeding and Ranging
Behavior in Lemurs, Monkeys and Apes, New
York, Academic Press.

LINDEMAN, J. C., AND S. P. MOOLENAAR. 1959. Pre-
liminary survey of the vegetation types of
northem Suriname. The Vegetation of
Suriname, Vol. 1, Part 2, pp. 1-45.

MITTERMEIER, R. A. 1977. Distribution, synecology
and conservation of Surinam monkeys. Unpub-
lished Doctoral Dissertation, Harvard Univ.

MITTERMEIER, R. A., AND M. G. M. VAN ROOSMALEN.
1981. Preliminary observations on habitat
utilization and diet in eight Surinam monkeys.
Folia primatol. (in press).

MORENO-BLACK, G., AND W. R. MAPLES. 1977. Dif-
ferential habitat utilization of four Cercopith-
ecidae in a Kenyan forest. Folia primatol., 27:
85-107.

Quris, R. 1976. Données comparative sur la socio-
écologie de huit especes de Cercopithecidae
vivant dans une méme zone de forét primitive
periodiquement inondée (Nord-est du Gabon).
Terre et Vie, 30: 193-209.

RopMan, P. S. 1973a. Synecology of Bomean pri-
mates, with special reference to the behavior
and ecology of orang-utans. Unpublished
Doctoral Dissertation, Harvard Univ.

___. 1973b. Synecology of Bornean primates. 1. A
test for interspecific interactions in spatial dis-
tribution of five species. Amer. J. Phys.
Anthrop., 38(2): 655-659.

___. 1978. Diets, densities and distributions of
Bomean primates. In G. G. Montgomery (ed.),
Ecology of Arboreal Folivores. Smithsonian
Institution, Washington, D.C.

STRUHSAKER, T. T., AND L. LELAND. 1979. Socio-
ecology of five sympatric monkey species in
the Kibale Forest, Uganda, pp. 159-227. In
Advances in the Study of Behavior, 9. New
York, Academic Press.

Courtship Behavior in Varanus bengalensis

(Sauria: Varanidae)
WALTER AUFFENBERG!

ABSTRACT. Varanus bengalensis utilizes 19 be-
havioral acts during courtship. These are classified
into contextural and functional classes. Extensive
analysis is carried out on the sequence of adjacent
acts both in two-step and three-step interactions.
Courtship behavior appears to select the most
powerful males for breeding. Though courtship is
initiated by males, most subsequent male acts are
determined solely by the female’s immediately
previous act.

INTRODUCTION

The only published information on
varanid courtship behavior is that of
Varanus komodoensis (Auffenberg,
198la). The distantly related Varanus
bengalensis is generally placed in a sep-
arate subgenus (Indovaranus, fide
Mertens, 1942), and behavioral compari-
sons would be useful. Individuals of
Varanus bengalensis maintained in cap-
tive facilities in Gainesville, Florida,
afforded an opportunity to study the
sexual behavior of adult Varanus bengal-
ensis. Other contributions deal with
sexual differences in nonsexual contexts
(Auffenberg, 1979); agonistic behavior in
adolescents (Auffenberg and Ganci, in
press); and adult combative behavior
(Auffenberg, 1981b).

The primary purposes of this contribu-
tion are: 1) to describe courtship in this
Species in respect to its component be-
havioral acts; 2) to determine the degree
of behavioral act complexity; 3) to deter-

1Florida State Museum, University of Florida,
Gainesville, Florida 32611, U.S.A.

mine the size of the act repertoire; and 4)
to describe how these factors are related
to the reproductive strategy, through
courship, of this species.

MATERIALS AND METHODS

The data comprising the basis of this
study were obtained by review and anal-
ysis of closed circuit TV tapes. The
equipment included one remote camera
appropriately mounted in each of two
greenhouses (56 m? each). The cameras
ran continously for eight hours each day
(0800-1600 hrs) for a total of 720 days,
commencing in June 1977 and continu-
ing to the end of July 1979. Each camera
was connected to a time lapse VIR
automatic recorder set to advance the
tape at 0.3 sec intervals. This allowed for
rapid review when set at a normal replay
speed, yet preserved sufficient detail of
the recorded social interactions. In addi-
tion to this analysis system, motion pic-
tures (Super-8) were made whenever
practical, often within a 2 m range, so that
details of the behavioral acts were clearly
evident when reviewed by a stop-motion
film projector, using a projection system
similar to that described by Jenssen
(1977).

Each of the two greenhouses usually
contained two adult males and four adult
females. During the reproductive season
the males were sometimes shifted from
one greenhouse to another to encourage
courtship interactions with the “resi-

336

dent” females. The four adult males had
an average total length of 149 cm, (X
SVL xX 58.0 cm; X weight = 2743 g), and
the eight mature females had an average
length of 119.3 em (X SVL = 46.3 cm; X
weight 1452 g).

Of 351 sexual interactions recorded,
only 217 proceeded through the mount-
ing phase, including 21 in which coition
was completed. These 217 completed
courtship sequences form the basis of
this report.

Behavioral Acts. Courtship is defined
as comprising those acts leading to, but
excluding, intromission itself. In this
paper the term “act” is used to designate
a particular behavior pattern or display
performed; the term “step” denotes the
place in the sequence in which an act
occurred.

Analyses. Many recent investigators of
animal communication have used transi-
tion matrices to describe their data.
Among reptile behaviorists only Cooper
(1977) and Gillingham (1979) have ana-
lyzed their data by this method so far.
Transition matrices were constructed by
enumeration of all observed one-step
transitions. This method assumes that
both the transition probabilities and the
unconditional probabilities of single acts
are constant over the length of the inter-
action; that is, the probabilities exhibit
“stationarity.”

The analysis technique follows that
outlined by Oden (1977), in which sta-
tionarity is not assumed, and which
allows assessments of the significance of
dependence between acts separated by
an arbitrary number of steps.

In this type of study it is important to
determine whether the observed number
of adjacent lower order interactions (one-
step transitions; i.e., step 1 followed by
step 2, step 2 followed by 3, etc.) is signif-
icantly different from expected. Expect-
ed frequencies for each of the possible
act pairs in adjacent steps were calcu-
lated by the formula, Expected =
(A,)

Sa where A represents one act

Advances in Herpetology and Evolutionary Biology

of the pair and B the other and the sub-
scripts 1 and 2 the adjacent step numbers.
Expected frequencies of higher order
interactions (three steps at a time) were
calculated by use of the formula:

_ observed A,B, (observed B,C,)
Expected A, B,C,= wed Bt an B, ;
where A, B, C are left, and subscripts 1, 2,
3 are the step numbers. Rejection of the
assumption means that dependence of C,
on A, is not transmitted through B,.

RESULTS

There are 19 behavioral acts (= motor
patterns of some behaviorists, including
Gillingham, 1979) that are common com-
ponents of the courtship of Varanus
bengalensis. These are defined below
(letters in parentheses are abbreviations
used). More elaborate descriptions and
figures of most of these acts are found in
Auffenberg (1981b).

Following an earlier paper (Auffen-
berg, 1981b), behavioral acts included in
courtship are categorized into several
contextural and functional classes. These
are weaponry use, warning, and investiga-
tory tactics and signals used in social
(dominance-subordinance) or reproduc-
tive contexts.

WEAPONRY CLASS

Biting (Bi). In an agonistic context
during the early phases of courtship bit-
ing is used rarely by females. Bites
administered by females are indiscrimi-
nately administered on the head, body,
neck, or legs. Males mounted on females
also bite, but only in this context and only
in a specific place: females are bitten on
the dorsal skin of the neck, where the
scales tend to be largest. It appears to be
an immobilization technique and is com-
mon in many lizard groups (for review
see Carpenter and Ferguson, 1978).

Tail Slap (TS). A rapid lateral swing of

the tail, often, but not always, following
tail coiling (see below). The tail slap
usually terminates the interaction. The
blow is swift and smartly delivered,
usually striking the other individual on
the side of its body, or less often the
head. Tail slapping in combat has been
reported in V. niloticus, varius, spenceri,
mertensi, and salvadori (for review see
Auffenberg, 198 1a).

WARNING CLASS

Tail Coil (TC). Partially or completely
coiling the tail in the horizontal plane,
usually preceding and leading to a tail
slap. Reported in V. komodoensis
Auffenberg, 198 1a).

Hiss (Hi). An auditory cue, used dur-
ing courtship only in a defensive context
by females when approached by males.

DOMINANT CLASS

Topping (Tp). One individual (usually
the female in a courtship context) puts
one of its feet on the back of the male or
even climbs upon him (= riding of
Carpenter and Ferguson, 1978). It only
occurs during or after courtship and com-
bat.

SUBORDINANT CLASS

Walks Away (W). One individual
walks (or runs) away from the other; may
be a neutral signal in some contexts, but
clearly subordinate in most. Also re-
ported in a subordinate context in V.
komodoensis (Auffenberg, 1978b, 198 1a)
and V. salvator (Vogel, 1979).

The following three acts may be func-
tionally identical, i.e., they simply serve
to move the head away from the tongue of
the initiator (avoidance = “turning away”
of Vogel, 1979).

Head Down (HD). Head tilted down-
ward and held lower than the body axis.
Occurs in investigative context.

Head Away (HA). Head moved lateral
to central body axis and held there.
Occurs in investigative context.

VARANUS BENGALENSIS - Auffenberg 537

Head Up (HU). Head lifted above the
trunk and held there. Occurs in several
contexts, but most common in investiga-
tive context. In an agonistic context sig-

nifies social dominance (Auffenberg,
1981b; Greenberg, 1977; Weaver, 1970).

INVESTIGATIVE CLASS

Tonguing (To). Touching or extending
the tongue toward another individual. It
occurs prior to combat or courtship. Con-
comitant pressing of the snout (= touch
of Gillingham, 1979, in snakes) against
the body is particularly common in
Varanus komodoensis (Auffenberg,
1978b), but rare in V. bengalensis.

Following (F). Pursuit of one individ-
ual by another (= chase of Gillingham,
1979; Vogel, 1979); chiefly in courtship.
It is frequent in the courtship of V.
komodoensis (Auffenberg, 198la) and
in an agonistic context in V. salvator
(Vogel, 1979).

STRESS REACTION CLASS

Gular Extension (G). Enlarging the
throat by both lowering the hyoid appa-
ratus and inflating it with air, often
accompanying a hiss (= inflated throat of
Carpenter and Ferguson, 1978; Vogel,
1979). Common in most varanid species
(Auffenberg, 1981a).

REPRODUCTIVE CLASS

Mount (M). One individual (in this
context always the male) assumes a
superior position on the other (female),
with two or all four legs positioned on the
opposite sides of the lower one (female).

Scratching (S). The male (in courtship
context) scratches feebly with one foot,
usually on the shoulder or middle of the
back of the female (Fig. 1A), often when
the latter is walking away. The identical
acts occurs in Varanus komodoensis
(Auffenberg, 198 1a).

Chin Rubbing (C). Male in mounted
position jerkingly rubs his chin on
the dorsal surface of the neck and/or

6) |
Oo
Co

head of the female (Fig. 1B). This may be
functionally related to the chin ad-
pression of male snakes to the head or
body of females (see Gillingham, 1979).
It also occurs in Varanus komodoensis
(Auffenberg, 198 1a).

Head Twitch (HT). When mounted by
a male, the female often twitches the
head laterally in a jerky pattern. It is also
common during the courtship of V.
komodoensis. It may function as a sexual
rejection signal, for (1) the male often
follows this with vigorous chin rubbing,
and (2) head twitching by the female after
a long courtship sequence usually leads
to the male walking away, or doing noth-
ing (see below). Not observed in Varanus
komodoensis. The “head jerk” described
by Vogel (1979) in agonistic encounters
of V. salvator is apparently not the same.

Tail Lift (TL). A mounted male tries to
lift the tail base of the female with one of
his hind feet by stroking upward with the
claws (Fig. 1B). The identical act occurs
in Varanus komodoensis.

MISCELLANEOUS

Do Nothing (N). While this act is intui-
tively neutral in effect, I agree with
Hazlett and Bossert (1965) that for a re-
spondent individual to “do nothing” may
be a very meaningful act in many con-
texts. As will be shown below, it is
apparently important in the courtship of
Varanus bengalensis, for the signal of a
female making no response to a courting
male seems to encourage further court-
ship acts by the male.

Brachial Embrace (Br). In the
agonistic context, this act (see Auffen-
berg, 1981b) is used by bipedal males
in ritualistic combat, whereby both com-
batants clasp one another with their front
legs. Its major function is to hold both
individuals in a ventrally adpressed
bipedal position during the wrestling
phase. Brachial embrace of the female by
the mounted male is also a conspicuous
part of the later phases of courtship prior
to copulation (Fig. 1C). As in the agon-

Advances in Herpetology and Evolutionary Biology

istic context, the primary function seems
to be to immobilize the outer individual,
in this case by pressing the front legs of
the female to her sides and making it dif-
ficult for her to move forward out from
under the mounted male. In a sense, it
provides the same function as the partial-
ly immobilizing xiphiplastral ramming in
land tortoises (e.g., Geochelone, Auffen-
berg, 1964, 1978a), the biting of female
lizards on the neck by mounted males
(Carpenter and Ferguson, 1978), and the
laterally draped body of the male over
that of the female during snake courtship
(Carpenter, 1977).

Table 1 shows that the most intensive
activity in captivity in Florida begins in
early April and ends in late May, with
peak activity during the middle of this
period. At the same time, it is clear that
the number of behavioral acts from court-
ship initiation to mounting decreases
steadily from six to one as the season
progresses, so that courtship is both more
common and shorter later in the season.
Coition was noted from as early as June
11 to as late as August 30. Eggs were laid
from August 16 to October 3, with most
during the height of the Florida rainy
season (late August through September).

In Florida, captive-reared females laid
their first eggs during their fourth year of
life, though the males engaged in some
courtship behavior (unsuccessfully) dur-
ing the preceding year. Among mature
adults, unsuccessful courtship was
apparently due to the female slipping out
from under the mounted male, caused
mainly by the fact that the females were
not being held tightly enough during the
brachial embrace. Thus, the success in
courtship is at least partially dependent
on the size (= strength) of the males.
Smaller males not only had difficulty im-
mobilizing females with the brachial
embrace, but their hind feet were too far
forward to lift the tail base of the females
prior to coition. In general, the larger the
male in respect to the female, the better
chance for successful courtship to have
taken place.

VARANUS BENGALENSIS - Auffenberg 539

ae MLZ hin,

Figure 1. Major behavioral acts comprising the reproductive class in Varanus bengalensis. (A) Scratching of the
back of the female by the male, usually as she walks away. (B) When mounted the male rubs chin horizontally
over neck and head of female as she twitches head back and forth. (C) The female is grasped in a brachial
embrace by which her front legs are pinioned to the sides of her body by the front legs of the male. At the same
time the male uses one of his hind feet to stimulate the female to lift the base of her tail by raking upwards with the
claws.

TABLE 1. DATES AND NUMBER OF MALE BEHAVIORAL ACTS FROM INITIATION OF COURTSHIP TO THE FIRST MOUNTING
ATTEMPT IN CAPTIVE VARANUS BENGALENSIS.

Average Number Number
Inclusive Dates Male Behavioral Acts Observations
March 1-14 6.2 + 3.1 21
15-30 Hpll se HLS) 37
April 1-14 42+2.5 59
15-30 Do 28 ile} 111
May _—‘i1-14 2.4 + 2.0 63
15-30 1.7 + 0.8 35
June 1-14 1.8) 2 1® 16
15-30 1.0 + 0.8 5
July 1-14 1.0 + 0.0 1
15-30 0.0 0
August 1-14 1.1+ 0.6 2
15-30 1.0 + 0.0 Il

540

Of 351 early courtship steps observed,
79.9% were initiated by the male when
the female was lying still; 16.1% when
the female was walking; 0.0% when she
was running; and 4.0% when the activity
of the female was nonclassifiable. Of the
courtships observed, 3.6% resulted when
a female walked in front of, or next to, the
winning male of a ritual combat only
moments before. Females often seemed
to watch these contests and showed signs
of increasing interest during the phase
when one male performed a number of
dominant acts against the other, usually
moving closer to the interacting males at
that time.

The length of time from courtship ini-
tiation to the first copulation attempts in
121 timed sequences varied from 3.2 to
44.4 sec (X = 25.5 + 22.2 sec). Complete
time in copula varied from 10.3 to 87.4
sec, and total courtship plus intromission
varied from 20.8 to 123.1 sec (X = 101.8
+ 60.3 sec).

The following brief description is typ-
ical of courtship behavior in the early
part of the season; later, as mentioned
above, it tends to include fewer acts. In
this case the courtship pattern passed
through nine steps (in parentheses).

A basking female is approached from behind by
an adult male. When approximately even with her
shoulder, the male (1) flicks his tongue onto her
occiput one time, her neck two times, and her
tympanum region three times, as she (2) turns her
head away. The male (3) places his chin on the
frontal area of the female, pushing downward as he
jerks his head from side to side (see Fig. 1). The
female (4) slips away and the male (5) follows, (6)
mounts her, (7) applies a strong brachial embrace,
and (8) engages in a vigorous tail lift, to which she
responds by (9) lifting her tail. Intromission
occurs.

The modal number of steps in success-
ful courtship is 10 (range = 3-16), with
the longest sequences tending to be more
unsuccessful—probably because it be-
comes apparent to the female that the
male is not an appropriate partner. Short
unsuccessful courtships are usually ter-
minated by a female that becomes agres-
sive and/or runs away. Many unsuccess-

Advances in Herpetology and Evolutionary Biology

ful courtship sequences contain a series
of repetitive acts occurring during the
middle or near the end of the sequence,
most often following step 5. On this basis
I assume the most important discrimina-
tory communication interchange in this
species takes place during steps 1-5. It is
for this reason that the following analysis
concentrates on the acts of the first five
steps. Because the male always initiates
courtship behavior, acts 1, 3, and 5 are
always those of the male, and acts 2 and 4
are those of the respondent female.

To determine the question of stationar-
ity during the first five behavioral steps,
the frequencies of each of the 20 acts
noted in courtship were considered first.
Table 2 shows that the probability of any
act occurring is probably dependent on
what had occurred previously, for within
both male and female sequences certain
patterns are more common in the earlier
steps and others in the later ones. As
examples of temporal dependence, male
chinning and tonguing are notably re-
duced from steps 1 to 5, whereas walking
away is increased. Sex is also obviously
significant, for mounting, chinning, and
scratching are only found in the male,
whereas others (usually agonistic), such
as indiscriminate biting, topping, head
away, and head twitching are restricted
to the female. While the remainder are
used by both interactants, some are more
common in males (brachial embrace, tail
lifting, or tonguing), and others in fe-
males (head up, do nothing, or walking
away). As pointed out under methods a-
bove, it is for this reason that the prob-
abilities of specific step acts occurring
cannot be stationary, and this important
fact is taken into consideration in the
analysis.

ADJACENT LOWER ORDER INTERACTIONS

The unconditional probabilities of the
acts characteristic of one sex or the other
are not constant with time during the
courtship bout (Table 2).

Steps I-2. In this initial phase of

VARANUS BENGALENSIS - Auffenberg 541
TABLE 2. THE ACTS OCCURRING IN STEPS 1-5 OF 217 COURTSHIP BOUTS IN VARANUS BENGALENSIS.
Acts*
fp Diem Ge HA HD) Hi Hi BU SM NSS —iG fe io) py as) WW
1 38 1 47 8 1 122:
2 1 1 1 Se 2 20 8 59 2 4 1 110
3 Is BPA MG 36 9 le pr OO Ee eA, mil: 1
4 2 Sy Il) OR onelo 42 1 2 6 100
3) 4 1 18 8&4 8 1 BY 4 56

Note: Even-numbered steps were performed by respondent females and odd-numbered steps by initiator-

males.

*Bi = Biting; Br = Brachial Embrace; C = Chin Rubbing; F = Following; G = Gular Extension; HA =
Head Away; HD = Head Down; Hi = Hiss; HT = Head Twitch; HU = Head Up; M = Mount; N = Do
Nothing; S = Scratching; TC = Tail Coil; TL = Tail Lift; To = Tonguing; Tp = Topping; Ts = Tail Slap; W =

Walks Away.

courtship the male initiates the interac-
tion in step 1 and the female responds in
step 2. In general, the acts of steps 1-2
show much less variability than steps
later in the courtship sequence. Of a total
of 21 expressed act pairs in steps 1-2,
only 12 are significantly more common
than expected; however, in steps 2-3
there are 19 of 30 act pairs that are signif-
icantly more frequent than expected, in
steps 3-4 there are 27 of 39 act pairs that
are significantly higher than expected,
and in steps 4-5 there are 19 of 27 act
pairs that are significantly higher than
expected. Thus, in the early phase of
courtship there is low variability in the
adjacent acts, which then increases dur-
ing the middle phase and becomes lower
again as the interaction becomes de-
graded prior to separation of the interact-
ing individuals.

About one half of the significantly high
step 1 male acts of the pair belong in the
investigatory class (tonguing), and the
remainder include only acts in the repro-
ductive class (chinning, mounting,
scratching). Male acts that are pair com-
ponents not significantly different than
expected include some of the reproduc-
tive class (mounting), some usually used
in an aggressive context (gular extension,
tail coil), or are investigatory (tonguing).

Respondent female acts (step 2) of act
pairs significantly higher than expected

are “do nothing” acts usually used in an
agresssive context (tail coil), or of impor-
tance socially (head away, head up, walk-
ing away), or (?) sexual acts (head twitch).
Female acts included in act pairs not
significantly different from expected are
usually either used in an agressive con-
text (gular extension, biting, topping,
brachial embrace), in a social one (head
down), in a reproductive context (head
twitch), or “do nothing.”

Of those act pairs that occur significant-
ly more frequently than expected, the
most common (highest significance) in-
clude reproductive acts (chinning, moun-
ting, scratching); the less common
(lowest significance) include an inves-
tigatory component (tonguing).

Within the act pairs of significantly
higher occurrence, there are only four
different male acts—almost always in the
reproductive class (chinning, mounting,
scratching). However, there are seven
different female respondent acts—almost
twice the level of act variability than in
the male acts in this initial stage of court-
ship. The usual female respondent act
involves one of the head movement in
the social class, thus apparently reflect-
ing an emphasis by the female on agon-
istic rather than reproductive motivation
at this early stage of courtship.

Steps 2-3. In this act pair the male’s
response (step 3) to step 2 of the female is

542

analyzed. First, the most obvious differ-
ence when compared with steps 1-2 is
that the variety of acts included in steps
2-3 is increased by approximately 50%
(30, as opposed to 21, acts significantly
higher than expected). Second, the levels
of significance of steps 2-3 are generally
higher than those of steps 1-2, suggesting
a closer association between the com-
ponent acts of the act pairs than in steps
1-2. This is largely due to the apparently
more specific response of the male to the
often agonistic acts of the female. Third,
the component female acts of the signifi-
cantly higher act pairs are largely social,
involving head movements (head down,
away, up, twitch), “do nothing,” walking
away, Or aggressive or investigatory acts.
Of the respondent male acts in the sig-
nificantly high act pairs, approximately
one half are investigatory (tonguing) and
the other half are either in the reproduc-
tive class (chinning, brachial embrace,
mounting, tail lift, scratching) or follow-
ing. Thus, though the courtship tactic of
the male has not changed appreciably
from his initial step 1 phase, this, his first
response to the female’s initial act, shows
greater association with it.

The highest departures from expected
values are found in act pairs, includ-
ing female head movements, suggesting
the highest association occurs between
male responsive acts (often reproductive
or investigatory) and female social in-
dicators. The least significant act pairs
(= lowest association between adjacent
acts) comprise those in which the male is
responding to the female walking away.
Mounting becomes common, but is
higher than expected only when the
female has “done nothing,” though al-
most all female “do nothing” acts are
followed by some form of male reproduc-
tive act.

Included within the significantly
associated acts in steps 2-3 are eight
different component acts and seven com-
ponent responses. Thus, male act varia-
tion becomes almost doubled from step 1.

There is no obvious pattern in the act

Advances in Herpetology and Evolutionary Biology

pairs exhibiting less than expected fre-
quencies, except that many contain
agonistic acts. This suggests that these
probably reduce the frequency of associ-
ative acts and that their expression tends
to reduce the possibility of further com-
municative sequencing toward success-
ful coition.

Steps 3-4. In this set of adjacent acts
the female (step 4) is responding to the
previous male act (step 3). It constitutes
her second act in the courtship sequence.
In steps 3-4, 39 act pairs occur in signifi-
cantly higher frequencies than expected,
so that variability in potentially com-
municative adjacent acts is again in-
creased by about 30% from steps 2-3. I
believe this part (steps 34 and nearby) to
be the most important of the courtship
sequence because (1) there is a gradual
reduction in the number of significantly

Figure 2. Common behavioral acts in the first six steps
of courtship, with acts significant at the 5% level shown
as thin arrows and those significant at the 2% level
shown as thick arrows. Steps 1, 3, and 5 are male acts;
steps 2, 4, and 6 are female acts. Heavily bordered
area includes those acts most commonly associated
with successful courtship; lighter bordered area repre-
sents less commonly successful sequences.

associated acts in later steps, and (2) re-
maining steps are often repetitive.

Within the significantly high associa-
tions, about one half of the male com-
ponent acts are reproductive, a third are
investigatory, and a few are of the “‘fol-
lowing’ type. Common associated fe-
male responding acts are walking away,
social acts involving head movements,
and aggressive types (biting, gular exten-
sion, topping, hissing). Female reproduc-
tive acts (head twitch) make their first
significant appearance at this stage, but
they are not associated with a broad
range of male acts.

Unlike the two earlier steps, there is no
clear grouping of act pairs into those that
are very closely associated and others
that are much less associated. All associa-
tions tend to be high, with the highest
between the tail lifting act of the male
and the corresponding respondent tail lift
of the female prior to coition.

The female response to tonguing (in-
vestigation by the male), though variable,
is usually a social act. She may also walk
away, but the tonguing-walk away act
pair has one of the lowest associations in
the entire step 34 phase. Of probably
high communicative importance is that
the female rarely “does nothing” after
being scratched, but walks away much
more commonly than expected.

There are 7 different component male
and 12 different component female acts
in the highly significant act pairs. Thus
variation in component acts of signifi-
cantly high act pairs increased from the
first to the second communicative act in
both males (steps 1 and 3) and females
(steps 2 and 4). The result is that the
range of probably important communica-
tive acts is greater at this stage than any
preceding or succeeding ones.

Steps 4—5. In this set of adjacent acts
the male acts for the third time (step 5)
and is responding to a female act for the
second time. Interestingly, there is a sig-
nificant reduction in the number of asso-
ciative acts from 39 in steps 2-3 to only
19. Thus communication potential, as

VARANUS BENGALENSIS - Auffenberg — 543

suggested by the breadth of associative
acts, is greatly reduced, to the level of
that of steps 2-3, illustrating a general
deterioration of the interaction at this
stage if mounting has not yet taken place.

The female acts of those significantly
associated remain about one half social
(head away, head up, walk away) or
neutral (do nothing) (total 11), with about
one quarter reproductive (tail lift, head
twitch) and one investigative (tonguing).
Female reproductive acts appear in sig-
nificantly high frequency for the first
time in the courtship sequence. Yet the
male acts have not changed greatly, for
about one half of the dominant elements
among the significantly associated acts
continue to be reproductive (mounting,
scratch, chinning, tail lift), while inves-
tigatory acts (tonguing) are still common.
However, a significantly larger propor-
tion of socially “neutral” acts (do noth-
ing, walking) is apparent, often asso-
ciated with preceding similar or socially
subordinant acts by the female. These
high associations may reflect increasing
loss of interest on the part of the male.

The highest level of association is
found in the precopulatory tail lift act by
both individuals (as in steps 2-3). Female
aggressive acts remain much higher than
expected, often followed by rather
neutral acts by the male (walk away,
tonguing, do nothing). The weakest asso-
ciations are again seen in act pairs in
which the female walked away, suggest-
ing the male may become confused or
disinterested. However, in general, the
level of association between male and
female acts remains high. During this
phase there are 10 different female acts
included in the significant associations
and 8 in the following male acts.

HIGHER ORDER INTERACTIONS

I have shown in the preceding section
that certain initiation and respondent acts
are often associated in higher frequen-
cies than expected on the basis of chance
alone, and that the probabilities of specif-

544

ic associations change within a temporal
and sexual framework, dependent to a
large measure on the immediately pre-
ceding act. In this section I offer the re-
sults of analyses concerned with the
probability of a certain act being ex-
pressed following a specific prior act of
the same individual, or following a spe-
cific act pair of preceding acts by the
same individual and its interacting part-
ner. Analyses of higher order interactions
consider three rather than two steps at a
time.

Binomial tests for each of the 3-step
sequences show that many of them differ
significantly at the = 0.05% level (Table
3). Of the 81 possible three combinations
in steps 1-3, only 20 occur significantly
more frequently (P = < 0.05, binomial
test) than expected. Of the remainder,
there is a much lower association than
expected in the 3-step acts To,N.Ms;,
To,W,C3.

The fact that a complete list of ob-
served and expected frequencies is avail-
able for all the steps in the courtship
sequence permits the use of a chi-
squared test to estimate the significance
of the deviation from expected for each
3-step act. Pearson’s chi-squared approx-
imation was used to perform the chi-
squared test (degress of freedom r—] +
el):

The highly associated acts of steps 1-3
include female step 2 acts of walk away
(W,), do nothing (N,), head twitch (HT,),
tonguing (To,), and head away (HA,). Of
these, only the first three occurred with
sufficiently varied male acts 1 and 3 to
make chi-squared tests useful estima-
tions of association. W, occurred 110
times, and when compared with the
preceding (No. 1) and succeeding (No. 3)
male acts, the analysis provided a x2 of
0.99 (df = 10). Thus the male act 3 occurs
as part of the W, n-step sequence (C3, S3,
F, Br,, To;, and M;) and is not depend-
ent on W,, or on W, preceded by male
acts C,, M,, or To,. The same results
were obtained for C, and To, following
C, Ht, (df = 1, x2 = 0.34), and for C;, M;,

Advances in Herpetology and Evolutionary Biology

TL,, and To, following MIN,, S,N,, and
To,N, (df = 6, x2 = 5.45). The same con-
clusions can be drawn for the transitions
in steps 24: HA,, W;, To;, and N; are
not dependent on To, or HT,To,,
HU,To,, N,To,, N, Tos, or To, To, (df =
9, x2 = 1.71); or HT;, W;, or N; after C,,
Ht, C,, N,G,, or W,C, df = 4, y2=
1.54); or steps 4-5 here.

Thus, the courtship interaction data do
not require more than 2-step transition
probabilities to describe them; i.e., no
more than secondary nonstationary
Markov processes need to be postulated.
The behavioral acts during courtship are
evidently largely dependent on the im-
mediately preceding acts. This conclu-
sion is additionally strengthened by a
series of G-tests (likelihood ratio test, see
Oden, 1977), which are more conserva-
tive than x2 tests based on Pearson’s
method. The results for all significantly
associated acts in all steps are shown in
Figure 2.

DISCUSSION

Figures 3 and 4 diagram the conclu-
sions of the analyses conducted on the
courtship of Varanus bengalensis. Figure
3 diagrams the highly significant associa-
tions among the adjacent acts comprising
the 10 most highly significant act pairs
listed in Table 3. The number 10 is arbi-
trary, but includes what appear to be the
most significant adjacent acts in each step
to intromission. In that part of the dia-
gram covering steps 1-2 it is obvious that
the first act of the male in the courtship
sequence is usually a_ reproductive
type—usually scratching, mounting, or
chinning—and that these are significant-
ly associated with “do nothing” or walk
away by the female. The male acts often
lead to a rejection act (head twitch) by the
female or other head movements, usually
denoting social subordination (head
away, head up). The latter are probably
most important in the successful comple-
tion of the courtship sequence, for in the

Nothing

Walk Away | Reproductive Re Social

WwW Ha,Hu

|

|

Warning
STEPS) 152 Tc
Walk Away
Ww
Social
wr Tp,Hu
Reproductive

S,Br,TI,M,

<

Warning
G,Hi

SIER'S #93) (4

Nothing ~<a
N

VARANUS BENGALENSIS - Auffenberg 545

Nothing
N

Reproductiv

Ht

A

Social Investigative Investigative|

Ha,Hu To

Warning
SUERS 2=3 Tc

Warning ae Reproductive
Tc,Hi,G TI1,M

Investigative Investigative Reproductive
To Ht
Weaponry a
Bi
Social
STEPS 4 5 Tp,Hu

Figure 3. Sequence diagram arbitrarily based on the ten highest 2-tuple act behavioral patterns. Circles repre-
sent male acts, squares female acts. Arrow widths represent proportion (in %) of associated act pairs in the steps
shown.

546 Advances in Herpetology and Evolutionary Biology

TABLE 3. UNUSUALLY FREQUENT N-TUPLES FOR 217 LENGTH 5 COURTSHIPS OF VARANUS BENGALENSIS.
Columns show starting step number, step acts (n-tuple), expected and observed frequencies, z-values, and
one-tailed significance (binomial distribution). Only PS 0.05 deviations from expected are shown. See p.
541 for act abbreviations Step acts are arranged from most to least significant within each step number
category.

Step Frequency
No. Act Expected Observed z-value Significance
1 S-N 0.30 8 15.16 0.0066
M-HA 0.65 3 14.16 0.0071
M-W 6.28 29 13.77 0.0073
C-HT 2.98 17 12.53 0.0080
C-W 3.68 21 12.46 0.0080
M-N 2.38 iil TET 0.0083
To-TC 1.12 2 9.67 0.0103
To-HT 1.12 2 9.67 0.0103
To-To 2.25 5 9.05 0.0111
To-HA Soil 6 EO 0.0141
To-HU 4.50 8 6.89 0.0145
To-N 20.25 36 6.54 0.0153
To-W 33.73 60 6.25 0.0150
2 HD-To 0.02 2) 20.63 0.0049
TC-To 0.02 2 20.63 0.0049
HA-C 0.07 2; 20.10 0.0050
HU-C 0.07 2 20.10 0.0050
To-To 0.07 4 16.72 0.0060
HT-Br 0.28 3 16.39 0.0061
HU-To 0.22 6 15.54 0.0064
HT-C 0.65 7 14.43 0.0069
HT-To 0.74 8 13.64 0.0073
N-TL 1.60 6 iL 0.0085
N-To 3.80 14 11.14 0.0090
N-M 4.62 17 11.05 0.0090
N-C Spal 21 10.99 0.0091
HA-To 0.18 5 10.53 0.0095
W-S 4.56 9 Coli 0.0130
W-Br 4.56 9 Corll 0.0130
W-F 3.11 16 7.50 0.0133
W-M 8.61 17 7.49 0.0134
W-To 19.26 38 7.36 0.0136
3 TL-TL 0.07 2 20.10 0.0050
F-N 0.15 2 19.27 0.0052
M-G 0.33 1 17.40 0.0058
M-Hi 0.33 1 17.40 0.0058
Br-N 0.21 3 16.57 0.0060
S-W 0.17 4 16.30 0.0061
TL-N 0.13 4 16.04 0.0062
Br-W 0.27 4 15.87 0.0063
C-Tp 0.48 2 15.83 0.0063
C-HU 0.48 2 15.83 0.0063
Br-HT 0.55 8 14.67 0.0068
F-W 0.88 12 14.26 0.0070
M-N 1.00 0 13.44 0.0074
To-HT 0.74 2) 13.13 0.0076
To-Tp 0.74 2 13.13 0.0076
To-HU 0.74 2 13.13 0.0076
To-Bi 0.74 D2 IS} 1S) 0.0076
To-Hi 0.74 2; 13.13 0.0076
M-W 4.15 25 12.56 0.0080
C-HT 3.35 14 11.63 0.0086

ORI
z
a

QO
<
ees

-W-To

N-To-HA

Frequency

Expected

VARANUS BENGALENSIS - Auffenberg 547

Observed
24

_
UPR TROWWWRWWNN W

Oe w ee il ond
DMPROWOOMDWOHhANWNOWF COS

— —

CORWNAKRORNWAHNKH DOD AO OO

z-value

11.44
11.34
10.40
9.57
9.57
6.27
5.36
20.63
20.63
17.71
17.65

TABLE 3. CONTINUED.

Significance

0.0087
0.0088
0.0096
0.0104
0.0104
0.0160
0.0187
0.0049
0.0049
0.0056
0.0057
0.0060
0.0061
0.0063
0.0063
0.0064
0.0064
0.0065
0.0066
0.0066
0.0067
0.0069
0.0070
0.0073
0.0078
0.0083
0.0084
0.0102
0.0120
0.0124
0.0126
0.0069
0.0101
0.0107
0.0121
0.0131
0.0107
0.0145
0.0145
0.0546
0.0149
0.0164
0.0169
0.0192
0.0223
0.0261
0.0278
0.0319
0.0400
0.0426
0.0483
0.0063
0.0087
0.0087
0.0098
0.0106

TABLE 3. CONTINUED.

diagram representing steps 2-3 these
same acts are highly significantly asso-
ciated with further reproductive acts by
the male in step 3. Note that “do noth-
ing’ on the part of the female is also sig-
nificantly associated with further male
reproductive interest. Investigatory,
social, warning, and sexual rejection fre-
quently lead to further investigation by
the male. However, rejection by the
female is often ignored, with the male
pressing the courtship further. During
steps 34 success is often achieved via
mutual tail lifts in both interactants.
However, tail lifts by the male more fre-
quently lead to “do nothing” or social
acts, including topping which is clearly a
dominant act. It also frequently elicits
warning behavior and rejection in the
form of the female walking away. In steps
4—5 we see that female warning acts are
common, though these seem largely ig-
nored by the male, for tail lifting and
mounting are closely associated with

548 Advances in Herpetology and Evolutionary Biology
Step Frequency
No. Act Expected Observed z-value Significance
W-C-HT 4.30 3 7.83 0.0128
W-To-HA 7Al 13 7.78 0.0129
W-M-N 2.68 5 7.57 0.0132
N-M-W 7.19 12 4.87 0.0205
N-C-HT 4.88 7 4.77 0.0210
HT-C-W 2.69 2 3.33 0.0300
W-To-HA 6.75 6 2.02 0.0495
W-To-W 10.50 12 1.92 0.0521
W-M-W 8.91 8 Neat 0.0565
3 Br-HT-M 1.00 +) 13.33 0.0075
G-W-N 2.39 17 13.05 0.0077
To-W-N 1.77 11 13.00 0.0077
C-HT-M 2.04 6 10.70 0.0093
F-W-W 6.23 9 9.86 0.0101
To-W-N 2.79 6 8.68 0.0115
M-W-W 4.53 10 8.61 0.0116
To-W-To 2.05 4 8.13 0.0123
To-N-N 2.57 5 8.10 0.0123
To-HA-To 4.89 10 7.98 0.0125
C-W-W 3.41 6 7.00 0.0143
M-W-To 4.51 9 2.57 0.0389
Br-HT-S 0.40 2 2.39 0.0418

them. On the other hand, female weapon-
ry use leads to “do nothing” by the male.
Sexual rejection by the female also leads
to a breakdown of the sequence at this
stage (though not in earlier steps). The
male reacts similarly to female social
acts—exactly the opposite of his reaction
in steps 2-3.

Figure 4 is a diagrammatic conclusion
of the more conservative G-tests. Again,
the significant male step 1 acts are chin-
ning, mounting, or tonguing. These are
variously, but significantly, associated
with the respondent female acts of head
twitch, walking away, or doing nothing.
If the female tongues the male at this
stage, or turns her head away, there is a
very strong association with a responding
tonguing act by the male in step 3. There
is also a significant association between
his tonguing act and her preceding head
twitch, and her possibly doing nothing.
His chinning in step 3 is also closely as-
sociated with her doing nothing in step 2.

Clin CIHT2

Re 20 oe
Poe) Fs || M3l_|¢

S| N2

MI We2 MI W2

Amacs Hl2TO3s

Cl W2

N2 C3

VARANUS BENGALENSIS - Auffenberg 549

MI N2

N2
toal_|S!] toal_|¢|

TOIHA2 = TOIN2

Cl We

STEPS 1-3

TO2 TO3.) =We2 M3

STEPS 2-4

Figure 4. Behavioral act sequence summary based on G-test analysis.

Tail lifts and mounting by the male at this
stage are also closely associated with her
doing nothing. Tonguing by the female
in step 4 is associated only with similar
behavior on the part of the male. Head
twitching by the female is associated
with his chinning. Her tail lift is closely
associated with his previous similar
activity. There is a very strong associa-
tion between her walking away in this
step and the male’s previous mounting
attempt. According to G-test analysis,
successful courtship, in that it leads to
coition, is most probable in the sequence
M, N, TI, Tl,, or the same sequence be-

ginning one step later. Thus, the social
head movements, which seemed so im-
portant on the basis of highly significant-
ly associations in Pearson’s x2 estimates,
are here indicated to be less significant in
the successful completion of courtship,
though undoubtedly of considerable
importance in establishing rapport be-
tween the interacting individuals.
Because of the data requirements, the
G-tests included more courtship se-
quences during the late part of the court-
ship season than the data shown in
Figure 3. This may help explain the more
simplified act system suggested, as well

550

as the short, direct sequence leading to
successful coition. These short, usually
more successful, sequences occur later in
the same season, after attainment of
familiarity between individuals seems to
have been accomplished during earlier
courtship sequences of the type illus-
trated in Figure 3. Specific acts frequent-
ly deleted in these later phases are those
that fall into the agonistic and social
classes.

Most female acts in unsuccessful se-
quences are agonistic and may include
actual weaponry use (biting, tail slap).
Continued association with particularly
large, powerful males able to restrain the
aggressive females brings about a reduc-
tion in female agonistic behavior. As I
pointed out in my earlier study of
Varanus komodoensis (Auffenberg,
198la), the large size of this species,
plus the aggressive nature of the females
and their capability of inflicting serious
wounds and of even killing smaller indi-
viduals, have undoubtedly contributed to
the development of a reproductive be-
havioral strategy that stresses courtship
throughout the year among the mature
male and female pairs who share the
greater part of the same foraging area
(Auffenberg, 1978b). As a result, familiar-
ity is assured among individuals poten-
tially dangerous to one another, thereby
reducing stress during the actual breed-
ing period itself. This system apparently
tends to develop a long-term pair bond,
for individuals of Varanus komodoensis
seem to inhabit the same area for many
years (Auffenberg, 1978b, 198la). The
probability for courtship success is thus
enhanced by what amounts to an almost
year-round courtship with otherwise very
aggressive and powerful females. The
same strategy in its basic format is seen in
the present study of Varanus bengalen-
sis. Though it is not known what role, if
any, pair-bonding plays in this species in
the wild, it is obvious that a long associa-
tion (several months in captivity) effected
through repeated courtship by the male
is usually necessary for successful mat-
ing.

Advances in Herpetology and Evolutionary Biology

The entire social system of Varanus
bengalensis seems to select the most
powerful males and females for ultimate
breeding. I base this on the fact that
visual displays intimating social rank are
of much less importance in varanids than
in agamids or iguanids. On the other
hand, warning and actual weaponry use
are common components of even adol-
escent interactions (Auffenberg and
Ganci, in press). Secondly, unlike lizards
in other families, female varanids are
much more apt to engage in agonistic
displays and use their teeth and tail more
often than do males of the same species
(Auffenberg, 1981b). This means that
male Varanus bengalensis interested in
copulation must be able to so completely
restrain the females that they remain un-
injured after breeding. The females show
an inclination to breed with the males
that are capable of doing this. Third, male
varanids engage in a test of strength, the
evident purpose of which is to determine
the one that has the greater ability to
physically manipulate but not harm the
other. This is exactly what is required in
courtship within the same species (harm-
less manipulation), for at this time it is
the females that must be restrained. Fin-
ally, the females themselves often select
by overt behavior those males most able
to restrain and otherwise manipulate the
other; i.e., those males most probably
capable of restraining the females them-
selves. The entire system is based on
proven performance, not on bluff, and
seems analagous to that in many un-
gulates (see Wilson, 1975). I believe it is
the basis for much of what we know
occurs in male-male combat and male-
female courtship of snakes (for review
see Carpenter and Ferguson, 1978).

The emphasis of the female Varanus
bengalensis on agonistic behavior during
the early part of the courtship season and
much of each courtship bout may help
explain the fact that the male acts are
associated only with the preceding
female acts and in no way dependent on
his previous acts, either separately or in
combination with those of the female. If

my interpretation is correct, the female is
less predictable in her response and cer-
tainly potentially dangerous in view of
her tendency to become aggressive. The
latter is clear, while the former is sug-
gested by the greater variability of her
responses and the lack of significant
association of most of her acts with any
specific male initiating act. In view of
this, the male probably determines most
of what he does during courtship on the
basis of the female’s immediately
previous act alone. His courtship be-
havior is not rigidly programmed, but
remains flexible to contend with more
varied reciprocal acts of the female.

ACKNOWLEDGMENTS

This work was made possible through
the support of the Florida State Museum,
which purchased most of the equipment
used; the New York Zoological Society,
which provided the greenhouse facilities
and funds for the field work in 1974 when
some of the individuals were captured;
and the United States Customs Office,
which permitted use of individuals con-
fiscated as the result of illegal shipments.

LITERATURE CITED

AUFFENBERG, W. 1964. Notes on the courtship of
Geochelone_ travancorica (Boulenger). J.
Bombay Nat. Hist. Soc., 61: 247-253.

—__. 1978a. Courtship and breeding behavior in

Geochelone radiata (Testudines: Testudini-

dae). Herpetologica, 34: 277-287.

1978b. Social and feeding behavior in

Varanus komodoensis. pp. 301-331. In N.

Greenberg and P. MacLear (eds.), Behavior

and Neurology of Lizards, Natl. Inst. Mental

Health.

—__.. 1979. Intersexual differences in behavior of
captive Varanus bengalensis (Reptilia, Lacer-
tilia, Varanidae). J. Herpetol., 13: 313-315.

VARANUS BENGALENSIS - Auffenberg 551

—___.. 198la. Behavioral Ecology of the Komodo
Dragon (Varanus komodoensis). Univ. Presses
of Florida, Gainesville, pp. 1-406.

1981b. Combat behaviour in Varanus
bengalensis (Sauria: Varanidae). J. Bombay

Nat. Hist. Soc., 78: 54-72.

AUFFENBERG, W., AND S. Gancl. In press. Be-
havioral cues and displays in adolescent
Varanus bengalensis. J. Herpetol.

CARPENTER, C. C. 1977. Communication and dis-
plays of snakes. Amer. Zool., 17: 217-223.

CARPENTER, C. C., AND G. FERGUSON. 1978. Varia-
tion and evolution of stereotyped behavior in
reptiles, pp. 138-308. In C. Gans and D. W.
Tinkle (eds.), Biology of the Reptilia. Vol. 7,
Ecology and Behavior. London, Academic
Press.

Cooper, W. E. 1977. Information analysis of agonis-
tic behavioral sequences in male iguanid
lizards, Anolis carolinensis. Copeia, 1977:
721-735.

GILLINGHAM, J. C. 1979. Reproductive behavior of
the rat snakes of eastern North America, Genus
Elaphe. Copeia, 1979: 319-331.

GREENBERG, N. 1977. An ethogram of the blue spiny

lizard, Sceloporus cyanogenys (Reptilia,
Lacertilia, Iguanidae). J. Herpetol., 11(2): 177-
195.

HAZLETT, B. A., AND W. H. BOssERT. 1965. A statis-
tical analysis of the aggressive communication
systems of some hermit crabs. Anim. Behav.,
13: 357-373.

JENSSEN, T. A. 1977. Display diversity and anoline
lizards and problems of interpretation, pp.
269-285. In N. Greenberg and P. MacLean
(eds.), Behavior and Neurology of Lizards,
Natl. Inst. Mental Health.

MERTENS, R. 1942. Die Familie der Warane
(Varanidae). Abh. Senckenb. Naturforsch.
Ges., Pts. 1-3: 1-391.

ODEN, N. 1977. Partitioning dependence in nonsta-
tionary behavioral sequences, pp. 203-220. In
B. A. Hazlett (ed.), Quantitative Methods in
the Study of Animal Behavior. New York,
Academic Press.

VoGEL, P. 1979. Zur biologie der Bindeswarens
(Varanus_ salvator) im Westjavanischen
Naturschutsgebiet Ujung Kulon. Ph.D. disser-
tation, Univ. Basel, pp. 1-139.

WEAVER, W. G., JR. 1970. Courtship and combat be-
havior in Gopherus berlandieri. Bull. Florida
State Mus., Biol. Sci., 15(1): 1-43.

WILSON, E. O. 1975. Sociobiology: The new synthe-
sis. Cambridge, Harvard Univ. Press, pp.
1-607.

Display Behavior of Two Haitian Lizards,
Anolis cybotes and Anolis distichus

THOMAS A. JENSSEN!

ABSTRACT. The display repertoires of two common
and frequently sympatric congeners, Anolis cybotes
and A. distichus, were quantitatively analyzed. The
displays for each species were moderately stereo-
typed, being consistent within a population as well
as shared by conspecifics between populations.
Comparisons between the two species showed
completely different signature display patterns, and
a striking difference in the size of their respective
repertoires. Anolis cybotes has only a single
species-unique signature display. In contrast, A.
distichus has six display types, some with unusual
characteristics, and the largest repertoire yet de-
scribed for an anoline species. An evolutionary
scheme is provided, giving the hypothetical steps
required to derive the display-rich distichus reper-
toire, and some socio-ecological factors which may
be selective for this enlarged repertoire.

INTRODUCTION

Hispaniola supports a complex Anolis
fauna of at least 34 species (Williams,
1976), with 20 of those species having
been described since Cochran’s (1941)
checklist appeared. Many of the newly
recognized taxa are cryptic species, being
distinguished primarily by their electro-
phoretic profiles, collection localities,
and a few subtle external characteristics,
while others have restricted distributions
and were discovered through new and
more numerous sampling sites.

With the proliferation of anoline forms,
the cybotoid and distichoid species have
had their share of revision. Anolis

1Biology Department, Virginia Polytechnic In-
stitute and State University, Blacksburg, Virginia
24060, U.S.A.

cybotes has recently yielded a cryptic
species, Anolis marcanoi (Williams,
1975), which has been established to a
large measure by electrophoretic criteria
(Webster, 1975). Another cybotoid,
Anolis whitemani, was separated by
Williams (1963) from its sibling species
A. cybotes with whom it shares almost
identical albumin immunological char-
acteristics (Wyles and Gorman, 1980);
now based on more collection data, A.
whitemani has been subdivided into
three subspecies (Schwartz, 1980). The
Hispaniolan Anolis distichus represents
as many as 12 subspecies (Schwartz,
1968), with the intraspecific nature of
some of the forms still in question
(Williams, 1977a). Anolis brevirostris, a
distichoid species, has been recently
divided into four cryptic species (Arnold,
1980; Williams, 1976), chiefly through
the electrophoretic studies of Webster
(1977) and Webster and Bums (1973).
The continued discoveries of sibling
species, cryptic species, and extensive
variation within species have elicited an
exciting and in some ways perplexing
perception. Anolis systematics of the
Hispaniolan fauna seems to become less
stable with increased information, and
the evolutionary events which shaped
the fauna become even more enigmatic
with the improved resolution of the
taxonomic picture. To achieve any
measure of sense to the emerging scene,
more morphological, karyotypic, electro-
phoretic, ecological, and behavioral data
will be needed. As Williams (1977b: 135)

points out, even the most simple ques-
tion of how many anoline species there
are on the island defies an answer; too
frequently one is not sure where one
species ends and another begins.

In this paper, I present the display
repertoires of important representatives
of the cybotoid and distichoid groups, A.
cybotes and A. distichus dominicensis.
The display repertoires of these two
sympatric species are quite different in
size, with one being the smallest possible
and the other the largest yet described for
Anolis. The signature displays of A.
cybotes and A. distichus are quite differ-
ent, while the displays of both species
show some remarkable and unusual qual-
ities. These display descriptions serve as
a baseline study for future comparative
behavioral investigations of Hispaniolan
anoles. The presented display data
should aid in taxonomic decisions as well
as possibly assisting evolutionary inter-
pretations of the genus.

MATERIALS AND METHODS

CYBOTES SAMPLE

Sixty-two displays were filmed in the
field. Five of these displays were from a
male in Petion-ville, and 57 were by
seven males from a population 2.5 km
south of Montrouis on Route 100 (Fig. 1).
Films were taken of males during undis-
turbed social activities and during elic-
ited aggressive encounters. The latter
were created by: 1) releasing an unre-
strained intruder in a resident’s territory
and 2) placing a male confined within a
transparent cubicle into an occupied ter-
ritory.

DISTICHUS SAMPLE

One hundred ninety-three displays
were filmed. Of these, 29 were filmed in
the field; 23 were from three males in
Petion-ville and six from males in
Montrouis. These displays were given

LIZARD DISPLAYS - Jenssen 593

within naturally occurring social con-
texts.

Five males from Montrouis had 164
displays recorded in my lab. The lizards
were housed separately in large enclo-
sures (1.3 X 0.7 X 0.3 m) containing sim-
ulated natural habitat and one to two
females. The enclosures were lit with
fluorescent lights (12L:12D photoperiod)
and maintained at 23 to 25 C. At the time
of filming, a subject was transferred to a
filming enclosure identical to the holding
enclosures. After a day or more of resi-
dency, the subject’s activities were
filmed from a blind with the added illum-
ination from two quartz iodine flood-
lights. Each subject initially displayed
after perch changes (usually non-
directed) and when interacting with resi-
dent females. Later they displayed dur-
ing encounters with a released male
intruder.

TECHNIQUES

All films were taken with a Nizo S80
super 8 movie camera, using a film ad-
vance setting of 18 fps. Only those dis-
plays known to be complete and of good
resolution were used in the analysis. The
resulting sample was analyzed frame-by-
frame with a Kodak MFS-8 Ektagraphic
projector. Details of the analysis methods
are given by Jenssen and Hover (1976).

To analyze the complex bob patterns, I
arbitrarily divided each kind of stereo-
typed head bob pattern (display type)
into units. Odd-numbered units were
chosen to give the duration of a bob and
even-numbered units delineated the
duration of an inter-bob pause. For some
odd-numbered units, a single bob was
delineated by a unit. For the other odd-
numbered units, a series of uninterrupt-
ed bobs comprised the unit; for such odd-
numbered units, a bob count was also
made. In addition, counts were made of
the number of dewlap_ extension-
retractions (pulses) which followed most
of the displays.

Quantification of the data was aided by

554

MONTROUIS

&

&

Archaie

Duvalierville
@

Advances in Herpetology and Evolutionary Biology

N
l

Port-au-Prince

¥ PETION-VILLE

Figure 1. Collecting areas in Haiti, indicated by asterisks.

computer analysis. All data of a display
were punched on an IBM card; these
included: 1) displayer’s location; 2) dis-
player's species; 3) displayer’s individual
identification if known; 4) display type;
5) social context if known; 6) frames of
film for each unit; 7) number of bobs for
units depicting a bob series; 8) number of
dewlap pulses if present; and 9) whether
filmed in the field or lab. An IBM 370
computer then: 1) converted fps to real
time for each unit; 2) computed descrip-
tive statistics (mean, standard deviation
of the mean, standard error of the mean,
and minimum and maximum values) for
each display variable; 3) calculated the
coefficient of variation (standard devia-
tion X 100/mean) for each variable; and
4) performed tests of significance.

DAP GRAPHS

The visual summaries resulting from
the display analysis appear as display-
action-pattern graphs (DAP _ graphs).
These graphs show head amplitude
movement along the y-axis and the pas-
sage of time along the x-axis. Thus a
quick up-and-down head bob will look
like a “spike” when graphed, and a bob
which is held before the head is lowered
will have a “plateaued” appearance. If
the dewlap is extended during the dis-
play, it appears on the DAP graph under
the head bobs; the dewlap extension is
shown as a downward vector along the
y-axis, dewlap retraction as an upward
vector along the y-axis, and passage of
time follows the x-axis.

STATISTICAL TABLES

Tables enumerating the descriptive
statistics and statistical test results are
available from the author and the Depart-
ment of Herpetology, Museum of Com-
parative Zoology upon request.

RESULTS

SIGNATURE DISPLAY PATTERN-CYBOTES

Anolis cybotes has but a single stereo-
typed head bob display in its repertoire.
This display pattem is used during non-
directed displaying while a male patrols
his territory; during male-female court-
ship, copulatory, and postcopulatory dis-
playing; and during male-male agonistic
disputes. The head bob pattern is pro-
duced by action of the forelimb and neck
muscles (Fig. 2). The bobs of the signa-
ture display are rapidly sequenced and of
relatively low amplitude. Without bene-
fit of frame-by-frame analysis, the display
does not appear to have much pattern to
it. However, once the pattern is known, it
is readily recognized and quite predict-
able.

The cybotes signature display is com-
posed of two sequences of bobbing (the A
and B Elements), with the second se-
quence repeated during some perform-
ances (Fig. 3). The A Element begins
with three large bobs (Units 1, 3, and 5).
Occasionally a fourth was included (Fig.
4B). This variation was performed by five
of the eight subjects and involved 23% of
the displays. The A Element concludes
with a variable number (3-9, x = 7) of
small rapid bobs (Unit 7). Ifa male began
a signature display, it always stopped at
the end of an element. Usually the dis-
play concluded after a B Element, but 12
displays (18%) finished after the initial A
Element (Fig. 5).

The B Element is similar to the A, but
tends to have one less large bob and
fewer smaller rapid bobs (Fig. 3). The B
Element is initiated with two large bobs
(Units 9 and 11); occasionally two lizards

LIZARD DISPLAYS - Jenssen D200

of my sample performed three (16% dis-
plays with a B Element) (Fig. 4C). The
concluding series of small bobs (Unit 13)
were variable in number (2-7) and aver-
aged four bobs.

A large number of the displays (48%)
ended with the first B Element. Some
displays, however, concluded after the B
Element was repeated two, three, or four
times (Fig. 5). Although based on anecdo-
tal observation, it seems that the more
aroused a male is, the more likely he will
repeat the B Element within a given dis-
play. There was no statistical difference
in unit durations or number of bobs when
comparing between B,-B, Elements.

The Petion-ville signature displays
possessed the same pattern as those by
males outside Montrouis in the north.
The Wilcoxon Two-Sample test (Sokal
and Rohlf, 1969) showed no significant
difference between any of the compar-
able display variables from the two local-
ities. Of interest is the display pattern of
two A. cybotes males (Fig. 4) filmed dur-
ing a 1970 trip to Etang Saumatre in the
Cul-de-Sac area east of Port-au-Prince
(Fig. 1). Their pattern also falls within
the signature display characteristics of
the males from the Montrouis and Petion-
ville regions. This suggests that the A.
cybotes in at least central Haiti have a
fairly stable display pattern across geo-
graphic localities.

i

distichus

cybotes

Figure 2. Minimum (stippled) and maximum (solid)
amplitude during head bob displaying by Anolis disti-
chus and Anolis cybotes.

556 Advances in Herpetology and Evolutionary Biology

A element B, element Bo element Bs element

SECONDS

Figure 3. Summary DAP graph of Anolis cybotes signature display. Upper edge of upper solid block shows head
amplitude change (y-axis) through time (x-axis), upper edge of stippled block shows head amplitude changes of
optional introductory movements, and lower edge of lower solid block shows dewlap extension through time.

SECONDS

Figure 4. Examples of different Anolis cybotes head bob behavior. A. Head dipping sequence with intervening
quiescent periods. B. Stippled area indicating where a major bob of the A Element was performed as two. C.
Stippled area indicating where the usual two major bobs of the B Element are performed as three. D. DAP graph

of the signature display from near Etang Saumatre.

Z BO
S
Qa.
Q
20
v=
eo)
i=
2
10
E
=)
Zz

hes. Bs Bs Ba

Last Completed Element

Figure 5. Frequency of filmed Anolis cybotes signa-
ture displays which ended after a particular element.

DISPLAY STEREOTYPY

The first step in a quantified analysis of
ritualized behavior is to ask to what
degree the behavior is stereotyped. To
answer this, a coefficient of variation
(CV) is computed for each of the be-
havior’s variables. The CV is an index of
variability, where CV values increase as
variability increases. Therefore, it can
also be used to gauge relative stereotypy,
where the smaller the CV, the greater the
stereotypy. Most quantified ritualized
behaviors which Barlow (1977: 103)
found reported in the literature have CV
values within a range of 15 to 35%. From
this Barlow suggested that behavioral
components having CV values less than
15% should be considered highly stereo-
typed. Test results in this study show that
cybotes unit durations and number of
bobs within Units 7 and 13 fall primarily

LIZARD DISPLAYS - Jenssen DoT

within the 15 to 35% range.! Therefore,
the signature display is classified as
stereotyped, but certainly not devoid of
variability.

The next question is how this variabil-
ity is partitioned within the sample. To
do this a one-way Analysis of Variance
(ANOVA) (Sokal and Rolf, 1969) was
used to compute total variance and then
to partition the proportion of variance
attributed to comparisons between
lizards (among component) and _ that
found between displays of the same
lizard (within component). Test results
show that most of the variance appears in
the within component. This means the
lizards of the sample tended to share a
common range of variability for each unit
duration and number of bobs. There was
little indication of highly stereotyped
individual-unique display characteris-
tics.

The ANOVA (F value) did find signifi-
cant differences for several variables
when comparing between lizards. How-
ever, when Duncan’s Multiple Range
test (Steele and Torrie, 1960) was applied
to the display variables, the results
showed broadly overlapping variances
between lizards, with no lizard or group
of lizards being significantly different
from the rest of the lizards for any given
variable. Again, there was no indication
of individual-unique display characteris-
tics.

OTHER BOB MOVEMENTS

When a male was strongly aroused, as
during territorial defense, he sometimes
prefaced his signature display with an
introductory “head roll.” This head
movement was not particularly stereo-
typed and consisted of a moderate to slow
raising and lowering action of the head
(Fig. 3, stippled area adjacent to A Ele-

1 Tables of descriptive statistics and test results of
the display analyses are available upon request from
the author and the Department of Herpetology,
Museum of Comparative Zoology.

Ol
Ol

ment). This is a common introductory
movement to stereotyped displays of
many anoles (Jenssen, 1979).

Another bob movement was also seen
to occasionally precede the signature dis-
play. This was a quick inverted bob, best
described as a “head dip” (Fig. 3 and 4A).
Head dipping was also performed inde-
pendently of the signature display. Soli-
tary males were frequently seen head
dipping as they took a new perch or when
they appeared to be surveying their sur-
roundings. The dips came in long series,
with each separated by an extremely
variable time interval (2-30 sec). If this
behavior has a social significance, it very
likely is not completely emancipated
from its original nonritualized source be-
havior. I am guessing the noncommuni-
catory function of the head dip behavior
is to enhance visual discrimination of dis-
tant objects. Head dipping is not per-
formed by the other Anolis species I have
studied, which leads me to suspect the
behavior is unique to A. cybotes and pos-
sibly its close relatives.

DISPLAY REPERTOIRE-DISTICHUS

In contrast to the single display reper-
toire of A. cybotes, A. distichus exhibits a
set of six display types (a, A, B, C, D, and
E). This is the largest repertoire yet re-
ported for an anoline species. Anolis limi-
frons had the largest previously known
repertoire which contained five different
display patterns (Hover and Jenssen,
1976). Below is the display-by-display
description of the distichus repertoire.

TYPE a DISPLAY

Type a display is not represented by a
single display pattern, but by three (Fig.
6). It is likely that the list could be
expanded with the observation of more
lizards. The criteria used to lump the
patterns under the Type a category were:
1) extremely low bob amplitude (approx.
¥3 that of other display types); 2) very
short duration (< 2 s); and 3) a general
similarity between the patterns which in

8 Advances in Herpetology and Evolutionary Biology

no way resembled the other display
types.

The a displays are simplistic in com-
parison to the other five display types,
giving me the impression that they rep-
resent display fragments. These related
patterns are composed of 6 to 7 quick,
simple bobs which are really more like
nods. The low amplitude nods associated
with this display type were produced
solely by the neck muscles. Dewlap
pulses following head bobbing ranged
from one to seven, with 12% of the dis-
plays having no pulses. On occasion, the
a displays were not completed, especial-
ly the a, and a, displays.

Stereotypy was fairly strong for most of
the a units, especially when considering
how short the unit durations are. The
coefficient of variation values for many of
the units fall close to or within Barlow’s

ay

DMRS bis (7!

SEGONDs

Figure 6. Summary DAP graphs of three versions of
the Type a display by Anolis distichus. Numbered
boxes immediately under each DAP graph indicate
duration and identity of bob and inter-bob units.

(1977) O-15% range for highly stereo-
typed behavior.

All but three a displays were filmed in
the lab. The context was usually associ-
ated with exploration; mostly solitary
males recently released into a new enclo-
sure performed the a displays and, in this
situation, the displays were non-directed.
The other context for the a displays
was then two males were in close prox-
imity to each other (< 10 cm). At this
time five such displays (11% of the
sample) were recorded. They were per-
formed by only one member of the pair.
It is very likely the a display carries a
connotation of uncertainty or subordina-
tion.

TYPE A DISPLAY

Type A display is the species’ signa-
ture display and is represented by a
single consistent pattern (Fig. 7A). The
pattern is composed of 11 units. Three
units delineate single bobs (Units 1, 5,
and 9) and three units contain multi-bob
series (Units 3, 7, and 11). Separating the
bob units are inter-bob pauses. Thus all
odd-numbered units depict bobs and
even-numbered units are pauses (Fig.
7A). In this display type and in display
types B, C, and D, one sees a construct of
single bob units alternating with multi-
bob units.

All the bobs of the Type A display are
quick up and down movements which
graph as “spikes.” Head amplitude
movement of these bobs is created by
simultaneous use of neck and limb
muscles. In over half the A displays all
four limbs came into play (Fig. 2). The
use of four legs rather than just forelimbs
seems to relate to increased arousal of the
displayer. Although anecdotal for A. dis-
tichus, this correlation has been more
firmly established in Anolis opalinus
(Jenssen, 1979). Four-legged bobs were
even more frequently incorporated into
the distichus Types B, C, D, and E dis-
plays.

As is true of all of the distichus display
types, dewlap pulsing following the head

LIZARD DISPLAYS - Jenssen 509

bob sequence was the usual case, the
number of pulses being variable. For
82% of the A displays, concluding dew-
lap pulses were performed; when pres-
ent, they ranged from 1-8 pulses.

Display stereotypy was moderate, with
9 to 11 of the 14 display variables (ex-
cluding dewlap pulses) having their CV
values within the 0 to 35% range for
stereotyped behavior.

Comparisons of the Type A display pat-
tern were made between the Montrouis
and Petion-ville samples. The two sum-
mary DAP graphs are very similar (Fig.
7A, and 7A,.). When comparing homol-
ogous display variables for the two locali-
ties, Wilcoxon’s Two-Sample tests (Sokal
and Rolf, 1969) indicated statistically
significant differences for many unit
durations and bob numbers within multi-
bob units. Thus there are _inter-
populational differences, but these di-
vergences are minor on a qualitative
level. They have not produced two dif-
ferent display patterns since bob cadence
and the shape of the bobs are shared by
the two populations. An A display by any
of the sampled lizards fits the summary
pattern.

Besides comparing between popula-
tions, I also compared the A displays
between individuals within a population
by using the 42 A displays of five lab-
held males from Montrouis. One-way
ANOVA tests showed significant differ-
ences (P < 0.05) for all but three display
variables. To see if one or more lizards
had display characteristics distinctly dif-
ferent from the rest of the sample, a
Duncan’s Multiple Range test was also
used to compare among lizards. Test re-
sults showed there were significantly dif-
ferent groups of subjects for all even-
numbered units and Unit 3. This same
generalization is reflected in the parti-
tioned variance, where those variables
which reflected individual-unique dif-
ferences (e.g., even-numbered units and
Unit 3) would also tend to have most (=
60%) of the partitioned variance in the
among group category.

560

Advances in Herpetology and Evolutionary Biology

SECONDS

Figure 7. Summary DAP graphs of Anolis distichus Type A and B displays. A, and B, are the patterns from
Montrouis in north and A, and B, are from Petion-ville in the south. Numbered boxes immediately under each
DAP graph indicate duration and identity of bob and inter-bob units.

Examples of interindividual display
differences are given in Figure 8; these
DAP graphs are summaries for three
males for which I had the most displays.
The display durations for the three males
were made comparable by computing
unit durations as a percentage of the total
display duration. One sees that the posi-
tion of homologous bobs between lizards’
displays is somewhat variable and the
average number of bobs performed for
the multi-bob units also differs be-
tween lizards. Yet it is evident that these
DAP graphs are all Type A displays and
cannot be confused with any of the other
distichus display patterns (e.g., the Type
B displays also depicted in Fig. 8).

The Type A display is considered the
species’ signature display by previously
defined criteria (Jenssen, 1978). First, it
is a well-defined stereotyped pattem
shared by the population. Second, it was
the most frequently performed display
type; from the film record there were 60
A, 43 a, 29 B, 28 E, 18 C, and 14 D dis-
plays. Last, it occurred during non-
directed displaying associated with terri-
torial patrol. The Type A display, as
expected of the signature display, ap-
peared in all other major social contexts:
during courtship, copulation, and the im-
mediate post-copulatory period, and dur-
ing aggressive interactions between
males.

TYPE B DISPLAY

Type B display pattern was quite con-
sistent and closely resembled the Type A
display, the primary difference being the
shape and duration of the single bobs of
Unit 1 and Unit 5 (Fig. 7B). There was
also a trend for more bobs in the multi-
bob Units 3 and 11 and fewer in Unit 7.
Dewlap pulses usually followed the head
bobbing (23 of 29 displays), and when
present, ranged from 1-7.

Display stereotypy was moderate, with
11 to 12 of the 14 display variables having
CV values within the 0 to 35% range for
stereotyped behavior.

When comparing between Montrouis
and Petion-ville Type B_ displays,
Wilcoxon’s Two-Sample tests resulted in
nonsignificance for all but four display
variables. As with the Type A display,
there were only minor quantitative dif-
ferences between the B displays of the
two populations. In essence, they shared
the same general pattern.

The primary context in which the B
displays were observed was that of male-
male aggression. However, on several
occasions males were filmed giving B
displays to females during courtship.

TYPE C DISPLAY

The C display type is also related to the
Type A display, and is composed of the
alternation of single and multi-bob units
(Fig. 9C). However, there are some large
departures from the A pattern, much
more so than seen in the B pattern. The
major differences are: 1) a greater separa-
tion of bob Units 1 and 3; 2) a change in
the shape of the head bobs of Units 1, 3,
and 7; 3) the appearance of the dewlap
during the head bobbing sequence (rare-
ly during Unit 4, always during Unit 6);
and 4) marked qualitative variability in
the display pattern. Like the Type A dis-
play, dewlap pulses usually followed the
head bobs; 79% of the C displays con-
tained post head bob dewlap pulses, and
when present they ranged from | to 9.
Major amplitude movement was fre-

LIZARD DISPLAYS - Jenssen 061

quently produced by four-legged push-
ups.

The C displays were qualitatively
variable, but tended to be of two versions
which I have designated as C, and C,
(Fig. 9C). Individual lizards performed
one of the two versions. My sample size
was too small, however, to confirm
whether there was a polymorphism in the
population for these two C sub-patterns,
or if lizards simply had individual-typical
versions of the Type C display. A C dis-
play was classified as one of the two ver-
sions on the basis of the bob shapes in
Units 1 and 3 (see Fig. 9C). Besides C,
and C,, no other pattern variants were
detected in my sample.

Contexts for the C displays included
courtship as well as male-male disputes.
The film record contains fights where
paired male contestants were very

aroused (i.e., displaying many postural
modifiers such as mouth open and orbed
eyes) and exchanging C displays at close
range. These were followed by one of the
males lunging forward and locking jaws
or biting his adversary on the body.

20 40 60 80 100

PERCENT DURATION
Figure 8. Summary DAP graph of Type A and B dis-
plays for three male Anolis distichus from Montrouis.
Display duration is expressed as a percentage.

562

Advances in Herpetology and Evolutionary Biology

A du A 6
— Dy

SECONDS

Figure 9. Summary DAP graphs of Anolis distichus Type C and D displays. C, and C, are two versions of the C
display type. Stippled dewlap extension indicates the pulse was not always present. Numbered boxes indicate

duration and identity of bob and inter-bob units.

TYPE D DISPLAY

As is true of display Types A-C, the
Type D pattern is also based on single
bob units altemating with multi-bob
units (Fig. 9D). However, the D pattern
least resembles the signature display,
having several unique characteristics.
First, the D pattern is “indeterminate”
because it has no fixed end point. In this
sense it is similar to the cybotes signature
display. Second, the multi-bob units 7,
11, and 15 tend to have only two bobs per
unit, rather than a wide range of possible
bob numbers as seen in the multi-bob
units of the A-C patterns. Third, dewlap
pulses occur throughout the head bob
sequence.

The large number of units for the Type
D display reflects the repetition of the
single and multi-bob units (essentially
the bobs of Units 5 and 7). Most D dis-
plays were performed through Unit 11,
but some ceased as early as Unit 7, some
as late as Unit 15. There was a weak
tendency to stop after a multi-bob unit.

With the repetition of the single and
multi-bob units, the “plateau-spike”
shaped bobs of the D display tend to
grade into “spike-like” bobs toward the
end of a display (Fig. 9D). Dewlap pulses
always appeared after the head bob se-
quence. In the 58% of the D displays
which concluded with dewlap pulses,
the number of pulses ranged from | to 9.
The context for the Type D display was
primarily male-male interactions, but two
displays were filmed during male to
female displaying. The D display is
obviously an agonistic signal as it is
accompanied by display modifiers (e.g.,
tongue out) commonly associated with
aggressive behavior (Jenssen, 1979).
How the D displays function in male to
female relationships is uncertain.

TYPE E DISPLAY

Type E display does not share the
construct of alternating single and multi-
bob units as do the A-D display types.
After Unit 3 there is an indeterminate

series of single bobs (Fig. 1OE). This dis-
play type appears to be a simplified ver-
sion of the Type D pattern.

As in the C and D displays, the dewlap
is pulsed during head bobbing (i.e., in-
ternal dewlap pulses). For most of the E
displays, there was a single internal dew-
lap pulse just before the last head bob,
and in some performances there were
multiple internal dewlap pulses (Fig.
10E, stippled outline of dewlap move-
ment). The dewlap was also pulsed after
the last head bob in all E displays, rang-
ing from | to 7 extensions per display. All
E displays were performed through Unit
7, but many contained additional single
bobs. Covarying with this indeterminate
bob series were changes in bob shape
(Fig. 10E..) as seen to a lesser degree in
D displays. The addition of single bobs
during repeated E displays by a male,
covaried with the bob of Unit 1 changing
from a “spike” to a “plateau-spike.” The
number of bobs in Unit 3 also increased,
and the single bobs tended to metamor-
phose into the plateau-spike configura-
tion. This transformation of the E pattern

SECON Ds

Figure 10. Summary DAP graph of Anolis distichus
Type E display (E), with an example of four sequenced
displays by a single male (E.,). Stippled dewlap exten-
sion indicates the pulse was not always present. Num-

bered boxes indicate duration and identity of units.

LIZARD DISPLAYS - Jenssen 963

during repeated performances is unique
among previously studied lizard species.
The closest resemblance to this phenom-
enon is a shifting pattem within display
performances by the banded iguana,
Brachylophus fasciatus (Greenberg and
Jenssen, 1983).

Because the Type E display pattem
was not qualitatively stable, unit dura-
tions and bob shape of given bob units
experienced a wide range of variation.
Therefore, the DAP graph of the Type E
display (Fig. 10E) is not a good gener-
alization for all E displays. It simply
reflects mean unit durations and inter-
mediate bob shapes. That is why DAP
graphs for four consecutive E displays
were provided in Figure 10.

The context for all E displays was
resident-intruder interaction in enclo-
sures. No E displays were filmed in the
field. Their occurrence in the field may
require intense territorial defense which
is a rare event. In enclosures, high levels
of agonistic behavior are more easily
induced since there are no potential
escape routes available. The progressive
transformation of the E pattern may indi-
cate the displayer’s shift to a more in-
tense level of agonistic intent.

DISCUSSION

Anolis cybotes and A. distichus domi-
nicensis are two common species on
Hispaniola; the first belongs to the
cybotes species group and the second to
the stratulus species group (Williams,
1976). A. cybotes is a medium-sized
anole (males ~ 65 mm SVL), while A.
distichus is smaller (males ~ 50 mm
SVL). The two species are sympatric in
many areas. Their structural habitat
preferences are quite similar. Adults of
both species utilize broad perch sub-
strates, primarily tree trunks and rock
outcroppings. They are usually found
from ground level up to about 3 m for A.
cybotes and up to 8 m for A. distichus
(Moermond, 1979).

564

There was no a priori suspicion that
these two sympatric congeners would
have such radically different display
repertoire sizes. Anolis cybotes has only a
single species-unique head bob display
type, its signature display. Other social
behaviors observed, but not included in
the repertoire count, were display modifi-
ers (see Jenssen, 1978, 1979), some
weakly stereotyped head dipping, and
“rapid head bobs,’ sometimes called
“jiggling” (see Jenssen, 1977). This last
behavior has been reported by almost all
studies of iguanid social behavior; it is
most frequently performed by a courting
male on the move. In contrast, A. dis-
tichus has six species-unique display
types in its repertoire, excluding its own
display modifiers and rapid head bob-
bing.

DISPLAY EVOLUTION

For the display-rich A. distichus, it is
tempting to speculate on the evolution of
its contemporary set of display types. As
previously proposed for the evolution of
anoline displays (Jenssen, 1977), it is
parsimonious to assume that the primi-
tive distichus repertoire had but one dis-
play type, the signature display. This is a
situation not unlike that of present day A.
cybotes. Since the A-D display types of A.
distichus are based on alternating single
and multi-bob units, the ancestral signa-
ture display probably shared this theme.
The ancestral signature display is de-
picted as three series of alternating single
and double bobs (Fig. 11). I am curious
about the bob cadence of the other A.
distichus subspecies. It will be instruc-
tive to see if their signature displays can
be derived from my hypothetical signa-
ture DAP, or if some other ancestral pat-
tern will be more appropriate.

The current Type A display was
derived from the ancestral signature
DAP. The Type B display evolved next
(Figs. 11 and 12). The B pattern is very
similar to that of the Type A display. One
need only visualize the lengthening of

Advances in Herpetology and Evolutionary Biology

the bobs in Units 1 and 5 to derive the B
display from the A (Fig. 11). It is not un-
usual for anoles and for some other
iguanids to have a two display type reper-
toire in which the second display pattern
is very similar to the species’ signature
display (Jenssen, 1975, 1981; Ortiz, 1979;
Rothblum and Jenssen, 1978); this is also
true for anoles with more than two dis-
play types in their repertoires (Hover and
Jenssen, 1976; Jenssen and Rothblum,
1977).

After the Type B evolved, the Type A
display also provided the basis for the
Type C display. This took place with the
additions of an internal dewlap pulse
(i.e., a pulse during the head bobbing
sequence) and the re-shaping of the
multiple bob series of Units 3 and 7
(Figs. 11 and 12). Like the A and B dis-
plays, the Type C display is determinate,
having no bobs after Unit 11. With the
evolution of its internal pulses, the Type
C display was most likely the source for
the Type D display which also has in-
ternal pulses, but is indeterminate. It is
most probable that the D pattern was
derived from the C in two steps; the bob
of Unit 1 was deleted and the D pattem
became indeterminate by indefinitely
repeating Units 5 and 7 (Figs. 11 and 12).
The E display type came from the D
through a process of reduction; the Type
E pattern has fewer internal pulses and
has lost the alternating single and multi-
bob units (Fig. 11 and 12). The Type E
displays continue the indeterminate con-
struct of the D display type.

The last display type of the repertoire
is composed of the a displays. These
seem to be fragments of either the ances-
tral signature display or of the current
signature pattern. Because of their very
low amplitude and simplicity of pattem,
the a displays do not fit into the A-E flow
of the above hypothetical scheme.

I am left with the question of why A.
distichus has such a large repertoire of
social signals, while A. cybotes, another
quite successful congener, suffices with a
simple repertoire. To correlate large rep-

565

LIZARD DISPLAYS - Jenssen

‘ailoede Aeidsip snyoysip syouy eu) BAUEp 0} pasinbes sdays AreuoNNjons ey} HuysebH6ns awayos jeoneujodAH “41 aunbi4

ONILVSd3y 2 HUN

!
ONILVSdSY 2 8G Stuf ae, a 9 t !

dvd
SYNLVNSIS
WHYLSSAONV

(dvVd e4njoubis) Vv

\
. fs
: /
l e Lzysfs] + Pe yeti
'
\
\

\ '
'

Zz e ([s]_> Te yeti)
f

\
'
1
1
t

566

eine

pattern
fragments

ANCESTRAL
SIGNATURE
DAP
internal
pulses

0
deleted
Wroakys I

e
pattern
reduction

Advances in Herpetology and Evolutionary Biology

A-D composed of
alternating
single and

multi-bob units

C-E

contain

internal
dewlap
pulses

D-E have
indeterminate
construct
(repeating units)

Figure 12. Diagram showing the relationships between Anolis distichus display types and the hypothesized
evolutionary events producing the multiple display repertoire (refer also to Fig. 11).

ertoire anoles with socio-ecological fac-
tors not shared with congeners of limited
repertoires is still premature, since there
are still so few anoles known to have
repertoires of three or more display
types. Nevertheless, using my experi-
ence with A. cybotes and A. distichus,
some tentative factors can be suggested.

DISTICHUS HYPOTHESIS

All of the distichus displays are used in
agonistic encounters (Table 1). How
these six display types function in ag-
gressive interactions is not well known.
However, in A. limifrons, the displays of
its large repertoire are used sequentially
(Hover and Jenssen, 1976). As the dis-
tance decreases between contesting
males, their exchanged displays progres-
sively shift through the inventory of dis-
play types in a hierarchial manner. This
phenomenon is an elaborate ritualized
aggressive system which seems to allow
males to assess each other’s willingness
to fight; it actually decreases the likeli-

hood of fighting by extending the period
for withdrawal without risk of injury. I
suspect the distichus repertoire is used
in the same way.

Possible factors favoring a more elab-
orate bluff system are: 1) shortlived, high
turnover species active throughout the
year; 2) relatively high density per unit
area with numerous male neighbors; 3)
relatively small body size with concomi-
tant small territory size; and 4) structural-
ly complex microhabitat with widely and
continuously available perch substrates.

Under this set of conditions, a male
would experience frequent contacts with
its neighboring males, some being new to
the group of associates. A complex micro-
habitat, such as a large fig tree, would
provide a substrate for many territories
with no natural boundaries between ter-
ritories. A single male could not control
the entire microhabitat complex, and
would need to continuously patrol to
maintain his residential boundaries,
especially against intrusions by new
neighbors. Conversely, the same exten-

LIZARD DISPLAYS - Jenssen 567

TABLE 1. GENERAL ASSOCIATION OF DISPLAY WITH GROSS CONTEXTUAL SITUATIONS.
An “X” indicates a strong association, an “x” a weak association, and a “?” an uncertain association.

General Context

Species Display Type Non-directed Courtship Aggression
cybotes Signature xX ™
distichus A (Signature) x x

a x
B x xX
GC x Xx
D re xX
E xX

sive habitat complex would provide liv-
ing space for many lizards (i.e., not be a
significant limiting resource), although it
would certainly vary in preferred quality.
The male population within this patch of
habitat could easily reflect a shifting
strategy between strict _ territoriality
(specific defended area) and a social
hierarchy. A few dominant males would
have fairly stable territories, while less
dominant males would inhabit less pre-
ferred parts of the substrate complex and
have unstable or floating territories.
Since many male-male contacts would
be experienced throughout the day, en-
forcement of territorial borders would be
best served by graded ritualized aggres-
sion, modulated by the potential threat of
the intruder. If all members of the popu-
lation were quick to fight, the incidence
of injury would be great; this is not likely
to be selected for as indicated below.
Applying game theory to animal con-
flicts, Maynard Smith and Price (1973)
concluded that there should be no stable
pure strategy in aggressive contests; a
mixed strategy of at times “hawk” and at
times “mouse” would be stable (evolu-
tionary stable strategy-ESS). An ESS
requires that contestants respond to es-
calated aggressive behavior by escalating
in return, and that individuals vary their
behavior from contest to contest depend-
ing on the opponent’s fighting power and
willingness to fight (Dawkins and Krebs,
1978). Where fighting is avoided because

the cost of injury outweighs the benefits
(e.g., being injured while maintaining a
portion of a relatively abundant re-
source), one would expect an extensive
and graded repertoire of agonistic signals
by which to assess an opponent's fighting
ability. Thus the ESS becomes one of
“assessor/retaliator,’ using conventional
or ritualized displaying; escalation to
actual fighting should occur only be-
tween evenly matched contestants
(Parker, 1974).

How well A. distichus fits my sug-
gested scheme requires still ungathered
data on its behavioral ecology. However,
collected males showed no fight marks,
no naturally occurring fights were ob-
served, and only in enclosures (where
living space was truly limited) was fight-
ing induced.

CYBOTES HYPOTHESES

The simple repertoire of A. cybotes
suggests a different approach to social
interactions than that of A. distichus.
Factors likely to maintain a more primi-
tive repertoire are: 1) a relatively low
density per unit area, with few male
neighbors; 2) a relatively low tumover
rate, with a stable set of nearest neigh-
bors; 3) relatively large body size, with
concomitant large territory size; and 4)
territories which encompass entire or
nearly entire patches of preferred micro-
habitat.

568

Under these conditions a territorial
male would not be frequently challenged
by its neighbors. Strips of unsuitable
habitat forming an ecological “no man’s
land” would frequently separate terri-
tories. Territories could be rigidly de-
fended and controlled because natural
breaks in the microhabitat would delin-
eate territories and decrease casual con-
tacts. Intrusion into a territory would
represent a serious threat to a resident,
since eviction would mean a significant
search time for a new territory by the ex-
resident. If resident and intruder were
moderately matched for fighting ability
(i.e., similar body size), a rapid escalation
to fighting is predicted because of the
high benefit to holding or acquiring the
territory. by a contestant (suitable habitat
may be a more limited resource than in A.
distichus). A large repertoire of ritualized
signals would be unnecessary because of
the relatively rare event of male-male
contact and a greater tendency to risk
injury to protect a more limited resource.

My observations of A. cybotes indicate
that in comparison with A. distichus they
are very aggressive and prone to injuri-
ous fighting. Anolis cybotes were seen
engaged in naturally occurring fights.
Many (~ 20%) of the observed and col-
lected males bore scars and fresh head
and jaw wounds. Several males had very
badly damaged jaws which were obvi-
ously broken.

ACKNOWLEDGMENTS

I am indebted to a number of people:
Chris Pague and Lars Jenssen assisted
me in the field; Teri Brentnall and Mike
Menjak conducted initial film analyses
on parts of the film record; Ned Fetcher
helped with the computer programming;
and Pam Pettry typed the manuscript.
Financial assistance came from a Na-
tional Science Foundation Grant DEB
7420143.

Advances in Herpetology and Evolutionary Biology

LITERATURE CITED

ARNOLD, D. L. 1980. Geographic variation in Anolis
brevirostris (Sauria: Iguanidae) in Hispaniola.
Breviora Mus. Comp. Zool. No. 461, pp. 1-31.

BarRLOw, G. W. 1977. Modal action patterns, pp. 98-
134. In T. A. Sebeck (ed.), How animals com-
municate. Bloomington, Univ. Indiana Press.

CocuHRAN, D. M. 1941. The herpetology of His-
paniola. Bull. U.S. Nat. Mus., 177: 1-398.

DAWKINS, R., AND J. R. KREBS. 1978. Animal signals:
Information or manipulation?, pp. 282-309. In
J. R. Krebs and N. B. Davies (eds.), Behav-
ioural ecology, an evolutionary approach.
Sunderland, Massachusetts, Sinauer Associ-
ates.

GREENBERG, N., AND T. A. JENSSEN. 1983. Social
behavior of captive banded iguanas, Brachy-
lophus fasciatus. In G. M. Burghardt and A. S.
Rand (eds.), [guanine lizards of the world: Be-
havior, ecology, and conservation. New Jersey,
Noyes Publ. In Press.

Hover, E. L., AND T. A. JENSSEN. 1976. Descriptive
analysis and social correlates of agonistic dis-
plays of Anolis limifrons (Sauria, Iguanidae).
Behaviour 58: 173-191.

JENSSEN, T. A. 1975. Display repertoire of a male
Phenacosaurus heterodermus (Sauria: Iguani-
dae). Herpetologica 31: 48-55.

__.. 1977. Evolution of anoline lizard display be-
havior. Amer. Zool., 17: 203-215.

____. 1978. Display diversity in anoline lizards and
problems of interpretation, pp. 269-285. In N.
Greenberg and P. MacLean (eds.), Behavior
and neurology of lizards: An interdisciplinary
conference. N.I.M.H., Rockville, Maryland.

___. 1979. Display modifiers of Anolis opalinus
(Sauria, Iguanidae). Herpetologica, 35: 21-30.

__. 1981. Unusual display behavior by Anolis
grahami from westem Jamaica. Copeia, 1981:
728-733.

JENSSEN, T. A., AND E. L. Hover. 1976. Display
analysis of the signature display of Anolis limi-
frons (Sauria, Iguanidae). Behavior, 57: 227-

240.
JENSSEN, T. A., AND L. ROTHBLUM. 1977. Display
repertoire analysis of Anolis townsendi

(Sauria, Iguanidae) from Cocos Island. Copeia,
1977: 103-109.

MAYNARD SMITH, J., AND G. R. PRICE. 1973. The
logic of animal conflict. Nature, 246: 15-18.

MOERMOND, T. C. 1979. Habitat constraints on the
behavior, morphology, and community struc-
tures of Anolis lizards. Ecology, 60: 152-164.

OrTIz, P. 1979. Behavioral comparison of anoline
competitors: Anolis cooki and Anolis cristatel-
lus (Reptilia: Sauria: Iguanidae). M.S. Thesis.
Virginia Polytechnic Institute and State
‘University, Blacksburg, Virginia.

PARKER, G. A. 1974. Assessment strategy and the

evolution of fighting behaviour. J. Theor. Biol.,
AT: 223-243.

ROTHBLUM, L., AND A. JENSSEN. 1978. Display
repertoire analysis of Sceloporus undulatus
hyacinthinus (Sauria, Iguanidae) from south-
westerm Virginia. Anim. Behav., 26: 130-137.

ScHWaRTZ A. 1968. Geographic variation in Anolis
distichus Cope (Lacertilia, Iguanidae) in the
Bahama Islands and Hispaniola. Bull. Mus.
Comp. Zool., 137: 255-310.

____. 1980. Variation in Hispaniolan Anolis white-
mani Williams. J. Herpetol., 14: 399-406.
SOKAL, R. R., AND F. J. ROHLF. 1969. Biometry. San

Francisco, W. H. Freeman and Co.

STEEL, R. G. D., AND J. H. TorrIE. 1960. Principles
and procedures in statistics. New York,
McGraw-Hill.

WEBSTER, T. P. 1975. An electrophoretic compari-
son of the Hispaniolan lizards Anolis cybotes
and A. marcanoi. Bull. Mus. Comp. Zool., 431:
1-8.

____. 1977. Geographic variation in “Anolis brevi-
rostris’: Evidence from proteins and a consid-
eration of dewlap color, pp. 153-165. In E. E.
Williams (ed.), The Third Anolis newsletter.
Mus. Comp. Zool., Harvard Univ., Cambridge,
Massachusetts.

LIZARD DISPLAYS - Jenssen 569

WEBSTER, T. P., AND J. BURNS. 1973. Dewlap color
variation and electrophoretically detected sib-
ling species in a Haitian lizard Anolis breviros-
tris. Evolution, 27: 368-377.

WiLuiaMs, E. E. 1963. Anolis whitemani, new
species from Hispaniola (Sauria, Iguanidae).
Breviora Mus. Comp. Zool. No. 197, pp. 1-8.

____.. 1975. Anolis marcanoi new species: sibling to
Anolis cybotes: Description and field evi-
dence. Breviora Mus. Comp. Zool. No. 430, pp.
1-9.

___. 1976. West Indian anoles: A taxonomic and
evolutionary summary. I. Introduction and a
species list. Breviora Mus. Comp. Zool. No.
440, pp. 1-21.

__. 1977a. Additional notes on distichoids, pp.
174-182. In E. E. Williams (ed.), The Third
Anolis newsletter. Mus. Comp. Zool., Harvard
Univ., Cambridge, Massachusetts.

____. 1977b. Species problems, pp. 132-151. In E.
E. Williams (ed.), The Third Anolis newsletter.
Mus. Comp. Zool., Harvard Univ., Cambridge,
Massachusetts.

WYLES, J. S., AND G. C. GoRMAN. 1980. The classifi-
cation of Anolis: Conflict between genetic and
osteological interpretation as exemplified by
Anolis cybotes. J. Herpetol., 14: 149-153.

The Social Behavior of Anolis valencienni

ROBERT A. HICKS!
ROBERT L. TRIVERS?

ABSTRACT. Anolis valencienni is highly cryptic
and displays many features atypical of other Anolis.
Adult females are completely nontervitorial and
nonaggressive, frequently passing within centi-
meters of each other. Adult females overlap with
each other extensively in their daily movements.
Sometimes as many as 40 females may occupy a
common feeding space. Males are aggressive to
each other but show broad overlap in the space
occupied. It is not uncommon for five adult males or
more to share much of the same space. Consistent
with their lack of territories, adult females wander
slightly greater distances than do adult males, even
though they are smaller. Males appear to suffer dif-
ferential mortality, and the adult sex ratio is biased
toward females. Both sexes feed by searching slow-
ly over the substrate of trees and bushes for insects,
many of which are cryptic and at rest. Because many
adults of both sexes encounter each other daily,
there are unusual opportunities for female choice.
Females copulate often during the summer, some-
times, and probably often, more than once a day.
Over a period of six weeks a female may copulate
with five or more males. Experiments were con-
ducted to test the criteria females use in choosing
males, but these were not successful. Field data
show that both adult males and adult females copu-
late more often as they increase in size, but for
males this relationship appears to be weaker than in
the more size-dimorphic Anolis garmani. It is pro-
posed that the many unusual features of A. valen-
cienni result from the primary feeding adaptation of
searching for stationary, cryptic prey.

INTRODUCTION

In appearance, Anolis valencienni is
the most cryptic of Jamaica’s seven
species of anoline lizards. It is also the

1 P.O. Box 420, Ipswich, Massachusetts 01938,
U.S.A.

2 Applied Sciences, University of California, Santa
Cruz, California 95064, U.S.A.

slowest moving, crawling deliberately
and very slowly over the trees it inhabits.
Both in size and in secondary sexual
structures such as the dewlap, it is like-
wise unique in that females are never
aggressive toward each other and tolerate
the presence of other females at distances
of centimeters. More than a year of field
work on this species suggest to us that the
above attributes of valencienni are all
interrelated and ultimately explained by
valencienni’s unique feeding specializa-
tion: individuals search actively for and
feed upon cryptic insects, many of which
may be diurnally immobile. We describe
here the results of this field work.

METHODS

Anolis valencienni was studied in two
different localities and at several differ-
ent times in each locality.

Study site 1 (Maryfield) was located in
a suburban area of Kingston and con-
sisted of the three acre grounds of the
Maryfield Guest House. These, in turn,
consisted of scattered large trees (usually
over 10 m high) typical of Jamaica, mostly
mangos (Mangifera indica) but also
Persea americana, Guaiacum officinale,
Delonix regia, Terminalia catappa,
Haematoxylum campechianum, Swie-
tania mahogoni, Gliricida sepium, and
Brya ebenus. During 12 days in Septem-
ber 1968, feeding behavior was studied
by following (through binoculars) un-
marked lizards. From June 15 to Septem-
ber 14, 1969, the social behavior of these

lizards was studied for a total of eleven
weeks by capturing, sexing, measuring,
marking, releasing and later following
265 lizards in the study area. Brief visits
to this study area (to recapture marked
individuals) were made in January 1970,
and again in April 1970. All observations
in study site 1 were made by Trivers, and
no assistants were employed. A. valen-
cienni were abundant in this study site,
but because the site consisted of many
large trees, the lizards were difficult to
resight (and recapture) and most of our
data on social behavior come from the
second study site.

Study site 2 (Southfield) was located in
rural St. Elizabeth and was chosen be-
cause valencienni were abundant and be-
cause the study site consisted of many
bushes and short trees (most were under
6 m in height). Typical tree species, in
addition to mangos, were Pimenta
officinalis, Cocos nucifera, Manilkara
zapotilla, Blighia sapida, Chrysophyl-
lum cainito, and Persea americana.
Social behavior was studied for two
weeks in June 1971, by Trivers with the
help of field assistants, and for seven
weeks in July and August 1973, by Hicks
with help of field assistants. All resight-
ings of individuals, including copulations
and aggressive encounters, were con-
firmed by Trivers or Hicks even if first
made by one of the assistants. The
second site was visited by Trivers for
eleven days in December 1973 to Jan-
uary 1974 in order to recapture marked
valencienni. In addition, Trivers studied
Anolis garmani for a year in Southfield,
where valencienni were abundant, and
he made numerous casual observations
during this year. Finally, five months of
field work in the spring of 1976 and in
July 1977, concentrated on measuring
female choice.

Social behavior was studied as in A.
garmani (Trivers, 1976); that is, an effort
was made to capture, sex, measure, and
mark all adults in the study area. Individ-
uals were given a permanent mark by
clipping two or three toes in a pattern
unique for each lizard. Likewise, a num-

ANOLIS VALENCIENNI « Hicks and Trivers

ay

ber was painted on each lizard’s back,
permitting identification in the field
without recapture. As molting takes place
about every three weeks in valencienni,
individuals were often recaptured and
repainted. Such recapture work permit-
ted some systematic data on growth rates,
and these are analyzed below as a func-
tion of size, sex, and rate of dispersal.
When an effort was made to gather copu-
lation data (see below), the study area
was systematically searched for copulat-
ing lizards. If these were marked, num-
bers were recorded. If one or both were
unmarked, the unmarked individuals
were captured. It was not possible to esti-
mate precisely to what extent the sample
was a biased sample of those copulations
actually occurring, but some biases were
apparent. These, and other methodolog-
ical problems are discussed, where rele-
vant, below. Resightings were recorded
while searching for unmarked lizards and
for copulating pairs. The data from re-
sightings are analyzed below. In addi-
tion, observations of aggressive, sexual,
and neutral encounters were made in the
process of the work described.

RELATIVE CRYPTICITY

Anolis valencienni is unusually cryptic
(Lynn and Grant, 1940; Underwood and
Williams, 1959; Rand, 1967a; T. W.
Schoener and A. Schoener personal com-
munication), and experienced Jamaicans
routinely assert that valencienni is the
most cryptic anole on the island. Several
lines of evidence support this opinion. In
what follows, we emphasize the compari-
son between valencienni on the one hand
and A. lineatopus and A. grahami on the
other; because the latter two species are
similar in size to valencienni, they often
occupy similar perches, and they both
occur in our two study areas.

Sightings. 1) Ona typical day in either
of the study sites, individual lineatopus
and grahami were sighted first, although
we were searching for valencienni. 2)
Marked grahami and lineatopus are

Ol
=]
bo

much more frequently resighted than
marked valencienni. (This is partly due to
the greater crypticity of valencienni and
partly due to its less predictable choice of
perches.) 3) In both study sites, valen-
cienni initially appeared to be rare, but
marking revealed valencienni to be more
common than grahami in both study
sites, more common than lineatopus in
the second study site, and only slightly
less common than lineatopus in the first
study site. 4) Rand’s (1967a) census study
of Jamaican anoles found valencienni to
be less common than any other species
except garmani. This is expected since
lizards were not marked and each study
area was traversed only once per day.
Underwood and Williams (1959) likewise
point out that valencienni is much more
common than the number of collected
specimens would suggest “for it is very
clever at concealing itself.” 5) It is a com-
mon experience to be attracted to a tree
because one spots one or more linea-
topus or grahami and after watching
these for several minutes to notice an
overlooked valencienni, often because it
finally moves.

Morphology. 1) A. valencienni is usu-
ally grey in color (sometimes brown) with
irregular dark markings which tend to
obliterate body outline. This contrasts
with the solid blue (or dark brown) color
of grahami and the regular, reticulated
pattern of lineatopus. 2) In contrast to
grahami and lineatopus, the splotched,
irregular pattern of valencienni covers
the entire body, including the belly, the
face and lower jaw. 3) All other Jamaican
anoles have a bright, yellow eye-ring.
This eye-ring is obliterated when the
lizards are frightened in social en-
counters and their entire bodies darken.
A. valencienni lacks the eye-ring en-
tirely.

Behavior. Individual valencienni
move very close to the surface they are
climbing on. When they do move, they
move very slowly, so that they appear to
glide quietly and nearly invisibly from
place to place.

Advances in Herpetology and Evolutionary Biology

SEXUAL DIMORPHISM

Both in size and in secondary sexual
characteristics, valencienni is, along with
A. reconditus (Hicks, 1973), the least
dimorphic Jamaican anole.

Data on adult size dimorphism from
different visits to the two study areas give
consistent results. Since the data from the
summer of 1973 (Southfield study site)
are the most complete, they are pre-
sented here. The snout-vent length of the
largest five males averages 85.2 mm
while that for the largest five females
averages 73.8 mm. The smallest adult
male seen copulating was 54 mm in size
(= snout-vent length), while the smallest
female seen copulating was 50 mm in
size. Using these sizes as size of sexual
maturity, the average adult male size (N
= 93) is 72.8 mm, while the average adult
female size (N = 98) is 64.8 mm. At all
sizes for which data are available, males
grow faster than females of similar size
(Fig. 1). Like other anoles, male valen-
cienni have larger jaws than females of
similar snout-vent length.

It is of particular interest to know
whether the size dimorphism in valen-
cienni results entirely from differential
growth rate or whether males hatch out at
a larger size than females. A sample of 55
valencienni hatched from eggs (and then
preserved) is available at the Museum of
Comparative Zoology. These were
measured and then sexed. Forty-nine of
these could be sexed unambiguously (by
the enlarged post-anal scales of the
males), but six individuals had post-anal
scales intermediate in size. Although
these are probably males, they have been
treated separately. No sex difference in
size at hatching is apparent. Mean size of
known males (N = 22) is 23.8 mm, while
the mean size of known females (N = 27)
is 23.7 mm. If the six uncertain individ-
uals are indeed males then the mean size
of males (N = 28) is 24.0.

There are no pronounced secondary
sexual structures in valencienni except
the dewlap, which is very large and

20

@ MALE
* FEMALE

a

GROWTH RATE (MM/100 DAYS)
a fo)

30 40 50 60 70 80 S/O)
SIZE (MM)

Figure 1. Growth Rate (mm/100 days) is plotted as a
function of initial size (Snout-vent length in mm) for
males and females between summer 1973 and winter
1973-74 in Southfield. On average each individual of
both sexes was recaptured six months after initial cap-
ture.

bright purple in the male and smaller and
less brightly colored in the female, often
showing many white scales.

ADULT SEX RATIO

In all visits to both study areas more
adult females were captured than adult
males. We divide the data by study area,
however, because at Maryfield females
were easier to capture (Table 1) while
males were easier at Southfield (Table 2).
In order to make inferences about differ-
ential mortality by sex, we present not
only the number of adult males captured
(= 55 mm), but also the number of males
= 50 mm captured, since the latter cate-
gory certainly contains individuals
younger than the youngest adult females
(see below).

During the summer of 1969 at Mary-
field 198 adult females (= 50 mm) were
captured compared to 69 adult males (=
50 or 55 mm). These data are certainly
biased, because males were much less
often sighted and were less often cap-
tured when sighted (since they stayed
higher in the trees). As evidence of this
bias, for example, the first 29 individuals
captured in the study area were females.

ANOLIS VALENCIENNI + Hicks and Trivers

573

TABLE 1. NUMBERS OF MALES AND FEMALES CAPTURED
AT MARYFIELD.

males
females (=55 mm or
(250mm) = 50 mm)
Summer 1969 199 62
Winter 1969-1970 26* OF

*Of these 26 females, 20 had been marked in the
summer of 1969.

tOf these 9 males, 3 had been marked in the sum-
mer of 1969.

Likewise, of the 121 adult females cap-
tured on or before August 6, 48% (58)
were, at one time or another, recaptured,
but of the 41 adult males marked on or
before August 6, only 31.7% (13) were
ever recaptured. Thirty-seven females
marked on three large connected trees on
or before August 6 (see “Female Move-
ments ) were resighted a total of 19 times
after August 6 for an average of 0.51 re-
sighting per female, while 12 males
marked on or before August 6 were re-
sighted thereafter a total of only once, for
an average of 0.08 resighting per male.
Recapture data permit a crude estimate
of the real adult sex ratio at Maryfield.
During seven days in December 1969 to
January 1970 and during three days in
April 1970, a total of 35 adults were cap-
tured of whom 23 had been marked in the
summer of 1969. Of the 26 females cap-
tured 88.5% (23) were marked the pre-
ceding summer, while of the nine adult
males captured only 33% (3) had been
marked. It is unlikely that many lizards
migrated in and out of the study area,
which was very large and bounded on
three sides by roads. A very crude esti-
mate of the real adult sex ratio at
Maryfield (assuming no _ differential
migration) is 186 males and 259 females.
At Southfield males appeared to be
slightly easier to spot than females and,
once sighted, very few individuals of
either sex eluded capture. The 92 adult
males captured were resighted a total of
158 times for an average of 1.72 resight-

574

ing per male while the 97 adult females
were resighted a total of 151 times for an
average of 1.56 resighting per female. Yet
each visit to Southfield showed more
adult females than adult males (Table 2).
None of the differences are significant;
nor, when combined, are the totals signif-
icantly different.

Recapture data appear to confirm that
more females than males remained un-
captured during the summer of 1973. Of
the 31 females captured during the
winter of 1973 to 1974, 67% (21) had been
marked during the preceding summer,
while of the 24 males captured 75% (18)
had been marked in the preceding sum-
mer. These data may partly reflect
greater female mobility, as data from the
summer of 1973 show a slightly greater
distance between successive sightings of
females than of males (see below).
Assuming no difference in mobility, a
crude estimate of the real adult sex ratio
in the summer of 1973 is 92 males and
123 females. The male : female sex ratio
for animals of all sizes at Southfield
(1973) was 110: 121.

Males grow faster than females at all
sizes for which we have data. If, as in A.
grahami (Trivers, 1976) and A. linea-
topus (Rand, unpublished data), the sex
difference in growth rates extends to
small individuals (as appears likely) then
a female 50 mm in length is in fact, older
than a male of the same size. Likewise, a
newly adult female is almost certainly
older than a newly adult male. Thus, the

TABLE 2. NUMBERS OF MALES AND FEMALES CAPTURED
AT SOUTHFIELD.

males
females (=55 mm or
(=50mm) = 50 mm)
June 1971 44 35
Summer 1973 97 92
Winter 1973-1974 31* 24F

* Of these 31 females, 21 had been marked in the
summer of 1973.
+ Of these 24 males, 18 had been marked in the
summer of 1973.

Advances in Herpetology and Evolutionary Biology

sex ratio for adults of the same age is
expected to be more strongly biased
toward females than our data on all adults
(and on all individuals = 50 mm) show. If
the sex ratio of hatchlings is 50/50 (as
expected, since there is no size differ-
ence between the sexes at hatching),
then our data suggest differential male
mortality. What data we have suggest that
the sex ratio at hatching may, indeed, be
50/50 but the data are few and our ability
to recognize males, imperfect. Of the 24
individuals < 35 mm caught at Southfield
during the summer of 1973 and winter
1973 to 1974, 11 were males and 13,
females.

FEEDING BEHAVIOR

Individual valencienni capture insect
prey from the ground at the base of trees
to the outer edges of leaves forty feet
above the ground. They do so by search-
ing slowly and carefully throughout the
tree (or bush) in which they live. When a
lizard first spots a prey item, the item is
usually within 8 cm of the lizard’s face
and is usually captured once spotted.
These findings emerge from 84 hours of
systematic observation of unmarked
valencienni in 1968 and are supported by
numerous incidental observations made
during the summers of 1969 and 1973.

In September, 1968, 36 feeding
attempts were observed. An attempt at
feeding was assumed to occur if an indi-
vidual moved quickly for several inches
and snapped its jaws. The feeding
attempt was gauged to be a success if the
jaw snap was followed by chewing (i.e.,
by jaw movements). In the majority of
unsuccessful attempts, a small prey item
(assumed to be an insect) fluttered away.
Likewise, in the majority of successful
attempts, an insect was seen in the
lizard’s mouth before being consumed.

The distance at which the lizard first
spotted the prey item was recorded as the
distance from which the lizard began its
lunge. There is no evidence that individ-
ual valencienni ever creep up on prey.

Usually, the lizard is moving slowly
over an area, constantly searching it
visually, when it interrupts the searching
to lunge and snap, often at something
hidden in the bark under its nose. Of the
36 feeding attempts, 27 were successful.
Of the 34 for which the distance the prey
item was spotted was recorded, all but
three times, the prey item was less than
12 cm from the lizard when first spotted.
The furthest a prey item appeared to be
spotted was 45 cm. The feeding attempts
ranged from ground level to 13 m up.
Twelve took place 6 m above the ground
or higher, but these data are strongly
biased toward observations close to the
ground, since lizards were more easily
spotted and followed at low heights. Be-
cause valencienni takes prey at short dis-
tances with high success, once prey is
spotted, the species is, in the terminology
of Schoener (1971) a searcher (as op-
posed to a percher, such as lineatopus).
The finding that valencienni takes food
from throughout the tree is consistent
with the data of Rand (1967a) and
Schoener and Schoener (1971), namely,
that valencienni has the most variable
“perch” height of a Jamaican anole, i.e.,
it is sighted from the ground to the tree-
tops.

In behavior, valencienni is clearly a
searching species, in the sense that most
individuals can be observed most of the
time searching for prey. They do so in
five different places: 1) on the ground; 2)
on leaves; 3) on the trunk of a tree and its
limbs; 4) in holes; and 5) on epiphytes
growing on trees, epiphytes such as
Tillandsia recurvata. The searching
behavior itself consists of moving very
slowly while frequently cocking the head
so as to look at the substrate. The lizards
are especially likely to search irregular
places in the bark, large holes in a tree
and places where limbs have broken off.
Likewise, they search out all epiphytes
very carefully and on broad-leaved trees
such as mangos will go out onto leaves to
search their surface.

Typical searching behavior of two dif-

ANOLIS VALENCIENNI +: Hicks and Trivers

575

ferent individuals are presented here,
condensed from Trivers’ field notebook
(measurements converted to metric
equivalent).

While drawing close to a tree to ob-
serve two lineatopus at 8:30 am on
September 16, 1968, Trivers almost step-
ped on a valencienni (assumed to be a
female) who was on the ground a foot
from the tree. Trivers withdrew and
watched through binoculars:

8:33 lizard jumps onto tree and climbs straight
up to 5m.

8:35 stops at series of bromeliads (Tillandsia
recurvata)

8:38 moves in among them, searching

8:40 carefully, slowly climbing through a brome-
liad

8:43 has moved a foot higher since entering
bromeliads

8:46 stealthily moves another three inches

8:49 another three inches, then suddenly moves
45 cm very quickly, as if it has spotted something. It
then darts, snaps something off a bromeliad and
eats.

8:50 moves several inches through bromeliad
and stops

8:55 now 6 m up, turns to move out a horizontal
branch which is about 7.6 cm thick

9:05 upside down ina large bromeliad, searching

9:10 same place

9:18 still upside down, searching, sometimes on
a bromeliad, sometimes on the branch itself

9:25 upside down on a bromeliad when it seems
to spot something on a neighboring bromeliad, also
upside down. I too spot something on the second
bromeliad. Starts to dart the 5 cm to the neighboring
bromeliad but—as if forgetting it is upside down—it
steps into thin air and falls 6 m to the ground. It
appears to be uninjured.

In the second case, an individual was
watched continuously for three hours and
forty minutes. The individual was almost
certainly a female as judged by her
appearance, the size and coloration of her
dewlap, and by the nonaggressive way
she interacted with an adult male who
appeared to be courting her. During the
time she was observed, she moved from
the foot of the tree to the foliage 11 m up
and back down to the ground. She fed
three times (once at ground level, twice
at 11 m). All three prey items were
spotted at distances of 15 cm or less. She

576

used her dewlap on six different occa-
sions, on three of which the adult male
was within a couple of meters of her. Ob-
servations on this animal are as follows: —
At 9:40 on September 6, 1968 an adult
valencienni was spotted upside down on
a horizontal branch 5 m above the
ground. Assumed to be a female, the
lizard was searching an irregular area in
the bark. She moved on to the trunk, and
in the next ten minutes moved down to
1 meter above the ground. At 9:56 she
moved down to 0.5 m, then down to 7.5
cm. She paused, dashed about 15 cm out
onto the ground, caught a prey item, re-
turned to 15 cm up the tree and ate what
appeared to be a small, white moth.

10:13 a cat appears, circles the tree rapidly, and
scares the female up to 5 m where she stops to
search a warty, cracked area of the bark

10:10 1 m higher; a male who was also scared up
by the cat is 1 m higher than the female, facing away
from her and dewlapping repeatedly

10:14 male 1.5 m higher than female and still
dewlapping; she jerks her head as if about to dew-
lap but no dewlap

10:23 she is now 10 m up going out a 2.5 cm thick
branch

10:27 11 m up at end of branch, looking over a
whorl of leaves

10:35 goes out another twig which ends in a
whorl of over 20 leaves. Pulls itself into leaves and
searches visually for over a minute, suddenly darts
straight up 7.5 cm and catches a large insect, which
she chews

10:40-10:57 searches other leaves in area

11:00 facing down on a branch 0.3 m below the
leaves; male is 1 m above her dewlapping

11:04 female just dewlapped (small dewlap,
many scales)

11:08 both male and female are 1.5 m lower.
Male is dewlapping. Male moves rapidly down the
branch, past where the female is and further down

11:13 female is 5m up slowly searching the bark;
for the next five minutes she does not move

11:28 she is 0.3 m lower, facing down, head
raised, often looking around (at what?)

11:30 still looking around, moves sideways and
dewlaps again, then moves to another branch

11:35 5.4 m up, slowly moving up

11:38 6 m up, moving up. At 11:42 she dewlaps
several times (again, small dewlap, much white
showing)

For the next 45 minutes she spent most
of her time searching several clumps of

Advances in Herpetology and Evolutionary Biology

leaves (without success). At one point, a
large wasp landed within 8 cm of her
without eliciting any response. At 11:56
she dewlapped twice; no other lizard was
visible near her. At 12:20 she fed on a
small insect which was about 2 cm back
of her head when she spotted it (10 m
above ground on a twig). At 12:26 she
dewlapped again and within a minute the
large male was spotted about one meter
above her, dewlapping. He moved to
within a few cm of her and dewlapped
repeatedly. Then he ran by her. For the
next twenty minutes she and he descen-
ded to 5 m up, and the entire time he was
within a meter of her, often dewlapping.
She dewlapped at 12:53, and he moved
rapidly up the tree away from her. For
the next half hour she descended slowly
to a height of less than a meter. Observa-
tions were discontinued at 1:25.

FEMALE-FEMALE INTERACTIONS

Two or more marked adult females
were seen within centimeters of each
other over 100 times during all visits to
the two study areas, and no aggressive
interaction was ever observed. Marked
females were never seen to display to
each other, nor to make any physical con-
tact nor to show any clear avoidance be-
havior. On four occasions unmarked
adults who appeared to be females dis-
played toward other adults assumed to be
females. In three cases, a presumptive
female dewlapped at a second presump-
tive female which was several centi-
meters away. There was no response nor
any follow-up behavior. It is very likely
that the dewlapping individual in these
three cases was a female, but it is not cer-
tain that the recipient was also a female.
In the fourth case a presumptive female
bit at her own reflection in a mirror. (All
other females ignored their mirror im-
ages.)

In virtually all interactions between
marked females, the relations appeared
to be neutral, but in certain situations,
described below, females may have been

positively attracted to each other (or to
some common stimulus). The neutral
relations between adult female valen-
cienni contrast strongly with the aggres-
sive interactions that characterize
female-female relations in A. lineatopus
(Rand, 1967b), A. garmani (Trivers,
1976), A. opalinus (Jenssen, 1973) and A.
grahami (Trivers, unpublished data). To
illustrate the nonaggressive nature of
valencienni female relations, we give
here two examples.

1) On July 22, 1969, three adult females
(snout-vent lengths: 60, 65, and 66 mm)
were seen within centimeters of each
other at a hole about 5 m up a mango tree.
The three arrived at the hole within fif-
teen minutes of each other from three dif-
ferent directions, each stayed several
minutes and left independently of the
others. Two other adult valencienni,
which were unmarked and assumed to be
females, passed within centimeters of
two of the marked females. At no time did
any of the individuals clearly alter be-
havior in response to the others. The hole
which attracted the females was one near
which females were often seen and one
which contained valencienni eggs. It is
not known whether the sight of a female
near such a hole attracts other females.

2) Females (like males) are reluctant
to leave bushes or trees for open ground.
They appear to be much less reluctant if
another individual has preceded them.
On August 6, 1969, at 2:15 pm an un-
marked adult valencienni left a mango
tree to climb among some rocks at the
base of the tree. Within about 20 minutes,
four other unmarked adults left the tree
to search among the same rocks. At 2:45,
all were caught. Two were then 4 m from
the base of the tree but within centi-
meters of each other. Both were 66 mm
females. Two others closer to the tree and
within one meter of each other were a 64
mm female and a 62 mm female. And,
finally, the individual nearest the tree
was a 62 mm male. All appeared to be
searching for food, and at least one fed
successfully while among the rocks. No

ANOLIS VALENCIENNI + Hicks and Trivers

577

interactions were apparent. The pres-
ence of an adult male is not typical, but
other such cases have been observed:
one male within a meter of one or more
females, with no obvious interaction. For
example, at 10:30 am July 6, 1969, four
adults were caught while searching for
food on the trunk of a royal poincianna
tree, all within 1 m of the ground and of
each other. Three were females (59, 60,
and 65 mm), and one was a male (68 mm).

The above example is only atypical in
that more than two individuals were in-
volved. In the majority of female-female
interactions, one female passes close to
another while both are searching for
food. It would be interesting to know if
females tend to avoid searching through
areas recently searched by others, but no
good evidence exists one way or the
other. Certainly females do not show
strong avoidance, and we have often seen
one female search the general area re-
cently searched by another. Females may
be actively attracted to each other 1) at
holes containing eggs and 2) at the base
of trees, but this is merely an impression
based on the frequency with which we
observed more than one female together
at such places.

DAILY FEMALE MOVEMENTS AND
DISPERSAL

Since females do not interact aggres-
sively and since there is no evidence
they avoid each other, it is of interest to
know how extensively their home ranges
overlap. The best evidence for degree of
overlap comes from three large contigu-
ous trees whose occupants were watched
during the summer of 1969. (Data from
other trees give similar, but much less
detailed, results.) The trees were two
mangos connected together by a royal
poincianna. Each tree stood about 13 m
tall and had extensive foliage. The three
trees were connected to each other in the
foliage but to no other trees. It was pos-
sible, then, for a female to wander

378

throughout the three trees. Distances be-
tween the trees and the number of adult
females captured on each are given in
Table 3.

Between June 16 and September 6,
1969, 43 adult females were caught and
marked on the three trees. The great
majority of these (37) were marked by
August 6. Since the six females caught
after August 6 had little opportunity to be
resighted (only one of them was, in fact,
resighted), the analysis here is limited to
the 37 marked by August 6. Of these 37,
29 were resighted at least once for a total
of 69 resightings, or almost two resight-
ings per individual marked on or before
August 6. Of these 69 resightings, 64
found the lizard on one of the three con-
nected trees. (The other five resightings
involved nearby trees and they are dis-
cussed below.)

The 64 resightings are analyzed ac-
cording to each lizard’s tree of origin in
Table 3. Just as more females were cap-
tured on mango-1| than on the other two
trees, more females were resighted on
that tree as well. This may partly indicate
a preference by the lizards for mango-1,
but it may result entirely from the greater
ease with which lizards could be re-
sighted and caught on mango-1. The tree
had a large hole containing over 50 valen-
cienni eggs located only one meter above
the ground, and females were often
sighted near this hole or near a second
egg hole 4 m above ground. (The poin-
cianna was not known to contain an egg

TABLE 3. THE NUMBER OF TIMES 37 FEMALES WERE RE-
SIGHTED ON THREE TREES AS A FUNCTION OF THE TREE
ON WHICH EACH WAS INITIALLY SIGHTED.

M-2 POINCI M-1 TOTAL

TREES M-1 5 3 36 44
FIRST (21) 2
CAUGHT Poinci 2, 2 3) $)
ON (No. (5)
inparen- M-2 3 2 6 11

theses) (11)

TOTAL 10 9 47

Advances in Herpetology and Evolutionary Biology

hole and mango-2 was only known to
have one, 6 m above ground.)

Of the 21 lizards caught on mango-1, 18
were resighted and 82% of the resight-
ings found the females on that same
mango. Of the five females caught on the
poincianna, all were resighted, but only
22% of the resightings found the lizards
on the poincianna. Of the eleven females
caught on mango-2, only six were re-
sighted and only 27% of these resightings
occurred on that mango. By treating each
resighting as an independent event, it is
possible to test for heterogeneity in the
data. The data almost show a significant
tendency for lizards to be resighted more
often on the tree of their capture than
expected by chance (x? = 8.13; 0.05 < p
< 0.10). What is clear is that the females
wander very widely. Only one female (60
mm) showed a marked tendency to re-
main on the tree of initial capture
(mango-1): during a three month period,
she was resighted seven times, all of
them on mango-1. If data on her are re-
moved from Table 3, no trend toward
localized movement is apparent.

Movement from one tree to another can
be rapid, taking place in less than two
hours. Four cases of females moving
between two of the three trees within a
day are presented here. 1) On July 8 a 60
mm female was captured and released on
mango-2. An hour and a half later, she
was spotted at the base of the poincianna.
The following day she was seen again on
the poincianna, but two days later than
this she was seen on mango-1. On July 14
she was again seen on mango-l. 2) On
July 18 a 64 mm female was caught at
10:00 am on mango-! and was resighted
that afternoon at 4:15 at the base of
mango-2, 29 m away. She presumably
reached mango-2 by way of the poincian-
na. 3) On July 21 a 65 mm female was
caught on the poincianna. The following
day she was seen at the base of mango-1.
4) On July 24 a 61 mm female was caught
on the poincianna and a day later was
seen on mango-2.

It might be supposed that these rapid,

long-range movements are a response to
the trauma of capture, but we doubt this
for several reasons. 1) Similar movements
have been recorded between resightings
(where no capture was involved). For
example, a 62 mm female was resighted
on August 5 on the poincianna and two
days later was resighted again on mango-
1 (where it had originally been captured).
Likewise, in the data above, the 60 mm
female moved to a new tree on the day of
capture, where she remained for at least a
day, before moving on to the third tree. 2)
By following individual unmarked
females for several hours through binoc-
ulars, two such long-range movements
were observed on the same trees in 1968,
although no lizards were captured during
that visit. 3) In the Maryfield study site,
from which these data on movements
come, lizards were caught only at ground
level, so that a lizard could easily evade
capture by remaining above 4 m yet the
movements described involve moving to
a height of 10 or more meters, crossing toa
neighboring tree and sometimes return-
ing to ground level. Within a half hour of
capture, a female’s behavior could not be
distinguished from that of an uncaptured
female; that is, a female recently captured
rapidly returned to the slow, methodical
search for food characteristic of the spe-
cies. 4) Such long movements in response
to capture have not been observed in
highly territorial species (A. lineatopus,
Rand, 1967b; A. garmani, Trivers, 1976;
A. grahami, Trivers and Hicks, unpub-
lished data).

In addition to the 64 resightings re-
corded in Table 3, five additional resight-
ings were recorded in trees near (but not
connected in the foliage to) mango-1. In
one case a female was captured on
mango-2 and resighted only once, nearly
two months later, on a mango 30 m from
mango-1. This may have been a perma-
nent move. In the second case, a female
moved a short distance (about 12 m)
across open ground to a neighboring tree
and returned to her tree of capture
(mango-1) all within two weeks. The

ANOLIS VALENCIENNI «+ Hicks and Trivers

579

third case involved a 66 mm female who
was caught at the base of mango-2 on
August 6. Ten days later she was re-
sighted 56 m away on an avocado tree
near mango-l. Thirteen days after this
she was again seen on mango-2, but
twelve days later (on September 10) she
was seen back on the avocado tree (as she
was on September 13). The female pre-
sumably moved from mango-2 onto the
poincianna, from there onto mango-1 and
then moved 10 m across open ground to
some bushes which connected with the
avocado tree. Although it is much more
likely that the female crossed by way of
the poincianna and mango-! than that she
moved 55 m across open ground, such
inferential data have not been included
in Table 3. (Similarly, one can safely
infer presence on the poincianna when a
lizard is sighted on mango-1 and then on
mango-2, or vice-versa, but five such
inferences were not included in Table 3).

Even these examples do not give an
adequate picture of valencienni female
movements. Observations of female
movements within a given tree indicate
that all females spotted low on a tree soon
move high into the foliage. Observations
of the same female on successive days
reveal no tendency for a female to
emerge from the same part of the foliage
nor to return to the same part. Although it
is likely that more detailed data will re-
veal some localization, it is clear that the
overlap of female home ranges within a
tree, or within contiguous trees, is enor-
mous.

The three trees also contained at least
12 adult males. This is certainly a serious
underestimate, since many unmarked
adult males were seen but never cap-
tured, primarily because males stayed
considerably higher in the trees than did
the females. Based on several lines of
evidence reviewed in the “Sex Ratio”
section, we estimate that the trees con-
tained at least 25 adult males. Of the 12
actually caught, eight were captured on
or before August 6, but only one of these
was resighted, so that almost nothing is

380

known about these males. It is clear,
however, that even if males defended
nonoverlapping territories, no female
would live exclusively within one male’s
territory and most females would wander
(in the space of about two weeks) through
virtually every male territory. In the
space of a single day a female would be
expected to encounter at least five adult
males.

In other Jamaican Anolis, the larger
size of the adult male is associated with a
larger territory or home range than seen
in the adult female. In A. valencienni, we
may expect that this relationship will not
hold. Females are not territorial, while
males, as we shall see, are repelled by
each other. Thus, females may wander
more widely than do males.

In the summer of 1973, we had a suffi-
cient number of resightings to test this
possibility. Lizards were watched for six
weeks. The maximum distance between
resightings was calculated for each indi-
vidual who was observed more than one
time. Forty-seven adult females were re-
sighted an average of 3.1 times each,
while 30 adult males were resighted an
average of 4.7 times. Although males
were sighted more often than were fe-
males, their average maximum distance
between resightings, 26.8 m, was slightly
shorter than that for females, 28.7 m. If
we imagine that each resighting contri-
buted equally to the maximum distance
(which is unlikely), then for each resight-
ing, males moved 4.7 m, while females
moved 7 m. What seems clear is that
adult females move somewhat greater
distances during six weeks time than do
adult males. These data certainly include
some examples of dispersal to new areas,
but these instances are few. Most of the
data described movements typically
made during a period of two or three days
wandering.

Individual A. valencienni take very un-
willingly to the ground. They are slower
on the ground than any other Jamaican
anole and are found there less often. Like
other Jamaican anoles, they descend to

Advances in Herpetology and Evolutionary Biology

the ground to feed, but we have never
seen an individual move more than 3 m
from the base of its tree in search of food.
We have only twice seen individuals on
the ground who were not feeding. One
had fallen accidentally out of its tree, and
the other was in the process of dispersing
22 m to a new tree. Data on 257 marked
individuals indicates that some individ-
uals do move as much as 35 m over
ground to reach a new tree. Of the 257
marked individuals, 49 were recaptured
and, of these, eight had dispersed across
ground to new trees.

Comparing those who dispersed with
those who did not shows that those who
did not disperse grew more than those
who did. Of the 41 individuals who did
not disperse, 34 showed measurable in-
creases in size (2 mm or more) in periods
ranging from one half month to two
months, whereas only two of the nine
who dispersed showed significant in-
creases in size in similar time periods
(p = 0.01). The average growth rate
for nondispersers was 2.34 mm/30 days
and for dispersers it was 1.06 mm/30 days
(p = 0.001).

When growth rate is plotted as a func-
tion of female size, it is seen that dispers-
ing individuals tend to have lower
growth rates for two reasons. Dispersing
females tend to be larger and to show
smaller growth rates for a given size.
Neither effect is in itself significant.

Each effect could be explained as fol-
lows. Larger females may be less vulner-
able to predation and able to traverse
more quickly the open space. Females
with low growth rates may be selected to
take a greater risk in finding a new feed-
ing area. Of course, females who have
low growth rates and disperse, may do so
because dispersal causes a subsequent
low growth rate in a new and unfamiliar
environment. But several lines of evi-
dence support the former interpretation.
1) On August 18, 1969, Trivers predicted
that either no. 76 or no. 134 or both would
soon disperse, as neither had grown in
more than a month. No. 134 was never

found again, but no. 76 was next seen 11
days later on a tree 24 m across open
ground from its original one. 2) Four of
the eight dispersals were off of one tree
which supported a large population of
which those remaining behind showed a
lower than average growth rate (1.67 mm/
30 days). Six dispersals went to a mango,
which had been almost empty until dis-
persing individuals reached it. These
facts suggest that the dispersers may be
leaving overcrowded trees for less pop-
ulated ones. 3) The only disperser caught
at least two weeks after dispersing
showed no significant growth while dis-
persing, but a 3 mm growth during the six
weeks after dispersal.

COPULATORY BEHAVIOR

Female and male valencienni appear to
copulate wherever they find themselves.
We cannot think of a single place (other
than inside holes) where we have seen
valencienni without also seeing them
copulate there. They copulate along the
thin outer branches of a bush, anywhere
along the trunk of a tree, facing up or
down, one centimeter off the ground to
10 m up a tree. We have even seen them
copulating on the outer leaves of a mango
tree 6 m off the ground. There is no evi-
dence that females exert choice over
place of copulation, as they appear to in
A. garmani in which individuals almost
invariably copulate face down on the
exposed trunk of a tree (Trivers, 1976).

The most striking feature of female
copulatory behavior is that females will
copulate with more than one male on the
same day and with the same male (and
different ones) on successive days. Since
it is unlikely that females are producing
eggs at a rate of two or more per day,
sperm competition is expected to be an
important determinant of male reproduc-
tive success in valencienni. The key
fact—that females appear to copulate
more than once per egg fertilized—is
supported by a number of observations,
summarized here.

ANOLIS VALENCIENNI + Hicks and Trivers

581

In the summer of 1969 only four copu-
lations were observed involving marked
females. Two of the copulations were
performed on the same day, by the same
female, with two different males. The
female was no. 85 (58 mm). She was
watched continuously from 10:35 am to
1:15 pm on July 5. For two hours she
searched for food at the base of her tree,
often in the company of two other marked
females. At 12:25 male no. 66 (71 mm)
rushed down from a branch and caught
no. 85 on a small branch, where they
mated. Shortly thereafter, male no. 72 (75
mm) appeared. He dewlapped at the
smaller male, no. 66, who jumped on to a
neighboring branch and moved up into
the foliage. Although male no. 72 and
female no. 85 began moving toward each
other, no. 72 suddenly jumped on to a
nearby branch and moved rapidly up the
tree. Within 15 minutes he retumed to
within 30 cm of no. 85, but at 1:05 male
no. 95 (81 mm) rushed down the tree and
caught no. 85. Male no. 72 crouched and
dewlapped at no. 95 but then jumped out
of the way. No. 95 and no. 85 copulated,
for her the second copulation in 40 mi-
nutes.

During the July-August, 1973 visit to
the Southfield study site a special effort
was made to observe copulations, and 10
females were seen to copulate more than
once, usually twice within a week and
sometimes twice within 24 hours. 1) Fe-
male no. 171 (60 mm) copulated on July
27 first with male no. 78 (83 mm) and
then with male no. 134 (84 mm) and on
the following day again with male no.
134. The first copulation took place on
one tree, and the next two on a pimento
(Pimenta officianalis) separated from the
first tree by open space. 2) Female no. 24
(57 mm) copulated with male no. 45 on
July 27 and on the following day with
male no. 165 (73 mm). The copulations
took place on contiguous, small trees.
3) Female no. 151 (60 mm) copulated on
July 21 with male no. 88 (76 mm) and the
following day copulated again with the
same male on the same tree. 4) Female

382

no. LF (68 mm) copulated three days
apart on the same tree with the same
male (no. 134: 84 mm). 5) Two females
copulated four days apart with different
males on nearby trees. 6) Three more
females accounted for eight more copu-
lations, each separated by between six
and nine days from a neighboring copu-
lation. 7) Finally, one 69 mm female cop-
ulated 14 days apart on different trees
with different males. In all, there were 23
multiple copulations out of a sample of 50
copulations (in which the female was
marked).

How is the size of the female associ-
ated with her tendency to recopulate?
The frequency with which females seen
to copulate once copulated additional
times was plotted as a function of size of
female. Above 59 mm females tended to
recopulate less the larger they became.
In addition, the larger a female becomes
the greater the span of time separating
recopulations. For example, the four re-
copulations performed within 24 hours of
an earlier copulation were all performed
by females 60 mm in size or smaller.
Since it is very unlikely that a female
matures an egg more often than once a
week, these additional copulations
should be unnecessary to the female. It is
tempting to suggest that the smaller a
female is the less control she has over
whether she copulates. Data on the per-
centage of females who copulate at least
once (as a function of size) make it un-
likely that smaller females are maturing
eggs faster than large females (see Figs. 2
and 3). If these data are correct, then
sperm competition is expected to be
more intense in small females while
female choice is expected to be a more
important determinant of male reproduc-
tive success for large females.

We have timed copulations in nature
and in our female choice experiments
(see below), and these show remarkable
consistency, averaging about 2 minutes,
rarely less than 1.5 minutes or more than
2.5 minutes. There is no relationship
between size of the lizards and length of
copulations.

Advances in Herpetology and Evolutionary Biology

45 FEMALES

it

FREQUENCY OF COPULATION
ie)
|

i 2/10 | 3/22 | 4720 | 4/13
Oss 60 65 70=
SIZE (MM)
4 MALES
Zz
ie)
&
3.35
2 ae
8
u 2
>
oO
i
3 |
2 1/4 \/7 1/7 1/6 4/\2 6/22 1/6
Oss 60 65 80 85 =

70 15
SIZE (MM)

Figure 2. Frequency of copulation as a function of size
(snout-vent length in mm) for females and males.
These data were gathered during two weeks in June,
1971. Each fraction gives the number of copulations
observed by members of the size class, divided by the
number of individuals in that class.

Copulations sometimes appear to be
somewhat aggressive. Males often bite
females on the skinfold on the top of the
neck. This may be done at the time of
intromission, and the grip may be main-
tained during the entire copulation.
Sometimes copulations are preceded by a
very active stage, in which the male and
female run a third of a meter or more
while he attempts to cover her. These
have the appearance of aggressive
chases, in which the male adds an ele-
ment of force to the persuasion.

Beyond this, in August 1973, Hicks
observed an unusual pattern in one small
adult male, which he entered in his notes
under the title, “The bizarre behavior of
Male no. 115, or rape and attempted forc-
ible entry among the lower animals.” No.
115 was a 57 mm male. On the 9th of
August he copulated with a 50 mm fe-
male while they locked jaws. They con-

47 FEMALES
5
QO 3
ee
—)
=)
o)
© 2
LL
re)
6
Zz 4
Z.
=)
Fs}
WwW 10/27 13/35 10/21
jag
LL
50 55 60 65 70 =
SIZE (MM)
Zz
6
& 3
=)
=)
foe
O
O 2
LL
o)
>
O
a 3
2 |
=)
G
LWW 2/12 6/18 5/23 6/20
ag
LL
55 60 65 70 75 80 >

SIZE (MM)

Figure 3. Frequency of copulation as a function of size
(snout-vent length in mm) in females and males. Data
for females include all copulations seen except those in
June 1971. Data for males include all copulations
except those seen in June 1971 and those seen while
individual males were observed for three-hour periods.

tinued to lock jaws and fight after copula-
tion was complete. Three days later,
Hicks observed courting-fighting be-
tween the same male and a 66 mm fe-
male. No. 115 dewlapped at the female,
but when he approached her closely, he
would snap at her with his jaws. She
chased him, he chased her. Chasing and
dewlapping occurred around the trunk
and up and down the tree. Finally, she
half hid in a hole at the top of a dead
branch. He approached within centi-
meters several times, once coming up
right behind her. She did not move, and
he seemed unable to do anything. Final-
ly, he departed.

ANOLIS VALENCIENNI + Hicks and Trivers

5983

FEMALE CHOICE

In the highly territorial A. garmani,
male courtship followed by female rejec-
tion has almost never been observed
(Trivers, 1976). In garmani, females
appear to choose the place to copulate
and signal their accessibility by going to
this place. In addition, the territorial sys-
tem strongly reduces the opportunities
for female choice. By contrast, female
valencienni are exposed to many adult
males each day and we have often ob-
served male courtship and female rejec-
tion. In such a situation, a dewlapping
male will typically approach a female
who will move away from him. If he
pursues her, she may run away from him,
or she may dewlap back at him and
attempt to keep her rear oriented away
from him.

In the summer of 1973, we observed
eleven cases of courtship and rejection
involving marked valencienni. In four
cases, the female who rejected the male
was observed to copulate with that same
male at some other time. In one case, she
copulated with a male a day after reject-
ing him. In a second case, she copulated
with a male two days after rejecting him.
In a third case, a male courted the same
female he had copulated with an hour
and a half earlier, and she appeared dis-
interested in his second advance. Final-
ly, a female ran from a courting male who
had 16 days earlier copulated with her.
These cases suggest that the female may
be rejecting the time or the place as much
as the male himself or that the female
may succeed in rejecting some but not all
of the advances of a particular male.

It seems certain that some cases of
female rejection resulted because we
were watching the lizards at very close
quarters, but unambiguous cases of fe-
male rejection were also observed at con-
siderable distances.

The most dramatic case of female
avoidance took place on July 27, 1973. An
81 mm male was observed from 9:30 until
12:30 during which time he courted two
females, both of whom appeared to reject

584

him. The male first courted a 61 mm fe-
male who ran from him. He did not pursue
her. He then courted a 65 mm female who
also ran from him. He chased the second
female who continued to run from him.
Finally, forced out on the end of a branch,
she leapt from a height of 6 m to the
ground in order to escape his attentions.
He did not leap after her.

Our data on female rejection are too
scanty to show whether there is any rela-
tionship between female size and ten-
dency to reject, or between male size and
tendency to be rejected, or between rela-
tive size of male and female and female
tendency to reject.

To find out what criteria females use in
choosing males, a series of experiments
were run during the summer of 1976. At
issue was whether females preferred
males who differed by size or recent rate
of growth. To ascertain the latter, nearly
300 lizards were captured between
February and May. These individuals
were then recaptured in June and July
and used in choice experiments. Thus,
for all individuals used size was known,
and for many of these individuals, recent
rate of growth to achieve that size was
also known.

Initially the plan was to house each
adult male in a 40-liter fish tank. All sur-
faces of the fish tank were painted
opaque except one of the two large sur-
faces. The two tanks were placed next to
each other, so that the males could not
see each other, but so that a single female
in a neighboring 80-liter fish tank could
see each of the two males. The plan was
then to measure the relative amount of
time the female spent near each male. In
actual fact, the fish tanks appeared far too
confining on the individuals of both
sexes. Most females spent the first hour
trying to escape from the tank, moving
from corner-to-comer. When they real-
ized that they were not going to escape,
they settled down in one comer and
typically stayed there for dozens of mi-
nutes on end, sometimes for as long as
two hours. It was difficult to convince

Advances in Herpetology and Evolutionary Biology

ourselves that females retained any
interest in males under these conditions
of confinement.

Two other experiments were attempt-
ed which were equally unsuccessful. In
the first, females were released from a tin
can onto a 3 m stick which separated at its
end onto two sticks moving apart at a 90°
angle. On each of these two sticks were
tethered an adult male. Most females
moved very slowly along the first 3 m,
then chose which tum to make and ran
very rapidly undemeath the tethered
male. Females often seemed to choose
the smaller or less active of the two
males, as if considering the male only as
an obstacle to her own escape.

In the other kind of experiment, a
series of eight small bushes were ar-
ranged in a row so that the vegetation of
each bush connected its neighbors. On
several of the bushes we tethered a
single adult male per bush. In this exper-
iment, females were provided a more
natural and safe environment within
which to choose a male. None of the
females chose any of the tethered males.
Most easily avoided them, and the males
were rarely able to act in a natural way.
Most moved as far as the tether would
permit and strained against it. Some lost
their grip on the substratum and hung
from their tether. These had to be placed
back on the bushes. In short, the females
were exposed to the novel opportunity of
meeting a series of restrained males, but
this was of no use in discerning female
choice.

In order to gain some value from our
investment in capturing and recapturing
the lizards, 108 experiments were run of
the following sort. A female was released
onto a bush which was placed in a large
cleared area. Five minutes later an adult
male was released onto the same bush.
The two were permitted to remain to-
gether for an hour, or until they copu-
lated, whichever came first. These exper-
iments produced 36 copulations. It was
clear from the behavior of the lizards that
these experiments did not necessarily

measure a female choice alone. Since the
bush on which the lizards found them-
selves was isolated from neighboring
vegetation, the female was reluctant to
leave the bush, and the male had, in ef-
fect, a partly imprisoned female to court.
Females often resisted the advances of
males by moving away from them, by
squirreling behind a limb so as to be out
of sight, and by running rapidly when
pursued, but females only rarely left the
bushes under male duress. Larger males
copulated significantly more often than
smaller males, and larger females copu-
lated more often than smaller females.
Indeed, frequency of copulation for indi-
viduals of both sex varied as a function of
size in the same way as observed in na-
ture (Figs. 2, 3). The data relating size to
frequency of copulation are as follows.
For males < 64 mm, 0.17; for successive-
ly larger 5 mm categories: 0.25, 0.31,
0.36, 0.38, and 0.42. For females <= 59
mm, 0.24; for successively longer 5 mm
categories: 0.30, 0.37, 0.38, and 0.40.

Most copulations took place in the
second half-hour of the experiment, but
there was considerable variation. This
allowed us to see whether size of the
male, size of the female, or relative size
of the two had any influence on the time
elapsed until copulation. Relative size
had no effect, but larger males copulated
significantly quicker and larger females
copulated quicker, though not signifi-
cantly so (see Fig. 4). These data do not
support the possibility that relatively
larger males were forcing copulations.
Instead, larger females appear more will-
ing to copulate and do so more often,
while the same is true of males.

For those wishing to pursue the prob-
lem, two suggestions are made. One is to
build a large natural enclosure into
which are released a number of adults of
both sexes. A cage 5 m tall, 4 m wide, and
2 m deep enclosing small trees and
bushes might permit a population of 10 or
20 adults to be observed simultaneously.
Measurements could be taken of be-
haviors that correlate with female rejec-

ANOLIS VALENCIENNI + Hicks and Trivers

585

tion in order to see whether male char-
acteristics can be correlated with these
attempts at rejections. This kind of ar-
rangement would have the virtue of en-
closing the adults in a natural setting, but
would have the disadvantage that fe-
males are not exposed to clear, binary
choices. Thus it would be preferable to
house adult females in naturalistic cages
until their behavior had settled down and
then to place two contiguous male cages
next to the female cage for a period of
time such as an hour, to see if the female
will actively demonstrate a preference
for remaining near one of the two males.
Her behavior can be monitored to see
whether this choice represents avoidance
of the one male or choice of the other.

D
oO

2
S
Zz
Q
& 40
=!
ra
fe)
Oo
a
= 20
Zz
5
Ww
2
|
65 70 75 80 85
SIZE (MM)

607 Females seh 8 2 ge y
e

TIME UNTIL COPULATION (MIN)

SIZE (MM)

Figure 4. Time until copulations (minutes) as a func-
tion of size (Snout-vent length in mm) for males and fe-
males during the mating experiments of 1976. The
regression line fitting the male points is y = 97.02
—0.754x which yields an r of 0.397 with 35 degrees of
freedom (p < 0.05). The regression fitting the female
points is y = 94.12 —0.8454x, r = 0.302 with 35 de-
grees of freedom (0.05 < p < 0.10).

586

THE USE OF THE FEMALE DEWLAP

It seems incongruous that female
valencienni should possess a dewlap.
The dewlap is normally found only on
the more aggressive sex, i.e., on males,
yet female valencienni are far less aggres-
sive than female grahami, garmani, and
lineatopus, all of whom lack the dewlap.
Furthermore, female valencienni rarely
use their dewlaps (in contrast to males of
their own and other species). To resolve
this anomoly, a special effort was made in
the summer of 1969 to record all cases in
which females dewlapped.

In addition to the case described under
“Feeding Behavior,” we twice observed
a female dewlapping back at a male
valencienni who approached the female
closely. Both times the female was as
large or larger than the male, and in both
cases the male moved off without copula-
ting. Except for three possible cases of a
female dewlapping at another female
(see “Female-Female Interactions’’), all
other observations show females dewlap-
ping at members of other species. Fe-
males were observed three times to
dewlap at adult male A. grahami, two of
whom appeared to be courting the fe-
males; that is, the male grahami were
dewlapping while approaching the fe-
male valencienni. In addition, female
valencienni were observed to dewlap at
1) an adult male lineatopus who was
about 30 cm away; at 2) a small female
grahami who was centimeters away and
who leapt over the dewlapping valen-
cienni and ran away; and at 3) one of us
when he approached the female to within
a distance of one meter to photograph
her.

These scanty observations do not
permit firm conclusions, but it appears
that females primarily employ their dew-
lap to discourage courting males, of their
own and closely related species.

MALE-MALE INTERACTIONS

A male is rarely seen in the presence of
another male, but when two are seen to-

Advances in Herpetology and Evolutionary Biology

gether, they are always fighting, display-
ing aggressively, or one is trying to hide
trom the other. In contrast to fights in A.
lineatopus (Rand, 1967; personal obser-
vation), A. grahami (personal observa-
tion) and A. garmani (Trivers, 1976),
valencienni fights appear tame. Males
commonly dewlap toward each other and
then one male darkens in color and
attempts to hide. Movements are slow
and chasing is rare. Of nine fights ob-
served in the summer of 1973 (all involv-
ing marked males), only one involved
actual contact, namely jaw-locking. Of
the approximately ten male-male en-
counters observed in previous visits,
none involved any body contact.

Data from eight fights observed in 1973
were reviewed. In only one of the six
fights in which the two males were with-
in 4 mm of each other in size did the
larger of the two win. In four fights, the
smaller appeared victorious (as judged by
which male darkened and attempted to
hide), and in one the outcome was un-
clear. Rand (1967) has shown in A. linea-
topus that when the smaller of two
lizards wins a fight it is usually because
the smaller is the territory holder and the
larger is an intruder.

Although male no. 92 was 4 mm larger
than male no. 45, he lost to no. 45 on two
separate occasions (separated by a week)
(see Fig. 5). Both fights took place within
the home range of no. 45, while no. 92
commonly occupied a neighboring home
range. Indeed, the only two days no. 92
was spotted within no. 45’s home range
were the days he was seen fighting with
no. 45 (see Fig. 8). It is worth emphasiz-
ing, however, that no. 92 copulated in no.
45’s home range after one of the fights
(see Fig. 5, black dot almost touching
triangle). (No. 92 was only seen to copu-
late once in his own home range.) This is
consistent with other observations of
male-male encounters: a fight does not
usually result in the ouster of one of the
two males; instead, one male retreats and
hides, remaining in the same area to feed
and, sometimes, to copulate.

Male no. 45 also appeared to vanquish

One meter
SS

A FIGHTS WITH #45
© COPULATIONS. OF #45
@ OTHER COPULATIONS

Figure 5. Male no. 45’s territory during July and
August, 1973. Pictured are the locations of fights (4)
and copulations of Male no. 45 (0), and copulations of
other males within no. 45’s territory (e).

another male who was larger than he was,
no. 114 (79 mm). No. 114 was first seen in
mid-July nearly 45 m from no. 45’s home
range; when sighted, no. 114 was copula-
ting. No. 114 was next seen on August 10
in male no. 45’s home range; he was seen
there August 11 and 12 also but not seen
anywhere afterwards. On August 11 he
appeared to lose a long fight, involving
jaw-locking, to male no. 45. Note that all
three of no. 45’s fights occurred at the
edges of his home range (Fig. 5).

For the two other fights in which a
smaller male appeared to vanquish a
larger one, the relevant data are even less
detailed. In both cases, the loser was
seen only once, namely on the day of the
fight. By contrast, one winner (male no.
172) was seen on seven different days on
a set of eight neighboring trees. The
second winner (male no. 107) was seen
on eleven different days on two sets of
trees. In both cases, the evidence sug-
gests that the loser was either an outsider
or sufficiently subordinate to be rarely
sighted. Given the frequency with which
many small males are resighted, we lean
toward the former interpretation.

ANOLIS VALENCIENNI - Hicks and Trivers

587

MALE HOME RANGE OVERLAP

In sharp contrast to more sexually
dimorphic Jamaican anoles (A. garmani,
Trivers, 1976; A. lineatopus, Rand, 1976;
A. grahami, Trivers and Hicks, unpub-
lished data), male home range overlap in
A. valencienni is substantial. No male in
either study area was known to occupy an
exclusive area and several males were
commonly seen within any one male’s
home range. To illustrate the extent of
the overlap, we present detailed data on
the home ranges of two males, one from
each study area.

The Home Range of Male No. 72 (74
mm). Between June 25 and July 12
(1969) male no. 72 was seen eleven sep-
arate times on one of two connected
small trees in the Maryfield study site.
These two trees were part of a clump of
four small connected trees. On six days,
no. 72 was watched for at least an hour
and seen to travel through a portion of
one, or both, trees. On June 26 he copu-
lated within his home range with female
no. 46 (57 mm). Yet during this period of
17 days, five other adult males were seen
within no. 72’s home range and two of
these males copulated with females resi-
dent within no. 72’s home range.

Between July 1 and July 12 male no. 63
(73 mm) was seen seven times, each time
on one of the two trees occupied also by
male no. 72. On two of these occasions,
no. 63 was watched for over two hours
and each time he wandered throughout
the two trees much as no. 72 did. On July
10 male no. 63 and male no. 72 encoun-
tered each other and no. 72 (although
larger) appeared to be subordinate. Male
no. 63 was about 4 m up on one of the
trees when he spotted no. 72 on a stick
60 cm from the base of the tree. No. 63
dewlapped toward no. 72, then rushed
down to a height of one meter and stood
for several minutes looking in the direc-
tion of no. 72, who, meanwhile, had
squirreled around on the stick so that the
stick was between him and no. 63. No. 63
slowly started back up the tree dewlap-
ping repeatedly. Fifteen minutes after

588

spotting no. 72, no. 63 had moved into the
canopy out of sight, at which point no. 72
righted himself on the stick.

No. 63 was observed to court a female
but was not seen to copulate before July
12. He alone of the males was seen after
July 12 (although the trees were searched
virtually daily). In the next two months
no. 63 was seen six times on the same two
trees, the last time of which he was copu-
lating with an unmarked female.

Male no. 66 (71 mm) appeared to occu-
py one of the two trees contiguous with
male no. 72’s home range. That is, three
of the four times male no. 66 was sighted,
he was on this neighboring tree. The
fourth time, he was seen on one of male
no. 72’s trees where he copulated with
female no. 85 (58 mm) (see “Female
Copulatory Behavior’). After the copula-
tion, he walked onto a branch male no. 72
was sitting on. No. 72 dewlapped at him
and no. 66 jumped to a neighboring
branch and moved rapidly away.

Two males (63 mm and 81 mm) were
seen only once each, and both were seen
on one of the two trees that comprised no.
72’s home range. The larger male copu-
lated with female no. 85 on July 5 and
appeared to frighten male no. 72 shortly
afterwards (see “Copulatory Behavior’).

In summary, for at least 16 days male
no. 72 occupied a home range which he
shared with a second male (no. 63) who,
although smaller, appeared to be domi-
nant over him. In addition, at least three
other adult males were observed within
this home range and two copulated with-
in it. All interactions observed between
the males were characterized by display
and avoidance. Unfortunately, the diffi-
culty of spotting valencienni in this study
area does not permit us to infer anything
from the absence of sightings after July
12. It is possible, for example, that after
July 12 only male no. 63 remained within
the two trees and that the high overlap
observed prior to that day was merely
temporary. Detailed data from the sum-
mer of 1973 render that an unlikely
assumption, however. Throughout the

Advances in Herpetology and Evolutionary Biology

summer of 1973 the same degree of over-
lap was observed in all cases for which
there were sufficient observations to plot
home ranges. An example is the home
range of male no. 45.

The Home Range of Male No. 45. Data
on male no. 45 give a detailed picture of
the overlap in male home ranges and the
partly correlated overlap in access to
females. Male no. 45 (77 mm) occupied a
large home range containing nine small
trees and many bushes and covering an
area of about 25 square m (see Fig. 5, 6).
First captured on July 7, 1973, he was
watched throughout the summer visit to
the Southfield study area and was last
seen on August 18. He was seen on 16
different days and on five of these days
his movements were followed for about
three hours each day. On three of these
three-hour watches, he wandered
throughout about a third of his home
range. On the other two three-hour
watches, he restricted his movements to a
corner of the home range.

Twenty-six adult females were seen at
one time or another within no. 45’s home

one meter
jo)

@ TREES

Figure 6. Male no. 45’s movements in his territory dur-
ing five days in which he was observed continuously
each day. e = the trees in his territory. Note that in
three days he travelled widely, while in two he moved
little.

range. These females were sighted a total
of 51 times within no. 45’s home range
and a total of 42 times outside his home
range. Sixteen of the 26 females were
seen to copulate, and they were seen to
copulate a total of 24 times (only seven of
which took place outside of no. 45’s home
range). Male no. 45 accounted for eight of
these copulations, each with a different
female (see Fig. 5). He twice copulated
with two females on the same day. He
copulated on five of the seven days in
which his movements were followed for
three hours. Indeed, six of his eight copu-
lations were seen during these 21 hours
of observation.

Five of the females with whom no. 45
copulated also copulated with other
males. The five copulated seven other
times with a total of five different males.
Of the 16 copulations with males other
than no. 45, nine took place inside no.
45’s home range. A comparison of the
location of these nine copulations with
the location of no. 45’s eight copulations
shows broad overlap (Fig. 5).

The degree to which females seen in a
male’s territory will also be seen outside
his territory is shown in Figure 7. The
home ranges of six females, who were
seen at least once in male no. 45’s terri-
tory and were sighted at least five times
in total, are shown in Figure 7. As we can
see, even if male no. 45 maintained ex-
clusive sexual access to these females
when they were in his territory, which he
does not, they would still wander widely
outside of his territory.

The variety of copulations a male may
enjoy in a short period of time are sug-
gested by observations of male no. 134
(see Table 4).

Male no. 45 shared his home range
with at least ten other adult males. The
overlap was considerable. For example,
on August 10 four adult males (in addi-
tion to no. 45 himself) were seen within
no. 45's home range. The day before,
three females copulated within no. 45’s
home range, only one of them with no.
45. The males overlapping with male no.

ANOLIS VALENCIENNI + Hicks and Trivers

589

45 can be divided into four categories.
Three were only sighted once or twice (if
twice, both sightings were within no. 45’s
home range). Of those sighted more than
twice, two lived entirely within no. 45’s
home range, two lived more within no.
45’s home range than outside it, and
three lived primarily outside no. 45’s
home range (see Fig. 8).

Males completely within no. 45’s home
range: 1) No. 96 (72 mm) was spotted on
four different days on six separate trees
within no. 45’s home range. No. 96 copu-
lated with three females (on different
days) only one of whom was seen to cop-
ulate with no. 45. 2) No. 115 (57 mm) was
spotted on ten different days and on four
different trees, all within no. 45’s home
range. No. 115 copulated once with a
female, and she was not seen to copulate
with no. 45.

Males primarily within no. 45’s home
range: 1) No. 164 (73 mm) was seen eight
different times, of which five sightings
were within no. 45’s home range and
three were at its edge. He copulated
three times—all of them in no. 45’s home
range. Only one of the females was also
seen to copulate with no. 45. He fought
once with male no. 115 (57 mm). 2) Male
no. 114 (79 mm) was seen on three
successive days in no. 45’s home range
(August 10-12). He was not observed to
copulate there. The only other time he

TABLE 4. HISTORY OF COPULATIONS FOR MALE NO. 134
(84 MM) INSUMMER, 1973 .*

SIZE OF
DATE FEMALE FEMALE (mm)
IL. 7/22 LF 68
2. 7/25 LF 68
3. 7/26 157 70
4. 7/27 171t 60
5. 7/27 Si 67
6. 7/28 171 60
he 7/29 182 70
8. 8/5 194 62

*A1l took place in the same pimiento tree.
*No. 171 copulated earlier that day with male no.
78 on a nearby achee tree.

590

Advances in Herpetology and Evolutionary Biology

|
one meter

Figure 7. The overlap of Male no. 45’s territory is pictured with the home ranges of six adult females observed at
least once in his territory, for each of which there were at least five sightings. The mean number of sightings per

female was 8.2.

was seen was nearly a month earlier, 43
m away, at which time he was seen cop-
ulating.

In summary, it is clear that adult males
overlap with each other tremendously,
both in space occupied (Fig. 8) and in
access to members of the opposite sex
(Figs. 5, 7). Commonly several males
copulate with the same female and
several males copulate with different
females in the same space. Fights are
short, mild and lead to withdrawal and
hiding. Some males, at least, are highly
variable in their movements, during
some periods moving considerable dis-

tances, during others remaining mostly
in one place (Fig. 6). Male-male compe-
tition must take an interesting and com-
plex form in A. valencienni that would
repay careful study.

FEMALE AGGRESSIVENESS IN A
SEARCHING SPECIES

Why are female valencienni not ag-
gressive? Imagine the contrary. Imagine
that each female defends a territory with-
in which to feed. Assume a female com-
pletes the search of her territory for food

ANOLIS VALENCIENNI « Hicks and Trivers

591

\ one meter
Se

Figure 8. The overlap of Male no. 45’s territory with those of seven other adult males seen at least once in his
territory and sighted altogether more than two times. The mean number of sightings for these seven males is 6.7.

in less than a day. She must now decide
whether to research her own territory or
to move into a neighbor's. To predict her
behavior we should know the answers to
three questions. 1) At what rate do re-
sources within her territory renew them-
selves? 2) Have her neighbors completed

the search of their own territories? 3) Is
she more likely to find items her neigh-
bors missed or items she herself missed?
What is the long-term variance in good-
ness of territory?

If resources renew themselves slowly
(for example, overnight) then there will

592

be no strong incentive for a female to
research her own territory. Indeed, under
such conditions, there should be a
stronger incentive to search her neigh-
bors’ territories, since 1) she knows she
has completely searched her own terri-
tory and she does not know that her
neighbors territories have also been
completely searched and since 2) it is
more likely that she will find items her
neighbors have overlooked than that she
will find items she herself has over-
looked (on the assumption that females
show individual variability in searching
behavior, e.g., in specific search images).
Since the same arguments apply to other
females, there is, under these assump-
tions, no incentive for females to guard
territories once they have been searched.

FEEDING BEHAVIOR AND CRYPTICITY

For a series of reasons that stem from
its feeding adaptation valencienni has
(we believe) been strongly selected for
crypticity. The fact that it is a searcher,
means that individuals must constantly
be on the move, thereby becoming more
visible to predators. At the same time,
however, they have not been selected for
speed once they sight prey, so they are
more vulnerable once sighted by a preda-
tor. In addition, they search by scanning
very closely the surface centimeters in
front of them. This narrows their field of
vision considerably compared to perch-
ing lizards, making it less likely they will
sight a predator early.

One could also argue that differential
visibility toward potential prey may have
selected for some of valencienni’s cryp-
ticity. If true, this should only apply to
the face and neck and not to the dorsal
surface of legs. It would also require that
potential prey be able to respond to
valencienni at distances of centimeters
and this is not unlikely. The use of cryp-
ticity in social encounters is discussed
below.

Our only actual data on predation are
four observations of small valencienni

Advances in Herpetology and Evolutionary Biology

being eaten by other anoles. Two were
hatchlings at Mayfield in August 1968,
eaten apparently within an hour of hatch-
ing, one by an adult male A. lineatopus,
the second by an adult male A. grahami.
The others were eaten at Southfield 1) on
July 21, 1973 by an 82 mm male valen-
cienni and 2) on August 8, 1973 by a 56
mm male valencienni. In the first case of
cannibalism the eaten lizard had been
marked and was probably a 43 mm male.
It was eaten head first. The second was
unmarked and was eaten tail first, the
head being bitten off without being swal-
lowed.

COMMUNAL EGGLAYING, SEASONAL
REPRODUCTION, AND PREDATION ON
YOUNG

Anolis valencienni females lay eggs
communally (Rand, 1976a). In Maryfield
we found five communal egg deposits
and in Southfield, one. These were locat-
ed in holes in tree trunks from ground
level to 6 m up. The number of eggs
ranged from only two to well over 30.
The largest was in a large hole one meter
up the trunk of a mango tree. Although
females often frequented this hole, and a
mirror was arranged to view their activi-
ties inside the hole, egg laying was never
observed. (Feeding attempts within the
hole were observed, but never directed
against the eggs.)

Anolis valencienni appears to breed
seasonally. Although all five egg deposits
contained eggs during the summertime,
none contained eggs in late December
and early January. Since the only egg
hatched in captivity required 53 days
from laying to hatching (Underwood and
Williams, 1959), we conclude that no
eggs are laid during at least November,
December, and early January, but the
periods of observation have been short
enough so that copulations could have
been overlooked by chance. (Except for
seven days in December 1969, and
eleven days in December 1973 to Jan-

uary 1974, valencienni has only been
studied in the summertime). However, in
two different summers of work at South-
field (June 1971, and July and August
1973) copulations were observed at a rate
of slightly more than one per day, so that,
at the very least, frequency of copulation
must be strongly reduced in wintertime.

The size distribution of lizards caught
strongly suggests that there is a peak in
lizards hatching in the late summer and
fall and that very few hatch between
January and March. Of 231 lizards caught
in July and August of 1973 (at Southfield)
only thirteen (5.7%) were under 35 mm
in size, while of 72 lizards caught in the
same study site in December 1973 to
January 1974, eleven (15.3%) were under
35 mm. The small lizards caught in the
summer were probably laid as eggs as
early as the beginning of April. A. S.
Rand (personal communication) collect-
ed over 50 valencienni eggs during
November and early December, and all
of these hatched by mid-December (they
are all the hatchlings on which Figure 1
is based).

Casual observations suggest that
predation on hatchling valencienni is
heavy and that hatching may be synchro-
nous (perhaps to swamp predators). Dur-
ing a one hour and 40 minutes period on
August 5, 1969, four hatchlings emerged
from the large egg hole three feet up a
mango tree at Maryfield. The first
emerged at 1:50 pm, was captured and
died while being measured (male = 25
mm). The second was eaten by an adult
male A. grahami who had just entered
the hole seconds before. It took the male
two and a half minutes to swallow the
hatchling. The third emerged at 3:10 and
slowly (in short bursts of running) made
its way up to 3 m up the trunk. During its
last burst of running it was spotted by an
adult male A. lineatopus, who seized it
and ate it head first, requiring again two
and a half minutes to consume the lizard.
The fourth emerged from the hole, rested
for ten minutes, then passed by an adult
female grahami (47 mm). She saw the

ANOLIS VALENCIENNI -« Hicks and Trivers

593

hatchling but ignored it. In the next 20
minutes the hatchling made its way up to
about 5 m, after which it could no longer
be seen. When last seen, it was higher on
the tree than where the male lineatopus
was perched. (The male grahami was out
of sight high in the tree.)

The synchrony of emergence (and
presumably hatching) is striking. Al-
though the hole was searched daily (be-
fore and after August 5), no other hatch-
lings were seen. Between August 4 and
August 6, about 20 eggs appear to have
hatched in this one hole. On August 6
there were no more eggs stuck to the wall
but the bottom of the hole was littered
with opened eggs.

One other instance of predation was
seen at Maryfield. On August 9 in a dif-
ferent part of the study area a 62 mm
adult male grahami caught and ate head
first a very small valencienni who was 1.3
m above the ground on the tip of a one
centimeter thick twig. One case of can-
nibalism was observed at Southfield. On
July 21, 1973, an 82 mm male valencienni
ate a small marked valencienni head first
(probably no. 95, a 43 mm male).

THE EVOLUTION OF SOCIAL BEHAVIOR
AND SEXUAL DIMORPHISM IN
A. VALENCIENNI

In summary, we argue that the unusual
features of A. valencienni probably
evolved in the following fashion. At some
point, individuals switched from perch-
ing to searching as a means of acquiring
food. Since this new source of food was
only slowly renewable within the day,
females were no longer selected to main-
tain exclusive feeding territories, but
were instead selected to wander widely
and to interact nonaggressively with
other females. This new style of move-
ment rendered individuals more vulner-
able to predation, since they were now
continuously on the go and preoccupied
with searching the substratum. Hence
they became unusually cryptic.

594

Once a system of female territoriality
had broken down, males were selected to
tolerate much greater degrees of terri-
torial overlap. This is because 1) in their
search for food, males were also selected
to wander widely; 2) excluding all males
from a territory no longer gave exclusive
access to the females residing in that ter-
ritory; 3) the individuals of both sexes
were more difficult to detect, thus ag-
gressive control more difficult to impose;
and 4) females, in their choice of mating
partners, may have placed less emphasis
on large male size than when they were a
perching species.

We must confess, however, that we are
ignorant concerning the role that female
choice plays in generating male behavior
and male size. Certainly A. valencienni
seems, among lizard species, unusually
suited to intense female choice, but we
have been unable to design the proper
experiment for measuring female prefer-
ence. What evidence exists suggests that
females may prefer larger than average
males, but not strongly so. We do not
know why sexual dimorphism is reduced
in A. valencienni. In particular, we can-
not discem the relative roles of male-
male competition and female choice. We
have been unable to penetrate the system
of male-male competition. Do males
move frequently, relatively large dis-
tances? Or do they patrol key crossing
points, as if to waylay caravans?

ACKNOWLEDGMENTS

For help spotting and catching lizards,
we thank Raymond Simpson, Glenroy
Ramsey, Leeman Wynter, Veron Elliott
and Paul Bartels. For hospitality in
Southfield, we are grateful to Dorel
Staple. For financial support we thank
the American Philosophical Association,

Advances in Herpetology and Evolutionary Biology

Ernest E. Williams, and the Milton Fund
(Harvard).

DEDICATION (TRIVERS)

Ernest E. Williams took me to Jamaica
in 1968. He was on a collecting expedi-
tion; I was the driver. We arrived in
Kingston in the evening and drove in a
blinding rainstorm to the Maryfield
Guest House, a decaying English great
house set on three acres. The beautiful
old trees and well-tended garden attract-
ed a big population of lizards.

We began our field work over breakfast
on the veranda watching Anolis linea-
topus. The males warmed themselves in
the sun and then engaged in display and
aggressive encounters as they reoccupied
their territories. These bright, active, lit-
tle lizards reminded one almost of pup-
pies or kids, enjoying a little social play
in the early moming hours.

Ernest soon drew my attention to a
more sinister species, which seemed to
hide in the background. This was Anolis
valencienni, which moved in a very
distinctive fashion. Structurally, it was
sufficiently aberrant to have appeared in
earlier classifications as a separate genus
Xiphocircus (Underwood and Williams,
1959). Individuals of this species seemed
unusually abundant at the Guest House,
and since an Anolis of this type had not
been studied, I soon concentrated on
figuring out its social system.

Ernest impressed me very warmly on
that trip to Jamaica. He traveled in a very
calm, quiet, unpretentious style. When
we filled out our immigration cards and
were asked to state our occupation, my
natural impulse was to jack up the de-
scription as high as I could. I expected
Ernest to do likewise. Nothing less than
“Alexander Agassiz Professor of Zoology
at Harvard University” seemed appro-
priate to the occasion. Instead he wrote,
“Teacher.” Simple, unpretentious, and

when need be, nonspecific or ambig-
uous. He also seemed fearless in his
travels through Jamaica, and I much
respected him for this. He neither swag-
gered nor scraped, but held himself calm-
ly at all times.

In taking me to Jamaica, Ernest intro-
duced me to more than A. valencienni. I
have now lived nearly five years in
Jamaica. I am married to a Jamaican and
have four children by her. I own an acre
of land and have planted 250 trees,
mostly fruit. I have studied the butter-
flies of Southfield and collected data on
an interesting symbiosis between a car-
penter ant and a membracid. The trees I
have planted are now beginning to attract
anoles, including A. valencienni, and I
look forward to many more years of
watching.

I dedicate this paper to Emest E.
Williams with affection and respect. He
was my graduate advisor and well loved
at the time. Those were good years, now
warmly remembered. What he _intro-
duced me to has become an enduring
part of my life.

ANOLIS VALENCIENNI «+ Hicks and Trivers

595

LITERATURE CITED

HICKS, R., AND JENSSEN, T. A. 1973. New Studies on
a montane lizard of Jamaica, Anolis reconditus.
Breviora Mus. Comp. Zool. No. 404, pp. 1-23.

LYNN, W. G., AND C. GRANT. 1940. The herpetology
of Jamaica. Bulletin of the Institute of Jamaica.
Science Series, No. 1.

RAND, A. S. 1967a. The ecological distribution of
the anoline lizards around Kingston, Jamaica.
Breviora Mus. Comp. Zool. No. 272, pp. 1-18.

____. 1967b. Ecology and social organization in the

iguanid lizard Anolis lineatopus. Proceedings
of the United States National Museum 122 No.
3595: 1-79.

SCHOENER, T. W. 1970. Size patterns in West Indian
Anolis lizards. III. Sexual dimorphism in rela-
tion to differential resource utilization and
sexual selection. Manuscript pp. 1-113.

____. 1971. Theory of feeding strategies. Ann. Rev.
Ecol. Syst., 2: 369-404.

SCHOENER, T. W., AND A. SCHOENER. 1971. Struc-
tural habitats of West Indian Anolis lizards. I.
Lowland Jamaica. Breviora Mus. Comp. Zool.
No. 368, pp. 1-24.

TRIVERS, R. L. 1976. Sexual selection and resource-
accruing abilities in Anolis garmani. Evolu-
tion, 30: 253-269.

UNDERWOOD, G., AND E. E. WILLIAMS. 1959. The
anoline lizards of Jamaica. Kingston, The Insti-
tute of Jamaica.

Conditional Relatedness, Recombination, and
The Chromosome Numbers of Insects

JON SEGER!

ABSTRACT. If two polymorphic loci are out of
phase equilibrium, a homozygote at one of these
loci is more highly related to its kin, at the other
locus, than is an equivalent heterozygote. As a re-
sult, selection can favor (1) phenotypic responses to
relative heterozygosity, and (2) increased re-
combination between the loci inducing these re-
sponses. Selection is expected to have these con-
sequences only to the extent that kin strongly affect
each other's fitnesses. The chromosome numbers of
social insects appear to be higher, on average, than
those of allied solitary species, which is consistent
with this model‘on the assumption that chromo-
some numbers are selected in part for their effects
on recombination.

INTRODUCTION

Coefficients of relatedness are always
conditional on something. In the usual
formulations they are conditional on
structural relatedness, that is, on the
pedigree connections between the indi-
viduals whose relatedness is to be evalu-
ated. Full siblings are related by one half,
aunt and nephew by one fourth, first
cousins by one eighth, and so on, in
randomly mating diploid populations of
infinite size. At first it seems strange to
speak of structural relatedness as some-
thing that conditions genetic relatedness,
since we usually treat it as that which
sie genetic relatedness. Structural
relatedness is certainly one of the most
important determinants of genetic

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

relatedness, but it is not the only one. For
example, phenotypic similarity is treated
as the conditioner of genetic relatedness
in so-called “spotter gene” or “green
beard” models. Genetic relatedness at a
given locus may be conditional on both
structural relatedness and gene fre-
quency if the fitmesses of the related
individuals are determined according to
one of a number of nonadditive schemes
(Seger, 1981, and references therein).
Genetic relatedness may also be condi-
tional on heterozygosity, in species with
certain kinds of population structures
(Seger, 1976). Here I show that where
relatedness is conditional on _hetero-
zygosity, relatively high rates of recom-
bination may be favored by natural selec-
tion. This idea survives a comparative
test in which chromosome number is
used as a proxy for the average rate of
recombination.

Ernest E. Williams became my thesis
adviser some time after I first saw the
outlines of this argument, but some time
before I first attempted to publish it. This
is nearly the twentieth version of a paper
that has had a long and erratic develop-
ment. Ernest encouraged me to stay with
it on many occasions when I found my-
self stalled and confused. Several im-
portant changes resulted directly from
his criticisms. From time to time he
asked me whether I thought the model
might help to explain lizard micro-
chromosomes. I always said that I sup-
posed it might, but that I was still unable
to see how. In fact, I found the question

RELATEDNESS AND RECOMBINATION - Seger

extremely aggravating. Micros obviously
posed a beautiful problem, closely re-
lated to all my other interests, and yet I
had no idea what to do with them.
Couldn’t he see that? Why then did he
persist in bringing them up? I now sus-
pect that Emest sets out quite deliberate-
ly to tie knots like this in the minds of all
of his students. Those who manage to get
rid of the things, by solving them, will
advance science. The rest will suffer
from recurring and unusually vivid
flashes of ignorance, no matter what
progress they may make on other prob-
lems. In either case, Ermest succeeds as a
teacher. We may not have been exploited
as students, but if this suspicion is cor-
rect, most of us were thoroughly outwit-
ted.

RELATEDNESS CONDITIONAL ON
HETEROZYGOSITY

Consider two loci each with two alleles
(A, and A, at A, B, and B, at B). Let p, be
the frequency of A, at A, and let p, be the
frequency of B, at B. Let x,, x,, x,, and x,
be the frequencies of the gametes A,B,,
A, B,, A,B,, and A,B,, respectively. Then
we can express the gametic frequencies
as

= p,p,+D,

Se PGlp,)

% = (lp) 2:

ae (py) (=p) >
by letting

DRX, XX,

It follows from the definition of condi-
tional probability that

P(A,=A,|B,=B,) = p,+D/p, [1]
and

597

Thus the presence or absence of allele B,
affects the probability that A, occurs on
the same gamete if D is anything other
than zero.

One of the classic results of population
genetics theory states that D decays to
zero from any initial value, in a closed
randomly mating population, if all four
alleles are neutral and there is any
recombination between the A and B loci
(Robbins, 1981; Malécot, 1948). D is
traditionally referred to as the “coeffi-
cient of linkage disequilibrium,’ because
the rate of decay depends on the tight-
ness with which the loci are linked. But
D can take nonzero values for unlinked
loci, even under random mating, and so it
is now increasingly referred to as the
“coefficient of gametic phase disequili-
brium.”

Kimura (1956) and many others have
studied the ways in which epistatic fit-
ness interactions can generate nonzero
values of D, at gene frequency equili-
brium, in an infinite randomly mating
population. Hill and Robertson (1968)
were the first to show that phase dis-
equilibrium is also generated at gene
frequency equilibrium in the absence of
epistasis, even between neutral loci, if
the population is finite in size. The
process is difficult to model, but easy to
grasp intuitively. Gametes are sampled
out of the parental generation, and there-
fore some are inevitably chosen in pro-
portions that do not perfectly reflect the
frequencies of their constituent alleles.
On average there will be either an excess
of “coupling” gametes (D positive) or an
excess of “repulsion” gametes (D nega-
tive). These random perturbations of the
gametic frequencies tend to cancel, giv-
ing an expectation of D equal to zero. But
since D is very seldom exactly zero, the
expectation of D? is greater than zero.

The expected magnitude of D depends
strongly on the gene frequencies p, and
p,. It can be shown that

D(p,.p,) = [o2p,p,(1—p, )(1-p, * [3a]

598

gives a reasonable approximation to the
expected absolute value of D if selection
is weak (Seger, 1980). In equation [3a]

o2 = 1/[3+4N(c+k)-2/(2.5+N(ct2k))],
[3b]

where N is the effective population size,
k is the sum of the four forward and
backward mutation rates, and c is the re-
combination fraction (Ohta and Kimura,
1969). Since k is usually much smaller
than c, mutation rates have little effect on
o% (and thus on D(p,,p,)), except for very
tightly linked genes. Selection also has
little effect unless it is very strong (Ohta
and Kimura, 1969). Linkage increases the
expected magnitude of D, but even for
unlinked genes D takes values approxi-
mately 0.02, 0.10, and 0.20 times its pos-
sible maxima (given p, and p,) in popula-
tions of sizes N=1,000, N=50, and N=10,
respectively. Such population sizes ap-
pear to be typical (Jain, 1976; Lande,
1979; Levin and Kerster, 1969; Loukas et
al., 1979; Schaal, 1980; Schaal and Levin,
1978; Wright, 1978). Thus, although the
sign of D is unpredictable unless there
are epistatic fitness interactions between
the A and B loci, the expected magnitude
of D may be substantial, even when A
and B are unlinked.

Given the conditional probabilities [1]
and [2], it is easy to find the correlation
between an individual’s A-locus alleles,
conditional on its B-locus genotype. As-
sign the arbitrary genic values 1 and 0 to
alleles A, and A). The joint genic distri-
bution of the gametes that united to form
the individual can be written

: [4]

where a,, is the probability that the indi-
vidual received A, from both parents,
aj) = ap, is the probability that it received
A, from one parent and A) from the other,
and a,, is the probability that it received
A, from both. Letting X denote the genic

Advances in Herpetology and Evolutionary Biology

value of the individual’s maternal gamete,
and Y that of the paternal, it follows
immediately that

E(XY) = a,,
and

E(X) = E(X?) = E(Y) = E(¥2) =a,,+a,, =
M.,.
Suppose the individual is known to be
a heterozygote at the B locus. The proba-
bility of this event is k=2p,(1—p,). The
probability that the ander thal is both
B,B, and A,A, is 2x,x,. Thus the condi-
tional probability that it is A,A,, given

that it is B, B,, can be written as

2x, xX, /k
2 P,P. +D)[p, (1-p,)—-DI]/k
p?-2D[D +p, 2p, — Dik:

By similar reasoning

(x, +x, )/2k
p, — D(2p, — 1)/k.

Thus

Cov(X,Y) = —[D?/k][2+(2p, —1?/k], [5]

and

Var(X) Ti p,(1—p,)+[D(2p, —1)/k]
(2p; >! De@p,— iki: [6]
Given any intermediate values of p, and
p,, Cov(X,Y) is negative or zero. Thus
since we expect a nonzero value of D ina
finite population, we expect a negative
correlation between the A-locus gametes
of a B-locus heterozygote. The correla-
tion between an individual’s gametes is
the same as the coancestry of its parents,
so the parents of a B-locus heterozygote
have a negative coancestry and are nega-
tively related at the A locus.

Where mating is at random, there can
be no overall correlation between par-
ents at the A locus or at any other. This

RELATEDNESS AND RECOMBINATION - Seger

clearly implies that the parents of B-locus
homozygotes are positively related at the
A locus, by an amount sufficient to cancel
the negative relatedness between par-
ents of B-locus heterozygotes. By means
of an argument exactly like that leading
to equations [5] and [6] it can be shown
that

Cov(X,Y) = [D?/k][2—(2p,-1P/k] [7]

and

Var(X) = p,(1—p,)+[D(2p, — 1)/k]
[1-2p, -D@2p, — 1)/k], (8]
where k=1—2p,(1—p,) is the overall
probability of being a B-locus homo-
zygote. In this case Cov(X,Y) is always
zero or positive. Thus the parents of a B-
locus homozygote are positively related
at the A locus, in a finite population.

This implies, but does not prove, that a
B,B, heterozygote will be related to its
full ‘sibling by less than one half, across
the A locus, while a B,B, or B,B, homo-
zygote will be related to its sisi by
more than one half. The explicit demon-
stration is tedious but straightforward
(Seger, 1980). The conclusion can be
summarized by saying that A-locus re-
latedness is conditional on _ B-locus
genotype, for full siblings in a finite
population. Figure 1 shows these condi-
tional coefficients of relatedness as a
function of population size, for the case
in which p, =0.5 and the A and B loci are
unlinked. Linkage increases the ex-
pected magnitude of D, but decreases the
extent to which the conditional coeffi-
cients diverge from one half. This is il-
lustrated in Figure 2, for populations of
effective size 10, 50, and 250.

Suppose the phenotype affected by A-
locus genotype has fitness effects on both
ego and its sibling. Then ordinary kin
selection will favor A-locus alleles that
maximize ego's inclusive fitness, given
the siblings’ nominal relatedness of one
half, and given any well defined con-
straint on their possible joint fitnesses.

599

-20

Jl
00
49

48

2500 250 50 10

N

Figure 1. Conditional relatedness as a function of
population size.

The upper curve shows R*, a B-locus homozygote’s
A-locus relatedness to its sibling, as a function of N, the
effective population size. The lower curve shows R-,
the corresponding relatedness of a B-locus hetero-
zygote. The upper horizontal axis shows values of D’,
the standardized coefficient of gametic phase disequi-
librium, Corresponding to the values of N on the lower
axis. Loci A and B are unlinked, and py = 0.5.

But we know that ego’s A-locus related-
ness to its sibling is actually conditional
on ego's B-locus zygosity. Thus it seems
to follow that an A-locus allele that made
its phenotypic effect appropriately con-
ditional on ego’s B-locus genotype could
displace all alleles with unconditional
phenotypic effects. This conjecture can
be shown to hold under plausible as-
sumptions regarding the constraints on
joint fitness (Seger, 1980), and can be
summarized by saying that an uncondi-
tional A-locus phenotype is not in
general an ESS against a phenotype that
is conditioned on the zygotic states of
other polymorphic loci (Seger, 1976).
From the point of view of the A locus,
the zygotic state of B is a signal that con-
veys information about the A-locus re-

600

2 N=IO
5I
N=50
N=250
R 50
N=250
N=50
43
48
N=10
SS
Qn sulveee otro, sito mies

Figure 2. Conditional relatedness as a function of
linkage between A and B.

Increasing the linkage between loci A and B in-
creases the expected magnitude of D’, but decreases
the difference between R* and PR. Here pz = 0.5, and
c varies between 0 (no recombination) and 0.5 (free
recombination). Three different population sizes are
shown. Upper curves are R‘, lower curves are R’, as
explained in the legend to Figure 1.

latedness of ego and its sibling; geno-
types B,B, and B,B, indicate an inflated
A-locus 4 Fiediness. and B, B, indicates a
deflated relatedness, sella tne to the
unconditional relatedness of one half.
Thus we can refer to ego’s membership
in distinct relatedness sets which are
defined by its B-locus genotype. In the
example considered so far, in which
there is only one signalling locus, there
are only two conditional relatedness sets.
But in principle there could be a large
array of signalling loci and several dis-
tinct relatedness sets corresponding to
different configurations of homozygosity
and heterozygosity across the array of
signalling loci.

Advances in Herpetology and Evolutionary Biology

Figure 2 shows that linkage between A
and B reduces the distinctness of A-locus
relatedness sets defined on B. Thus a
gene that increased the rate of recom-
bination between A and B would in-
crease the distinctness of relatedness
sets, and would thereby promote the
spread of conditionally responding
alleles at the A locus. But would selec-
tion also favor the gene for increased
recombination?

SELECTION FOR INCREASED
RECOMBINATION BETWEEN
SIGNALLING LOCI

Consider a locus C, unlinked to either
A or B, with alleles C, and G,. C, pro-
motes recombination between A and B,
while C, suppresses it. Formally, B is a
signalling locus from the point of view of
C, just as it is from the point of view of A,
despite the fact that C is unable to bring
about any phenotypic response to the
zygotic state of B. If A were unlinked to
B, as C is, it and C would belong to the
same relatedness sets defined on B and
would therefore have the same inclusive
fitness interests. This implies that C,,
which reduces linkage, might indeed be
favored over C,. Consider a slightly dif-
ferent model in which there are many
polymorphic B loci, all linked to each
other but unlinked to A and C. In the
limit of complete linkage to each other
they constitute a supergene with many
pseudoalleles of different degrees of
complementarity. It is clear that the
average homozygosity over this set of B
loci carries less information about the
relatedness of an individual’s parents
when the loci are tightly linked than it
does when they recombine freely, which
implies that a gene promoting recombi-
nation between the B loci could be
favored.

Like the fitnesses of sex ratio pheno-
types, the fitnesses of recombination
phenotypes do not manifest themselves
until one or more generations following

RELATEDNESS AND RECOMBINATION - Seger

their expression. Thus there seems to be
no simple way to apply genotypic co-
variance methods (Price, 1970; Seger,
1980, 1981) to the evolution of recombi-
nation modifiers. But positive selection
for increased recombination can~— be
demonstrated by simulation of a particu-
lar model with many signalling loci. One
such simulation is described below.

The model species is a _ diploid
hermaphrodite in which self-fertilization
alternates with random mating. The
population is infinitely large, but the
effective population size is finite, owing
to the high frequency of selfing. Off-
spring are produced in pairs. Each pair is
given one unit of resource (x), and an
individual’s fitness (W) depends on the
amount of resource it consumes, accord-
ing to the function

W(x) = 1.0-—e™%.

Given this fitness function and _ total
shareable resource, the fitness set for
sibling pairs is slightly convex (Fig. 3).
One sibling, called “left,” divides the
unit of resource between itself and the
other sibling, called “right.”

Locus A controls the left sibling’s di-
vision of the unit of resource, and locus C
controls the rate of recombination at B, a
large array of signalling loci to which
neither A nor C is linked. In the simula-
tion, A and C are treated explicitly, but B
is treated in the aggregate. Typical loci in
B are imagined to be highly polymorphic,
such that outbred individuals are always
more heterozygous than inbred _indi-
viduals, if recombination between B loci
is free. C, causes free recombination
between all B loci, and C, suppresses
recombination between them. In some
runs C, was dominant to C,, and in others
it was recessive.

The behavior of this genetic system
can be described as follows. All of the
selfed progeny of recombining parents
are moderately homozygous at B, having
received the same allele in egg and
sperm at approximately half of their loci.

601

(a)

Figure 3. Fitness set for siblings.

The fitness of each sibling is determined by its con-
sumption of resource (x), according to the function
W=1.0—e~. This function is shown in (a). Given that
two siblings share a total of one unit of resource, the
set of their possible joint fitnesses is as shown in (b).
W_ is the fitness of the “left” sibling, and We, is the
fitness of the “right” sibling, as explained in the text.

Half of the selfed progeny of nonrecom-
bining parents are entirely homozygous,
having received an intact copy of the
same entire strand in egg and sperm, but
half of them are only slightly homozy-
gous, having received the parent’s
maternal strand in one gamete, and its
paternal strand in the other. A, causes the
left sibling to respond to its zygosity at B
and is codominant with A,, which when
homozygous causes a fixed division of
the resource. Left siblings of genotypes
A, A, keep 0.8 units of resource and give
the remaining 0.2 units to their right
siblings. This is the ESS nonresponding
phenotype, which was found by playing

602

A, variants against each other in the ab-
sence of A,. Lak siblings of genotypes
A, A, and A, A, keep relatively more for
themselv 7eS when heterozygous at B (0.85
and 0.9 units, respectively), but keep less
(0.75 and 0.7) when moderately or en-
tirely homozygous at B. This level of
response is derived from a_ simple
inclusive fitness model, based on the
formal relatedness of selfed siblings.

The simulation evolves as expected,
when allowed to run freely for many
generations. A, spreads from low fre-
quencies, even in populations fixed for
C,, the gene that blocks recombination at
B. ‘This happens because the zygosity of
B contains some information about the
siblings’ relatedness even when there is
no recombination. Consider an outbred
carrier of A,, the responding gene. Such
an individual is heterozygous at B, and
therefore gives relatively little of dhe
resource to its right sibling. This is ap-
propriate, owing to the relatively low
relatedness of outbred siblings. Now
consider the inbred (selfed) carriers of
A,. Half of them are heterozygous at B,
because there is no recombination, ancl
mistakenly act as if they were oon,
but half of them are homozygous andl
appropriately give a relatively large share
of the resource to their right siblings.
Thus carriers of A, respond in a way that
is appropriate to their actual relatedness
three quarters of the time, and this is suf-
ficient to give them a higher average in-
clusive fitness than that realized by the
nonresponding A, A, homozygotes, in this
particular model.

When C, and A, are introduced to-
gether they both | spread, and _ their
progress appears to be mutually reinforc-
ing. They go to fixation whether C, is
recessive or dominant and whether the C
and A loci are linked or unlinked (Fig. 4).
This happens because the inbred
progeny of recombining parents usually
carry C,, and those who also carry A,
never make the inappropriate response
made by half of the nonrecombinant in-

bred responders. C, increases the infor-

Advances in Herpetology and Evolutionary Biology

mation about relatedness that is con-
veyed by the zygosity of the signalling B
loci, and thereby improves the accuracy
of the response to zygosity. This benefits
both C, and A, because the siblings are
equally related across all loci.

This model differs in several reopeee
from previous models for the evolution of
recombination. It does not require high
fecundity, environmental variation, or
epistatic fitness interactions. It does not
require that the recombination modifier
be linked to the sites it affects, or even
that there be any average fitness differ-
ences between alleles at the affected loci.
But this model does require that the af-
fected loci (the B array in the realization
discussed here) exhibit the form of over-
dominance that is brought about by the
spread of genes that promote responses
to the zygosity of the affected loci.

SOCIALITY, RECOMBINATION, AND
CHROMOSOME NUMBERS

Selection for phenotypic responses to
zygosity should be much stronger in
some species than in others. Selection
should be weak where relatives seldom
compete or where there is never any
significant inbreeding or gametic phase
disequilibrium. Both of these conditions
are probably found in many species with
very large effective population sizes, for
example, those with planktonic larvae.
Selection should be strong where rela-
tives often compete and where condi-
tional relatedness sets are well differen-
tiated. These conditions are probably
found in a variety of species with small
effective population sizes, for example, at
all stages of the life cycle in many social
species, during the larval stage in species
with gregarious development, and during
the embryonic stage in many self
compatible hermaphrodites. The fact that
predominantly self-pollinating angio-
sperms tend to have higher chiasma
frequencies than do their predominantly
cross-pollinating relatives (Lewis and

RELATEDNESS AND RECOMBINATION - Seger 603
1.0 (a) (b)
OG
as
O
O 10 O 1.0
P(A,) P(A,)

Figure 4. Gene frequency trajectories at loci controlling recombination and sibling behavior in a deterministic
model.

A, promotes responses to zygosity, and C, promotes recombination. Figure 4.1a: C; is recessive to Co.
Trajectories terminate at generation 1350. Recombination fractions between A and C are zero (complete linkage),
0.005, and 0.5 (free recombination) in the longest, intermediate, and shortest trajectories, respectively. Figure
4.1b: C, is dominant to Cp. Trajectories terminate at generation 800. Recombination fractions are as in Figure
4.1a.

Both A; and C, spread more rapidly when linked to each other than when not linked, and both spread more
rapidly when C, is dominant than when it is recessive. Rates of gene frequency change along the trajectories
suggest that the marginal fitnesses of A, and C, change in complex ways during the course of selection. Apparent
selective advantages for A, range between 0.003 and 0.012. Those for C, range between 0.003 and 0.025. C,
benefits more from the advance of A; than A; does from the advance of C,, which is expected because

recombination affects only the responding phenotypes (those carrying A; ).

John, 1963; Murray, 1976; Ved Brat,
1965; Zarchi et al., 1972) is consistent
with this expectation, but it is also con-
sistent with other models for the evolu-
tion of recombination under selfing
(Charlesworth et al., 1977; Maynard
Smith, 1978).

In almost all social insect species there
are aspects of larval development and of
adult behavior that appear to reflect the
distribution of coefficients of relatedness
within colonies (Hamilton, 1972; Trivers
and Hare, 1976). In some species there is
also intense competition between colo-
nies (Wilson, 1971; Michener, 1974).
Many species of social insects appear to
have populations small enough to gener-
ate significant gametic phase disequili-

brium even under local random mating
(Talbot, 1965; Wilson, 1958, 1963, 1971;
Michener, 1974; Haskins and Wheldon,
1965; Haskins, 1970; Taylor, 1978). If so,
Isoptera (termites) and social Hymen-
optera (ants, bees, and wasps) should
have rates of recombination higher on
average than those of their closest soli-
tary relatives.

Unfortunately, there are no compara-
tive data on the recombination of
markers, or even chiasma frequencies, for
social insects and their relatives. But
chromosome number is a major determi-
nant of the average linkage within a
genome, and chromosome numbers are
known from many species of insects. Loci
at the opposite ends of a long chromo-

604

some may recombine almost freely, if
several crossovers usually occur between
them. Nonetheless, they remain some-
what linked. A fission occurring between
them reduces their linkage all the way to
zero. More importantly, other pairs of loci
at map positions between them but on
opposite sides of the point of fission,
which were formerly significantly linked,
also become unlinked. Thus given an
approximately constant number of cross-
overs per genome, the average linkage
between loci is a strongly decreasing
function of the chromosome number.
This implies that chromosome number
can be used as a proxy for the rate of
recombination in comparative studies, as
long as large numbers of species are
compared.

As expected, the average chromosome
numbers of eusocial taxa are consistently
higher than those of the most closely
related nonsocial taxa for which chromo-
some data exist (Sherman, 1979). Formi-
cidae, Apidae, and Vespidae represent
independent derivations of eusociality
within the aculeate Hymenoptera (Wil-
son, 1971). Although only a few solitary
aculeates have been karyotyped, their
chromosome numbers are similar to
those of the entirely solitary Symphyta
and Parasitica (Table 1). This suggests
that high numbers have evolved inde-
pendently in the social families. The
Isoptera are entirely eusocial, so no close
comparisons are possible in their case.
But their average chromosome number
exceeds those of the other orthopteroid
orders Blattodea, Orthoptera, Mantodea,
Dermaptera, and Embioptera (Table 1).

It should be emphasized that the pos-
tulated selection in favor of increased
recombination has nothing to do with
sociality as such. Sociality is merely
being used as an indicator of situations in
which interactions are expected to be
strongly conditioned on relatedness. In
such situations natural selection should
favor phenotypic responses to individual
zygosity, if at the same time the popula-
tion structure is such as to create signifi-

Advances in Herpetology and Evolutionary Biology

cant gametic phase disequilibrium.
Selection in favor of increased recom-
bination is thus an expected secondary
consequence of the establishment of sys-
tems of responding loci. It is therefore
expected in some solitary species, for
example, ones in which extended juve-
nile dependency leads to _ parent-
offspring conflict over the allocation of
resources (Trivers, 1974), and it is not
expected in some social species, for
example, those with very large effective
population sizes. The range of haploid
numbers in ants (3-46) exceeds that in all
other Hymenoptera. If the argument
made here is correct, some of this varia-
tion should be associated with differ-
ences of population structure. Unusually
high chromosome numbers occur in
Rhytidoponera, Myrmecia, and Notho-
myrmecia. Aptery, brachyptery, and
patchy distributions are common in these
genera (Haskins and Wheldon, 1965;
Haskins, 1970; Taylor, 1978), suggesting
that many of their species typically ex-
perience small effective population sizes.

The average chromosome numbers of
six species of blattid roaches believed to
exhibit larval aggregation are 60% higher
than the chromosome numbers of ten
species believed not to exhibit larval
aggregation (Sherman, 1979).

It is difficult to assign levels of statisti-
cal significance to the observed patterns
of chromosome number variation, be-
cause it cannot be assumed that the
chromosome numbers of related species
are statistically independent. Thus it is
not legitimate to perform an analysis of
variance on the chromosome numbers of
species, as was done by Sherman (1979).
The analysis given here treats genus
averages as primary observations. This
improves matters somewhat, but since
genera tend to be correlated within fami-
lies, as do species within genera, genera
cannot properly be treated as independ-
ent points in a comparison between fami-
lies. Thus the significance levels given in
the legend to Table 1 should be taken
only as indicators of the relative strengths

RELATEDNESS AND RECOMBINATION - Seger

TABLE 1. AVERAGE HAPLOID CHROMOSOME NUMBERS
OF SOCIAL INSECTS AND THEIR ALLIES .*

N n s.d.
Orthopteroids
DERMAPTERA 20 13.38 5.92
EMBIOPTERA 4 10.83 0.41
ORTHOPTERA 33 10.84 DUO)
PHASMATODEA 40 20.14 4.45
MANTODEA 69 13.20 2.44
BLATTODEA 64 19.05 6.62
Blattidae i 18.03 5.90
ISOPTERA* 27 20.71 1.95
Hymenoptera
SYMPHYTA 27 9.41 3.96
APOCRITA
PARASITICA 19 8.51 2.38
ACULEATA
Vespoidea
Formicidae* 69 15.76 6.82
Eumenidae 2 7.63 0.53
Vespidae* 2 14.30 2.40
Sphecoidea
Halictidae 4 11.50 5.26
Anthophoridae 3 13.67 4.04
Megachilidae 1 16.00 —
Apidae* 18 16.23 2.86

*Sources of data are listed in Literature Cited B.

Social taxa are indicated in the table by an as-
terisk. Most of the analysis was conducted using
genus averages rather than species counts, to in-
crease the independence and representativeness of
sample points, and to increase the normality of the
resulting distributions. Thus N is the number of
genera in a sample, and n is the mean of the genus
mean haploid chromosome numbers. The average
for Orthoptera, however, represents modal numbers
for 28 families and subfamiles of Caelifera (based
on hundreds of studied species [White, 1973]), and
average numbers for five families of Ensifera (based
on about 70 genera [Makino, 1951]). Data for
Isoptera include nine unpublished species counts
in five genera of Kalotermitidae (P. Luykx, personal
communication). One-tailed U-tests were applied to
comparisons of interest (t-tests with d.f. adjusted for
unequal variances were also applied and gave simi-
lar results).

Isoptera exceed the blattid roaches, to whom they
are believed to be most closely related (McKittrick,
1965), but the difference is not significant. The
comparison with Blattodea as a whole is significant
(p<0.005), as are those with Orthoptera, Mantodea,
Dermaptera (p<0.0005), and Embioptera
(p<0.001). The comparison with Phasmatodea is
almost significant (p<0.1).

Formicidae (ants) and Apidae (eusocial bees)
each exceed both Parasitica (chalcids, ichneumons,
gall wasps) and Symphyta (sawflies, horntails)
(p<0.0005). The small sample of halictid, antho-

605

phorid, and megachilid bees consists of five species
with haploid numbers of 16, and three species with
haploid numbers of 6, 8, and 9. The species with
low. numbers are communal to quasisocial
(Michener, 1974), and three of the species with high
numbers are solitary, contrary to expectation.
Owing to the complexity of social evolution in the
bees, a much larger and more representative sample
of chromosome numbers is required to determine
whether in fact they exhibit a pattern different from
the one seen elsewhere in the Hymenoptera.

The most appropriate comparison is that between
the Vespidae (eusocial paper wasps) and the closely
related Eumenidae (solitary potter wasps). Their
average numbers diverge in the expected direction,
but too few genera have been studied to allow a
statistical test of the generic averages. If all the
published species counts (7 and 5, respectively) are
used, the difference is significant (U=5, p=0.024,
t=2.8, 8 adjusted d.f., p<0.025). Formicidae exceed
Eumenidae (p<0.005) but not Vespidae, using
species counts for the wasps.

of different associations, not as statistical
tests.

A conservative test of the overall pat-
tern can be carried out by comparing the
signs of the differences between groups
at the same taxonomic levels. There are
at least six relevant comparisons. Apidae
exceed all other bees. Vespidae exceed
Eumenidae. Aculeata as a whole, who are
mainly social in this sample, exceed
Parasitica. Isoptera exceed Blattodea.
Rhytidoponera, Myrmecia, and Notho-
myrmecia exceed other ants. Gregarious
blattids exceed nongregarious blattids.
The comparison between Formicidae
and Eumenidae could be added to this
list, as could separate comparisons be-
tween Apidae and the other families of
bees, but these will be omitted in the
interests of conservatism. The probabili-
ty that six independent comparisons all
fall in the same unspecified direction is
0.03125, and the probability that they all
fall in the same specified direction is
0.015625. Thus it appears that an associ-
ation does exist between high chromo-
some number and circumstances favor-
able to the evolution of responses to

zygosity.

606

DO BEES HAVE GREEN SETAE?

Sherman (1979) was the first to note the
association between eusociality and high
chromosome numbers. His explanation
for the association, which differs from the
one given here, can be summarized as
follows. Parent and offspring are always
related by exactly one half, because the
parent transmits to the offspring exactly
one member of each of its pairs of
homologous chromosomes. Full siblings
also expect to share half of their chromo-
somes identical by descent through the
parents, but owing to the uncertainties of
meiosis this expectation has a finite vari-
ance. Some siblings have more than half
of their DNA in common, identical by
descent, and others have less than half.
Thus some siblings are related by more
than one half, and others by less. If an
individual could discriminate between
the two kinds of siblings, favoring those
to which it was related by more than one
half, it could increase its inclusive fit-
ness. In a social insect colony such dis-
crimination is likely to work against the
reproductive interests of the queen and
her mate. The scope for such discrimina-
tion is a function of the variance of actual
relatedness between siblings, and the
variance of relatedness is an inverse
function of the number of chromosome
pairs. It is in the queen’s interest to re-
duce the variance of relatedness as far as
possible, so as to create a situation in
which workers are selected to minister to
their reproductive siblings “on the basis
of their need rather than kinship” (Sher-
man, 1979). Increasing the chromosome
number is probably one of the simplest
and most effective ways to reduce the
variance of relatedness. Thus chromo-
some numbers are expected to increase
in eusocial lineages and in others where
siblings interact intensely.

This argument turns on two critical as-
sumptions. First, the variance of the pro-
portion of DNA held in common identi-
cal by descent corresponds to a variance
of relatedness on which kin selection

Advances in Herpetology and Evolutionary Biology

could act. Second, although an increased
variance of relatedness would be in the
offspring’s interest, a reduced variance
will evolve because that is in the queen’s
interest. This second assumption is
problematic, but it will not be considered
here in any detail. The first assumption is
almost certainly incorrect, at least as
stated by Sherman (Dawkins, 1979). The
idea seems to arise naturally from the
identity-by-descent view of relatedness,
which emphasizes amounts rather than
covariances. The problem with it is that
identity at one locus does not predict
identity at another. As Dawkins puts it,
“If I want to guess whether your hand of
cards contains the ace of spades I would
be quite wrong to say: I already know
you have the 2, 3, 5, 6, 7, 9, 10, Jack and
King of spades; therefore you have a
strong hand in spades; therefore you
probably have the ace!” A rational gene
that controlled discrimination would gain
no information regarding its presence in
a sibling of its bearer by knowing that the
sibling, like the gene’s own bearer, had a
green beard. If the discrimination gene
were tightly linked to the green beard
gene the argument might hold, except
that selection at all unlinked loci would
oppose the favorable discrimination.
Few theoretical arguments are affected
by the fiction of the single omnipotent
gene, but this appears to be one that is.
Any act as complex as a comparison of
particular details of one’s own and one’s
sibling’s phenotypes must depend direct-
ly on scores or hundreds of genes, and
indirectly on thousands. It is asking a lot
to expect them all to be firmly linked to
the same phenotypic marker and to be
protected from interference on the part of
genes elsewhere in the genome.

It was pointed out above that correla-
tions of allelic state arise through drift in
finite populations, and that these correla-
tions can drive the evolution of pheno-
typic responses to zygosity. These same
correlations can be shown to underlie a
consistent version of Sherman’s argu-
ment for the evolution of discrimination

RELATEDNESS AND RECOMBINATION - Seger

based on phenotypic similarity. But as
will be seen, this version of the argument
may have implications somewhat differ-
ent from those suggested by Sherman.

Consider a genetic system consisting of
two unlinked loci, each with two alleles.
Locus B determines a visible phenotypic
attribute and has two distinct, fully pene-
trant, codominant alleles, B, and B,.
Locus A determines a response to the B-
locus similarity of its bearer and another
individual taken at random from the local
population. With an adjustment for struc-
tural relatedness the argument can easily
be adapted to the case of siblings, but the
case of unrelated individuals is both
simpler and more general than that of
structurally related individuals. Locus A
can be taken to represent every locus in
the genome that is unlinked to B. We
seek the A-locus relatedness of two indi-
viduals, conditioned on their B-locus
similarity.

Let p, be the frequency of A,, and let p,
be the frequency of B,. Then om equa-
tions [1] and [2] we awe that

q, = P(A, =A, |B, =B, ) = p,+D/p,

and

A P(A, =A, |B, =B,) = =D Ds)
where D is the coefficient of gametic
phase disequilibrium. Define as “alike”
those pairs of individuals in which both
are B,B, or both are B,B,. Assuming
Hardy-Weinberg neOrions at both loci,
and assuming that individuals meet at
random, the probability that two indivi-
duals are both B,B, is clearly

f, = p,
while the probability that both are B, B, is
fy = Ula)
Consider the A-locus genotypes of the

B,B, pair. The probability of finding A,
ah alban chromosome at the A locus in

607

either individual is just q,. Thus the
pu seledlasy that both individuals are A, A,
is q?. Similarly, if both individuals are
B,B, the peel eloiiisy that they are also
both A,A, is qj. Thus if two individuals
are known to be “alike” at B, the proba-
bility that they are both A,A, is

2 = (fai +fq5)/k

where k=f,+f,. The probability a,, is the
first element of the joint genotypic distri-
bution

Ags a) Ab M,
Ais ay) a9 M, [9]
Ano A) Ag M).

Proceeding in this manner it is easy to
show that the other elements of the joint
genotypic distribution can be _ repre-
sented as

2 = By
a, = 4h 2 +£2)/k,
M, = (hae +fae)/k,
M, = 2(F.¢, +f, c, )/k,

where c » =a, (1— q,) and c,=q)(1—q,). M
and the remaining a. are easily found ibs
subtraction. Substitution of the key ele-
ments of [9] into

R = [a,, +a,, +’4a,, —(M, +72M, ? ]/
[M,M, +14M,(1—M,)]

(Seger, 1980) gives R*, the A-locus re-
latedness of two randomly chosen indi-
viduals who are “alike” at B. If both loci
are polymorphic and if D is nonzero then
R* is positive (Table 2).

Define as “unlike” those pairs of indi-
viduals in which one is B,B, and the
other is B,B,. Define as “ gommaemlbnt”
those pairs in which at least one indi-
vidual is the heterozygote B,B,. Then
proceeding as above it is easy to Sal R,
the relatedness of “unlike” pairs, and Rs,
the relatedness of “somewhat” pairs.
These are both negative.

Like the conditional coefficients dis-

608

TABLE 2. CONDITIONAL RELATEDNESS IN A SPOTTER
GENE MODEL.*

D’=.05 Do
1 3 5
.00001 .00076 .00499
002TH —.00038 0
—.01430 —.00596 —.00499
.00001 .00075 .00499
D; 3 | —.00269 —.00038 0
—.01390 —.00594 —.00499
.00001 .00075 00499
5 | —.00269 —.00038 0
—.01385 —.00594 —.00499
D’=.1 Po
1 3 5
.00003 .00309 .01980
il || = OUI —.00154 0
—.06279 —.02398 —.01980
.00003 .00302 .01980
P 3 | —.01089 —.00153 0
—.05575 —.02361 —.01980
.00003 .00301 .01980
5 | —.01086 —.00153 0
—05501 — 02357 —.01980

*The upper number in each set of three is the A-
locus relatedness of two randomly chosen indivi-
duals that are “alike” at the B locus. The lower
number is the relatedness of “unlike” pairs, and the
middle number is the relatedness of “somewhat”
pairs. Conditional relatedness is shown for three
different frequencies of A, and B,, at two different
values of D’, the standardized goeitereat of gametic
phase disequilibrium. Each distribution is sym-
metrical about both p, =0.5 and p,=0.5.

cussed in the introduction, Rt and R™
diverge as D increases (Table 2). But R*+
and R~ also diverge as the A and B loci
become more tightly linked, unlike the
conditional coefficients discussed in the
introduction, which converge on the un-
conditional coefficient as linkage
tightens. This implies that conflicts
might arise between A-loci closely linked
to B, and A-loci unlinked to B, because
the former will tend to be farther out of
phase quilibrium with B than will the

Advances in Herpetology and Evolutionary Biology

latter. In view of this it is hard to see
whether a high rate of recombination
looks more like a way to frustrate dis-
crimination than it does like a way to
promote it. By reducing the expected
gametic phase disequilibrium between B
and a typical A locus, recombination
reduces the difference between R* and
R’, thereby reducing the potential ad-
vantage of discrimination. But recombi-
nation also increases the fraction of the
genome that is effectively unlinked to
any given B locus. By increasing the
similarity of the different values of R*
evaluated at different loci, recombination
might well increase the ease with which

mechanisms inducing discrimination
could evolve.
Phenotypic discrimination mecha-

nisms of the form envisioned by Sherman
and others require that an individual be
able to assess its own similarity to
another individual. Recent experimental
work on discrimination by means of indi-
vidual odor differences in Lasioglossum
demonstrates that these bees compare
strangers to the aggregate of their nest-
mates, and that they apparently do not
perceive their own odors (Greenberg,
1980; C. D. Michener, personal com-
munication).

CONCLUSIONS

It is possible that high chromosome
numbers are caused by some feature of
sociality that has nothing to do with kin-
ship. For example, the genomes of social
insects might be larger than those of soli-
tary insects, and there might be an opti-
mum distribution of chromosome sizes,
at least within particular lineages. If this
were true, chromosome number and
genome size would correlate positively,
but several lines of evidence suggest that
they do not. Chromosome numbers and
absolute DNA values for 45 species of
insects representing six orders do not
correlate either within orders or between
them (Fig. 5). Imai, Crozier, and Taylor

RELATEDNESS AND RECOMBINATION - Seger

pg DNA/C

HAPLOID NUMBER

Figure 5. Relationship between genome size and
chromosome number in six orders of insects.

Genome size is expressed as picograms of DNA per
haploid genome. The orders are Orthoptera (@),
Hemiptera (w), Coleoptera (©), Lepidoptera (A),
Diptera (O), and Hymenoptera (4). Spearman rank
correlations are nonsignificant within Orthoptera
(0.097), Diptera (0.006), and Hymenoptera (zero). The
entire sample is significantly correlated (r,=0.34,
p<0.05) because it is dominated by Orthoptera and
Diptera, whose ranges do not overlap on either vari-
able. Using mean values for each of the six orders, the
overall correlation is negative and nonsignificant
(r, =—0.086).

There is a large average difference in genome size
between the hemimetabolous and holometabolous
insects in this sample (represented by filled and by
Open symbols, respectively). The only social species
represented (Apis mellifera and A. cerana, n=16, 0.19
pg) have genomes that appear to be small even for
Hymenoptera. Other Hymenoptera are Megachile
rotundata (n=16, 0.3 pg), Habrobracon juglandis and
H. serinopae (n=10, 0.16 pg), and Mormoniella
vitripennis (n=5, 0.33 pg).

Sources of data are listed in Literature Cited B.

(1977) optically measured chromosome
lengths in 16 species of myrmeciine and
ponerine ants with haploid numbers of
four to 42. They found that average
chromosome length is inversely propor-
tional to chromosome number, and con-
cluded that the diversity of chromosome
number in these subfamilies does not
reflect any large or systematic differences

609

of genome size. Data for angiosperms
(Bennett and Smith, 1976) and for verte-
brates (White, 1973) also fail to suggest
an inherent’ relationship between
genome size and chromosome number.

The first argument outlined here im-
plies that selection will favor relatively
high overall rates of recombination in
species where it also favors phenotypic
responses to heterozygosity. Such spe-
cies are characterized by certain be-
havioral, ecological, and demographic
parameters. To the extent that sociality
indicates the required parameters, and to
the extent that chromosome number in-
dicates the rate of recombination, social
insects are expected to have elevated
chromosome numbers, as they do on
average. The second argument, that of
Sherman (1979), also leads to the expec-
tation that social insects will have ele-
vated rates of recombination. The two
arguments invoke nearly identical sets of
causal parameters and thus make similar
comparative predictions. There seems to
be no way to distinguish between them
on the basis of available evidence.

ACKNOWLEDGMENTS

I thank R. L. Trivers for encourage-
ment and for many stimulating discus-
sions during the early stages of this work.
I thank S. H. Bartz, M. E. Etter, W. D.
Hamilton, P. H. Harvey, R. C. Lewontin,
J. Maynard Smith, J. W. Stubblefield,
E. E. Williams, and the anonymous re-
viewer of a previous manuscript for help-
ful criticisms of various kinds. I thank
P. Luykx and R. M. Syren for permission
to use their unpublished termite chromo-
some numbers. This work was supported
in part by National Institutes of Health
Grant GM07620-02 to Harvard Univer-

sity.

NOTE ADDED IN PROOF

Tooby (J. theor. Biol., 97: 557-576
[1982]) has recently argued that eusocial

610

insects are expected to have high rates of
recombination (and thus high chromo-
some numbers) because they live at high
population densities and should therefore
be subject to particularly intense selec-
tion for resistance to rapidly evolving
pathogens.

LITERATURE CITED A
(TEXT)

BENNETT, M. D., AND J. B. SmMiTH. 1976. Nuclear
DNA amounts in angiosperms. Phil. Trans.
Roy. Soc. Lond. B, 274: 227-274.

CHARLESWORTH, D., B. CHARLESWORTH, AND C.
STROBECK. 1977. Effects of selfing on selection
for recombination. Genetics, 86: 213-226.

DAWKINS, R. 1979. Twelve misunderstandings of
kin selection. Z. Tierpsychol., 51: 184-200.

HAMILTON, W. D. 1972. Altruism and related
phenomena, mainly in social insects. Ann.
Rev. Ecol. Syst., 3: 193-232.

HASKINS, C. P. 1970. Researches in the biology and
social behavior of primitive ants. In L. R.
Aronson, et al. (eds.), Development and Evolu-
tion of Behavior. Freeman.

HASKINS, C. P., AND R. M. WHELDEN. 1965.
“Queenlessness,” worker sibship, and colony
versus population structure in the formicid
genus Rhytidoponera. Psyche, 72: 87-112.

HILL, W. G., AND A. ROBERTSON. 1968. Linkage
disequilibrium in finite populations. Theoret.
Appl. Genet., 38: 226-231.

IMAI, H. T., R. H. CROZIER, AND R. W. TayLor. 1977.
Karyotype evolution in Australian ants.
Chromosoma, 59: 341-393.

JAIN, S. K. 1976. Patterns of survival and micro-
evolution in plant populations. In S. Karlin and
E. Nevo (eds.), Population Genetics and
Ecology. Academic Press.

Kimura, M. 1956. A model of a genetic system
which leads to closer linkage by natural selec-
tion. Evolution, 10: 278-287.

LANDE, R. 1979. Effective deme sizes during long-
term evolution estimated from rates of chromo-
somal rearrangement. Evolution, 33: 234-251.

LEVIN, D. A., AND H. KeErsTER. 1969. Density-
dependent gene dispersal in Liatris. Am. Nat.,
103: 61-73.

Lewis, K. R., AND B. JOHN. 1963. Chromosome
Marker. Little, Brown. nae

Loukas, M., C. B. KRIMBAS, AND Y. VERGINI. 1979.
The genetics of Drosophila subobscura popu-
lations. IX. Studies on linkage disequilibrium
ee four natural populations. Genetics, 93: 497-
523.

Advances in Herpetology and Evolutionary Biology

MALECOT, G. 1948. Les Mathématiques de |’Héré-

dité. Masson et Cie.

MAYNARD SMITH, J. 1978. The Evolution of Sex.

Cambridge University Press.

MICHENER, C. D. 1974. The Social Behavior of

Bees. Harvard University Press.

Murray, B. G. 1976. The cytology of the genus
Briza L. (Gramineae) II. Chiasma frequency,
polyploidy and interchange heterozygosity.
Chromosoma, 57: 81-93.

OunTA, T., AND M. Kimura. 1969. Linkage disequili-
brium at steady state determined by random
genetic drift and recurrent mutation. Genetics,
63: 229-238.

PRICE, G. R. 1970. Selection and covariance. Na-
ture, 227: 520-521.

Rossins, R. B. 1918. Some applications of mathe-
matics to breeding problems. II. Genetics, 3:
73-92.

SCHAAL, B. A. 1980. Measurement of gene flow in
Lupinus texensis. Nature, 284: 450-451.
SCHAAL, B. A., AND D. A. LEvIN. 1978. Morpho-
logical differentiation and neighborhood size
in Liatris cylindracea. Amer. J. Bot., 65: 923—

928.

SEGER, J. 1976. Evolution of responses to relative
homozygosity. Nature, 262: 578-580.

__. 1980. Ph.D. thesis. Department of Biology,
Harvard University.

___. 1981. Kinship and covariance. J. Theor. Biol.,
91: 191-123.

SHERMAN, P. W. 1979. Insect chromosome numbers
and eusociality. Am. Nat., 113: 925-935.
TALBOT, M. 1965. Populations of ants in a low field.

Insectes Sociaux 12: 19-48.

TAYLOR, R. W. 1978. Nothomyrmecia macrops: a
living-fossil ant. Science, 210: 979-985.
TrIvERS, R. L. 1974. Parent-offspring conflict.

Amer. Zool., 14: 249-264.

TRIVERS, R. L., AND H. Hare. 1976. Haplodiploidy
and the evolution of the social insects. Science,
191: 249-263.

VED BraT, S. 1965. Genetic systems in Alliwm III.
Meiosis and breeding systems. Heredity, 20:
325-339.

Waite, M. J. D. 1973. Animal Cytology and Evolu-
tion, 3rd edn. Cambridge University Press.

WILSON, E. O. 1958. Patchy distributions of ant
species in New Guinea rain forests. Psyche,
65: 26-38.

____.. 1963. Social modifications related to rareness
in ant species. Evolution, 17: 249-253.

___. 1971. The Insect Societies. Harvard Univer-
sity Press.

WRIGHT, S. 1978. Evolution and the Genetics of
Populations, Vol. 4. University of Chicago
Press.

ZARCHI, Y., G. SIMCHEN, J. HILLEL, AND T. SCHAAP.
1972. Chiasmata and the breeding system in
wild populations of diploid wheats. Chromo-
soma, 38: 77-94.

RELATEDNESS AND RECOMBINATION - Seger

LITERATURE CITED B

(FIGURE 5 AND TABLE 1)

COHEN, S., AND L. M. ROTH. 1970. Chromosome
numbers of the Blattaria. Ann. Ent. Soc. Am.,
63: 1520-1547.

Crain, W. R., E. H. DAVIDSON, AND R. J. BRITTEN.
1976. Contrasting pattems of DNA sequence
arrangement in Apis mellifera (Honeybee) and
Musca domestica (Housefly). Chromosoma,
39: 1-12.

Crozier, R. H. 1975. Animal Cytogenetics 3:
Insecta 7: Hymenoptera. Gebruder Bom-
traeger.

DANEHOLT, B., AND J.-E. EDSTROM. 1967. The con-
tent of deoxyribonucleic acid in individual
polytene chromosomes’ of Chironomus
tentans. Cytogenetics, 6: 350-356.

EFSTRATIADIS, A., W. R. CRAIN, R. J., BRITTEN, E. H.
DAVIDSON, AND F. C. KaratTos. 1976. DNA
sequence organization in the lepidopteran
Antheraea pernyi. Proc. Natl. Acad. Sci. USA,
73: 2289-2293.

FAHRENHORST, H. 1977. Nachweis wbereinstim-
mender chromosomen-zahlen (N=16) bei
allen 4 Apis-arten. Apidologie, 8: 89-100.

FONTANA, F., AND M. AMORELLI. 1975. Chromosome
complement of Kalotermes flavicollis Fabr.
(Isoptera: Kalotermitidae). Boll. Zool., 42: 99-
100.

Fox, D. P. 1970. A non-doubling DNA series in
somatic tissues of the locusts Schistocerca
gregaria (Forskal) and Locusta migratoria
(Linn.). Chromosoma, 29: 446-461.

GALL, J. G., H. C., MacGrecor, AND M. E.
KIDSTROM. 1969. Gene amplification in the
oocytes of dytiscid water beetles. Chromo-
soma, 26: 169-187.

GLANCEY, B. M., M. K. St. ROMAIN, AND R. H.
CROZIER. 1976. Chromosome numbers of the
red and black imported fire ants, Solenopsis
invicta and S. richteri. Ann. Ent. Soc. Am., 69:
469-470.

GOopDPASTURE, C. 1974. Karyology and taxonomy of
some species of eumenid wasps. J. Kansas Ent.
Soc., 47: 364-372.

GOoDPASTURE, C., AND E. E. GRrISSELL. 1975. A
karyological study of nine species of Torymus
(Hymenoptera: Torymidae). Can. J. Genet.
Cytol., 17: 413-422.

HInps, T. R., AND R. W. NEwBurGH. 1972. The DNA
of the thoracic muscle of Manducta sexta.
Biochim. Biophys. Acta, 281: 514-521.

HUNTER, K. W., AND A. C. BARTLETT. 1975. Chromo-
some number of the parasitic encyrtid Cop-
idosma truncatellum (Dalman). Ann. Ent. Soc.
Am., 68: 61-62.

611

IMAI, H. T., R. H. CROZIER, AND R. W. TaYLor. 1977.
Karyotype evolution in Australian ants.
Chromosoma, 59: 341-393.

JOHN, B., AND G. M. HeEwitr. 1966. Karyotype
stability and DNA variability in the Acrididae.
Chromosoma, 20: 155-172.

JORDAN, R. A., AND R. W. BROSEMER. 1974. Charac-
terization of DNA from three bee species. J.
Insect Physiol., 20: 2513-2520.

Jost, E., AND M. MAMELI. 1972. DNA content in
nine species of Nematocera with special refer-
ence to the sibling species of the Anopheles
maculipennis group and the Culex pipiens
group. Chromosoma, 37: 201-208.

LaGowskI, J. M., M.-Y., WANG Yu, H. S. FORREST,
AND C. D. LairD. 1973. Dispersity of repeat
DNA sequences in Oncopeltus fasciatus, an
organism with diffuse centromeres. Chromo-
soma, 43: 349-373.

LAIRD, C. D., AND B. J. MCCartTuHy. 1969. Molecular
characterization of the Drosophila genome.
Genetics, 63: 865-882.

LIMA-DE-FariA, A., T. GUSTAFSSON, AND H. JAWoR-
SKA. 1972. Amplification of ribosomal DNA in
Acheta. Il. The number of nucleotide pairs of
the chromosomes and chromomeres involved
in amplification. Hereditas, 73: 119-142.

MAKINO, S. 1951. An Atlas of the Chromosome
Numbers in Animals, 2nd edn. Iowa State
University Press.

MAXWELL, D. E. 1958. Sawfly cytology with empha-
sis upon the Diprionidae (Hymenoptera:
Symphyta). Proc. Tenth Int. Cong. Ent., 2:
961-978.

MCKITTRICK, F. A. 1965. A contribution to the under-
standing of cockroach-termite affinities. Ann.
Ent. Soc. Am., 58: 18-22.

Rascu, E. M. 1974. The DNA content of sperm and
hemocyte nuclei of the silkworm, Bombyx
mori L. Chromosoma, 45: 1-26.

Rascu, E. M., H. J. BARR, AND R. W. Rascu. 1971.
The DNA content of sperm of Drosophila
melanogaster. Chromosoma, 33: 1-18.

Rascu, E. M., J. D. Cassipy, AND R. C. KING. 1977.
Evidence for dosage compensation in
parthenogenic Hymenoptera. Chromosoma,
59: 323-340.

ROBERTS, B., AND S. ROBERTS. 1972. The DNA con-
tent of testes and malpighian tubule nuclei of
Tricholioprocta impatiens (Sarcophagidae,
Diptera). Chromosoma, 39: 83-91.

SMITH, S. G., AND N. VIRKKI. 1978. Animal Cyto-
genetics 3: Insecta 5: Coleoptera. Gebruder

Borntraeger.
SWIFT, H., AND R. KLEINFELD. 1953. DNA in grass-
hopper spermatogenesis, oogenesis, and

cleavage. Physiol. Zool., 26: 301-311.

SYREN, R. M., AND P. Luykx. 1977. Permanent
segmental interchange complex in the termite
Incisitermes schwarzi. Nature, 266: 167-168.

VINKE, P. P., AND J. P. TILQUIN. 1978. A sex-linked

612 Advances in Herpetology and Evolutionary Biology

ring quadrivalent in Termitidae (Isoptera). WHITE, M. J. D., AND G. C. WEBB. 1968. Origin and

Chromosoma, 67: 151-156. evolution of parthenogenetic reproduction in
Wuite, M. J. D. 1973. Animal Cytology and Evolu- the grasshopper Moraba virgo (Eumasticidae:
tion, 3rd edn. Cambridge University Press. Morabinae). Aust. J. Zool., 16: 647-671.

___. 1976. Animal Cytogenetics 3: Insecta 2: Blat- WILMORE, P. J., AND A. K. BROWN. 1975. Molecular
todea, Montodea, Isoptera, Grylloblattodea, properties of orthopteran DNA. Chromosoma,
Phasmatodea, Dermaptera and Embioptera. D1: 337-345.

Gebruder Borntraeger.

Is Sphenodon punctatus a Maladapted Relict?

CARL GANS!

ABSTRACT. The tuatara, Sphenodon punctatus, is
the only Recent member of the Rhynchocephalia.
Many aspects of its morphology suggest that it is
primitive among the reptiles, and the fossil record
indicates that its morphology has hardly changed
since the Triassic. Consequently, its peculiar phys-
iology (low metabolic rate correlated with low activ-
ity temperature), the relatively long period before
sexual maturity, and its longevity have sometimes
been interpreted as relicts of past conditions, pre-
sumably characterizing the ancestral group. Review
of recent work suggests that many of these special-
izations represent obvious adaptations to current
conditions. The tremendous diversity of adaptive
patterms seen in Recent lepidosaurians makes it
risky to use any small cohort of these animals to
infer conditions in various extinct forms.

INTRODUCTION

The tuatara, Sphenodon punctatus,
sole surviving member of the Rhyncho-
cephalia, is the only living non-squamate
lepidosaurian. Textbooks and the semi-
popular literature emphasize this point
and often note that it is a “living fossil,” a
“remnant” or “relict”; some even suggest
that it (i.e., its skeleton) has remained
unchanged since the Triassic. Other
comments give one the impression that
the occasional specimens of Sphenodon
seen in zoos were sampled from remnant
populations of slow and somehow inade-
quate relict reptiles, clinging for dear life
to a few cold and barren, windswept
rocks off the New Zealand coast.

It seems to be tempting to speculate
that the physiology of such relict animals

1 Division of Biological Sciences, The University
of Michigan, Ann Arbor, Michigan 48109, U.S.A.

might in some way have retained an
ancestral condition rather than being
clearly matched to present circum-
stances. For instance, Bogert (1953b)
notes that the very low body temperature
at which tuataras are active may have
implications for the thermal pattern that
existed in extinct members of the group.
More recently, Bakker (1971) suggests
that the thermal pattem probably reflects
a primitively low thermal preference.
Stebbins (1958) wonders whether parie-
tal eye function in the tuatara might have
a bearing on this function in “ancient
reptiles.” In contrast, Dawbin (1962)
notes that the physiology of Sphenodon
might well have developed quite inde-
pendently from its skeleton.

I here suggest that the astonishing
ecological and physiological traits of
Sphenodon are neither maladaptive nor
necessarily primitive, but are clearly
matched to current conditions; they need
not have implications for the specializa-
tions exhibited by the ancestral popula-
tions. It is my pleasure to dedicate this
essay to Professor Emest E. Williams, for
personal reasons and because of his ef-
forts at fostering an increased under-
standing of the biology of the lower tetra-
pods.

MORPHOLOGICAL ASPECTS

Sphenodon was early recognized as
morphologically primitive, because its
temporal region was diapsid and _ its
quadrate fixed (Gunther, 1867); this pat-
tern, as well as the acrodont jaws and the

614

absence of male intromittent organs, was
then considered to have been that of the
earliest reptiles. Comparison showed
that the Recent form closely resembles
the various Triassic fossils (Romer, 1956),
not only in the architecture of the skull,
but also in aspects of the postcranial
skeleton, such as number and propor-
tions of skeletal elements. This level of
similarity between Triassic and Recent
forms has held up as more rhyncho-
cephalian fossils became available
(Carroll, 1977; Romer, 1956, 1966). By
the Triassic, sphenodontids were recog-
nizably distinct from lizards (Robinson,
1973), which were already recorded in
the Upper Permian (Carroll, 1977).

Various diapsids were initially and in-
appropriately assigned to the Rhyncho-
cephalia (Romer, 1956). However, the
early rhynchocephalian forms Palacro-
don (South Africa, Lower Triassic), Poly-
sphenodon (Europe, Upper Triassic), and
Opistias (Wyoming, Upper Jurassic;
Simpson, 1926) are remarkably similar to
Sphenodon, as are the excellently pre-
served Homoeosaurus (late Jurassic of
Europe; cf. Hoffstetter, 1955; Romer,
1966) and Toxolophosaurus (Lower
Cretaceous, Montana; Throckmorton et
al., 1981). Indeed, the various rhyncho-
saurs and several levels of advanced
forms within the group differ in their
structure both from the early rhynchoce-
phalians and the modern form. Recent
work suggests that the rhynchosaurs may
be quite distinct from the Rhynchocepha-
lia sensu stricto (Carroll, 1977). However,
the Rhynchocephalia (including Spheno-
don) retain their assignment separate
from the lizards, which has been based
on such cranial characteristics as the
diapsid skull, the quadratojugal, the rigid
attachment of the quadrate, and the
acrodont dentition.

Perhaps the most important point in
the present context is that even the
Triassic sphenodontids were functionally
specialized, particularly in terms of their
feeding system. Their dentition consists
of three rows on each side, with the

Advances in Herpetology and Evolutionary Biology

dentary teeth closing between the maxil-
lary and palatine rows (Robinson, 1976).
Functional analysis of their masticatory
system indicates that the power stroke or
slow-closing component of each mastica-
tory cycle, is divided into two phases. An
initial vertical movement is followed by a
secondary anterior sliding movement of
the mandible (Gomiak et al., 1982;
Rosenberg and Gans, 1977). As the upper
and lower surfaces of the food object are
each held by the tooth cusps, the bolus is
bent or sheared as the jaws slide. The
triangular shape of the cusps permits the
teeth to dig into the prey; thus one sees
mice being decapitated (in captive ani-
mals) and obtains an understanding of
how these animals decapitate seabirds
(Crook, 1975). One further discovery is
the remarkably limited degree of free-
dom of the sphenodontid mandible.
There is almost no opportunity for lateral
movements, even when the mouth is
partially open.

The pattern of jaw action emphasizes
the extreme specialization of the mandi-
bular system. No lizard shows such cut-
ting ability. The most that has been noted
in the Sauria is a crushing or piercing ac-
tion. One superficially parallel action is
the leaf-cropping of agamid and iguanid
lizards (Throckmorton, 1976); however,
in that case the very small and multi-
lobed teeth serve as “pinking shears.”
Furthermore, the pattern in rhynchosaurs
appears to have shifted to grinding the
food; the various later forms are broad-
headed and often show multiple-rowed
and widened sets of teeth in the palatal
arrays (Sill, 1971).

ECOLOGY AND PHYSIOLOGY

Some years of laboratory observation
and a brief visit to the spectacular colony
on Stephens Island, New Zealand, indi-
cate that, whatever the physiology of
their remote ancestors, tuataras are cur-
rently successful animals, well adapted
to the circumstances of their current

SPHENODON: A MALADAPTED RELICT? -

environment. On Stephens Island, one of
some dozen sites where viable popula-
tions now survive, Sphenodon occurs at
an extremely high density (estimates
range from 20,000 to 100,000 specimens
on the 150 hectares, higher by orders of
magnitude than densities elsewhere [D.
G. Newman in litt.]). They are clearly the
top nonavian predators, feeding on
ground-dwelling arthropods, amphi-
bians, reptiles and birds (Crook, 1975;
Dawbin, 1962; Newman, 1977; Newman
et al., 1979; Walls, 1978). They crush and
dismember hard prey, crunching large
arthropods and carving up sea birds.
Their hunting pattem is more than a
passive sit-and-wait strategy. Individuals
were observed at night inspecting
crevices containing tree weta (Hemi-
deina crassidens, a 10+ cm long, armored
orthopteran); tuataras in such positions
immediately moved to attack other
insects when these were dropped close
to them.

The capacity of Sphenodon to operate
at low temperatures has often been
stressed, particularly after it was shown
that many other reptiles can raise their
body temperatures to levels from 25° to
42°C and that certain physiological
processes of other forms may not even
respond below 25°C (Dawson, 1975).
There can be little question that Spheno-
don are active nocturnally at body tem-
peratures between 6.2° and _ 16.0°C
(Bogert, 1953a, personal observation);
the hunting tuataras mentioned in the
preceding paragraph had body (cloacal)
temperatures near 14°C (and force-plate
measured activity in the laboratory at the
University of Illinois was found to be
highest at dusk and dawn; J. Heath and
R. G. Northcutt, personal communica-
tion). However, it has recently been con-
firmed that Sphenodon will bask, both in
the field and in the laboratory (Dawbin,
1949; Stebbins, 1958; Wemer and
Whitaker, 1978; Barwick, in litt.; Gans
and Regal, in litt.), and that basking
animals raise their body temperature to
28°C or more. Still, the energy require-

Gans 615

ments during activity appear to be very
low (Milligan, 1923; Wilson and Lee,
1970). Though tuataras have low body
temperatures, they are hardly lethargic,
thus Gunther’s (1867: after Smith) com-
ment that they are “slow and sluggish”
should be set aside.

In the densest populations, burrows
are packed at 2 to 3-meter spacings; this
apparently involves a complex social
interaction (Newman, 1978). Tuataras
produce vocalizations within their best
hearing range (Gans and Wever, 1976),
suggesting that these may be involved in
communication. The burrows that birds,
such as the fairy prion, Pachyptila turtur,
and the diving petrel, Pelecanoides uri-
natrix, construct during the nesting
season, are shared by Sphenodon. At
lower densities, numbers of petrels and
tuataras are correlated; however, the rela-
tive density of tuataras first becomes
constant and then drops as the concentra-
tion of petrels and burrows increases
further (> 0.3/ m?; Crook, 1974, 1975).
While this may indicate interaction, it
may also reflect different requirements.
The animals interact within their colo-
nies. Thus resident tuataras would some-
times emerge from tunnels or other sites
to attack and chase other tuataras that had
been displaced (being released away
from their homesite during the night).
Body temperatures of such animals range
between 14° and 16°C. Recent laboratory
observations document that males bite
each other when establishing dominance,
and bite females during mating (Gilling-
ham et al., ms.).

Many reports comment that tuataras
rarely bask in open areas during the day,
whether such areas are due to bird nest-
ing activity or cattle grazing. They do
bask in the open shade of the low forest,
but do not seem to move far from the
mouth of their burrows. They are cryp-
tically colored and may show some color
polymorphism yielding green and occa-
sional orange specimens (the latter color-
ation seemingly matching the patterns of
some local fungus Favolaschia caloce-

616

rata, that may have been recently intro-
duced). This suggests that sight-hunting
predators, such as hawks (Saint Girons et
al., 1981), may be significant. Even adult
tuataras are small enough (< 1 kg), so that
predation remains a potential problem.

The present range of the tuatara is
clearly less than that of even a thousand
years ago, when the form was widespread
on the main islands of New Zealand
(Robb, 1973, 1977). The arrival of man
and of the various rats that he introduced
changed the situation. Today tuataras
have been exterminated on the mainland
and on many of the coastal islands they
once occupied (Crook, 1973a, b). It is
uncertain how far inland the distribution
of Sphenodon originally extended. There
are numerous inland records, but most of
these derive from Quaternary (Rich,
1974, 1975) kitchen middens (Crook,
1975). Clearly the Polynesian invasion
affected these animals in three ways. The
first is that Sphenodon and other local
vertebrates- were hunted for food
(Davidson, 1972; Scott, 1973). The sec-
ond is that the invaders produced signifi-
cant vegetational and associated ecolog-
ical change (Jones, 1975; McGlone,
1978). The third is that the invaders intro-
duced Rattus exulans, which competes
for the same food objects and also preys
on the tuatara, particularly on its eggs and
most juvenile stages (Whitaker, 1978). On
some islands, these rats are obviously
limiting the reproductive capacity, as the
resident populations of Sphenodon con-
sists mainly of adults and may no longer
be self-sustaining (Crook, 1973a;
Whitaker, 1978). Nevertheless, tuataras
seem to do very well except in zones in-
vaded by the rats.

Another set of characteristics about
which there is substantial interest and
which may be associated with the low
body temperature is the longevity of the
form. The values of twenty years to
achieve sexual maturity and fifty years
during which growth continues (Crook,
1970; Dawbin, 1962) rank Sphenodon
among those reptiles with the slowest

Advances in Herpetology and Evolutionary Biology

development, though not necessarily
among those with the longest lifespan
(Case, 1978). In any case, the structure of
populations seems curious for a lepido-
saurian and is more like that noted in
island tortoises (or in trees). The large
number of first-year animals, noted in
some reptilian populations (Richmond,
1965), does not seem to be in evidence,
even on islands that lack rats (Crook,
1970, 1973a). Though growth is slow, not
all of their physiological processes are
slowed down, as Sphenodon are clearly
capable of rapid and protracted move-
ment.

INFERENCE FROM SPHENODON

The above set of circumstances permits
us to pose two obvious questions. 1) Is
Sphenodon well adapted to its current
environment (i.e., that which it occupied
before the disturbance induced by the
arrival of man)? 2) Are any of the phys-
iological specializations of this animal
likely to have been retained from its an-
cestral forms, and might they permit con-
clusions about adaptation or climates in
the Triassic?

The first question revolves around the
low body temperature of these animals.
The suggestion that this represents an
ancestral condition in a relict species is
probably unjustified for various reasons.
Recently, we have had to depart from the
earlier concept that all reptiles “prefer”
and are primarily active at body tempera-
tures above 30°C. Increasing numbers of
reptiles have been shown to be active at
body temperatures in the high 10’s and
low 20’s (cf. Gans, 1976; Avery, 1982):
other forms, such as geckos, feed nightly
at low temperatures and bask under
cover during the day. Also, certain pop-
ulations of lizards selectively occupy
environments in which they can never
achieve the basking temperatures prefer-
red by or available to populations occu-
pying adjacent sites (Huey and Slatkin,
1976). In such environments, the bio-

SPHENODON: A MALADAPTED RELICT? -

topes either lack accessible sources of
radiant energy or such radiant energy is
only available at times and at locations
where its cost, perhaps in energy re-
quired to reach it or exposure to preda-
tors while doing so, may be higher than
the benefit obtained.

What are the factors that restrict
Sphenodon to a low temperature regime?
The first factor appears to be the occur-
rence of predation. The hawks and other
birds of prey are generalist predators that
do not depend upon Sphenodon as a food
resource; thus the population size of
these predators is established, in part, by
other factors. However, these birds
represent a continuing threat to any
tuatara and hence limit the net gain, in
terms of increased digestive or other
physiological rate, achievable by bask-
ing. Reduction of the “preferred” body
temperature reduces the need to bask.

Another possible set of reasons for a
reduced body temperature reflects the
adaptive zone occupied by Sphenodon.
On Stephens Island, tuataras harvest the
fabulous energy resource reflected by the
enormous number of seabirds, the prions
and petrels that nest on coastal rocks and
islands. A direct harvest occurs by preda-
tion on the birds themselves; thus tua-
taras take significant quantities of their
eggs and offspring (Newman, 1978).
Adult birds are more vulnerable at night
than in the daytime (Walls, 1978). An
indirect harvest occurs, as the tuataras
feed mostly on the large arthropods that
are a by-product of the nesting colony.
The annual reworking of the soil during
the formation of nesting tunnels and the
deposit of masses of guano, food scraps
and dead chicks provide resources for
arthropod scavengers, as well as fertilizer
for the island vegetation that supports
crickets and beetles. The various prey
species are also nocturnal and must be
hunted when ambient temperatures are
low. Thus, Sphenodon appears to be
adapted for nocturnal predation.

Perhaps the longevity of Sphenodon
also represents an adaptation to an envi-

Gans 617

ronment in which all of the resources are
at the mercy of occasional shifts in the
patterns of oceanic currents. Very low
energy requirements, not only for metab-
olism but also for growth, could reduce
the effect of unpredictable “bad sea-
sons, when changes in ocean currents
cause a crash of the prey populations on
which the seabirds subsist.! (On steep,
high islands, such as Stephens, the vast
majority of individual Sphenodon would
lack access to the seashore, or to alterna-
tive food sources, during periods when
fewer or no birds breed. At the least, a
shift of populations toward the shore
would increase their exposure to preda-
tors. While the availability of varied
herbivorous insects might tend to buffer
short-term effects, all of the biota must
reflect the annual influx of birds.) How-
ever, prolonged immaturity would also
slow the recovery of populations de-
pleted during “bad seasons,” would pro-
long the time that the sensitive eggs and
juveniles are exposed to predation and
may simply be a “cold island” strategy,
possible only in the face of a most limited
suite of predators.

Whatever the basis of some of the
specializations now seen, the tuatara was
clearly a well-adapted form at the time
man first arrived in New Zealand and
remains sO now in a more restricted
range.

1 There seem to be relatively few data on long term
stability of the breeding success for most oceanic
birds (Murphy, 1936). Dunnet et al. (1979) docu-
ment gradual changes (over 28 years) in an Orkney
fulmar population, with the hatching success in
good years being more than twice that in bad ones.
On Peruvian islands (Hutchinson, 1950) breeding
may be interrupted and the changes in the numbers
of birds may be even more marked. On the islands
in question, the birds “always re-appear, although
their breeding success may vary from place to place
and year to year’ (J. Warham, also J. A. Bartle, both
in litt.; for similar comments on other species, see
Richdale, 1963, 1965). Comments suggest that New
Zealand’s rather drastic geological history
(Fleming, 1975a, b) may have led to past interrup-
tion of breeding on coastal islands.

618

The second question is whether the
pattern discerned may be deemed a
retention from an ancestral condition.
Such a retention would imply that
Sphenodon derives from a “low tempera-
ture’ stock that survived only in the cool
climate of these islands. Two kinds of
evidence regarding this point exist. First,
there is the documentation that New
Zealand underwent considerable clima-
tic variation during the Quaternary and
earlier (Fleming, 1979). Secondly, there
is the record of the unique lizard fauna of
New Zealand which is replete with
exceptional life histories. For instance,
an unpublished study by R. Barwick,
reports that the New Zealand gecko,
Hoplodactylus duwauceli, also has de-
layed sexual maturity (7 years) and an
extremely slow growth rate, with growth
continuing for more than 19 years (D. G.
Newman, in litt.) while another New
Zealand gecko, H. maculatus, lived more
than 30 years in captivity (D. G. Newman
in litt.). In these aspects, these animals
differ significantly from other geckos for
which such data are available (Bowler,
1977). Similar curiosities abound in other
New Zealand lizards. Thus Towns
(1975a, b) notes that the small noctumal
skink, Leiolopisma suteri, has one of the
longest developmental times in its fami-
ly, and the local geckos are unique in in-
cluding viviparous species (Robb, 1973).
It is of further interest that the tree weta
are also long-lived (D. G. Newman, in
litt.), but so are some other members of
this primitive orthopteran group (T. H.
Hubbell, personal communication).

Neither of these sets of evidence
would eliminate the concept that the
tuatara had retained its cold-temperature
physiology from its ancestral stock, or
that the Triassic or Cretaceous spheno-
dontids were cool-temperature animals.
On the other hand, they do not offer any
support for these hypotheses; indeed the
modifications of the local lizard fauna
make it unlikely that Sphenodon only

Advances in Herpetology and Evolutionary Biology

survived in New Zealand because of its
temperate climate (Bakker, 1971). In-
stead it suggests that these specializa-
tions most likely reflect current adapta-
tions (to a cool climate with both short-
term and long-term environmental fluc-
tuations). The specializations of Spheno-
don and local lizards also document the
extreme diversity seen in Recent
Lepidosauria. If Hoplodactylus were the
“lone survivor,” it would provide a most
inadequate indication of the physiologi-
cal responses of other geckos. This
example emphasizes the cautions neces-
sary in utilizing growth data, population
structure, and predator to prey ratios
pertinent to any one set of lepidosaurians
as a basis for inferring a uniform situation
in major groupings of reptiles now ex-
tinct.

ACKNOWLEDGMENTS

I am delighted to thank the New
Zealand Wildlife Service, in particular
Donald G. Newman, for arranging a trip
to Stephens Island and for his friendly
assistance and advice. J. Warham shared
his unpublished data on the breeding
success of sea birds. Many others partici-
pated in the discussions that helped
formulate and refine these ideas. C. M.
Bogert, R. L. Carroll, G. C. Gomiak,
Allen Greer, J. A. Hopson, R. W. Marlow,
D. G. Newman, J. Ostrom, R. Payne,
F. H. Pough, P. Pridmore, R. Shine, and
E. E. Williams read and commented on
the draft manuscript, though the respons-
ibility for its final formulation is clearly
my own. These travels were carried out
while a participant in a U.S./Australia
Cooperative Science Project; the travel to
Stephens Island was supported by a John
Simon Guggenheim Memorial Fellow-
ship and preparation of this manuscript
by National Science Foundation Grant
DEB 77-02605 and DHEW G-IROIDE
Onli:

SPHENODON: A MALADAPTED RELICT? -

LITERATURE CITED

AvERY, R. A. 1982. Field studies of body tempera-
tures and thermoregulation. In C. Gans and
H. F. Pough (eds.), “Biology of the Reptilia,”
12(4): 93-166. London, Academic Press.

BAKKER, R. T. 1971. Dinosaur physiology and the
origin of mammals. Evolution, 25(4): 636-658.

BOGERT, C. M. 1953a. Body temperatures of the
tuatara under natural conditions. Zoologica,
38(4): 63-64.

____. 1953b. The tuatara: Why is it a lone survivor?
Sci. Mo., 76: 163-170.

BOWLER, J. K. 1977. Longevity of reptiles and
amphibians in North American collections as
of 1 November 1975. Soc. Stud. Amph. Rept.
Herp. Circ., (6): i-iv + 32.

CaRROLL, R. L. 1977. The origin of lizards. Linn.
Soc. Symp. Ser., (4): 359-396.

CasE, T. J. 1978. On the evolution and adaptive
significance of postnatal growth rates in the
terrestrial vertebrates. Quart. Rev. Biol., 53(3):
243-282.

Crook, I. G. 1970. Tuatara. Wildlife (Wellington),
2: 16-19.

—.. 1973a. The tuatara, Sphenodon punctatus
Gray, on islands with and without populations
of the Polynesian rat, Rattus exulans (Peale).
Proc. N. Z. Ecol. Soc., 20: 115-120.

___. 1973b. Tuatara and the Polynesian rat. Wild-
life (Wellington), 4: 47-48.

__. 1974. Are tuataras dependent on petrels?
Wildlife (Wellington), 5: 43-46.

___.. 1975. The tuatara, pp. 331-352. In G. Kuschel
(ed.), “Biogeography and Ecology of New
Zealand.” The Hague, W. Junk, B. V.

DaviDson, J. M. 1972. Archaeological investiga-
tions on Motutapu island, New Zealand. Rec.
Aukland Inst. Mus., 9: 1-14.

DAwBIN, W. H. 1949. The tuatara . Tuatara, 2: 91-96.

1962. The tuatara in its natural habitat.
Endeavour, 21: 16-24.

Dawson, W. R. 1975. On the physiological signifi-
cance of the preferred body temperatures of
reptiles, pp. 443-473. In D. M. Gates and R. B.
Schmerl (eds.), “Perspectives in Biophysical
Ecology.” New York, Springer Verlag.

DUNNET, G. M., J. C. OLLASON, AND A.ANDERSON
1979. A 28-year study of breeding fulmars
Fulmarus glacialis in Orkney. Ibis, 121(3):
293-300.

FLEMING, C. A. 1975a. The geological history of
New Zealand and its biota, pp. 190. In G.
Kuschel (ed.), “Biogeography and Ecology in
New Zealand.” The Hague, Junk.

___. 1975b. The Quaternary record of New Zea-
land and Australia, pp. 1-20. In R. P. Suggate
and M. M. Cresswell (eds.), “Quaternary
Studies.” Wellington, Roy. Soc. New Zealand.

—__.. 1979. “The Geological History of New Zea-

Gans 619

land and its Life.” New Zealand, Auckland
Univ. Press.

— Gans, C. 1976. Aspects of the biology of uropeltid

snakes, pp. 191-204. In A. d’A. Bellairs and
C. Barry Cox (eds.), “Morphology and Biology
of Reptiles.”’ London, Linn. Soc. Symp. Ser (3).

GANS, C., AND E. G. WEVER. 1976. The ear and hear-
ing in Sphenodon punctatus. Proc. Nat. Acad.
Sci., 73(11): 4244-4246.

GILLINGHAM, J. C., C. GANS, AND D. L. CLark. Ms.
Courtship, mating and male combat in tuatara,
Sphenodon punctatus. J. Herpetol., submitted.

GORNIAK, G. C., H. ROSENBERG, AND C. Gans. 1982.
Muscle activity during mastication in the
tuatara, Sphenodon punctatus. J. Morphol.,

GUNTHER, A. C. L. G. 1867. Contribution to the

anatomy of Hatteria (Rhynchocephalus,
Owen). Phil. Trans. Roy. Soc. London., 157(2):
595-629.

HOFFSTETTER, R. 1955. Rhynchocephalia, pp. 556—
576. In J. Piveteau, (ed.), “Traité de Paléontol-
ogie. Tome V. Amphibiens, Reptiles,
Oiseaux.” Paris, Masson et Cie.

HUuEY, R. B., AND M. SLATKIN. 1976. Costs and bene-
fits of lizard thermoregulation. Q. Rev. Biol.,
ol: 363-381.

HUTCHINSON, B. E. 1950. Survey of contemporary
knowledge of biogeochemistry. 3. The bio-
geochemistry of vertebrate excretion. Bull.
Amer. Mus. Nat. Hist., 96: xviii + 554.

JONES, R. 1975. The Neolithic, Paleolithic and the
hunting gardeners: Man and land in the Anti-
podes, pp. 21-34. In R. P. Suggate and M. M.
Cresswell (eds.), “Quaternary Studies.” Wel-
lington, Roy. Soc. New Zealand.

MCGLONE, M. S. 1978. Forest destruction by early
Polynesians, Lake Poukawa, Hawkes Bay,
New Zealand. J. Roy. Soc. New Zealand, 8(3):
275-281.

MILLIGAN, R. R. D. 1923. The respiration and
metabolism of the tuatara. Austr. Assn. Adv.
Sci. Rept., 16: 404406.

Murpny, R. C. 1936. Oceanic Birds of South Amer-
ica. 2 Vols. New York, Amer. Mus. Nat. Hist.

NEwMaN, D. G. 1977. Some evidence of the preda-
tion of Hamilton’s frog (Leiopelma hamiltoni
(McCulloch)) by tuatara (Sphenodon puncta-
tus (Grey)) on Stephens Island. Proc. N. Z.
Ecol. Soc., 24: 43-47.

____. 1978. Tuataras and petrels. Wildlife (Welling-
ton), (9): 16-23.

NEWMAN, D. G., I. G. CROOK, AND L. R. MORAN.
1979. Some recommendations on the captive
maintenance of tuataras Sphenodon punctatus
based on observations in the field. Int. Zoo
Yearb., 68-74.

Ricu, T. H. 1974. Soc. Vert. Paleontol. News Bull.,
(102): 43-44.

___.. 1975. Soc. Vert. Paleontol. News Bull., (105):
35-36.

620

RICHDALE, L. E. 1963. Biology of the sooty shear-
water, Puffinus griseus. Proc. Zool. Soc. Lond.,
14]: 1-117.

____. 1965. Biology of the birds of Whero Island,
New Zealand with special reference to the div-
ing petrel and the broad-billed prion on Whero
Island, New Zealand. Trans. Zool. Soc. Lond.,
31: 87-155.

RICHMOND, N. D. 1965. Perhaps juvenile dinosaurs
were always scarce. J. Paleontol., 39(3):
503-505.

Ross, J. 1973. Reptiles and Amphibia, pp. 285-303.
In G. R. Williams (ed.), “The Natural History
of New Zealand.” Wellington, Reed.

___. 1977. “The Tuatara.” Durham, Meadowfield.

ROBINSON, P. L. 1967. The evolution of the Lacerti-
lia, pp. 395407. In “Problemes actuels de
Paléontologie.” (Evolution des Vertébrés).
Col. Int. Cent. Natl. Rech. Sci.

____. 1973. A problematic reptile from the British
Upper Trias. J. Geol. Soc. London., 129: 457-
459.

___. 1976. How Sphenodon and Uromastyx grow
their teeth and use then, pp. 43-64. In A. d’A.
Bellairs and C. Barry Cox (eds.), “Morphology
and Biology of Reptiles.” Linn Soc. Symp. Ser.
(3), London.

RoMER, A. S. 1956. “Osteology of the Reptilia.”
Chicago, Univ. Chicago Press.

__. 1966. “Vertebrate Paleontology. Third Edi-
tion.” Chicago, Univ. Chicago Press.

ROSENBERG, H. I. AND C. Gans. 1977. Preliminary
analysis of mastication in Sphenodon puncta-
tus. (Abstract). Amer. Zool., 17(4): 871.

SAINT GIRONS, H., B. D. BELL AND D. G. NEWMAN.
1981. Brief observations on the activity and
thermoregulation of the tuatara, Sphenodon
punctatus (Reptilia: Rhynchocephalia), on
Stephens Island, Cook Strait. New Zealand J.
Zool.,

ScoTT, S. D. 1973. Excavations on the “Sunde Site”
N38/24, Motutapu island, New Zealand. Rec.
Auckland Inst. Mus., 7: 13-30.

Advances in Herpetology and Evolutionary Biology

SILL, W. D. 1971. Functional morphology of the
rhynchosaur skull. Forma et functio, 4: 303—
318.

SIMPSON, G. G. 1926. American terrestrial rhyn-
chocephalia. Amer. J. Sci., 12(5): 12-16.
STEBBINS, R. 1958. An experimental study of the
“third eye” in the tuatara. Copeia, 1958(3):

183-190.

THROCKMORTON, G. S. 1976. Oral food processing in
two herbivorous lizards, Iguana iguana
(Iguanidae) and Uromastix aegyptius
(Agamidae). J. Morphol. 148: 365-390.

THROCKMORTON, G. S., J. A. HOPSON, AND P. PARKS.
1981. A_ redescription of Toxolophosaurus
cloudi Olson, a lower Cretaceous herbivorous
sphenodontid reptile. J. Paleontol. (in press).

Towns, D. R. 1975a. Ecology of the black shore
skink Leiolopisma suteri __ (Lacertilia:
Scincidae), in boulder beach habitats. New
Zealand J. Zool., 2(4): 389-407.

___. 1975b. Reproduction and growth of the black
shore skink Leiolopisma suteri (Lacertilia:
Scincidae), in north-eastem New Zealand.
New Zealand J. Zool. 2(4): 409-423.

WALLS, G. Y. 1978. The influence of the tuatara on
fairy prion breeding on Stephens island, Cook
Strait. New Zealand J. Ecol., 1: 91-98.

WERNER, Y. L., AND A. H. WHITAKER. 1978. Observa-
tions and comments on the body temperatures
of some New Zealand reptiles. New Zealand J.
Zool., 5: 375-393.

WHITAKER A. H. 1978. The effects of rodents on
reptiles and amphibians, pp. 75-88. In “The
Ecology and Control of Rodents in New Zea-
land Nature Reserves.” Wellington Dept. of
Lands and Survey Information, ser. 4.

WILSON, K. J. AND A. K. LEE. 1970. Changes in
oxygen consumption and heart-rate with activ-
ity and body temperature in the tuatara
Sphenodon punctatum. Comp. Biochem.
Physiol., 33: 311-322.

Chromosomal C-Banding in Anolis grahami

J. A. BLAKE?

ABSTRACT. The technique of C-banding is applied
to chromosomes of Anolis grahami, an iguanine
lizard showing large amounts of chromosomal varia-
tion within and between populations. Differential
staining of one chromosome pair distinguished one
sample from others. Variation both in diploid num-
ber and in amount and distribution of heterochro-
matin is noted among the three populations samp-

led.

INTRODUCTION

Although banding techniques have
been used extensively in chromosomal
analyses of mammals since 1969, they
have not been readily adaptable to other
groups. Within the reptiles, banding pro-
cedures have been utilized to examine
sex chromosomes in snakes (Becak,
1972), and have revealed cryptic sex
chromosomal differentiation of lizards
(Bull, 1978). Comparative karyology
of snakes (Stock and Mengden, 1975;
Mengden and Stock, 1980) and turtles
(Bickham and Baker, 1976; Sites et al.,
1979) have been discussed in light of C-,
G-, and AgNOR chromosomal banding
studies. The identification of chromo-
some arms (linkage groups) greatly en-
hances their usefulness as taxonomic
characters revealing information about
the evolution of the genome. C-banding,
or centromeric banding, allows deter-
mination of the location and amount of
heterochromatin present (Arrighi and
Hsu, 1971; Chen and Ruddle, 1971).

1Museum of Comparative Zoology, Harvard Uni-
versity, Cambridge, Massachusetts 02138, U.S.A.

In this paper, C-banding is used to
identify chromosome groups in the
iguanine lizard, Anolis grahami, a beta-
anole native to Jamaica. Two subspecies
are recognized on the basis of distinct
color and morphological differences
(Lynn and Grant, 1940; Underwood and
Williams, 1959). Anolis g. grahami is
distributed throughout most of Jamaica,
excluding the eastern coast. Anolis g.
aquarum is restricted to the moist north-
eastern and eastern areas of Jamaica.
There are complex interactions in areas
of sympatry (Underwood and Williams,
1959). Chromosomal variation within
Anolis grahami was first described by
Hall (1974), who reported extensive poly-
morphism reflecting a ‘pattern of con-
siderable complexity’ involving both
occasional accessory chromosomes and
fission of up to three pairs of macrochro-
mosomes. My work has confirmed and
extended the observations of Hall and his
student Espinoza revealing chromosome
numbers ranging from 2n = 30 to 2n = 37
(Blake, 1981). The purpose of this com-
parison of C-banding is to examine the
types of information available from C-
banding studies in anoles, and to initiate
discussion of the heterochromatic varia-
tion which is apparent in Anolis grahami.
This report of the use of banding tech-
niques to study chromosomal variation in
anoles is a logical extension of the great
amount of work concerning chromosomal
diversity and its contribution to the study
of systematic relationships in lizards
which has emanated from the laboratory
of Ernest E. Williams (e.g., Gorman et al.,
1968; Gorman and Atkins, 1966, 1968;

6 99

Webster et al., 1972; Hall and Selander,
1973; Paull et al., 1976).

MATERIALS AND METHODS

Specimens of Anolis grahami were col-
lected on Jamaica from the vicinity of
Anchovy (N = 6, St. James Parish, A. g.
grahami), Mandeville (N = 5, St. Thomas
Parish, A. g. grahami), and Bath (N = 3,
St. Thomas Parish, A. g. aquarum) during
March 1980 (Fig. 1). All animals used in
this study are deposited in the MCZ. One
male from each population was C-banded
(MCZ 158586, 158594, 158643).

Chromosome samples were prepared
in the field according to procedures de-
scribed by Hall and Selander (1973).
Animals were injected with colchicine 12
to 24 hours before dissection. Bone mar-
row, testes, and spleen were removed
and macerated in 2.2% NaCitrate. Sus-
pensions were centrifuged and hypotonic
solution (1.8% NaCitrate) added to each
pellet, which was then suspended and
held for 12 minutes before centrifuga-
tion. After centrifugation (700 rpm for 7
minutes in an IEC tabletop centrifuge),
fresh Carnoy’s fixative was added drop
by drop to the pellet. This preparation
can be stored in 2 mls of fixative at 4°C for
at least 6 months. A drop of suspension in
fresh fixative was deposited on an acid-
cleansed slide which had been drained
after being held in ice-cold 50% metha-
nol. The slide was then tilted and air
dried.

Diploid number was determined from
a minimum of 20 spreads in each animal.

C-banding technique is modified from
that of Sumner (1972) and Schmid (1978).
Four- to seven-day-old slides are treated
in 0.2N HC1 for 45 minutes and rinsed in
distilled deionized water (d.d. H,0). A
series of slides is subjected to 5%
Ba(OH), at 37°C for times ranging from 1
to 5 minutes at one minute increments.
After rinsing in d.d. H,O, the slides are
incubated in 2XSSC (1X SSC is 0.15M
NaCl, 0.015M NaCitrate) pH = 7.0 at

Advances in Herpetology and Evolutionary Biology

*Mandeville

Figure 1. Map of Jamaica showing localities where
animals used in this study were collected.

60°C for 1 hour. The slides are finally
rinsed and stained for 10 minutes in
Giemasa/citric acid stain. Selected
spreads were photographed on a Leitz
compound microscope using Kodak
Tech-Pan film.

RESULTS

Diploid Number. Five _ individuals
from Anchovy showed a diploid number
of 36 with 20 macrochromosomes and 16
microchromosomes. One individual from
Anchovy had a diploid number of 37.
Individuals from Mandeville showed di-
ploid numbers of 30(2), 31(2), and 32(1)
(Table 1). Of the three individuals from
Bath, one had a diploid number of 30; the
other two had 31. All individuals have 16
microchromosomes. The _ differences
appear in the number of macrochromo-
somes. As the number of macrochromo-
somes increases, so does the number of
telocentrics.

Centromeric Heterochromatin. All
metacentrics showed centromeric band-
ing indicative of centromeric hetero-
chromatin (Figs. 2, 3, 4). In the Anchovy
individual, the largest pair of metacen-
trics shows a difference in intensity of
staining in the centromeric region (Fig.
2). One chromosome is heavily stained
at the centromere. The other is less in-
tensely stained, and the centromere and
blocks of heterochromatin on either side
of the centromere are distinguishable
from each other. This heteromorphism
was observed in five different C-banded
prints, as well as in C-band spreads of

TABLE 1. GENERAL KARYOTYPE DESCRIPTIONS FOR

ANOLIS GRAHAMI FROM THREE JAMAICAN LOCALITIES.

V = metacentrics
sV = small metacentrics
I = telocentrics
A = acrocentrics

Locality N 2n m = microchromosomes

Anchovy 6 36(5) 4V + 4sV + 121+ 16m
37(1) 3V +4sV + 141 + 16m
Mandeville 5 30(2) 10V + 4sV + 16m
31(2) 9V+4sV+ 21+ 16m
32(1) 8V+4sV+ 41+ 16m
Bath 3 30(1) 10V + 4sV + 16m
31(2) 1A+ 8V+4sV+ 21+ 16m

CHROMOSOMAL C-BANDING IN ANOLIS GRAHAMI + Blake 623
~» | f
| id
e * r)
; ® | * :
. ° ,
# °
Cd Lei

he *
; *2
‘ % =
ig 4 & v7 a
a © ~~. ™ = ~
2 *& = rd
a * bl * ¢
se 9 * @ " nt
¥ " ‘i
a ,* «a &
& e ge
ie
‘A
a *
*

Figure 2. C-Banded karyotype of a 6 Anolis grahami
from Anchovy. Arrows show differential centromeric
staining of large metacentric pair. 2n = 36.

individuals of another Anolis g. grahami
population at Laughlands, St. Ann Parish
not reported here. The differential stain-
ing is not observed in the specimens from
Mandeville (Fig. 3) or Bath (Fig. 4). The
telocentrics also show centromeric stain-
ing with two prominent blocks of hetero-
chromatin distinguishable in the centro-
meric region (Fig. 2).

Telomeric Heterochromatin. All telo-
mers are stained, although some are
stained more intensely than others. This
can be diagnostic, as in the case of the
largest metacentrics of the Anchovy
karyotype (Fig. 2). Here the largest pair is
distinguishable from other metacentrics
not only on the basis of size, but also on

Figure 3. C-Banded karyotype of a 3 Anolis grahami
from Mandeville. Arrows show absence of differential
centromeric staining in large metacentric pair. 2n = 30.

Figure 4. C-Banded karyotype of a ¢ Anolis grahami
from Bath. Only 15 microchromosomes are observed,
the 16th is presumed to be covered by other chromo-
somes. The arrow indicates the heterochromatic
acrocentric arm. 2n = 31.

the basis of telomeric staining. One
telomeric region of the chromosomes is
heavily stained, the other is not. This is
repeatedly observed, and is apparent also
in the individuals from Mandeville and
Bath. In Mandeville and Bath such dis-
tinction is more important in identifying

624

chromosomes as there are three pairs of
metacentrics of approximately the same
size in these populations.

Whole-Arm Heterochromatin. In all
cases, the two pairs of small metacentrics
are distinguishable on the basis of
whole-arm staining (Figs. 2, 3, 4). One
pair is heavily stained, the other is not. In
many spreads, differences in staining be-
tween members of the darkly staining
pair was often observed. One member of
the pair stained heavily over both arms,
the other seemed to stain only over one
arm. This distinction may be important as
it is these small metacentrics which are
involved in sex chromosome arrange-
ments in other grahami series species.

One of the macrochromosomes in the
Bath spread appears to have a hetero-
chromatic arm. The presence of an acro-
centric chromosome with a heterochro-
matic short arm in A. g. aquarum popula-
tions is under further investigation.

Microchromosomes. The micro-
chromosomes are stained in all spreads.
The heterochromatic nature of the micro-
chromosomes is especially apparent in
meiotic metaphase I spreads (Blake,
1981). Microchromosomes are an impor-
tant part of sex chromosome systems in
Anolis when these have been observed
(Gorman et al., 1968; Gorman and Atkins,
1969), and the lack of distinction among
the microchromosomes is therefore of
special interest.

DISCUSSION

Anolis grahami presents a complex
case of chromosomal polymorphism.
Subspecies designation does not clarify
the variation observed. Anolis g. grahami
diploid numbers range from 30 to 32 at
Mandeville, to 37 at Anchovy. Popula-
tions with intermediate numbers of
chromosomes exist, and some have been
extensively studied (Blake, 1981).
Whether the polymorphism evident at
Mandeville is balanced and stable can
not be decided here. The presence of

Advances in Herpetology and Evolutionary Biology

such polymorphism, both at Mandeville,
and in the Anolis g. aquarum at Bath, is
indicative both of the widespread ten-
dency for karyotypic change in this
species, and of the presence of chromo-
somal polymorphisms throughout its
range.

Such conditions of chromosomal poly-
morphism are not unknown in Anolis.
Anolis monticola provides a_ parallel
example of fissioning of the karyotype
and presence of polymorphisms (Webster
et al., 1972). In that case, the primitive
karyotype of 12M + 24m is altered
through fission events to 48M + 24m.
Other monticola series species also
appear to have undergone karyotypic
changes through fission events (Webster
et al., 1972).

Anolis grahami is a member of the
grahami series, a group of closely related
beta-anoles inhabiting Jamaica. Within
this series, A. lineatopus, A. opalinus,
and A. garmani have diploid numbers of
30, 14M + 16m, although some poly-
morphism in A. opalinus is suspected
(Gorman and Atkins, 1968a; Gorman,
1969). Anolis conspersus, another mem-
ber of this series from nearby Grand
Cayman Island, shows sex chromosomal
heteromorphism, with the males having a
small acrocentric and metacentric in
contrast to two small metacentrics in the
female (Gorman and Atkins, 1968b). Most
of these determinations were based on
small sample sizes and, in light of current
information concerning the variable
Anolis grahami karyotype, warrant fur-
ther investigation.

The presence of differential staining of
the centromere of the large metacentric
of the Anchovy individual is strongly
reminiscent of the kind of variation re-
ported in the lizard Cnemidophorus
tigris by Bull (1978). In that case, Bull
demonstrated a sex difference in the
staining pattern and designated the
chromosomes as cryptic sex chromo-
somes, an intermediate stage in sex
chromosomal differentiation. The obser-
vation that such differential staining is

CHROMOSOMAL C-BANDING IN ANOLIS GRAHAMI « Blake

not observed in the three males studied
here makes such an argument unlikely in
this case. In addition, the C-banding pat-
tern in female Anolis grahami is not
known. The absence of repeatability of
such differential staining in different
populations is warning that the karyo-
typic description of a species must be
based on a fair sampling of the popula-
tion.

ACKNOWLEDGMENTS

I thank S. Abmeyr for her assistance in
micro-photographic procedures and S. M.
Case for critical reading of the manu-
script. This work was supported by Na-
tional Science Foundation Grant DEB 79-
05837 to E. E. Williams, and by the
Anderson Fund.

LITERATURE CITED

ARRIGHI, F. E.., AND T. C. Hsu. 1971. Localization of
heterochromatin in human chromosomes.
Cytogenetics, 10: 81-86.

Becak, W. 1972. W-sex chromatin fluorescence in
snakes. Experentia, 28: 228-229.

BICKHAM, J. W., AND R. J. BAKER. 1976. Chromo-
some homology and evolution of emydid
turtles. Chromosoma, 54: 201-219.

BLAKE, J. A. 1981. Chromosomal variation in the
Jamaican lizard Anolis grahami. Ph.D. Thesis,
Dept. of Biol., Harvard Univ., Cambridge.

BULL, J. 1978. Sex chromosome differentiation: An
intermediate stage in a lizard. Can. J. Genet.
Cytol., 20: 205-209.

CHEN, T. R., AND F. H. RUDDLE. 1971. Karyotype
analysis utilizing differentially stained consti-
tuitive heterochromatin of human and murine
chromosomes. Chromosoma, 34: 51-72.

Gorman, G. C. 1969. Mammalian Chromosome
Newsletter. Vol. 10, pp. 222-224.

GorMaAN, G. C., AND L. ATKINS. 1966. Chromosomal
heteromorphism in some male lizards of the
genus Anolis. Am. Natur. 100: 579-583.

____. 1968a. New karyotypic data for 16 species of
Anolis from Cuba, Jamaica, and the Cayman
Islands. Herpetologica, 24: 13-20.

625

____. 1968b. Confirmation of an X-Y sex determin-
ing mechanism in lizards (Anolis). Copeia, 1:
159-160.

____. 1969. The Zoogeography of Lesser Antillean
Anolis lizards. An Analysis Based upon
Chromosomes and Lactic Dehydrogenases.
Bull. Mus. Comp. Zool., 138: 53-80.

GORMAN, G. C., R. THOMAS, AND L. ATKINS. 1968.
Intra- and interspecific chromosome variation
in the lizard Anolis cristatellus and its closest
relatives. Breviora Mus. Comp. Zool. No. 293,
pp. 1-13.

HALL W. P. 1974. 2nd Anolis Newsletter. E. E.
Williams (ed.), 2 pp.

HALL, W. P., AND R. K. SELANDER. 1973. Hybridiza-
tion of karyotypically differentiated popula-
tions in the Sceloporus grammicus complex
(Iguanidae). Evolution, 27: 226-242.

Lynn, W. G., AND C. GRANT. 1940. The Herpetology
of Jamaica. Bull. Inst. Jamaica Sci. Ser. No. 1,
pp. 1-148.

MENGDON, G. A., AND A. D. Stock. 1980. Chromo-
somal evolution in serpentes: A comparison of
G and C chromosome banding pattems of
some colubrid and boid genera. Chromosome,
79: 53-64.

PAULL, D., E. E. WILLIAMS, AND W. P. HALL. 1976.
Lizard karyotypes from the Galapagos Island:
(Chromosomes in phylogeny and evolution.)
Breviora, Mus. Comp. Zool. No. 441, pp. 1-31.

SCHMID, M. 1978. Chromosome banding in
Amphibia I. Constitutive heterochromatin and
nucleolus organizer regions in Bufo and Hyla.
Chromosoma, 66: 361-388.

SITES, J. W., J. W. BICKHAM, M. W. HAIDUK, AND
J. B. IBERSON. 1979. Banded karyotypes of six
taxa of kinosternid turtles. Copeia, 4: 692-698.

Stock, A. D., AND G. A. MENGDON. 1975. Chromo-
some banding pattern conservatism in birds
and non-homology of chromosome banding
patterns between birds, turtles, snakes and
amphibians. Chromosoma, 50: 69-77.

SUMNER, A. J. 1972. A simple technique for demon-
strating centromeric heterochromatin. Exp.
Cell Res., 75, 304-306.

SUMNER, A. T., H. J. EVANS, AND R. A. BUCKLAND.
1971. A new technique for distinguishing be-
tween human chromosomes. Nature N. B., 232:
31-32.

UNDERWOOD, G., AND E. E. WILLIAMS. 1959. The
Anoline Lizards of Jamaica. Bull. Inst. Jamaica
Sci. Ser. No. 9, pp. 1-48.

WEBSTER, T. P., W. P. HALL, AND E. E. WILLIAMS.
1972. Fission in the evolution of lizard karyo-
type. Science, 177: 611-613.

The Relationships of the Puerto Rican Anolis:
Electrophoretic and Karyotypic Studies

GEORGE C. GORMAN!
DONALD BUTH?
MICHAEL SOULE?

S. Y. YANG4

ABSTRACT. Relationships of the Puerto Rican anoles
were investigated through both phenetic and clad-
istic treatments of two independently generated
multi-locus allozyme data sets. These findings are
compared with a phylogeny developed from pre-
viously reported karyotypic data. Phylogenetic
treatment of the data supports a rather conventional
view of the relationships of the Puerto Rican ano-
line radiation. There is electrophoretic support for
the recognition of three major karyotypically
limited lineages: 1) Anolis evermanni and A.
stratulus; 2) ‘A. cooki and A. cristatellus; and 3)
A. gundlachi, A. poncensis, A. pulchellus, and A.
krugi, although A. gundlachi clusters with the latter
group based on a karyotypic rather than any elec-
trophoretic character. Within the third lineage, A.
poncensis is the sister-taxon of a group comprised of
A. krugi and A. pulchellus, and these three “grass-
bush” anoles are in a sister group relationship to A.
gundlachi.

INTRODUCTION

There are ten species of Anolis lizards
on Puerto Rico, all endemic to the island
bank. This adaptive radiation has re-
ceived considerable attention from

12 Department of Biology, University of Califor-
nia (UCLA), Los Angeles, California 90024, U.S.A.

3 Department of Biology, University of California
(UCSD), La Jolla, California 92037, U.S.A. Present
Address: Institute for Transcultural Studies, 905
South Normandie Avenue, Los Angeles, California,
90006, U.S.A.

4Department of Biology, University of California
(UCSD), La Jolla, California 92037, U.S.A. Present
Address: Department of Biology, Inha University,
Incheon, Korea.

evolutionary biologists in the past two
decades. Rand (1964) studied ecological
niche partitioning among the species,
and Schoener and Schoener (1971) fur-
ther elaborated upon this.

In 1972, Ernest E. Williams published
a stimulating paper on the anoline radia-
tion of Puerto Rico. In that paper he
presented a model based largely upon
ecological principles that described how
the Puerto Rican anoline fauna may have
evolved in place following an_ initial
colonization event. Subsequent analyses
tend to support much of the model (e.g.,
Gorman et al., 1980a). It now appears that
there is a primary radiation accounting
for eight of the species. Two additional
species (Anolis cuvieri and A. occultus)
that appear to have more primitive rela-
tives living on the island of Hispaniola
may have independently colonized
Puerto Rico.

The morphology of most Greater
Antillean Anolis is highly correlated with
their ecology. Williams (1972) assigned
the Anolis of Puerto Rico to various
“ecomorph”’ categories. Anolis cuvieri is
a “crown-giant” and A. occultus is a
“twig-dwarf’ species. The other
ecomorph categories each have several
species which replace one another by
climatic niche. The three “trunk-ground”
species, A. gundlachi, A. cristatellus, and
A. cooki, are distributed in the rainforest,

RELATIONSHIPS OF THE PUERTO RICAN ANOLIS - Gorman et al.

lowland mesic areas, and the xeric
southwestern part of the island, respec-
tively. The three “grass-bush” species, A.
krugi, A. pulchellus, and A. poncensis,
replace one another similarly. Finally,
there is a pair of “trunk-crown” species:
Anolis evermanni is the typical upland
species, and A. stratulus is widespread in
the lowlands.

Only the three widespread lowland
(mesic) species occur on the islands east
of Puerto Rico that form part of the Puerto
Rico Bank. (Politically, some of these is-
lands are part of Puerto Rico, the re-
mainder being the United States and
British Virgin Islands. The island of St.
Croix, of the U.S.V.I., is not part of the
Puerto Rico Bank, and has an endemic
anole.)

In 1976 Williams published a formal
classification of West Indian Anolis. The
classification closely follows the oste-
ological interpretations of Etheridge
(1960). In this classification, the genus is
divided into two sections (alpha and
beta); the alpha section is divided into
two subsections (carolinensis and
punctatus). All of the Puerto Rican
anoles belong to the alpha section, and
all but one belong to the punctatus sub-
section. The diminutive Anolis occultus
is the only Puerto Rican member of the
carolinensis subsection.

The remaining nine species were
placed in three different series. Anolis
cuvieri belongs to the cuvieri series,
along with several Hispaniolan giant
anoles, and the possibly extinct giant
anole A. roosevelti of Culebra (a small
island that is part of the Puerto Rico
Bank). The “trunk-ground” and “grass-
bush” anoles were placed in the crista-
tellus series. This series was divided into
two subseries—the cristatellus subseries
and the cybotes subseries. The former
contains only the six Puerto Rican spe-
cies plus two extralimital forms that are
closely allied with cristatellus (Gorman
et al., 1980b). The cybotes subseries is a
Hispaniolan radiation of a single super-
species. The “trunk-crown” anoles of

627

Puerto Rico were placed in the bimacu-
latus series. This series, as defined by
Williams (1976), is represented through-
out the northern part of the Lesser
Antilles, the island of St. Croix, and on
Hispaniola where there is a single radia-
tion of closely related species (the dis-
tichoids). In addition, the Hispaniolan A.
distichus has colonized some of the
Bahamas and Florida.

Interpretation of data obtained from
albumin immunology, as well as inter-
pretation of karyotype data, has raised
some questions about the classification
proposed by Williams (1976). Shochat
(1976) and Shochat and Dessauer (1981)
questioned the alpha-beta dichotomy;
Wyles and Gorman (1980) raised some
questions about the division of the alpha
section into two subsections. The latter
study revealed that A. cybotes is not
closest in albumin immunological dis-
tance or karyotype to the cristatellus
radiation, but that its closest affinities
appears to be with other Hispaniolan
anoles, some of which were placed by
Williams in the carolinensis subsection.

Gorman et al. (1980a) presented a
cladistic analysis of eastern Caribbean
Anolis based upon albumin immuno-
logical data. This analysis did not deal
with the alpha-beta dichotomy, or the
carolinensis-punctatus dichotomy; these
investigators chose to restrict the analysis
exclusively to the punctatus subsection
of the alpha section as defined by Wil-
liams (1976). The analysis of immuno-
logical data led to a proposed classifica-
tion that differs somewhat from the one
proposed by Williams (1976). Specifi-
cally, the cybotoids were removed from
the cristatellus series and placed in their
own series. The cristatellus series was
then divided into two subseries, desig-
nated cristatellus and bimaculatus. The
latter is restricted to the Lesser Antilles;
the former now includes the Greater
Antillean (and St. Croix) forms that Wil-
liams placed in both the bimaculatus and
cristatellus series. Thus the “trunk-
crown’ anoles of Puerto Rico are placed

628

in the same series and subseries as the
“trunk-ground” and “grass-bush”’ radia-
tions. The revised classification is thus
completely in accord with Williams’
(1972) idea of a “radiation in place.”

The purpose of the present paper is to
utilize previously reported karyotypic
data combined with new data obtained
from multi-locus electrophoretic studies
to help us further analyze the relation-
ships of the Puerto Rican anoles. Our
primary concern is with lower-level
patterns of relationships, not with the
classification of the genus; however, we
have chosen certain “outside” forms for
comparison specifically to examine some
of the controversial questions in anole
systematics.

There are two independent allozyme
data sets utilized in this paper. Almost
ten years ago, three of us (Yang, Soulé,
Gorman) collaborated on a 19 locus elec-
trophoretic study of all species of Puerto
Rican Anolis, plus several selected out-
side representatives. The original study
failed to resolve clearly some of the pat-
terns of relationship among the eight
Puerto Rican species assigned by Gor-
man et al. (1980a) to the cristatellus sub-
series. Thus a second data set was ob-
tained, based on fresh material, for the
eight species. The number of loci ex-
amined was expanded to 35 and different
outside reference species were utilized
in the quest for clearer patterns to
emerge. Nei and Roychoudhury (1974)
pointed out that genetic distance esti-
mates are far more sensitive to the num-
ber of loci examined than to the number
of specimens examined. This has been
confirmed empirically for Anolis by
Gorman and Renzi (1979).

MATERIALS AND METHODS
ALLOZYME DATA SETS

The first electrophoretic study in-
volved 22 population samples represent-
ing 15 species of Anolis, including all ten

Advances in Herpetology and Evolutionary Biology

Puerto Rican species, which were ex-
amined for gene products of 19 loci. The
electrophoretic methods have been pre-
viously presented (Yang et al., 1974;
Gorman and Kim, 1976) as have been the
allelic data for the populations of A.
scriptus, A. cooki, A. monensis, and the
representative population of A. crista-
tellus (#8 of Gorman et al., 1980b), which
are also compared in the present study.
Sample size and locality data for all
populations in this first study are listed in
Appendix I.

The second electrophoretic study in-
volved 12 population samples repre-
senting 11 species of Anolis, including
eight species from Puerto Rico, which
were examined for gene products of 35
loci. The electrophoretic methods for the
second study are essentially those of
Buth et al. (1980) with the exceptions that
adenosine deaminase Ada-A, esterase
Est-2, peptidase Pept-3, superoxide dis-
mutase Sod-2, and xanthine dehydro-
genase Xdh-A of the latter study have
been replaced by creatine kinase (Ck-C;
EC 2.7.3.2), dihydrolipoamide reductase
(Dlr-1; EC 1.6.4.3), fumarate hydratase
(Fum-A; EC 4.2.1.2), L-iditol (sorbitol)
dehydrogenase (Iddh-A; EC _ 1.1.1.14),
and a second “malic enzyme” (Me-2; EC
1.1.1.40). Staining procedures for these
five systems were modified from Shaw
and Prasad (1970), Brewer (1970), Lin et
al. (1969), Kaplan and Beutler (1967), and
Ayala et al. (1972), respectively. Sample
size and locality data for all populations
in this second study are listed in Appen-
dix II.

TREATMENT OF ALLOZYME DATA

Phenetics. Differentiation among
populations was quantified for both of
the allozyme data bases by computing
Nei’s (1972) genetic distance (D) and
similarity (I) coefficients between each
pair of populations. The distance coeffi-
cients were then clustered using the
unweighted pair group method with

RELATIONSHIPS OF THE PUERTO RICAN ANOLIS - Gorman et al.

arithmetic averages (Sneath and Sokal,
1973).

Cladistics. Phylogenetic relationships
within the cristatellus subseries (Gor-
man et al., 1980) and the Puerto Rican
representatives of this subseries were
estimated using the Wagner procedure
(Farris, 1970) as applied to both of the
allozyme data bases. Allelic products
were encoded as “present” or “absent”
following Mickevich and Johnson (1976).
While the latter study employed a fre-
quency minimum of 0.05 for allelic
“presence, our treatment for the two
data sets employed a much higher fre-
quency requirement (0.25) to minimize
the resolution limitations set by the gross
sample size differences between the two
studies. Most of the uncommon alleles
omitted in this coding procedure were
population-specific and contributed little
to the estimation of relationships be-
tween taxa. Initially estimated unrooted
Wagner networks (Lundberg, 1972) were
rooted by including species presumed to
be external to the phyletic radiation in
question (see Farris, 1972). Thus, the first
electrophoretic study employed A. cu-
vieri, A. cybotes, A. occultus, and A. ocu-
latus for rooting purposes whereas the
second study employed A. gadovi, A.
grahami, and A. carolinensis for this pur-
pose. The first two members of the root-
ing array used in the second study are
members of the beta section; the third
species is a North American form of
Cuban affinity (Buth et al., 1980) that
represents the other (i.e., carolinensis)
subsection of alpha Anolis.

RESULTS

The allozyme data bases for the first
and second electrophoretic studies are
presented in Tables 1 and 2, respec-
tively.

Phenetic relationships derived from
both allozyme data sets are depicted in
Figures | and 2. In all cases where more
than a single population was examined

629

from a given species, the intraspecific
genetic distances are uniformly low and
the populations of a species cluster
together (e.g., A. acutus, A. cristatellus,
A. krugi, A. poncensis, A. pulchellus, A.
stratulus). Among the eight Puerto Rican
species examined in common between
the two electrophoretic studies, the
average genetic distance value (D) for
pairwise interspecific comparisons is
similar. The average absolute difference
between the studies is D=0.07. Despite
these small average differences, there are
certain trends in distances of the second
study that provide a different perspective
regarding interspecific similarities. The
most striking discrepancy is in the rela-
tionship of A. evermanni and A. stratu-
lus. In the first study, the mean genetic
distance between this pair was approxi-
mately D = 0.5, and the two species were
not closely associated (Fig. 1). In the
second study, D = 0.27 between these
forms and this pair clusters together (Fig.
),
The cladistic analyses were limited to
the members of the cristatellus subseries
of Gorman et al. (1980a). The cladistic
treatment of both allozyme data sets
yielded Wagner trees with levels of
homoplasy that might be considered
high. Consistancy indices (Farris, 1969;
Mickevich and Johnson, 1976) of c = 0.64
and c = 0.61 were computed for Wagner
trees based on the first and second elec-
trophoretic studies, respectively. Thus, a
large number of equally parsimonious
minimum-length Wagner trees may be
constructed. Evaluations of several such
trees revealed repeated clustering of cer-
tain groups of species in both cases.
These conservatively estimated phylo-
genetic relationships are depicted in
Figures 3 and 4.

The phylogenetic relationships of the
eight Puerto Rican species in the second
electrophoretic study were examined
with reference to a comparison of karyo-
typic evolution previously reported by
Gorman et al. (1968) and Gorman and
Atkins (1969). These karyotype descrip-

630 Advances in Herpetology and Evolutionary Biology

TABLE |. ALLELIC VARIABILITY WITHIN AND AMONG SAMPLES OF 22 POPULATIONS OF 15 SPECIES OF ANOLIS.
Allelic products are lettered corresponding to increasing anodal electrophoretic mobility and are mono-
morphic within samples unless otherwise noted. Allelic notation is equivalent to the data base of Gorman et
al. (1980b) which includes additional populations of A. cristatellus; however, allele letters used here do not
necessarily correspond to those in other tables in the present study. Populations are numbered in the order
they are listed in Appendix I. * = no expression of the locus (null allele?). ** = equivalent to creatine kinase

(Ck-A).
Locus 1 2 3 4 5 6 7
1. Adh-A f f f c(0.17) c(0.02) £f(0.92) (0.
f(0.83) (0.98) h(0.08) (0.
2. S-Aat-A d(0.97) d d d(0.94) b(0.94) ¢(0.13) d
g(0.03) (0.02) d(0.06) 4d(0.87)
f(0.04)
3. Est-1 c(0.07) b(0.04) j k (0.04) i e(0.07) c(0.
d(0.03) (0.96) m(0.52) £(0.04) e(0.
f (0.10) p(0.46) h(0.05) j(0.
h(0.73) j(0.84) 1(0.
j (0.07)
kh. Fum-A d d d(0.87) d a(0.02) d(0.75) d
£(0.13) d(0.98) e(0.25)
5. Gp-l d(0.07) e c(0.20) e £(0.02) d(0.98) d(0.
e(0.93) e(0.80) h(0.98) (0.02) e(0.
£(0
6. Gp-2 b b b b b b b
i-mGp-3 a a e ic a a a
8. Gp-4 b b b b b b b
9. Gpi-A F f a(0.07) c b(0.04) c¢(0.05) e
c(0.93) c(0.02) e(0.92)
£(0.94) £(0.03)
10. G-3-pdh-A e e b(0.02) e a(0.04) e(0.98) e
e(0.98) e(0.96) (0.02)
11. M-Icdh-A c(0.97) c c (0.73) c a(0.71) b(0.90) (0.
e(0.03) e(0.27) c(0.29) c(0.10) e(0.
12. S-Icdh-A c c a(0.15) € c Cc c
c(0.85)
13. Ldh-A c c c c c c (0.90) Cc
f (0.10)
14. Ldh-B c(0.67) c(0.98) f e(0.98) f a(0.57) a(0.
e(0.33) (0.02) f (0.02) £(0.43) (0.
15. M-Mdh-A d d d d a(0.12) d d
e(0.88)
16. S-Mdh-A d d d d d(0.94) d d(0.
e(0.06) e(0.
17. Pept-1 b(0.10) d(0.98) d d b(0.96) d d
d(0.90) (0.02) e(0.04)
18. Pam-A d(0.64) g d d d a(0.02) a(0
g(0.33) d(0.87) d(o
h(0.03) £(0.08) e(0
h(0.03) g(0
19. Pgdh-A a(0.07) d d a(0.02) d(0.02) d d(0
d(0.83) d(0.87) i(0.98) e(0
e(0.03) e(0.11)
f(0.07)

74)
08)

.18)

95)
05)

.10)
-81)
02)
-07)

-80)
.20)

8 ) 10 in
f(0.98) (0.83) c(0.05) f
h(0.02) h(0.17) (0.77)

h(0.18)
c(0.02) ¢(0.02) d d(0.96)
d(0.98) 4d(0.98) g(0.04)
e(0.02) b(0.02) a(0.12)
£(0.12) e(0.98) b(0.02)
j (0.72) e (0.86)
1(0.14)

d d d d
c(0.02) e (0.02) F
e(0.96) (0.98)

g(0.02)

b b b b

a a a a

b b b b
d(0.12) e c(0.08) c
e (0.88) e(0.90)

(0.02)
e(0.98) e e e
£(0.02)

b c c c

Cc Cc Cc (
a(0.02) Cc c Cc
c(0.98)

(0.62) a a(0.95) f
(0.38) £(0.05)

d d d d

d d(0.63) d(0.25) d
e(0.37) e(0.75)

d d d d
a(0.06) a(0.04) a(0.17) d(0.50)
d(0.86) (0.79) d(0.83) g(0.50)
g(0.08) (0.17)

a(0.02) c(0.20) c(0.17) d
b(0.02) d(0.04) e(0.83)
d(0.88) e(0.76)

e(0.02)

RELATIONSHIPS OF THE PUERTO RICAN ANOLIS - Gorman et al. 631

Locus

1. Adh-A

2. S-Aat-A

3a ESiEo1

4. Fum-A

6. Gp-2
7. Gp-3"
8. Gp-4
9. Gpi-A

10. G-3-pdh-A

11. M-Icdh-A

12. S-Icdh-A

13. Ldh-A

14. Ldh-B
15. M-Mdh-A
16. S-Mdh-A

17. Pept-1

18. Pgm-A

19. Pgdh-A

d(0.93)
g(0.05)

b
a
b

a(0.11)
c (0.89)

c(0.89)
e(0.11)

d(0.44)
g(0.56)

c(0.98)
e(0.02)

a(0.04)
d(0.94)
£(0.02)

b(0.03)
d(0.64)
(0.33)

e

d(0.92)
f (0.08)

c (0.06)
e(0.94)

TABLE 1. CONTINUED.

b(0.04)
(0.96)

d(0.98)
g(0.02)

16 17 18
f f c (0.06)
£(0.94)
b b c(0.96)
d(0.04)
k(0.11) p(0.95) (0.98)
p(0.89) q(0.05) q(0.02)
d d d
e(0.78) c(0.10) a(0.17)
(0.22) e(0.90) g(0.83)
b b b
a a a
b b b
c ¢(0.95) c(0.04)
£(0.05) (0.96)
b b c (0.04)
e(0.96)
d d c
c c b(0.23)
c(0.77)
c c a(0.01)
c (0.98)
d(0.01)
d d g
a(0.05) c c
c(0.95)
b b d
d d d
d d d
d(0.98) d a(0.04)
f (0.02) d(0.29)
e(0.67)

c(0.19)
£(0.81)
b
a
b

a(0.04)
c(0.96)

c(0.60)
e(0.40)

c(0.24)
d(0.76)

(

£(0.88)
h(0.12)

20

a
b
b

c (0.95)
£(0.05)

e

d

d(0.77)
g (0.23)

d(0.97)
e(0.03)

Cc

b(0.50)
d(0.47)
£ (0.03)

d

22

b

c(0.13)
e(0.84)
g (0.03)

tions are summarized in Table 3, and the
karyotype phylogeny is shown in Figure
5. Unfortunately, there have been no fur-
ther chromosome studies on these anoles

incorporating modern banding _tech-
niques. Thus, the karyotype phylogeny is
based upon several assumptions. The
presence of six pairs of metacentric

e e e q q e e e e e e e V-4PD “LI
(t1°0)@ (£0°0)® (Z1°0)® (€g°0)P (£€8°0)P
p p p p q e (gg-0)2_ (€6"0)2 (ggo)P © (LL0)e (LL ‘o)e v-!d9 “91
e q p S) eS) fo) ) ) c) fe) ) i) €-d9 “Sl
fs) q e e e e e e e e e e 7-4) “41
e e 2 e e e e e e e e e 1-d9 °€1
q e e e e e e e e e e e y-wn4y “ZL
e e e e e e e e e e e e 1-354 ‘LL
e e e e e e e e e e e e l-41d ‘OL
q e eS) q q q q q q q q q I-49 6
e e e e e e e e e gq e e W-19 °8
e q q e e e e e e e e e y-7e9 ‘2
e e e e e e e e e e e e 1-dq9 °9
(Z1°0)P
e 3 (88°0)> q q e e e ) q q q V-3eV-S °S
e eS) q q q q q q q q q q y-dy *4
q e e e e e e e e e e e V-4PY “€
(Z1°0)2 (€€°0)9
(8€°0)9 (05°0)49 (S2°0)9 (Z1°0)9 (ZE°0)9 (49°0)> (Z1°0)9
e P (05°0)4 (0S°0)e® (S2°0)4 q (88°0)4 q (€9°0)4 =) (€€°0)4 (0S°0)e Y-UOSY-S °C
(S0°0)
2 2 -) q q e q -) P (S6°0) 9 s) V-UOOY-W “1
Z1 Lt Ol 6 8 L 9 S 4 € Z l sno07

‘TJ xipuoddy ut pays] ore Aay} Japi0 ouj ul pozequinu are suonejndog ‘Apnys Juasaid ay} Ul Sa[qe} I9y{O UT asOYy 0} puodsa1109
A[Liessa0au JOU Op aay pasn sioye] VTaTTe “at ‘oUOTe aSeq eJeP SITY} OF DANLIAI SI UOHLIOU OI[I[[Y ‘pse}OU ssTAIOYy}ZO ssoyuNn
sojdures uryyim orydioulouour are pure AqWI{IGout OMe10Yydooa]e [epour SuTSKaIOUT 0} SUIPUOdSaLI0D potoyey] ore sjonpod ora] [V

Advances in Herpetology and Evolutionary Biology

‘SITONY AO SAIONdS Tl JO SNOLLV1NdOd ZI AO SATA WVS ONOWY GNV NIHLIM ALITIGVIUVA OITA TTY GA TAV LT

633

RELATIONSHIPS OF THE PUERTO RICAN ANOLIS - Gorman et al.

(ZE°0)4
(€9°0)4

(S2°0)2
(52°0)4

oO

P
(S9°0)4
(82°0)2
(Z0°0)P

4

rc)

(9€°0)e
(79° 0)?

q

(Z0°0)@
(€6°0)P

1-POS “SE

y-dug “7

v-4pbd “€€

y-wbg °7e

GonGeag 1G

l-ideq “0€

Y-!dw °6Z

c-OW BC

L-9W ~LZ

Y-4PW-S ~92

Y-U4PW-W SZ

SII “7

V-4P1 “EZ

Y-4PI1-S “cd

Y-Y4PII-W “172

WPI] “OG

V-4PPH “61

W-upd-€-9 81

634 Advances in Herpetology and Evolutionary Biology

1 A. cristatellus

ae 2 A. scriptus

{1
} A. poncensis
{2

3 A. cooki
“--4 4 monensis
6
1 >A pulchellus
8
I
as A. krugi
_Arugl
10
alo
BoosoSsose A. acutus
114

{8 A. evermanni
9 A. gundlachi
wo ---------- 19 A. ocu/atus
15
{7 ) A. stratulus
16
bess 5 S555 Se 5555555 555555=5- 22 A. cyboles
21 A. occultus

20 A. cuvieri

cae eee ee
1.2 1.0 0.8 0.6 0.4 0.2 0

GENETIC DISTANCE (J)

Figure 1. Phenogram of genetic relationships among populations of ten species of Puerto Rican Anolis (solid
lines) plus five non-Puerto Rican species (dashed lines). Relationships are derived from calculations of Nei’s
(1972) coefficient of genetic distance (D) clustered using the unweighted pair-group method with arithmetic
averages (UPGMA: Sneath and Sokal, 1973: 230). Calculations are based on the analysis of 121 allelic products
encoded by 19 presumptive loci. Populations are numbered as listed in Appendix |.

RELATIONSHIPS OF THE PUERTO RICAN ANOLIS - Gorman et al.

A. cooki
A. cristatellus #\
A. cristatellus #2
A. poncensis
A. krugi
= 4 See A. pulchellus
| A. gundlachi
! A. evermonni
A. stratulus
OO a a aoe aan A. grahami

Dims = oS SS STS SSS A. carolinensis

L n Ee gy
1.2 1.0 0.8 0.6 0.4 0.2 0)
GENETIC DISTANCE (0)

Figure 2. Phenogram of genetic relationships among
nine populations of Puerto Rican Anolis (solid lines)
plus A. carolinensis, A. gadovi, and A. grahami
(dashed lines). Relationships are derived from calcula-
tions of Nei’s (1972) coefficient of genetic distance (D)
clustered using the unweighted pair-group method with
arithmetic averages (UPGMA: Sneath and Sokal,
1973: 230). Calculations are based on the analysis of
154 allelic products encoded by 35 presumptive loci.

Figure 3. Cladogram of the species of the cristatellus
subseries based on the allozyme data of the first elec-
trophoretic study (see Table 1). A number of minimum-
length Wagner trees were evaluated to conservatively
estimate the relationships depicted here although
many relationships remain unresolved.

macrochromosomes, 24 microchromo-
somes and the absence of sexual hetero-
morphism is considered primitive
(Gorman, 1973; Paull et al., 1976). Re-
duction in diploid number and the evolu-
tion of X,X,Y heteromorphism are de-
rived conditions. Major macrochromo-

635

Figure 4. Cladogram of eight Puerto Rican species of
Anolis based on the allozyme data of the second elec-
trophoretic study (see Table 2). A number of minimum-
length Wagner trees were evaluted to conservatively
estimate the relationships depicted here although
many relationships remain unresolved.

somal rearrangement affects only the
Lesser Antillean bimaculatus subseries
(group III). Anolis distichus has the
highest diploid number in the cristatel-
lus series but also is the only member of
the series with possible loss of macro-
chromosomes; thus its position (IIG) in
Figure 5 is open to other interpretations.
Anolis evermanni (IIF in Figure 5) has
the lowest diploid number in the crista-
tellus series (and in the genus) and also
has XY rather than X,X,Y heteromor-
phism. Except for the sex chromosomes
and the reduction of a pair of micro-
chromosomes, its karyotype is identical
to A. stratulus (IIB); it is thus considered
derived from the IIB state even though
its sex chromosomes appear simpler.
States IIB, C, and D all have one fewer
pair of microchromosomes than IIA.
They differ in relative size of smaller
chromosomes. It is not certain that they
represent three independent lineages,
but it is impossible to suggest a single
sequence of derivation based simply on

636 Advances in Herpetology and Evolutionary Biology

TABLE 3. DESCRIPTIVE SUMARY OF KARYOTYPES OF PUERTO RICAN AND OTHER EASTERN CARIBBEAN ANOLIS -

Group I.

Primitive karyotype, 2n=36; 6 pairs of metacentric macrochromosomes; 12 pairs of distinctly smaller
microchromosomes. No obvious sex chromosome heteromorphism.

Included species:

A. occultus Puerto Rico
A. cuvieri Puerto Rico
A. cybotes Hispaniola

Group II.

Derived karyotype, usually involving only the microchromosomes which are reduced in number. Sex
chromosome heteromorphism is obvious.

A. 2n=31, 2n=32
Autosome pairs 1-6 metacentric, large
Autosome pair 7 metacentric, distinctly intermediate in size
Seven pairs of microchromosomes
X, X,Y male heteromorphism. Y chromosome is metacentric, approximately same size as autosome 7.

Included species:
A. acutus St. Croix

mo

2n=29, 2n=30
Same as II-A except six pairs of microchromosomes

Included species:
A. stratulus Puerto Rico

Cc: 2n=29, 2n=30
Same as II-B except that there are two pairs of intermediate sized metacentric autosomes (pairs 7 and
8), and thus only five pairs of microchromosomes

Included species:

A. gundlachi Puerto Rico

A. krugi Puerto Rico

A. pulchellus Puerto Rico

A. poncensis Puerto Rico
ID}. 2n=29, 2n=30

Same as II-C except no distinct size break between intermediate autosome pair 8, and largest of
microchromosomes, pair 9

Included species:

A. cooki Puerto Rico
A. monensis Mona
E. 2n=27, 2n=28

Same as II-D except one fewer pair of microchromosomes

Included species:

A. cristatellus Puerto Rico
A. scriptus Southern Bahamas
F. 2n=26, 2n=26

Autosome pairs 1-7 as in II-A and II-B

Five pairs of microchromosomes

X-Y male heteromorphism (the only species in the group that is not X, X,Y).
The Y chromosome is acrocentric, slightly smaller than autosome 7.

Included species:
A. evermanm Puerto Rico

RELATIONSHIPS OF THE PUERTO RICAN ANOLIS - Gorman et al.

637

TABLE 3. CONTINUED.

G. 2n=33, 2n=34

Macrochromosomes differ from all of the above in that there are essentially 7 pairs, with a sharp
break in size between pairs 5 and 6 (alternatively we might say there are 5 pairs of metacentric
macrochromosomes, and 2 intermediate sized pairs of metacentric autosomes).

Eight pairs of autosomal microchromosomes

X, X,Y sex chromosome heteromorphism. Y chromosome is metacentric, just smaller than autosome

Included species:
A. distichus

Group III.

Hispaniola

Derived karyotype. Microchromosomes and sex chromosomes (X,X,Y) as in group II, but macrochromo-

somes show major rearrangement.

Included species:
A. oculatus Dominica
All northem Lesser Antillean
anoles from the Anguilla Bank
to Dominica

*For figures and detailed descriptions of karyotypes, see Gorman et al. (1968) and Gorman and Atkins (1969).

karyotype morphology. State IIE appears
to be derived directly from IID by a loss
of a pair of microchromosomes.

DISCUSSION

A phenetic treatment of the first allo-
zyme data set (including all ten Puerto
Rican species, plus five additional spe-
cies of Anolis) reveals that two of the
extralimital species, A. scriptus and A.
monensis, are very similar to A. crist-
atellus and A. cooki, respectively (Fig. 1).
These relationships have been previ-
ously discussed (Gorman et al., 1980b);
they will receive no further comment.
Two Puerto Rican species, A. occultus
and A. cuvieri, and the Hispaniolan A.
cybotes are not similar to one another or
to any of the remaining species that were
considered. This lack of similarity is fur-
ther support for our decision (Gorman et
al., 1980a) to remove the Hispaniolan
cybotoids from the cristatellus series as
defined by Williams (1976), and also sup-
ports the idea that neither A. cuvieri nor
A. occultus is intimately related to the
remainder of the Puerto Rican anoline

radiation. Further discussion of the
Puerto Rican anoline radiation will ex-
clude consideration of A. occultus and A.
cuvieri.

Five species from Puerto Rico, A.
cristatellus, A. cooki, A. pulchellus, A.
krugi, and A. poncensis, are very similar
and all phenetically cluster at about D =
0.2. The three remaining species, A.
gundlachi, A. evermanni, and A. stratu-
lus, are phenetically distinctive, cluster-
ing with the previously mentioned spe-
cies at about D = 0.5. The electrophoretic
distinction of A. gundlachi is unexpected
given that its karyotype is indistinguish-
able from that of the “grass-bush” species
(Gorman et al., 1968), and that its al-
bumin immunological distances to A.
cristatellus and the “grass-bush” species
are low and indistinguishable, respec-
tively (Shochat, 1976). The apparent lack
of congruence among these phenetically-
treated data sets may be the result of
independent rates of relative evolution-
ary change; allozymes may be evolving
faster in A. gundlachi than albumin or
chromosome structure. Anolis gundlachi
phenetically clusters with A. oculatus of
Dominica, Lesser Antilles (Fig. 1), but

Advances in Herpetology and Evolutionary Biology

State Il F
A. evermanni
2n= 26

State IT B
A. stratulus
2n=29

State IL C

A. gundlachi
A. poncensis

State LE

A. cristatel/us
A. scriptus
on=27

State II D
A. cook/

A. monens/s
2n=29

State II G
A. arstichus
en=33

A. pulchellus
A. krug/
2n=29

State I A
A. acutus

en= 31

cristatellus subseries

bimacu/atus subseries

cristatellus series

State I
2n= 36

Figure 5. Phyletic relationships of karyotypic character states described in Table 3. Species of the cristatellus

series expressing each state are indicated.

this similarity is not necessarily indica-
tive of an evolutionary relationship
between these two taxa. Further studies
may elucidate the relationship of A. ocu-
latus to the Puerto Rican assemblage.
Anolis acutus of St. Croix exhibits
greater similarity to the main body of the
Puerto Rican radiation than do the elec-
trophoretically distinctive Puerto Rican
forms—A. gundlachi, A. evermanni, and
A. stratulus. The phenetic relationship of
A. acutus to these forms differs from the
cladistic relationship discussed below.
Anolis grahami was chosen for com-
parison in the second electrophoretic
study specifically because Shochat (1976)
suggested that the beta anoles of the
Jamaican grahami series were intimately

associated with the eastern Caribbean
cristatellus-bimaculatus radiation. The
genetic distances between A. grahami
and the species in the cristatellus sub-
series prove to be quite large, but A.
grahami does exhibit a slightly greater
similarity to these species than do A.
gadovi or A. carolinensis (Fig. 2).

By combining the implications of both
Wagner trees (Figs. 3, 4), we can attempt
to assess some of the patterns of evolu-
tionary relationships. In Figure 3, the
species A. krugi and A. pulchellus form a
sister group, the third “grass-bush” spe-
cies, A. poncensis, arises from a common
node with the previous pair plus A. crist-
atellus and A. cooki. However, in Figure
4 the three “grass-bush” anoles are a unit

RELATIONSHIPS OF THE PUERTO RICAN ANOLIS : Gorman et al.

forming an unresolved trichotomy. Thus
we believe A. krugi and A. pulchellus to
be a sister group, and A. poncensis as the
sister group of this pair.

Anolis gundlachi is one of the most
problematic species in the Puerto Rican
radiation. One reason is that it has many
autapomorphic alleles. Although Figures
3 and 4 show unresolved polychotomies,
A. gundlachi could be the sister group of
the “trunk-ground” and “grass-bush”
ecomorphs. This is concordant with the
thoughts of Etheridge (1960), Williams
(1972) and Gorman et al. (1968) that A.
gundlachi represents the descendent of
the lineage that later radiated into the
other “trunk-ground” and “grass-bush”
species.

In both Figures 3 and 4, A. evermanni
and A. stratulus form a pair, and their
lineage is not cladistically associated
with any of the remaining Puerto Rican
species. Some of our equally parsimo-
nous Wagner trees associated A. acutus
with A. evermanni and A. stratulus while
other trees show no relationship of A.
acutus to the other Puerto Rican anoles.
We believe there is some relationship
among these three species but the rela-
tionship is not clear.

SUMMARY CLADOGRAM AND
CLASSIFICATION

In Figure 6 we present our best esti-
mate of a summary cladogram. Following
the proposed terminology of Williams
(1976), and our recent proposals (Gorman
et al., 1980a), we are discussing the crist-
atellus subseries of the cristatellus
series. Within the subseries, we recog-
nize two species groups, the acutus
species group and the cristatellus spe-
cies group. The groups may be distin-
guished by an osteological character;
there are 23 presacral vertebrae in the
cristatellus species group and 24 in the
acutus species group. The acutus species
group is not further divided into sub-
groups because, despite the fact that A.

639

ANCESTOR

Figure 6. Summary cladogram of relationships in the
cristatellus subseries. Allelic synapomorphies are indi-
cated by number: (1) Pept-2 allele c, (2) Me-1 allele e,
(3) Gdh-A allele b, (4) Pnp-A allele b, (5) S-Aat-A allele
a, (6) Me-2 allele e, (7) Ldh-B allele c. Allele letters are
those used in Table 2. One karyotypic synapomorphy
is indicated as IIC, corresponding to the state designa-
tions of Table 3 and Figure 5. The six species of the
cristatellus species group share an_ osteological
synapomorphy (23 presacral vertebrae).

evermanni and A. stratulus appear to be
a natural subgroup, we cannot presently
place A. acutus and A. distichus in this
radiation.

The cristatellus species group is di-
vided into two subgroups. The crist-
atellus subgroup contains the cristatellus
superspecies (A. cristatellus and A.
scriptus) and the monensis superspecies
(A. cooki and A. monensis). Thus what
Williams (1976) called the cristatellus
species group is now reduced one rank to
a subgroup. The second unit is the
gundlachi subgroup. It contains A.
gundlachi and the “grass-bush”’ anoles.
Since the Williams classfication (1976)
includes no divisions between subgroup
and superspecies, and since the “grass-
bush” anoles are not a superspecies by
conventional definition, there are no
further available categories to separate A.
gundlachi from the “grass-bush” anoles.
Although it is hard to imagine that any-
one will accept yet another division in
this hierarchical classification, we could
define a gundlachi infragroup and a
pulchellus infragroup. This would unite
the “grass-bush” anoles and distinguish

640

TABLE 4. PROPOSED CLASSIFICATION OF THE CRIST-
ATELLUS SUBSERIES.

cristatellus subseries
acutus species group
A. evermanni
A. stratulus
A. distichus superspecies
A. acutus

cristatellus species group
cristatellus subgroup
A. cristatellus superspecies
A. monensis superspecies

gundlachi subgroup
A. gundlachi
A. poncensis
A. pulchellus
A. krugi

them as a lineage separate from A.
gundlachi. Table 4 summarizes the pro-
posed classification of the cristatellus
subseries.

ACKNOWLEDGMENTS

This study was supported by National
Science Foundation Grants GB-27358
(M. Soulé), B-019801 (E. E. Williams),
and DEB 77-03259 (G. C. Gorman).
Field and logistic assistance of W. P.
Maclean, III of the College of the Virgin
Islands, and Jose A. Colon and the sup-
porting staff of the Puerto Rican Nuclear
Center greatly facilitated our work. C. S.
Lieb, R. D. Orton, and T. L. Vance as-
sisted in the laboratory. Genetic dis-
tances and similarities were calculated
using a program supplied by E. G.
Zimmerman. Wagner trees were esti-
mated with the HENNIG MEMORIAL
PROGRAM written by J. S. Farris. We
also thank P. Fuerst, M. Nei, and W. J.
Rainboth, Jr. for additional data analysis
and discussion. Finally, we are indebted
to E. E. Williams and Susan Case for
thoughtful comments on an earlier draft
of this paper.

Advances in Herpetology and Evolutionary Biology

APPENDIX |

Specimens and localities for the first
electrophoretic study. No voucher
specimens were deposited. Sample sizes
are indicated in parentheses.

Population 1. Anolis cristatellus PUERTO RICO:
Rio Piedras (15). 2. A. scriptus TURKS AND
CAICOS ISLANDS: Grand Turk (25). 3. A. cooki
PUERTO RICO: Hwy. 333 between km posts 2.7
and 3.1 near Guanica (20). 4. A. monensis ISLA
MONA (23). 5. A. gundlachi PUERTO RICO:
Luquillo National Forest, near La Mina recreation
area (30). 6. A. pulchellus PUERTO RICO: Hwy.
186, km 22 near El Verde (30). 7. A. pulchellus
PUERTO RICO: Hacienda Roses near Utuado (30).
8. A. pulchellus U.S.V.1.: St. Thomas: Estate Tutu
(24). 9. A. krugi PUERTO RICO: Hacienda Roses
near Utuado (30). 11. A. poncensis PUERTO RICO:
2.3 miles W of La Parguera (9). 12. A. poncensis
PUERTO RICO: La Parguera (27). 13. A. acutus
U.S.V.L.: St. Croix (23). 14. A. acutus U.S.V.L:
Green Cay near St. Croix (18). 15. A. stratulus
B.V.I.: Virgin Gorda (26). 16. A. stratulus PUERTO
RICO: Rio Piedras (27). 17. A. stratulus PUERTO
RICO: Hacienda Roses near Utuado (10). 18. A.
evermanni PUERTO RICO: Luquillo National
Forest, near La Mina recreation area (30). 19. A.
oculatus DOMINICA: Roseau (26). 20. A. cuvieri
PUERTO RICO: Hwy. 186, near km 19 [El Verde
Field Station] (11). 21. A. occultus PUERTO RICO:
Luquillo National Forest, base of El Toro Trail (16).
22. A. cybotes HAITI: 9 km W of Marmelade (15).

APPENDIX Il

Specimens and _ localities for the
second electrophoretic study. Vouchers
are deposited with the Natural History
Museum of Los Angeles County.

Population 1. Anolis cristatellus PUERTO RICO:
Hwy. 186, km 22 (3, LACM 130688-90). 2. A. crist-
atellus PUERTO RICO: Jct. Hwys. 191 and 988 (3,
LACM 130691-93). 3. A. cooki PUERTO RICO:
Salt Plant near Pta. Jaguey (5, LACM 130683-87). 4.
A. gundlachi PUERTO RICO: Jct. Hwys, 191 and
988 (4, LACM 130694-97). 5. A. pulchellus
PUERTO RICO: Hwy. 3, NE edge of Luquillo (7,
LACM 130676-82). 6. A. krugi PUERTO RICO: Jet.
Hwys. 191 and 988 (4, LACM 130698-701). 7. A.
poncensis PUERTO RICO: Lighthouse, Pta. Jaguey
(2, LACM 130667-68). 8. A. stratulus PUERTO
RICO: Hwy. 186, km 22 (4, LACM 130669-72). 9. A.
evermanni PUERTO RICO: Jct. Hwys. 191 and 988
(3, LACM 130673-75). 10. A. grahami JAMAICA:
Discovery Bay (6, LACM 13070409). 11. A. gadovi
MEXICO: Guerrero: Tierra Colorado (1, LACM

RELATIONSHIPS OF THE PUERTO RICAN ANOLIS - Gorman et al.

129728). 12. A. carolinensis [purchased, no locality
data] (2, no vouchers).

NOTE ADDED IN PROOF

Mickevich and Mitter (1981) have
noted that the “independent alleles
model” for encoding allozyme data, as
used in this study, usually yields lower
consistancy indices compared to other
methods such as Transition Series Analy-
sis. The independent alleles model may
also be biologically unrealistic in some
situations as it does not prohibit the for-
mation of a hypothetical intermediate
lacking alleles at a locus. Readers should
be aware of the potential limitations of
our allozyme analyses.

LITERATURE CITED

AYALA, F. J., J. R. POWELL, M. L. Tracey, C. A.
MOuRAO, AND S. PEREZ-SALAS. 1972. Enzyme
variability in the Drosophila willistoni group.
IV. Genic variation in natural populations of
Drosophila willistoni. Genetics, 70: 113-139.

BREWER, G. J. 1970. An Introduction to Isozyme
Techniques. New York, Academic Press.

BuTH, D. G., G. C. GORMAN, AND C. S. Lizs. 1980.
Genetic divergence between Anolis
carolinensis and its Cuban progenitor, Anolis
porcatus. J. Herpetol., 14: 279-284.

ETHERIDGE, R. E. 1960. The relationships of the
anoles (Reptilia: Sauria: Iguanidae): an inter-
pretation based on skeletal morphology. Ph.D.
thesis, University of Michigan. University
Microfilms, Inc., Ann Arbor, Michigan, xiv +
236 pp.

Farris, J. S. 1969. A successive approximations
approach to character weighting. Syst. Zool.,
18: 374-385.

—__. 1970. Methods for computing Wagner trees.
Syst. Zool., 19: 83-92.

—___.. 1972. Estimating phylogenetic trees from dis-
tance matrices. Amer. Natur., 106: 645-668.

FILDES, R. A., AND H. Harris. 1966. Genetically
determined variation of adenylate kinase in
man. Nature, 209: 261-263.

GoRMAN, G. C. 1973. The chromosomes of the Rep-
tilia, a cytotaxonomic interpretation, pp.
349-424. In A. B. Chiarelli and E. Capenna
(eds.), Cytotaxonomy and Vertebrate Evolu-
tion. London and New York, Academic Press.

GorRMAN, G. C., AND L. ATKINS. 1969. The zooge-
ography of Lesser Antillean Anolis lizards—an
analysis based upon chromosomes and lactic

641

dehydrogenases. Bull. Mus. Comp. Zool., 138:
53-80.

GORMAN, G. C., AND Y. S. Kim. 1976. Anolis lizards
of the eastern Caribbean: A case study in evo-
lution. II. Genetic relationships and genetic
variation of the bimaculatus group. Syst. Zool.,
20: 167-185.

GorMaAN, G. C., AND J. R. RENZI, JR. 1979. Genetic
distance and heterozygosity estimates in elec-
trophoretic studies: Effects of sample size.
Copeia, 1979: 242-249.

GORMAN, G. C., R. THOMAS, AND L. ATKINS. 1968.
Intra- and interspecific chromosome variation
in Anolis cristatellus and its closest relatives.
Breviora Mus. Comp. Zool. No. 293, pp. 1-12.

GorMaN, G. C., D. G. BUTH, AND J. S. WYLES. 1980a.
Anolis lizards of the eastern Caribbean: A case
study in evolution. III. A cladistic analysis of
albumin immunological data, and the defini-
tion of species groups. Syst. Zool., 29: 143-158.

GORMAN, G. C., D. G. BuTH, M. SOULE, AND S. Y.
YANG. 1980b. The relationship of the Anolis
cristatellus species group: Electrophoretic
analysis. J. Herpetol., 14: 269-278.

KAPLAN, J. C., AND E. BEUTLER. 1967. Electro-
phoresis of red cell NADH- and NADPH-
diaphorases in normal subjects and patients
with congenital methemoglobinemia. Bio-
chem. Biophys. Res. Commun., 29: 605-610.

Lin, C. C., G. SCHIPMANN, W. A. KITTRELL, AND S.
OHNO. 1969. The predominance of heterozy-
gotes found in wild goldfish of Lake Erie at the
gene locus for sorbitol dehydrogenase.
Biochem. Genet., 3: 603-607.

LUNDBERG, J. G. 1972. Wagner networks and ances-
tors. Syst. Zool., 21: 398-413.

MICKEVICH, M. F., AND M. S. JOHNSON. 1976.
Congruence between morphological and al-
lozyme data in evolutionary inference and
character evolution. Syst. Zool., 25: 260-270.

MICKEVICH, M. F., AND C. MITTER. 1981. Treating
polymorphic characters in systematics: A
phylogenetic treatment of electrophoretic data,
pp. 45-58. In V. A. Funk and D. R. Brooks
(eds.), Advances in Cladistics. Bronx, N.Y.,
The New York Botanical Garden.

NEI, M. 1972. Genetic distance between popula-
tions. Amer. Natur., 106: 283-292.

NEI, M., AND A. K. RoyCHOUDHURY. 1974. Sampling
variances of heterozygosity and genetic dis-
tance. Genetics, 76: 379-390.

PAULL, D., E. E. WILLIAMS, AND W. P. HALL. 1976.
Lizard karyotypes from the Galapagos Islands:
chromosomes in phylogeny and evolution.
Breviora Mus. Comp. Zool. No. 441, pp. 1-31.

RAND, A. S. 1964. Ecological distribution in anoline
lizards of Puerto Rico. Ecology, 45: 745-752.

SCHOENER, T., AND A. SCHOENER. 1971. Structural
habitats of West Indian Anolis lizards. II.
Puerto Rican uplands. Breviora Mus. Comp.
Zool. No. 375, pp. 1-39.

SHAw, C. R., AND R. PRASAD. 1970. Starch gel elec-

642

trophoresis of enzymes—A compilation of
recipes. Biochem. Genet., 4: 297-320.

SHOCHAT, D. 1976. Comparative immunological
study of albumins of Anolis lizards of Carib-
bean Islands. Ph.D. Thesis, Louisiana State
Univ., School of Medicine, New Orleans, x +
129 pp.

SHOCHAT, D., AND H. C. DESSAUER. 1981. Compara-
tive immunological study of albumins of
Anolis lizards of the Caribbean islands. Comp.
Biochem. Physiol., 68A: 67-73.

SNEATH, P. H. A., AND R. R. SOKAL. 1973. Numerical
Taxonomy. San Francisco, W. H. Freeman and
Co.

WILLIAMS, E. E. 1972. The origin of faunas: Evolu-

Advances in Herpetology and Evolutionary Biology

tion of lizard congeners in a complex island
fauna: A trial analysis. Evol. Biol., 6: 47-89.

___.. 1976. West Indian anoles: A taxonomic and
evolutionary summary. I. Introduction and a
species list. Breviora Mus. Comp. Zool. No.
440, pp. 1-21.

WYLES, J. S., AND G. C. GORMAN. 1980. The classifi-
cation of Anolis: Conflict between genetic and
osteological interpretation as examplified by
Anolis cybotes. J. Herpetol., 14: 149-53.

YANG, S. Y., M. SOULE, AND G. C. GORMAN. 1974.
Anolis lizards of the eastern Caribbean: A case
study in evolution. I. Genetic relationships,
phylogeny, and colonization sequence of the
roquet group. Syst. Zool., 23: 387-399.

Modes of Speciation and Evolution in the
Sceloporine Iguanid Lizards.

I. Epistemology of the Comparative Approach
and Introduction to the Problem

WILLIAM P. HALL!

ABSTRACT. Comparative studies of speciation
mechanisms and their evolutionary consequences
in the radiation of +120 sceloporine lizard species
are introduced. The epistemological foundations of
the comparative approach are explained. II]lustra-
tions from the sceloporine study show how the ap-
proach guided the work and aided interpretation. A
foundation problem in evolutionary biology over
methods for generating and testing explanatory
models traces from one in the philosophy of science
over the values of different “logics” of discovery
and heuristics for “verification.” Popper’s model of
scientific discovery and deductive falsification is
inappropriate for processes like speciation—which
are not universal, and where it cannot be assumed
either that one set of causes is responsible or that
they are deterministic. The worth of the compara-
tive approach is evaluated and its heuristic struc-
ture is outlined. In a collection of case histories of a
phenomenon, the methodology finds modes of cor-
relation between observations on the cases and
markers identifying the phenomenon which may be
obscured by randomly unrelated features. If the
number of different mechanisms causing the
phenomenon is small relative to the case histories,
each mode should result from one causal mecha-
nism. Models explaining the observed correlations
are realistic to the degree that they logically explain
the observed correlations. Their realism may be
increased or diminished by further testing, but their
“truth” or “falsity” can never be proved.

1Queen’s College and Department of Genetics,
University of Melbourne, Parkville, Vic. 3052, Aus-
tralia.

INTRODUCTION

Mayr (1942, 1947, 1954, 1963, 1970)
has argued that essentially all animal
speciation results from genetic changes
which can evolve between populations of
a species only when these populations
become allopatrically isolated from one
another by geographic barriers to gene
flow. Bush (1975b), Bush et al. (1977),
Wilson et al. (1975), and White (1978),
among others, using modes of argument
which differ from Mayr’s, concluded that
the majority of animal speciation does not
require geographic isolation. They also
suggested that much of this non-
allopatric speciation involves fixation of
chromosomal modifications in initially
small, inbred, but not geographically
isolated local populations. Bush and
Wilson, particularly, also observed that
rapid morphological (but not biochemi-
cal) evolution is associated with rapid
chromosomal evolution, and suggested
Goldschmidt’s (1955) idea of macromu-
tation as a possible explanation for the
putative relationship, resurrecting a con-
troversy laid to rest 20 years previously.

Aside from proposing obvious and non-
controversial models for speciation by
parthenogenesis and polyploidy, other
workers, many following still other and
less readily definable modes of argu-
ment, have proposed a multitude of
models for non-allopatric speciation (e.g.,

644

Bush, 1969, 1974, 1975; Ehrlich and
Raven, 1969; Endler, 1973, 1977; Gold-
schmidt, 1955; John and Lewis, 1965,
1966; Lewis, 1966, 1973; Lewis and
Raven, 1958; Matthey, 1964, 1966; Mur-
ray, 1972; Spurway, 1953; Spurway and
Callan, 1960; Todd, 1970; Wallace, 1959,
etc.). Thus, evolutionists face a Babel of
different and frequently incommensur-
able attempts to explain speciation and to
account for its significance in evolution.
This situation resembles the crisis which
Kuhn (1962) believes precedes a revolu-
tionary change in the paradigm of a scien-
tific community.

The present work will approach the
problem of speciation in still another,
and clearer way—one which should help
reduce the confusing Babel. This series
of papers will report systematically con-
trolled comparative studies of mecha-
nisms of speciation and evolution in the
sceloporine branch of the Iguanidae
(Savage, 1958; Etheridge, 1964; Presch,
1969). The sceloporines include some
120 species (Hall, part II), and enough
evidence now exists to allow most specia-
tion events in their history to be recon-
structed with some degree of plausibility.
From its inception, this research program
has had three major goals: 1) to deter-
mine how many qualitatively different
modes of speciation have occurred and
how frequent each was in the prolifera-
tion of the sceloporines; 2) to develop
and critically test models for the mecha-
nism of each mode, where this has not
already been done; and 3) to develop and
critically test models to predict the evolu-
tionary potentials for each of the modes.
The results then provide a solid founda-
tion from which to extrapolate to explain
speciation patterns in more distantly re-
lated organisms.

In this, the introductory paper to the
series, the characteristics of the scelo-
porines which led them to be studied are
described, and I explain my plan of attack
for the study. The organization of this
plan is relatively novel and uses unusual
but powerful forms of logical argument.

Advances in Herpetology and Evolutionary Biology

These must be explained and justified
scientifically and epistemologically. To
do so involves confronting major para-
digmatic crises both in evolutionary
biology and in the philosophy of science.
The second paper (Hall, part II) will
present an overview of sceloporine evo-
lution and cytosystematics which is
needed to orient the more detailed re-
ports to follow.

THE SPECIES PROBLEM: A CRISIS IN
EVOLUTIONARY BIOLOGY

Ghiselin (1974) and others have ob-
served that evolutionary biology now ~
faces a crisis (Kuhn, 1962) in its historical
development similar to that before Dar-
win published his “On the Origin of
Species’ in 1859, or before Dobzhansky
(1937), Huxley (1940, 1942), Mayr (1942),
and Simpson (1944) achieved the con-
sensus of ideas known as the “synthetic
theory of evolution”. These past crises
were both concemed primarily (but not
exclusively) with major revisions in
understandings of the mechanisms by
which species evolved through time
(e.g., anagenesis; White, 1978). The
present crisis, as indicated by the confu-
sion of papers cited in the introduction,
relates primarily to the species problem
(e.g., cladogenesis—White, 1978). To
what does the species concept actually
refer (e.g., Hull, 1976)? How are these
species formed? What effects, if any, does
the mode of species formation have on
the evolution of the speciating lineage?
The philosophical problems which
underlie these questions and make them
so difficult to answer also involve the
whole of evolutionary biology (e.g.,
Lévetrup, 1975; Peters, 1976; Platnick
and Gaffney, 1978). The array of ideas on
species and speciation already proposed,
and the obvious problems in answering
questions they raise, suggest that some-
thing more than a new paradigm in
Kuhn’s (1962) rather loose sense is
needed. Rather, more efficient heuristic

USE OF THE COMPARATIVE APPROACH - Hall

schemata are needed for rejecting un-
realistic or patently unscientific pro-
posals.

Kuhn (1962, 1970a) publicized the im-
portance of “paradigms” in scientific
research and communication. At the
same time, he admits to having hope-
lessly confused usage of the word para-
digm (Kuhn, 1969, 1970b; Masterman,
1970), a confusion not entirely resolved
by his introduction of the substitute
terms “disciplinary matrix” and “exem-
plar’ (Kuhn, 1969). Kuhn’s ideas most
relevant here concern scientific revolu-
tions, incommensurability, and com-
munication within and between scienti-
fic communities following different
paradigms. These ideas also seem to
apply to a more precisely defined con-
cept, the “heuristic schemata” of a re-
search program. Based on usage of the
two words separately in Suppe [ed.]
(1977), the term “heuristic schemata”
will be used here to refer to a particular
framework of initial metaphysical as-
sumptions, and the pattern of logical
argument which is followed within those
assumptions, either to test or to discover
a theory or putative knowledge. This is
not to be confused with specific observa-
tions of nature which form the basis of
the knowledge, the mechanical apparatus
used to collect the observations, or with
the theory or knowledge which is dis-
covered or tested by following the
heuristic schemata.

Reviewers of my earlier attempts to
describe the comparative studies of
sceloporine speciation and _ evolution
(e.g., Hall, 1977)! have had conspicuous

l1Reviewers were uniformly impressed by the
models proposed in Hall (1977) but seemed unable
to follow my arguments. Only when one well-
intentioned reviewer took the extraordinary effort
to completely rewrite my awkward prose did it
become unavoidably apparent that he and I were
seeing the same data from incommensurable van-
tages. At this point I withdrew the manuscript and
attempted to understand the problem. The present
series of papers is intended to cover the same
ground, but with this understanding in hand. Un-

645

difficulties in following what I thought
were clearly organized arguments used
to discover, develop, and test hypothe-
ses. Conversely, I have also found it ex-
ceptionally difficult to either review
other work intelligibly or to follow it as a
model for writing these arguments. From
trying to find the source of these long
standing communication difficulties, I
have concluded that the heuristic
schemata used in the sceloporine re-
search, which I collectively referred to as
the comparative approach (Hall, 1977),
are unusual and poorly understood by
many evolutionists. Once I extracted
them from my own work I have been
unable to find sources where these
schemata either are described or are laid
out clearly in forms which could usefully
serve as exemplars for the approach I
have followed. Most importantly, more
detailed, but still incomplete studies to
understand the problems raised by this
apparent incommensurability (Kuhn,
1962) of heuristic schemata suggest that
the approach followed here should be
more efficacious than others applied to
the species problem.

However, although I have modified
and/or used some in new contexts, the
schemata used in the sceloporine re-
search are not entirely original with me.
For instance, as his logical methodology
is described by Ghiselin (1969) and Hull
(1973), Darwin probably used similar
approaches to guide the development
and testing of his many revolutionary

fortunately, because of the wide interest in its con-
clusions, the 1977 manuscript has already been
effectively published by unauthorized but wide-
spread photocopying from copies I circulated for
informal reviews. Given the situation, I have no
choice but to declare this manuscript to be in the
public domain. I hereby grant permission to anyone
concerned to either make further copies or to cite
the paper. I request that the following citation style
be used: “Hall, W. P. 1977. Cascading chromosomal
speciation and the paradoxical role of contact hy-
bridization as a barrier to gene flow. Informally
published, 91 pp. [cited by author’s permission, see
Hall, (= full citation to the present paper)].”

646

theories. Unfortunately, Darwin did not
explain or justify his schemata clearly
enough for his critics to understand or
accept his arguments (Ghiselin, 1969;
Hull, 1973) although many of Darwin’s
most revolutionary theories have been
fully substantiated with time. Ghiselin
also asserts that many present workers
still do not understand Darwin’s
heuristic schemata. I would add here, to
emphasize the problems involved in
extracting such schemata from the inter-
stices of the work which they organize,
that if I have used a Darwinian approach
in the sceloporine studies, and if I under-
stand my own approach, then even
Ghiselin (1969, 1974) fails to recognize
all of the roles or significances of all of
Darwin’s schemata. This is especially
true for Ghiselin’s understanding of the
approach to originating hypotheses, as
compared to practices for testing them,
which are described in more detail. Hull
(1973) makes the distinction between the
two functions of Darwin’s research pro-
gram more clearly than Ghiselin does,
but Hull describes neither set of
schemata in useful detail.

More recently, Bush (1975); Bush, et
al. (1977); Wilson et al. (1975), and per-
haps White (1978) have used comparative
approaches which have some similarities
with the ones here; but in none of their
works are the heuristic procedures ade-
quately explained, justified, or used with
full epistemic power. Bush especially
can also be criticized for logical and syn-
tactical errors in his probably intuitive
usage of these schemata.

Thus, given the present crisis in evolu-
tionary biology over the species problem,
the history of my own previous attempts
to report my sceloporine findings, and
the variety of communication problems
which result from the implicit use of dif-
ferent heuristic schemata (Kuhn, 1962), it
is desirable to take the unusual step to
explain and justify philosophically the a
priori assumptions and logical structures

Advances in Herpetology and Evolutionary Biology

of the schemata which I have followed. I
illustrate this discussion with examples
from the historical development of my
sceloporine studies to demonstrate how
the schemata have been used.

Because of the problems dealt with, I
will unavoidably raise many issues in
this paper which demand more complete
treatment than I can provide here. For
example, the heuristic schemata I
present should be axiomatized and tested
for logical consistency. Their historical
origins should be traced in detail, and
they should be compared with schemata
used by other evolutionists. However,
these issues are beyond the scope of the
present work. Such projects would
minimally take several years’ work, and
the present paper is primarily intended:
1) to explain why and how I have organ-
ized my sceloporine studies as I have; 2)
to explain the unusual logical organiza-
tion of the arguments I use to reach and
defend the conclusions of these studies;
and 3) to provide enough of a philosophi-
cal justification for this approach to show
that it is scientifically valid.

To successfully meet these intentions,
certain pitfalls must be avoided. To illus-
trate some of the traps I will begin with
some (probably oversimplified) com-
ments on relevant trends and problems in
the discipline of the philosophy of sci-
ence. Most of my references will be to
Popper (1972a,b,c); and to the informa-
tion on recent developments in the
philosophy of science in Suppe [ed.]
(77).

At the outset, despite my harsh
comments, I subscribe to most of
Poppers philosophy—as philosophic
ideals. However, the self-limitations of
these ideals should be examined very
critically before any attempt is made to
apply them to a discipline as complex as
evolutionary biology. At least some of the
confusion already rampant in the field is
due to oversimplistic attempts to use
Popperian ideals.

USE OF THE COMPARATIVE APPROACH ° Hall

A CRISIS
SCIENCE

IN THE PHILOSOPHY OF

PROBLEMS WITH INDUCTION AND
SEPARATING SCIENCE FROM FANTASY

Epistemology is concerned with the
problem: How and how much do we
truly know of that which we think we
know? The question is central to all
science, and it still lacks a generally satis-
factory answer. Logic is concerned with
developing rules of argument, which if
properly followed, will yield a true or at
least a probabilistically true output from
a true input. The rules of logical argu-
ment are central to the program of gain-
ing scientific knowledge, but even here,
there are important and unsolved prob-
lems. This is particularly true of induc-
tive reasoning.

Induction, as used in science, is the
process of inferring conclusions about
the existence and details of natural laws
from specific observations of the conse-
quences of these laws. In other words, it
is the logic of making generalisations
from “facts.” Popper (e.g., 1972a, b, c)
and many others have unquestionably
shown that conclusions reached by
inductive methods can never be proven
to be absolutely true. Yet, how does
science gain “knowledge” of natural laws
except through the specific observations
it makes? This paradox is known as the
problem of induction, or Hume’s
problem (Popper, 1934). A related, and
even more important epistemic question
is: how does one distinguish factual,
rational, or scientific knowledge from
fantasies? Popper calls this the “problem
of demarcation’.

Many philosophers assert that because
one cannot argue logically from specific
facts to certain truths about natural laws,
then no method or schema of inductive
discovery can be justified philosophi-
cally as a reliable procedure to produce
probably, or certainly true conclusions

647

(Suppe, 1973). For example (Popper,
1972a: 31, 32): “... The act of conceiving
or inventing a theory, seems to be neither
to call for logical analysis nor be suscep-
tible of it... . There is no such thing as a
logical method of having new ideas, or a
logical reconstruction of this proc-
ess... Or, similarly (Hempel, 1965: 7):
“What determines the soundness of a
hypothesis is not the way it is arrived at
(it may even have been suggested by a
dream or hallucination), but the way it
stands up when tested, i.e., when con-
fronted with the relevant observational
data.”” Thus, many of these workers have
explicitly washed their hands of dis-
covery as being “irrational”, and have
left it to “psychologists, historians, and
sociologists” to explain how scientists
find new ideas (Hanson, 1958; Suppe,
1973; and others in Suppe [ed.], 1977).

Somewhat contradictorily, given their
position on discovery, many philoso-
phers have argued that repeatedly suc-
cessful predictions can statistically (or
inductively) verify the truth of a hy-
pothesis: specific and observable con-
sequences can be deduced with logical
certainty from the hypothesis and stated
initial conditions. If these predictions are
confirmed consistently by a variety of
tests, then it can be assumed that the
theory is probably true (Hempel, 1965).
However, Popper (1972a, b, c) shows in
well-justified arguments that no finite
number of specific facts can ever abso-
lutely prove the universal theory or
generalization to be true. It makes no dif-
ference whether these facts come before
or after the statement of theory. Arguing
from metaphysical assumptions (uniform-
itarianism for the future is not accepted),
Popper goes even further to assert that
repeated substantiations do not even
show a theory to be “‘probably” true. This
is the problem of induction in another
form.

Popper (1972c: chapter 1) claims to
have solved the problem of induction in

648

1933 by accepting that all knowledge
remains theoretical, and that although
some of this knowledge may be true, its
truth can never be proved, even to a
degree of probability. This solution
makes distinguishing science from fan-
tasy all the more important. How then do
we account for the obvious practical suc-
cesses of science versus fantasy in pro-
viding useful understandings of how the
world works? Popper proposes an
“evolutionary theory of knowledge,
which also answers his problem of
demarcation.

According to him, the only logically
defensible test of knowledge is that of
falsification. If knowledge can be stated
in the form of a theory, then specific
predictions can be deduced logically
from it and given initial conditions. (I
pass over the critical and unresolved
problem of what a theory is—cf. Suppe
[ed.], 1977.) With a proper test, the
disconfirmation of a single prediction
logically proves the theory to be false,
but no number of confirmatory tests will
ever prove its truth. Popper then answers
his problem of demarcation by conclud-
ing that no idea is scientific or rational if
it cannot make potentially falsifiable
predictions about the empirical world.
Thus defined, scientific understanding
grows only through proposing bold
hypotheses—hypotheses which cover or
include the “known” phenomena, but
which are “improbable” (or information
rich) because they deductively predict
heretofore unexpected relationships or
situations. These new predictions are
then subjected to empirical tests, or in
Popper's words “ingenious and severe
attempts to refute them.“ As bold hy-
potheses are proposed and tested ever
more stringently against reality, those
which contain high contents of untruth
are falsified and selectively eliminated.
Thus, a bold hypothesis which survives
criticism contains additional information
which is not demonstrably false, even
though its absolute truth cannot be
proved. The net result of Popper’s pro-

Advances in Herpetology and Evolutionary Biology

gram to “discover” scientific knowledge
is then to “evolve” selectively an increas-
ing and empirically realistic understand-
ing of the world (Kuhn, 1965; Popper,
1970, 1972a, b, c).

Caveat emptor. As an ideal, Popper’s
program has much to recommend it, but
as a scientist following Popper's advice to
critically test generalizations against
empirical reality, I find it difficult to ap-
ply. There are several traps in the philo-
sophical ideals abstracted above, but a
digression into metaphysics and seman-
tics is needed before I can adequately
discuss them.

My most basic metaphysical assump-
tions for my practice as a scientist are uni-
formitarianism, empiricism or realism,
and nondeterminism. I see no reason to
assume that anything has happened in
the past, or will happen in the future,
which cannot at least in principle happen
in the present. Nor do I see any justifica-
tion for assuming the existence behind
the empirical world of an ultimate reality
of deterministic causality and essences.
Many aspects of the empirical world are
usefully and pragmatically explained by
the theoretical world of nondeterministic
quantum physics. For those unfamiliar
with the physical evidence for the exist-
ence of this world, I recommend Ejisberg
and Resnick (1974) as a nontrivial intro-
duction to the effects of quantum level
uncertainty on more familiar levels of
matter. Although fairly rigorous, the
nonphysicist should still find this read-
able. If this quantum physical world is
accepted as the basis for reality, the
deduced conclusion is that no causal rela-
tionship involving matter is determin-
istic, except to a (sometimes nearly exact)
approximation. Quantum level uncer-
tainty leads to uncertainty in radioactive
decay and Brownian motion. In tum,
these and other nondeterministic effects
on molecules directly influence such
macroscopically important phenomena as
neuron firing; fertilization; and the
mutation, recombination, and assortment
of chromosomes. Thus, nondeterministic

USE OF THE COMPARATIVE APPROACH : Hall

processes significantly affect the predict-
ability of the development and behavior
of individual organisms.

This world view has profound implica-
tions for the natural philosophy of evolu-
tionary processes. However, one thing
not implied is that these processes must
be completely unpredictable. All that is
indicated is that they cannot be predicted
exactly. Both theory and practice suggest
that stochastic or probabilistic predic-
tions are still possible (e.g., Monod, 1971;
Morowitz, 1968).

To avoid misunderstanding, either
here or in later discussions, the terms
deterministic, nondeterministic, stochas-
tic, random, and cause should be defined
as I use them. A deterministic relation-
ship between two events or things signi-
fies that the occurrence of one specific
event or thing requires that the other also
occur with exact and unchanging quali-
ties and properties in comparison to the
first. The deterministic relationship is
assumed to apply universally. Non-
determinism simply implies that the
relationship between two events or
things is no longer exactly deterministic.
Nondeterministic relationships may be
either stochastic or random. A stochastic
relationship is where one specified event
or thing has a definite but not determinis-
tic effect on the probability of the occur-
rence, qualities, or properties of the
second. In general this relationship can
be described by some probability distrib-
tion function which may reflect the ac-
tions or characteristics of some natural
law. A random relationship is where one
event or thing has absolutely no effect on
the probability of occurrence, qualities,
or properties of the second. A causal re-
lationship between two events or things
bearing a temporal relationship to one
another assumes that the antecedent
event influences the occurrence, quali-
ties, or properties of the subsequent
event or thing through the action of
natural law. This influence may be either
stochastic or deterministic, depending on
the law. The antecedent event or thing is

649

the cause, and the influenced character-
istic(s) of the subsequent event or thing is
the effect.

Given this groundwork, the scientist
wishing to use the ideals of the philoso-
phers of science should beware of the fol-
lowing difficulties:

1) From Plato and Aristotle to at least
the 1960’s most philosophers have been
searching for a logical methodology to
reveal absolute truth in an ideal world of
universal and deterministic laws—a
world which science has convincingly
demonstrated simply does not exist in
empirical reality. By 1969 the philoso-
phers were just getting this message
(Suppe [ed.], 1977).

2) Many situations which scientists are
concemed with are not universal by any
definition of the term (e.g., Kitts, 1977).
Many logical arguments philosophers
use to support their theses depend cru-
cially on the assumption of universal
determinism. For instance, although
Popper accepts nondeterminism in his
world view (Popper, 1972c), his argu-
ment that a generalization can be abso-
lutely falsified by a single disconfirma-
tion is logically invalid if the prediction
to be falsified is nondeterministic. For
stochastic predictions, no number of dis-
confirmations will ever absolutely prove
falsity. All arguments, whether inductive
or deductive, must be criticized on their
initial assumptions and statistical merits,
not by some ideal of absolute truth or
falsity. (But note that the logic of statisti-
cal inference also has critical and unre-
solved foundation problems—Salmon,
1966, 1967.)

However, that hypotheses should be
empirically realistic is still the principle
which makes science different from fan-
tasy. The idea of a test assumes that the
generalization predicts empirically ob-
servable conditions which may in prin-
ciple be distinguished from imaginable
alternative conditions. If such predic-
tions are confirmed, they support the
belief that the generalization is realistic.
If they fail, the generalization is probably

650

unrealistic. However, there are also other
ways to eliminate unrealistic generaliza-
tions which will be discussed in more
detail below.

3) Many philosophers of science, in-
cluding Popper, may be criticized be-
cause they have chosen to ignore proces-
ses used to discover generalizations. Al-
though Popper gave his main thesis the
title, “The Logic of Scientific Dis-
covery, his program of “discovery” is no
more than his evolutionary theory of
knowledge as I have outlined it above.
His “logic” is simply to be “bold’’—to
include as many untried predictions as
possible under a single covering gen-
eralization, and to test as many different
bold generalizations as possible—to en-
tail the maximum information content.
The logical result of this boldness should
be that some increase in realistic infor-
mation will survive testing. Thus, despite
Popper's lip service to the logic of mak-
ing generalizations, he dismisses the
process of discovery as irrational and not
worth considering because it involves
induction.

My conclusions amplify those of Suppe
(1977) and others in Suppe [ed.] (1977):
at least until recently, most philosophers
of “science” have paid scant attention to
how science actually works. Aside from
their quixotic search for truth in a non-
existent ideal world, they have assumed
that one set of heuristic schemata (e.g., the
Hypothetico-Deductive schema) is ade-
quate for all science. Most have also ig-
nored the logic and epistemic signifi-
cance of making generalizations (except
for example, Hanson, 1958). Yet real
science inescapably involves both mak-
ing generalizations and assessing
whether they are realistic. Both compo-
nents yield information on the empirical
realism of the putative knowledge
(Salmon, 1966, 1967; Suppe, 1977), but
philosophers have not even fully clarified
how realistic generalizations are tested,
let alone generated. Nickles (1973) and
Suppe (1977) indicate the path to resol-
ving this situation. It is to accept that each

Advances in Herpetology and Evolutionary Biology

scientific discipline tends to have its own
armory of heuristic schemata, which
evolve as new practices prove better than
older ones in producing empirically re-
alistic knowledge. Philosophers of sei-
ence should then develop a taxonomy. of
the various kinds of problems scientists
wish to solve, identify the specialized
schemata used to solve them, and ascer-
tain the logical and epistemic values of
these schemata.! Then, perhaps philoso-
phers can innovate and justify still better
logics for problem solving. This work is
only beginning (Suppe, 1977).

My study of heuristic schemata used in
evolutionary biology is incomplete.
However, I have isolated those used in
my own work from the data and hypothe-
ses within which they are embedded.
Once isolated, their functions can be
described and their logical and epistemic
values for solving particular kinds of
problems assessed. The schemata are
surprisingly general: they should apply
with little modification to a wide range of
problems with similar logical structures
in a variety of fields beyond biology. My
discussion of the schemata is broken into
four sections: in the first, speciation
represents a class of problems which
share many characteristics. Given these
characteristics the possible forms of a
solution and the possible ways this solu-
tion can be reached are constrained.
Solving a problem of this type is equiva-
lent to locating and then understanding
an unknown but repeated signal hidden

1Chorley and Haggett [eds.] (1967) and Harvey
(1969) have done almost exactly this for geography,
which studies some problems that are structurally
very similar to those of evolutionary biology. Har-
vey (1969) reviews and develops the relevant
epistemological theory much more completely than
I have. These works are commendable exemplars
for what could be done with the heuristic methods
of evolutionary biology. Also interesting is that they
seem to result from a major revolution in the logical
methodology of problem identification and solving
in geography which is at least as fundamental as
that evolutionists now seem to face (see also foot-
note page 660, below).

USE OF THE COMPARATIVE APPROACH : Hall

in the random fluctuations of a noisy
communication channel. The second
section describes schemata used to locate
those features in the noise which should
be included in an explanatory model to
solve the problem. This may be called a
program for discovery. The third section
then discusses how models may be con-
structed once the components of the
potential solution have been located. The
final section then discusses several heur-
istic schemata which may be used to
further test the empirical realism of the
hypothesis. Each of these will be illus-
trated by examples of how I have used
them in the sceloporine program.

THE PROGRAM FOR DISCOVERY BY
COMPARISON

THE PROBLEM TO BE SOLVED: TO EX-
PLAIN COMPLEX STOCHASTIC BUT ITER-
ATED HISTORICAL PROCESSES

The nature of the problem of specia-
tion. Speciation is typical of problems
many evolutionists study. Most evolu-
tionists probably would accept the fol-
lowing generalizations about speciation:
the formation of two or more present
species from one past species results
from historical processes comprised of
individually unique and unrepeatable
combinations of events which bear spa-
tial, temporal, and perhaps causal rela-
tionships to one another. When the ques-
tion of how one past species became two
present species is considered, the only
evidence usually available will be ob-
servable effects of the past history of
events on the present. The effects of
some, or many of these past events, may
be partially or completely obscured by
the effects of other historical events hav-
ing nothing to do with the case of specia-
tion being considered (i.e., are randomly
related to it). Speciation is also a
phenomenon where most causal relation-
ships can be assumed to be nondetermin-
istic (i.e., involving mutation, recombina-

651

tion, migration, selection, and sampling
errors of reproduction and death—all of
which trace stochastic properties from
quantum physics). Although some evolu-
tionists might disagree, there is also no
reason to believe a priori that there is
only one set of qualitatively similar
processes which form new species. The
desired solution to the problem of specia-
tion is to find the most realistic under-
standing of the laws of causation govern-
ing processes which result in speciation.

As defined here with respect to histori-
cal processes, a law is a fundamental
property of nature. Although the scientist
can never prove that the law is known
exactly, it may at least be understood to
an empirically testable approximation.
The most useful understandings of laws
are those which prove to have the great-
est explanatory or predictive power. Ex-
planations should logically account for
observations already made. Predictions
should specify observations to be ex-
pected in unexamined sets of the
phenomenon in question. Predictions
may concern either unexamined proper-
ties of the initial cases studied, or they
may apply to the same types of properties
originally studied but in cases not
studied initially. Obviously, if the ex-
planation is based on unrealistic assump-
tions, if it is not logical, or if its predic-
tions are not substantiated, the under-
standing is poor and should be revised or
replaced.

Generalized statement of the problem.
In its most general sense, the problem of
speciation is to identify those kinds of
past events which result in the formation
of two or more descendent species from
one ancestral species and to explain how
this result is achieved. Thus it belongs to
a general class of fundamentally similar
problems characterized by at least most
of the following assumptions about
properties of the phenomenon to be
explained, “P’, and of the world in which
P occurs:

1) P results from complex historical
processes which are evolutionary

652

bo
—

7)

9)

10)

(i.e., they involve a series of re-
lated events and changes spread
through time) and which produce
the observable consequence(s)
by which P is recognized.

Each process producing P may
be resolved into a finite number
of unit events which are unique
for that particular example in
detail, number, and _= spatio-
temporal relationships (i.e., the
process is “unrepeatable’’).
Causal relationships among the
unit events producing P are
governed by stochastic laws.
These laws derive from funda-
mental properties of matter,
space, and time and may be ex-
plained by them.
Uniformitarianism applies: noth-
ing has happened in the past
which is in principle not observ-
able in the present.

Several causes may affect one
event, and one cause may affect
several events. Such multiple
causes or causations will obey
rational laws of interaction as in 3
and 4, above.

The only evidence of past events
in processes which result in P is
their empirically observable con-
sequences in the present.
These consequences may be
obscured to some unknown ex-
tent by the effects (or “noise’’) of
events unrelated to P.

Although processes producing P
are unique because of their dif-
ferences in detail, P’s are iter-
ated, and similar consequences
produced by similar processes
may be examined independent-
ly.

Processes producing P’s may be
found in various stages of com-
pletion.

The desired solution is to under-
stand how the consequences by
which P is recognized are pro-
duced and to determine the

Advances in Herpetology and Evolutionary Biology

range of initial conditions which
result in P. The understanding
should be as realistic as possible.

Besides their frequency in evolution-
ary biology, problems with these charac-
teristics (or simpler relatives) are found
in disciplines as diverse as the history of
science, geology, economics, and psy-
chology, as recognized by Ghiselin
(1969, 1974). I would add political sci-
ence, ecology, geography, and cosmology
to this list. The comparative approach
should be efficacious for achieving
understandings in all of these fields. It
involves the successive and/or parallel
use of various heuristic schemata and
also includes several stages of corrective
feedback from nature which help the
evolving understanding become more
realistic in its explanations of nature.
Figure | provides a flow chart to indicate
the informational relationships among
the various components of the entire
program of the approach.

SIGNAL AVERAGING SCHEMAS

Overview. The major source of induc-
tive power in the comparative approach
is a logical methodology which facilitates
generalizations about causal relation-
ships among present variables. It works
with sets of facts obtained from observa-
tions of the present characteristics of
multiple examples of P. The core of the
method is a statistical signal averaging or
cross correlation procedure: if multiple
cases of P exist and can be studied, and if
the causes of these P’s are iterated,! then
their effects projected onto present fea-
tures should be detectable as modes of
correlation with one another and with the
features that identify P. The “noise” due
to random features should not correlate
significantly. The added information

1If P results from more than one set of causes, at
least the number of these sets should be small with
respect to the number of cases available for analy-
sis, and there should be several cases of each set.

USE OF THE COMPARATIVE APPROACH °: Hall

2. SELECT APPROPRIATE
TO ILLUSTRATE PROBLEM

SIGNIFICANTLY CORRELATED PHENOME

se
MODEL LOGIC

7. TEST ASSUMPTIONS

a. DEMONSTRATIONS
b. H-D EXPERIMENTS
ce. SIMULATIONS

8. TEST PREDICTIONS

3. COLLECT DATA FROM EXPERIMENTS AND CONTROLS

5. GENERATE MODEL(S) THROUGH ANALOGY,
ETC. WHICH PROVIDE CAUSAL EXPLANATIONS FOR

a. SAME PHENOMENA OF NEW CASES
b. OTHER PHENOMENA OF ORIGINAL CASES
c. OTHER PHENOMENA OF OTHER CASES

653

1. PROBLEM IDENTIFICATION AND INITIAL SPECULATIONS

"EXPERIMENTS" AND "CONTROLS"

COLLECT OTHER NEEDED DATA

4a. FURTHER CROSS CORRELATION
ANALYSES WITH NEW DATA

INDUCTION, 5a. REVISE AND/OR REPLACE
MODEL AS INDICATED BY

NEW CORRELATION ANALYSES

9. TEST RECONSTRUCTIONS

DO MODELS PLAUSIBLY
RECONSTRUCT CASES
ACCORDING TO EVIDENCE

10. A NATURAL PHENOMENON HAS BEEN IDENTIFIED AND UNDERSTOOD, BUT THIS UNDERSTANDING SHOULD
BE HELD ONLY AS LONG AS IT PROVIDES REALISTIC EXPLANATIONS OF OBSERVATIONS ABOUT NATURE

Figure 1.

Flow chart tracing the informational relationships of the heuristic schemata used to discover and test

understandings about the causal processes involved in the evolutionary development of classes of historical

phenomena.

from correlated features should limit
“guessing for an explanation, since the
explanation should be able to account
deductively for the coexistence of the
newly identified features along with
those used to identify P. In other words,
the cross correlation procedure provides
a logical basis for developing bold
hypotheses in Popper’s sense, which will
contain an information content already
shown to be realistic.

More power is gained if stages in the
development of P can be correlated
across similar developmental stages.

Presumably, initial events will be less
obscured in the earlier stages by noise
and subsequent events than they will be
in later ones. Dated stages should also
provide evidence on the temporal evolu-
tion of the processes involved.

It is even better if a group of geneti-
cally or otherwise related cases can be
found which resemble the P cases, but
which lack the features of P itself. These
may be considered as “controls” for the
“experimental” cases which do exhibit
features of P. If the experimental or P
cases show correlated features not found

654

among the selected controls, this pro-
vides strong evidence that these are
causally related to P.

The signal averaging principle will be
amplified and exemplified, as I describe
and discuss the individual steps in the
discovery program (Fig. 1). These ex-
amples serve to introduce the scelopo-
rine studies.

STEP 1: PROBLEM IDENTIFICATION AND
INITIAL SPECULATIONS

General principle. Contrasted with
mere data gathering, any scientific re-
search should begin with a problem. The
problem will generally first be seen by
the scientist as a situation or phenome-
non which is inadequately explained by
his knowledge of the world. The more
clearly the problem can be demarcated
and questions relating to it formulated,
the easier it will be to develop specific

methodologies to find the desired
knowledge.
Background information on __ the

sceloporines. While I was an under-
graduate. in biology at San Diego State
University, from 1961 through 1964, I
became interested in lizard biology and
studied some simple field ecology prob-
lems with local species under the super-
vision of Don Hunsaker II. From my
courses (including Richard Etheridge’s
comparative anatomy of the vertebrates)
and informal contact with Etheridge and
Hunsaker, both experts on sceloporines,
I became familiar with these animals and
the problem they posed.

According to Smith and Taylor (1950),
the lizard fauna of continental North
America included 54 species of Scelo-
porus and perhaps 50 Anolis (also
iguanids). The next most speciose genera
in North American representation were
Eumeces (Scincidae) with 22 species, and
Cnemidophorus (Teiidae) with about 17
species. No other genus contained more
than about 15 species by this taxonomy.
Aside from being remarkably numerous,
Sceloporus species are also ecologically

Advances in Herpetology and Evolutionary Biology

remarkably diverse. They are found in
almost every habitat available to lizards;
over a range from the Canadian border to
Panama, from below sea level in desert
basins to above timberlines in high
mountains, and from extreme tropical
rainforests to extreme temperate deserts.
In many areas three to seven species
occur sympatrically. No other North
American genus approaches this ecologi-
cal diversity or species density,! although
both Eumeces and Cnemidophorus have
wider geographic ranges due to their
occurrence on other continents.

Savage (1958) grouped Sceloporus
with eight other genera (Phrynosoma,
Uta, Urosaurus, Petrosaurus, Callisau-
rus, Holbrookia, Uma, and Sator) to form
the sceloporine branch of the family.
Etheridge (1964) excluded Phrynosoma
from this assemblage, but acknowledged
its close relationship. Etheridge’s (1964)
osteological observations indicated that
Sceloporus are not primitive scelopo-
rines. Sceloporus, Uta, Urosaurus, and
Sator were grouped in a relatively de-
rived position within the sceloporines
(see Fig. 2, from Presch, 1969, based
primarily on Etheridge, 1964). Although
osteology does not differentiate among
these four genera, external morphology
(development and imbrication of body
scales, and the loss of the gular fold)
suggests that Sceloporus are probably
derived even with respect to Uta and
Urosaurus. Also, a comparatively recent
derivation is supported by the fact that
morphological variation among the 54+
species of Sceloporus does not exceed
that found in other iguanid genera with
only 10 to 15 species. A thesis that Scel-

1Cnemidophorus, as it is presently known, may
have up to 5 or 6 species occurring sympatrically in
areas of the Rio Grande Valley of New Mexico.
However, in these cases, all but one or two of the
“species” prove to be diploid or triploid unisexual
parthenospecies, which clearly are the products of
instantaneous speciation. Most of these are proven
to be of hybrid origin (Cole, 1975; Parker and
Selander, 1976).

PHRYNOSOMA (15 sp.) UMA (4 Sp.)

CALLISAURUS (1 Sp.)
HOLBROOKIA (4 Sp.)

e. First cervical

rib lost

USE OF THE COMPARATIVE APPROACH « Hall 655

SCELOPORUS (54 Sp.)
SHOR (2 SS.)
UROSAURUS (10 Sp.)
URAMCESISipey)

f. Clavicles develop
hooks

d. Frontals covered
anteriorly in some

Cc. Lacritial and postfrontal
lost

d. Frontals covered anteriorly

g. Interclavicle wedian process

Shortened

“XN

a. Sternal fontanella increased

in size

b. Sternal ribs reduced to 3

PETROSAURUS (incl.

A. Sternal
B. Sternal ribs 4

STREPTOSAURUS) (4 Sp.)

fontanella of moderate size

C. Lacrinal ana postfrontal present
D. Frontals exposed anteriorly

E. Five pairs of cervical ribs

F. Clavicles without hooks

G. Interclavicle wedian process long.

Figure 2. Phylogenetic relationships and numbers of species of the sceloporine genera. Phylogeny based on
Presch (1969: 286), after Etheridge (1964). Primitive character states designated by capital letters. Small letters
indicate derived character states. Species numbers are from Smith and Taylor (1950). By 1966 taxonomic
revisions changed Holbrookia to 3 species, Petrosaurus to 3, Phrynosoma to 14, and Sceloporus to 57.

oporus had more opportunities for speci-
ation simply because it is an older radia-
tion cannot reasonably explain the strik-
ing species diversity. The conclusion
from these data showed that the species
diversity of Sceloporus was extraordi-
nary compared to other North American
genera, and particularly to their close
relatives.

Hunsaker continually reminded me
that this anomaly presented a fascinating

problem to be explained. However, no
opportunities in biogeography or special-
izations in ecology or morphology ex-
plained why Sceloporus, instead of some
combination of the related genera,
should have achieved such a striking
proliferation of species. The anomalous
species diversity most likely was a result
rather than a cause.

For a cytogenetics course project (Hall,
1964) also suggested by Hunsaker, I col-

656

lected all information then available on
iguanid lizard cytogenetics (Painter,
1921: Matthey, 1931; Cavazos, 1951;
Hunsaker, personal communication;
Schroeder, personal communication;
Zeff, 1962). Data were found for 11
species (Table 1): five were from genera
distantly related to Sceloporus and to one
another. All were reported to have 2n=36
karyotypes, which I argued (Hall, 1964)
were primitive in the Iguanidae (see
Paull et al., 1976, for a discussion of the
logic followed). The remaining six spe-
cies, all Sceloporus or their closest rela-
tives (Fig. 2), were reported to have 2n’s
from 22 to 30 or more. Many of these data
proved to be wrong, but the concentra-
tion of chromosomal diversity in Scelop-
orus has been fully validated (Paull et al.,
1976).

First statement of the sceloporine
problem. As I finished my undergraduate
degree in January 1964, I planned to spe-
cialize in community ecology, but my
ecological interests did not suggest prac-
tical master’s thesis projects. With the
question in mind—“what explains the
extraordinary species diversity of Scelop-
orus?’—I saw that the correlation of
chromosomal diversity and_ species
diversity might lead towards this ex-
planation. I did not expect that the prob-
lem could actually be solved, but that the

Advances in Herpetology and Evolutionary Biology

attempt would help me to understand
better the species concept as a useful
background for studying species interac-
tions in ecological communities. The
comparative approach provided the ideal
tool for attacking the problem.

Even though 2n’s were available for
only 11 of more than 500 species in the
family, because I could see no other ob-
vious correlation with prolific speciation,
I thought that the chromosomal diversity
might offer a clue (Hall, 1964). Chromo-
somal differentiation might allow some
form of non-allopatric “chromosomal
speciation,’ while more conservatively
evolving genera could form species only
by slower allopatric speciation (Mayr,
1963). Thus, chromosomally variable
lineages might form species besides
those formed allopatrically. These extra
species would provide added opportuni-
ties to evolve a wider variety of ecologi-
cal specializations. Sceloporus was parti-
cularly suitable for intensive study, not
only because of the possible association
between species diversity and chromo-
somal diversity, but also because its
many closely related species offered the
possibility of finding relatively early
stages in the process of chromosomal dif-
ferentiation and the possibly associated
speciation (i.e., where morphospecies
were still “polymorphic’’). The compara-

TABLE |. IGUANID KARYOTYPE FORMULAS AVAILABLE IN 1964.

species 2n:formula*
Anolis carolinensis 36:12MM,24m
Dipsosaurus dorsalis 36:12MM,24m
Crotaphytus collaris 36:12MM,24m

Phrynosoma cornutum 36:12MM,24mt
Holbrookia texana
Urosaurus ornatus
Sator angustus
Sceloporus graciosus
Sceloporus undulatus

30:12MM, 18mt
~28:12MM,16m?t

30:12MM, 18m$

Sceloporus occidentalis 22:12MM,10m
Sceloporus olivaceus 22,:12MM,10m
*MIM = Metacentric Macrochromosome, m = microchromosome.

t Actual formulas = 34:12MM,22m.
Actual formula = 30:12MM,18m.
?Actual formula = 22:12MM,10m.

34-36: 12MM,22-24mt

30-36: 12MM, 18-24m!

source

Painter, 1921; Matthey, 1931

Zeff, 1962

Painter, 1921; Zeff, 1962

Cavazos, 1951

Painter, 1921

Painter, 1921

Hunsaker, pers. comm.

Schroeder and Hunsaker, pers. comm.
Painter, 1921

Schroeder and Hunsaker, pers. comm.
Painter, 1921

USE OF THE COMPARATIVE APPROACH - Hall

tive approach might reveal details that
would show how chromosomal specia-
tion differed from other kinds. I also
mentioned but did not discuss in detail
that variation should be examined in
other genera, such as Anolis (not closely
related to Sceloporus) to test further the
reality of the correlation between
chromosomal differentiation and specia-
tion.

STEP 2: SELECT APPROPRIATE “EXPERI-
MENTAL AND “CONTROL” CASES TO
ILLUSTRATE THE PROBLEM

General principle. The principle of
“controlling” a comparative study is
analogous to that of controlling a labora-
tory experiment. In the laboratory, the
experimenter uses an apparatus where an
independent variable can be manipu-
lated to learn its effects on dependent
variables whose variations are presumed
to be causally connected to those of the
independent variable. To exclude effects
in the output of the experimental appara-
tus from unknown or extraneous inputs
not under the experimenter’s control, the
experiment is controlled by an apparatus
which is as similar to the experimental
apparatus as possible, except that the
independent variable is held constant.
Any variation in the output of the control
apparatus is assumed to be experimental
artifact, and subtracted from the output of
the experimental apparatus. The varia-
tion in the dependent variable remaining
after subtracting the artifact should result
from the causal influences of the parame-
ter manipulated by the experimenter.

The principle in a comparative study is
the same, except that “nature” is the
experimenter. The investigator selects a
set of natural experiments, or cases, Cx,,
from nature which exhibit the diagnostic
variables, V*, which identify the
phenomenon of interest, P; and a set of
similar control cases, Co, that do not
show the V* of P. Many features, Vc, will
be constant, or correlate strongly across
both sets of cases. These presumably are

657

not causally related to V* but result from
the selection procedure. Other features,
Vx, will show correlations within the set
of Cx,, but not among the Co.. These Vx
are assumed to be causally related to V*
in the production of P.

Comparative studies differ from the
laboratory experiment in that experimen-
tal cases may be selected either because
they exhibit a presumed “cause” or in-
dependent variable, e.g., the fixation of a
certain kind of chromosomal rearrange-
ment, or because they exhibit the pre-
sumed “effect” or dependent variable,
e.g., speciation. Also, the heuristic
schema requires no _ hypothesis to
account for the relationship of the vari-
ables, V* and Vx, or even an idea of what
the Vx should be. The relationship is
demonstrated as a logical consequence of
the causal connections between the vari-
ables, and not by preconceived beliefs of
the investigator. In practice, there may
be a working hypothesis to explain P,
which can help to select potentially im-
portant variables to study. This kind of
selection is useful, if not too restrictive,
since it is impractical to study all aspects
of the cases (this is the difficulty with
Baconian induction). On the other hand,
preconceptions should not influence
selection of which experimental and
control cases to study. This should be
determined arbitrarily by the way the
problem is defined. In biological sys-
tems, the most obvious and probably
least biased controls would be lineages
which ecologically parallel the experi-
mental lineages, and which are phylo-
genetically closely related to them. If
such controls are not practical, depend-
ing on the problem, controls should be
selected which meet at least one of these
criteria.

Sceloporine experiments and controls.
The research question formulated to
guide the selection of cases for the com-
parative study of sceloporine speciation
was based on the anomalous species
diversity of Sceloporus relative to the
other North American lizard genera: is

658

this anomaly explained by the fact that
Sceloporus formed “extra” species by
nonallopatric speciation mechanisms not
available to conservatively speciating
genera? Does this involve chromosomal
differentiation, and if so, what other fea-
ture(s) result from this speciation me-
chanism (or mechanisms), and how is this
constellation of features produced? The
obvious experimental cases were specia-
tion events in Sceloporus, and the obvi-
ous controls were speciation events in
the other eight sceloporine genera.
Although there was no evidence, ini-
tially, to determine which Sceloporus
species may have resulted from the sup-
posed non-allopatric speciation mecha-
nism(s), it was assumed that Sceloporus
probably included many cases of non-
allopatric speciation in the evolution of
its 54+ species. However, some Scelop-
orus species were presumably also
formed allopatrically. Although later in
the study these would provide the best
controls for nonallopatric speciation, ini-
tially, the cases of allopatric speciation

would represent noise or artifact among

the experimental cases. Thus appropriate
control genera should be studied to
identify features that correlate only with
allopatric speciation, so they could be
subtracted from the Sceloporus data base,
to leave primarily those cases likely to
have resulted from nonallopatric specia-
tion.

The other sceloporine genera serve as
controls for identifying the allopatrically
speciating Sceloporus. These are Petro-
saurus (2 species), Phrynosoma (14 spe-
cies), sand lizards—Callisaurus, Hol-
brookia, and Uma (~10 species), Uta (6
species), Urosaurus (~10 species), and
Sator (2 species). According to Savage
(1958), Etheridge (1964), and Presch
(1969), all share a close common ancestry
with Sceloporus. The biogeographical
opportunities for speciation appear to
have been similar for all of the genera
except the insular Sator: all have distri-
butions centering on the North American
deserts, and belong to the “New North-

Advances in Herpetology and Evolutionary Biology

em Faunal Element” of the North
American herpetofauna (Savage, 1960,
1966) which evolved in situ along with
the development of the North American
deserts (Axelrod, 1950, 1958; Axtell,
1958; Norris, 1958; see also Morafka,
1977). Within this desert environment,
the sceloporine genera show three main
ecological specializations: Phrynosoma,
which are specialized nomadic anteaters;
the sand lizards, which are cursorial
insectivores that run after their prey, and
which normally perch directly on the
ground or small stones; and the Uta-
Sceloporus assemblage, which normally
perch off the ground (on rocks, trees,
bushes, etc.), and wait for prey to come
within easy striking distance. At least Uta
and Urosaurus should _ ecologically
parallel Sceloporus. Other controls are
needed to distinguish speciation of any
kind from non-speciation. These are
provided by studying different popula-
tions included within operationally de-
fined species.

Extending the idea of the natural ex-
periment, if the correlation between
chromosomal diversity and_ species
diversity seen in the sceloporines results
from generally applicable natural laws,
rather than from some unique specializa-
tion of Sceloporus, then the “experi-
ments” should be repeatable: similar
patterns of correlation should be seen
when other prolifically speciose genera
such as Anolis and Liolaemus are com-
pared with their appropriate controls.

STEP 3: INITIAL DATA COLLECTING

General principle. Once appropriate
experimental cases and their suitable
controls have been selected, then as
much information as practical relating to
each should be collected: a “history”
should be developed for each case. The
inductive power of the methodology
depends on not limiting observations to
those parameters which would support
the preconceived speculations. Every
attempt should be made to survey a

USE OF THE COMPARATIVE APPROACH - Hall

variety of parameters, some of which
might potentially relate to P, and some of
which should not.

Data collected or available from the
sceloporines. In my _ studies of the
sceloporines, besides collecting informa-
tion on a variety of cytogenetic parame-
ters from as many species as practical, I
have observed reproductive biology,
anatomy, population biology, and ecol-
ogy. Also, I have tried to keep track of all
published work dealing with any aspect of
the biology of the genera being studied.
These data will be presented in the more
detailed papers to come later in the
present series.

STEP 4: CROSS CORRELATION ANALYSES

General principles: Signal averaging
schemata. The major inductive power in
the comparative approach comes from
heuristic schemata which facilitate gen-
eralizing from specific observations to
assumptions about causal realtionships
among the variables studied from the
case histories. At the same time the
methodology keeps these generalizations
in touch with reality. Modes of inter-
correlated variables which are probably
causally involved with P can be identi-
fied independently from any a _ priori
speculations. The schemata are de-
veloped by analogy from an inductive
statistical procedure known as _ signal
averaging.

The concept of signal averaging is well
known in neurophysiology (Glaser and
Ruchkin, 1976) where a frequently en-
countered problem is to extract iterated,
but otherwise unknown signals, which
are individually completely hidden with-
in random noise in an input channel.
This is the simple signal averaging
schema.

Assume that a noisy input includes a
signal in the form of a voltage fluctuation
which has a fixed time relationship to
some identifiable marker event. A
computer or tape storage can then be
triggered by the marker event to record

659

the voltage fluctuations during the time
sequence believed to contain the un-
known signal. The marker triggered
recording sequence is repeated many
times with the same time delay relative
to the marker event. Each recorded
sequence is placed in register with the
triggering events and added algebraically
to sum the recordings. As many recorded
cycles are added, voltage deviations
which are randomly related to the marker
will tend to an average or null value,
since positive deviations for a given time
delay after the marker will tend to be as
frequent for negative deviations at the
same delay. However, any repeated sig-
nal embedded in the noise which shows
a constant voltage fluctuation relative to
the marker, will always show the same
slight positive or negative deviation
added to the noise. With enough repeti-
tions, a generalized signal will eventu-
ally emerge from the masking noise as a
statistical average of the repeated signals.
Assuming unlimited repetitions and a
constant signal, the signal-to-noise-ratio
can be increased to any desired value by
continued averaging, even though the
signal is undetectable in any one repeti-
tion.

A slightly different schema—an auto-
correction procedure—can be used to
extract unmarked signals from a noise
channel if they are repeated at a constant
interval, even if nothing is known about
the signal’s periodicity or other charac-
teristics. Here, one long recording from
the input channel is chopped into many
short segments of a given duration, and
the segments are then added together in
register relative to the chopping points.
Any recurring signal in the chopped
recording which has a simple harmonic
relationship relative to the period deter-
mined by the duration of the chopped
segment will add algebraically. With the
summing of enough segments, the signal
will emerge from the masking noise as in
the simple signal averaging schema. With
a computer, a long recording from the
channel can be systematically chopped

660

and added this way for a variety of dif-
ferent duration periods until a signal with
the highest signal-to-noise ratio is found
for a constant number of additions. This
presumably occurs when the length of a
single chopped segment exactly coin-
cides with the actual repetition period of
the original signal. Other procedures for
extracting signals from noise have also
been developed, but the basic principles
of signal averaging illustrated above are
all that are needed for the analogies to be
developed here.

Cross-correlation schemata. Analysis
of data from complex, nondeterministic,
but iterated historical processes involves
extracting signals produced by causal
events in the past. These signals are
transmitted from the past through the
noisy communication channel of time
onto the event surface of the present, and
need to be extracted from the variety of
“noise” generated by random historical
events. Simple signal averaging from a
communication channel involves making
essentially longitudinal correlations of
input voltage deviations stored along the
linear axis of a recording tape. By an-
alogy, in a comparative study, when case
histories are made of the present states of
parameters of the selected experiments
and controls, information on the variables
is stored on various parameter axes (one
for each parameter examined) of a data
matrix for each case. The stored informa-
tion in these data matrices or case histo-
ries can then be averaged to search for
correlations among their parameters.
This is exactly analogous to what is done
with the one dimensional data matrices
of the signal averaging schema. Any vari-
ables bearing relatively constant rela-
tionships among the data sets of the case
histories should stand out as strong addi-
tive correlations against the “noise” due
to unrelated events. I call this the n-
dimensional cross-correlation schema.!

1As part of their revolution in heuristic method-
ology (see footnote p. 650), geographers have re-

Advances in Herpetology and Evolutionary Biology

To clarify the analogy, in signal averag-
ing the input triggered by a marker event
is recorded as a voltage deviation along
the linear time axis established by the
uniform movement of a tape past a
recording head, or by the sequential fil-
ling of adjacent “bins” in a computer
memory. If records associated with trig-
gering events are stored individually for
later off-line analysis, each recorded time
deviation sequence is associated with a
unique address for each specific trigger-
ing event. Thus, a one dimensional cross-
correlation matrix is formed. For each
specific triggering event, the input devi-
ations are recorded along the one dimen-
sional time axis. Note that the filled data
matrix is already three dimensional in a
physical sense: one dimension is estab-
lished by a sequence of addresses cor-
responding to the sequence of individual
triggering events. A second is established
by the time axis of the recorded sequence
for each triggering event. And the third is
required to hold the specific deviations
for each given instant of delay along this

cently begun to think and work with multi-
dimensional “data matrices” very similar to the idea
I propose here. Matrix manipulations and multi-
variate statistical procedures (many of them from
numerical taxonomy!) are applied to identify modes
of correlation and relationships among the various
parameters of these matrices (e.g., Berry, 1964;
Haggett and Chorley, 1967; Harvey, 1969; King,
1969). To quote from Harvey (1969: 347-348):

We are concemed to find general classes and
general relationships among attributes, we are
concerned with finding comparable underlying
structures in complex data matrices, and, above
all, we are concerned with identifying a theory
about structures which can command our confi-
dence as an analytic, retrodictive, or predictive
device. Quantitative techniques for classifica-
tion bode well as search procedures. They can
lead us to new ideas, new frameworks for analy-
sis, and so on. [My italics]

Although all of these points are touched on by
Harvey (1969), the geographers have not focused on
the ideas of selecting natural experiments and their
controls, the epistemic contributions of the process
of discovery to the realism of explanations derived
from the process, or the ways the empirical content
of the explanation can be increased through differ-
ent kinds of attempts to refute it.

USE OF THE COMPARATIVE APPROACH - Hall

time axis. The average signal is then ex-
tracted from the matrix by averaging all of
the individual input voltage x time his-
tories to form the single 2-dimensional
plot of the iterated signal.

When the signal averaging schema is
modified to extract information from the
effects of past events on the present, the
method of selecting the cases to be ex-
amined is exactly analogous to the trig-
gering event in the one dimensional sig-
nal averaging paradigm. Clearly, just as
in triggering a recording from a noisy
communication channel, the success of
the n-dimensional procedure will de-
pend on how well selection has limited
the cases to be studied to one, or at mosta
few, underlying causes relative to the
number of cases to be studied. In the
sceloporine lizards, by present count
(Hall, part II) there are about 57 species
which are known to be, or probably are,
chromosomally derived. Each of these
derived species may then be the reflec-
tion on the event surface of the present of
one or more instances of a past event of
“chromosomal speciation”. The remain-
ing approximately 65 species in the radia-
tion all have, or are suspected to have,
identical, presumably conservative 2n=
34 karyotypes. These almost certainly do
not reflect past chromosomal speciation.
The radiation includes approximately
120 species in total. Each pair of species
will include one or more speciation
events in its derivation from a common
ancestor. When averaged, the cases
should form at least two modes of correla-
tion—one resulting from allopatric
speciation and the other(s) from non-
allopatric speciation. Differences be-
tween these pairs can also be compared
with differences between paired popula-
tions belonging to a single species. Once
cases have been selected, an _ n-
dimensional data matrix is established for
each case history. A different axis of the
standard data matrix is used to store ob-
servations for each kind or parameter
being studied across the selected cases.

If features of the selected cases reflect

661

iterated but still unknown underlying
mechanism(s), and if all cases have simi-
lar ages relative to P (i.e., if the triggering
event bears a relatively precise time re-
lationship to the signal), then when a
given axis is averaged from the data
matrices from the selected cases, states of
a parameter having a specific causal
(either caused or causing) relationship to
the process should frequently coincide.
Conversely, states of a parameter only
randomly associated with the process
should be randomly distributed and
should average to some neutral value. In
some cases, effects due to unrelated
processes may also add to the value of a
causally related parameter, but presum-
ably the deviations from these other ef-
fects would also be random with respect
to the value causally related to P. In other
words, the effects of random events
should be exactly analogous to the noise
in a communication channel. For cau-
sally unrelated parameters, observations
would include only noise, and should
average null as the number of record
matrices examined is increased. For a
causally related parameter, the observa-
tions will include a signal embedded in
the noise, which should give increasingly
stronger correlations as the sample size is
increased. Similarly, causal relationships
between different parameters will show
up as cross correlations across the
parameter axes involved.

Here the importance of having “con-
trols” for experimental cases should be
obvious. The controls establish a null or
baseline value for a parameter which
contains only noise relative to experi-
mental cases. Strongly intercorrelated
values from experimental cases which
differ from this null are fairly conclusive-
ly related causally to the factor(s) which
led these cases to be designated as “ex-
periments.

However, actual historical processes
will rarely be of the same age or at the
same degree of completion, and values of
important parameters may vary causally
according to the stage of completion of

662

the process. This presents a problem
which makes the schema more complex,
but its resolution contributes importantly
to the inductive power of the approach. If
available information (e.g., taxonomy and
phylogenetic reconstruction in the scel-
oporine example) allows the cases to be
segregated into subsets that can be
ranked according to approximately
equivalent stages in the process, then it is
probably most efficient to proceed by
doing so. Each subset of case histories
resulting from this segregation is then
averaged to identify correlated parame-
ters associated with particular stages of
the process. A tentative model generated
from such ranked subsets may suggest
better criteria for re-segregating and re-
ranking; which may, in turn, suggest
revisions in the model. This negative
feedback process should lead through a
series of successive approximations to a
stable relationship between the ar-
rangement of the data and the theoretical
generalizations drawn from the data. If
the process does not converge to a stable
relationship, the basic model is probably
fundamentally unrealistic and _ other,
completely different models should be
considered.

In many instances it may not be ob-
vious how to segregate the cases, or the
attempt to do so may have given no
useful results. An auto-correlation
schema may then be used to randomly
segregate the cases into variously sized
subsets. Each possible combination of
cases is then independently checked for
cross correlation of information. The
procedure is analogous to auto-correlating
from a communication channel where
both the periodicity and characteristics of
the signal are originally known. As ap-
plied to historical problems, the schema
may be termed auto-correlation by ran-
dom association.

If many cases are available for study,
and data have been collected for many
parameters of each, auto-correlation by
random association would be extremely
laborious if performed manually. But

Advances in Herpetology and Evolutionary Biology

data involved in such historical studies
may not lend themselves to computer
coding and analysis either. It is in-
triguing to speculate that human memo-
ries may be organized like the n-
dimensional data matrices described for
the cross-correlation and auto-correlation
schemas (Harvey, 1969). Possibly human
memories automatically correlate these
matrices by random association as part of
the unconscious filing system. I would
even suggest that this process might
account for the powerful inductive “in-
tuition” of some comparative biologists.
However, the auto-correlation process
has enough logical rigor so computer
processing could be used where observa-
tions can be suitably quantified, and it is
also one which can be performed manu-
ally by hand sorting where few enough
recording matrices (e.g., data entered in a
standard format on a page) are involved
to make it feasible.

In summation, signal averaging
schemata provide logical methodologies
to reveal correlations among variables
from a given number of empirical ob-
servations of real-world situations. These
correlations provide objective statistical
evidence to support the inductive
generalization that the correlated vari-
ables are causally related. Similarly, the
methodology will also provide objective
statistical evidence on the absence of
causal relationships among other vari-
ables which do not show correlation.
These generalizations are based directly
on the evidence, completely indepen-
dently from any preconception which
may have been held before the observa-
tions were made. The problem of model
building then becomes a greatly simpli-
fied one of plausibly explaining causal
relationships which are corroborated by
already existing statistical evidence. This
aspect of the comparative methodology
may be sufficient to explain “the truly
amazing feature of Darwin’s intellect”
which accounted for “the frequency with
which he was able to ‘guess’ correctly,
even though he lacked the requisite data

USE OF THE COMPARATIVE APPROACH - Hall

and anything like an adequate theory
governing the phenomena” (Hull, 1973:
Ma).

CONSTRUCTING EXPLANATORY MODELS

STEP 5: MODEL GENERATION

General principles. When certain vari-
ables correlate across cases selected be-
cause they demonstrate some particular
P, this correlation provides evidence to
suggest that the variables represent ef-
fects of past events in the causation of P.
Model building then involves develop-
ing a covering explanation that requires
the observed correlations as logical con-
sequences. This resembles Hanson’s
(1958, 1961) program of retroduction
(Achinstein, 1971). Obviously, explana-
tions can require conditions other than
those already observed. An explanation
may depend on various assumed initial
conditions, and unanticipated additional
consequences may follow logically from
these assumed initial conditions. These
dependencies and requirements beyond
the original observations provide oppor-
tunities for testing the realism of the ex-
planation. However, at the very least, any
realistic model should account logically
and simply for those conditions already
shown to correlate with P. Similarly, a
realistic model should not require condi-
tions already observed not to exist.

Developing a realistic explanation
does not test its realism, yet a program
which discovers information helping to
explain P contributes epistemically to the
realism of any explanation which logi-
cally requires this information. Examin-
ing the alternative, to have no discovery
program, demonstrates this: on proba-
bilistic grounds, explanations which
logically require consequences already
observed to correlate with P should en-
tail more empirically realistic informa-
tion than would any single attempt to
explain P by a random guess, no matter
how “bold” the guess. Irrationally gener-

663

ating and then testing many different
bold hypotheses might eventually
produce an empirically realistic explana-
tion for P which entailed as much infor-
mation as an explanation yielded by the
discovery program. However, for such
explanations with similar contents, it
should make no epistemological differ-
ence whether observations corroborating
the required circumstances are made
before or after developing the explana-
tion. Note that this argument assumes
that the information content of the ex-
planations derives from examining the
same number of variables in the same
number of cases (cf. Popper, 1972c). In
practice, guessing for fruitful explana-
tions is difficult when few data limit the
guesses. It is more efficient (and more
justifiable epistemologically) to collect a
series of related case histories in advance
of guessing, and scan them for correla-
tions. Large domains of possible guesses
may then be avoided because observa-
tions already made contradict some con-
ditions demanded by the domains. Thus,
attention for guessing is focused logically
on what is frequently a very much
smaller domain of possible explanations
not already contradicted by observations.

Models to explain modes of scelo-
porine speciation. I will show in detail
later in this series that features of the
cases of sceloporine speciation group
into two particularly distinct modes. The
characteristics of the two modes will be
abstracted here from Hall (1973, 1977) to
provide background for illustrating the
logic followed to explain them. In one
mode, pairs of closely related species
show a sibling type relationship and
belong to conservatively evolving radia-
tions containing few species. The paired
species frequently have an obvious his-
tory of allopatric isolation from one
another and do not differ chromosomally.
Usually, where species are presently in
geographic contact they are so similar
ecologically that close sympatry is pre-
vented. The allopatric speciation model
(Mayr, 1963, etc.) easily explains all of

664

these features, so this mode will not be
considered further here.

In the second mode, pairs of closely
related species have ancestor-descen-
dent relationships and belong to pre-
dominantly linear sequences of phy-
logenetic derivation. These lineages fre-
quently contain many species. Paired
species show little or no evidence of past
allopatric isolation from one another, and
they frequently differ enough ecologi-
cally to coexist in extensive sympatry or
syntopy. Paired species in lineages fre-
quently differ by chromosomal re-
arrangements which can _ potentially
cause meiotic malassortment, thereby
reducing fitness of heterozygotes (i.e.,
heterozygotes show negative heterosis).
Chromosomally primitive species near
the origins of sequences of chromosomal
derivation tend to be ecologically con-
servative, while highly derived species
towards the ends of sequences either
have extreme ecological specializations
or show impressive ecological domi-
nance. Most sequences of chromosomal
derivation involve only one type of
chromosomal rearrangement (e.g., all
fissions or all fusions), and it seems that
rates of speciation accelerate towards the
terminations of the sequences of deriva-
tion. Terminal species at the ends of
sequences of derivation frequently have
either exhausted the karyotypic substrate
for the particular type of chromosomal
rearrangement involved in the sequence,
or show polymorphisms for that kind of
rearrangement. The most closely related
populations known to differ chromo-
somally, form narrow hybrid zones where
the populations meet geographically
(e.g., Hall and Selander, 1973). Paradoxi-
cally, although hybrids in hybrid zones
are fertile and backcross, there seems to
be a complete block to gene flow be-
tween the chromosomally different popu-
lations. No sceloporine species outside of
Sceloporus shows any evidence for
chromosomal differentiation; and even
within Sceloporus, speciation events
between close relatives that differ

Advances in Herpetology and Evolutionary Biology

chromosomally are considerably less fre-
quent than allopatric speciation events
not involving chromosomal differentia-
tion. Yet, a greatly disproportionate
number of Sceloporus species have a
background of recent chromosomal
derivation in their phylogenetic_ his-
tories.

As observations forming this second
mode accumulated, I developed a
chromosomal speciation model to ac-
count logically for the correlations that
emerged (Hall, 1973, 1977). The com-
plete explanatory model resolves into
three major components, each of which
explains some of the features listed
above. One part of the model describes
how negatively heterotic chromosomal
rearrangements can become fixed in
populations and _ initiate speciation.
Another explains how hybrid zones be-
tween chromosomally differentiated
populations function to block gene flow.
The third part of the complete chromo-
somal speciation model explains: 1) the
disproportionate evolutionary successes
of chromosomally derived species in
comparison to species formed allopatri-
cally, 2) the predominantly linear nature
of sequences of chromosomal derivation,
and 3) details of patterns of chromosomal
variation within the lineages. These par-
tial explanations were developed in the
order listed, to account for particular cor-
relations in the growing body of informa-
tion on sceloporine speciation. Without
attempting to detail their actual historical
development, I abstract the reasoning
followed to reach these explanations, to
illustrate the logic of model generation.

The chance fixation model. The first
and most obvious correlation seen with
the prolific speciation of Sceloporus re-
lated the fixation of chromosomal dif
ferences between species to prolific
speciation. Species differ by Robert-
sonian mutations, which, at least theo-
retically, are negatively heterotic be-
cause they cause meiotic malassortment
(White, 1973, 1978). Thus, heterozygous
hybrids and_ backcrosses between

USE OF THE COMPARATIVE APPROACH - Hall

chromosomally different populations
should be less fit than either homozygous
parental type. Two consequences follow
directly from this assumption: 1) Re-
duced hybrid fitness should serve as a
partial barrier to reduce gene flow be-
tween the chromosomally different
homozygous populations. This will make
it easier for them to evolve indepen-
dently. 2) Reduced hybrid fitness should
also selectively favor individual geno-
types which avoid hybridization. Thus,
the chromosomally differentiated popu-
lations should rapidly evolve isolating
mechanisms to prevent hybridization.
Both consequences would work in the
absence of allopatric isolation to speed
the evolution of a completed barrier to
gene flow between the chromosomally
different populations. Neither conse-
quence would occur if the populations
were geographically isolated from one
another.

Thus, in theory, negatively heterotic
chromosomal differences can aid specia-
tion; but how are the differences first
established? Any mutation which has a
strong enough negative heterosis to favor
the evolution of a barrier between incipi-
ent species could not become fixed in a
large, randomly breeding population.
This is because the Hardy-Weinberg
equilibrium frequency equals the fre-
quency of the new (rare) chromosome,
while the frequency of the neutral or
advantageous homozygous mutant
genotype equals the square of the fre-
quency of the new arrangement. In a
large population, essentially all of the
chromosomes carrying a new rearrange-
ment will occur in the selectively disad-
vantageous heterozygous state. Hence
selection will quickly eliminate it, even if
it potentially has a strong selective ad-
vantage in the homozygous state.

However, if the negatively heterotic
mutations should occur in a very small
population, statistical sampling errors in
mating (= genetic drift) become evolu-
tionarily significant. Here, a negatively
heterotic mutation has an evolutionarily

665

interesting possibility of becoming fixed
by chance (about one chance in 10 fora
population size of 10 and a 50% reduction
in heterozygote fitness [Wright, 1941]).
Also, selection against negatively hetero-
tic mutants constantly works to eliminate
whichever chromosome arrangement is
rarest in a population. If a new mutation
once passes 50% frequency by drift,
selection will work to push it to complete
fixation, even against some immigration
of individuals carrying the ancestral
chromosome. Hence, as long as the small
population remains predominantly in-
bred, small amounts of outcrossing per
generation (say up to 10 or 15%) will not
greatly affect the probability of fixing a
new mutation. Given reasonable condi-
tions of population structure, fixation of
the chromosomal rearrangement may
occur entirely without allopatric isolation
of the populations in the classical sense.

In this partial model, initiation of
chromosomal speciation requires certain
antecedent conditions: 1) Chromosomal
heterozygotes are substantially less fit
than either homozygote. 2) Negatively
heterotic mutations occur with appreci-
able frequency. 3) These negatively
heterotic mutations occur in populations
which include many small inbred demes.
It follows from these initial conditions
and the chance fixation model that: 1)
Chromosomally derived species will be
“founded” by initially very small demes
within which the chromosomal re-
arrangements first become fixed. 2) Such
founder populations occur within the
geographical range of the chromosomally
more primitive ancestral stock. 3)
Competitive interactions between the
founder population of the chromosomally
derived incipient species and the much
more massive ancestral stock should
force the derived species to differentiate
ecologically away from the niche of the
more massive ancestral stock. Thus,
given only the correlation between
chromosomal diversity and _ species
diversity, other observed correlations fol-
low logically from the most straight-

666

forward guess to explain the first correla-
tion: species will be formed in an ances-
tor-descendent relationship, and chromo-
somally derived species will be eco-
logically derived relative to their ances-
tral species.

However, predictions of this simplistic
chance fixation model seem to be con-
trary to the evidence in at least one re-
spect: selection against parents which
hybridize to form the negatively hetero-
tic chromosomal heterozygotes should
lead to the rapid evolution of premating
isolating mechanisms to prevent this
hybridization. Evidence from the narrow
hybrid zones in Sceloporus grammicus
(Hall, 1973; Hall and Selander, 1973),
and similar studies in other organisms
(e.g., Nevo and Bar-El, 1976; Szymura,
1976a, b; Szymura et al., in preparation;
White et al., 1964, 1967, 1969), suggests
that selection in these hybrid zones does
not lead to the evolution of premating
isolation. Paradoxically, although the
failure to evolve premating barriers
allows frequent hybridization and back-
crossing rather than just reducing gene
flow, the hybrid zones appear to com-
pletely block it. The second partial
model was developed to explain this
paradox.

The hybrid sink model. If populations
differ by a negatively heterotic mutation
and hybridize, their hybrid zone will in-
clude a relatively high frequency of less
fit heterozygotes. Because of the negative
heterosis of the hybrids, populations in
the center of the zone will have a re-
duced productivity compared with pure
homozygous populations outside of it.
Thus, hybrid populations will put less
pressure on the carrying capacity of their
environment than will the pure popula-
tions. Consequently, there should be a
net migration or diffusion of individuals
from pure populations towards the re-
duced pressure of the hybrid zone. Mi-
grating individuals will carry their genes
towards the “sink” for gene flow, repre-
sented by the chromosomally un-
balanced (and therefore genetically

Advances in Herpetology and Evolutionary Biology

lethal) gametes of the chromosomal
heterozygotes and the balanced gametes
that may combine with them to form
lethally unbalanced zygotes. It seems
reasonable that the diffusion gradient
towards the hybrid zone may become
steep enough to prevent any gene from
leaving the vicinity once it enters the
zone, and that this may happen well
before a complete sterility barrier
evolves. Hence, the hybrid zone may
completely block gene flow between the
pure populations even though hybrids
and backcrosses retain appreciable fer-
tility. It also follows that the residence
time of any gene in the hybrid zone
should be relatively short, as long as
individuals carrying this gene cannot
discriminate absolutely against mis-
mating with chromosomally different
individuals. Thus, the sink would prob-
ably prevent the evolution of premating
isolation in the hybrid zone, even though
selection favors this. No single gene
mutation would be likely to confer its
carrier with a complete discriminatory
ability to avoid mismating, and any gene
conferring only partial discriminatory
ability would almost certainly be lost in
the sink before other mutations allowing
partial discrimination could combine
with it to produce a perfected isolating
mechanism.

Many details of the fully developed
hybrid sink model have been suggested
by field observations of the sceloporine
hybrid zone. These will be discussed in
later papers of this series. However, of
concern here is that hybrid sinks should
exhibit surface tension (Key, 1968, 1974).
Populations on the concave side of a
curved hybrid zone will have a shorter
contact front than will those on the con-
vex (or outer) side of the curvature. If
other circumstances are equal, the popu-
lation on the concave side cannot feed as
many immigrants into the sink as can the
other population. Consequently, the
larger number of immigrants from the
convex side will push the center of the
zone towards the concave side. Where

USE OF THE COMPARATIVE APPROACH - Hall

populations meet on a broad front, this
surface tension effect will tend to
straighten out any kinks in the contact.
But where a newly differentiated founder
population occupies only a very small
area within the range of its parental
stocks, the surface tension will press
inward around this entire circumference.
Many newly differentiated populations
must be quickly swamped by this effect,
thus raising the question: could any
chromosomally differentiated popula-
tions survive swamping in this critical
early state? Attempts to answer this
question and to account for the impres-
sive ecological dominance of the pro-
ducts of one chain of derivation in
Sceloporus led to guessing for the third
part of the chromosomal speciation
model.

The cascading or chain speciation
model. Three sets of assumed or demon-
strated circumstances provide the basis
for this explanation, which also accounts
for the remaining features correlated
with rapid speciation in Sceloporus: 1)
Chromosomally differentiated species
originate as very small founder popula-
tions. 2) The probability that chromo-
somal differentiation will occur at all is
profoundly affected by: a) rates of chro-
mosomal mutation, b) behavior of the
chromosomes in meiosis and the genetic
consequences of any meiotic errors, c)
details of the species’ mating system, and
d) details of its population structure. 3)
The aspects of a species’ genetic system
listed above are genetically controlled.
These circumstances allow evolution by
two unusual mechanisms, a group selec-
tion effect and a positive feedback or
deviation amplifying process (Szarski,
1971). These may counteract natural
selection working at the level of single
individuals.

The genetic system parameters listed
above can vary considerably and still
have minimal effects on individual fit-
ness. Hence, if loci determining them are
polymorphic, demes small enough for fix-
ing chromosomal rearrangements may

667

also show substantial variation in these
parameters due to genetic drift. Over the
range of a species which has an appro-
priate population structure, through
chance some demes will have a much
more favorable genetic background for
chromosomal speciation than will others.
It follows that chromosomal speciation
will more likely begin in these “fa-
vorable’ demes than in others. Founder
populations that survive as new species
will tend to perpetuate the favorable
genetic backgrounds. Thus, chromo-
somally derived species will, on the
average, offer more favorable circum-
stances of further chromosomal specia-
tion than will ancestral species. This
positive feedback amplification process
may work in each chromosomal specia-
tion event in a sequence to produce a
predominantly linear and increasingly
rapid chain of chromosomal speciation.
(see Hall, 1973, 1977, and later papers in
this series). Three circumstances (2 in-
trinsic and | extrinsic) can terminate such
a chain of speciation.

1) If chromosomal fissions and fusions
are qualitatively different kinds of muta-
tions, different genes should affect their
mutation rates. Similarly, different genes
may control their meiotic behavior. Thus,
if a speciation event involved centric fis-
sions, then positive feedback effects
would result in more favorable condi-
tions for further speciation involving fis-
sions. The chain of derivation may then
proceed at an accelerating rate until all
metacentric chromosomes in the karyo-
type were fissioned. At this point, further
chromosomal speciation involving fis-
sions would be blocked, and the genetic
system might not be conditioned to favor
speciation involving other kinds of
chromosomal mutations.

2) One variable which has a directly
proportional effect on the probability that
chromosomal speciation will occur is the
rate of chromosome mutation for a given
kind of rearrangement. This rate could
easily be raised by the positive feedback
process to such a high level that many

668

individuals in a population would suffer
from the negative heterotic effects of
chromosomal heterozygosity. If this
happened, individual selection would
favor evolving modifications in the
mechanics of meiotic assortment to
prevent malassortment in the large
number of heterozygotes carrying new
mutations. Chromosomal heterozygosity
would no longer show negative heterosis
and could no longer produce the hybrid
sink required for successful speciation.
Consequently, the species forced into
this adaptation could form no more spe-
cies chromosomally, but would be left
with a high rate of chromosome mutation,
and should be polymorphic for the kinds
of rearrangements involved in the chain
of derivation in its ancestry.

3) The third type of termination in-
volves extrinsic circumstances. If the
chain speciation process works as I have
suggested, a fairly large number of spe-
cies may form in a geologically short time
under circumstances where they are not
geographically isolated from one another.
If species need an ecological niche of
some minimum width, and if their basic
adaptations allow them to use only a cer-
tain subdivision of the environment, then
as the lineage proliferates new species,
they would be limited to progressively
more restricted niches or geographic
areas. Eventually a point would be
reached where further speciation either
would be impossible or could not occur
without extinction of other forms. If fur-
ther speciation was blocked for long
enough, individual selection would
eventually force genetic system parame-
ters to revert to their “unamplified” ori-
ginal conditions, which would not favor
chromosomal speciation even with new
ecological opportunities.

Group selection in this model is asso-
ciated with the positive feedback
mechanism, but its effects show pri-
marily in the ecological consequences of
chromosomal speciation. The major
“selective” force involved is the surface
tension produced by the hybrid zone sur-

Advances in Herpetology and Evolutionary Biology

rounding newly differentiated founder
populations. According to the competi-
tive exclusion principle, two species
cannot coexist in the same ecological
niche. If a nascent species is formed
within the range of its ancestral stock, the
ancestral and derived forms will in-
evitably compete ecologically. Three
outcomes are possible: 1) One of the
species becomes extinct—most probably
this will be the nascent species. 2) The
two species remain in similar niches but
displace one another geographically
along some environmental gradient (i.e.,
geographical exclusion). 3) The two spe-
cies coexist geographically but displace
one another along one or more resource
axis of the environment (i.e., character
displacement). As mentioned above in
the discussion of the random fixation
model, in character displacement the
ancestral population will almost always
displace the derived population, thereby
forcing it into a new and probably more
specialized niche. However, in geo-
graphic exclusion, exclusion of the ini-
tially very small founder population will
be equivalent to extinction.

When a lineage invades a new en-
vironment or adaptive range, and its
species have few competitors, character
displacement is probably easy (e.g., Wil-
liams, 1972). The ready availability of
vacant sympatric niches and the initial
selection against hybridization in the
random fixation model may allow the
rapid evolution of premating isolation.
One can also imagine that establishment
of a hybrid sink situation along some
resource axis of the environment would
prevent gene flow along the axis so that
disruptive selection would favor the
rapid evolution of sympatric species.
However, once an environment becomes
fairly saturated with related species (as in
the current situation for the Sceloporus
grammicus complex), most chromosomal
speciation probably involves a long
period of geographic exclusion. The
consequences logically required by the
group selection model are interesting.

USE OF THE COMPARATIVE APPROACH : Hall

Because of the surface tension of its
hybrid sink, which occurs independently
of any environmental circumstances, the
nascent chromosomally differentiated
species will be at an immediate disad-
vantage. If the risk of hybridization can-
not be reduced quickly by character dis-
placement and expansion into a sympa-
tric niche (or an uncontested geographic
range) the founder population will prob-
ably be consumed by its own hybrid sink.
Where the founder population cannot
escape hybridization its survival is likely
only if the parental stock can be pushed
back to increase the hybrid zone’s radius
of curvature enough to reduce the surface
tension. Two kinds of chance circum-
stances allow the founder population to
achieve this initial expansion. The first
depends entirely on extrinsic events,
such as a catastrophe, to eliminate an
adjacent ancestral population so the
nascent species can spread into the
depopulated area. The other circum-
stance depends on intrinsic aspects of the
genetic system.

The founder population required for
the chance fixation of a negatively
heterotic chromosomal rearrangement is
ideal for other stochastic phenomena
such as the random fixation or drift of
alleles at a variety of polymorphic gene
loci. Also, founder populations will be
exposed to specific local circumstances
which may result in selection pressures
that differ considerably from the average
selective environment of the ancestral
species. The ancestral species must
maintain an adaptation to some average
habitat occupied by populations ex-
changing genes with one another, while
the initial barrier provided by chromo-
somal differentiation allows the founder
population to adapt to its own specific
local habitat. Thus, although most incipi-
ent chromosomally differentiated species
will be quickly consumed by their own
hybrid sinks, one may rarely achieve a
fortuitous genotype which is sufficiently
superior in its local environment so that it
can feed enough migrants into the hybrid

669

sink to counteract its surface tension. If
the founder population once manages to
push the sink back to a greater radius of
curvature, the pressure will be reduced
and the founder population’s superior
fitness may then allow it to displace the
ancestral stock from a wide geographic
range. Chromosomal _ differentiation
events may become frequent towards the
end of a chain of derivation, but the en-
vironment will probably be too saturated
to allow easy character displacement.
The vast majority of these incipient spe-
cies will be immediately consumed by
their hybrid zones. Barring the lucky
environmental circumstances, the only
populations to survive will be those
which achieve an especially superior
adaption which allows them to over-
whelm the surface tension of their hybrid
zone.

Note that the positive feedback in
chain speciation will also work here.
Each founder population must be fit
enough to be able to displace its im-
mediate ancestor against the added
impediment or pressure of the hybrid
sink’s surface tension. Thus, species at
the ends of chains of geographic dis-
placement should be especially effective
competitors against species formed early
in the chain. Chromosomally conserva-
tive species should survive competition
with their derivatives, because they
presumably held extensive geographic
ranges before chromosomal speciation
began. However, intermediate species in
a chain may easily be excluded into ex-
tinction by their competitively superior
derivatives, because they may have little
chance to expand geographically before
producing further derivatives.

In sum, the discovery program of the
comparative approach has identified in
the sceloporine radiation, independently
of models, a variety of circumstances of
cytogenetics, ecology, and phylogenetic
relationships which appear to correlate
with prolific speciation in Sceloporus.
These are not adequately explained by
the allopatric speciation of Mayr (1963,

670

etc.). The “guessing” to explain the
totality of these relationships has been
very straightforward and has led in easy
steps to a three-part model which fully
explains all of the phenomena revealed
so far by the discovery program. By con-
trast, none of the guesses made in ad-
vance of collecting these data make
predictions which approach the richness
or realism of the circumstances logically
required by the complete model and
reasonable initial conditions. The com-
plete model is, of course, fully corrobo-
rated by the information content of the
conditions it has been developed to ex-
plain. The question remains—can the
realism of these explanations be tested
further? The answer provided by the next
section is yes.

TESTING EXPLANATORY MODELS

General principles. The logic of testing
the realism of explanatory models for
historical processes is not difficult. How-
ever, it is my impression that many
workers who fail to understand how the
models are generated, fail also to under-
stand all of the tests which can be used to
further corroborate them (e.g., Peters,
1976). Under any circumstances it is logi-
cally impossible to prove any universal
explanation to be absolutely true; al-
though if uniformitarianism is accepted,
the explanation can be demonstrated to
be realistic to some degree of statistical
confidence. If the world is fundamentally
nondeterministic, falsity cannot be abso-
lutely proven either. However, the idea
of empirically testing and criticizing the
logically derived statements a theory
makes about nature still must form the
basis that makes science different from
fantasy.

If the program of discovery described
above is followed it will lead to general-
izations which already contain a large
quantity of empirically realistic (or pre-
tested) information. However, the pro-

Advances in Herpetology and Evolutionary Biology

gram of science is always to extend the
frontiers of knowledge, and Popper's
evolutionary theory of knowledge sug-
gests how this should be done, except
rather than beginning by testing wild
guesses, the program begins with testing
generalizations which already include a
large body of pretested and empirically
realistic information. Following Popper
the content of realistic information en-
tailed by the explanation can be in-
creased by attempting to refute it.

There are at least three ways to refute
an explanation (most workers think only
of testing “predictions”, and assume that
an explanation which cannot be attacked
in this way must be metaphysical, e.g.,
Peters, 1976): 1) An explanation may be
refuted if its predictions are consistently
unrealistic. 2) If an explanation depends
logically on certain initial conditions or
assumptions and it can be shown that
these are unrealistic, then the explana-
tion is refuted (Ghiselin, 1969). 3) Also, if
it can be shown that supposed final con-
ditions actually do not follow logically
from supposed initial conditions, the ex-
planation is obviously faulty. Full cor-
roboration of an explanation should re-
quire that it survive all of these attempts
to refute it. Each of these tests will be
explained in more detail and illustrated
with examples from my research pro-
gram.

STEP 6: TEST LOGIC

Principle. In principle, testing the
logic of an explanation requires no addi-
tional data gathering or processing. This
should be the first and most obvious type
of test to apply, but judging by many of
the published attempts to explain non-
allopatric speciation, it is frequently
overlooked. The critical question to be
asked in this kind of test is, do the sup-
posed consequences the model proposes
to explain actually follow logically from
its supposed initial conditions. It should

USE OF THE COMPARATIVE APPROACH - Hall

be possible to reduce deterministic
models to mathematical or logical sym-
bols to see if the consequences do follow.
For stochastic models, some of which can
become quite complex, it should be pos-
sible to test their logic by computer simu-
lation to see if the supposed conse-
quences follow from the initial condi-
tions and the explanation.

Application. Several aspects of the
chromosomal speciation model are sus-
ceptible to testing via computer simula-
tion. The most critical points to attack are:

1) Fixation of negatively heterotic
mutations in small populations: the
probability of fixing a negatively heter-
otic mutation should be calculated or
determined by simulations for a variety
of population sizes, migration rates, and
relative fitnesses of the three possible
genotypes (two homozygous conditions
versus heterozygotes).

2) The hybrid sink effect: for various
conditions of population structure and
density, vagility, and degrees of negative
heterosis, how effective is the hybrid
sink as a block to the penetration of un-
linked or linked alleles? What relative
fitmesses of the two populations are re-
quired to maintain a curved hybrid sink
in equilibrium and how do these vary as a
function of the radius of curvature of the
center of the sink? What conditions
would be required to form a hybrid sink
along an environmental resource axis
rather than a geographical axis?

3) Chain speciation and positive feed-
back amplification: how reasonable is the
group selection argument? For a plau-
sible random fixation model, what muta-
tion rate would be required to generate
enough incipient species to provide a
plausible frequency of species able to
displace their ancestors against the ini-
tially high surface tension pressure?
What changes in the frequency of alleles
in controlling genetic system parameters
are plausible in chromosomal speciation
and how rapidly could selection at the

671

individual level change these _ fre-
quencies? Many other aspects could also
be modeled, but the tests listed here
would seem to have the greatest power to
corroborate the proposed explanations.

STEP 7: TEST ASSUMPTIONS

Principle. Any demonstration that a
necessary d priori assumption of the ex-
planation is unrealistic, immediately
makes the model unrealistic to the extent
that it depends on the assumption, and to
the extent that the assumption is shown
to be realistic. There are many ways to
provide such demonstrations. Any model
which assumes a “fact” of nature is falsi-
fied by demonstrating that the assumed
“fact” is in reality not true.

Applications. The most critical as-
sumption in the chromosomal speciation
model is that the chromosomal re-
arrangements fixed between species
were negatively heterotic when they
originally occurred as new mutations.
Presumably, where they are involved in
narrow hybrid zones which appear to
function as sinks, such as found in the
Sceloporus grammicus complex, they
should still be negatively heterotic.
Thus, examining meiotic assortment in
heterozygotes from these hybrid zones
should test this important assumption.
Such tests have already been done for a
similar situation of rapid chromosomal
differentiation in European mice of the
genus Mus (Cattanach and Moseley,
1973), where single metacentric chromo-
somes have been backcrossed for the test
into essentially pure Mus musculus geno-
types to isolate them from possible genic
problems in the hybrids. It should also be
possible to do the same with Sceloporus
grammicus. Another critical assumption
concems population subdivision. In
sceloporines the most recent chromo-
somal speciation appears to have been in
the Sceloporus grammicus complex.
Superficially it appears to have an ideally

672

subdivided population. However, the
assumption can be tested further by mak-
ing detailed observations on population
structure, mating system and vagility by
using mark-observation and recapture

methods.

STEPS 8 AND 9: TESTING PREDICTIONS

Principles. If an explanation requires
potentially observable circumstances
(conditions which should be causally
related to P) which have not been ex-
amined, then attempts to demonstrate
these circumstances will test the expla-
nation. For stochastic models, confirma-
tion or disconfirmation of these predic-
tions will support or reject the explana-
tion with some degree of statistical con-
fidence. Such tests can take at least two
qualitatively different forms: testing
predictions in a relatively strict meaning
of the term, and testing the ability of the
model to reconstruct case histories re-
alistically.

Testable predictions can be sub-
divided into three categories: 1) The
model may predict already observed
classes of phenomena in new cases (i.e.,
the natural experiments should be re-
peatable). 2) The model may predict new
classes of phenomena in cases already
observed. 3) The model may predict new
classes of phenomena in new cases.
These are listed here in the order of the
epistemic values of the tests from least to
most (cf. Popper, 1972a, b, c). However, it
should be recognized that an efficient
discovery program may already have
found all of the classes of phenomena
causally related to P. If the explanation
covers these, the failure of the model to
predict “new classes of testable
phenomena” does not necessarily sug-
gest that the model is unscientific as
Popper might claim. If both the logic and
assumptions of the model have been
tested, the failure to predict new phe-
nomena suggests that the discovery pro-
gram has already provided the model

Advances in Herpetology and Evolutionary Biology

with nearly the maximum content of in-
formation available in the domain it
covers. That is, the model has been
proven to be realistic within the limita-
tions of the real world. This would cor-
respond to Kuhn’s (1962, 1970a, b) “‘nor-
mal science’.
Historical explanations can be used to
make “predictions” of still another kind.
Models will usually be expressed in the
form of generalizations about assumed
initial conditions and the kinds of effects
expected to follow from them. Aside from
predicting general classes of phenomena,
as discussed above, the explanations
should also allow the past histories of
specific cases to be reconstructed in de-
tail. If evidence not already included in
the model by the discovery program al-
lows an independent reconstruction of
the past histories of cases, then the
realism of the model is tested to the
degree that the independent reconstruc-
tions are realistic and coincide with the
reconstructions provided by the model.
Applications. Except for requiring cer-
tain initial conditions which can be
tested empirically, the chromosomal
speciation model predicts no phenomena
not already suggested by the discovery
program. However, these “predictions”
can be tested over a wider range of scel-
oporine species than have been ex-
amined to date. Also, the predicted pat-
terns of chromosomal variability and
phylogenetic relationships should be
found in other speciose genera of the
Iguanidae (i.e., in Anolis and Liolaemus)
and they should not occur in other small
genera (Paull et al., 1976 have already
begun this test). Also, can these patterns
be found in other lizard families, other
classes of vertebrates, and other phyla?
The most critical tests available are those
of the accuracy of historical explanations.
The chromosomal speciation model
greatly constrains possible phylogenies
which can be reconstructed from karyo-
typic evidence. Do these karyotypic
phylogenies correspond to phylogenies

USE OF THE COMPARATIVE APPROACH - Hall

which can be reconstructed from evi-
dence provided by independent charac-
ter sets?

OTHER CONSIDERATIONS

Corrective feedback and the growth of
knowledge. Tests provide new empirical
observations which tend to falsify, sup-
port, or suggest extending the tested
model. Whatever the “favorable” or “un-
favorable” import of the observations for
the model being considered, these new
data increase the information content of
the domain of the explanation which in-
cludes the observations by giving a more
complete view of the world than existed
previously (Popper, 1970, 1972c). A
methodology which does not take ad-
vantage of this input is less than fully
effective. Where some assumption of a
model has been proven false, the obvious
response is to find another explanation
for the available data which does not
depend on this assumption. If this cannot
be done, perhaps the problem is poorly
defined and the new data may suggest a
clearer formulation of the research ques-
tion, further data collecting, and/or look-
ing at the available case histories in new
combinations. Predictive or reconstruc-
tive tests automatically provide new data
which can be entered in the cross correla-
tion analyses to improve their inductive
precision, and the evidence provided by
the new data should clearly indicate the
strengths and weaknesses of the tested
model. Assuming that a model has at least
some contact with reality, this feedback
should lead through successive ap-
proximations of data gathering, testing,
and revisions to ever more realistic
understandings of the fundamental laws

governing the processes modeled
(Popper, 1972).!

1See Lakatos and Musgrave [eds.] (1970) for a
debate between Popper, Kuhn, and their various
followers and critics about their respective ideas
concerning how scientific “knowledge” grows and

673

Semantics and syntax. The final
epistemological difficulties of a method-
ology for the comparative study of evolu-
tionary problems concem the precision
of its tools. Since the comparative ap-
proach seeks to understand processes
which cannot be manipulated readily in
the laboratory, information and specula-
tions concemming them must be ab-
stracted through several levels of lan-
guage before they can be manipulated
logically in building and testing explana-
tory models. The logical sequences of
abstracting inductively from reality to
formulating a model, and reflecting de-
ductively back to reality for testing the
model, both depend on the precision of
the symbols used for these manipula-
tions. The conclusion of any logical
argument of this nature can be no
stronger than the weakest or foggiest de-
finitions used in the chain of reasoning.
(Note that this is a very different problem
from determining the “essential” mean-
ing of a word; cf. Popper, 1976.) Many
proposed explanations of non-allopatric
speciation are unrealistic because of
logical problems introduced by using
common words with semantically or
syntactically faulty definitions. These
faults are easily made by using as de-
scriptors of nature (i.e., “facts’) terms
which were originally defined on the
basis of some unsubstantiated model of
nature. As Popper (e.g., 1970, 1976) and
others (e.g., Kuhn, 1962; and others in
Suppe [ed.], 1977) have shown, all words
are theory laden, so the problem is un-
avoidable to some degree; but the utmost
should be done to minimize it, and it is
certainly unwise for an investigator to
forget that it exists. Another reason for
the confusion of proposals cited in my
introduction results from inconsistent
use of terms in discussing speciation
phenomena between the different

can be defined through its contacts with empirical

reality.

674

“scientific communities” (Kuhn, 1970:
176) concerned with the problems of
species and speciation.

Thus, although many before me have
discussed these definitions, I think it
essential for the following papers in this
series to redefine much of the vocabulary
concerned with geographic and genetic
relationships of populations and species,
and with isolating mechanisms and their
failures. A glossary of these definitions
will be provided later in this series.

CONCLUSIONS

Solving the problem of speciation. To
understand realistically how new species
form is probably the most difficult task
evolutionists have attempted to achieve.
Besides confronting the unresolved prob-
lem of what species are, these attempts
confront the fundamental epistemologi-
cal problem of determining how realistic
the explanations are, when speciation
cannot be directly manipulated or ob-
served. I propose that the comparative
methodology described above answers
the epistemological problem and _ pro-
vides an efficacious approach towards
explaining how species are formed, and
thus also, what they are. Many evolution-
ists claim to use comparative method-
ologies but few appear to understand
clearly the epistemic power of the ap-
proach; so far as I know, none have ade-
quately formulated or justified the logical
methodology. From this, and my own
experience with the approach, I conclude
that it is sufficiently novel to be ex-
plained in detail and justified philo-
sophically. Therefore, in the present
paper I have presented and justified
epistemologically the logical schemata I
have followed in my own attempts to
solve the problem of how species are
formed. These schemata offer seemingly
novel tools, both for discovering causally
related effects of historical processes
hidden among the noise of unrelated
events, and for empirically testing the

Advances in Herpetology and Evolutionary Biology

realism of any models proposed to ex-
plain the process and its causally related
effects. In justifying these schemata I
confront and tentatively answer two ma-
jor problems in the philosophical founda-
tions of scientific methodology: 1) Is
there an inductive logic which can be
used to infer realistic generalizations
about nature from specific observations
of nature, and if so what is the epistemic
value of this logic? 2) How can one test
the empirical realism of generalizations
about historical processes, when the
processes cannot be observed or manipu-
lated experimentally? The answers to
these questions provide a methodology
which appears to be effective for under-
standing speciation, and a surprisingly
wide diversity of other historical prob-
lems ranging from cosmology through
economics and possibly even to funda-
mental particle physics. The approaches
to answering these questions also seem
to indicate the way to solutions of some of
the foundation problems of the philoso-
phy of science.

The inductive power of the compara-
tive approach is most effective if it can be
applied to a compact phylogenetic
radiation which can be subdivided into a
series of genetically and ecologically
related cases, some of which serve as
“natural experiments” to demonstrate
the phenomenon to be explained, and
others which serve as “controls” because
they are similar to the “experiments” in
most respects, but do not show the ex-
perimental phenomenon. The 120+
species of the sceloporine radiation of the
lizard family Iguanidae provide an ideal
radiation for this comparative approach.
The radiation involves at least two quali-
tatively different modes of speciation
which may be compared and contrasted
by the comparative methods to help iso-
late specific features that help to explain
the differences. I abstract my research
program on the comparative cytogene-
tics, speciation, and evolution of the
sceloporine lizards to illustrate how the
comparative methodology has actually

USE OF THE COMPARATIVE APPROACH - Hall

been used in my own research program.
Thus, the present paper serves to intro-
duce the series of papers to follow, which
will present in detail the results of the
research program on sceloporine specia-
tion. It also serves as a methodology sec-
tion for this program.

Modes of speciation: where we stand.
To illustrate and recapitulate the epis-
temological problems the present paper
faces and attempts to solve, I offer the
following quotations as a necessary
background to the attempt to understand
sceloporine speciation:

It is rather discouraging to read this perennial
controversy because the same old arguments are
cited again and again in favor of sympatric specia-
tion.... In the last analysis, all the various
schemes make arbitrary postulates that at once
endow the speciating individuals with the attri-
butes of a full species. They attempt thus to by-
pass the real problem of speciation. One would
think that it should no longer be necessary to
devote so much time to this topic, but past ex-
perience permits one to predict that the issue will
be raised again at regular intervals. Sympatric
speciation is like the Lemmaean Hydra which grew
two new heads whenever one of its old heads was
cut off. There is only one way in which final
agreement can be reached and that is to clarify
the whole relevant complex of questions to such
an extent that disagreement is no longer possible.
(Mayr, 1963: 451)

False facts are highly injurious to the progress
of science, for they often endure long; but false
views, if supported by some evidence, do little
harm, for everyone takes a salutory pleasure in
proving their falseness; and when this is done,
one path toward error is closed and the road to
truth is often at the same time opened. (Darwin,
1889: 606, as quoted in Ghiselin, 1969—
emphasizing the importance of distinguishing
syntactically between observations and explana-
tions)

For evolutionary biology... history... pro-
vides the needed key to scientific knowledge.
Our approach to evolutionary and population
problems should involve historical reconstruction
as well as more traditional comparative and ex-
perimental techniques. We must account for what
actually has occurred: what might take place
under ideal conditions will never do. Otherwise
we shall justify our conceptions of things from
our ideas, rather than letting the way things are
determine how we shall conceive of them. [Italics
mine] (Ghiselin, 1974: 27, stressing the impor-

675

tance of not begging the conclusion by working
logically from unrealistic initial assumptions [i.e.,
to avoid the error of petitio principii])

Two alternatives seem open to us at this junc-
ture: either revert to a hypothetic-inductive
model of science or argue that evolutionary
theory after a century is still inadequately formu-
lated and that in a more finished form will con-
form to the H-D model. The problem with the
first alternative is that there is no H-I model.
Thus, this alternative reduces to the admission
that there is no reconstruction of science appro-
priate to evolutionary theory as it now stands.
Biologists are currently working on the second
alternative. Some are attempting to reformulate
macro-evolutionary theory more rigorously so
that deductive confirmation or disconfirmation is
possible. Others find evolutionary theory in terms
of organisms and their interactions too crude to
permit an adequate formulation of evolutionary
theory. Instead, they call for a molecular version
of evolutionary theory, hoping in this manner
to fulfill the requirements of the H-D model. ...
(Hull, 1973: 34-35)

I would claim that the comparative
approach outlined in this paper is as logi-
cally rigorous both for making inductive
generalizations and for deductively test-
ing them as is possible, given a funda-
mentally stochastic reality. It remains to
demonstrate the utility of the method-
ology, as will be attempted in the follow-
ing papers.

ACKNOWLEDGMENTS

The heuristic schemata trace from
diverse exemplars of methodology, be-
ginning with my father’s geology back-
ground. My formal training started with
physics—to statistical mechanics and
atomic physics. I changed to biology at
UCLA in 1960 with Raymond Cowles’s
natural history, which taught “population
thinking” and anti-essentialism, and with
William Beckwith’s comparative eth-
ology, which taught the “natural experi-
ment.” G. A. Misrahy (Sensory and
Developmental Physiology Lab, Chil-
dren’s Hospital, Los Angeles—NIH
Training Grant 2B-5264) taught “black-
box”” systems analysis, controlling ex-

676

periments, and how to use signal averag-
ing computers.

The sceloporine study owes its genesis
to Don MHunsaker, II and Richard
Etheridge, who taught me at San Diego
State University from 1961-1965.
(Etheridge also demonstrated the com-
parative approach in his teaching.) Dis-
cussions and/or manuscript reviews by
Don Gartside, George Gorman, Max
Hecht, Bernard John, Max King, Emst
Mayr, Jeffry Mitton, Hobart Smith, Jacek
Szymura, Henryk Szarski, Bruce Wallace,
Michael White, Max Whitten and many
students, among others, helped greatly.
Special thanks go to Ralph Axtell and
Ernest E. Williams, at Southern Illinois
University-Edwardsville and Harvard,
where I did most of the field work,
and to Scott Moody, who identified
communication problems in my 1977
paper. Roslyn Dunn (now my wife) typed
the manuscript and eliminated some of
my worst prose. Hall (1977) was com-
pleted at the University of Colorado,
Boulder, and the University of Mel-
bourne Research Fellowship supported
the present work. Field work was sup-
ported by grants from the Society of the
Sigma Xi (1966 and 1968), Evolutionary
Biology Committee Training Grants
(Harvard, 1968 and 1969), and National
Geographic Society Grants 864 and 972
(1970-1971).

LITERATURE CITED

ACHINSTEIN, P. 1971. Law and Explanation: An
Essay in the Philosophy of Science. London,
Oxford University Press, 168 pp.

AXELROD, D. I. 1950. Evolution of desert vegetation
in western North America, pp. 215-306. In
D. I. Axelrod (ed.), Studies in Late Tertiary
Paleobotany. Contrib. Paleont., Carnegie Inst.
Washington, Publ. No. 590.

—__. 1958. Evolution of the Madro-Tertiary Geo-
flora. Bot. Rev., 24: 433-509.

AXTELL, R. W. 1958. A monographic revision of the
iguanid genus Holbrookia. Ph.D. Thesis,
University of Texas, Austin, 222 pp.

Berry, B. J. L. 1964. Approaches to regional analy-

Advances in Herpetology and Evolutionary Biology

sis: A synthesis. Ann. Assoc. Amer. Geogr., 54:
2-11.

BusH, G. L. 1969. Sympatric host race formation and
speciation in frugivorous flies of the genus
Rhagoletis (Diptera, Tephritidae). Evolution,
23: 237-251.

____. 1974. The mechanism of sympatric host race
formation in the true fruit flies (Tephritidae),
pp. 3-23. In M. J. D. White (ed.), Genetic
Mechanisms of Speciation in Insects. Sydney,
Australia and New Zealand Book Co.

___. 1975a. Sympatric speciation in phytophagous
parasitic insects, pp. 187-206. In P. W. Price
(ed.), Evolutionary Strategies of Parasitic In-
sects. London, Plenum.

___. 1975b. Modes of animal speciation. Ann. Rev.
Ecol. Syst., 6: 339-364.

Busu, G. L., S. M. Case, A. C. WILSON AND J. L.
PATTON. 1977. Rapid speciation and chromo-
somal evolution in mammals. Proc. Nat. Acad.
Sci. USA 74: 3942-3946.

CaIN, A. J. 1953. Geography, ecology and coexis-
tence in relation to the biological definition of
the species. Evolution, 7: 76-83.

CATTANACH, B. M., AND H. MOSELEY. 1973. Non-
disjunction and reduced fertility caused by the
tobacco mouse metacentric chromosomes.
Cytogenet. Cell Genet., 12: 264-287.

Cavazos, L. F. 1951. Spermatogenesis of the
homed lizard Phrynosoma cornutum. Amer.
Nat., 75: 373-379.

CHORLEY, R. J., AND P. HAGGETT (EDS.). 1967.
Models in Geography. London, Methuen &
Co., 816 pp.

CoLe, C. J. 1975. Evolution of parthenogenetic
species of reptiles, pp. 340-355. In R. Reinboth
(ed.), Intersexuality in the Animal Kingdom.
New York, Springer-Verlag.

Darwin, C. 1889. The Descent of Man, and Selec-
tion in Relation to Sex, 2nd. Ed. New York, D.
Appleton.

DOBZHANSKY, T. 1937. Genetics and the Origin of
Species [lst. Ed.]. New York, Columbia Uni-
versity Press.

___. 1970. Genetics of the Evolutionary Process.
New York. Columbia University Press, 505 pp.

EHRLICH, P. R., AND P. H. RAVEN. 1969. Differentia-
tion of populations. Science, 165: 1228-1232.

EISBERG, R., AND R. RESNICK. 1974. Quantum Phy-
sics, Atoms, Molecules, Solids, Nuclei, and
Particles. New York, John Wiley & Sons, 713
pp. + Appendices and Index.

ENDLER, J. A. 1973. Gene flow and population dif-
ferentiation. Science, 179: 243-250.

____. 1977. Geographic Variation, Speciation, and
Clines. Monogr. Pop. Biol. no. 10, Princeton,
Princeton University Press, 246 pp.

ETHERIDGE, R. 1964. The skeletal morphology and
systematic relationships of sceloporine lizards.
Copeia, 1964: 610-631.

USE OF THE COMPARATIVE APPROACH - Hall

GHISELIN, M. T. 1969. The Triumph of the Dar-
winian Method. Berkeley, Univ. of California
Press, 287 pp.

____. 1974. The Economy of Nature and the Evolu-
tion of Sex. Berkeley, Univ. of California Press,
346 pp.

GLASER, E. M., AND D. S. RUCHKIN. 1976. Principles
of Neurobiological Signal Analysis. New York,
Academic Press, 471 pp.

GOLDSCHMIDT, R. 1955. Theoretical Genetics.
Berkeley, Univ. of Califormia Press.

HAGGETT, P., AND R. J. CHORLEY. 1967. Models,
paradigms and the new geography, pp. 19-41.
In Chorley and Haggett (eds.).

HALL, W. P. 1964. Cytogenetic studies in the family
Iguanidae. Unpublished manuscript. San
Diego State University Dept. Zoology, San
Diego.

___.. 1973. Comparative population cytogenetics,
speciation, and evolution of the iguanid lizard
genus Sceloporus. Ph.D. Thesis, Harvard
Univ., Cambridge, Massachusetts, 193 pp. +
addenda.

____. 1977. Cascading chromosomal speciation and

the paradoxical role of contact hybridization as

a barrier to gene flow. Informally Published

[see footnote 1, p. xxx, this paper], 91 pp.

. Modes of speciation and evolution in the
sceloporine iguanid lizards. II. An overview of
chromosomal variation and phylogeny. In Pre-
paration.

HALL, W. P., AND R. K. SELANDER. 1973. Hybridiza-
tion of karyotypically differentiated popula-
tions in the Sceloporus grammicus complex
(Iguanidae). Evolution, 27: 226-242.

HANSON, N. R. 1958. Patterns of Discovery, An
Inquiry into the Conceptual Foundations of
Science. London, Cambridge University Press,
241 pp.

____. 1961. Is there a logic of discovery?, pp. 20-42.
In H. Feigl and G. Maxwell (eds.), Current
Issues in the Philosophy of Science. New York,
Holt, Reinhart and Winston.

Harvey, D. 1969. Explanation in Geography. Lon-
don, Edward Amold, 521 pp.

HEMPEL, C. G. 1965. Aspects of Scientific Dis-
covery and Other Essays in the Philosophy of
Science. New York, The Free Press, 504 pp.

HULL, D. L. 1973. Introduction, pp. 1-77. In D. L.
Hull (ed.), Darwin and His Critics, The Recep-
tion of Darwin’s Theory of Evolution by the
Scientific Community. Cambridge, Harvard
Univ. Press.

___. 1976. Are species really individuals? Syst.
Zool., 25: 174-191.

HUunr, W. G., AND R. K. SELANDER. 1973. Biochemi-
cal genetics of hybridization in European
house mice. Heredity, 31: 11-33.

HUux.ey, J. S., (ED.). 1940. The New Systematics.
Oxford, Clarendon Press.

677

____. 1942. Evolution, the Modem Synthesis. Lon-
don, Allen and Unwin.

JOHN, B., AND K. R. LEwis. 1965. Genetic speciation

in the grasshopper Eyprepocnemis plorans.
Chromosoma, 16: 308-344.

____.. 1966. Chromosome variability and geographic
distribution in insects. Science, 152: 711-721.

Key, K. H. L. 1968. The concept of stasipatric speci-
ation. Syst. Zool., 17: 14-22.

___. 1974. Speciation in the Australian morabine
grasshoppers—taxonomy and ecology, pp. 43—
56. In M. J. D. White (ed.), Genetic Mecha-
nisms of Speciation in Insects. Sydney, Aus-
tralia & New Zealand Book Co.

KING, L. J. 1969. Statistical Analysis in Geography.
Englewood Cliffs, Prentice-Hall, 288 pp.
Kitts, D. B. 1977. Karl Popper, verifiability, and

systematic zoology. Syst. Zool., 26: 185-194.

KOESTLER, A. 1967. The Ghost in the Machine.
London, Pan Books, Ltd., 384 pp.

Kuun, T. S. 1962. The Structure of Scientific Revo-
lutions. Chicago, University of Chicago Press,
173 pp.

____. [1965.] Logic of discovery or psychology of
research, pp. 1-23. In Lakatos and Musgrave
(eds.), 1970.

____. [1969.] Second thoughts on paradigms [Ses-
sion VI of Symposium, March 26 to 29, 1969],
pp. 459-519 [incl. commentary]. In F. Suppe
(ed.), 1977.

__.. 1970a. Postscript, 1969, pp. 174-210. In T. S.
Kuhn. The Structure of Scientific Revolutions,
2nd Ed., Enlarged. Chicago, University of
Chicago Press.

___. 1970b. Reflections on my critics, pp. 231-278.
In Lakatos and Musgrave (eds.), 1970.

LAKATOS, I., AND A. MUSGRAVE (EDS.). 1970. Criti-
cism and the Growth of Knowledge. Proceed-
ings of the Intemational Colloquium in the
Philosophy of Science, London, 1965, Vol. 4.
London, Cambridge Univ. Press, 282 pp.

Lewis, H. 1966. Speciation in flowering plants.
Science, 152: 167-172.

___. 1973. The origin of diploid neospecies in
Clarkia. Am. Nat., 107: 161-170.

LEwIs, H., AND P. H. RAVEN. 1958. Rapid evolution
in Clarkia. Evolution, 12: 319-336.

LOVETRUP, S. 1975. Letter. Syst. Zool., 24: 507-511.

MASTERMAN, M. 1970. The nature of a paradigm, pp.
59-89. In Lakatos and Musgrave (eds.), 1970.

MATTHEY, R. 1931. Chromosomes de reptiles—
sauriens, ophidiens, cheloniens. L’evolution
de la formule chromsomiale chez les sauriens.
Rev. Suisse Zool., 38: 117-186 + 8 pl.

___. 1964. La signification des mutations chromo-
somiques dans les processus de speciation:
Etude cytologique du sous-genre Leggada
Gray (Mammalia—Muridae). Arch. Biol., 75:
169-206.

____. 1966. Genetic mechanisms and speciations of

6/5

mammals, pp. 1-11. In G. Gergamian (ed.),
The Chromosomes, Structural and Functional
Aspects. In Vitro, Publ. Tissue Culture Associ-
ation.

Mayr, E. 1942. Systematics and the Origin of Spe-
cies. New York, Columbia University Press,
334 pp.

____. 1947. Ecological factors in speciation. Evolu-

tion, 1: 263-288.

1954. Change of genetic environment and
evolution, pp. 157-180. In A. C. Hardy and
E. B. Ford (eds.), Evolution as a Process.
London, Allen and Unwin.

____. 1963. Animal Species and Evolution. Cam-
bridge, Harvard Belknap Press, 797 pp.

___. 1970. Populations, Species and Evolution.
Cambridge, Harvard Belknap Press, 453 pp.

Monop, J. 1971. Chance and Necessity: An Essay
on the Natural Philosophy of Modem Biology
[Trans. from the French by A. Wainhouse].
New York, Vintage Books, 199 pp.

MoraFka, D. J. 1977. A Biogeographic Analysis of
the Chihuahuan Desert through its Herpeto-
fauna. Biogeographica Vol. 9. The Hague, Dr.
W. Junk B. V., Publ., 313 pp.

MorowiTz, H. J. 1968. Energy Flow in Biology:
Biological Organization as a Problem in
Thermal Physics. New York, Academic Press,
179 pp.

Murray, J. 1972. Genetic Diversity and Natural
Selection. Edinburgh, Oliver and Boyd, 128

pp.

NEvo, E., AND H. Bar-EL. 1976. Hybridization and
speciation in fossorial mole rats. Evolution, 30:
831-840.

NEVO, E. Y., J. Kim, C. R. SHAW, AND C. S. THAELER
JR. 1974. Genetic variation, selection and
speciation in Thomomys talpoides pocket
gophers. Evolution, 28: 1-23.

NICKLES, T. [1973.] Heuristics and justification in
scientific research: Comments on Shapere, pp.
571-589. In Suppe (ed.), 1977.

Nornris, K. S. 1958. The evolution and systematics
of the iguanid genus Uma and its relation to
the evolution of the other North American
desert reptiles. Bull. Amer. Mus. Nat. Hist.,
114: 247-326.

PAINTER, T. S. 1921. Studies in reptilian spermato-
genesis. I. The spermatogenesis of lizards. J.
Exptl. Zool., 34: 281-327.

PARKER, E. D., JR., AND R. K. SELANDER. 1976. The
organization of genetic diversity in the par-
thenogenetic lizard, Cnemidophorus tessel-
atus. Genetics, 84: 791-805.

PATTON, J. L., R. K. SELANDER, AND M. H. SmITH.
1972. Genic variation in hybridizing popula-
tions of gophers (genus Thomomys). Syst.
Zool., 21: 263-270.

PAULL, D., E. E. WILLIAMS, AND W. P. HALL. 1976.
Lizard karyotypes from the Galapagos Islands:
Chromosomes in phylogeny and evolution.
Breviora Mus. Comp. Zool. No. 441, 31 pp.

Advances in Herpetology and Evolutionary Biology

PETERS, R. H. 1976. Tautology in evolution and
ecology. Am. Nat., 110: 1-12.

PLATNICK, N. I., AND E. S. GAFFNEY. 1978. Evolu-
tionary biology: A Popperian perspective. Syst.
Zool., 27: 137-141.

PopPer, K. R. 1934. Logik der Forschung. Vienna.
Julius Springer Verlag.

___. 1970. Normal science and its dangers, pp. 51—
58. In Lakatos and Musgrave (eds.), 1970.
—___. 1972a. The Logic of Scientific Discovery,
Sixth Impression (Revised) [First Impression,
1959]. London, Hutchinson & Co., Ltd., 480

pp.

—___. 1972b. Conjectures and Refutations: The
Growth of Scientific Knowledge, 4th Ed.
(Revised) [lst Ed., 1963]. London, Routledge
and Kegan Paul, 431 pp.

—___.. 1972c. Objective Knowledge: An Evolution-
ary Approach. London, Oxford Univ. Press,
380 pp.

___. 1976. Unended Quest: An Intellectual Auto-
biography., Revised Ed. [first publ. 1974]. La
Salle, Illinois. Open Court Publ. Co., 255 pp.

PRESCH, W. 1969. Evolutionary osteology and rela-
tionships of the horned lizard genus Phryn-
osoma (family Iguanidae). Copeia, 1969: 250-
DS,

Rivas, L. R. 1964. A reinterpretation of the concepts
“sympatric” and “allopatric” with proposal of
the additional terms “syntopic” and “allo-
topic.” Syst. Zool., 13: 42-43.

SALMON, W. C. 1966, 1967. The Foundations of
Scientific Inference. Pittsburg, Univ. of Pitts-
burg Press, 157 pp.

SAVAGE, J. M. 1958. The iguanid lizard genera
Urosaurus and Uta with remarks on related
groups. Zoologica, Sci. Contrib. New York
Zool. Soc., 43: 41-54.

___. 1960. Evolution of a peninsular herpetofauna.
Syst. Zool., 9: 184-212.

___. 1966. The origins and history of the Central
American herpetofauna, Copeia, 1966:
719-766.

Simpson, G. G. 1944. Tempo and Mode in Evolu-
tion. New York, Columbia Univ. Press.

____. 1961. Principles of Animal Taxonomy. New
York. Columbia Univ. Press, 247 pp.

SMITH, H. M., AND E. H. Tayitor. 1950. An anno-
tated checklist and key to the reptiles of Mexi-
co, exclusive of the snakes. Bull. U.S. Nat.
Mus., 199: 1-253.

SmiTH, S. G. 1966. Natural hybridization of the
coccinellid genus Chilocorus. Chromosoma,
18: 380-406.

Spurway, H. 1953. Genetics of specific and sub-
specific differences in European newts. Symp.
Soc. Exptl. Biol., 7: 200-237.

Spurway, H., AND H. G. CALLAN. 1960. The vigor
and male sterility of hybrids between the
species Triturus vulgaris and T. helveticus. J.
Genet., 57: 84-118.

SupPE, F. [1973.] The search for philosophic under-

USE OF THE COMPARATIVE APPROACH - Hall

standing of scientific theories, pp. 1-241. In
Suppe (ed.), 1977.

___. 1977. Afterword, pp. 615-730. In Suppe (ed.),
1977.

___.. (ed.) 1977. The Structure of Scientific Theo-
ries, 2nd Ed. Urbana, Univ. of Illinois Press,
818 pp.

SzARSKI, H. 1971. The importance of deviation am-
plifying circuits for the understanding of the

course of evolution. Acta Biotheor., 20:
159-170.

SZYMURA, J. M. 1976a. Hybridization between dis-
coglossid toads Bombina bombina and

Bombina variegata in southem Poland as
revealed by the electrophoretic technique. Z.
f. Zool. Syst. Evol.-Forsch., 14: 227-236.

____. 1976b. New data on the hybrid zone between
Bombina bombina and Bombina variegata
(Anura, Discoglossidae). Bull. Acad. Polon.
Sci. Cl 2, 34: 355-363.

THAELER, C. S., JR. 1974. Four contacts between
ranges of different chromosome forms of the
Thomomys_ talpoides complex (Rodentia:
Geomyidae). Syst. Zool., 23: 343-354.

Topp, N. B. 1970. Karyotypic fissioning and canid
phylogeny. J. Theor. Biol., 26: 445-480.
WALLACE, B. 1959. Influence of genetic systems on
geographic distribution. Cold Spring Harbor

Symp. Quant. Biol., 24: 193-204.

WHITE, M. J. D. 1973. Animal Cytology and Evolu-
tion, 3rd. Ed. London, Cambridge University
Press., 961 pp.

__. 1978. Modes of Speciation. San Francisco,
W. H. Freeman and Co., 455 pp.

679

WHITE, M. J. D., R. E. BLACKITH, R. M. BLACKITH,
AND J. CHENEY. 1967. Cytogenetics of the
viatica group of morabine grasshoppers. I. The

. “coastal” species. Aust. J. Zool., 15: 263-302.

Waite, M. J. D., H. L. Carson, AND J. CHENEY.
1964. Chromosomal races in the Australian
grasshopper, Moraba viatica in a zone of
geographic overlap. Evolution, 19: 417-429.

Waite, M. J. D., K. H. L. Key, M. ANDRE, AND J.
CHENEY. 1969. Cytogenetics of the viatica
group of morabine grasshippers. II. Kangaroo
Island populations. Aust. J. Zool., 17: 313-328.

WILLIAMS, E. E. 1972. The origin of faunas. Evolu-
tion of lizard congeners in a complex island
fauna: A trial analysis. Evol. Biol., 6: 47-89.

WILSON, A. C., G. L. BusH, S. M. CAsE, AND M.-C.
KING. 1975. Social structuring of mammalian
populations and rate of chromosomal evolu-
tion. Proc. Nat. Acad. Sci. USA, 72: 5061-5065.

WRIGHT, S. 1941. On the probability of fixation of
reciprocal translocations. Am. Nat., 75:
513-522.

WoopkruFF, D. S. 1973. Natural hybridization and
hybrid zones. Syst. Zool., 22: 213-218.

ZASLEVSKY, V. A. 1963. Hybrid sterility as a limiting
factor in the distribution of allopatric species.
Dokl. Akad. Nauk SSSR, 149: 470-471.

ZEFF, E. W. 1962. A technique for delineation of
chromosomal constitution of reptilian leuco-
cytes grown in culture. Unpublished manu-
script, UCLA Dept. Zool., Los Angeles.

Immigration and the Dynamics of

Peripheral Populations

ROBERT D. HOLT!

ABSTRACT. The influence of immigration upon the
abundance and dynamics of peripheral populations
should reflect the degree of genetic differentiation
between residents and immigrants. The interplay of
immigration, local dynamics, and natural selection
in models of population growth is examined. Given
that no genetic variation exists upon which local
selection may act, and that a population is numeri-
cally stable, immigration should increase popula-
tion size. In populations with discrete-time dynam-
ics, immigration may either destroy or create local
stability. If a population is unstable, an increase in
the rate of immigration may reduce its average
density. The introduction of haploid genetic varia-
tion dramatically changes the character of the rela-
tion between immigration rate and density. If fit-
nesses are .density-dependent, in a stable, poly-
morphic population total density is independent of
the rate of immigration. Decreasing the fitness of
immigrants relative to residents may enhance the
stabilizing influence of immigration. It is briefly
argued that frequency-dependent fitnesses or di-
ploid genetic variation with density-dependent fit-
nesses can produce an inverse relation between the
rate of immigration and population size. These
theoretical results suggest that the consequences of
dispersal for population dynamics may be strongly
influenced by the degree of local genetic adaptation
that exists within a species.

INTRODUCTION

Patterns of dispersal are central to both
island and continental biogeography, for
the range ultimately occupied by a newly
formed species depends upon that spe-
cies’ success as a colonizer (Williams,
1969). The importance of dispersal in

! Museum of Natural History and the Department
of Systematics and Ecology, University of Kansas,
Lawrence, Kansas 66045, U.S.A.

other phenomena of ecology and evolu-
tion is not as well understood and has
been the subject of considerable debate.
Over the past two decades, evolutionists
and ecologists have moved in curiously
opposite directions in their views on the
significance of dispersal in natural pop-
ulations. In 1963, Ernst Mayr in Animal
Species and Evolution strongly argued
that “gene flow is the main factor respon-
sible for genetic cohesion among the
populations of a species ... [and] one of
the principal reasons for the slow rate of
evolution of common widespread spe-
cies.” This view of gene flow as a signifi-
cant homogenizing force was accepted
evolutionary doctrine for many years but
has recently come under increasing
attack. The empirical and _ theoretical
studies of Antonovics (1968), Endler
(1977), Levin and Kerster (1974) and
Slatkin (1978) (to note just a few promi-
nent landmarks) have demonstrated that
gene flow does not as a general rule ef-
fectively counter spatially varying selec-
tion. Many evolutionary biologists are
now persuaded that even though gene
flow can provide an important source of
variation upon which selection might act,
in most natural populations rates of gene
flow are usually not great enough to
prevent the evolution of local adaptation.

Until the early 1970’s_ theoretical
ecology consisted largely of glosses on
the elegant mathematical edifice erected
by Vito Volterra and Alfred Lotka fifty
years earlier. This body of theory, de-
signed to analyze the dynamics of spa-
tially homogeneous populations, neglec-

MODELS FOR PERIPHERAL POPULATIONS -: Holt

ted the possible significance of move-
ment patterns in population processes.
But in recent years the role of dispersal in
local dynamics and interspecific interac-
tions has become the focus of much
research activity by theoretical ecologists
(e.g., Levin, 1978; Okubo, 1980). Empiri-
cal evidence is steadily mounting that
movement patterns affect both the mean
density of populations and the pattern of
fluctuations around that mean (Gaines
and McClenaghan, 1980; Taylor and
Taylor, 1977; Tamarin, 1977). We are left
in the slightly awkward position of ap-
pearing to believe that dispersal is impor-
tant in local ecological dynamics, but not
in local evolutionary dynamics!

There is one circumstance for which all
seem to agree that dispersal can be an
important conservative evolutionary
force. Peripheral populations in an
ecotone often exist in low densities ad-
jacent to populations at much higher
densities. In such peripheral popula-
tions, natural selection favoring locally
adapted alleles can be readily swamped
by rates of immigration which are low in
absolute numbers of individuals per unit
time, but high in proportion to the peri-
pheral population’s low abundance
(Antonovics, 1976). Mayr (1963) sug-
gested that such regular disruption of
local adaptations by immigration should
be expected wherever a border popula-
tion exists at low densities, and he argued
that this process could produce evolu-
tionary stasis at a species’ border.

It is premature to judge the empirical
adequacy of Mayr’s model, as too little is
known about the causes and dynamics of
species borders. However, the scenario
is consistent with analytic studies of the
countervailing influence exerted by gene
flow upon local selection in low-density
populations (e.g., Nagylaki, 1978).
Hence, it is possible that gene flow
neither accounts for the genetic cohesion
of a species over its entire range, nor as a
rule prevents local adaptation, yet does
confer a kind of evolutionary stability
upon that species’ border.

In this paper I explore an ecological

681

question suggested by Mayr’s evolution-
ary hypothesis. What is the relationship
between the rate of immigration into a
population and that population’s size? I
will argue that the answer to this ques-
tion should depend upon the existence
and character of genetic variation in the
species. Given that residents and immi-
grants are genetically identical, it seems
intuitively reasonable to expect that as
the number of immigrants increases, so
should population size. Conversely, if
immigrants and residents differ geneti-
cally, the average immigrant should be
less fit in the local environment than is
the average resident. The more immi-
grants there are, the lower the mean fit-
ness of the population should be. We may
reasonably conjecture that a sufficient
reduction in mean fitness will decrease
the size of the population.

Using verbal arguments alone it is dif-
ficult to gauge the relative magnitude of
these two opposing consequences of im-
migration. I have therefore explored the
effect of immigration upon the dynamics
of peripheral populations in three classes
of population models. In order of increas-
ing complexity, these are continuous-
time and _ discrete-time ecological
models: 1) without genetics, 2) with
haploid genetic variation, and 3) with di-
ploid genetic variation. This paper dis-
cusses the first two classes of models. My
studies of the third class of models—by
far the most complicated—will be pres-
ented elsewhere, along with certain tech-
nical details of the models described
below.

PERIPHERAL POPULATION MODELS:
GENETICALLY HOMOGENEOUS
POPULATIONS

The two general growth models to be
discussed are

dN = F(N) +1

ai (1)

and N(t + 1) = G(N(t)) + I, or, equival-
ently,

682

AN = G7 (N@®) +1 (2)

where N is population density, F(N)
and G(N) are growth functions, G’ (N) =
G(N) — N, and I denotes immigration.
Throughout this paper the rate of immi-
gration is assumed to be temporally in-
variant, and the growth functions F and G
are assumed to be continuous. A contin-
uous-time differential equation such as
(1) is strictly appropriate as a growth
model only for populations with com-
pletely overlapping generations, neglig-
ible time lags and no variation in
survivorship or fecundity with age. The
quantity I is an instantaneous rate of im-
migration. The discrete-time difference
equation (2) best describes populations
in which age-classes or growth stages are
temporally segregated, so that only one
class is present at any time. If dispersal
occurs during a single stage in the life
history of the organism, and the popula-
tion is censused immediately following
each pulse of dispersal and immigration,
equation (2) is an appropriate model for
studying the influence of immigration
upon population dynamics. Both models
assume that there are no qualitative dif-
ferences, including genetic differences,
between residents and immigrants.

The three standard steps in the analy-
sis of growth models such as (1) and
(2) are as follows: 1) solve for point equil-
ibria, N*; 2) determine the local stability
character of each equilibrium; 3) for
model (2), attempt to understand the
often complex dynamical behavior that
may exist when populations do not have a
stable point equilibrium. For any bio-
logically reasonable model, in the ab-
sence of immigration there will be an
equilibrium N* of maximal size—the
carrying capacity, K—above which the
population has a negative growth rate.
The number of equilibria found at densi-
ties below K is determined by the speci-
fic details of particular models.

Advances in Herpetology and Evolutionary Biology

IMMIGRATION AND POPULATION
EQUILIBRIA

In model (1), equilibria are found by
solving F(N*) + I = 0. The local stability
of each equilibrium is determined by the
slope of the growth curve,

where the derivative is evaluated at N =
N*. If A < O, the equilibrium is locally
stable, and if \ > 0 it is unstable. A popu-
lation nudged away from an unstable
equilibrium will converge asymptotically
to a stable equilibrium. How does im-
migration shift the position of each equi-
librium? By applying the chain rule of
differentiation to (1) we have

dl (aN? d > GN ee

Hence, the slope A of the growth curve at
an equilibrium determines both the sta-
bility of the equilibrium population and
the way in which immigration shifts its
density. Immigration increases the dens-
ity of populations at locally stable equi-
libria and decreases the densities at
which unstable equilibria occur. The
quantity |A| measures the strength of
density-dependence in the population;
increasing |A| reduces the sensitivity of
population size to changes in immigra-
tion rate. Regardless of the detailed form
of the growth model, however, if a peri-
pheral population is at its carrying capac-
ity, 4 < O and immigration will always
increase that population’s size.

A similar relation between the rate of
immigration and equilibrial density
holds in the discrete-time growth model.
We find the equilibria by solving G’ (N*)
+ I = 0. By linearizing around N* in the
usual way it can be shown that the local
stability properties of each equilibrium
are set by the sign and magnitude of

MODELS FOR PERIPHERAL POPULATIONS - Holt

» —_ dG’
SSN

where the derivative is evaluated at N*.
In general, three classes of equilibria are
possible: 1) A’ > 0. A population devia-
ting by a small amount from N* will
monotonically diverge away from N*. 2)
—2 < X’' < 0. The equilibrium is locally
stable. 3) \’ < —2. The negative density-
dependence at the equilibrium is so
severe that the population repeatedly
overshoots the equilibrium then crashes
to low densities. The population ulti-
mately settles into sustained oscillations
around the locally unstable equilibrium;
the character of these fluctuations de-
pends upon the specific details of the
full, nonlinear model. (May, 1976).

Applying the chain rule of differentia-
tion again, we have

UJ

a is evaluated at N*. Hence, if

d’ > 0, immigration reduces the density
at which equilibrium occurs, and if )’
< 0, population size increases with immi-
gration.

A simple graphical illustration of this
conclusion is shown in Figure 1. The
solid line marks a growth curve G(N) of
an isolated peripheral population. There
are two stable nontrivial equilibria
(points 1 and 3 in the figure) and one
unstable equilibrium (point 2). In a com-
parable but less isolated population with
I immigrants entering per generation, the
growth curve is G(N) + I. Adding a con-
stant rate of immigration I to the growth
equation is geometrically equivalent to a
rigid, vertical translation of the growth
curve graph by I units. The dashed line
in the figure portrays the growth curve of
the less isolated population. It can be
seen that the influx of immigrants has
pushed the population well above its

where

683

carrying capacity and has caused the
paired equilibria at lower densities to
disappear entirely. As the rate of immi-
gration is increased, populations are less
likely to be trapped at such low-density
equilibria.

IMMIGRATION AND POPULATION
STABILITY IN DISCRETE-TIME GROWTH
MODELS

Given that a population is locally
stable, we have seen that immigration
should increase population size. But if
negative density-dependence is too se-
vere at high densities, populations may
exist in a permanent state of fluctuation
around an unstable point equilibrium.
This suggests the following two ques-
tions about the impact of immigration
upon a peripheral population. Can immi-
gration destabilize an otherwise stable
population, or, conversely, can immigra-
tion impose stability onto an unstable,
fluctuating population? Given that a

Ne

Figure 1. The effect of immigration upon equilibrium
density in a discrete-time model. Equilibria occur
where G (N), the curved line, crosses the straight, 45°
line. Adding a constant amount of immigration raises
the growth curve uniformly and equilibrium 3 in-
creases, whereas the other equilibria vanish.

D
(ea)
Hom

population is varying cyclically or cha-
otically, how does immigration modify
the distribution of densities shown by the
population over time?

As immigration increases, the slope of
the growth curve at an equilibrium
where \’ < 0 changes according to

=

d2G) /dN*
aI oe Ges)

dI

Given a growth curve that is concave
downward,

BGS Zi).
dN?

immigration will push the population
toward levels of increasingly severe den-
sity-dependence and may even destabil-
ize an intrinsically stable peripheral
population. Conversely, immigration
may stabilize a population whose growth
curve is concave upward—

d2€ 0)
dN2

Therefore the effect of immigration upon
population stability is determined by the
concavity of the growth curve G(N).
Standard discrete-time models provide
ready examples of both effects. One
model for which immigration is destabi-
lizing is the discrete logistic equation

N(t+1) = N(t) (1 + ta) +I (3)

where r is the intrinsic growth rate. The
stability properties of this model without
immigration are well-understood (May
and Oster, 1976; Roughgarden, 1979). In
the Appendix, I outline the stability char-
acter of the discrete logistic with immi-
gration. Figure 2 depicts the stability
domains of this model. (The dashed and
dotted lines are explained in the Appen-
dix.) The overall impression from this
figure is that immigration reduces or
eliminates stability in the peripheral
population.

Advances in Herpetology and Evolutionary Biology

By contrast, in other models immigra-
tion may stabilize an intrinsically un-
stable population. Figure 3 shows the re-
sults of a local stability analysis of the
following model:

N(t+1) = N(t) exp[x(1 — N(®/K)] +-L-(4)

With no immigration and r < 3, the dy-
namics of this model resemble those of
(3) May (1976). As immigration increases,
the domain of unstable behavior dimin-
ishes. The contrast between Figures 2
and 3 is striking. In a more detailed
analysis it can be shown that the oppos-
ing consequences of immigration for the
two models are due to the downward
concavity of (3) and the upward concavity
of (4) at high values of r and N.

Numerical studies of a number of dis-
crete-time models were carried out in
order to ascertain how immigration modi-
fies the temporal distribution of densities
in cyclic or chaotic populations. In such
populations, a reasonable measure of
abundance is the arithmetic time-average
of densities

Figure 2. Stability regions for the discrete logistic (3).
S = monotonic return to equilibrium. os = oscillatory
return to equilibrium. u = locally unstable (stable
cycles or chaos). The solid line is the largest value of I
allowing non-negative numbers for G(N). The dashed
line is the outer bound of the unstable region for the
modified logistic model discussed in the Appendix.

MODELS FOR PERIPHERAL POPULATIONS : Holt 685

Figure 3. Stability regions for the exponential logistic
(equation (6)). s = monotonic stability. o = oscillatory
stability. u = unstable. The central line in the o region
separates parameter choices where an increase in I
decreases the rate of return to an equilibrium (to the
left) from those parameters where I effects a faster
return (to the right).

In cyclic populations the appropriate
value for T is one cycle length, whereas
in a chaotic population a large number of
generations per run starting from a num-
ber of initial conditions may be required
to fully characterize <N>.

What is the relationship between aver-
age density and the rate of immigration?
An example of a pattern that emerged
repeatedly in the simulations is depicted
in Figure 4. The model used for this
figure has been extensively exploited by
insect ecologists (e.g., Hassell, 1975).
With an added immigration term the
growth model is

N(t + 1) = N(tet (1 + dN(t))~P + 1.(5)

The four curves in the figure correspond
to four values for the intrinsic growth
rate, r. At I=0 and high r, populations
obeying equation (5) fluctuate, some-
times greatly, around an unstable point
equilibrium. In these unstable popula-
tions, <N> does not increase mono-
tonically with I. The influx of a few immi-
grants per generation may dramatically
increase <N>, and a yet greater rate of
immigration may actually decrease <N>.

This nonmonotonic relation between
<N> and I has a simple explanation. In

discrete-time growth models such as (5),
populations with high intrinsic growth
rates tend to exhibit chaotic behavior
(May, 1976). Time-series of populations
in chaos typically show overshoots of K
followed by precipitous declines in
abundance. Following each population
crash, several generations may elapse
before population numbers are suffi-
ciently large to produce a high total
growth rate, culminating in another
explosive overshoot and crash. Even a
slow trickle of immigrants can greatly
reduce the number of generations
between successive overshoots. In model
(5), population growth is essentially
geometric at densities well below K. If
N(O) is the number in the population
immediately following a crash, the
population size t generations later is
approximately

<N>

Figure 4. Time-averaged densities for model (7). The
numbers labelling the curves are the intrinsic growth
rates. The other parameters are d = 1, and b = r. For
populations exhibiting chaotic behavior, the lines
represent the average over many runs of 5,000 gener-
ations length, starting over a range of initial densities.

686 Advances in Herpetology and Evolutionary Biology

N(t) = N(O)ett+ I

given that numbers remain low enough
for negative density-dependence to be
negligible. From this expression it is
clear that the time required to reach a
given N (e.g., an N such that G(N) > K)
on N(O) may be substantially shortened
by regular immigration. This reduction in
the amount of time needed for recovery
from a population crash increases the
number of generations at high densities,
and therefore increases <N>. At higher
rates of immigration, density-depend-
ence among the immigrants themselves
tends to diminish the magnitude of the
overshoot. In model (5), this diminution
leads to a decrease in <N>.

We can summarize the above results as
follows: 1) Given no genetic differentia-
tion between immigrants and residents,
immigration should increase the density
of stable peripheral populations. 2) The
concavity of the growth curve determines
the relation between immigration rate
and stability. And finally, 3) in fluctuat-
ing peripheral populations, an increase
in the rate of immigration may decrease
the average size of the population.

PERIPHERAL POPULATION MODELS:
THE EFFECT OF HAPLOID GENETIC
VARIATION

The conjecture to be examined is that
genetic differentiation between immi-
grants and residents alters the functional
relationship between the rate of immigra-
tion and population size. In the re-
mainder of this paper I explore the
properties of models which incorporate a
particularly simple kind of genetics—
haploid variation with two alternative
alleles at a single locus. One allele—type
l—is fixed within the species’ main
range, and so all immigrants are type 1,
whereas the other allele—type 2—is
selectively favored in the peripheral
population. To explore how the balance

between selection and immigration af-
fects both the gene frequency and popu-
lation size, we modify models (1) and (2)
as follows.

Counting gene numbers, the appro-
priate generalization of model (1) is

dN,
a = N, 9, (Ny, No) + I
(6)
dN
aa = No 05 (Nj, No)

where N, is the density of type i (i=1,2),@
is the per capita instantaneous growth
rate or the absolute fitness of type i, and I
is the immigration rate of type 1. The total
pope size is defined as N = N, +

, the gene frequency of allele 1 is p =
N “IN, and the gene frequency of allele 2
i @g = Il = p. The mean fitness of
the population is @ = p@, + q@,. An alter-
native representation of system (6) is ob-
tained by differentiating N and p with
respect to t

dN na

4 (6’)
p

APR pq(M, — @,) + at

In like manner, we may embody haploid
genetic variation within the framework of
the discrete-time model (2) as follows:

Ni + Dy Ni@en(NeeNe) re re)
N,(t + 1) = N,(t)g,(N,, N,)
or, equivalently,

AN = N(g — 1) + 1

pa(e, — go) ae qn

where g, is the per-capita growth rate or
absolute fitness of type i and g = pg, +

MODELS FOR PERIPHERAL POPULATIONS : Holt

qg, is the mean fitness of the population.
The analysis of these two models pro-
ceeds through the usual steps of first
solving for equilibria (N*, p*), determin-
ing the local stability of each equilib-
rium, and then examining the behavior of
the model away from equilibria. Models
(6)’ and (7)’ explicitly display the genetic
character implicit in models (6) and (7).

DENSITY-DEPENDENT SELECTION

It is difficult to make much headway
without specifying in more detail the fit-
ness functions @ and g. A selective
regime that has received a great deal of
attention from population biologists
is density-dependent selection, for which
Q@ and g, may be written as functions of
total density N, @(N) and g(N) (Rough-
garden 1971, 1976; Charlesworth 1971,
1980; Asmussen, 1979). A significant
finding of these theoretical studies is that
if fitnesses are strictly density-depend-
ent, and the population is stable, natural
selection adjusts gene frequencies in
such a way that population density is
locally maximized at the joint demo-
graphic and genetic equilibrium.

First consider the continuous-time
model (6). To simplify the analysis, I as-
sume that each allele has a carrying ca-
pacity K, such that @(K,) = 0, and that
there is negative density-dependence at
all densities:

aN, <0.

In the absence of immigration, it is
straightforward to show that the allele
with higher K excludes the alternate
allele. Since type 2 is assumed to be lo-
cally favored, K, > K, and the population
equilibrates at K,.

As the rate of immigration increases, so
does the frequency of type 1. However,
given that the population is polymorphic
for both alleles, population size is inde-
pendent of the rate of immigration. For
the population to be at its equilibrium

687

(N*, p*) both growth rates in (6)
must equal 0, so N%*@,(N*) = 0. As the
population is assumed to be _ poly-
morphic, N*¥ > 0. Thus 0,(N*) = 0. But by
assumption, @, = 0 only when N* = K,.
Therefore, the total population size re-
mains at K, as the rate of immigration
increases. Immigration does not change
the total population size, but instead
shifts the relative proportions of the two
types. From (6’) we find that

29 (Kg)

(since K, > K,, 0(K,) < 0). Hence, a poly-
morphic equilibrium exists only if I <
|K,,(K,)|. In the Appendix it is shown
that the equilibrium

jo =

=

N*, p*) = (Ko,
is both locally and globally stable.
Thus we have completely char-

acterized the interplay of immigration
and selection in a general, haploid model
of density-dependent selection. As long
as I < |K,@(K,)|, immigration merely
shifts the genetic composition of the
peripheral population without changing
its total density at all. The opposing
ecological and genetic effects of immi-
gration discussed in the introduction
exactly cancel each other out.

By a parallel argument it can be shown
that the comparative statistics of the dis-
crete-generation model (7) are identical
to those of the continuous-generation
model. Given fitnesses that are density-
dependent, equilibrial densities are in-
dependent of the rate of immigration in
polymorphic populations. As with the
purely ecological models discussed
earlier, the principal difference between
models (6) and (7) is that the latter may
exhibit sustained oscillatory behavior.
The point equilibrium of (7) is locally
stable only if the magnitude of the real
part of the dominant eigenvalue ) is less
than unity. Let

688

Then the two eigenvalues are
A= [aj] = ag99 ar ((aqq = ago)” ote 4ay9a91)) 7/2.

As both a,, and a,, are negative, the dis-
criminant is positive and both eigen-
values are real numbers. We wish to
understand how immigration alters the
stability properties of a peripheral pop-
ulation, and how immigration perturbs
average abundance in unstable popula-
tions. Substituting explicit fitmess func-
tions such as (3) or (4) leads to cumber-
some masses of algebra from which it is
difficult to extract necessary and suffi-
cient conditions for local stability.
However, it is plausible that if immigra-
tion reduces the frequency of a locally
favored allele, the peripheral population
could be made more stable. In discrete-
time population models, the intrinsic
growth rate r typically “tunes” the
dynamic behavior of the model—the
magnitude and period of oscillations
usually increase with r. As immigrants
are assumed to be less fit than residents,
the values of r, or K, or both, for the
immigrant should be lower. Diluting a
population at high r with an admixture of
immigrants with lower r should reduce
the average r of the population as a
whole, and therefore tend to stabilize an
unstable peripheral population.

In like manner, a lower tolerance of
crowding in immigrants (K, < K,) may
enhance the stabilizing fad oo mr
gration. As a particular example, consider
a population with discrete generations in
which the fitnesses of each type are
described by g,(N) = exp[r(1 — N/K,)].
The two alleles share the same r but dif-
fer in K. If the rate of immigration is too
great, the polymorphism will not persist.
As depicted in Figure 5, the maximum I
consonant with polymorphism varies
with both r and K,. The local stability

Advances in Herpetology and Evolutionary Biology

Figure 5. Maximum rate of immigration I permitting
polymorphism in model (9) with equation (6) as a fit-
ness function. K, = 10. r, = r,. K, is denoted by the
number marking each curve.

Figure 6. Stability regions of model (9) with fitnesses
given by (6). s = monotonically stable. os = damped
oscillations. u = unstable. The solid lines mark the
edges of stability regions for K, = 1, and the dashed
lines for K, = 5. The dotted lines are from Figure 3 and
depict the stability character of a population in which
the immigrants are type 2.

properties of this model are displayed in
Figure 6. We have already observed that
immigration is weakly stabilizing in this
model when immigrants are of the same
genetic type as the residents; the dotted
lines, lifted from Figure 4, demarcate
transitions from monotonic to oscillatory
stability (left line) and from locally stable
to unstable point equilibria (right line).
The solid lines demarcate the edges of

MODELS FOR PERIPHERAL POPULATIONS : Holt

the corresponding stability regions for
immigrants with K, = 1, and in like man-
ner the dashed lines for K, = 5. Decreas-
ing the carrying capacity of the immi-
grants heightens the stabilizing influence
of immigration.

In discrete-time haploid models such
as (6), as in the purely ecological model
(2) discussed above, instability may
complicate the relation between immi-
gration and population size. In numerical
studies I have found that polymorphisms
are difficult to maintain if the demo-
graphic attributes of each type lead to
severe oscillations in monotypic popula-
tions. In a stable population, a low im-
migration rate produces a low frequency
of the immigrant type. The same rate of
immigration can lead to a much higher
gene frequency and even fixation in an
unstable population, for the simple
reason that following a population crash a
few immigrants may comprise a sizeable
fraction of the total population. During
these crashes the locally adapted type
can go extinct because of genetic drift,
while immigration steadily replenishes
the less fit type. Gene flow should
hamper local adaptation more readily in
unstable than in stable peripheral popu-
lations.

If the polymorphism does persist, over
the course of population fluctuations
high densities are correlated with low
frequencies of the immigrant type. In the
particular model used for Figures 5 and
6, the time-average of density, <N>,
seems to be independent of I, but in
other haploid models <N> may decrease
with increasing I.

From this analysis of density-
dependent selection in haploid popula-
tions, we may conclude that 1) the rate of
immigration does not affect the total
density of stable, polymorphic popula-
tions; 2) decreasing the fitness of immi-
grants relative to residents strengthens
the stabilizing effect of immigration upon
population dynamics; and 3) the main-

689

tenance of local adaptation in the face of
gene flow is unlikely in severely fluctu-
ating populations.

OTHER GENETIC MODELS

Two obvious generalizations of this
model are to employ more general fitness
functions in (5) and (6) and to develop
comparable diploid models. Space limi-
tations preclude a full treatment of these
extensions here, so I will merely outline
a few salient changes in the results
produced by these modifications in the
models. If the absolute fitness functions
in models (5) or (6) are functions of the
separate densities of the two types rather
than of their summed density, the selec-
tive regime is a mixture of density-
dependent and _ frequency-dependent
selection. Alternative stable states may
exist (Fenchel, 1975). For the con-
tinuous-time model (6), if the population
is at a stable equilibrium, it can be shown
that

_ 9Og/ON j

dNn* _ 4N;* ( as
405/0N 5

lec )

where the partial derivatives are eval-
uated at (N,*, N,*). Since immigration in-
creases the density of N,, the overall
effect of immigration upon the total
population hinges upon the relative
magnitudes of the derivatives 00,/dN, and
00,/8N,. If absolute fitnesses are deter-
mined solely by the total density of the
population,

and N* is independent of I. Immigrants
that exert a disproportionate effect on the
fitness of residents, compared with the
effect of residents upon themselves,

690

aN,

aN’

decrease the total size of the population.

Even without immigration the diploid
equivalents of (6)' and (7)' can manifest a
rich panoply of behaviors. Asmussen
(1979) has analyzed a discrete-generation
model of density-dependent selection in
which (4) served as a fitness function.
She demonstrated the existence of regu-
lar and chaotic cycling in both population
size and gene frequency. In contrast to
classical selection models, heterozygote
superiority in crowding tolerance is not
required for the maintenance of genetic
variation in these cyclic populations. All
these features are retained in the models
with immigration. An additional feature
of the diploid models is that the relation-
ship between immigration and popula-
tion size depends upon the dominance
relations of alternative alleles in the
peripheral population. For instance,
consider an allele that is simultaneously
genetically dominant yet locally unfit. An
increase in the rate of immigration of in-
dividuals bearing that allele has three
distinct effects, two of which were dis-
cussed above. First, there is the ecologi-
cal effect of increasing density by in-
creasing the influx of individuals into the
population; this corresponds to a purely
ecological model such as (1). Second, if
fitnesses are density-dependent, this
increase in density lowers the fitness of
the locally favored genotype, the num-
bers of which will decline until the
overall density of the population is un-
changed; this accounts for the uncou-
pling of immigration rates and equilibrial
densities in the haploid models (6) and
(7). Third, since in a Mendelian popula-
tion the average fitness of the offspring
produced by a genotype is partially
determined by the array of genotypes
available for mating, the more immi-
grants there are, the more the fitness of
the resident’s offspring will be diluted by
cross-matings. The number of residents

Advances in Herpetology and Evolutionary Biology

declines even further to compensate for
this mating effect. The final pattern is an
inverse relation between immigration
rate and population density. By contrast,
if the immigrant allele is recessive the
mating effect is greatly diminished, and,
just as in the haploid model, population
size may be essentially independent of
the rate of immigration.

A third theoretical approach, and one
that ultimately might be the most profit-
able, is to extend quantitative genetic
models of phenotypic evolution (Lande,
1976) in order to examine how immigra-
tion by individuals with nonoptimal
phenotypes in one or more character
states influences population size in den-
sity-regulated populations. Antonovics
(1976) and Slatkin (1978) suggest some
promising lines of development for such
a theory.

DISCUSSION AND CONCLUSIONS

I have argued that two characteristics
of a peripheral population should mold
the functional relation between immigra-
tion rate and population size—its dynam-
ic stability, and the degree of its genetic
adaptation to the local environment. To
begin understanding the dynamics of a
peripheral population we must first iso-
late those factors responsible for its low
population density. It is widely believed
that density-independent factors act
more severely in peripheral than in cen-
tral populations. For instance, Mayr
(1963) asserts that “the border region is a
place in the area of a species where den-
sity-dependent factors are of minor im-
portance.” Alternatively, the peripheral
population may be rare because of an in-
crease in the intensity of density de-
pendence. In populations with delayed
density-dependence, these two explana-
tions for rarity at a species’ border have
fundamentally different implications. If
rarity results from a high rate of density-
independent mortality the intrinsic
growth rate of the population will be low

MODELS FOR PERIPHERAL POPULATIONS - Holt

and the population will stably persist at
its carrying capacity, at least in temporal-
ly constant environments. By contrast, if r
is high but K is low, the population may
exhibit fluctuations of considerable
magnitude around K, sometimes to the
point of extinction. As the demographic
and genetic consequences of immigra-
tion depend upon the stability character
of the peripheral population, the first
step in gauging the role of immigration
should be to try to understand the causal
basis for population fluctuations at the
species border.

Immigration may not change, or may
even decrease, the density of a popula-
tion that has adapted to its local environ-
ment. Scant data exists for testing this
idea, in part because the requisite
evidence is technically difficult to obtain.
As noted above, there recently has been a
de-emphasis of gene flow as a constraint
on the evolution of local adaptation. To
the extent that this view is valid, we
should expect the predicted phenome-
non to be rarely observed. An additional
reason for the rarity of relevant data,
however, is that many ecologists simply
assume that organisms are well adapted
to their environment. Although usually
reasonable, in peripheral or ecologically
marginal populations this assumption
may well be false. I know of two possible
examples of local maladaptation in
peripheral populations resulting from
gene flow. Camin and Ehrlich (1958)
argued that on islands in Lake Erie a
balance between migration and selection
maintained a polymorphism in banding
patterns of water snakes (Natrix sipedon).
A substantial fraction of the island popu-
lations had locally unfit banding patterns.
Stearns and Sage (1980) have suggested
that the life-history traits of one fresh-
water population of the mosquito-fish
(Gambusia affinis) might be best inter-
preted as maladaptation to the fresh-
water environment. Gene flow from a
nearby brackish-water population may
account for the apparent lack of local
adaptation in this population. In neither

691

instance is anything known about the ef-
fect of the apparent maladaptation upon
population density. Comparable situa-
tions might well exist in many organisms
with passive dispersal or territoriality
residing in locally heterogeneous en-
vironments. For instance, were the
habitat of a species of Anolis a mosaic of
high and low quality patches, territorial
behavior could lead to a steady flow of
individuals into low density patches.
This influx could reduce the efficacy of
selection for better adaptation to the low-
quality patch type.

A critical and very difficult empirical
problem is to gauge the relative magni-
tudes of spatial scales associated with
dispersal, selection, and nonselective
determinants of density. Slatkin (1973)
has suggested that a useful quantity for
understanding evolution in _ spatially
structured populations is the ratio be-
tween the average dispersal distance and
a measure of the gradient in selection. If
this ratio is small, a species should close-
ly track spatial variation in selection
(Roughgarden, 1979); since immigrants
into a population are drawn primarily
from nearby populations, and nearby
populations experience similar selective
regimes, immigrants should genetically
resemble residents. In this case the pure-
ly ecological models discussed above are
appropriate. By contrast, if the ratio is
large, many immigrants will be from rela-
tively distant populations with different
selection pressures, and immigrants may
differ greatly from the locally favored
type. Understanding the role of immigra-
tion in population dynamics in this case
requires a blending of ecology with
genetics.

For the rate of immigration to be large
enough, relative to carrying capacity, to
qualitatively perturb the abundance or
dynamics of a peripheral population,
there usually must be a second, more
abundant population in _ reasonable
proximity to the first—a sharp spatial
gradient in densities should exist. This
density gradient need not correspond to a

692

comparable selection gradient, since
many factors that affect density do so
indirectly. For instance, a consumer such
as a detritivore may be able to alter the
abundance, R, but not the rate of renewal
of its resource. Individual consumers are
food—limited, and there is no direct inter-
ference among them, we can write the
per capita consumer growth rate as some
increasing function of resource abun-
dance, Y(R). The consumer population
should grow until its collective demand
has reduced resource availability to some
level, C, at which each consumer just
replaces itself with a single descendent.
If the parameters determining the form of
Y and the quantity C are spatially invari-
ant, selection will act uniformly on these
parameters throughout the consumer’s
range. However, the number of con-
sumers present at equilibrium is deter-
mined indirectly by the renewal rate of
the resource, which might well vary
greatly over space. Hence, a consumer's
density may be highly variable in space
even though selection is not; dispersal
could then be of more consequence for
the ecology than for the evolution of the
species. In other circumstances, of
course, density gradients may closely
match selection gradients.

The degree to which ecological phe-
nomena are influenced by intraspecific
genetic variation is a problem of great
current interest. I have here argued that
the ecological consequences of dispersal
into a population depend upon whether
or not immigrants differ genetically from
residents. In the study of species’
borders, in particular, it should be real-
ized that ecological and genetical phe-
nomena may be inextricably entangled.

ACKNOWLEDGMENTS

Ermest E. Williams has long empha-
sized, both in his own work and in his
guidance of graduate students, the poten-
tial value of species’ borders as loci for
the study of significant ecological and

Advances in Herpetology and Evolutionary Biology

evolutionary problems. My own work on
models of peripheral populations, some
of which is presented in this paper, re-
ceived its initial stimulation from conver-
sations we had on the question of the
ecological and evolutionary stability” of
species borders.

I would like to express my gratitude to
Ernest and to William Bossert for patient-
ly listening to my ruminations about
species borders while I was a graduate
student at Harvard. I thank Steve
Hubbell, Phil Hedrick, Jim Hamrick,
Mike Gaines, Norman Slade, and
Gregory Glass for comments on the work
reported here, and Russell Lande and
Juli Armstrong for valuable comments on
the manuscript. My graduate assistant
Juli Armstrong kindly assisted with the
numerical simulations, and the Univer-
sity of Kansas supplied the requisite com-
puter funds. I thank Jan Elder, Coletta
Spencer, and my wife Lynne for assisting
with manuscript preparation.

APPENDIX

THE DISCRETE LOGISTIC MODEL

Without immigration, the single non-
trivial equilibrium is N* = K, and the
stability-setting eigenvalue is \ = 1—r. N
converges asymptotically to K if r < 2.
Stable limit cycles occur when 2 < r <
2.828, whereas for 2.828 < r <= 3 the
population exists in a chaotic regime in
which the long-term behavior of any par-
ticular population depends upon its ini-
tial conditions; the behavior may be
either periodic or aperiodic (May and
Oster, 1976). Given r > 3, the model can
predict negative densities; to preclude
this biological absurdity we require that r
<= 3. In like manner, as I increases we
would like to ensure that population
production G(N) is never negative. For
the discrete logistic, a population
minimum occurs during the generation
immediately following a maximum. The
maximum value of N(t + 1) occurs at the
critical point (N(t) = (K(1 + r)/2 of (8),

MODELS FOR PERIPHERAL POPULATIONS : Holt 693
K(1 + d d
ey SP re we 01, SL ee
4r dN dN
° > fi
eqiaue Gul (ilacass IN) = Us wae ine ae are all negative at N = K,, a < 0. This

the ratio I/N is bound by the following
expression:

lg JU ee

<< Oe

This bound on I/K is the dotted line of
Figure 2. We can avoid the annoying pos-
sibility of negative densities by simply
setting N(t + 1) = I whenever G(N) is a
negative quantity, which occurs when-
ever

N(t) > K (++).

For this modified discrete logistic, the
dashed line in the Figure is an upper
bound on the parameter values that allow
fluctuating densities.

Let Y=(1+4I/rK)!?. Then N*=
K(1 + Y)/2, and the eigenvalue is
N= hia G

As N* increases with increasing I, \ de-
creases. The maximum r consistent with
local stability is

= 2((2/K2 + 1)!2 — I/K).

THE STABILITY OF THE
CONTINUOUS-TIME HAPLOID MODEL (6)

The two eigenvalues are

h = [a + (a2 — 40)No oo 2/2,
where

d
a=) +N{ oI +N ie

Since the quantities

implies that Re(A) < 0. Monsovex straight-
forward manipulations show that the
separatrix

om do

a 40)No9 IN

is positive, so (N*, p*) is a stable node.
The per capita growth rates of type 1 and
2 were assumed to be negatively density-
dependent at all densities. Applying
Theorem (2) of Hastings (1978) we find
that the equilibrium is globally stable.
The critical assumption required for this
result is that the fitness of both types
depends only upon the total density of
the population.

LITERATURE CITED

ANTONOVICS, J. 1968. Evolution in closely adjacent
plant populations. VI. Manifold effects of gene
flow. Heredity, 23: 507-524.

____. 1976. The nature of limits to natural selection.
Ann. Missouri Bot. Gard., 63: 224-247.

ASMUSSEN, M. A. 1979. Regular and chaotic cycling
in models of ecological genetics. Theor. Pop.
Biol., 16: 172-190.

CAMIN, J. H., AND P. R. EHRLICH. 1958. Natural
selection in water snakes (Natrix sipedon L.)
on islands in Lake Erie. Evolution, 12:
504-511.

CHARLESWORTH, B. 1971. Selection in density-reg-
ulated poplulations. Ecology, 52: 469-474.

____. 1980. Evolution in age-structured populations.
Cambridge, Cambridge University Press.

ENDLER, J. A. 1977. Geographic variation, specia-
tion, and clines. Princeton, Princeton Univ.
Press.

FENCHEL, T. 1975. Factors determining the dis-
tribution patterns of mud snails (Hydrobiidae).
Oecologia, 20: 1-17.

GAINES, M. S., AND L. R. MCCLENAGHAN, JR. 1980.
Dispersal in small mammals. Ann. Rev. Ecol.
Syst., 11: 163-195.

HASSELL, M. P. 1975. Density dependence in
single-species populations. J. Anim. Ecol., 44:
283-295.

HastTIncs, A. 1978. Global stability of two-species
systems. J. Math. Biol., 5: 399-403.

LANDE, R. 1976. Natural selection and random >

694

genetic drift in phenotypic evolution. Evolu-
tion, 30: 314-334.

LEVIN, D. A., AND H. W. KERSTER. 1974. Gene flow
in seed plants. Evol. Biol., 7: 139-220.

LEVIN, S. A. 1978. Population models and commun-
ity structure in heterogeneous environments,
pp. 439-476. In S. A. Levin (ed.), Studies in
Mathematical Biology Vol. 2: Populations and
Communities. Washington: Math. Assoc.
Amer.

May, R. M. 1976. Simple mathematical models with
very complicated dynamics. Nature, 261: 459-
467.

May, R. M. AND G. F. OsTER. 1976. Bifurcations and
dynamic complexity in simple ecological
models. Amer. Nat., 110: 573-599.

Mayr, E. 1963. Animal species and evolution.
Cambridge, Harvard Univ. Press.

NaGyLakI, T. 1978. Clines with asymmetric migra-
tion. Genetics, 88: 813-827.

Oxuso, A. 1980. Diffusion and ecological problems:
mathematical models. Berlin, Springer-Verlag.

ROUGHGARDEN, J. 1971. Density-dependent natural
selection. Ecology, 52: 453-468.

____. 1976. Resource partitioning among competing
species—A coevolutionary approach. Theor.
Pop. Biol., 9: 388-424.

Advances in Herpetology and Evolutionary Biology

____. 1979. Theory of population genetics and evo-
lutionary ecology: An introduction. New York,
MacMillan.

SLATKIN, M. 1973. Gene flow and selection in a
cline. Genetics, 75: 733-756.

___. 1978. Spatial patterns in the distributions of
polygenic characters. J. Theor. Biol., 70: 213—
228.

SPIETH, P. T. 1979. Environmental heterogeneity: A
problem of contradictory selection pressures,
gene flow and local polymorphism. Amer. Nat.,
113: 247-260.

STEARNS, S. C., AND R. D. SAGE. 1980. Maladapta-
tion in a marginal population of the mosquito
fish, Gambusia affinis. Evolution, 34: 65-75.

TAMARIN, R. H. 1977. Dispersal in island and main-
land voles. Ecology, 58: 1044-1054.

TayLor, L. R., AND R. A. J. TAYLOR. 1977. Aggrega-
tion, migration, and population mechanics.
Nature, 265: 415421.

WILLIAMS, E. E. 1969. The ecology of colonization
as seen in the zoogeography of anoline lizards
on small islands. Quart. Rev. Biol., 44: 345—
389.

Distribution and Conservation of the
Black Caiman (Melanosuchus niger)

MARK J. PLOTKIN’
FEDERICO MEDEM?2
RUSSELL A. MITTERMEIER?
ISABEL D. CONSTABLE*

ABSTRACT. The black caiman (Melanosuchus
niger) once ranged throughout much of Amazonia,
but its populations have decreased sharply due to
hunting for the skin trade and for food, incompati-
bility with man and his domestic animals, and habi-
tat destruction. Viable populations of black caiman
apparently still exist in Ecuador, French Guiana,
and Peru, but only in Peru do they occur within
protected areas. Establishment of reserves for pro-
tection of the Ecuador and the French Guiana pop-
ulations is recommended as well as better enforce-
ment of existing legislation in all of the countries in
which the black caiman occurs.

INTRODUCTION

Most of the world’s living crocodilian
species are rapidly disappearing, mainly
as a result of hunting for the skin trade,
hunting for food, incompatibility with
man and his domestic animals, and, in
some areas, habitat destruction or mod-
ification. Five crocodilian species are
found in the Amazonian region of north-
erm South America (including the
Amazon and Orinoco drainages and the

1 Harvard Botanical Museum, Cambridge, Massa-
chusetts 02138, U.S.A.

2Universidad Nacional de Colombia, Bogota,
Colombia.

3 World Wildlife Fund, 1601 Connecticut Avenue,
NW, Washington, D.C. 20009, U.S.A.

4Brown University, Providence, Rhode Island
02912, U.S.A.

Guianas). The three smaller species
(Caiman crocodilus, Paleosuchus trigo-
natus, Paleosuchus palpebrosus), are still
relatively abundant over much of their
range and are probably among the
world’s least endangered crocodilians. In
striking contrast, the two larger species,
Crocodylus intermedius and Melano-
suchus niger, are both considered en-
dangered and are already extinct over
much of their former range. Here we re-
view the conservation status of the black
caiman, Melanosuchus niger and discuss
several key areas important for its con-
tinued survival.

PAST DISTRIBUTION

The black caiman was once wide-
spread in Amazonia. It has been recorded
from Brazil, Bolivia, Peru, Colombia,
Ecuador, Guyana, French Guiana, and
even Paraguay. In fact, the only Amazon-
ian countries from which there are no
authenticated records are Surinam and
Venezuela. The historical distribution of
the species, based on museum localities,
literature records, and sightings by the
authors or other reliable researchers, is
shown in Figure 1.

Within this vast area, the black caiman

696

S
eo i @

wie
S
KO

Figure 1. Historical distribution of Melanosuchus
niger. & = populations currently considered to be
viable.

appears to prefer quiet waters to the cur-
rents of the larger rivers. In large rivers
like the Amazon it is found in quieter
water around bends and especially in the
lakes and lagoons beside the mainstream.
It is also fond of the creeks and swamps
associated with the Rio Branco and the
Amazon delta islands of Marajo,
Mexiana, and Caviana. In the dry months
of November and December in lower
Amazonia, large individuals sometimes
estivate until the beginning of the wet
season in January.

The black caiman was once incredibly
abundant in many areas. Bates (1863) re-
corded their high density in the upper
Rio Solimoes, and enormous numbers
were also to be found around the delta
islands of Mexiana, Marajo, and
Caviana (Wallace, 1853; Goeldi, 1898;
Hagmann, 1902).

Advances in Herpetology and Evolutionary Biology

Today, however, the black caiman is
either extinct or very rare in areas where
it once was common. It is no longer pos-
sible to see them along the mainstream of
the Rio Solimoes where Bates was so
impressed by their numbers, and they
have all but disappeared from the delta
islands. They are now rare or absent in
most other parts of Amazonia, and are
regularly encountered only in the more
remote lake and tributary rivers—and
never in numbers approaching those of
the past century.

REASONS FOR DECLINE

The main reasons for the decline of the
black caiman are large scale hide-hunting
and incompatibility with man and his
domestic animals, with hunting as a food
source playing a role in some areas. Hab-
itat destruction does not appear to have
been a major factor in the past, although
it could well become one in the near fu-
ture.

Hide-hunting was responsible for tre-
mendous slaughter during this century,
and the black caiman was long the spe-
cies of choice among the caimans be-
cause of its size and abundance. Com-
mercial hide-hunting apparently began
in the late 1930’s and 1940's, and by the
1950’s enormous numbers of caiman
skins were being exported from Ama-
zonia. For instance, Fittkau (1973) esti-
mated that up to 5,000,000 caiman skins
were removed from Brazilian Amazonia
each year in the early 1950’s, and that one
or two additional animals were killed for
each of the officially reported skins.
Between 1960 and 1970, the total number
of skins was still between 200,000 and
700,000 per year. However, many of
these skins were of the smaller Caiman
crocodilus which became increasingly
important as Melanosuchus niger popula-
tions declined in Brazil. Export figures
vary somewhat (e.g., Melo Carvalho,
1967; Medem, 1972), but it is clear that

many millions of black caimans were

CONSERVATION OF THE BLACK CAIMAN - Plotkin et al.

killed for their hides
decades of this century.

Incompatibility with man and _ his
domestic animals was another problem.
Though not nearly as dangerous as some
of the larger Asian, African, and South
American crocodiles (e.g., C. porosus, C.
niloticus, C. intermedius), the black
caiman occasionally attacks man (Bates,
1863; P. Soini, personal communication).
As a result of their potential danger to
man, black caiman have often been wip-
ed out in the vicinity of human settle-
ments—even if their hides aren’t neces-
sarily used.

Black caiman were considered a major
threat to cattle ranching activities, es-
pecially in the seasonally flooded campos
or grasslands like those on Marajo or
Mexiana. Ranchers once held great
caiman drives in the dry season to exter-
minate as many of the crocodilians as
possible. Hagmann (1902) reported a
drive in which about 800 animals be-
tween | and 4.2 m were killed in two
days. Similar drives were also held in
other parts of Amazonia, the most recent
we know of in the 1940’s in the Brazilian
state of Amapa (R. Renau Ferrer,
personal communication).

On top of this, the black caiman was
and sometimes still is hunted for its meat
or fat in parts of its range (Wallace, 1853).
Black caiman meat, especially that of the
tail, is still eaten in some parts of the
animal’s range (e.g., the Kaw region of
French Guiana, the Rio Tapiche in
Peru, and several tributaries south of
Belem, Para).

in the middle

CURRENT STATUS

The tremendous slaughter of black
caimans over the past century resulted in
a steep population decline, and this spe-
cies is now either extinct or rare in most
of its former range. Only a few areas ap-
parently harbor substantial populations,
although none of these approach those
observed by the nineteenth-century na-
turalists.

697

The black caiman is officially protected
in most of countries in which it occurs
and appears in the Red Data Book as an
endangered species and on Appendix 1
of the Convention on International Trade
in Endangered Species, and the U.S.
Endangered Species List. The hide trade
has clearly diminished from the levels of
the 1950’s, but this is probably as much
due to large scale decline of populations
as to protective legislation. Although ex-
port is now illegal, black caimans con-
tinue to be killed, and enforcement in the
remote areas where they still occur is vir-
tually nonexistent. Although the vastness
of the species range will probably pre-
vent it from disappearing completely
before the turn of the century, it is cer-
tainly not exaggerating to say that the
present population is probably less than
1% of what existed a century ago, and the
decline still continues.

In this section, we review the current
situation and existing protective legisla-
tion on a country by country basis.

BOLIVIA

The black caiman once occurred in
much of northern and eastern Bolivia,
apparently as far south as Laguna
Caceres on the Brazilian border near
Corumba, and Puerto Grether, further in-
land on the Rio Mamore.

Hide-hunting was particularly heavy
during the period 1942-1960, and contin-
ues today. In 1973, Medem found over
2,000 skins (including 120 over 3 m in
length) in tanneries in Cochabamba and
Trinidad. He also saw a shipment of 500
black caiman skins in Riberalta in June
1973, all of them from the middle course
of the Beni. Some 70,000 Melanosuchus
skins were exported between 1967 and
1973 and Campbell (1977) reported 4580
skins exported in 1976.

The hide industry in Bolivia has de-
veloped sophisticated methods for ex-
ploiting caimans. Small planes are used
to locate isolated lakes and ponds in the
forest, the position is fixed on aerial

698

photographs, and well-equipped hunting
crews then cut trails or travel by canoe to
the area and exterminate the caiman pop-
ulation. Indications are that helicopters
may be used in the future if enough
caiman are present to make it commer-
cially feasible.

Since the tanning industry in Bolivia is
second only to mining and oil and gas in
export and since nature legislation is
largely ineffective, the black caiman does
not seem to have much of a future in this
country and has already been wiped out
in areas in which it once occurred. It is
clearly a highly endangered animal with-
in Bolivia.

BRAZIL

Once widespread in Brazilian Ama-
zonia, the black caiman is now rare or
locally extinct over most of its historical
range. A small population occurs in the
Amazonas National Park (Magnusson,
1979) and this may also be true in the
other large parks that have been estab-
lished in Brazilian Amazonia in the past
few years, but survey work is needed to
see if they are large enough to be viable
over the long term.

Illegal hunting continues and Magnus-
son (1980) reports that most river pas-
senger/freight boats carried caiman
(Melanosuchus and Caiman) hides in
1979 and 1980. In 1973, Mittermeier
found skulls of black caiman that had
been eaten in small villages along the Rio
Negro and the Rio Solimoes.

COLOMBIA

The black caiman was once abundant
in the Colombian Amazon, from Leticia
to the Rio Atacuari at the Peruvian bord-
er, and along the Rio Putumayo and the
Rio Caquetd, major Colombian tributar-
ies of the Amazon. Along the upper
Putumayo, its range was limited by the
falls and rapids of Araracuara; on the Rio
Yari, a tributary of the Caqueta below
Araracuara, it was limited by the rapids of
La Gamitana; and on the Rio Mesai, a

Advances in Herpetology and Evolutionary Biology

tributary of the Yari, it was limited by the
falls known as E] Bufeo. It occurred in
the lower Rio Apaporis, largest tributary
of the Rio Caquetd, but its spread up this
river was prevented by the rapids of La
Libertad as well as two other sets of
rapids.

Although larger falls and rapids proved
a barrier to Melanosuchus distribution
in Colombian Amazonia, smaller ones
did not. The falls known as Chorro de
Cordoba on the Caqueta, above the vil-
lage of La Pedrera, did not limit its
spread into the Rio Miriti-Parana and the
Rio Cahuinart, and later up to the Ser-
rania de Araracuara (Medem, 1963).

In the 1940’s some attempts were made
to introduce black caiman into the upper
Rio Caqueta, above Araracuara (Medem,
1960, 1963). For instance, in 1943, a ser-
geant of the Colombian navy released 25
juvenile M. niger into the Caqueta near
the village of Tres Troncos. These an-
imals apparently did not reproduce.
However, some may have survived for at
least a while since an individual approx-
imately 300 cm in length was killed along
the Rio Sencella, a tributary of the upper
Caqueta, and it may have been one of the
last of these 25 juveniles.

Commercial hide-hunting began along
the Rio Amazonas before 1945 and on the
lower Rio Caqueta and the Rio Putumayo
in 1950 and resulted in a rapid decline of
Melanosuchus populations throughout
the region. By the late 1950’s, the species
was already rare. Medem, for instance,
did not observe any specimens during
expeditions to the La Pedrera region and
the lower Rio Apaporis. He only saw six
juveniles in Lago Tarapoto and the Rio
Loreto-yacu. above Puerto Narino
(October, 1958) and saw none at La
Concepcion on the Putumayo (Novem-
ber, 1958). In 1968, Medem also did not
see any black caiman during a three
month expedition on the Rio Caqueta,
Miriti-Parand, and Cahuinart, although
local Indians claimed that a few widely-
scattered and shy individuals still exist-

ed.

CONSERVATION OF THE BLACK CAIMAN ° Plotkin et al.

Few black caiman were seen in
Colombia in the 1970’s. Roger Foote ob-
served black caiman in the wild only
once between 1973 and 1975. This was a
sighting of four individuals under 120 cm
in length in a shallow lake about one
hour from Leticia and 30 minutes from
the town of Benjamin Constant (R. Foote,
in litt. to Medem, July 5, 1975). In 1977,
two Colombian biologists, Hernando
Chirivi Callego and Jorge Morales, made
a two-month survey of M. niger popula-
tions along the Amazon and the Putum-
ayo. They found only one specimen, a 62
cm juvenile kept by a settler at the mouth
of the Rio Caucaya. Finally, in 1977 a
female about 350 cm in length was re-
ported killed by a settler on the Rio
Caqueta, close to La Pedrera (P. von
Hildebrand, personal communication).
No more recent records exist.

Despite the passage of laws designed
to protect the black caiman, hunting con-
tinued at least until 1972. Statistics ob-
tained from INDERENA and from the
Instituto de Comercio Exterior (INCO-
MEX) reveal that a total of 66,369 hides
and live animals were illegally exported
from Leticia, Bogota, Barranquilla, and
Cali in 1970 (Medem, 1971) and 1972
(Table 1). After 1972, official export
statistics were no longer available.

It is clear that hide hunting has all but
exterminated the black caiman _ in
Colombia, and that this animal is highly
endangered having been reduced to
small numbers of widely dispersed indi-
viduals. A major conservation program
would be necessary to restore it to even a
fraction of its former abundance, and it
may already be too late for such a pro-
gram to be undertaken.

ECUADOR

In Ecuador, the black caiman is found
in the drainage of the Rio Napo, extending
east almost to the foothills of the Andes.
It is expecially fond of lakes and lagoons
and sometimes occurs in large rivers at
places where the current is weak. It has

699

been recorded from several tributaries of
the Curaray, the Conambo, and the
Pindo, themselves tributaries of the
Maranon (upper Rio Amazonas). Black
caiman occur at relatively high densities
in Limon Cocha (K. Miyata, personal
communication) and may also occur in
established conservation units in
Amazonian Ecuador such as Yasunr
National Park, Cuyabeno Faunal Produc-
tion Area, and Lagartococha Ecological
Reserve (C. Freese, personal communi-
cation).

There appears to be little active hide-
hunting in Ecuador at this time. How-
ever, there was once a very active hide
industry for the American crocodile
(Crocodylus acutus) in Ecuador (C.
Freese, personal communication), but we
were unable to determine if the black
caiman was also included.

FRENCH GUIANA

The black caiman occurs only in the
extreme northeastern corner of French
Guiana. In this region, it can still be seen
in the seasonally flooded grasslands (les
Marais de Kaw) of the remote Kaw River
and in the largely inaccessible swampy
area known as la Savanne Angelique,
which is bordered by the Kaw River, the
Montagnes de Kaw, and the Montagne de
Gabrielle. Black caiman are also some-
times found to the east of Kaw between
the lower Approuague River and the
Ounary River (J. Lescure, personal com-
munication), at a few sites along the
lower Approuague River (particularly at
the mouth of the Kaw Canal and by the
Mantouni and Aipoto Islands, J. Lescure,
personal communication), and in the
small Ouapou Creek to the south of the
Montagnes de Kaw. They once occurred
in the Gabrielle Creek, the Savanne
Gabrielle, and the Mahury River to the
west of Kaw, but are now extinct there (J.
Lescure, personal communication). They
also were once common along the lower
Oyapock and its tributaries, the Gabaret
and the Ouanary River, but Brazilian

700 Advances in Herpetology and Evolutionary Biology

TABLE 1. EXPORT DATA FOR MELANOSUCHUS NIGER SKINS FROM COLOMBIA.

Year Hides/Live caimans
1970 30,105 (H)

1970 436 (H)

1970 7,633 (H)

1970 259 (L)

1970 43,417 (H&L)
1972 22,254 (H)

1972 698 (H)

1972 22.952 (H)

Total for 1970 and 1972: 66,369 (H & L)

poachers have almost completely wiped
out the animal along the Oyapock (al-
though occasional individuals can still be
seen there—e.g., Condamin, 1975; R.
Renau Ferrer, personal communication).
Stragglers even appear from time to time
on the beaches of Cayenne and nearby
Montjoly. A 3 m specimen was killed by a
gendarme in 1918, and two others were
shot on the Montjoly beach in June 1972
and January 1975. A distribution map for
the species in French Guiana is given in
Condamin (1976).

Laws forbidding the hunting of black
caiman have not been effective as ev-
idenced by the fact that a local business-
man requested permission to export 500
black caiman hides in March 1975 and
apparently was not prosecuted. French
officials started to seize in-transit ship-
ments, however, and there have not been
any more since 1977. Nonetheless, there
is a small-scale trade carried out by local
hunters, and some animals are killed by
Brazilian hunters or by tourists from
Cayenne (J. Lescure, personal communi-
cation).

Fair numbers of Caiman crocodilus
hides are shipped from French Guiana to
tanneries in France, and the tanners are
trying to open a trade in Melanosuchus as
well. Such shipment is not international

Site Source
Leticia Medem, 1971
Tarapaca INDERENA, March, 1971
Bogota, Medem, 1971
Baranquilla
Bogota, Mendal Bros., Lao,
Baranquilla INDERENA
(Total, 1970) Medem, 1971
Bogota, Cali INCOMEX
Leticia Scheuerman and Foote,
in litt.
(Total, 1972)

since French Guiana is a department of
France rather than a separate country.
France has taken a reservation on
Melanosuchus under CITES so there is
every reason to believe that French tan-
ners will continue to push for the French
Guiana hides. They are already supply-
ing French Guiana papers on shipments
of Caiman crocodilus yacare hides orig-
inating from Paraguay and U.S. officials
are accepting the papers for U.S. imports
(F. Wayne King, personal communica-
tion).

Stuffed juvenile spectacled caiman are
sold in fairly large numbers in the mark-
ets and tourist shops of Cayenne, but pro-
tective legislation has at least been effec-
tive enough to exclude the black caiman
from this trade (although the remoteness
of the animal’s habitat has undoubtedly
played a role as well). In three surveys
conducted in 1975, 1976, and 1979, we
saw hundreds of spectacled caiman but
not a single black caiman for sale in the
shops in Cayenne.

Although the Kaw population may be
one of the largest enclaves of black
caiman left in South America, it is a rel-
atively small area and still not a park or
reserve. We therefore believe that the
black caiman should be considered an
endangered species in French Guiana.

CONSERVATION OF THE BLACK CAIMAN : Plotkin et al.

GUYANA

In Guyana, the black caiman is found
in the Rupununi, the Essequibo, and
the Berbice Rivers. The black caiman
was apparently once abundant in the
Rupununi District, but hide-hunting was
a major cause of decline. People from
Guyana would apply for permits to hunt,
and would then arrange for Brazilian skin
dealers in Boa Vista to cross the border
and organize local Amerindians to under-
take the actual hunting and skinning. A
handful of skins were declared to cus-
toms officials situated at Lethem, and
much larger numbers were sent across
the Tacutu River at uncontrolled points.
According to a manager of the Balata
Company (personal communication to F.
Medem), the commercial hunting result-
ed in such a wholesale slaughter “that
the lower reaches of the river at Apoteri
stank for weeks, and a large number of
piranhas suddenly appeared to feed on
the numerous black caiman carcasses.” It
was impossible to estimate the number of
skins shipped out to Brazil, but Medem
learned that one hunter obtained 5,000
skins on his own and a second got about
700. As a result of the slaughter, the
Guyana government in 1965 declared a
total ban on caiman hunting for a period
of five years.

We do not have any information on
further protection in Guyana, but it is
clear that the species has declined drasti-
cally in that country and must be consid-
ered endangered there.

PARAGUAY

During his 1973 survey in Paraguay,
Medem obtained unexpected informa-
tion suggesting that the black caiman
might occur in that country. Although he
did not himself see live individuals or
hides, two reliable informants told him
that they had seen or shot several spec-
imens 3 to 5 m in length, which, by their
descriptions (e.g., white ventral surface,
size), are unlikely to have been Caiman
crocodilus yacare or Caiman latirostris,

701

the other caiman species in Paraguay.
The localities in question included a
stretch between the Rios Tebicuary and
Tebicuary-mi and the village of Villa del
Pilar (1950), and the Laguna Ypoa (1965),
both within the drainage of the Rio
Paraguay.

The black caiman is not mentioned by
Bertoni (1913) in his list of Paraguayan
reptiles, but this is not surprising given
what would be a very low black caiman
population density in a little-known area.
Populations of black caiman exist (or
existed) nearby in the upper Rio Para-
guay of Brazil's Mato Grosso (e.g., Sao
Luis de Caceres, Corumba, the town of
Mato Grosso), so it would not have been
difficult for the animal to enter Paraguay
from these areas.

The black caiman would be’ protected
under the general decree (Decreto No.
18.796 of 1975) prohibiting hunting,
commercialization, and export of Para-
guayan wildlife. Since this law was
passed, numerous warehouses of ani-
mals have been shut down in the capital
city of Asuncion. However, one can still
buy novelty items such as stuffed
Caiman latirostris in tourist shops locat-
ed along Asuncion’s Calle Colon (Jody
Stallings, personal communication). In
addition, illegal shipments of caimans
from Paraguay still enter the U.S. and
Europe (F. Wayne King, personal com-
munication), although it is not known
whether Melanosuchus has been _ in-
cluded among these.

Recently, the Servicio Forestal Nacion-
al has initiated a national biological
survey in Paraguay. This project will add
to our knowledge of caiman distribution
within Paraguay. In any case, if the black
caiman does occur, it would definitely
have to be considered highly en-
dangered.

PERU

The black caiman was once abundant
throughout the upper Amazon drainage
in Peru. However, commercial hide-hun-

702 Advances in Herpetology and Evolutionary Biology

ting, which started around 1950 took a
tremendous toll with at least 26,463 still
being exported as recently as 1970 (FAO
statistics, quoted in Medem, 1971).

This crocodilian is now on the verge of
extinction in Peru, with good populations
existing only in the Manu National Park
and possibly in the large swampy area on
the lower Rio Madre de Dios. A popula-
tion also exists in the Pacaya-Samiria
Reserve, largest of Peru’s protected
areas, and its status is currently being in-
vestigated by Pekka Soini.

IMPORTANCE OF THE BLACK CAIMAN
IN THE AMAZONIAN ECOSYSTEM

The black caiman is not only important
as a wildlife heritage. It also occupied an
important position in the Amazonian eco-
system and its elimination in many areas
has resulted in a number of ecological
changes, some of them affecting man.

Fittkau (1970, 1973), reported that a de-
cline in caiman populations has brought
about an accompanying decline of fish
populations, apparently due to interrup-
tion of a delicate nutrient chain. Caiman
excrement in the lakes and swamps ap-
parently provides a key food source for
zoo- and phytoplankton, which in turn
serve as food for fish fry of the many
species that use these lakes as their major
egg-laying sites. When the caiman is
eliminated, its excrement no longer
forms the basis of the nutrient chain, and
the fish species dependent on it drop
significantly. This effect is accentuated
in the nutrient poor “black water’ areas
where living animals constitute a major
portion of the biomass and are respon-
sible for bringing allochthonous nutri-
ents into the partially closed system.

Overexploitation of caimans therefore
results not only in the loss of the valuable
resource that these animals themselves
represent, but also can have a significant
effect on fish populations, which are the
major protein source for man _ in
Amazonia. (This effect, however, may at

least be partially offset by the increase in
the population of spectacled caiman [C.
crocodilus] that has apparently accom-
panied the decline of the black caiman
[Magnusson, 1980].)

Black caiman also play an important
role in control of animals harmful to
man’s agricultural activities in Amazonia.
In Bolivia and part of Brazilian
Amazonia, for instance, it is reported that
the widespread extermination of the
black caiman has resulted in an increase
of their herbivore prey species, notably
capybara and other smaller rodents, and
these have become serious agricultural
pests. In some parts of Bolivia, an in-
crease in the piranha (Serrasalmus spp.)
population was noted, to the extent that
these fish frequently attacked cattle cros-
sing the flooded grasslands.

The message here is that black caiman
conservation is not only of interest to
preservationists. It is also an essential
part of habitat management and helps as-
sure the continued usefulness of the
Amazonian riverine ecosystem for man.

THE FUTURE OF THE BLACK
CAIMAN—KEY CONSERVATION AREAS

Although small populations or isolated
individuals undoubtedly hang on in
many remote comers of Amazonia (e.g.,
Magnusson, 1979), few substantial con-
centrations of black caiman remain. We
know with certainty of only three, and
briefly discuss them here.

LIMON COCHA, ORIENTE PROVINCE,
ECUADOR

Limon Cocha is a lake in Amazonian
Ecuador, which harbors a large popula-
tion of black caiman. During the dry sea-
son (December-—February), it is possible
to see over 100 individuals in several
hours of night searching (Ken Miyata,
personal communication). Limon Cocha
was proposed for ecological reserve
status in the late 1970’s, but we were un-

CONSERVATION OF THE BLACK CAIMAN : Plotkin et al.

able to learn whether any official action
has been taken.

MANU NATIONAL PARK, DEPARTMENT OF
MADRE DE DIOS, PERU

One of the largest, most important
and most remote protected areas in Ama-
zonia, the 1,500,000 ha Manu National
Park has a large and undisturbed popula-
tion of black caiman that was estimated to
number some 150 adult individuals in
the early 1970’s (K. C. Otte, personal
communication to Medem). In the Manu
River, the animals can still be seen dur-
ing the day, a habit that remaining indi-
viduals in heavily hunted areas have
abandoned long ago. The Manu Mela-
nosuchus were studied by a German
ecologist, Dr. K.-C. Otte, but thus far only
a brief report has appeared on this work
(Otte, 1974).

Otte (1974) also surveyed a number of
other rivers in Peru, among them the
Sotileja, the Piedras, the Heath, and the
Pariamanu. Apart from the upper Rio de
las Piedras, he did not encounter black
caiman in any of these. Based on reports
from hunters and skin dealers and from
local government statistics, he concluded
that the black caiman existed in econom-
ic numbers only on the upper courses of
the Rio Tambopata, Manu, Piedras, and
Amigo. Viable populations have recently
been seen in lagoons along the Rio
Tambopata (K. Miyata, personal com-
munication), and Plotkin found evidence
of continued hunting of black caiman
along the eastern border of Manu Park in
July of 1980. It is clear that continued
protection of the important Manu Na-
tional Park population of black caiman is
essential to the survival of this animal in
Peru.

THE KAW REGION, FRENCH GUIANA

The Kaw region is evidently the last
major black caiman concentration in the
Guianas, and perhaps even in all of lower
Amazonia. The vast swamps to the west
of the Kaw River and north of the Monta-

703

gnes de Kaw are inaccessible, and the
Kaw River itself has a very small human
population (not exceeding 100 individ-
uals) in the village of Kaw and on two
nearby cattle ranches.

A former animal dealer familiar with
Kaw estimates that there may still be
1,500 to 2,000 large black caiman in the
whole region. However, the French
herpetologist J. Lescure (personal com-
munication) believes that 1,000 should
be considered a maximum for all of
French Guiana. In any case, the Kaw
region certainly still harbors some of the
5 m giants common in the past. Towards
the end of the dry season, when most of
the swamps are without water, a large
pond known locally as the “Mare aux
Caimans’’ forms in the uninhabited area
to the west and it is said that individuals
6 to 7 m in length can still be seen.

Small individuals are commonly seen
at night in the Kaw River. During a brief
expedition to this river in July, 1979,
Plotkin, Mittermeier, and Constable saw
a number of small black caiman one
meter in length and under, and a single
large adult of about 5 m. Since it was the
wet season, most of the larger caiman
were dispersed throughout the flooded
grasslands, but local informants told us
that many concentrated in the river dur-
ing the drier months.

Black caiman are occasionally hunted
for food by the inhabitants of the region,
but this is rare and other mammal and
avian game are much preferred. A greater
threat is the seasonal buming of the
dried-out swamps around Kaw, a practice
that probably results in the destruction of
many caiman nests. Furthermore, the
regions inaccessibility will soon end
with the construction of a road to Kaw (J.
Lescure, personal communication).

The area should obviously be protect-
ed, and this was already suggested by M.
Condamin, a French zoologist working in
the Cayenne branch of ORSTOM, in
1975. In a report entitled “Projets de
Reserves Naturelles sur le Littoral
Guyanais’ (Condamin, 1975, see also

704 Advances in Herpetology and Evolutionary Biology

Condamin, 1976), he suggested four
areas for reserves along the French
Guiana coast. These include Mana
(22,000 ha) for sea turtles, Sinnamary
(15,000 ha) for scarlet ibis and other
coastal birds, Le Grand Connetable (1
ha), a tiny island important for nesting
marine birds, and Kaw (49,000 ha) for
black caiman and hoatzin. The first three
were established, but Kaw still has not
because of lack of funds. The Society for
Nature Protection in French Guiana
(SEPANGUY) and its director, Dr. Leon
Sanite, are very concerned that Kaw be
fully protected as soon as possible, and
we are recommending to the French
government that Condamin’s plan for
Kaw be implemented as soon as possible.

ACKNOWLEDGMENTS

We would like to thank the following
people for the information that they pro-
vided: J. Dorst, R. Ferrer, R. Foote, C.
Freese, P. von Hildebrand, F. W. King,
J. Lescure, K. Miyata, P. Soini, and
J. Stallings.

APPENDIX

South American legislation conceming
Melanosuchus niger:

BOLIVIA - Officially protected by
Decreto Supremo No. 08063 of August
16, 1967 (Article 20) which imposes a
total ban on hunting from July 31 to
January 1.

BRAZIL - The 1967 Fauna Protection
Law (Lei No. 5.197/67) officially protects
all forms of Brazilian wildlife. The black
caiman does not appear on the Brazilian
List of Endangered Species.

COLOMBIA - Officially protected by a
total ban on hunting (Resolution No. 411,
July 6, 1969) passed by the Division de
los Recursos Naturales of the Ministerio
de Agricultura. In 1969, the Instituto de
Recursos Naturales Renovables (INDER-
ENA) also declared a ban on black

caiman hunting and egg collection (Res-
olution No. 573, July 24, 1969).

ECUADOR - The black caiman was
not included in the wildlife protection
Resolution No. 818 of 1970, but was
covered by the total ban on wildlife ex-
port declared in 1972. :

FRENCH GUIANA - First protected in
French Guiana in 1968 (Arrete Prefector-
al No. 68-719 ID/2B - January 31, 1975).
Melanosuchus received stronger protec-
tion in 1975 (Arrete Prefectoral No. 172
ID/2B - January 31, 1975).

GUYANA - The government passed a
total ban on caiman hunting in 1968 for a
period of five years.

PARAGUAY - All wildlife is protected
by general decree (Decreto No. 18.796 of
1975) which prohibits hunting, commer-
cialization, and export of Paraguayan
fauna.

PERU - The black caiman is officially
protected by law No. 21147 which pro-
tects all forms of wildlife other than those
of importance to local people as food
sources.

LITERATURE CITED

Bates, H. W. 1863. The Naturalist on the River
Amazons. London, John Murray.

BERTONI, A. W. 1913. Fauna paraguaya. Catalogos
sistematicas de los vertebrados del Paraguay.
In: Descripcion Fisica y Economica del
Paraguay. Asuncion, Moises S. Bertoni.

CAMPBELL, H. 1977. SSS/IUCN Croc. Gp. Newsl. 12.

CONDaMIN, M. 1975. Projects de reserves naturelles
sur le littoral Guyanais. Unpublished report.

—_. 1976. Faunes originales (plaines cOtieres).
Atlas des Départements d’Outre Mer. IV. La
Guyane, 1-2. Edit. CNRS. ORSTOM.

FITTKAU, E. J. 1970. The role of caimans in the nu-
trient regime of mouth-lakes of Amazon afflu-
ents. Biotropica, 2(2): 138-142.

____.. 1973. Crocodiles and the nutrient metabolism
of Amazonian waters. Amazoniana, IV(I): 103—
133.

GOELDI, E.A. 1898. Die Eier von 13 brasilianischen
Reptilien, nebst Bemerkungen ueber Lebens-
und Fortpflanzugsweise letzterer. Zool. Jahrb.
(Syst.), 10: 640-676.

HAGMANN, G. 1902. Die Eier von Caiman niger.
Zool. Jahrb. (Syst.), 16: 405-410.

MAGNUSSON, W. E. 1979. The distribution of caiman
within the Parque Nacional da Amazonia

CONSERVATION OF THE BLACK CAIMAN - Plotkin et al.

(Tapajos). Unpublished report to the National
Amazon Research Institute, Manaus.

____. 1980. Report on the status of Amazonian croco-
dilians. Unpublished report to the IUCN/SSC
Crocodile Specialist Group.

MEDEM, F. 1960. Datos zoo-geograficos y ecoldgicos
sobre los Crocodylia y Testudinata de los rios
Amazonas, Putumayo, y Caqueta. Caldasia,
8(38): 341-351.

____. 1963. Osteologia craneal, distribucion geograf-
ica, y ecologia de Melanosuchus niger (Spix),
(Crocodylia, Alligatoridae). Rev. Acad.
Colomb. Ci. Exact., Fis., Nat., 12(45): 5-19.

____. 1970. El estado actual respeto a la extermina-
cidn de los crocodilideos en la hoya Orinoco
Colombiano. Unpublished report.

____. 1971. Situation report on crocodilians from
three South American countries. In: Croco-

705

diles. Proc. First Work. Meeting IUCN/SSC

Croc. Spec. Group. IUCN Publ. (32): 54-71.
1972. Situation report on South American

crocodilians. Unpublished report.

____. 1973. Summary of the surveys of the status of
crocodilian species in South America. In: Proc.
Sec. Work. Meeting IUCN/SSC Croc. Spec.
Group. IUCN Publ. (41): 33-36.

MELO CARVALHO, J. C. 1967. A conservacao da nat-
uraleza e recursos naturais na Amazonia
Brasiliera. Atas Simposio Biota Amazonica, Rio
de Janeiro, (7): 32-33.

OTTE, K.-C. 1974. Research programme Melano-
suchus niger in the Mant National Park. World
Wildlife Yearbook 1973-1974: 257-260.

WALLACE, A. W. 1853. A narrative of travels on the
Amazon and Rio Negro. London, Reeve and Co.

The Leatherback Sea Turtle, Dermochelys
coriacea, Nesting in the Dominican Republic

JAMES PERRAN ROSS 1

JOSE ALBERTO OTTENWALDER 2

ABSTRACT. An estimated 300 female Dermochelys
nest each year in the Dominican Republic, and 37%
of nestings occur on four beaches. Nearly 100% of
nesting females and eggs are taken by local people
for food. Favored beaches have no offshore fringing
reef and few human inhabitants nearby. Conserva-
tion measures need implementation and enhance-
ment.

INTRODUCTION

The leatherback sea turtle Dermo-
chelys coriacea is the largest of the sea
turtles and is considered to be an en-
dangered species by both the United
States Department of the Interior and the
International Union for the Conservation
of Nature. Leatherbacks nest widely on
mainland shores of the tropics, but large
concentrations are only reported from
Trengganu, Malaysia, French Guiana
and Surinam, and Pacific Mexico. Few
nesting populations have been adequate-
ly surveyed or receive protection
(Pritchard, 1971; Ross, 1982).

In the Caribbean there are small pop-
ulations of leatherbacks nesting in Flori-
da, Vieques, Culebra, the United States
Virgin Islands, Trinidad, and Costa Rica.
Only the Costa Rican population is
thought to number more than a few
dozen individuals. Some of these small

' Museum of Comparative Zoology, Harvard Uni-
versity, Cambridge, Massachusetts 02138, U.S.A.
“Museo Nacional de Historia Natural, Plaza de la
Cultura, Santo Domingo, Dominican Republic.

populations have been proposed for
protection (Dodd, 1978).

Leatherbacks are known to migrate in-
to temperate waters from their tropical
breeding grounds. They are commonly
reported in Massachusetts waters (Lazell,
1980) and as many as several hundred
sightings per year have been made in re-
cent years. (C. R. Shoop, personal com-
munication)

The origin of these leatherbacks is un-
known, but there appear to be too few
adults in the reported nesting areas to ac-
count for the sightings made in temperate
waters. It appears likely that there are
unreported nesting sites in the Carib-
bean area.

Recently, a previously unreported
nesting site for leatherbacks was reported
in the northeastern Dominican Republic
(D.R.) on the island of Hispaniola (A.
Carr; T. Carr, personal communication).
The aim of the present survey was to ac-
curately locate the nesting beaches in the
D.R., to estimate how many leatherbacks
nest there; to evaluate the activities of
people taking eggs and killing adults for
food; to initiate contact with the D.R.
authorities; and to assess the population
for conservation action.

METHODS

To achieve these aims the senior auth-
or visited the Dominican Republic be-
tween March 24th and April 13th, 1980.

DERMOCHELYS IN THE DOMINICAN REPUBLIC - Ross and Ottenwalder

On March 27th, the authors flew in a
chartered aircraft around the eastern part
of the D.R. to search for beaches and
signs of nesting. In the next three weeks
we traveled by car to many coastal loca-
tions. At each place we talked with local
people and examined the beaches for
signs of recent nesting and skeletal mate-
rial.

AERIAL SURVEY

We flew around the coast from Miches
to La Romana, maintaining a height of
31 mand a speed of 160 km/hr., position-
ed slightly offshore of the beach. We
noted beaches that appeared suitable for

Puerto Plata

: DOMINICAN

0 REPUBLIC

Barahona

: “isla Beata

707

nesting and looked for tracks, body pits,
and carcasses on the beaches. Locations
were established using time of flight, a
1:600,000 map, and prominent land-
marks.

GROUND SURVEY

Areas surveyed in detail on the ground
are shown in Figure | and Table 1.

INTERVIEWS

Interviews were conducted in Spanish
through a fluent translator. No formal in-
terview sequence was used but conversa-
tion led into asking what types of sea

@ concentrated nesting

=
Zo

low density nesting
C.Samana

Pd MUERTO
P.MAC AO

C.Engano

> Isla Saona

| 60 km

Figure 1. Nesting of Dermochelys coriacea in the Dominican Republic. P = Playa; P.d. = Playa del; C. = Cabo

(Cape).

708

turtles occur nearby? Do they nest?
Where? How many nest in one night?
Are the eggs and meat good to eat? How
many are taken? Are turtles more or less
abundant than in previous years? Care
was taken to maintain neutrality to an-
swers and avoid leading people into
statements meant to please us. A total of
twenty-four interviews were used as a
data base after eliminating those contain-
ing clearly erroneous or inaccurate in-
formation. The distribution of these in-
formants by area is given in Table 1.

RESULTS

The local name for Dermochelys is
“Tinglar,” similar to names used in Pacif-
ic Mexico, “Tinglada” and _ Peru,
“Tinglado.”

NESTING SEASON

Only one recent nest was seen, on 30
March at Playa del Muerto in Altagracia
province. Local informants generally
agreed that most Dermochelys nesting
takes place from mid-April to mid-June.

DISTRIBUTION OF NESTING

Most coastal people recognized the lo-
cal name for Dermochelys, “Tinglar,”
and say that they are occasionally en-
countered. However, in most locations
people agreed that Dermochelys were
only seen from once/year to once/life-
time—in other words, they are uncom-
mon. Dermochelys were reported to nest
occasionally in very low densities
throughout the D.R. wherever suitable
sandy beaches occur. Three sections of
coastline support more regular nesting al-
though the individuals are widely dis-
persed. Within these regions four loca-
tions were identified that support modest
concentrations of nesting Dermochelys
(see Fig. 1 for locations).

Advances in Herpetology and Evolutionary Biology

TABLE 1. AREAS SURVEYED AND INTERVIEWS WITH
LOCAL INFORMANTS.

Punta Rucia -Luperon (20 km) — 3 interviews
Puerto Plata - Sosua (20 km) — 1 interview
Nagua - Las Terrenas (37 km) — 5 interviews
Miches - Boca Maimon (64 km) — 6 interviews

Macao

Barahona -Enriquillo
Cabo San Luis

Cabo Falso - Cabo Rojo

(5 km) — 1 interview
(140 km) — | interview
(13 km) — 2 interviews
(30 km) — 5 interviews

Low DENSITY NESTING
CONCENTRATIONS OF DERMOCHELYS

1. Nagua - Cabo Samana (North side of
Peninsula) (88 km).

There are many fine beaches along the
north coast of the Samana Peninsula.
These are difficult to approach by car, but
there are many small settlements with
almost continuous cultivation of coco-
nuts. The beaches at Las Terrenas and
Coson are narrow and backed by coconut
palms. Nesting is said to occur through-
out the area in low density. Local people
estimate from on to two per year up to
one per night on various sections of the
beach. Particular locations are Matan-
citas, El Pozo, Las Terrenas, Boca Rio
Limon (Puerto Escondido), and Galeras
(Bahia del Rincon).

2. Miches - Cabo Engano (94 km).

A series of narrow beaches backed almost
entirely by coconut plantations supports
low density nesting throughout its
length. The beach is accessible by a
sandy track through the coconuts. Esti-
mates of local informants range from one
to two per year up to 10 to 12 per year of
various sections. Included within this
section are two concentrations of nesting
that are discussed below.

3. Enriquillo - Pedernales (140 km).

This dry area with relatively low density
of human population has extensive
beaches lying around the tip of the
Barahona Peninsula. Two concentrations
are reported below. It is highly likely that

DERMOCHELYS IN THE DOMINICAN REPUBLIC - Ross and Ottenwalder

there is more Dermochelys nesting in
this remote area, particularly between
Cabo Falso and Punta San Luis.

Other occasional sightings of nesting
Dermochelys were reported as occurring
at Isla Beata, Bahia de Ocoa, and Isla
Saona.

HIGH DENSITY NESTING
CONCENTRATIONS OF DERMOCHELYS

1. Playa del Muerto (- de los Muertos)
Altagracia province.

Approximately 4 km of excellent nesting
beach is located 10 km south of the
mouth of the Rio Nisibon. The beach is
accessible by motor vehicle from the
main road at Las Vacamas. The beach is
40 m wide with a wide, well-defined
beach platform stabilized by Ipomea and
grasses. There is no fringing reef, and
consequently there is heavy surf on the
beach. Behind the beach is a strip of
coconut plantation 200 m wide, and be-
hind this is a mangrove swamp. Local in-
formants including two nearby residents
agree that this beach is the major concen-
tration of Dermochelys between Rio
Nisib6n and Rio Maimon. This location
corresponds to the location near Nisibon
reported by T. Carr (personal communi-
cation) who visited it in 1978. Estimates
vary from two to six turtles nesting each
night in April and May. Knowledgeable
informants were properly cautious about
estimating annual populations, citing the
difficulties of multiple nesting and in-
complete observations. When pressed,
informants guessed from 40 to 60 turtles
nesting each year.

Informants readily admitted to taking
all the eggs they found and killing the
females whenever they were encounter-
ed. Three separate informants agreed
that between 12 and 15 Dermochelys
were killed here in 1979. Examination of
the beach revealed several old nest pits
and bones from two separate individuals.
A freshly killed carcass, 4 to 10 days old,

709

was found on the beach with plastron,

limb bones, and meat removed.
Nesting at lower densities is reported

on the narrow beaches between Rio

Nisibon and Rio Maimon (approx.
32 km).

2. Playa Macao Altagracia province.
Approximately 5 km of excellent nesting
beach lies south of Punta Macao. The
track to Cabo Engano lies 250 m inland of
the beach separated from it by low dunes
and scrub. The beach is 40 m to 120 m
wide with a high beach platform stabil-
ized by dune vegetation. Local inform-
ants report that 3 to 4 Dermochelys can
be seen each night during April and May
and that an experienced collector work-
ing on foot from El Macao can collect 500
eggs in one night. Local people claim to
only occasionally kill females for meat.
Examination of 2.5 km of the beach did
not reveal any recent nesting or turtle
specimens.

Dermochelys nesting is said to occur in
lesser density on the narrower beaches
southward to Cabo Engano (approx. 20
km).

3. Playa San Luis Pedernales province.
Approximately 15 km of nesting beach lie
between Cabo San Luis and Cabo
Inglesa. There is a narrow strip of palm
scrub behind the beach, then a large sal-
ine lagoon that effectively isolates the
beach. Access is difficult, on foot or
across the lagoon by small boat from the
main road at Oviedo Viejo. This location
is the site of a new tagging and conserva-
tion program of the Museo Nacional.
Under the direction of Ottenwalder, a
local resident patrols about 5 km of the
beach each night and is instructed to tag
and measure all Dermochelys. This per-
son estimates that between 1 and 3
Dermochelys come ashore most nights in
April and May. In 1979 he observed 15
nesting turtles. Recommendations for
enhancing the effectiveness of this pro-
gram are given below.

710 Advances in Herpetology and Evolutionary Biology

4. Playa Aguilas Pedernales province.
About 5 km of remote beach is located 9
km south of Cabo Rojo in Bahia de las
Aguilas. Another small beach (Playa
Majer or Playa Caliente) is said to be
further south. The beach lies in a sparse-
ly inhabited region and is accessible by a
very rough track from Cabo Rojo or by
boat from the nearby fishing village of
Cueva. Fishermen at Cueva (population
about 80) report that one to two Dermo-
chelys can be seen any night in April—
June. I found parts of two old carcasses
butchered for meat, and fishermen say
they regularly take the eggs. The beach is
rather narrow (5-15 m) but slopes steeply
up to a stabilized dune platform.

Topographical details for these areas
can be seen on U.S. Army Topographic
maps: Dominican Republic 1:50,000
Series E 733 Ed. 2. Sheets: 64721,
647211, 6572111, and Ed. 3. Sheets:
58691, 586911, 5969111.

PREDATION BY PEOPLE

Predation by people on the eggs and
nesting females of Dermochelys is com-
mon throughout the Dominican Re-
public. A striking feature of most of the
coastline is the high density of the human
population. Every single beach visited
showed signs of at least weekly visita-
tion, and most beaches are patrolled daily
by people who live nearby. As a result a
very high proportion of eggs of all turtle
species are found and taken. The eggs are
both for home consumption and for sale.

People mark the location of nests and
return between 10 and 30 days later in an
attempt to intercept the female on a later
nesting visit. The females are slaughter-
ed without regard for whether they lay or
not, and the meat is transported to nearby
villages for sale. Dermochelys eggs retail
for 30¢ (U.S.) each and the meat for $1/Ib.
and thus represent a sizeable income for
the vendor.

Coastal dwellers frequently claimed to
kill all the Dermochelys they encounter-

ed. All the Dermochelys specimens
found in this study (five individuals) ap-
peared to have been killed for meat. The
vigor of persecution seems to be most in-
tense on the east coast in the region of
Nisib’¥3on and more casual at Macao, in
the Samana Peninsula and in the south-
west. However, exact data are lacking.
The taking of eggs and the killing of nest-
ing females constitute the most serious
threat to the survival of Dermochelys in
the Dominican Republic.

Skin divers, who prey heavily on
hawksbill turtles (Eretmochelys imbrica-
ta), claim to rarely encounter Dermoche-
lys and do not spear them because they
are too large to handle.

Accidental capture in fisheries aimed
at other species was not reported and ap-
pears to be negligible.

POPULATION ESTIMATES

Population numbers of sea turtles are
notoriously difficult to assess as several
of our informants were aware. The fol-
lowing estimates are included to give
‘order of magnitude’ understanding of
the importance of D.R. nesting beaches
and to show the relative importance of
different areas. In the absence of nightly
data on nesting at the height of the sea-
son, these estimates are based on infor-
mation from informants and are conse-
quently of low reliability. We believe
most informants overestimate numbers of
turtles because they remember occasion-
al nights of heavy nesting and do not cal-
culate in many nights when few turtles
nest. On the other hand, individuals’ est-
imates are usually limited to short sec-
tions of coast that they know well and
thus estimates from distant informants
can be summed. The highly dispersed
low density of nesting in much of the
D.R. makes population assessment even
more difficult.

Population numbers are estimated in
two ways. We take the estimated number
of turtles nesting each night and, assum-
ing a season 60 days long, and that each

DERMOCHELYS IN THE DOMINICAN REPUBLIC - Ross and Ottenwalder

turtle nests three times, extrapolate to an
annual nesting population. We also give
the informants’ estimates of annual nest-
ing. These data are shown in Table 2.

The estimates show that each year per-
haps 300 Dermochelys nest in the
Dominican Republic. The estimate is
highly tentative and needs refining by
further research (see recommendations).

A considerable proportion of the est-
imated nesting is concentrated on four
beaches.

TURTLE CONSERVATION IN THE
DOMINICAN REPUBLIC

PROBLEMS

The major problem for the continued
existence of Dermochelys nesting popu-
lations in the D.R. is the uncontrolled
taking of eggs and killing of nesting
females for food by local people. The
problem is rendered complex because
the rural population is very poor, widely
dispersed, and uniformly dense. Turtles,
including Dermochelys, constitute an
important cash source for some members
of coastal communities.

A secondary problem is the develop-
ment of isolated beaches as tourist resorts

711

with a consequent increase in back light-
ing and disturbance.

Laws pertaining to wildlife and fisher-
ies conservation are contained in several
Presidential decrees of the Dominican
Republic. These are given in detail in
Ross (1980). In summary, there are provi-
sions for closed seasons and lower size
limits for sea turtles. However, this body
of legislation has some deficiencies and
is inadequate for protection of all sea
turtles, particularly Dermochelys.
Nomenclature in the legislation is un-
clear, and there is little biological justifi-
cation for the size limit and closed sea-
son. All of our informants agreed that
there is no policing of these regulations.

This situation is currently under re-
view and revised protective legislation is
pending implementation.

RESEARCH IN PROGRESS

There is a small but active group of re-
searchers, headed by Ottenwalder, that is
working to improve conservation in the
D.R. and is collecting data. The Museo
Nacional de Historia Natural (MNHN) is
designing a series of aerial surveys to
evaluate sea turtle nesting. MNHN has
begun a small tagging and protection pro-
ject on Dermochelys at Playa San Luis. A

TABLE 2. FIRST APPROXIMATION NESTING POPULATION ESTIMATES. FEMALE DERMOCHELYS CORIACEA IN THE

DOMINICAN REPUBLIC.

Location
nightly

Playa del Muerto 2-6
Playa Macao 3-4
Playa San Luis 1-3
Playa de Aguilas 1-2
Miches - Cabo Engano —

(excluding concentrations

above)
Nagua - Cabo Samana 4

Dispersed other —

Informants Estimated annual
annual*
40-60 40-120
100 60-80
45 20-60
= 20-40
2-12 20
less than 100 80
240
pal, 60 (?)
300

*The range of values estimated by local informants is given. See Table 1 for distribution of informants.

712

local man patrols the beach nightly, tag-
ging turtles and discouraging people
from taking eggs or killing turtles. These
studies are greatly handicapped by an
almost total lack of funding and suffer be-
cause the few trained and interested per-
sonnel in the D.R. need to spread their
activities over several conservation
projects in addition to their routine work.
Nevertheless there are promising begin-
nings.

CONSERVATION PLANNED

A pan-Caribbean turtle conservation
symposium is planned for 1983 in San
Jose, Costa Rica sponsored by
IOCARIBE and NOAA. It is hoped that
this conference will stimulate participat-
ing nations to establish an expanded data
base on sea turtles.

DISCUSSION

The documentation of a large number
of Dermochelys nesting in the Domin-
ican Republic is a useful addition to our
knowledge of the distribution of this spe-
cies, although further work is necessary
to improve the estimate of numbers. It is
likely that some of the leatherbacks nest-
ing in the D.R. migrate up the U.S. coast
to temperate waters. Tagging programs in
the D.R. will provide a means of verify-
ing this postulated pathway.

Within the three general areas where
Dermochelys are more common, there
are four beaches where concentrations of
nesting occur. These beaches share two
factors that appear to make them more
suitable for Dermochelys. All are backed
by infertile land with reduced human
population density. Playa del Muerto and
Playa Macao are backed by dunes and
mangrove swamp, Playa San Luis is
backed by a saline lagoon and Playa
Aguilas by barren limestone karst. This
may serve to reduce the regularity of dis-
turbance and predation by local people.

Each of the beaches where Dermo-
chelys nest in greater density lacks an off-

Advances in Herpetology and Evolutionary Biology

shore fringing reef and has heavy surf. As
a result the beaches are wide, rise steeply
to stable beach platforms, and provide
unimpeded access to the sea. In contrast,
much of the remainder of the D.R. coast-
line has a fringing reef 100 to 500 m off-
shore, with beaches that are uniformly
narrow (5-10 m) with coastal vegetation
forming a dense barrier at the top. Much
of this coastline supports a strip of coco-
nut palm plantations and the dense root
masses of the coconut trees form a barrier
just above the high tide line.

Hawksbill turtles (Eretmochelys im-
bricata) nest intermittently along the
narrow beaches, although their numbers
have been greatly reduced in recent
years. However, the narrow low profile
beaches are apparently poorly suited to
the nesting requirements of Dermo-
chelys. Sand quality seems to be relative-
ly unimportant to nesting turtles within
broad limits (Stancyk and Ross, 1978).
However, the extremely large Dermo-
chelys does require more open, unim-
peded beaches than the smaller, more
agile Eretmochelys, a conclusion similar
to that of Hendrickson and Balasingam
(1966). Unfortunately, remote beaches of
the sort favored by sea turtles are in great
demand for resort developments. Two
such developments were noted in the
aerial survey. Care should be taken that
similar development does not occur on
the very few beaches where concentrated
Dermochelys nesting occurs.

The widely dispersed nature of
Dermochelys nesting in much of the D.R.
and the complex socio-economic prob-
lems surrounding the use of turtles by
people will make conservation of Dermo-
chelys difficult. However, the small num-
ber of nesting concentrations provides an
opportunity for effective action. We sug-
gest that conservation activities be focus-
ed on the four important nesting beaches:
Playa del Muerto, Playa Macao, Playa
San Luis, and Playa Aguilas. We also re-
commend the present MNHN survey and
tagging program should be enhanced to
accurately assess the numbers and distri-

DERMOCHELYS IN THE DOMINICAN REPUBLIC - Ross and Ottenwalder

bution of nesting turtles. The basic re-
quirement is money to pay salaries and
provide very basic equipment. A limited
amount of expert advice on the expansion
of the tagging program will ensure that
useful biological data are accumulated.

Aerial surveys would be of limited use
in D.R. In addition to the usual problems
of ground truth calibration and observer
error, the narrow, littered and overgrown
beaches make accurate observations of
nesting difficult.

A large scale public relations and edu-
cational program on endangered sea
turtles needs developing in rural areas.
Something like the WWF mobile audio
visual center would be ideal, as most
coastal villages are accessible by road,
have electric power, and lack competing
entertainment on most evenings. Such a
program should be developed by Domin-
ican personnel, perhaps as an extension
service of the National Zoo which has al-
ready assumed a burden of public educa-
tion on natural resources.

As a further conservation measure, it
may be possible to buy fresh Dermo-
chelys eggs collected from areas of dif-
fuse nesting and hatch them from natural
beach hatcheries at a central protected
location.

ACKNOWLEDGMENTS

Ernest E. Williams’s early published
work concerned the taxonomy and sys-
tematics of turtles, including the Dermo-
chelyidae (Williams, 1950). Turtles con-
tinued to attract his interest, resulting in
over 20 papers and articles spanning 17
years. In reference to the extinct tortoises
of the West Indies he wrote, “... much
remains to be learned about the West
Indian fauna” (Williams, 1952), and he
has continued to show a lively interest in

713

turtles of the Caribbean region. His con-
cerns extend beyond systematics to the
problem of conservation of the present
fauna, and in early 1980 he encouraged
the senior author to visit the Dominican
Republic to assess the status of popula-
tions of Dermochelys coriacea reported
to be nesting there.

This work was carried out under con-
tract to the U.S. Fish and Wildlife Ser-
vice. Additional support was received
from Ocean Research and Education
Society, Chelonia Institute, Fauna Pre-
servation Society and Emest E. Williams.
We thank K. Musik for assistance with
translation in the field.

LITERATURE CITED

Dopp, C. K. 1978. Terrestrial critical habitat and
marine turtles. Bull. Maryland Herpetol. Soc.,
14(4): 223-240.

HENDRICKSON, J. R., AND E. BALASINGAM. 1966.
Nesting beach preferences of Malayan sea
turtles. Bull. Nat. Hist. Mus. Singapore, 33: 69-
76.

LAZELL, J. D., JR. 1980. New England waters: criti-
cal habitat for marine turtles. Copeia, 1980:
290-295.

PRITCHARD, P. C. H. 1971. The leatherback or
leathery turtle. [UCN Publications, monogr. 1:
39 pp. Switzerland, Morges.

Ross, J. P. 1980. Report on the leatherback sea turtle

(Dermochelys coriacea) nesting in the

Dominican Republic. U.S. Fish and Wildlife

Service, mimeo report, 17 pp.

1982. Historical decline of loggerhead,
ridley and leatherback turtles, pp., 189-195. In
Bjorndal, K. A. (ed). Biology and Conservation
of Sea Turtles. Washington, D.C., Smithsonian
Institution Press.

STANCYK, S. E., AND J. P. Ross. 1978. An analysis of
sand from green turtle nesting beaches on
Ascension Island. Copeia, 1978(1): 93-99.

WILLIAMS, E.. E. 1950. Variation and selection in the
cervical central articulations of living turtles.
Bull. Am. Mus. Nat. Hist., 94(9): 509-561.

___. 1952. A new fossil tortoise from Mona Island,
West Indies and a tentative arrangement of tor-
toises of the world. Bull. Am. Mus. Nat. Hist.,
99(9): 545-559.

SYSTEMATIC INDEX

PHYLUM PROTOZOA
Plasmodium balli 458, 460
colombiense 456, 466
floridense 456, 458
mexicanum 457
minasense 458
sp. 458
tropiduri 456, 458, 460

PHYLUM PLATYHELMINTHES

helminths 456

PHYLUM ARTHROPODA
Aculeata 605
Anthophoridae 605
ants 517, 603
Apidae 604
Apis cerana 609
mellifera 609
arthropods 445, 451, 510
Blattidae 605
Blattodea 604
Coleoptera 278, 609
crickets 445.
Dermaptera 604
Diptera 609
Drosophila 506
Embioptera 604
Eumenidae 605
Formicidae 605
Habrobracon juglandis 609
serinopae 609
Halictidae 605
Hemideina crassidens 615
Hemiptera 609
Hymenoptera 604, 609
insects 596
Isoptera 604
Kalotermitidae 605
Lasioglossum 608
Lepidoptera 609
Lutzomyia vextrix 457
Luzomya trinidadensis 459
Mantodea 604
Megachile rotundata 609
Megachilidae 605
Mormoniella vitripennis 609
Myrmecia 604
Nothomyrmecia 604
Orthoptera 604, 609
Parasitica 604
Phasmatodea 605
Rhytidoponera 604

roaches 604

Symphyta 604

Vespidae 604

PHYLUM CHORDATA

SUBPHYLUM VERTEBRATA

SUPERCLASS AGNATHA
acanthodians 223
placoderms 223

SUPERCLASS GNATHOSTOMATA

SERIES PISCES
crossopterygians 223
fish 517

Gambusia affinis 691
Ichthyostegalia 223
labyrinthodonts 223
Lepisosteus 223
paleoniscoids 223
Serrasalmus 702
sharks 223

teleosts 223

SERIES TETRAPODA
CLASS AMPHIBIA

ORDER GYMNOPHIONA
Gymnophiona 186

ORDER URODELA
FAMILY PLETHODONTIDAE
Aneides lugubris 9
Batrachoseps 11
Batrachoseps campi 11
wrighti 11
Bolitoglossa 2, 5, 8
Bolitoglossa cuchumatana 5
hartwegi 5
sp. 5
Chiropterotriton 2
Hydromantes 11
Lineatriton 2, 9
Nyctanolis, gen. nov. 2, 11
Nyctanolis pernix, sp. nov. 2
Oedipina 2, 9
Parvimolge 2
Pseudoeurycea 2
Thorius 2
ORDER ANURA
FAMILY BUFONIDAE
Ansonia 419
Ansonia megregori 420
mulleri 420
Bufo biporcatus 419

kisoloensis 352
lemur 104
marinus 419
punctatus 400
Bufonidae 426
Pelophryne albotaeniata 419
brevipes 419
lighti 419
FAMILY CENTROLENIDAE
Centrolenidae 426
FAMILY DENDROBATIDAE
Dendrobatidae 426
FAMILY DISCOGLOSSIDAE
Barbourula busuangensis 418, 427
FAMILY HYLIDAE
Hyla arenicolor 400
bogerti 35
carnifex 35
charlesbogerti 35
columbiana 33, 40, 46
crucifer 400
labialis 44, 46
leucophyllata 46
marmorata 46
microcephala 46
minuta 36, 46
parviceps 36, 46
praestans, sp. nov. 43
variabilis 40
Hylidae 426
Smilisca 48
FAMILY LEIOPELMATIDAE
Leiopelmatidae 393
FAMILY LEPTODACTYLIDAE
Eleutherodactylus 33, 110
Eleutherodactylus anolirex, sp. nov. 52
antillensis 104
briceni 53, 56
cochranae 104
devillei 56
schwartzi 104, 111
supernatis 52, 56
unistrigatus 54
vertebralis 52, 56
Leptodactylidae 426
Leptodactylus albilabris 104
Syrrhophus guttilatus 400
marnockii 402
FAMILY MICROHYLIDAE
Chaparina fusca 426
Kalophrynus pleurostigma 426
Kaloula baleata 426
conjuncta 426
picta 416, 426
rigida 426
Microhylidae 426
Oreophryne annulata 426
nana 426
FAMILY PELOBATIDAE
Leptobrachella mjobergi 27
Leptobrachium 419
Leptobrachium gracilis 23

SYSTEMATIC INDEX

hasselti 16, 20, 420

hendricksoni 19

minimum 26

montanum 16

nigrops 18

oshanense 26

pelodytoides 26

pullum 16, 21

sp. 21
Leptolalax 13, 31
Megophrys 30, 419
Megophrys minor 30

monticola 420
Scaphiopus bombifrons 30
FAMILY PIPIDAE
Pipidae 426
Xenopus laevis victorianus 352

ruwenzoriensis 352

vestitus 352

wittei 352
FAMILY RANIDAE
Arthroleptis adolfifriederici 353
Cardioglossa cyaneospila 354

cyaneospila inornata 355
Micrixalus mariae 418, 422
Ooeidozyga diminutiva 422

laevis 421
Phrynobatrachus acutirostris 353

asper 353

bequaerti 352, 353

graueri 353

pedropedetoides 352

versicolor 352
Platymantis 425
Platymantis cornutus 418, 422

corrugatus 421

dorsalis 422

guentheri 422

hazelae 422

ingeri 422

insulatus 422

laevigatus 422

lawtoni 422

polilloensis 418, 422

subterrestris 418, 422
Rana angolensis chapini 353

cancrivora 421

clamitans 488

diwata 422

erythraea 420

everetti 421

leytensis 421

limnocharis 421

magna 421

melanomenta 422

micrixalus 418, 422

microdisca parva 420

nicobarensis 420

pipiens 400

ruwenzorica 353

sanguinea 420

signata 421

715

716 Advances in Herpetology and Evolutionary Biology

woodworthi 422
Ranidae 427
Schoutedenella globosa 354
hematogaster 354
pyrrhoscelis 354
schubotzi 354
Staurois natator 421
FAMILY RHACOPHORIDAE
Afrixalus laevis 353
orophilus 353
Callixalus pictus 353
Edwardtayloria picta 426
spinosa 426
Hyperolius castaneus 352
castaneus submarginatus 355
chrysogaster 353
discodactylus 353, 355
frontalis 353
leucotaenius 353
leucotaenius allogynus 355
xenorhinus 358
Leptopelis kivuensis 352
Philautus 426
Philautus acutirostris 422
alticola 418, 426
aurifasciatus 422
emembranatus 422
leitensis 422
lissobrachius 421
schmackeri 421
sp. 421
surdus 422
williamsi 418, 426
Polypedates leucomystax 416, 426
macrotis 426
Rhacophorus appendiculatus 426
everetti 426
hecticus 418, 422
pardalis 426
zamboangensis 426

CLASS REPTILIA
SUBCLASS ANAPSIDA

ORDER COTYLOSAURIA
Captorhinus 191
Eocaptorhinus 192

ORDER TESTUDINES
Cryptodira 179, 191
Pleurodira 179, 180, 191
FAMILY CHELIDAE
Chelidae 58
Chelodina 160, 170, 179, 181
Chelodina expansa 179
longicollis 180
novaeguineae 174, 179
oblonga 160, 180
rugosa 179
steindachneri 180
Chelus 59, 180, 184
Chelus fimbriatus 179, 431
Elseya 66, 160, 177
Elseya dentata 169

latisternum 170
novaeguineae 170
Emydura 66, 169, 186
Emydura australis 171
dentata 170
kreffti 171
latisternum 177, 179
macquaria 170, 178
subglobosa 170, 176
Hydraspis geoffroyana 59, 70
Hydromedusa 59, 180
Phrynops 59, 186
Phrynops dahli 59, 70
geoffroana 60
geoffroana geoffroana 59, 70
geoffroanus 58, 179, 184
geoffroanus geoffroanus 59
geoffroanus sspp. 60
geoffroyana 59
gibbus 59, 65, 70, 179, 181, 184, 430
hilarii 59, 65, 69
hogei 59, 65, 70
nasutus 59, 70, 160, 430
rufipes 59, 65, 70
tuberculatus 59, 70
tuberosus 59, 70
vanderhaegei 59, 65, 70
wermuthi 59, 70
williamsi, sp. nov. 59, 65, 70
Platemys 59, 179, 181
Platemys geoffreyana 59, 70
geoffroyana 59
platycephala 160, 179, 184, 429
Pseudemydura 66, 179
Pseudemydura umbrina 169, 178
FAMILY CHELONITIDAE
Caretta caretta caretta 150
Chelonia mydas 186
Eretmochelys imbricata 440, 710, 712
FAMILY DERMOCHELYIDAE
Dermochelys coriacea 440, 706
FAMILY EMYDIDAE
Chrysemys picta 150
scripta ssp. 400
Clemmys guttata 186
Ocadia sinensis 186
Pseudemys dorbignyi 69
scripta elegans 150
FAMILY GLYPTOPSIDAE
Glyptops 192
Mesochelys 192
FAMILY KINOSTERNIDAE
Kinosternon hirtipes 186
hirtipes murrayi 400
FAMILY PELOMEDUSIDAE
Dacquemys 75
Erymnochelys 76, 160
Erymnochelys madagascariensis 75
Eusarkia rotundiformis 160
Kenyemys, gen. nov. 74
Kenyemys williamsi, sp. nov. 75
Latisternon microsulcae 79

Pelomedusa 76, 160, 181
Pelomedusa subrufa 161, 179
Pelusios 78, 159, 179, 181, 184
Pelusios adansonii 161
bechuanicus 159, 163
bechuanicus bechuanicus 161
upembae 162
blanckenhorni 163
carinatus 162
castaneus 159, 166
castaneus castaneus 159, 162
castanoides 162
chapini 159
derbianus 159
kapika 159, 162
lutescens 162
rhodesianus 159
williamsi 162
derbianus 179
dewitzianus 163
gabonensis 161
nanus 161
niger 161
rhodesianus 162
rudolphi 161, 163
rusingae 161
sinuatus 160, 162
sinuatus leptus 160
zuluensis 160
subniger 161
williamsi lutescens 159
williamsi 159
Podocnemis 69, 160, 179, 181, 184
Podocnemis aegyptiaca 79
antiqua 75
blanckenhorni 75
bramlyi 77
congolensis 75
fajumensis 75, 79
madagascariensis 75
podocnemoides 79
sextuberculata 65, 69, 78
unifilis 78, 179
venezuelensis 75, 160
Shweboemys antiqua 75
Stereogenys cromeri 80
libyca 79
Sternothaerus derbianus 160
Taphrosphys congolensis 75

FAMILY PLATYSTERNIDAE
Platysternon megacephalum 186

FAMILY PROGANOCHELYIDAE
Proganochelys quenstedti 190
Stegochelys dux 190
Triassochelys 191

FAMILY TESTUDINIDAE
Geochelone 538

Geochelone carbonaria 101, 111
FAMILY TRIONYCHIDAE
Trionyx 181, 186

Trionyx ater 186

SYSTEMATIC INDEX

SUBCLASS LEPIDOSAURIA

ORDER RHYNCHOCEPHALIA
FAMILY SPHENODONTIDAE
Homoeosaurus 614

Meyasaurus 376, 382

Opistias 614

Palacrodon 614

Polysphenodon 614

Sphenodon 186, 196

Sphenodon punctatus 201, 224, 613
Sphenodontidae 393
Toxolophosaurus 614

ORDER SQUAMATA

SUBORDER EOLACERTILIA
Cteniogenys 373, 376

Fulengia 367

Icarosaurus 367

Kuehneosaurus 367

Kuehneosuchus 367

Palaeagama 367

Paliguana 367

Saurosternon 367

SUBORDER LACERTILIA
FAMILY AGAMIDAE
Acanthosaura armata 209
crucigera 209
lepidogaster 209
Agama 208, 368, 380
Agama agama 209
anchietae 209
atra 209
caudospinosa 209
doriae 209
hartmanni 209
hispida 209
kirkii 209
mossambica 209
planiceps 209
rueppelli 209
sankaranica 209
weidholzi 209
Amphibolurus barbatus 209
caudicinctus 209
clayi 209
cristatus 209
diemensis 209
femoralis 209
fionni 209
fordi 209
inermis 209
isolepis 209
maculatus 209
microlepidotus 209
minimus 209
minor 209
muricatus 208, 209
nobbi 209
ornatus 209
pictus 209
reticulatus 209
rufescens 209

lil

718 Advances in Herpetology and Evolutionary Biology

scutulatus 209
Aphaniotis 203
Aphaniotis acutirostris 209
fusca 209
Aporoscelis 208
Brachysaura minor 209
Bronchocela cristatellus 209
jubatus 209
marmoratus 209
Caimanops amphiboluroides 210
Calotes calotes 210
elliotti 210
emma 210
jerdoni 210
maria 210
mystaceus 210
nigrilabris 210
rouxii 210
versicolor 210
Ceratophora stoddartii 210
tennentii 210
Chelosania brunnea 210
Chlamydosaurus 208
Chlamydosaurus kingii 200, 210
Cophotis ceylanica 210
sumatrana 203, 210
Coryophophylax subcristatus 210
Dendragama boulengeri 210
Diporiphora australia 210
bilineata 210
winneckei 210
Draco blanfordii 210
fimbriatus 210
formosus 210
lineatus beccarii 210
maculatus 210
maximus 210
quinquefasciatus 210
spilopterus 210
taeniopterus 210
volans 210
Gonocephalus bellii 210
beyschlagi 210
chamaeleontinus 210
godeffroyi 438
grandis 210
interruptus 210
liogaster 210
miotympanum 210
robinsonii 210
semperi 210
sophiae 210
Harpesaurus 203
Hydrosaurus 195, 204
Hydrosaurus amboinensis 210
pustulatus 204, 210
weberi 210
Hylagama borneensis 210
Hypsilurus (Arua) auritus 210
modestus 210
(Hypsilurus) binotatus 210
boydii 210

bruijni 210
dilophus 210
godeffroyi 210
nigrigularis 210
papuensis 210
spinipes 210
Japalura brevipes 210
dymondi 210
flaviceps 210
planidorsata 210
polygonata 210
splendida 210
swinhonis 210
tricarinata 210
variegata 210
yunnanensis 210
Lophocalotes ludekingi 210
Lophognathus 208
Lophognathus gilberti 210
longirostris 210
temporalis 210
Lyriocephalus scutatus 210
Mictopholis austeniana 203
Mimeosaurus 368, 379
Moloch horridus 210
Oriocalotes paulus 210
Otocryptis wiegmanni 210
Phoxophrys borneensis 210
nigrilabris 210
Phrynocephalus 203
Phrynocephalus arabicus 210
axillaris 210
clarkorum 210
forsythii 210
guttatus 210
helioscopus 210
interscapularis 210
luteoguttatus 210
maculatus 210
mystaceus 210
ornatus 210
przewalskii 210
reticulatus 210
rossikowi 210
scutellatus 210
versicolor 210
vlangalii 210
Physignathus 195, 203

Physignathus cocincinus 204, 206, 210

gilberti 208
lesueurii 201, 210
longirostris 208
temporalis 208
Psammophilus blanfordanus 210
dorsalis 210
Pseudocalotes 203
Pseudocalotes floweri 210
tympanistriga 210
Ptyctolaemus gularis 210
Salea 203
Salea anamallayana 210
horsfieldii 210

kakhiensis 210
Sitana 337
Sitana ponticeriana 210
Stellio 208, 368
Stellio agrorensis 210
atricollis 210
caucasica 210
cyanogaster 210
erythrogaster 210
himalayana 210
lehmanni 210
melanura 210
nupta 210
stellio 210
stoliczkana 210
tarimensis 210
tuberculata 210
Thaumatorhynchus brooksi 203
Tinosaurus 368, 379
Trapelus (Pseudotrapelus) sinaita 210
(Trapelus) agilis 210
flavimaculata 210
impalearis 210
mutabilis 210
persicus 210
rubrigularis 210
ruderata 210
savignii 210
Tympanocryptis lineata 210
tetraporophora 210
Xenagama batillifera 208, 210
taylori 210
FAMILY AIGIALOSAURIDAE
Proaigialosaurus 386
FAMILY AMPHISBAENIDAE
Amphisbaena fenestrata 111
FAMILY ANGUIDAE
Anguidae 371, 389
Anguis 371
Anguis fragilis 201
Anniella 371
Barisia 5
Eoglyptosaurus 371
Gerrhonotus 487
Gerrhonotus multicarinatus 222
Glyptosaurus 371
Helodermoides 371
Melanosaurus 371, 387
Odaxosaurus 371
Ophisaurus 371
Ophisaurus apodus 201
Paraglyptosaurus 371
Peltosaurus 371
Placosaurus 371, 387
Xestops 371, 387
FAMILY ANGUOIDEA
Ilaerdesaurus 376
FAMILY ARDEOSAURIDAE
Ardeosauridae 376
Ardeosaurus 368
Eichstaettisaurus 368
Yabeinosaurus 368

SYSTEMATIC INDEX

FAMILY ARRETOSAURIDAE
Arretosaurus 368
FAMILY BAVARISAURIDAE
Bavarisauridae 376
Bavarisaurus 368
Palaeolacerta 368
FAMILY CHAMAELEONIDAE
Bradypodion 203
Bradypodion pumilus 209

ventralis 209
Brookesia 203
Brookesia stumpffi 209

superciliaris 209
Chamaeleo 201, 203, 368
Chamaeleo adolfifriederici 354

africanus 209

bitaeniatus 209

bitaeniatus ellioti 354

brevicornis 209

carpenteri 354

chamaeleon 209

fallax 209

gracilis 209

jacksonii 209

johnstoni 354

lateralis 209

namaquensis 209

oustaleti 209

owenti 209

parson 200

rudis 354

rudis schoutedeni 355

tigris 209

verrucosus 209

xenorhinus 354
Rhampholeon 203
Rhampholeon boulengeri 209

spectrum 209
FAMILY CORDYLIDAE
Cordylidae 386
Cordylus giganteus 196
Gerrhosaurus 370
Pseudolacerta 370
FAMILY DIBAMIDAE
Dibamidae 382
FAMILY DORSETISAURIDAE
Dorsetisauridae 370
FAMILY EUPOSAURIDAE
Euposauridae 368, 379
FAMILY GEKKONIDAE
Cnemaspis quattuorseriatus 354
Cyrtodactylus 276
Gecko japonicus 200
Gekko 230
Gekkonidae 369, 381
Hemidactylus mabouia 100, 111, 115
Hoplodactylus duvauceli 618

maculatus 618
Platydactylus japonicus 200
Sphaerodactylus altavelensis 86

altavelensis altavelensis 89

brevirostratus 93

719

720 Advances in Herpetology and Evolutionary Biology

enriquilloensis 91
lucioi, ssp. nov. 94
asterulus 95, 97
brevirostratus 89
brevirostratus brevirostratus 93
enriquilloensis 91
cinereus 95, 97
cryphius 86
difficilis 86, 96, 97
difficilis randi 89
elegans 97
macrolepis 110
macrolepis macrolepis 115
notatus 86
nycteropus 86
ocoae 86
parthenopion 111
randi 86, 89
shrevei 97
williamsi, sp. nov. 96
zygaena 86
Thecadactylus rapicaudus 111
FAMILY GYMNOPHTHALMIDAE
Gymnophthalmidae 369, 383
FAMILY HELODERMATIDAE
Eurheloderma 371, 389
Heloderma 371, 375, 389
Heloderma suspectum 200
Paraderma 371, 389
FAMILY IGUANIDAE
Aciprion 368, 378
Agama marmorata 119
Amblyrhynchus cristatus 209
Anisolepis grillii 127
undulatus 120, 209
Anolis 5, 33, 203, 245, 284, 363, 365, 406, 472, 499,
507, 514, 626, 654, 672, 691
Anolis acutus 480, 629, 638
aeneus 451
aequatorialis 259, 270
aliniger 511
allisoni 478
allogus 479
angusticeps 209, 288, 482
annectens 274, 277
auratus 257, 274, 278, 456, 466
baleatus 285, 288
barkeri 259, 276
bimaculatus 209, 288, 505, 627, 635
biporcatus 285
boettgeri 406
bombiceps 406
brevirostris 552
carolinensis 209, 288, 456, 482, 491, 627, 638, 656
chlorocyanus 511
christophei 511
chrysolepis 247, 257, 274, 276, 285, 288
chrysolepis planiceps 209, 295
scypheus 407
cochranae 511
coelestinus 209, 288, 511, 513
conspersus 624

cooki 479, 626, 637

cristatellus 102, 109, 209, 273, 473, 484, 504, 626,
629

cristatellus cristatellus 107
wileyae 101, 107, 115

cupreus 451

cuvieri 101, 246, 252, 273, 626, 637

cybotes 280, 288, 478, 511, 513, 552, 627, 637

desechensis 108, 109 i

dissimilis 407

distichus 294, 482, 511, 513, 552, 627, 635

distichus dominicensis 553, 563

equestris 209, 288, 515

ernestwilliamsi, sp. nov. 102

etheridgei 511

eugenegrahami 276

evermanni 273, 474, 627, 629

fuscoauratus 209, 285, 288

fuscoauratus fuscoauratus 407

gadovi 629, 638

garmani 285, 571, 624

gingivinus 499

grahami 209, 482, 571, 621, 629, 638

grahami aquarum 621
grahami 621

gundlachi 474, 626, 637

hendersoni 294, 511, 513

homolechis 479

koopmani 294, 512

krugi 627, 629

laevis 407

latifrons 209, 288

leachi 515

limifrons 441, 455, 511, 558, 566

lineatopus 571, 624

lucius 479

macrolepis 276

marcanoi 552

marmoratus 478

meridionalis 127

monensis 108, 109, 481, 628, 637

monticola 294, 511, 513, 624

nebulosus 285, 288

occultus 247, 252, 257, 274, 626, 637

oculatus 473, 478, 629, 637

onca 209, 274, 276, 285, 288

opalinus 559, 577, 624

ophiolepis 288

ortoni 407

oxylophus 276

pentaprion 273

petersi 209

poncensis 627, 629

porcatus 504

pulchellus 110, 262, 273, 627, 629

punctatus 627

punctatus boulengeri 407
punctatus 407

reconditus 572

roosevelti 627

roquet 474

roquet salinei 480

zebrilus 480
sagrei 209, 288, 478, 491
scriptus 108, 628, 637
semilineatus 294, 511, 513
sheplani 247, 252, 257, 274
shrevei 479
sp. n. near eulaemus 257, 265, 274, 280
stratulus 115, 273, 473, 627, 629
trachyderma 408
transversalis 408
tropidonotus 262, 273
tropidonotus spilorhipis 249
tropidonotus 249
valencienni 209, 247, 252, 257, 269, 274, 288, 515,
570
wattsi 499
whitemani 552
Aperopristis catamarcensis 209
Aptycholaemus longicauda 209
Audantia 284
Basiliscus basiliscus 209
galeritus 209
Brachylophus 110
Brachylophus fasciatus 209, 563
Callisaurus 654, 658
Callisaurus (Cophosaurus) texanus scitulus 402
draconoides 209
Chalarodon 202
Chalarodon madagascariensis 209
Chamaeleolis 245, 284, 288
Chamaeleolis chamaeleontides 209, 273
porcus 209, 273, 295
Chamaelinorops 245, 284
Chamaelinorops barbouri 209, 248, 285
Conolophus subcristatus 209
Cophosaurus 403
Cophosaurus texanus 209
Corythophanes cristatus 209
Crotaphytus 377
Crotaphytus collaris 209, 656
Ctenoblepharis 203
Ctenoblepharis adspersus 209
Ctenosaura pectinata 209
Cyclura 110
Cyclura cornuta stejnegeri 209
Deiroptyx 284
Diaphoranolis 284
Diplolaemus bibroni 209, 273
Dipsosaurus 368, 487
Dipsosaurus dorsalis 209, 656
Ecphymotes acutirostris 120
pictus 120
plica 120
undulatus 120
Enyaliodes 203
Enyaliodes heterolepis 209
laticeps festae 209
laticeps 209
oshaughnessyi 209
palpebralis 209
Enyaliosaurus quinquecarinatus 209
Enyalius 280

SYSTEMATIC INDEX

Enyalius bilineatus 209
brasiliensis 209
iheringii 209
pictus 120
Gambelia wislizenni 209
Geiseltaliellus 368, 379
Holbrookia 403, 654, 658
Holbrookia maculata 209
texana 656
Hoplocercus 203
Hoplocercus spinosus 209
Iguana 110, 367
Iguana iguana 101, 200, 209
pinguis 111
tuberculata 200
Lacerta marmorata 118, 122
Laemanctus longipes 249
serratus 209, 249
Leiocephalus 368, 379
Leiocephalus carinatus 209, 491
vinculum altavelensis 89
Leiolepis auduboni 120, 122
Leiosaurus belli 209
Liolaemus 203, 658, 672
Liolaemus alticolor walkeri 209
Mariguana 284
Morunasaurus 203, 259
Morunasaurus annularis 209, 249
groi 209, 249
Norops 284
Ophryoessoides ornatus 209
Oplurus 202
Oplurus cyclurus 209
fierinensis 209
grandidieri 209
quadrimaculatus 209
saxicola 209
Paradipsosaurus 368, 378
Parasauromalus 367, 378
Pelusaurus 203
Petrosaurus 654, 658
Petrosaurus mearnsi 209
Phenacosaurus 245, 274, 284, 288, 293
Phenacosaurus heterodermus 269, 295
richteri 209
Phrynosoma 654, 658
Phrynosoma cornutum 209, 656
coronatum 200
Phymaturus palluma 209
Platynotus semitaeniatus 209
Plica plica 120, 209
Polychroides peruvianus 121
Polychrus acutirostris 118, 209
angustirostris 121
anomalus 120, 122
fasciatus 120, 122
femoralis 121, 128
geometricus 120
gutturosus 120
liogaster 121
marmoratus 118
marmoratus acutirostris 121

721

722 Advances in Herpetology and Evolutionary Biology

femoralis 121
liogaster 121
marmoratus 121
spurrelli 121
neovidanus 120, 122
peruvianus 121, 128, 209
spurrelli 121
strigiventris 120, 122
virescens 119, 122
Pristidactylus scapulatus 209
Pristiguana 367, 377
Pristodactylus achalensis 273
Proctotretus 203
Proctotretus azureus 209
pectinatus 209
Psilocercus 119
Psilocercus marmoratus 122
Sator 654, 658
Sator angustus 209, 656
Sauromalus 368
Sauromalus obesus 209
Sceloporus 5, 203, 230, 487, 654, 656, 664
Sceloporus formosus 209
graciosus 209, 656
grammicus 666, 668, 671
grammicus disparilis 209
horridus albiventris 209
occidentalis 457, 656
olivaceus 656
poinsetti 209
scalaris 209
undulatus 209, 656
undulatus consobrinus 400
uniformis 209
variabilis 209
Sphaerops anomalus 120, 122
Stenocercus guentheri 209
Streptosaurus 655
Strobilurus torquatus 209
Tropidodactylus 284
Tropidurus 128
Tropidurus peruvianus 209
torquatus 456
Uma 404, 654, 658
Uma notata 209
Uracentron flaviceps 209
Uranoscodon superciliaris 209
Urosaurus 404, 654, 658
Urosaurus bicarinatus 209
ornatus 656
Urostrophus torquatus 209
vautieri 209
Uta 654, 658
Uta stansburiana 209
Vilcunia 203
Xiphocercus 284
FAMILY LACERTIDAE
Algyroides vauereselli 354
Cabrita 218
Durotrigia 373, 382
Eolacerta 370
Lacerta 196, 230, 370

Lacerta agilis 150, 196, 199
jacksoni 354
lepida 196
viridis 196, 201
Ophisops 218
Plesiolacerta 369
Podarcis muralis 196, 201
FAMILY NECROSAURIDAE
Colpodontosaurus 371
Eosaniwa 371
Necrosauridae 389
Necrosaurus 371
Parasaniwa 371, 389
Provaranosaurus 371
FAMILY PARAMACELLODIDAE
Paramacellodidae 376
Paramacellodus 370
Saurillodon 370
FAMILY PYGOPODIIDAE
Pygopodiidae 213, 381
FAMILY SCINCIDAE
Ablepharus 218
Afroablepharus 217
Carlia amax 214
novaeguineae 216
Chalcides chalcides 201
Contogenys 370, 385
Corucia zebrata 435
Cryptoblepharus 215
Cryptoblepharus plagiocephalus 214
Dasia smaragdinum 360
Emoia arnoensis 360
boettgeri 360
boettgeri orientalis 360
cyanura 216, 360
Eugongylus albofasciolatus 359
Eumeces 230, 370, 384, 386, 487, 654
Lamprolepis smaragdinum 360
Leiolopisma greeni 219
pretiosum 219
suteri 618
trilineatum 215
Leptosiaphos blochmanni 354
graueri 354
hackarsi 354
meleagris 354
Lerista 215
Lerista elegans 215
Lipinia noctua 359
Mabuya megalura 354
sloanei 110, 115
Macroscincus coctei 201
Menetia 217
Morethia 215
Neoseps 386
Notoscincus 216
Panaspis 217, 354
Panaspis cabindae 216
Prasinohaema virens 275, 276
Proablepharus 215
Sauriscus 370, 385
Scelotes 201, 218

Scincella 5, 218, 386
Scincidae 384
Sphenomorphus tympanum 214
Tiliqua 198, 201, 437
Tiliqua scincoides 196
Typhlacontias 218
FAMILY TEIIDAE
Ameiva ameiva 197
exsul 104, 115
Anotosaura 218
Bachia 218
Callopistes 369
Chamops 369
Cnemidophorus 369, 487, 654
Cnemidophorus tigris 624
Gymnophthalmus 218
Heterodactylus 218
Iphisa 218
Meniscognathus 369
Micrablepharus 218
Peneteius 369
Polyglyphanodon 369
Teiidae 383
Tretioscincus 218
Tupinambis 369
FAMILY UROMASTYCIDAE
Leiolepis 195, 380
Leiolepis belliana 200, 202, 206
belliana belliana 210
ocellata 210
guttata 210
peguensis 210
reevesti reevesii 210
rubritaeniata 210
triploida 210
Uromastyx 195, 368, 380
Uromastyx acanthinurus 205
acanthinurus acanthinurus 210
werneri 210
aegyptius 201, 210
asmussi 204, 205, 210
benti 210
dispar 203
geyri 210
hardwickii 201, 205, 210
loricatus 204, 205, 210
macfadyeni 210
microlepis 210
ocellatus 210
ornatus 206, 210
philbyi 210
princeps 208
princeps princeps 210
scortecci 210
spinipes 201
thomasi 210
FAMILY VARANIDAE
Iberovaranus 371
Indovaranus 535
Lanthanotus 371
Palaeosaniwa 371
Saniwa 371, 379, 390

SYSTEMATIC INDEX

Varanidae 390, 535
Varanus 226, 371
Varanus bengalensis 535, 544

komodoensis 535, 537, 550

mertensi 537

niloticus 537

salcadori 537

salvator 537

spenceri 537

varius 537
FAMILY XANTUSIIDAE
Lepidophyma 5, 370
Palaeoxantusia 370
Xantusiidae 207, 386
FAMILY XENOSAURIDAE
Shinisaurus 370, 386
Xenosaurus 386
SUBORDER OPHIDIA
FAMILY BOIDAE
Candoia carinata 440
Chondropython viridis 437
Epicrates monensis 104
FAMILY COLUBRIDAE
Alsophis 110
Arrhyton 110
Conophis 296
Dasypeltis ater 354

Dipsadoboa unicolor viridiventris 355

Dromicus 110
Duberria lutrix 352
Gonyosoma 303
Heterodon 296

Leimadophis epinephelus ecuadorensis 138

Leptodeira 5
Liophis 110
Liophis alticolus 132
bimaculatus 132
bimaculatus lamonae 132
bipraeocularis 132
cobella alticolus 132
epinephelus 132
epinephelus albiventris 138
bimaculatus 138
ecuadorensis 132
epinephelus 137
fraseri 138
juvenalis 139
lamonae 139
opisthotaenia 139
pseudocobella 139
exiguus 104
fraseri 132
opisthotaenia 132
portoricensis 104, 112
pseudocobella 132
reginae albiventris 132
quadrilineata 132
taeniurus juvenalis 132
williamsi 146
Natrix natrix 304
sipedon 691

723

a7 £ )

Opheomorphus alticolus 138
Philothamnus heterodermus ruandae 354
Pituophis melanoleucus 489
Psammophylax variabilis 354
Synophis bicolor 294
Thamnophis sirtalis 296
sirtalis parietalis 296
Zamensis ater 132
FAMILY CROTALIDAE
Crotalus 128
FAMILY TYPHLOPIDAE
Typhlops angolensis irsaci 354
polylepis 354
richardi 104, 112
FAMILY VIPERIDAE
Atheris nitschei nitschei 354
Causus 304
Cerastes cornutus 304

SUBCLASS ARCHOSAURIA
ORDER THECODONTIA

aetosaurs 322

parasuchians 322

phytosaurs 322
proterosuchians 322
Stagonolepis 322
Stegomosuchus 323
ORDER CROCODILIA
Crocodilia 186, 305, 332

SUBORDER PROTOSUCHIA
Hemiprotosuchus 333

Orthosuchus 322

Protosuchus 322, 332

Protosuchus richardsoni 308, 310, 333

SUBORDER MESOSUCHIA
Bernissartia 323

mesosuchians 323

Teleosaurus 323

SUBORDER THALATTOSUCHIA
thalattosuchians 323

SUBORDER EUSUCHIA
FAMILY CROCODYLIDAE
Alligator 306, 333
Alligator mississippiensis 306, 320, 333
sinensis 319
Caiman 333, 335
Caiman crocodilus 306, 321, 695, 700, 702
crocodilus yacare 700, 701
latirostris 320, 701
Crocodylus 306, 333, 335
Crocodylus acutus 318, 699
cataphractus 314
intermedius 319, 695, 697
johnstoni 314
mindorensis 317
moreletii 318
niloticus 315, 697
novaeguineae 317
palustris 315
porosus 316, 440, 697
rhombifer 318

12 Advances in Herpetology and Evolutionary Biology

siamensis 316
Gavialis 306, 335
Gavialis gangeticus 313
Hassiacosuchus 323
Leidyosuchus 323
Melanosuchus niger 321, 695
Osteolaemus 306
Osteolaemus tetraspis 319

Paleosuchus palpebrosus 318, 321, 695

trigonatus 318, 321, 695
Procaimanoidea 323
Tomistoma 306
Tomistoma schlegelii 313
ORDER SAURISCHIA

Plateosaurus 191

ORDER ORNITHISCHIA

ornithischians 322
SUBCLASS SYNAPSIDA
ORDER THERAPSIDA
Glossopteris 379

Lystrosaurus 379
therapsids 223

REPTILIA INCERTAE SEDIS

Protolacerta 373

CLASS AVES
Apterygidae 186
Anatidae 186
Anthracothorax dominicus 509
birds 186, 223, 412, 507
Bucconidae 509
Burhinidae 186

Cariama 128

Coccyzus 517

Coereba flaveola 509, 511
Cracidae 186

Cuculidae 510
Dendrocolaptidae 509

Dendroica caerulescens 511, 514

coronata 511

discolor 511

palmarum 511, 514

petechia 511

tigrina 511
Dromaeus 186
ducks 517
Falco sparverius 110
Formicariidae 509, 514
Furnariidae 509
Geothlypis trichis 511
hoatzin 704
honeycreepers 456
ibis 704
Icterus dominicensis 509, 511
Mniotilta varia 511
Momotidae 509
Otididae 186
Pachyptila turtur 615
Parula americana 511
parulids 510, 512, 516
Pelecanoides urinatrix 615

Phaenicophilus palmarum 511

Phaethon aethureus 110

Phoenicopteridae 186

Rheidae 186

Saurothera 517

Seiurus aurocapillus 511
novaeborealis 511

Setophaga ruticilla 511

Struthionidae 186

Sturnella magna 412

suboscines 513

Thamnophilus punctatus 510

Threskiomithidae 186

Tinamidae 186

Todus 512

Todus angustirostris 509
subulatus 509, 511

Troglodytidae 514

Tyrannus 517

Vireo 516

Vireo altiloquus 509, 511

wrens 513

CLASS MAMMALIA

Alouatta seniculus 522

Artibeus harti 414

Ateles paniscus paniscus 522

bats 412

capybara 702

Cebus apella apella 522
nigrivittatus 522

Chiropotes satanas chiropotes 522

SYSTEMATIC INDEX

deer 456

Felis catus 112

Gorilla beringei 353

Lasiurus borealis borealis 413
cinereus 413
seminolus 413

Leptonycteris nivalis 413
sanborni 413
yerbabuenae 413

Lonchophylla robusta 414

monkeys 521

moose 456

Mus musculus 671

Myotis grisescens 413
keaysi 414
oxyotus 414

Pithecia pithecia 522

possum 438

Rattus exulans 616

rhesus monkey 208

rodents 517

Saguinus midas midas 522

Saimiri sciureus 522

Sturnira bidens 414
bogotensis 414
erythromos 414
ludovici 414

Tadarida brasiliensis 413

Vampyressa melissa 414

Vampyrops dorsalis 414
vittatus 414

725

ne OO ig 7 at
: e+ yher Shh PES
2 weet ny 4
oe ee
=) _ . Peon
bad Ag rr -

. 7

@

Se UL Mt

SHITE: