Problem Snake Management: The Habu and the Brown Treesnake 9781501737688

Citation preview

PROBLEM SNAKE MANAGEMENT

'

PROBLEM SNAKE MANAGEMENT Zr>

The Habu and the Brown Treesnake Editors Gordon H. Rodda

USGS Biological Resources Division Fort Collins, Colorado Yoshio Sawai

Director Emeritus The Japan Snake Institute Nitta-gun, Gunma Prefecture David Chiszar

Department of Psychology The University of Colorado Boulder, Colorado Hiroshi Tanaka

Professor Emeritus Institute of Medical Science The University of Tokyo Minato-ku, Tokyo

Comstock Publishing Associates

a division of Cornell University Press Ithaca and London

Copyright © 1999 by Cornell University Chapter 34, “Candidate repellents, oral and dermal toxicants, and fumigants for Brown Treesnake control,” is not subject to copyright. All rights reserved. Except for brief quotations in a review, this book, or parts thereof, must not be reproduced in any form without permission in writing from the publisher. For information, address Cornell University Press, Sage House, 512 East State Street, Ithaca, New York 14850. First published 1999 by Cornell University Press. Printed in the United States of America. Library of Congress Cataloging-in-Publication Data Problem snake management: the habu and the brown treesnake / Gordon H. Rodda ... [et al.] editors, p.

cm.

Includes bibliographical references (p.

) and index.

ISBN 0-8014-3507-2 (alk. paper) 1. Trimeresurus flavoviridis—Islands of the Pacific.

2. Brown

tree snake—Islands of the Pacific.

3. Trimeresurus flavoviridus

—Control—Islands of the Pacific.

4. Brown tree snake—Control

—Islands of the Pacific. SF810.7.T79P76

I. Rodda, Gordon H.

1998

639.97796—DC21

98-15664

Cornell University Press strives to use environmentally responsible suppliers and materials to the fullest extent possible in the publishing of its books. Such materials include vegetable-based, low-VOC inks and acid-free papers that are also either recycled, totally chlorine-free, or partly composed of nonwood fibers. Cloth printing

10

987654321

Produced with the asssistance of The M. M. Schmidt Foundation

Edmonds, Washington The Japan Snake Institute

Nitta-gun, Gunma Prefecture

'

'

CONTENTS Contributors

xi

Acknowledgments

xiv

Foreword: Living with snakes

xv

Harry W. Greene

Introduction Snake management 1 Gordon H. Rodda, Yoshio Sawai, David Chiszar, and Hiroshi Tanaka

Part I.

BASIC BIOLOGY

25

1. The biology of the Habu (Trimeresurus flavoviridis)

29

Shogi Mishima, Hiroshi Tanaka, and Yoshio Sawai 2. An overview of the biology of the Brown Treesnake (Boiga irregularis), a costly introduced pest on Pacific Islands

44

Gordon H. Rodda, Thomas H. Fritts, Michael J. McCoid, and Earl W. Campbell III 3. Seasonal changes of spermatogenesis and ultrastructural changes of spermatids during spermiogenesis in the Habu (Trimeresurus flavoviridis)

81

Masamichi Kurohmaru, Shosaku Hattori, AND YoSHIHIRO HAYASHI

4. Snakes on electrical transmission lines: Patterns, causes, and strategies for reducing electrical outages due to snakes

89

Thomas H. Fritts and David Chiszar

Part II.

VENOM AND HUMAN HEALTH

105

5. A historical outlook on studies of Habu (Trimeresurus flavoviridis) bites in the Amami and Okinawa Islands of Japan 107 Yoshio Sawai, Yoshiharu Kawamura, Yasutetsu Araki, AND YaSUHIRO ToMIHARA

6. The threat to humans from snakebite by snakes of the genus Boiga based on data from Guam and other areas

116

Thomas H. Fritts and Michael J. McCoid 7. Venom delivery by the Brown Treesnake (Boiga irregularis) and the Habu (Trimeresurus flavoviridis)

128

Kenneth V. Kardong 8. Factors affecting annual incidence of Habu bites, and how residents develop and transfer cognition of high-risk sites 139 Hiroshi Tanaka, Yoshihiro Hayashi, and Satoshi Nakamura VII

viii

Contents Part III.

BEHAVIORAL AND SENSORY BIOLOGY 9. Histology of the Habu’s sensory organs

147 149

Kazushige Hirosawa and Shigeru Takami

10. Repellents and use of prey items for delivering toxicants for control of Habu (Trimeresurus flavoviridis)

158

Masahiko Nishimura

11. Collection and analysis of airborne rat odors

168

Takashi Niwa, Shosaku Hattori, Hiroshi Kihara, Toshiya Sato, and Sizuaki Murata

12. Predatory behavior of Brown Treesnakes (Boiga irregularis): Laboratory studies of chemical attractants

187

David Chiszar, Thomas M. Dunn, and Hobart M. Smith

13. Integrated pest management: The case for pheromonal control of Habu (Trimeresurus flavoviridis) and Brown Treesnakes (Boiga irregularis)

196

Robert T. Mason

Part IV.

POPULATION BIOLOGY

207

14. Dispersal of snakes to extralimital islands: Incidents of the Brown Treesnake (Boiga irregularis) dispersing to islands in ships and aircraft 209 Thomas H. Fritts, Michael J.

McCoid,

and Douglas M. Gomez

15. Movements of Habu, as observed by radio tracking in the field

224

Hiroshi Tanaka, Yoshitake Wada, Yoshihiro Hayashi, AND KENJI IKEDA

16. Population density of Habu on the Amami Islands, as estimated by removal methods

230

Hiroshi Tanaka, Yoshihiro Hayashi, and Yoshitake Wada

17. Population trends and limiting factors in Boiga irregularis Gordon

H.

Rodda, Michael

and Earl W. Campbell

Part V.

J.

McCoid, Thomas

H.

Fritts,

III

CAPTURE AND DETECTION

255

18. Development of the box trap for Habu

257

Shosaku Hattori

19. Trap capture of Habu (Trimeresurus flavoviridis) with odor extracted from rats

264

Shosaku Hattori, Yoshihisa Noboru, Hiroshi Kihara, and Yoshihiro Hayashi

236

Contents 20. A state-of-the-art trap for the Brown Treesnake H.

Gordon Steve

W.

Rodda, Thomas

H.

Fritts, Craig

268

S.

Clark,

Gotte, and David Chiszar

Appendix. The effectiveness of snake traps worldwide Gordon

H.

285

Rodda and Masahiko Nishimura

21. Barriers to movements of the Brown Treesnake (Boiga irregularis) Earl

W.

Campbell

306 III

22. Structure and application of the slanting nylon-net fence to prevent dispersal of Habu (Trimeresurus flavoviridis)

313

Masahiko Nishimura

23. Development of electric fence barriers for Habu (Trimeresurus flavoviridis) in the Amami Islands

319

Yoshihiro Hayashi, Yoshio Sawai, Hiroshi Tanaka, AND SHOGI MlSHIMA

24. Complete removal of Habu (Trimeresurus flavoviridis) from a residential area by trapping

327

Hitoshi Shiroma and Hiroyuki Akamine

25. A ten-year trapping program to eradicate Habu (Trimeresurus flavoviridis) from Minnajima, a small island in the Okinawa Islands, lapan

340

Seiki Katsuren, Chokei Yoshida, and Masahiko Nishimura

26. Training a dog to detect Habu (Trimeresurus flavoviridis)

348

Hitoshi Shiroma and Hiromi Ukuta

27. A preliminary examination of public policy issues in the use of canine detection of Brown Treesnakes

353

Carolyn K. Imamura

Part IV.

BIOLOGICAL, ECOLOGICAL, AND CHEMICAL CONTROL

363

28. The effectiveness of habitat modifications for controlling Habu populations on Tokunoshima

365

Shogi Mishima, Yoshihiro Hayashi, Hiroshi Tanaka, and Yoshio Sawai

29. Food habits of feral mongoose (Herpestes sp.) on Amamioshima, lapan

372

Shintaro Abe, Yukari Handa, Yuko Abe, Yoshitaka Takatsuki, and Hideo Nigi

30. The possible use of haemogregarine parasites in biological control of the Brown Treesnake (Boiga irregularis) and the Habu (Trimeresurus flavoviridis)

384

Sam R. Telford Jr.

31. Biological control of Habu with Entamoeba invadens Akira Ishii and Yoshio Sawai

391

ix

x

Contents 32. Environmental risks of biological control of vertebrates Francis

399

G. Howarth

33. New dermal toxicants and methods of application for venomous snakes

411

Michihisa Toriba, Satoshi Senbo, and Yoshiaki Kosuge

34. Candidate repellents, oral and dermal toxicants, and fumigants for Brown Treesnake control Peter

417

J. Savarie and Richard L. Bruggers

35. An integrated management plan for the Brown Treesnake (Boiga irregularis) on Pacific islands 423 Earl W. Campbell III, Gordon H. Rodda, Thomas H. Fritts, and Richard

Part VII.

L. Bruggers

CONSERVATION BIOLOGY

437

36. Introduced amphibians and reptiles of the Ryukyu Archipelago, Japan

439

Hidetoshi Ota

37. Established exotic reptiles and amphibians of the Mariana Islands

453

Michael J. McCoid

38. A method for protecting nests of the Mariana Crow from Brown Treesnake predation 460 Celestino F. Aguon, Robert E. Beck Jr., and Michael

W. Ritter

39. The feasibility of controlling the Brown Treesnake in small plots 468 Gordon H. Rodda, Thomas H. Fritts, and Earl

W. Campbell III

Epilogue: Contributions of Brown Treesnakes and Flabu to science and society, 479 with commentary by D. Chiszar, T. H. Fritts, S. Hattori, S. A. Minton Jr., S. Mishima, M. Nishimura, G. H. Rodda, Y. Sawai, H. Shiroma, and 14. Tanaka

Index

523

CONTRIBUTORS Editors: Gordon H. Rodda

David Chiszar

USGS Biological Resources Division

Department of Psychology

4512 McMurry Avenue,

Campus Box 345, University of Colorado

Fort Collins, CO 80525 USA

Boulder, CO 80309 USA

([email protected])

([email protected])

Yoshio Sawai

Hiroshi Tanaka

Honshino Mansion 1-c

19-9 Izumi-Honcho 2, Komae City,

18-39 Hamacho, Ota City

Tokyo 201 Japan

Gunma 373-0853, Japan

([email protected])

Contributors, in alphabetical order (address correspondence to senior authors; junior authors’ addresses reflect institutional association at time of contribution):

W. Campbell III

Shintaro Abe

Earl

Wild Animal Research Center, Tatsugo, Amami,

National Wildlife Research Center, Hilo Field

Kagoshima 894-01 Japan

Station, P.O. Box 10880, Hilo, HI 96721 USA

Yuko Abe

Wild Animal Research Center, Tatsugo, Kagoshima Prefecture Japan

Craig S. Clark

National Biological Service, Guam USA Thomas

M. Dunn

Department of Psychology, University of Celestino F. Aguon

Division of Aquatic and Wildlife Resources, 192 Dairy Road, Mangilao, GU 96923 USA

Colorado, Boulder, CO 80309 USA Thomas H. Fritts

USGS Biological Resources Division, National Hiroyuki Akamine

Okinawa Public Health Association, Okinawa, Japan

Museum of Natural History, MRC 111, 10th and Constitution NW, Washington, DC 20560 USA

Yasutetsu Araki

Douglas

Habu Study Section, Okinawa Prefectural

Northern Mariana Islands Department of Fish

Institute of Health and Environment, Okinawa,

and Wildlife, Saipan MP USA

M. Gomez

Japan Steve

W. Gotte

Robert E. Beck Jr.

U.S. National Museum of Natural History,

Guam Division of Aquatic and Wildlife

Washington, DC 20560 USA

Resources, Guam USA Harry Richard

L. Bruggers

W. Greene

Museum of Vertebrate Zoology, 3101 Valley Life

Denver Wildlife Research Center, Denver, CO

Sciences Building, University of California,

USA

Berkeley, CA 94720 USA XI

XII

Contributors

Yukari Handa

Amami Animal Hospital, Kagoshima Japan

Masamichi Kurohmaru

Department of Veterinary Anatomy, Faculty of Agriculture, University of Tokyo, Bunkyo-ku,

Shosaku Hattori

Amami Branch Laboratory, Institute of Medical Science, University of Tokyo, 802 Sude, Te-an, Setouchi-cho, Oshima-gun, Kagoshima 894-15 Japan

Tokyo 113 Japan Robert T. Mason

Department of Zoology, Cordley Hall 3029, Oregon State University, Corvallis, OR 97331 USA

Yoshihiro Hayashi

Michael J. McCoid

Department of Veterinary Anatomy, University

Caesar Kleberg Wildlife Research Institute,

of Tokyo, Bunkyo-ku, Tokyo 113 Japan

Campus Box 218, Texas A&M University-

Kazushige Hirosawa

Kingsville, Kingsville, TX 78363 USA

Jr.

Department of Fine Morphology, Institute of

Sherman A. Minton

Medical Science, University of Tokyo, 4-6-1

4840 E. 77th Street,

Shirokanedai, Minato-ku, Tokyo 108 Japan

Indianapolis, IN 46250-2228 USA Shogi Mishima

Francis

G. Howarth

Department of Natural Sciences, Bishop Museum, 1525 Bernice Street, Honolulu, HI

Laboratory of Medical Sciences, Dokkyo University School of Medicine, Mibu, Tochigi 321-02 Japan

96817-0916 USA SlZUAKI MuRATA

Kenji Ikeda

Laboratories of Biophysics and Bioorganic

Institute of Medical Electronics, University of

Chemistry, Nagoya University, Nagoya Japan

Tokyo, Tokyo Japan

Satoshi Nakamura

Carolyn K. Imamura

Department of Bacteriology, University of the

Pacific Basin Development Council, 711

Ryukyus, Okinawa Japan

Kapiolani Boulevard, Suite 1075, Honolulu, HI

Hideo Nigi

96813-5214 USA

Division of Wild Animal Medicine, Nippon Veterinary and Animal Science University,

Akira Ishii

Department of Medical Zoology, Jichi Medical School, Minami-Kawachi Machi, Tochigi-Ken 329-04 Japan

Tokyo Japan Masahiko Nishimura

Habu Study Section, Okinawa Prefectural Institute of Health and Environment, Ozato

Kenneth

V. Kardong

Department of Zoology, Washington State University, Pullman, WA 99164-4236 USA

2003, Okinawa 901-12 Japan Takashi Niwa

Japan Snake Institute, Yabuzuka-honmachi,

Seiki Katsuren

Nitta-gun, Gunnma 379-23 Japan

Habu Study Group, Okinawa Prefectural

Yoshihisa Noboru

Institute of Health and Environment, Ozato

Institute of Medical Science, University of

2003, Ozato, Okinawa 902-12, Japan

Tokyo, Kagoshima Japan

Yoshiharu Kawamura

Hidetoshi Ota

Japan Snake Institute, Gumna Japan

Tropical Biosphere Research Center, University of the Ryukyus, 1 Senbaru, Nishihara-cho,

Hiroshi Kihara

Takara Institute of Bioscience, Shiga Japan

Okinawa 903-01 Japan Michael

W. Ritter

Yoshiaki Kosuge

Guam Division of Aquatic and Wildlife

Sumika Life-Tech, Osaka Japan

Resources, Guam USA

Contributors

xiii

Toshiya Sato

Yoshitaka Takatsuki

Takasago International Corporation, Tokyo

Nankai Nichi-nichi Daily Company, Kagoshima

Japan

Japan

Peter J. Savarie

Sam R. Telford Jr.

USDA/APHIS/WS, National Wildlife Research

Florida Museum of Natural History, University

Center, 1716 Heath Parkway, Fort Collins, CO

of Florida, Gainesville, FL 32611 USA

80524-2719 USA Yasuhiro Tomihara Satoshi Senbo

Habu Study Section, Okinawa Prefectural

Takarazuka Research Center, Sumitomo

Institute of Health and Environment, Okinawa

Chemical Company, Hyogo Japan

Japan

Hitoshi Shiroma

Michihisa Toriba

Chuo Health Center, Yogi 1-3-21, Naha,

Japan Snake Institute, Yabuzuka-honmachi,

Okinawa 9021 Japan

Nitta-gun, Gunma 379-23 Japan

Hobart M. Smith

Department of Environmental, Population, and

Hiromi Ukuta

Arakai Dog Training Center, Okinawa Japan

Organismic Biology, University of Colorado, Boulder, CO 80309 USA Shigeru Takami

Department of Fine Morphology, Institute of

Yoshitake Wada

Department of Parasitology, Tokyo Women’s Medical College, Tokyo Japan

Medical Science, University of Tokyo, Tokyo

Chokei Yoshida

Japan

Koza Public Health Center, Okinawa Japan

ACKNOWLEDGMENTS

A

work of this magnitude requires the efforts of an astonishing number of volunteers. Most are recognized in the chapters to which they contributed,

however, the contributions of several were unknown to the contributing authors. Foremost among these was Cynthia Melcher, whose editorial and organizational skills are both profound and deeply appreciated. Marie Timmerman, Kathy DeanBradley, Atsushi Sakai, and Hajime Moriguchi also assisted with editorial duties. Kathy Dean-Bradley, David Cunnington, Leilani Leach, Gad Perry, Carl Qualls, Fiona Qualls, and Kym Welstead assisted with proofreading. David Cunnington and Kym Welstead prepared the index. Cynthia Melcher, Dale Crawford, Kathy Dean-Bradley, and Debbie Aguilar prepared many of the illustrations. Terry Waddle and Masahiko Nishimura clarified many uncertainties of translation. Dr. Nishimura also corrected numerous factual errors throughout the text. Any he did not find are the sole responsibility of the editors. We greatly appreciate the encouragement and assistance provided by Peter Prescott and Mindy Conner of Cornell University Press. No scientific work is complete without the vital assistance of anonymous reviewers. Among those whose efforts are greatly appreciated but not credited in the corresponding chapters are Michael A. Bogan, Richard L. Bruggers, Earl W. Campbell III, David Duvall, Thomas H. Fritts, Douglas Gomez, David Hardy, Carolyn K. Imamura, Richard Jones, Kenneth Kardong, Steve Mackessy, Robert Mason, Russell Mason, Michael J. McCoid, Cynthia Melcher, Sherman A. Minton Jr., Masahiko Nishimura, Takashi Niwa, Renee J. Rondeau, Hobart Smith, Tim Smock, and Michihisa Toriba. The meeting that initiated this work was made possible by grants from Mr. Soken Oshiro and Mr. Masahide Ota. Their assistance is greatly appreciated by all. Preparation of the meeting was made possible by the efforts of Hiroyuki Akamine, Yasutetsu Araki, Tametsugu Hanasaki, Takao Kamura, Seiki Katsuren, Yoshiharu Kawamura, Tsuyoshi Kinjo, Masahiko Nishimura, Masatoshi Nozaki, Rinshin Shimabukuro, Hitoshi Shiroma, Keitetsu Sunagawa, Dr. Yasuhiro Tomihara, and Masanobu Yamakawa. The products and sources mentioned in this volume are for information only. Neither the authors, their institutions, nor the publisher endorse such products or judge them superior to those of alternate suppliers.

XIV

FOREWORD

Living with Snakes Harry W. Greene

T

he roughly 2700 species of extant serpents range from tiny, parthenogenetic Flowerpot Blindsnakes (Ramphotyphlops braminus) to giant Green Anacondas

(Eunectes murinus), from gentle Rosy Boas (Charina trivirgata) to deadly venomous Taipans (Oxyuranus scutellatus). Snakes inhabit terrestrial, aquatic, arboreal, fossorial, and marine environments, and eat animals as diverse as fish eggs, slugs, and porcupines. In addition to their important roles in natural ecosys¬ tems, serpents interact with humans in diverse ways, some “good” and some “bad,” and that’s where things get murky in terms of conservation. In this brief essay I affirm the necessity for management in a world that is nowhere truly pristine, underscore the many positive roles snakes play in nature and in human affairs, and emphasize some lessons learned from efforts to control snakes on Pacific islands. By the end of this century 6 billion humans will burden the earth, and the dev¬ astating impact of our expanding population on biological diversity is increas¬ ingly evident. In the face of this catastrophe, however, only about 10% of the world’s snake species receives special protection, a fraction that grossly under¬ estimates the number actually threatened with extinction. Beyond the fact that dozens of snakes are known only from type specimens (e.g., 7 species in Sri Lanka alone), the conservation status of most species of serpents has not yet been eval¬ uated. In Europe, where the snake fauna is small and relatively well studied, 19 of 27 species need active management; in contrast, the conservation status of almost all of the hundreds of species of Asian snakes is virtually unknown. Sustained persecution, habitat destruction, and restricted geographic distributions imply that perhaps 60% or more of 157 species of pitvipers are now threatened or endangered, but only about a dozen species in that group are formally singled out for protection. In fact, many snakes may be especially vulnerable to extinction because of their slow growth rates, small clutches, and infrequent reproduction— components of a general life history syndrome that hinges on high adult survivorship (Dodd, 1987, 1993; Greene and Campbell, 1992; Scott and Seigel, 1992). Blatant antisnake programs were routine earlier in this century, when similarly hostile attitudes prevailed even toward large mammalian predators. A book on Pennsylvania snakes written at that time says of the Copperhead (Agkistrodon

contortrix), “there is no creature more treacherous, despicable nor dangerous” xv

xvi

Foreword

(Surface, 1906), and Boy Scouts of the 1930s won special ‘ conservation medals for killing Northern Watersnakes (Nerodia sipedon), in the mistaken view that they were enhancing trout populations (Anon., 1938). Even in the 1970s, visitors at a national park in the United States overwhelmingly supported protection of “all wildlife,” but were in much less agreement about whether or not snakes should be killed in the park (Burghardt et al., 1972). Nevertheless, even granting wide¬ spread negative attitudes toward serpents, the present volume might seem para¬ doxical in the current climate of enlightened, conservation-minded biology. After all, efforts are under way in Europe and North America to conserve various snakes, but the authors of this book are concerned mainly with destroying snakes, or at least controlling their populations. Widespread mythology notwithstanding, we have been managing animal and plant populations for many thousands of years, and humans exterminated many species of vertebrates long before Europeans explored and colonized the globe (Nabhan, 1995; Steadman, 1995). Some snake populations warrant active management and sometimes even local extermination, for two broad classes of reasons. First, habitat destruction and human persecution threaten certain populations and even entire species with extinction, and their rescue may require some kind of intervention. In such cases, management might be contro¬ versial because of widespread antipathy toward snakes (if the merit of saving them is questioned) or because of a “hands-off” attitude toward conservation. Second, a few species of serpents cause problems for natural communities or human welfare. The Brown Treesnake (Boiga irregularis) is the only dramatic example of an artificially introduced snake causing widespread ecological damage, but clearly the potential exists for other such incidents. Perhaps 500 species of colubrids, atractaspidids, elapids, and viperids are potentially dangerous to humans; of those, about 50 species cause most snakebite accidents and fatalities, and today they remain serious public health problems in some tropical regions (e.g., Russell’s Viper, Daboia russelli, is responsible for about 1000 bites per year in Burma, and snakebite is the fifth most important cause of death in that country; see Warrell, 1989, 1991). Programs to conserve snakes and their habitats will not succeed unless these safety issues are successfully addressed, and the studies on Habu (Trimeresurus flavoviridis) summarized in this book exemplify ways in which a serious snakebite problem can be confronted by active, well-informed management. In addition to controlling or even eliminating problem snake populations, con¬ servationists must stress public education and ways we can live with these ani¬ mals. Given the emphasis in this book, I take this opportunity to underscore the many significant roles serpents play in their environments and in our lives. Many snakes eat proportionately very large items, and some serpents are competitors with and even prey on fairly large birds and mammals. Snakes are models for sev¬ eral insect mimicry systems (Pough, 1988), influence the evolution of nesting and

Foreword

xvii

social adaptations in birds (Jackson, 1974; Greene, 1989), and interact in diverse other ways with their ecological assemblages. For those who require utilitarian reasons to justify conservation, nature is essentially an enormous testing laboratory, in which over hundreds of millions of years organisms have developed solutions to countless environmental challenges. Accordingly, a component of pitviper venom, perhaps evolved for immobilizing or digesting prey, now is synthesized as a multimillion-dollar drug for high blood pressure (Ferguson et al., 1984). The shaker muscle of rattlesnake tails has several unique properties related to its natural defensive role and serves as a model system for studying muscle molecular physiology (Conley and Lindstedt, 1996; Schaeffer et al., 1996), in turn a prerequisite for developing cures for some diseases. From an aesthetic perspective, humans’ fascination with snakes is famously ambiguous, reflecting both fear and curiosity—perhaps in part because primates have both eaten and been killed by serpents since long before the origin of hominids (I explore these issues in more detail elsewhere; Greene et al., 1997; and also Kellert and Wilson, 1993). In any case, snakes haunt and inspire us, pro¬ voking outlandish legends, rational fears, and admiration. Even venomous snakes embody scaly symmetry and harmonious colors, and walking in their habitats fosters a special kind of respect for nature, an urgency that both humbles and enlivens us. Snakes offer some special lessons for conservation. First, these animals are especially problematic with respect to human attitudes, as evidenced by the per¬ sistence of rattlesnake roundups and other blatantly inhumane spectacles (Arena et al., 1995). If we can mobilize people to tolerate and even care about snakes, especially venomous species, then conserving vultures and badgers will be easy (for a particularly thorough account of the problems of conserving a venomous snake, see Brown, 1993). Second, management is inevitable if we are to preserve significant portions of the earth’s existing biotas, and management depends on science (Dodd and Seigel, 1991; Reinert, 1991). As repeatedly shown by the case studies in this book, decades of descriptive natural history studies underlie these activities (see also Greene, 1989, 1994). We especially need sustained monitoring of many animal and plant populations, yet support for such activities is widely available only for certain songbirds, raptors, and game species. There are woefully few mechanisms to sustain basic research on biological diversity, and everyone committed to conservation, from lay naturalists to leaders of environmental organizations, should work toward solving this problem. Third, and perhaps most important, the complex plight of snakes recalls other creatures that are danger¬ ous (e.g., sharks [Manire and Gruber, 1990]), secretive (e.g., velvet worms [New, 1995]), unpopular (e.g., centipedes), and relatively poorly studied. These animals remind us that we seriously underestimate the extent to which species and even entire natural communities are endangered—that in fact, a global biodiversity crisis is well under way.

xviii

Foreword

ACKNOWLEDGMENTS H. K. Reinert alerted me to the comment by Surface; D. L. Hardy Sr., G. H. Rodda, and K. R. Zamudio provided helpful comments on the manuscript.

LITERATURE CITED Anon.

1938. For better fishing—kill the watersnake. Pennsylvania Board of Fish Com¬

missioners, Combined Biennial Report [no volume], 71-75. P. C., C.

Arena,

D.

Warwick, and

Duvall.

1995. Rattlesnake round-ups. In R. L.

Knight and K. Gutzwiller, eds., Wildlife and Recreationist: Coexistence through Management and Research, pp. 313-324. Washington, D.C.: Island Press. Brown,

W. S. 1993. Biology, status, and management of the timber rattlesnake (Cro-

talus horridus): A guide for conservation. Herpetol. Circ. 22:1-72. Burghardt,

G. M., R. O.

Hietala, and

M. R.

Pelton.

1972. Knowledge and atti¬

tudes concerning black bears by users of the Great Smokey Mountains National Park. In S. Herrero, ed., Bears—Their Biology and Management, pp. 255-273. Morges, Switzerland: IUCN. Conley,

K. E.,

and

S. L.

Lindstedt.

1996. Minimal cost per twitch in rattlesnake tail

muscle. Nature 383:71-72. Dodd,

C. K. Jr. 1987. Status, conservation, and management. In R. A. Seigel, J. T.

Collins, and S. S. Novak, eds., Snakes: Ecology and Evolutionary Biology, pp. 478-513. New York: Macmillan. -. 1993. Strategies for snake conservation. In R. A. Seigel and J. R. Collins, eds., Snakes: Ecology and Behavior, pp. 478-513. New York: McGraw-Hill. Dodd,

C. K. Jr.,

and

R. A.

Seigel.

1991. Relocation, repatriation, and translocation

of amphibians and reptiles: Are they conservation strategies that work? Herpetologica 47:336-350. Ferguson,

R. K., P. H. Vlasses,

and

L. J.

Riley.

1984. Captopril in the treatment of

hypertension and congestive heart failure. Res. Staff Physician 30(ll):PC24-28. H. W. 1989. Ecological, evolutionary, and conservation implications of

Greene,

feeding biology in Old World cat snakes, genus Boiga (Colubridae). Proc. Calif. Acad. Sci. 46:193-207. -. 1994. Systematics and natural history, foundations for understanding and con¬ serving biodiversity. Am. Zool. 34:48-56. Greene,

H. W.,

and

J. A.

Campbell.

1992. The future of pitvipers. In J. A. Campbell

and E. D. Brodie Jr., eds., Biology of the Pitvipers, pp. 421-427. Tyler, Tex.: Selva. Greene,

H. W., M.

Fogden, and

P.

Fogden.

1997. Snakes: The Evolution of Mystery

in Nature. Berkeley: Univ. California Press. Jackson,

J. A. 1974. Gray Rat Snakes versus Red-cockaded Woodpeckers: Predator-

prey adaptations. Auk 91:342-347. S. R., and E. O. D.C.: Island Press.

Kellert,

Wilson,

eds. 1993. The Biophilia Hypothesis. Washington,

C. A., and S. H. Gruber. 1990. Many sharks may be headed toward extinc¬ tion. Conserv. Biol. 4:10-11.

Manire,

Foreword Nabhan,

xix

G. P. 1995. Cultural parallax in viewing North American habitats. In M.

Soule and G. Lease, eds., Reinventing Nature? Responses to Postmodern Decon¬ structionism, pp. 87—101. Washington, D.C.: Island Press. New, T. R. 1995. Onycophora in invertebrate conservation: Priorities, practice and prospects. Zool. J. Linn. Soc. 114:77-89. Pouch, F. H. 1988. Mimicry and related phenomena. In C. Gans and R. B. Huey, eds., Biology of the Reptilia. Vol. 16, Ecology B, Defense and Life History, pp. 153-234. New York: Alan R. Liss. Reinert, H. K. 1991. Translocation as a conservation strategy for amphibians and reptiles: Some comments, concerns, and observations. Herpetologica 47:357-363. Schaeffer,

P. J., K. E.

Conley, and

S. L.

Lindstedt.

1996. Structural correlates of

speed and endurance in skeletal muscle: The rattlesnake tailshaker muscle. J. Exp. Biol. 199:351-358. Scott, N. J., and R. A.

Seigel.

1992. The management of amphibian and reptile pop¬

ulations: Special priorities and methodological and theoretical constraints. In D. R. McCullough and R. H. Barrett, eds., Wildlife 2001: Populations, pp. 343-367. London: Elsevier. Steadman, D. W. 1995. Prehistoric extinctions of Pacific island birds: Biodiversity meets zooarchaeology. Science 267:1123-1131. Surface, H. A. 1906. The serpents of Pennsylvania. Mon. Bull. Pa. Dept. Agric. 4. Warrell,

D. A. 1989. Snake venoms in science and clinical medicine 1. Russell’s

Viper: Biology, venom and treatment of bites. Trans. Roy. Soc. Trop. Med. Hyg. 83:732-740. _. 1991. Snakes. In G. T. Strickland, ed., Hunter’s Tropical Medicine, pp. 877-888. Philadelphia: W. B. Saunders.

'

PROBLEM SNAKE MANAGEMENT

INTRODUCTION

Snake Management Gordon H. Rodda Yoshio Sawai David Chiszar Hiroshi Tanaka

WHAT IS SNAKE MANAGEMENT? Snake management is any attempt to systematically influence snake-human inter¬ actions. Snakes are usually managed to protect human health or to preserve bio¬ diversity, either of snakes or of their prey. For example, one might try to reduce the abundance of venomous snakes in the vicinity of schools by putting a snake barrier around the school yard, as the Japanese have done on Amamioshima, an island in the Ryukyu Archipelago. Or one might wish to increase the abundance of an endangered snake, as Antiguans have sought to do for the Antiguan Racer (Alsophis antiguae) in the West Indies. Or one might wish to eliminate an intro¬ duced snake that is preying on native wildlife, as the Jersey Wildlife Trust is propos¬ ing for the Wolf Snake (Lycodon aulicus) on the islets around Mauritius in the In¬ dian Ocean. Thus snake management may involve actions to increase or decrease the abundance of snake populations. Snake management excludes the clinical treatment of snakebite but includes public health actions that may be taken to avoid snakebite. Although serving diverse goals, snake management depends on common techniques and uses a common body of knowledge about human and snake ecology and behavior. This introduction gives many examples of the tools of snake management, a few of which are educational campaigns to inform peo¬ ple about behaviors that increase or decrease the risk of snakebite, regulation of the commercial harvest of snakes for leather or the pet trade, modification of ecosystems to increase or decrease the abundance of selected snakes, erection of barriers to discourage the presence of venomous species or the spread of exotic species, and eradication of introduced pests to promote the survival of endangered snakes. These and many other actions taken to modify humans’ relations with snakes constitute snake management, and the chapters in this book provide detailed accounts of selected approaches to snake management.

WHY MANAGE SNAKES? Informal snake management—efforts to avoid venomous snakebite—undoubt¬ edly predates humanity. Various vertebrates have been shown to fear and avoid snakelike forms and the color patterns associated with venomous species (S. M.

1

2

Rodda, Sawai, Chiszar, & Tanaka

Smith, 1975, 1977, 1978). Most human societies systematically disseminate infor¬ mation aimed at teaching people to avoid venomous snakes, the most basic form of snake management (Cummings, 1961). Some societies also spread information on how to avoid or kill nonvenomous snakes (Brock and Howard, 1962; Byford, 1986), although this goal of snake management may be unwise, and many soci¬ eties promote the conservation of snakes that eat rodents and other pests that harm human food supplies (Bogert, 1948; Greene and Campbell, 1992; Shine and Fitzgerald, 1996). Some religious or social taboos regarding the killing of snakes actually protect snakes that eat grain-damaging rodents (Howey, 1955; Morris and Morris, 1965; Mundkur, 1983). Thus the second oldest goal of snake management is to assist the production and storage of food. This includes not only the preser¬ vation of rodent-eating snakes, but also the protection of small prey animals, such as rabbits and chickens, that also provide food for humans. Until the late twentieth century, snake management was exclusively related to human health and agriculture. As the influence of human actions on the bios¬ phere becomes ever more pervasive, however, more subtle human-snake interac¬ tions have become important enough to warrant systematic management. For example, in some areas of the world (e.g., the south-central United States and Guam), arboreal snakes crawl on power lines with such frequency that they create repeated and expensive power outages (see Fritts and Chiszar, this volume, Chap. 4). In Guam in the early 1990s, there was a snake-induced power outage about every third day. As single outages may cause millions of dollars worth of damage and lost productivity, protection of power systems has become a small but influential part of snake management. Because all snakes are predators, some species inevitably prey on animals that are valued by humans for companionship and recreation. In the last few years, the Brown Treesnake on Guam has been managed to reduce predation on puppies, kittens, rabbits, cage birds, and fighting cocks (see Rodda et al., this volume, Chap. 2). Watersnakes at a goldfish hatchery in Missouri have been managed to reduce fish losses (see below). The most devastating predatory impacts of snakes have fallen not on domestic animals, however, but on wild prey that have been inad¬ vertent victims of snakes transported by humans to new environments, especially islands. Because they do not fly or swim great distances in salt water, snakes (excluding seasnakes) are naturally absent from many remote islands, and the small animals that live on such islands generally lack defense mechanisms to pro¬ tect them from snake predation. When a snake is transplanted to such an island, catastrophic loss of native species may occur, as it did on Guam and several other islands (see below). Guam has lost virtually all of its native vertebrates (see Rodda et al., this volume, Chap. 2), including birds, mammals, and lizards. Thus snake management on islands will include efforts to prevent the introduction of snakes and to protect wildlife that might be preyed on by nonnative snakes. But the introduced species problem can work the other way around as well: snakes also may lack defense mechanisms to protect them from predators that do

Introduction

3

not naturally occur where they live. Snake populations around the world have declined as a result of introduced predators, especially cats, rats, and mongoose (Sajdak and Henderson, 1991; Dodd, 1993). Dodd (1987) listed nearly 200 snake species for which conservation concerns have been published, with Caribbean islands identified as the geographic area in most immediate need of protection. Much of the world’s snake management is now directed at protecting snakes from inadvertent losses due to introduced predatory mammals. Many Caribbean members of the genera Alsophis (West Indian racers) and Liophis (also called racers) are threatened by introduced predatory mammals (Sajdak and Hen¬ derson, 1991; Sherriff et al., 1995). The Mona Island Boa (Epicrates monensis) has a very limited distribution around Puerto Rico, leading to several studies of this species at risk (e.g., Tolson, 1988; Chandler and Tolson, 1990; Tolson and Henderson, 1993). Under certain circumstances, snakes may suffer from the introduction of herbivores, which may so alter the vegetation that the habitats for snakes or their prey are adversely affected. The most famous example may be Round Island, in the Indian Ocean (Bullock, 1986). Unlike nearby Mauritius, which gained infamy for the extinction of the Dodo, Round Island is best known for its endemic rep¬ tiles. Not only is Round Island the last refuge of two endemic lizards, it is also the home of two species of snakes found nowhere else in the world. These snakes are so evolutionarily distinct that they have been placed in their own genera (Casarea and Bolyeria) and their own family (Bolyeridae) or subfamily (Bolyerinae), depending on one’s taxonomic persuasion. Among their other unique traits, these two snakes have not only the flexible joint at the meeting of the left and right halves of the jaw possessed by most snakes, but also a hinge halfway along the length of each lower jawbone (mandible), permitting the jaw to encircle prey held crosswise in the mouth. This feature differs from that found in all other higher vertebrates (amniotes). Unfortunately, the habitat of these unique lizard-eating snakes was devastated by the nineteenth-century introduction of goats and rab¬ bits, which reached densities sufficient to denude the island of most vegetation. The forest was eliminated entirely, and most of the island’s soil washed into the sea. No one yet knows the mechanism by which these habitat changes affected the snakes, but evidence of the snakes’ decline helped stimulate an international pro¬ gram to restore the island, which is now a Mauritius Nature Preserve, by eradi¬ cating the introduced rabbits and goats. The introduced mammals are gone now (Merton, 1987), but the recovery may be too late for Bolyeria, which is probably already extinct (Tonge, 1989). At first glance it might seem odd that actions to reduce snake abundance and actions to increase snake abundance are both included in the discipline of snake management. Both, however, depend on the same information and techniques. In practice, all management activities rely on biological information about snakes. The same epidemiological information that is collected about venomous snakes (see Part 2: “Venom and Human Health”) such as the Habu (Trimeresurus

4

Rodda, Sawai, Chiszar, & Tanaka

flavoviridis), which is managed for population reduction, is used to understand the role of envenomation by other snakes, such as the Timber Rattlesnake (Crotalus horridus), which is rare in parts of the northeastern United States and is therefore managed for population increase (Brown, 1993). Similarly, the infor¬ mation that has been collected about the sensory biology of snakes (see Part 3: “Behavioral and Sensory Biology”) is utilized for both concerns. For example, a biologist interested in preserving California’s endangered Giant Gartersnake (Thamnophis gigas; see U.S. Fish and Wildlife Service, 1991) might investigate feeding adaptations to understand the kinds of waterways the gartersnake requires for foraging, whereas a biologist searching for ways to control the Brown Treesnake might use information on foraging preferences to design an artificial lure to draw the snake to traps or toxicant bait stations. The techniques for pop¬ ulation enumeration (see Part 4: “Population Biology”) are similar for both endangered and pestiferous snakes, and so forth. Snake managers are united by their expertise, although they may at times have different tactical goals.

UNDER WHAT CIRCUMSTANCES MIGHT MANAGEMENT BE WARRANTED? Human intervention in the operation of an ecosystem, or “management,” is war¬ ranted in some ecosystems but not in others. Is there a pattern to the imposition of snake management? In principle, snake management might be justified in ecosystems that are functioning naturally—that is, as they evolved. Educational efforts are often directed at preventing snakebite in reasonably intact ecosystems. In the vast majority of cases, however, snake management has been made neces¬ sary by unintentional side effects of human actions, often habitat modifications or species introductions. Urbanization is one conspicuous form of habitat modification. How and Dell (1994) found that snakes were the reptiles most likely to disappear as a result of habitat fragmentation due to urbanization. The San Francisco Gartersnake, Thamnophis sirtalis tetrataenia, had the misfortune to evolve on a narrow penin¬ sula bordering a bay that would one day become some of the most sought-after real estate in the world (U.S. Fish and Wildlife Service, 1985). Most of the wetlands needed by this snake have been drained for urban development. Snake habitat may also be destroyed more intentionally, as in the destruction of rat¬ tlesnake dens (Jackley, 1947; Palmer, 1992; Brown, 1993). Road construction is a very widespread form of habitat destruction. Roads modify microenvironments and transport introduced species into many otherwise intact habitats (Harris and Silva-Lopez, 1992), but their effect on snakes is often more direct: snakes get run over. Although snakes may be tempted to bask, fatally, on warm roadways, many others are killed simply by their inability to cross a roadway fast enough to avoid collision with automobiles (Kaufman and Gibbons, 1975; Dodd et al., 1989; Rosen and Lowe, 1994). For example, R. Franz (in Harris and Silva-Lopez, 1992) found that approximately 13,000 snakes of 12 species, weighing an estimated 1.3 tons,

Introduction

5

were lost on a single 3.25 km stretch of Florida roadway over a 4.5 year period. Southall (1991) provided designs for barriers to divert snakes under roadways. Not all habitat disruption results in the diminution of snake numbers. Anthropogenic habitats are often rich in prey for snakes, especially rodents, lead¬ ing to artificially high numbers of snakes. Where such “subsidized” snake popu¬ lations are composed of dangerously venomous species, humans may choose to manage the snake populations as a way of minimizing health risks. An example is the Habu, a species highlighted in this book. The Habu abounds in sugarcane fields and other disturbed habitats on the Ryukyu Islands of Japan. The sugarcane ecosystem is an exceptionally favorable habitat for black rats (Rattus tanezumi), a food of primary importance for the Habu. Unfortunately for sugarcane workers, this snake has a highly lethal and myonecrotic (muscle-destroying) venom, and there are frequent and regrettable interactions between snakes and humans (see especially Chaps. 5 and 8). The adverse impacts of the introduced Brown Treesnake and the demise of many snakes brought about by introduced mammals are among the direct bio¬ logical costs of species introductions; but indirect effects may also be important. For example, we tend not to think of diseases as introduced species, although many are. Plague (Yersinia pestis) is a bacterium that was introduced into North America around 1900, with devastating consequences for various rodents, espe¬ cially prairie dogs (Oldemeyer et al., 1993). The great diminution of rodent pop¬ ulations has probably had a significant impact on rodent-eating snakes, although we are unaware of any attempts to quantify the population losses. Overharvest of snakes or snake prey for the pet trade is a factor that can potentially force humans to manage snake populations, especially when the species is rare and of great beauty (e.g., San Francisco Gartersnake). Snake species such as the Boa Constrictor are widely sold, not only for pets but also for leather products. To our knowledge, most of the snake leather trade is in species that are widely distributed and therefore potentially sustainable, although there are no data to tell us whether or not current practices occur on a sustainable level (Shine et al., 1995). Incidental or wanton killing of snakes is even more difficult to eval¬ uate, as much of it takes place in private settings. In addition, it is likely that snake populations are being altered unintentionally in undesirable ways by biocides and environmental contaminants (Hall, 1980; Dodd, 1987), although few studies of this phenomenon have been conducted, and none is required by any government, to the best of our knowledge.

DETERMINING WHEN MANAGEMENT IS NEEDED Analysis of the appropriateness of snake management can be considered by first collecting pertinent data and then comparing costs of snake management with prospective benefits; both processes are scientifically underdeveloped. The benefits of snake management have been difficult to quantify because snake man-

6

Rodda, Sawai, Chiszar, & Tanaka

agement techniques are too new and too experimental for extensive evaluation to have occurred. The evaluations that have been done have yielded mostly qualita¬ tive measures (e.g., Engeman et al., 1997a, 1997b). Most significant, there exists no widely accepted “currency” by which financial costs can be compared with nonmonetary benefits such as reduction in the numbers of crippling envenomations or avoidance of species extinctions. As an example, snake management at the Crescent Lake National Wildlife Refuge reduced duck nest losses attributable to Bullsnake (Pituophis melanoleucus) predation by 19.5% (Imler, 1945; see below), but refuge managers did not have a means to compare this result with the costs expended on Bullsnake control. Furthermore, they did not determine if re¬ duced nest losses translated into more ducks for hunting, or if compensatory nat¬ ural sources of mortality reduced the surplus generated through snake control. In principle, it ought to be possible to evaluate prospective snake management actions whose benefits are monetary and narrowly defined, such as power out¬ ages forestalled or agricultural losses avoided through control of pests. For example, Bauman and Metter (1975) estimated that goldfish losses valued at US$305,600 (corrected for inflation to 1996) were preempted by watersnake con¬ trol measures for which they paid US$2625. This is an unusual situation in which the costs and benefits accrued to a single entity (one fish hatchery). More often, control activities are characterized by a large number of beneficiaries (e.g., all power users benefit from uninterrupted power supplies) and a cost accruing exclusively to a single (usually) governmental management authority that has no direct way to recoup costs shared by diffuse beneficiaries. As more funds and energy are devoted to snake management, we expect that improved evaluative metrics will be developed and perfected. A necessary precursor to evaluative formulas is appropriate data collection. For example, one cannot determine the value of public health programs aimed at decreasing snakebites without accurate measures of snakebite frequency. Similarly, one cannot assess snake conservation needs if one has no indicators of wildlife population trajectories. In many parts of the world, accurate statistics on snakebite are not collected. In India, for example, private hospitals do not report visits of snakebite victims. Furthermore, many snakebite victims are not treated at formal facilities. Others travel long distances to urban centers to obtain treatment, implying that the patient pool of a large hospital is an indeterminate region. Thus the total popu¬ lation served by a hospital is unknown, and snakebite frequency (bites per person) cannot be accurately estimated. Another source of frustration to snakebite epidemiologists is that the species of the snake is generally unknown: Sawai (in Whitaker, 1978; see also Sawai and Honma, 1975,1976; Sawai, 1992) reported that the responsible snake was identified in only 6% of the cases receiving treatment in government hospitals in India. In most cases, physicians must treat patients on the basis of the identity of the snake that they presume was responsible, given the suite of symptoms presented. Herpetologists and snake epidemiologists estimate

Introduction

7

symptoms and patterns of snakebite on the basis of the species listed in the hos¬ pital reports. It is thus not surprising that authorities in India and many other parts of the world are still struggling to determine basic epidemiological patterns of snakebite. In Japan, the quest for epidemiological patterns has advanced well beyond the basic tabulation of human bite frequencies for various snake species. As illustrated by the chapters by Sawai et al. and Tanaka et al. in Part 2, Japanese researchers have examined in depth the specific localities where snakebite frequencies are high, the activities being performed when snakebite occurs, profiles of the victims in terms of age, gender, etc., and have even developed sophisticated models of how information about the risk of snakebite is transmitted through social channels. These data are invaluable for analyzing the costs and benefits of snake manage¬ ment. For example, the discovery that many Habu snakebite victims were in the vicinity of stone walls led to the highly cost-effective practice of eliminating snake refugia in masonry walls by plugging holes (Mishima et al., this volume, Chap. 28). On the other hand, the discovery that children were developing their aware¬ ness about sites with high risk of snakebite from other children rather than from their parents (Tanaka et al., this volume, Chap. 8) suggested a solution based on education rather than physical or biological manipulations. In contrast, serious bites of the Brown Treesnake occur primarily to newborns sleeping in their beds (Fritts and McCoid, this volume, Chap. 6), suggesting that physical barriers to keep out the snakes offered more promise for reducing bites than did education of the victims. Thus we see that epidemiological statistics both contribute to and benefit from a vigorous program to manage dangerously venomous snakes. A primary source of frustration for those who design programs to protect the biodiversity of snakes or their prey is the unavailability of population density estimates of the species of interest. It is impossible to know if a species is about to become endangered if no one is taking its “pulse.” Furthermore, to establish a trend in a population, measurements must be accurate and taken over a long enough span of years for seasonal, weather, and stochastic fluctuations to be filtered out (Pechmann et al., 1991). A drought cycle may last many years, requiring population estimates spanning a decade or more for most species. Longer series are needed for long-lived species with naturally low mortality rates. For many reptiles, not even a single rangewide estimate of species abundance is available. Many states in the United States have initiated nongame wildlife population monitoring programs in the last decade, but these are generally limited to highly visible species such as birds. National and international organizations tend not to become involved in monitoring population trajectories until the species in ques¬ tion has become critically endangered, by which time the process of enumeration is greatly complicated by the extreme rarity of the species. An example is the bio¬ diversity debacle in Guam. By the time the first quantitative bird population estimates were undertaken (Engbring and Ramsey, 1984), most of the birds were

8

Rodda, Sawai, Chiszar, & Tanaka

on the verge of vanishing (Engbring, 1983). Furthermore, not a single survey of Brown Treesnakes on Guam had been undertaken to determine if the snake was sufficiently abundant to have influenced the bird populations. Only after most of the birds had been extirpated was the first crude snake population estimate undertaken (Fritts, 1988), and by then it was impossible to observe the conditions under which the demise of the birds occurred. A major impediment to the enumeration of snakes is the fact that our tech¬ niques are ill suited to estimate population densities of species that are secretive and difficult to detect (Rodda and Fritts, 1992). Finding a single snake of a desired species is often a formidable challenge. To estimate the population den¬ sity, one must not only find a significant fraction (20-40% is recommended: White et al., 1982) of a reasonable-sized (e.g., 50-200) population, but one must do so within a period short enough to treat the sample as instantaneous, and one must develop a method for determining the geographic area covered by the sam¬ ple. Thus, there are many unresolved problems in the techniques for estimating population densities of snakes. Snake management has been a major stimulus for improvements in the science of reptile population enumeration techniques (see Part 4: “Population Biology”).

HOW ARE SNAKES MANAGED? Prevention of venomous snake-human conflicts is appropriate, regardless of whether the conflict is natural or exacerbated by anthropogenic landscape modifications. Furthermore, effective prevention often negates the need for more intrusive and difficult measures such as snake population manipulation. Thus prevention has been the frontline defense for virtually all types of snake management. The most widespread form of prevention is education. Informing people how they can effectively keep snakes out of their homes and how to distinguish haz¬ ardous species from those that are harmless and potentially beneficial to agricul¬ ture have obvious benefits. Repellents have not been widely used, but they are available for high-risk sit¬ uations in which repulsion of a snake is all that is needed to prevent a regrettable encounter (Secoy, 1979; Nishimura, this volume; Chap. 10; Toriba et al., this volume, Chap. 33). Detector dogs are in limited use for Brown Treesnake (Imamura, this volume, Chap. 27) and Habu (Shiroma and Ukuta, this volume, Chap. 26) detection. The dogs’ extremely sensitive sense of smell is used to check potential snake hiding places. In the case of the Brown Treesnake, the detector dogs alert handlers to the presence of snakes in cargo, thus preventing transportation of the snake to new islands. Barriers are beginning to play a major role in blocking the dispersal of snakes into high-priority areas (see, in this volume: Aguon et al., Chap. 38; Campbell,

Introduction

9

Chap. 21; Nishimura, Chap. 22; Hayashi et al., Chap. 23; and Shiroma and Akamine, Chap. 24). Japanese managers have used barriers to protect high-risk villages from the venomous Habu. American managers are now constructing bar¬ riers to limit the spread of the Brown Treesnake to new areas and to cordon off key areas of wildlife habitat to protect species that cannot coexist with introduced snakes. Barriers have also been used in highly localized situations to keep snakes from individual bird nests (Neal et al., 1993; Withgott et al., 1993; Aguon et al., this volume, Chap. 38). Interdiction by human searchers is another way to manage the movements of snakes. Interdiction can be used to protect native snakes from introduced pests as well. For example, keeping mongoose and rats off pest-free islands in the Caribbean will be of great value in preserving endangered snakes such as the Antiguan Racer, Alsophis antiguae (Day and Daltry, 1996). Prevention, as manifest in the above techniques, may be used to (1) protect people from dangerous snakes, (2) protect livestock from dangerous snakes, (3) protect snakes from dangerous people, (4) protect snakes from introduced preda¬ tors, and (5) protect native prey from introduced snakes. A special form of inter¬ diction is management of commercial trade. This has been done in an attempt to reduce harvesting to sustainable levels, often through CITES, the Convention on International Trade in Endangered Species (Brautigam, 1992). To the extent that unfortunate interactions with snakes cannot be avoided, management problems can be corrected by manipulating the densities of snake populations or those of their predators, prey, and competitors. How might snake numbers be manipulated upward? Habitat restoration is an obvious solution, but one done only rarely on behalf of snakes, in part because we know so little about their habitat requirements. In the extreme case of Round Island, the habitat is being restored on general principles, with good reason to believe that alternate factors are not responsible for the observed declines in Bolyeria and Casarea. Exotic predator control is an approach with the advantage that it directly reverses the circumstances that have so often been responsible for declines of snakes and other native wildlife. In recent years, total eradication of introduced pests has been attempted on a number of small islands. New Zealanders have spearheaded this approach as a means of conserving their endangered birds and lizards (Butler and Merton, 1992; Johnson et al., 1994). On the other hand, snakes are not native to New Zealand; we have only a few examples of case studies of exotic predator control expressly for snake management, mostly from Caribbean Islands (Day and Daltry, 1996). Prey enhancement involves the intentional manipulation of snake prey as a means of promoting the welfare of a species at risk. To our knowledge, this has not been done, except as a consequence of habitat restoration. •One popular approach to wildlife conservation is to capture animals for cap¬ tive propagation. This approach is generally coupled with habitat protection and restoration, such that the progeny of the wild parents can be returned or repatri-

10

Rodda, Sawai, Chiszar, & Tanaka

ated to native habitat (Dodd and Seigel, 1991; Snyder et al., 1996). Captive prop¬ agation has been attempted for a number of endangered snakes, such as the Aruba Island Rattlesnake (Crotalus unicolor, Honegger, 1975), Antiguan Racer (Day, 1996), Round Island Boa (Casarea dussumieri; Bloxam and Tonge, 1986), and San Francisco Gartersnake (U.S. Fish and Wildlife Service, 1985), but there are as yet no unequivocal examples of the restoration of endangered snake populations as a consequence of repatriated captive-reared snakes. Populations of the Aruba Island Rattlesnake, Northern Watersnake (Nerodia sipedon, not an endangered species), and Red-sided Gartersnake (Thamnophis sirtalis parietalis, not an en¬ dangered species) have been augmented by translocation of wild-caught indi¬ viduals (Cook, 1989; Peterson, 1992; Macmillan, 1995). Restoration through release of captive-reared animals has been performed too few times to determine if this is an appropriate conservation strategy. Captive propagation has engen¬ dered a large body of herpetoculture literature, which we do not cover in this book. Readers interested in this specialty are encouraged to consult a source such as Murphy et al., 1994, or one of the many herpetoculture manuals. The Red-sided Gartersnake management program mentioned in the preceding paragraph is an interesting one in that it reflects society’s changing attitudes toward snakes. One of the first papers about snake management (Flattery, 1949) reported with great pride the use of nicotine sulphate to poison the reviled gartersnakes near Inwood, Manitoba (“Although . .. harmless, they were a source of annoyance to the public”). Later, the abundance of gartersnakes near Inwood became a direct source of income for local residents, who sold so many snakes to pet dealers that the population was depleted (Macmillan, 1995). At present, Inwood uses its famous gartersnake aggregations as a drawing card for tourists and honors the snakes’ presence with a huge statue and civic events (Shine, 1991). Furthermore, wildlife authorities have translocated large numbers of gartersnakes to restore depleted or extirpated hibernacula around Inwood (Macmillan, 1995). Thus, attitudes toward these nonvenomous snakes have almost completely reversed within a single human generation. Snake management efforts to decrease the abundance of species that are introduced, dangerously venomous, or artificially subsidized by humans use the same techniques: habitat modification and prey manipulation (in this case depletion), but also biological control, toxicants, attractants, trapping, direct cap¬ ture, commercialization, and eradication (total elimination of a population). Some examples of these are mentioned above. To the best of our knowledge, biological control efforts directed at snakes have been largely unintentional or informal, and have often involved mongoose. Mam¬ mals have generally been introduced to control rats, although in the case of the Ryukyu Islands, some predators were reportedly brought in expressly for snake control. The effort does not appear to have been successful (Abe et al., this vol¬ ume, Chap. 29), as few snakes are eaten by the predators, but many nontarget wildlife species have suffered. Paradoxically, in the Caribbean, where mongoose

Introduction

11

were introduced for rat control, nontarget snake species have suffered from the predators (see above), as have other nontarget organisms. Thus, vertebrate bio¬ control agents have often done collateral damage to nontarget species (Howarth, this volume, Chap. 32). Toxicants have not been widely used for snakes, although a number of sub¬ stances highly toxic to snakes are known (e.g., nicotine; see Savarie and Bruggers, this volume, Chap. 34; Toriba et al., this volume, Chap. 33). There are two obsta¬ cles to the use of toxicants: the secretive nature of snakes and finding a method of delivering the toxicant exclusively to them. Due to their secretive nature, snakes are often more abundant than sightings of them would indicate. Thus, whereas rat poison often produces conspicuous carcasses; the victims of snake poisons will be difficult to find. To advance knowledge about toxicant target specificity, a sub¬ stantial effort has been made both in the United States and Japan (see Part 3: “Behavioral and Sensory Biology”) to learn about the sensory modalities and specific cues used by snakes. Although it has been possible to demonstrate that selected snakes use pheromones, surface chemicals, airborne chemicals, visual cues, vibrations, thermal cues, and motion to detect prospective prey or mates, presentations of these cues in isolation have not yielded attractants with poten¬ cies approaching that of natural prey or mates. The experiments have yielded a much richer understanding of the sensory world of snakes. We now know that prey (unintentionally) release a staggering number of chemicals to which snakes may respond. We also know that in certain contexts snakes may be largely insen¬ sitive to cues that would garner their full attention in other contexts. However, the individual facts have yet to be pulled together into a cohesive theory of prey recognition. The compilation of data has been greatly advanced by research efforts to synthesize attractants for both snake toxicants and snake traps. Direct capture of snakes is a control measure that has been used all over the world. Bounties have been used in a few areas to induce citizens to capture the maximum number of snakes. To the best of our knowledge, snake bounties have been subjected to systematic study only once (see below). Presumably, bounties suffer from the same ambiguity that has complicated most pest bounty programs: Do people actively search for the wildlife species as a result of the bounty, or do they merely collect the bounty as a windfall payment for opportunistic killings they would perform in the absence of the bounty? In the contexts in which the bounty is high enough to elicit new effort, do the aggregate captures result in an appreciable reduction in the target population? These questions were partially answered by a snake roundup that took place on Guam in the summer of 1990. As a promotion, a newspaper offered a new pickup truck and other consumer goods as prizes for the teams who captured the most Brown Treesnakes in one month. Although not a bounty, this contest was an attempt to stimulate snake catching by offering monetary rewards. Almost 1275 snakes were captured, nearly 85% of them by just three pairs of searchers (M. J. McCoid, 1990; pers. comm., 1996). In terms of snake control, the removal of 1275

12

Rodda, Sawai, Chiszar, & Tanaka

snakes from a population believed to total 0.5-2.5 million snakes would not yield a measurable change in the overall population. Excluding the cost of organizing, promoting, and managing the snake search, the US$20,000 in prizes induced fewer than 10 people to collect snakes intensively, at a cost of about US$16 per snake. It seems that few people are willing to spend their evenings combing jungles for venomous snakes, even if the financial rewards are substantial. Under these circumstances, bounties are more likely to provide a personal windfall than to stimulate significant snake control. Bounties may have greater utility in areas of the world with fewer opportuni¬ ties for remunerative employment than Guam. For example, in the late 1800s, when the bounty on Saw-scaled Vipers (Echis carinata) was tripled from about two cents to six cents, residents of the Ratnagiri district south of Bombay, India, turned in 115,921 snakes during an eight day period (Vidal, 1890, cited in Minton and Minton, 1969). The influence this action had on snake populations or snakebite frequency does not seem to have been documented. Commercialization of snakes has generally been stimulated by the opportunity for cash rewards, and might be encouraged as a way of decreasing the abundance of a target species (Klauber, 1982). In Japan, Habu are collected and sold to vendors who put them into bottles of liquor that are reputed to enhance the viril¬ ity of the consumers. This product commands extraordinary prices (around US$100 per bottle), producing high rewards for systematic collectors of Habu. Over the last several decades, Habu numbers seem to have declined in parts of their range, and it is possible that commercial collection has had a role in this de¬ cline. Thus commercialization may have some management value; however, it is a very crude tool, for at least three reasons: (1) nontarget species may also be de¬ pressed by this activity (other snakes are also taken and sold in large numbers); (2) once initiated, this activity could be impossible to modulate in response to changing wildlife management needs; and (3) commercialization creates an in¬ centive for the industry to pressure authorities to maintain a high density of snakes to accommodate future demand. Furthermore, if the price offered to hunters is stable, hunters are apt to limit their searching to areas where the snake is most common, leading to “commercial extinction” but not biological eradica¬ tion. Thus, commercialization is not likely to facilitate eradication if that is a goal. An additional problem of commercialization is that it may act as an incentive for proliferation: snakes may be planted in nonnative localities to produce revenue (see Part 7: “Conservation Biology”). To the best of our knowledge, systematic eradication of snakes has been attempted only a few times, and only for the species that are the focus of this book: the Habu (Shiroma and Akamine, this volume, Chap. 26; Katsuren et al., this volume, Chap. 25) and Brown Treesnake (Rodda et al., this volume, Chap. 39). The Brown Treesnake effort was a feasibility experiment for protection of native wildlife, and was successfully completed following preparation of that chapter. Subsequently, the barriers used in that experiment (Campbell, this

Introduction

13

volume, Chap. 21) were dismantled, and the snakes quickly repopulated the area. In contrast, the Habu initiatives were more permanent. An attempt to eradicate Habu from Minnajima, a small (37ha) islet near Okinawajima, resulted in sub¬ stantial reduction in the snake population, but not total elimination (Katsuren et al., this volume, Chap. 25). More complete success was achieved for a variety of small fenced villages on the main island of Okinawajima (Shiroma and Akamine, this volume, Chap. 24), although the continuous movement of people and cargo in and out of a village probably results in persistent leakage of snakes through the snake barrier fence. Thus eradication has been used for protecting native wildlife and humans, on both natural islands and artificially isolated “islands.”

WHERE IN THE WORLD WOULD ONE EXPECT SNAKE MANAGEMENT TO OCCUR? AND WHERE IS IT OCCURRING? Snakes are managed primarily to protect human health and biodiversity (Dodd, 1993). To a lesser extent, snakes are managed to protect agricultural products and power systems. Venomous snakes are present in most warm and temperate parts of the globe, as are snakes that climb on power lines and those that prey on agri¬ cultural pests. Biodiversity crises involving snakes are probably occurring world¬ wide, but are more pressing in island ecosystems. Yet, snake management has not reached advanced levels in most areas. The premier example of snake management for human health is the extensive Japanese program on the Habu. This program is characterized not only by welldeveloped snake management operations, but also by an extensive research pro¬ gram that has been the vanguard in virtually all aspects of snake management. Human health is also a concern in management of the Brown Treesnake. Are there no well-developed snakebite prevention programs elsewhere? Preparation of antivenin, evaluation of clinical snakebite procedures, education of physicians and residents to differentiate venomous from harmless species, and dissemination of basic information on the circumstances of snakebite are practiced worldwide; however, more active management of snakes seems to be limited to just a few localities. Is this because the risk of snakebite is higher in the intensively managed sites? Surprisingly, it does not seem possible to answer that question with the epidemi¬ ological information at hand. Snakebite incidences are reported as the number of bites per 100,000 people per year. Treatment efficacy is recorded as a percentage of total bites that are fatal. Although the number of fatalities can often be com¬ piled accurately, many bites go unrecorded, producing errors in estimating both the bite frequency and the treatment efficacy. In countries with advanced medical facilities, the fraction of bites that are fatal is relatively small. For example, Sawai (in Whitaker, 1978) estimated that only about 0.5% of those treated in Indian hospitals failed to recover (the mortality rate is lower for victims of venomous snakebite in more developed countries); however, Sawai also estimated that about

14

Rodda, Sawai, Chiszar, & Tanaka

90% of all snakebite victims in India did not receive hospital treatment, and of those about 5% died. What is the worldwide pattern of snakebite incidence? Bite rates in the United States are exceptionally low, about 3.5/100,000 (Parrish, 1966). In much of the Tropics, values are about 10 times higher. For example, an estimate of 24/100,000 was reported for both Central America (Clark, 1942) and South Africa (Visser and Chapman, 1978), whereas Sawai estimated a slightly higher value for India— about 35/100,000. Firm estimates for Southeast Asia do not seem to be available, although values in this range or slightly higher appear reasonable (Sawai, 1993). The highest published value seems to be for the Amami Islands, in the northern part of the Ryukyu Archipelago of lapan, where Habu bite incidences are as high as 200/100,000 (see Sawai et al., this volume, Chap. 5). One island in the Amami Islands (Tokunoshima) has experienced years with bite rates in excess of 400/100,000 (Tanaka et al., this volume, Chap. 8). Although clinical treatment of Habu bites has advanced to the point where fatal bites are exceedingly rare, the myonecrotic venom often causes permanent crippling. Thus it is not sur¬ prising that the lapanese have developed a strong research program in snake management. The agricultural and power system impacts of snakes do not seem to have been well quantified in any locality, with the exception of Guam, where records have been compiled on the frequency of power outages (about once per three days, see Fritts and Chiszar, this volume, Chap. 4) and qualitative loss rates for chickens (Fritts and McCoid, 1991). Gartersnakes (Thamnophis sp.) are responsible for many of the power outages that occur about once per week during the warm season in western Washington (R. A. Noble, pers. comm., 1996). Similar frequen¬ cies of power outages attributable to climbing ratsnakes (Elaphe obsoleta) have been reported for a variety of states in the southeastern and south-central United States (e.g., Georgia, Oklahoma, Texas), but no details have been published. With¬ out comparative information, it is impossible to determine if the Guam values are exceptional, although this is widely believed to be the case. Snake management is occurring in areas where losses in biodiversity can be attributed to snakes. Such spots include islands naturally lacking snake predators and unusual mainland situations where snake prey, especially fish and birds, and their snake predators are artificially concentrated. As example of the latter is the Crescent Lake National Wildlife Refuge in Nebraska (Imler, 1945), where ducks concentrated by artificial nest boxes and water level manipulations to facilitate their feeding were losing 38% of their nests to Bullsnakes. This loss rate was reduced to 18.5% following four years of intensive snake trapping (Imler, 1945). Another example was described by Bauman and Metter (1975), who reported extraordinary losses of hatchery fish to watersnakes (Nerodia sipedon). Hatchery managers offered a 50 cent bounty (corrected for inflation to

1996) for

snakes caught on the grounds, and collected 6328 snakes in a single year. The rate

Introduction

15

at which snakes were turned in did not decline over time however, suggesting that snake influx and recruitment were high enough to sustain this unusual “harvest” The biodiversity losses suffered as a result of the introduction of the Brown Treesnake on Guam are mentioned above. Less well understood are losses of bird and lizard species in the Mascarene Islands (Indian Ocean) arguably attributable to the introduction of the Wolf Snake (Lycodon aulicus), lizard losses in the Balearic Islands (Mediterranean Sea) arguably attributable to the introduction of the False Smoothsnake (Macroprotodon cucullatus), and the potential loss of lizards and native birds as a result of introductions of the Wolf Snake to Christmas Island (Indian Ocean), the Corn Snake (Elaphe guttata) to various islands in the West Indies (Schwartz and Henderson, 1991; R. W. Henderson, pers. comm., 1996), and a variety of snakes to Hawaii (which intercepts arriving snakes about once every two weeks, L. Nakahara, Hawaii Dept, of Agriculture, pers. comm., 1990). Because the histories of these introductions are not further described in this book, it is worthwhile to consider them here. The Wolf Snake is a species that readily enters houses throughout southern Asia (Leviton, 1965). This frequent association with people has probably increased the likelihood of accidental human transport, as the snake has colonized a number of islands, including many islands in the Philippines and Indonesia and both Australia’s Christmas Island (near lava, Indonesia) and Mauritius and Reunion (in the Mascarenes, east of Madagascar). The introduction of this saurophagous (lizard-eating) snake to the Mascarenes in the nineteenth century was followed shortly afterward by the disappearance of about half of the islands’ saurofauna (Cheke, 1987). In particular, the Wolf Snake has been associated with the demise of Bojer’s Skink (Gonygylomorphus bojerii; C. lones, pers. comm., 1993). A great many other ecological changes were occurring at the same time, however, and the data do not clearly demonstrate that the Wolf Snake was responsible for any of the losses. Nonetheless, the ecologists working with the lersey Wildlife Trust to restore the natural environment of an islet off Mauritius have concluded that eradication of the Wolf Snake will be an important com¬ ponent in making the lie aux Aigrettes Nature Reserve safe for repatriation of endangered lizards (C. Jones, in litt., 1994). A related situation has occurred with Wolf Snakes on Australia’s Christmas Island. In this case, the introduction is quite recent (L. A. Smith, 1988), and the impacts, if any, may not be manifest for several years (Fritts, 1993). However, there are five native reptile species found only on this island (Cogger et al., 1983), so much could be lost if the snake were to extirpate predator-naive species. The effect of the introduction of the False Smoothsnake on the Balearic Islands cannot be fully documented because it occurred prehistorically, in about the third century b.c. Of interest are the distributions of several wall lizards, especially Lilford’s Wall Lizard (Podarcis lilfordi). This species is found only on the islets sur¬ rounding the two main Balearic islands: Menorca and Mallorca. As this species is

16

Rodda, Sawai, Chiszar,

&

Tanaka

found nowhere other than in the Balearic Islands, it seems likely that it originated on the main islands, spread to the offshore islets, and then was extirpated on the main islands. This is the pattern we have observed with lizards that appear to have been recently extirpated by the Brown Treesnake on Guam: all of the extirpated species persist on tiny offshore islets. The long-term prospects of these tiny pop¬ ulations are poor, because chance weather events and population fluctuations are likely to cause the local extinction of such small populations. Thus the persistence of small populations on many offshore islets suggests that a species has been restricted to the islets for an ecologically short period of time. Similarly, most of the reptiles that have recently been extirpated from Mauritius and Reunion persist on tiny offshore islets. In all of these cases, ecologists are in agreement that the persistence of species on many offshore islets implies that extinction on the main island occurred recently. In the case of the Balearic Islands, however, there is no consensus as to which species was responsible for the recent extirpation of the wall lizards. Eisentraut (1950) proposed that the introduced False Smoothsnake was responsible. Mertens (1957) rejected this possibility but did not offer a concrete alternative, although he did attribute the strange distribution of the lizard to as yet undefined ecolog¬ ical forces. Colom (1957) argued that the extirpations on the main islands were most likely a result of adverse weather, although it is difficult to see how adverse weather could have totally eliminated the larger mainland populations while sparing the smaller populations on the tiny offshore islets. Alcover et al. (1981) attributed the loss from the main island of Menorca (but not Mallorca) to com¬ petition from another wall lizard, Podarcis sicula. In this regard it is notable that Case and Bolger (1991) reviewed the evidence for recent reptile extinctions on islands worldwide and found no strong evidence supporting competition as a cause for any reptile species extinction. They did find evidence that introduced predators had caused extinctions of reptiles on islands. Some examples of management programs for the conservation of snakes are given above. The tactics of greatest interest to snake conservationists seem to be associated with the locality of the ecosystem being managed. Conservation of mainland snakes seems to rely most heavily on purchase of key habitats and regulation of capture and harassment. For example, the emphasis in protection of the San Francisco Gartersnake (U.S. Fish and Wildlife Service, 1985) is on acquisition of essential wetlands and law enforcement of the regulations against capture of the snakes (primarily for the pet trade). Similarly, work being done to protect the Timber Rattlesnake in the northeast United States (Brown, 1993) focuses on protection of hibernacula and enforcement of laws against destruction of dens. In contrast, the efforts for island species, such as the Round Island bolyerids, Aruba Island Rattlesnake, and Antiguan Racer, emphasize captive pro¬ pagation and restoration of island habitats by eradicating introduced mammals such as mongoose and rats as a prelude to eventual repatriation or population augmentation by captive stock.

Introduction

17

The Aruba Island Rattlesnake program is an interesting one in that it grapples with two problems that may limit captive propagation and repatriation as a con¬ servation tool (Peterson, 1992). The first problem is that captive colonies are prone to suffer from diseases associated with captivity. In the case of the Aruba Island Rattlesnake, the captives have become infected with a paromyxo virus, which is nearly universal among colonies of captive vipers. Wild rattlesnakes on Aruba are free from the virus, and release of the captives might actually further endanger the species. Thus the value of the captive snakes is somewhat unclear. In the case of the San Francisco Gartersnake, captive propagation was undertaken not to augment the wild population but to satisfy the pet industry demand, thereby reducing poaching pressure on wild stocks (Tonge and Bloxam, 1986). The second problem experienced by the Aruba Island Rattlesnake may be even less tractable than the captive propagation dilemma: Aruba is a small island whose snake habitat has already been reduced to a marginal or submarginal amount, and the few undeveloped areas are under great pressure for development by tourism and industrial interests. Given a seemingly inevitable monotonic increase in the human population, what prospects exist for maintenance in perpetuity of a viable wild population? This is a problem that has concerned, or will concern, every manager of wildlife.

SUMMARY Snake management is the practice of systematically influencing snake-human interactions. Snakes are most often managed to promote human health and preserve biodiversity, both of snakes and of prey species that may be affected by introduced or subsidized snakes. Snakes may also be managed to reduce damage to electrical power systems, agricultural products, or other infrastructure com¬ ponents. Snake management may be stimulated by the presence of dangerously venomous snakes in their natural ecosystems, but more often snakes become the subjects of management efforts because their numbers have been changed by anthropogenic influences such as species introductions to new habitats, sub¬ sidization through an exceptional abundance of commensal species such as rats, or destruction of snake habitat. Snake-human interactions are managed through both prevention and correc¬ tion. Prevention may include education, repellents, detector dogs, barriers, and interdiction. Common corrective measures include habitat modification, exotic predator control, prey management, biological control agents, toxicants, attractants, direct capture, eradication, and captive propagation. These techniques are applied both to situations in which the immediate objective is reduction in snake abundance and to those in which the goal is to increase snake abundance. These two goals of snake management are joined by a common body of herpetological knowledge and similar techniques. Although a great many snake species are being studied, and a number of species are being preserved by habitat protection,

18

Rodda, Sawai, Chiszar, & Tanaka

exotic predator control, and captive propagation, the best-developed scientific programs in snake management are for two venomous species in the Pacific region: the Brown Treesnake of Guam and the Habu of Japan. Because of lan¬ guage barriers, each research program operated with little information about the other until we began preparation of this volume. This book brings together, for the first time, the major findings in snake management.

ACKNOWLEDGMENTS We thank T. Fritts, R. Rondeau, G. Perry, and K. Dean-Bradley for suggesting improvements to the manuscript. M. McCoid, C. Jones, L. Nakahara, M. Jennings, R. Henderson, N. Scott, and others generously shared unpublished data with us. E. W. Campbell III introduced us to much of the literature on conservation of island snakes.

LITERATURE CITED Alcover, J. A., S. Moya-Sola, and J. Pons-Moya. 1981. Les quimeres del passat, els

vertebrats fossils del Plio-Quarternari de las Balears i Pitiuses. Ciutat de Mallorca: Editorial Moll. Bauman, M. A., and D. E. Metter. 1975. Economics, feeding, and population struc¬ ture of Natrix s. sipedon in a goldfish hatchery. Prog. Fish Cult. 37.197—201. Bloxam, Q. M. C., and S. J. Tonge. 1986. The Round Island Boa, Casarea dussumieri, breeding programme at the Jersey Wildlife Preservation Trust. Dodo, J. Jersey Wildl. Preserv. Trust 23:101-107. Bogert, C. M. 1948. The problem of snake control. How to tell whether the snake at your doorstep is really one of the dangerous ones, and what you can do about it if it is. Nat. Hist. 57:185-188. Brautigam, A. 1992. Conservation of Pacific islands flying foxes and the Convention on International Trade in Endangered Species of Wild Fauna and Flora. In D. Wilson and G. Graham, eds., Pacific Island Flying Foxes: Proceedings of an International Conservation Conference, pp. 155-159. U.S. Fish Wildl. Serv. Biol. Rep. 90(23). Brock, E. M., and W. E. Howard. 1962. How to control snakes. Pest Control

(August, 1962): 30-36. Brown, W. S. 1993. Biology, Status, and Management of the Timber Rattlesnake (Crotalus horridus): A Guide for Conservation. Soc. Stud. Amphib. Rept. Herpetol. Circ. 22. Bullock, D. J. 1986. The ecology and conservation of reptiles on Round Island and Gunner’s Quoin, Mauritius. Biol. Conserv. 37:135—156. Butler, D., and D. Merton. 1992. The Black Robin: Saving the World’s Most Endangered Bird. Auckland, New Zealand: Oxford Univ. Press. Byford, J. L. 1986. Non-poisonous Snakes. Lincoln: Nebraska Cooperative Extension Service.

Introduction Case,

T. J.,

and

D. T.

Bolger.

19

1991. The role of introduced species in shaping the

distribution and abundance of island reptiles. Evol. Ecol. 5:272-290. C. R.,

Chandler,

R J.

and

1990. Habitat use by a boid snake, Epicrates

Tolson.

monensis, and its anoline prey, Anolis cristatellus. J. Herpetol. 24:151-157. A. S. 1987. An ecological history of the Mascarene Islands, with particular

Cheke,

reference to the extinctions and introductions of land vertebrates. In A. Diamond, A. S. Cheke, and H. F. I. Elliott, eds., Studies of Mascarene Island Birds, pp. 5-89. Cambridge: Cambridge Univ. Press. H. C. 1942. Venomous snakes. Some Central American records. Incidence of

Clark,

snakebite accidents. Am. J. Trop. Med. 22:37-49. Cogger,

H. G., R. A.

Sadlier, and

E.

Cameron.

1983. The Terrestrial Reptiles of

Australia’s Island Territories. Austr. Natl. Parks Wildl. Serv., Spec. Publ. 11. Colom,

G. 1957. Biogeografia de las Baleares. La formacion de las islas y el origen de

su flora y de su fauna. Ciutat de Mallorca: Edicion Estudio General Luliano. Cook,

R. P. 1989. And the voice of the Gray Tree Frog was heard again in the

land ... Park Sci. 9:6-7. M. W. 1961. Rattlesnake Control in Residential Areas. Univ. Calif. Agric.

Cummings,

Exp. Sta. One Sheet Answers No. 110. Day,

M. 1996. World’s rarest snake arrives in Jersey. On the Edge 77:1-2.

Day,

M., and J.

3:14-15. Dodd, C. K.

Daltry.

Jr.

1996. Rat eradication to conserve the Antiguan Racer. Aliens

1987. Status, conservation, and management. In R. A. Seigel,

J. T. Collins, and S. S. Novak, eds., Snakes: Ecology and Evolutionary Biology, pp. 478-513. New York: Macmillan. -. 1993. Strategies for snake conservation. In R. Seigel and J. Collins, eds., Snakes: Ecology and Behavior, pp. 363-394. New York: McGraw-Hill. Dodd,

C. K.

Jr.,

K. M.

Enge, and

J. N.

Stuart.

1989. Reptiles on highways in north-

central Alabama, USA. J. Herpetol. 23:197-200. Dodd,

C. K.

Jr., and

R. A.

Seigel.

1991. Relocation, repatriation, and transloca¬

tion of amphibians and reptiles: Are they conservations strategies that work? Herpetologica 47:336-350. Eisentraut,

M. 1950. Das Fehlen endemischer und das Auftreten landfremder

Eidechsen auf den beiden Hauptinseln der Balearen, Mallorca und Menorca. Zool. Beitr. (n.f.) 1:3-11. Engbring, J. 1983. Forest birds of Guam in critical danger. U.S. Fish Wildl. Serv. Endang. Species Tech. Bull. 8:6-8. Engbring,

J., and F. L. Ramsey. 1984. Distribution and Abundance of the Forest Birds

of Guam: Results of a 1981 Survey. U.S. Fish Wildl. Serv., FWS/OBS-84/20. Engeman,

R. M., M. A.

Linnell,

P. A.

Pochop, and

J.

Gamboa.

1997a. Substantial

reductions of Brown Tree Snake (Boiga irregularis) populations in blocks of land on Guam through operational trapping. Int. Biodegrad. Biodeter., in press. Engeman,

R. M., S.

Sayama, and

M. A.

Linnell.

1997b. Operational utility of

perimeter trapping for removing Brown Tree Snakes (Boiga irregularis) from a defined area. Snake, in press. Flattery, M. 1949. An effective way to control snakes. Pest Control 17(2): 16—18.

20

Rodda, Sawai, Chiszar, & Tanaka T. H. 1988. The Brown Tree Snake, Boiga irregularis, a Threat to Pacific

Fritts,

Islands. U.S. Fish Wildl. Serv., Biol. Rep. 88(31). -. 1993. The Common Wolf Snake, Lycodon aulicus capucinus, a recent colonist of Christmas Island in the Indian Ocean. Wildl. Res. 20:261-266. Fritts,

T. H.,

and

M. J.

McCoid.

1991. Predation by the Brown Tree Snake on

poultry and other domesticated animals in Guam. Snake 23:75-80. H. W.,

Greene,

and

J. A.

Campbell.

1992. The future of pitvipers. In J.

Campbell and E. Brodie Jr., eds., Biology of the Pitvipers, pp. 421-427. Tyler, Tex.: Selva. Hall,

R. J. 1980. Effects of Environmental Contaminants on Reptiles: A Review.

U.S. Fish Wildl. Serv., Spec. Sci. Rep. Wildl. 228. L. D.,

Harris,

and

G.

Silva-Lopez.

1992. Forest fragmentation and the conservation

of biological diversity. In P. Fiedler and S. Jain, eds., Conservation Biology: The Theory and Practice of Nature Conservation Preservation and Management, pp. 197-237. New York: Chapman & Hall. R. E.

Honegger,

1975.

Red Data Book, vol.

3,

Amphibia and Reptilia. Morges,

Switzerland: IUCN. How, R. A.,

and

J.

Dell.

1994. The zoogeographic significance of urban bushland

remnants to reptiles in the Perth region, Western Australia. Pac. Conserv. Biol. 1:132-140. Howey, Imler,

M. O. 1955. The Encircled Serpent. New York: Arthur Richmond.

R. H. 1945. Bullsnakes and their control on a Nebraska wildlife refuge. J. Wildl.

Manage. 9:265-273. Jackley,

A. M. 1947. How to locate and trap dens of rattlesnakes. Mont. Farmer

34:1,3-5. J. P., D.

Johnson,

Anthony, and

Q. M. C.

Bloxam.

1994. Rat eradication from

Praslin Island: The St. Lucia Whiptail Lizard conservation programme. Dodo, J. Jersey Wildl. Preserv. Trust 30:114-118. Kaufman,

G. A.,

and

J. W.

Gibbons.

1975. Weight-length relationships in thirteen

species of snakes in the southeastern United States. Herpetologica 31:31-37. L. M. 1982. Rattlesnakes: Their Habits, Life Histories, and Influence on

Klauber,

Mankind. Berkeley: Univ. California Press. Leviton,

A. E. 1965. Contributions to a review of Philippine snakes, VIII. The snakes

of the genus Lycodon H. Boie. Philippine J. Sci. 94:117-140. Macmillan,

S.

1995. Restoration of an extirpated Red-sided Garter Snake

Thamnophis sirtalis parietalis population in the Interlake region of Manitoba, Canada. Biol. Conserv. 72:13-16. M. J. 1990. Job progress report, research project segment. In R. B.

McCoid,

Anderson, G. J. Wiles, and L. L. Mariano, eds., 1990 Annual Report, pp. 178-199. Agana, Guam: Guam Division of Aquatic and Wildlife Resources. R. 1957. Mallorca: Ein herpetogeographisches Problem. Zool. Beitr. Berlin (n.f.) 3:1-16.

Mertens,

Merton,

D. V. 1987. Eradication of rabbits from Round Island, Mauritius: A conser¬

vation success story. Dodo, J. Jersey Wildl. Preserv. Trust 24:19-43.

Introduction S. A.

Minton,

Jr., and

M. R.

Minton.

21

1969. Venomous Reptiles. New York: Charles

Scribner’s Sons. R., and D.

Morris,

Morris.

1965. Men and Snakes. New York: McGraw-Hill.

B. 1983. The Cult of the Serpent: An Interdisciplinary Survey of Its

Mundkur,

Manifestations and Origins. Albany: SUNY Press. J. B., K.

Murphy,

Adler, and

J. T.

Collins,

eds. 1994. Captive Management and

Conservation of Amphibians and Reptiles. Society for the Study of Amphibians and Reptiles. Neal,

J. C., W. G.

Montague, and

D. A.

James.

1993. Climbing by Black Rat Snakes

on cavity trees of Red-cockaded Woodpeckers. Wildl. Soc. Bull. 21:165-171. Oldemeyer,

J. L., D. E.

Biggins,

B. J.

Miller, and

R.

Crete.

1993. Proceedings

of the Symposium on the Management of Prairie Dog Complexes for the Reintroduction of the Black-Footed Ferret. U.S. Fish Wildl. Serv., Biol. Rep. 13. Palmer,

T.

1992.

Fandscape with Reptile: Rattlesnakes in an Urban World. New York:

Ticknor 8c Fields. Parrish,

H. M. 1966. Incidence of treated snakebites in the United States. Public

Health Rep. 81:269. Pechmann,

J. W.

J. H. K., D. E.

Gibbons.

Scott,

R. D.

Semlitsch,

J. P.

Caldwell,

F. J.

Vitt, and

1991. Declining amphibian populations: The problem of separat¬

ing human impacts from natural fluctuations. Science 253:892-895. K. 1992. Aruba Island Rattlesnake Studbook. Houston Zool. Gardens,

Peterson,

Houston, Tex. H.,

Rodda, G.

and

T. H.

Fritts.

1992. Sampling techniques for an arboreal snake,

Boiga irregularis. Micronesica 25:23-40. P. C.,

Rosen,

and

C. H.

Fowe.

1994. Highway mortality of snakes in the Sonoran

Desert of southern Arizona. Biol. Conserv. 68:143-148. Sajdak,

R. A., and R. W.

Henderson.

1991. Status of West Indian racers in the Fesser

Antilles. Oryx 25:33-38. Sawai,

Y. 1992. Venomous snakes and snakebite treatment in Asia. Snake 24:129-141.

Sawai,

Y.,

and

M.

Honma.

1975. Snakebites in India. Snake 7:1-16.

-. 1976. Snakebites in India. In A. Ohsaka et al., eds., Animal, Plant and Microbial Toxins, vol. 2, pp. 451-460. New York: Plenum Press. -. 1993. Clinical problem of snakebites in Southeast Asia. In A. T. Tu, ed., ToxinRelated Diseases, pp. 445-469. New Delhi: Oxford 8c IBH. Schwartz,

A.,

and

R. W.

Henderson.

1991. Amphibians and Reptiles of the West

Indies: Descriptions, Distributions, and Natural History. Gainesville: Univ. Press Florida. Secoy, D. M. 1979. Investigatory behavior of Plains Garter Snakes Thamnophis radix (Reptilia, Colubridae) in tests of repellent chemicals. Can. J. Zool. 57:691693. Sherriff,

D., Q. M. C.

Bloxam, and

R.

Gibson.

1995. A report on the reptiles of

Maria Island Major, St. Fucia, West Indies. Dodo, J. Jersey Wildl. Preserv. Trust 31:103—111. Shine, R. 1991. Australian Snakes: A Natural History. Ithaca: Cornell Univ. Press.

22

Rodda, Sawai, Chiszar, & Tanaka

Shine,

R.,

and

M.

1996. Large snakes in a mosaic rural landscape:

Fitzgerald.

The ecology of Carpet Pythons Morelia spilota (Serpentes: Pythonidae) in coastal eastern Australia. Biol. Conserv. 76:113-122. Shine,

R., P.

Harlow,

J.

S. Keogh, and Boeadi.

1995. Biology and commercial

utilization of acrochordid snakes, with special reference to Karung (Acrochordus javanicus). J. Herpetol. 29:352-360. L. A. 1988. Lycodon aulicus capucinus a colubrid snake introduced to

Smith,

Christmas Island, Indian Ocean. Rec. West. Aust. Mus. 14:251-252. Smith,

S. M. 1975. Innate recognition of coral snake pattern by a possible avian

predator. Science 187:759-760. -. 1977. Coral-snake pattern recognition and stimulus generalisation by naive Great Kiskadees (Aves: Tyrannidae). Nature (Lond.) 265:535-536. -. 1978. Predatory behaviour of young Great Kiskadees (Pitangus sulphuratus). Anim. Behav. 26:988-995. N. F. R.,

Snyder,

R.

S.

D. Toone, and

B.

Derrickson, S. Miller. 1996.

species policy. Conserv. Biol.

R.

Beissinger, J.

W.

Wiley, T. B. Smith,

W.

Limitations of captive breeding in endangered

10:338-348.

P. D. 1991. The Relationship between Wildlife and Highways in the

Southall,

Paynes Prairie Basin. Lake City, FL: Florida Department of Transportation, District 2. P. J. 1988. Critical habitat, predator pressures, and the management of

Tolson,

Epicrates monensis (Serpentes: Boidae) on the Puerto Rico Bank: A multivariate analysis. In R. C. Szaro, K. E. Severson, and D. R. Patton, eds., Management of Amphibians, Reptiles, and Small Mammals in North America, pp. 228-238. USDA Forest Serv., Gen. Tech. Rep. RM-166, Fort Collins, Colo. P. J.,

Tolson,

and

R. W.

Henderson. 1993.

The Natural History of West Indian Boas.

Taunton, Somerset, England: R 8c A Publications. Tonge,

S. 1989. A preliminary account of changes in reptile populations on Round

Island following the eradication of rabbits. Dodo, J. Jersey Wildl. Preserv. Trust 26:8-17. S. J.,

Tonge,

Q. M. C.

and

Bloxam.

1986. The San Francisco Garter Snake

Thamnophis sirtalis tetrataenia at the Jersey Wildlife Preservation Trust. Dodo, J. Jersey Wildl. Preserv. Trust 23:112-114. U.S.

Fish and Wildlife Service.

1985. Recovery Plan for the San Francisco Garter

Snake (Thamnophis sirtalis tetrataenia). Portland, Ore.: U.S. Fish and Wildlife Service. -. 1991. Proposed endangered status for the Giant Garter Snake. Fed. Regist. 56:67046-67052. Visser, J., and

D. S.

Chapman.

1978. Snakes and Snakebite. Capetown, South Africa:

C. Struik. Whitaker,

R. 1978. Common Indian Snakes: A Field Guide. Delhi, India: Macmillan

India. White,

G. C., D. R.

Anderson,

K. P.

Burnham, and

D. L.

Otis.

1982. Capture-

Recapture and Removal Methods for Sampling Closed Populations. Los Alamos, N.M.: Los Alamos National Laboratory.

Introduction Withgott,

J.

H.,

J.

C.

Neal,

and

W. M.

Montague.

23

1993. A technique to

deter climbing by rat snakes on cavity trees of Red-cockaded Woodpeckers. In D. L. Kulhavy, R. Costa, and R. G. Hooper, eds., Red-cockaded Woodpecker Symposium 3: Species Recovery, Ecology and Management. College of Forestry, Stephen F. Austin Univ., Nacogdoches, Tex.

.

Part I

BASIC BIOLOGY

^''"^This section describes the biology of two species of snakes, the Habu (Trimeresurus flavoviridis) and the Brown Treesnake (Boiga ir¬ regularis). These species are discussed in detail because the biologi¬ cal, socioeconomic, and medical research they have spawned defines the frontiers of knowledge regarding the epidemiology of snake envenomation and the biodiversity problems caused by introduced snakes. Collectively, the management programs for the Habu and the Brown Treesnake constitute the bulk of the world’s snake man¬ agement expertise. Unfortunately, the ability of scientists, policy makers, and snake managers to exchange ideas and improve techniques through mu¬ tual consultation has been hampered by the difficulties of language translation. Most Japanese scientists are well versed in written Eng¬ lish, but few English speakers can reciprocate. Thus, Englishspeaking scientists have generally been unaware of the full range of the studies conducted on the Habu. For example, although the Habu is arguably the world’s most thoroughly studied pitviper, a recent English-language compendium of pitviper knowledge (Campbell and Brodie, 1992) includes only four sentences that tap the vast corpus of knowledge on Habu. We hope this volume will redress this imbalance for English-language readers. Japaneselanguage readers interested in a popular account of the Habu may enjoy Yoshida, 1979; a popular account of the Brown Treesnake story can be found in Jaffe, 1994. A detailed lay account of the is¬ sues surrounding island extinctions, including those caused by the Brown Treesnake, may be found in Quammen, 1996. A recurrent theme throughout this volume is the importance of natural history particulars in explaining differences between the management programs for Brown Treesnakes and Habu, differences whose biological roots are introduced in Part 1. Public health issues are examined in greater detail in Part 2: “Venom and Human Health.” Two key areas of biological knowledge are developed in Part 3: “Behavioral and Sensory Biology” and Part 4: “Population Biology.” Subsequent parts focus directly on snake management. Crucial techniques for detecting, excluding, and capturing snakes are detailed in Part 5: “Capture and Detection.” Management ap¬ proaches that might be used on the scale of landscapes are

25

26

island is the only English translation available for several concepts that are distinguished in Japanese by word endings such as -oshima, -shima, and -jima. Thus, some Japanese island names are sometimes given without the ending; for example, “Okinawa” (island implied) is often used in place of “Okinawa;ima.”

Map 1 The Ryukyu Islands (southern Japan). The Habu is found in the Amami and Okinawa island groups. The word

Basic Biology *■

27

Farallon de Pajeros 20'

16

Map 2 The Mariana Islands (western Pacific Ocean). Extralimital populations of the Brown Treesnake are established on Guam and possibly Saipan.

described in Part 6: “Biological, Ecological, and Chemical Control.” Conservation biology is the subject of Part 7. Maps 1 and 2 show the Ryukyu and Mariana Islands, respectively. Just as genetic differentiation proceeds more rapidly in isolation, the isolation caused by language barriers has resulted in divergent management approaches for the two snakes. As the language barri¬ ers have fallen, the researchers working on each species have dis¬ covered each others’ work, and thus fresh approaches that can be adapted to the management of their own species. The insights pro¬ vided by this immigration of ideas are summarized in the Epilogue. Because ideas have migrated in both directions, two teams identify the highlights: one of English-speaking scientists, and the other of Japanese scientists. Together, their summaries highlight the most notable accomplishments, and unmet challenges, in each area of concentration.

28

Part I LITERATURE CITED Campbell, J. A., and E. D. Brodie Jr. 1992. Biology of the Pitvipers.

Tyler, Tex.: Selva. Jaffe, M. 1994. And No Birds Sing. New York: Simon 8c Schuster. Quammen, D. 1996. The Song of the Dodo. New York: Scribner. Yoshida, C. 1979. Habu and Man [in Japanese]. Naha City, Okinawa:

Ryukyu Shinpo-sha.

1 The Biology of the Habu (Trimeresurus flavoviridis) Shogi Mishima Hiroshi Tanaka Yoshio Sawai

T

he Habu, Trimeresurus flavoviridis, is a dangerously venomous crotalid snake of the Ryukyu Islands of southwest Japan (see Map 1). It is found on the high

islands of the Amami and Okinawa island groups. The absence of the Habu from low islands (e.g., Kikaijima, Okierabujima, Yoronjima) is thought to reflect extir¬ pation during a geologically recent time when the low islands were submerged (Hanzawa, 1935; Takara, 1962). Recolonization by the Habu of small islands may have been facilitated by land bridges that formed during recent glaciation. The Habu is absent from a number of the more isolated small islands in the Okinawa group, such as Agunijima, Akajima, Izenajima, Gerumajima, and Zamamijima (see Toyama and Ota, 1991, for an exhaustive distribution list). The areas occu¬ pied by the Habu have a mild, thermally seasonal climate, with the lowest mean monthly temperature being about 14°C in January on Amamioshima, and the highest about 28°C on Okinawajima in July (Nihei, 1977). Frost is rare, and snow¬ fall unknown. Relative humidities are consistently high (monthly means 69-86%) as a result of abundant year-round rainfall (annual totals 2000-3000 mm on most islands). Taxonomic characters of the Habu are described in Koba, 1962; Takara, 1962; and Nishida et al., 1986. Japanese-language literature reviews may be found in Koba, 1971; and Members of Habu Seminar, 1982, 1983. The Habu is widely feared by residents of the Ryukyu Islands because of the high frequency and severe consequences of its bite. People are most often bitten while tending crops during the day or walking along roads at night, but bites also occur around and inside homes. The severity of the bites is exacerbated by the large size of the Habu; most adults are 1.1-1.7 m in total length. The snake may inject copious quantities of venom, especially if a strike encounters soft tissue such as muscle. The venom causes hemorrhage and necrosis with swelling and subcutaneous bleeding, pro¬ gressing, in severe cases, to unconsciousness and death. In the 1960s, between 200 and 400 (mean 250) persons were bitten in the Amami Islands each year. The human population of the islands was about 124,000, giving a bite frequency of 200 per 100,000 persons per year, with a max¬ imum of 600 per 100,000 per year in the highest-risk locality (Tanaka and Wada,

29

30

Mishima, Tanaka, & Sawai

1977). Antivenin became available in 1905, and the fatality rate declined from 10-15% to 2-4%. After it was recognized that vascular collapse was the proximate cause of death (Tateno et al., 1960), symptomatic antishock treatment lowered the fatality rate to 0.5% or less. At present, reduced human populations in the highest-risk areas and improved clinical therapy have greatly reduced mortality from Habu bites (to ca. 0.25%). Despite its notoriety, the Habu is not aggressive toward humans. It is normally a retiring creature, moving slowly and intermittently in vegetation at night search¬ ing for wild rats. Most bites occur when a human unknowingly thrusts a hand or foot toward a hidden snake, stimulating a defensive strike. Although most human encounters with Habu occur in agricultural fields, Habu readily enter houses. In one sample on Tokunoshima, 48% of the homes had been visited by the snake (Miyashita and Wakisaka, 1979). Frequency of visitation was positively associated with the houses proximity to sugarcane fields and domestic animals, especially fowl (Wakisaka et al., 1979), and appeared to be reduced by masonry walls around the residence (Wakisaka et al., 1979; Mishima et al., 1981).

SIZE OF HABU Newly hatched Habu in a large (>7000) sample of snakes from the Amami Islands averaged 414 mm in total length; the longest adult measured 2.3 m in total length and weighed more than 1kg (Mishima, 1961). The modal class of adult males with uninjured tails had a total length of 1.4-1.5 m (23.1% of the sample). Most males (86.8% of the sample) were in the range 1.2-1.7 m. Comparable values for females were 1.3-1.4 m (29.6% of the sample) and 1.1-1.5 m (86.3%; Table 1.1). The Habu is relatively slender for a viper, and its body mass is pre¬ dictable from its length (Table 1.2; see also Nishimura and Kamura, 1994a). It is

Table 1.1 Frequency distribution of total body length (cm) of adult Habu on the Amami Islands. Males

Females

Total body length (cm)

Number

Frequency (%)

Number

Frequency (%)

110

409

8.2

334

14.5

120

577

11.6

449

19.5

130

866

17.3

679

29.6

140

1153

23.1

521

22.7

150

1113

22.3

227

9.9

160

622

12.5

65

2.8

170

196

3.9

13

0.6

180

37

0.7

6

0.3

190

18

0.4

0

0.0

200

2

0.0

2

0.1

Totals

4993

100

2296

100

Biology of the Habu

31

not known why the coefficients of variability for weights of snakes in the 60 cm class are about twice those of other size classes (Table 1.2). Relative tail length is ontogenetically stable in the Amami Islands (Table 1.3), or shows a slight decline with increasing body size (Okinawajima: Nishimura and Kamura, 1989). Tail length is significantly but not diagnostically shorter in females (14.5 vs. 16.1% of total length). The sex of Habu can be determined on the basis of tail shape: in females, the tail tapers dramatically near the cloaca, whereas the tails of males taper only gradually, presumably to accommodate the hemipenes.

DIET Mishima (1966) sampled 927 prey items (including 43 species in 38 genera of 27 families) from Habu stomachs. The majority (85.9%) were rodents, but birds (6.6%), reptiles (4.9%), amphibians (2.5%), and fishes (0.2%) were also found. Most prey items are eaten head first. Six of the 45 reptile prey items were Habu (cannibalism). Of 796 rodents consumed, 82.5% were Rattus (the species was for¬ merly identified as R. rattus, but karyotypic evidence indicates a cryptic sister taxon, R. tanezumi). On Minnajima, an island lacking rodents, the Habu feeds pri¬ marily on birds (Katsuren et al., 1979). Compared with the sample from the Amami Islands, Habu from Okinawajima ate fewer Rattus (Nishimura et al.,

Table 1.2 Body weight of Habu according to total body length. Males Size (cm)

M

Females SD

N

CV

M

N

SD

CV

20.0

1

20.5

33.5

10

11.6

34.6

45.8

84.2

52.4

17

36.7

70.4

60

14.1

25.2

64.7

49

13.8

21.3

93.3

147

26.3

28.2

98.0

107

20.8

21.2

90

126.5

287

29.2

23.1

132.6

194

30.2

22.8

100

171.6

363

38.3

22.3

183.7

171

80.1

43.6

110

232.8

390

50.4

21.6

254.5

321

67.8

26.6

120

319.9

558

66.8

20.9

346.2

428

88.8

25.7

130

401.8

841

80.7

20.1

435.9

639

106.3

24.4

140

488.3

1103

83.0

17.0

522.0

489

106.5

20.4

160

687.9

609

125.5

18.2

691.0

63

198.8

28.8

170

822.3

191

141.5

17.2

811.9

13

181.6

22.4

180

968.1

37

229.1

23.7

830.0

6

208.9

25.2

190

1135.8

18

135.2

11.9

0.0

0

>200

670.0

2

1620.0

2

Adult Habu

636.3

3749

688.9

1961

3

-

1010.40

-

1009.60

-

1008.80

o 0 _Q

E

3

30

Surface Pressure (mbs)

98

1008.00 2

4

6

8

10

12

14

16

18

20

22

24

Time (2 h increments) Outages

Pressure (mbs)

Figure 4.5 Variation in outages and atmospheric pressure over a 24 hour cycle with data expressed in 2 hour increments for 1978-1994.

than 50% in 1989-1991 (Fig. 4.6). In contrast, the percentage of outages that occurred in the morning hours rose during the years sampled; afternoon outages were unchanged (Fig. 4.6). Outages in morning hours probably represent nocturnal foraging that is prolonged into morning hours.

Variation in Relation to Rainfall To assess the relationship between frequency of power outages and rainfall, we grouped all months from January 1983 through December 1990 into successive quarters. The number of power outages and the amount of rainfall (cm) for each three month period was calculated, and these data were cast into two-by-two tabular formats dividing the periods by either the mean or the median, as shown for the median in Table 4.3. Table 4.3 reveals a correlation between rainfall and frequency of power outages. Below-average rainfall is associated with a belowaverage frequency of power outages, and above-average rainfall is associated with an above-average frequency of power outages. Pearsons product-moment corre¬ lation can be estimated from these tables (r = 0.65 based on means, and r = 0.49 based on medians; both Ps < 0.01, with the latter value being most accurate: Edwards, 1954). Clearly, there is a positive relationship between rainfall and frequency of power outages, but this climatological factor accounts for only 24% of the variation in outage frequencies. Hence, several biological factors such as prey availability, population size, reproductive success, and demography are likely to be affecting snake diel patterns, and thereby oscillations in power outage fre-

Snakes on Electrical Power Lines

—■—

morning

-

-

afternoon

—A—

99

night

Figure 4.6 The percentage of outages occurring in nighttime hours (evening and late night total), morning hours, and afternoon hours (1978-1991). Percentages calculated for each year.

Table 4.3 Distribution of power outages (below and above median frequencies) for 32 three month periods (1983-1990) as a function of rainfall (below and above median levels). Power outages (median frequency per quarter =11) Number of quarters (%) with 11 outages

Number of quarters (%) with 27.5 cm rain

7 (21.8)

9 (28.1)

Rainfall (median per quarter = 27.5 cm)

quencies. Accordingly, concentrating on any single predictive variable is likely to give only a small piece of the picture. It is possible that snakes experiencing reduced prey abundances after 1985 spent more time foraging, and thus caused more outages. In 1984 and 1985, the last individuals of the Guam Rail and Micronesian Kingfisher were removed from the island after precipitous declines in their numbers (Savidge, 1987), and one introduced species, the Black Drongo, is known to have declined in the late 1980s (Fritts, pers. observ.; Guam Div. Aquatic and Wildlife Resources, unpubl. data). The major shift from predominantly nocturnal outages to nocturnal and morning outages after 1984-1985 (Fig. 4.6) corresponded with the period when

100

Fritts & Chiszar

the forest birds disappeared from all of Guam, although these birds had disap¬ peared from central and southern Guam previously. In addition, the continued decline in nocturnal outages after 1985 may reflect continued declines in prey availability, changes in the population structure of the snake, or a relaxation of selection for nocturnality. Rodda et al. (1992) described a major decline in snake density at a site on northern Guam between 1985 and 1988, and declines during the same period were suggested for several other sites, especially on northern Guam (see Rodda et al., this volume, Chap. 17). The collapse of forest bird and introduced mammal populations has likely caused snakes to engage in foraging forays over longer distances and for longer periods, leading snakes to remain active into the morning hours. The failure of power outages to decline, despite efforts by the electrical com¬ pany and despite the decline in northern Guam snake populations, may be the result of several factors. Most important, the decline in prey species (forest birds, introduced rats and shrews, and large geckos) is most conspicuous in remote forested habitats. In the urban, suburban, and roadside areas where power lines are located, poultry, introduced birds, rats, shrews, and pets are present and are preyed on by snakes. Thus, snake populations near humans, and hence near power lines, may not have declined as much as populations in natural areas, where most studies of snake populations have been conducted. The snakes that produce electrical problems represent the smallest size classes (juveniles and subadults), suggesting that the reproductive output of snakes may be higher in suburban areas with large prey compared with forested areas with few large prey. Regardless of the relative distributions of snake outages in rural and urban areas, all evi¬ dence points to continued electrical problems until snake populations are reduced significantly or means are devised for excluding snakes from electrical conductors.

Conical Guy Wire Barriers The guy wire barrier developed at the University of Colorado and tested on Guam consisted of a cone fitted around a guy wire at the Naval Air Station. A 40 cm length of plastic tubing (4 cm diameter) connected the rear of the cone to a box. Hence, a snake that entered the cone could travel through the tube into the box, where it would remain until removed by the experimenter. If a snake climbed over the cone and continued up the guy wire, it was immediately captured by the experimenter. In either case, the snake was placed back into an aluminum can to await its next trial. Since B. irregularis almost always elects to go up when given a choice between up and down, the vast majority of our trials resulted in snakes moving up toward the cone. Considering only the first trial for each of the 25 snakes, 23 tests resulted in captures and only 2 resulted in escapes, with the snakes going over the cone and farther up the guy. Not all captured snakes entered the box, although 21 of them

Snakes on Electrical Power Lines

101

entered the tube. Frequently (11 times), snakes remained in the tube until they were removed by the experimenter. Two snakes entered the cone and remained coiled in its interior for 10 minutes without ever entering the tube. As the trials were repeated (10 times for each snake), we noticed a slight increase in the frequency of balking (i.e., failure to progress up the guy toward the cone). In such cases the snakes either attempted to climb down the guy toward the ground or simply dropped from the guy. Although these events occurred only 18 times out of 250 trials, they never occurred on initial trials, only during the later ones, suggesting that snakes were learning something about the procedure. (We are reluctant to speculate further on what this change in performance implies, as this would require several control conditions not included in our study.) The conical obstruction continued to be effective, in that only 25 of 250 trials resulted in escapes; 2 escapes occurred during the initial trials and 23 more occurred during trials 2-10. Hence, the obstruction was 92% effective in trial 1 and 89% effective thereafter. As with initial trials, many snakes remained in the tube or coiled in the interior of the cone rather than progressing into the box. Also, some snakes entered the cone, remained inside for several minutes, and then began crawling down the guy toward the ground. Although such cases cannot be called “captures,” they represent blockage of upward movement and are therefore indicative of a successful obstruction. While we take these results to be promising, several factors must be kept in mind. First, the 10 minute trial duration was a methodologically necessary but artificial limitation, and we are hesitant to generalize the success of the conical obstruction beyond the parameters of this study. Further experiments are needed under other temporal conditions before such generalizations can be permitted. Second, the fact that many snakes failed to enter the box, remaining either in the tube or in the cone, is troublesome. A maximally effective system would not only block upward movement but would also capture the snakes. Perhaps the capture success can be improved by redesigning the box according to the specifications of the state-of-the-art snake trap described by Rodda et al. in Chapter 20 of this volume. Our conclusion, therefore, is only that conical obstructions that capital¬ ize on the anachoretic instincts of B. irregularis are reasonable candidates for further development as barriers to snake progression on guy wires. They will probably never be 100% effective, but they are by far the most effective barriers tested by the University of Colorado team.

RECOMMENDATIONS FOR REDUCING OUTAGES The data show a strong correlation between the presence of guy wires and the incidence of snakes. In addition, the presence of guy wires influences the number of outages on adjacent poles. The relatively large number of snakes observed crawling on the guy wires of two poles monitored with traps document that snakes were common in the area and that they used guy wires. Behavioral observations

102

Fritts & Chiszar

suggest that snakes find it difficult to climb the vertical surfaces of concrete poles. The low incidence of snakes on poles located significant distances from guy wires lends further support to the conclusion that snakes do not climb concrete poles, or do so only infrequently and under exceptional conditions (i.e., poles with exceptionally rough surfaces and ancillary attachments providing unusual sup¬ port points for the snake). A few outages can be attributed directly to snakes gaining access to transmis¬ sion poles via vegetation close to cross arms on the poles or the conducting wires. Coconut palms (Cocos nucifera) and tangantangan (Leucaena leucocephala) under or adjacent to the X41-X226 circuit constituted the most probable access route to conducting wires on poles that were significant distances from guy wires. The attachment of extra wires (cable TV, electrical distribution, and telephone) accessible to snakes also poses a risk to transmission lines. The data strongly suggest that reductions in outages can be achieved by (1) eliminating guy wires, (2) lowering guy wires (preventing the snakes from reaching conductors), or (3) excluding snakes from guy wires using other barriers. Careful placement of extra wires and removal or trimming of nearby vegetation would similarly reduce access by snakes.

ACKNOWLEDGMENTS The assistance of GPA and NPWC electrical personnel in obtaining data on out¬ ages is appreciated. Oliver Wood and Annette Donner of the GPA were especially helpful and patient in the face of our numerous inquiries. Brian Smith and the personnel of the Guam Division of Aquatic and Wildlife Resources offered valu¬ able assistance during our fieldwork in Guam. Robert Anderson brought the power outages caused by snakes to our attention. Initial fieldwork was partially supported by the Guam Power Authority and the U.S. Department of the Inte¬ rior’s Technical Assistance Program. Robert K. K. Lee developed and tested the conical trap at the University of Colorado, and John Groves helped DC conduct the field tests on Guam. We thank personnel at the Naval Air Station for their hos¬ pitality and assistance.

LITERATURE CITED Chiszar, D. 1990. Behavior of the Brown Tree Snake, Boiga irregularis: A study in

applied comparative psychology. In D. Dewsbury, ed., Contemporary Issues in Comparative Psychology, pp. 101-123. Sunderland, Mass.: Sinauer. Edmunds, M. 1974. Defense in Animals. Harlow, Essex, England: Longman Group. Edwards, A. L. 1954. Statistical Methods for the Behavioral Sciences. New York: Rinehart. Fritts, T. H. 1988. The Brown Tree Snake, Boiga irregularis, a Threat to Pacific

Islands. U.S. Fish Wildl. Serv., Biol. Rep. 88(31).

Snakes on Electrical Power Lines

103

T. H., N. J. Scott Jr., and J. A. Savidge. 1987. Activity of the arboreal Brown Tree Snake (Boiga irregularis) on Guam as determined by electrical power outages. Snake 19:51-58. Fritts, T. FI., N. J. Scott Jr., and B. E. Smith. 1989. Trapping Boiga irregularis on Guam using bird odors. J. Herpetol. 23:189-192. Jackson, J. F. 1988. Crevice occupation by musk turtles: Taxonomic distribution and crevice attributes. Anim. Behav. 36:793-801. Mundy, J. F. 1985. Climatological data for Guam, Mariana Islands. Naval Oceanogr. Command Tech. Notes 85-1:1-41. Rodda, G. H., T. H. Fritts, and R J. Conry. 1992. Origin and population growth of the Brown Tree Snake, Boiga irregularis, on Guam. Pac. Sci. 46:46-57. Savidge, J. A. 1987. Extinction of an island forest avifauna by an introduced snake. Ecology 68:660-668. Fritts,

'

-

Part II

VENOM AND HUMAN HEALTH

In studies of snakebites and their implications for human health, it is important to consider the behavioral ecology of the species in question as well as the pathology of diseases caused by snakebites. For example, the probability that a snake will strike when encoun¬ tering a human has as much influence on the outcome of the en¬ counter as the toxin it injects. Also crucial to the outcome is the capacity of the venom glands and the toxicity of the venom. Symp¬ tomatology (how the toxin affects the victim) is another important component of studies on venom and human health that is analyzed in this section. The close association between Habu and humans results from the Habu’s affinity for areas used by people, including residential buildings and agricultural fields, where Habu find suitable refugia, prey (e.g., rats; Sawai et al., 1990), or both. The Brown Treesnake, however, primarily inhabits brush and woodland areas, where it searches the canopy for prey. Although this snakes habits tend to keep it segregated from humans, Fritts and McCoid (Chap. 6) docu¬ ment recent reports of Brown Treesnakes entering human dwellings and biting sleeping babies. Fritts and Chiszar (this volume, Chap. 4) suggest that drastic declines in natural prey populations on Guam since 1985 have forced Brown Treesnakes to forage more widely and in places closer to dwellings. Habu venom is a hemorrhagic toxin far more toxic than the venom produced by the Duvernoy’s glands of Brown Treesnakes. The Brown Treesnake s venom results in slight neurotoxicity and binding activity at acetylcholine receptors, particularly when the venom is from large specimens (>1 m total length; Weinstein et al., 1993). The great variation in the effects of snake venom is a recur¬ rent theme in the chapters of this section.

LITERATURE CITED Y., Y. Kawamura, and Y. Kurokouchi. 1990. Study on the en¬ vironment related with Habu (Trimeresurus flavoviridis) bite on the Amami Islands of Japan. Snake 22:87-92.

Sawai,

105

106

Part II A., B. G. Stiles, M. J. McCoid, L. A. Smith, and K. V. Kardong. 1993. Variation of lethal potencies and acetylcholine re¬ ceptor binding activity of Duvernoy’s secretion from the Brown Tree Snake, Boiga irregularis Merrem. J. Nat. Toxins 2:187-198.

Weinstein, S.

5 A Historical Outlook on Studies of Habu (Trimeresurus flavoviridis) Bites in the Amami and Okinawa Islands of Japan Yoshio Sawai Yoshiharu Kawamura Yasutetsu Araki Yasuhiro Tomihara

I

n 1905, nine years after cobra antivenin was first produced for clinical use (Calmette, 1896), Kitajima and his colleagues developed and began using Habu antivenin for treating people with Habu bites in the Amami and Okinawa Islands of Japan (Kitajima, 1908). This was the first step in protect¬ ing people from the most severe consequences of Habu bites. In 1959, the re¬ search group of the Institute of Infectious Diseases at the University of Tokyo developed a process for improving Habu antivenin; they purified the antivenin through fractionation with ammonium sulfate and pepsin digestion, and then produced it for treatment of Habu bites (Sawai et al., 1961, 1962). In the mid1960s, studies on the standardization of Habu antivenin were conducted (Kondo et al., 1965a, 1965b). Since 1965, extensive epidemiological studies of Habu bites have been con¬ ducted by several researchers (Sasa et al., 1959; Sawai et al., 1967, 1976, 1990; Sawai, 1973; Wakisaka et al., 1978; Sawai and Kawamura, 1983a, 1983b, 1984a, 1984b, 1985a, 1985b, 1986, 1988, 1989, 1990; Araki and Tomihara, 1988, 1989a, 1989b; Kawamura and Sawai, 1989). Studies of the pathology of Habu bites have included analyses of the venom’s toxicity, the severe necrosis of muscle tissue it can cause, and the actual causes of death among victims of Habu bites (Maeno et al., 1959; Ohsaka et al., 1960; Okonogi et al., 1960). As a second step in combating the severe consequences of Habu bites, researchers produced a prophylactic vaccine (Sawai et al., 1969a). The severe necrosis of muscle tissue that develops rapidly after the victim is bitten is difficult to reverse or recover from, even after repeated injections of antivenin. We concluded that prophylactic vaccines would result in more success with the antivenin therapy in cases where antivenin treatment might be delayed for any reason. From 1965 to 1967, a large-scale field trial was conducted to evaluate the effectiveness of the vaccine. Volunteers from the Amami and Okinawa

107

108

Sawai, Kawamura, Araki, & Tomihara

Islands were given injections of the Habu toxoid. The result was a decrease in the frequency of severe reactions to Habu bites (Sawai et al., 1969b; Sawai 1979). Thereafter, purification, inactivation, and standardization of the toxoid was investigated extensively (Kondo et ah, 1971a, 1971b, 1976). In the meantime, the annual rate of Habu bites on the Amami and Okinawa Islands was decreasing. It was suggested, however, that the third important step in protecting people from Habu bites would be to diminish populations of Habu in areas of human activity, thereby further decreasing the chances of being bitten. In 1977, committees charged with directing researching on Habu control were established on the Amami and Okinawa Islands. The results of the commit¬ tees fundamental research on Habu attractants and traps, the sensory physiology and ecology of Habu, biological controls for Habu, radiotelemetry systems used to track Habu, and environmental factors relevant to Habu bites have been used in field trials of Habu control methods (Ikehara et al., 1978-1982; Sawai and Tanaka, 1978, 1979, 1980-1984; Yoshida, 1983-1991; Tanaka and Hayashi, 1985—1988; Hirosawa and Hayashi, 1989—1992). Electric barrier fences and traps supplied with live rats as attractants appear to be effective for diminishing Habu populations in specific localities.

The remainder of this chapter presents the results of a study we conducted to elu¬ cidate the relationship between the Habu’s behavioral ecology and the temporal and geographical distribution of Habu bites. This type of information should be useful in programs for controlling Habu.

MATERIALS AND METHODS We reviewed epidemiological studies of Habu bites reported during 1987 and 1988 for the Amami and Okinawa Islands (Araki and Tomihara, 1988, 1989a, 1989b; Sawai and Kawamura, 1989, 1990; Sawai et al., 1990). The bites reported in these studies were summarized according to the region and habitat type where they occurred, as well as the time of year and day.

RESULTS Regional Occurrence of Habu Bites There was a notable difference in the occurrence of Habu bites reported from Amamioshima, Tokunoshima, and Okinawa islands during 1987-1988 (Table 5.1). The highest morbidity rate (disease-producing bites per 1000 people) occurred on Tokunoshima, where the human population is smallest and the area planted with sugarcane is relatively large. It appears, therefore, that the relation¬ ship between human population size and the number of Habu bites is slightly weaker than the relationship between hectares of sugarcane and the number of Habu bites (r2 = 0.89 vs. 0.99, respectively).

History of Habu Bites in the Amami and Okinawa Islands

109

Occurrence of Habu Bites by Habitat Type The data illustrate that Habu associate closely with areas also used by humans. Of the 673 Habu bites recorded during the period of this study, 327 (48.6%) took place in agricultural fields, 146 (22%) occurred in residential areas, and 77 (11%) took place on roads. Sugarcane fields constituted more than 70% of the agricul¬ tural sites where Habu bites occurred. In agricultural fields (including footpaths and drainage ditches), most bites occurred while farmers were harvesting sugarcane or hay. In residential locations, Habu bites occurred while people were inside their houses, barns, sheds, or gar¬ dens. People bitten by Habu on roads were either trying to catch Habu or were simply walking. The occurrence of Habu bites by habitat type differed slightly between regions. The percentages of Habu bites occurring in agricultural fields and on roads were greater on the Amami Islands (Amamioshima and Tokunoshima combined) than on Okinawajima, whereas the percentage of Habu bites occurring in residential areas was greater on Okinawajima than on the Amami Islands (Table 5.2).

Table 5.1 Occurrence of Habu bites in Japan, by region, 1987-1988. Amamioshima

Tokunoshima

Okinawa

Total (Ave.)

103*

188

382

673

80,924

32,443

1,105,572

1,218,939

Bites/1000 people

1.27

5.79

0.35

(0.55)

Sugarcane (ha)

1,028

5,751

12,906

19,685

Bites/ha sugarcane

0.100

0.033

0.030

(0.034)

Total bites Human population

* Three fatal cases.

Table 5.2 Occurrence of Habu bites in Japan, by habitat type, 1987-1988. Amamioshima & Tokunoshima

Okinawa

Habitat type where bite occurred Sugarcane only Other agricultural All agricultural

Total No. bites

% bites

No. bites

% bites

no. bites

117

125

242

51

34

85

168

57.7%

159

41.6%

327

In houses

18

25

43

Other residential

39

64

103

All residential areas

57

19.6%

89

23.3%

146

On roads

37

12.7%

40

10.5%

77

Other

29

10.0%

95

24.8%

124

Totals

291

100.0%

382

100.0%

673

Note: Percentage values express distribution of bites within each island group.

110

Sawai, Kawamura, Araki, & Tomihara f

Seasonal Patterns of Habu Bites More people were bitten by Habu during the warmer months (March-October), when Habu are most active, than at other times (Table 5.3, Fig. 5.1). There were three distinct peaks in the frequency of Habu bites during this period. The first peak occurred in March, when sugarcane was being harvested; the second peak occurred during May and June—including the Habu breeding season; and the third peak occurred in October, when cane fields were being weeded.

Time-of-Day Patterns of Habu Bites In general, Habu are nocturnal in their habits. Nevertheless, 462 of 673 bites (68.7%) occurred during daylight hours (0600-1800), and 198 (29.4%) happened during hours of darkness (1800-0600) (Table 5.4). A further breakdown of these data shows that 90.5% of the bites that occurred in agricultural fields happened during daylight hours, and a smaller fraction of the bites that occurred in resi¬ dential areas (43.8%) and on roads (22.1%) took place during nighttime hours.

NUMBER OF BITES

In agricultural fields, farmers were most often bitten when they accidentally

Figure 5.1 Location of Habu bite incidents by time of year in the Amami and Okinawa Islands (1987-1988). The seasons of weeding and harvesting sugarcane are indicated by arrows.

History of Habu Bites in the Amami and Okinawa Islands

111

Table 5.3 Occurrence of Habu bites in Japan, by month, 1987-1988. No. bites by habitat type Total bites Agricultural

Residential

Month

(%)

January

18 (2.7)

5

3

February

25 (3.7)

15

March

62 (22.7)

April

Roads

Other 7

0

3 2

48

2

4

49 (7.3)

16

7

4

3 22

May

95 (14.1)

51

26

9

9

June

83 (12.3)

45

16

10

12

July August

47 (7.0)

17

14

7

9

42 (6.2)

11

13

9

9

September

74 (11.0)

28

24

9

14

107 (15.9)

60

28

14

5

November

45 (6.7)

20

9

4

12

December

26 (3.9)

11

4

2

9

673 (100.0)

327

146

77

123

October

Totals

8

Table 5.4 Occurrence of Habu bites in Japan, by time of day, 1987-1988. No. bites (%) by habitat type Time

Total bites (%)

Agricultural

Roads

Other

38 (26.2)

14 (18.2)

Residential

0000-0600

56 (8.3)

0600-1200

218 (32.4)

136 (42.2)

34 (23.3)

10 (13.0)

3 64

1200-1800

244 (36.3)

156 (48.3)

30 (20.5)

7(9.1)

51

1800-2400

142 (21.1)

29 (9.0)

43 (29.5)

45 (58.4)

25

Unknown Totals

13 673 (100.0)

1 (0.3)

5

1

1

327

146

77

143

stepped on Habu that were hiding in their daytime refugia under the vegetation. At night, when Habu leave their refugia to forage, many probably left the agri¬ cultural fields and invaded residential buildings, possibly explaining the higher incidence of bites in residential areas during hours of darkness. The incidence of bites in gardens, barns, livestock pens, and residential dwellings during daylight hours (64), however, indicates that Habu also spend their daytime hours hiding in refugia afforded by these structures. The higher frequency of bites inflicted on people using roads at night is probably due to snakes using roads more during hours of darkness and to their reduced visibility to humans at night. Fifty-nine villagers were bitten while on a road at night as they walked along or tried to catch Habu. Analyses of Habu habits and habitats will lead to an understanding of where the Habu are coming from as well as when and where bites are likely to occur. We found that large numbers of Habu inhabit fields, where their favored prey

112

Sawai, Kawamura, Araki, & Tomihara

(rats) nest under the detritus of sugar cane. We also found that Habu use suitable refugia in rock walls made of coral and under hedges planted around residences.

DISCUSSION Overall, Habu control programs are effective for decreasing the probability that people will be bitten or will suffer severely if bitten by Habu. The results described in this chapter point to additional means of minimizing human-Habu contact. For example, the data indicate that agricultural settings, particularly sugarcane fields, are the places where people are bitten most frequently. Further evaluations of these data revealed that poor field maintenance creates better Habu habitat and leads to higher numbers of Habu inhabiting the field. Population densities of Habu are greater in unkempt fields than in well-maintained fields. Thus, it is important to remove from agricultural fields all materials that provide suitable hiding places for Habu, to remove rat nests, and to maintain the sugarcane fields, including the footpaths and drainage ditches associated with them. Farmland managed through the Farm Land Improvement Project (carried out on the Amami Islands) has been effective in removing Habu completely (Mishima et al., 1984b, this volume, Chap. 28). Electric barrier fences and Habu traps have also proved to be effective in preventing invasions of Habu (Hayashi et al., 1979). To minimize the number of bites occurring in residences, it is important to keep gardens tidy, to clean out barns and shacks, and to create a field edge bar¬ rier by converting rock or hedge fences to concrete block walls. This eliminates hiding places used by Habu (Mishima et al., 1984a, this volume, Chap. 28). Most hunting for Habu is done by professional Habu catchers along roads of the Amami Islands, but Habu should be hunted in agricultural fields and resi¬ dential areas as well. Another important part of any Habu control program should be the eradication of rats so that Habu will not be encouraged to reinvade fields and homes.

LITERATURE CITED Araki,

Y.,

and

Y.

Tomihara.

1988. Study on the Habu bites in 1987 in Okinawa

Islands of lapan [in Japanese], In C. Yoshida, ed., Reports of Studies to Develop Antivenin for Habu, 1987, pp. 33—44. Ozato, Okinawa, Japan: Okinawa Prefecture Dept, of Health. -. 1989a. Study on the Habu bites in 1988 in Okinawa Islands of Japan [in Japanese]. In C. Yoshida, ed., Reports of Studies to Develop Antivenin for Habu, 1988, pp. 21-32. Ozato, Okinawa, Japan: Okinawa Prefecture Dept, of Health. -. 1989b. Snake venom and snakebite. In M. Matsui, T. Hikida, and R. C. Goris, eds., Current Herpetology in East Asia, pp. 482-486. Kyoto: Herpetological Society of Japan.

History of Habu Bites in the Amami and Okinawa Islands Calmette,

113

A. 1896. Le venin des serpents, physiologie de Penvenomation. Traitment

des morsures venimeuses par le serum des animaux vaccines, Paris: Societe d’Editions Scientifiques. Hayashi, Y., Y. Noboru, Y. Wada,

and

H.

Tanaka.

1979. The method to prevent the

invasion of Habu {Trimeresurus flavoviridis) by an electric fence [in Japanese with English summary]. Snake 11:58-62, 127-128. Hirosawa,

K.,

and

Y.

Hayashi,

eds. 1989-1992. [Series of papers with various titles,

in Japanese]. In Report of Studies about Attraction and Capture of Habu (1988-1991). Kagoshima, Japan: Kagoshima Prefecture. S.,

Ikehara,

eds. 1978-1982. [Series of papers with

K. Teruya, and C. Yoshida,

various titles, in Japanese]. In Reports of Ecological Researches to Diminish Bites of Habu (Trimeresurus flavoviridis) in Okinawa Prefecture, vols. 1-5). Ozato, Okinawa, Japan: Habu Study Sect., Okinawa Prefectural Inst, of Health and Env. Y.,

Kawamura,

and

Y.

Sawai.

1989. Habu {Trimeresurus flavoviridis) bites on the

Amami Islands in 1985. Snake 21:108-110. Kitajima,

T. 1908. On “Habu” venom and its serum therapy. Philippine J. Sci.

3:151-164. Kondo, H., S.

Kondo, S. Sadahiro, K. Yamauchi,

A.

Ohsaka, and R. Murata.

1965a. Standardization of antivenine. 1. A method for determination of antilethal potency of Habu antivenine. Japan. J. Med. Sci. Biol. 18:101-110. -. 1965b. Standardization of antivenine. II. A method for determination of antihemorrhagic potency of Habu antivenine in the presence of two hemorrhagic principles and their antivenine. Japan. J. Med. Sci. Biol. 18:127-141. Kondo,

H.,

R.

S. Sadahiro, and

Murata.

1976. Relationship between the amount

of Habu toxoid injected into the monkey and the resulting antitoxin titer in the circulation. In A. Ohsaka et al., eds., Animal, Plant and Microbial Toxins, vol. 2, pp. 431-438. New York: Plenum Press. Kondo, H., S. Sadahiro, K. Yamauchi, S.

Kondo, and

R.

Murata.

1971a.

Preparation and standardization of toxoid from the venom of Trimeresurus flavoviridis (Habu). Japan. J. Med. Sci. Biol. 24:281-284. Kondo,

H., S.

Sadahiro,

K.

Yamauchi, and

R.

Murata.

1971b. Immunogenicity

of a toxoid of Habu (Trimeresurus flavoviridis) venom in various animal species including man. Japan. J. Med. Sci. Biol. 24:365-377. Maeno,

H., M.

Morimura,

S.

Mitsuhashi,

Y.

Sawai, and

T.

Okonogi.

1959.

Studies on Habu snake venom 2b. Further purification and enzymic and biological activities of Ha-proteinase. Japan. J. Microbiol. 3:277-284. Mishima,

S., Y.

Hayashi, and

Y.

Sawai.

1984a. Effect of the improvement of the

habitual environment on the population of Habu, Trimeresurus flavoviridis. Snake 16:115-118. Mishima, S., Y.

Sawai,

Y.

Chigusa, and

H.

Yamamoto.

1984b. Studies on control

of Habu {Trimeresurus flavoviridis). III. An analysis of the effect of land improve¬ ment on the population of Habus on the Tokunoshima Island by the information obtained through questionnaire [in Japanese with English summary]. Snake 16:63-70.

114

Sawai, Kawamura, Araki, & Tomihara

Ohsaka,

A., H.

Ikezawa,

S.

Kondo, and

N.

Uchida.

1960. Hemorrhagic activities

of the Habu snake venom and their relations to lethal toxicity, proteolytic activi¬ ties and other pathological activities. Br. J. Exp. Pathol. 41:478-486. Okonogi,

T.,

S. Hoshi, M. Honma, S. Mitsuhashi, M. Maeno, and Y. Sawai.

1960. Studies on the Habu snake venom. Ill-(2). A comparative study of histopathological changes caused by crude venom, purified Habu-proteinase and other proteinase. Japan. J. Microbiol. 4:189-192. Sasa,

M.,

K. Teruya, H. Uchiyama. and S. Iwai.

1959. Epidemiology of the poiso¬

nous snake bite in the Amami and the Ryukyu Islands. Japan. J. Exp. Med. 29:417-444. Sawai,

Y. 1973. Snakebites in Asia [in Japanese with English abstract]. Snake 5:29-75.

-. 1979. Vaccination against snake bite poisoning. In C. Y. Lee, ed., Handbook of Experimental Pathology. Vol. 52, Snake Venom, pp. 881-897. Berlin: SpringerVerlag. Sawai, Y., and Y. Kawamura. 1983a. Habu (Trimeresurus flavoviridis) bites on the Amami Islands of Japan in 1977. Snake 15:1-16. -. 1983b. Habu (Trimeresurus flavoviridis) bites on the Amami Islands of Japan in 1978. Snake 15:75-80. -. 1984a. Habu (Trimeresurus flavoviridis) bites on the Amami Islands of Japan in 1979. Snake 16:1-6. -. 1984b. Habu (Trimeresurus flavoviridis) bites on the Amami Islands of Japan in 1980. Snake 16:85-89. -. 1985a. Habu (Trimeresurus flavoviridis) bites on the Amami Islands of Japan in 1981. Snake 17:1-5. -. 1985b. Habu (Trimeresurus flavoviridis) bites on the Amami Islands of Japan in 1982. Snake 17:91-95. -. 1986. Habu (Trimeresurus flavoviridis) bites on the Amami Islands of Japan in 1983. Snake 18:65-69. -. 1988. Habu (Trimeresurus flavoviridis) bites on the Amami Islands of Japan in 1984. Snake 20:93-97. -. 1989. Habu (Trimeresurus flavoviridis) bites on the Amami Islands of Japan in 1987. Snake 21:1-5. -. 1990. Habu (Trimeresurus flavoviridis) bites on the Amami Islands of Japan in 1988. Snake 22:1-7. Sawai,

Y., Y.

Kawamura,

I.

Ebisawa,

T.

Okonogi,

Z.

Hokama, and

M.

Yamakawa.

1967. Studies on the improvement of Habu (Trimeresurus flavoviridis) bites. 6. Habu bites on the Amami and Ryukyu Islands in 1964. Japan. J Exp Med 37:51-59. Sawai,

Y., Y.

Kawamura,

T.

Fukuyama, and

T.

Okonogi.

1969a. Studies on the

improvement of Habu (Trimeresurus flavoviridis) bites. 7. Experimental studies on Habu venom toxoid by dihydrothioctic acid. Japan. J. Exp. Med. 39:109-117. Sawai,

Y., Y.

Kawamura,

T.

Fukuyama,

T.

Okonogi, and I. Ebisawa.

1969b.

Studies on the improvement of Habu (Trimeresurus flavoviridis) bites. 8. A field trial of the prophylactic vaccination of the Habu venom toxoid. Japan. J Exp Med 39:197-203.

History of Habu Bites in the Amami and Okinawa Islands Sawai,

Y., Y.

Kawamura, and

Y.

Kurokouchi.

115

1990. Study on the environment

related to Habu (Trimeresurus flavoviridis) bites on the Amami Islands of Japan. Snake 22:87-92. Sawai, Y., M. Makino, Y. Kawamura,

Z.

Hokama, and

M.

Yamakawa.

T.

Fukuyama, H. Chinzei,

T.

Okonogi,

1976. Epidemiological study of Habu bites on

the Amami and Okinawa Islands of Japan. In A. Ohsaka et al., eds., Animal, Plant and Microbial Toxins, pp. 439-450. New York: Plenum Press. Sawai, Y., M. Makino, S. Miyasaki, K. Kato,

T.

Okonogi.

H.

Adachi, S, Mitsuhashi, and

1961. Studies on the improvement of treatment of Habu snake

(Trimeresurus flavoviridis) bites. 1. Studies on the improvement of Habu snake antivenin. Japan. J. Exp. Med. 31:137-150. Sawai, Y.,

M.

Makino,

I.

Tateno, T. Okonogi, and

S.

Mitsuhashi.

1962. Studies

on the improvement of Habu snake (Trimeresurus flavoviridis) bites. 3. Clinical analysis and medical treatment of Habu snake bite. Japan. J. Exp. Med. 32:117-138. Sawai,

Y.,

and

H.

Tanaka,

eds. 1978. [Series of papers with various titles, in

Japanese with English abstracts]. Snake 10:1-104. -. eds. 1979. [Series of papers with various titles, in Japanese with English summary]. Snake 11:1-133. -. eds. 1980-1984. Reports of studies on control of Habu (Trimeresurus flavoviridis) on the Amami Islands in 1979-1983 [in Japanese]. Tanaka,

H.,

and

Y.

Hayashi,

eds. 1985-1988. Report of Studies to Develop

Attractant for Habu (1984-1987) [in Japanese]. Kagoshima, Japan: Kagoshima Prefecture. Wakisaka,

I.,

M. Miyashita, T. Ando, and

A.

Takano.

1978. An epidemiological

study of Habu bites in the Amami Islands, Japan. J. Hyg. 33:606-612. Yoshida,

C., ed. 1983-1991. [Series of papers with various titles, in Japanese]. In C.

Yoshida, ed., Reports of Ecological Researches to Diminish Bites of Habu (Trimeresurus flavoviridis) in Okinawa Prefecture. Ozato, Okinawa, Japan: Habu Study Sect., Okinawa Prefectural Inst, of Health and Env.

6 The Threat to Humans from Snakebite by Snakes of the Genus Boiga Based on Data from Guam and Other Areas Thomas Michael

H. J.

Fritts

McCoid

T

he danger posed to humans by the bite of snakes of the genus Boiga has received little attention, although most authors acknowledge that these snakes

are mildly venomous and could be dangerous to humans (e.g., De Silva and Aloysius, 1983; Fritts, 1988). In part, the lack of epidemiological interest in Boiga bites stems from the reduced risk these snakes pose relative to members of the snake families Elapidae and Viperidae with which most Boiga species are sympatric. The bites of eight species of Boiga are known to cause or potentially cause medically important symptoms in humans (see Fritts et ah, 1994, for summary), but considerable additional information is needed to assess the risks of human death or serious injury. Because it is the only large snake present on Guam, the Brown Treesnake, Boiga irregularis, provides an unusual opportunity to assess the medical importance of this species. In this chapter we summarize epidemiological information on B. irregularis, note parallels between bites of this species and those of snakes of the elapid genus Bungarus, and present evidence suggesting that dangerous bites from species in the genus Boiga may have been erroneously attributed to sympatric snakes, including Bungarus.

THE MEDICAL SIGNIFICANCE OF THE BROWN TREESNAKE Little was known about the symptoms and signs resulting from snakebite by the Brown Treesnake until intensive studies of this snake were initiated in Guam. Ref¬ erences to the venom and human risk posed by the Brown Treesnake in its native range (Australia, Indonesia, New Guinea, and Solomon Islands) characterize the venom as not very toxic (McCoy, 1980), not dangerous to man (Cogger, 1992; Ehmann, 1992); or causing moderate pain or stinging (Parker, 1983). Nearly all references mention the snake’s pugnacious disposition and tendency to strike repeatedly when threatened (e.g., McCoy, 1980; Cogger, 1992). Brown Treesnakes can be locally abundant in New Guinea (Parker, 1983) and Australia (Kinghorn, 1964), but nowhere in the native range is the species as abun¬ dant as it is in most areas of Guam (Rodda et al., 1992). The Brown Treesnake is the only large snake present on the island (Ramphotyphlops, a miniscule subter-

116

Boiga Bites on Guom

117

ranean species, is the only other snake), and any visits to emergency medical facilities for snakebite on Guam are due to B. irregularis. Approximately 1 in 1200 emergency room visitors in Guam are seeking treatment for a Brown Treesnake bite (Fritts et al., 1990). To date, we have information on more than 160 snakebites for which medical attention was sought; most are described in questionnaires completed by attend¬ ing physicians, emergency room nurses, or public health officials. This informa¬ tion constitutes a unique resource for study of the attributes associated with the bite of a rear-fanged snake; we know of no comparable data set for any rear-fanged colubrid snake, even though many species are considered potentially dangerous to man. The information solicited on the questionnaire was described by Fritts et al. (1994). The present analysis is based on a total of 166 bites occurring from 26 September 1989 through 4 September 1993; 94 of these bites were summarized by Fritts et al. (1994). The situations in which snakebites occur in Guam are exceptional; a prepon¬ derance of bites involve victims bitten while sleeping in their homes. The persons bitten represent an equal sex ratio of males to females. The site of the bite is often the fingers and hand, but bites also occur on the face, neck, body trunk, and gen¬ itals. In addition, infants constitute a disproportionately high number of the chil¬ dren taken to emergency medical facilities in Guam, suggesting that the snake is selecting infants over older children on the basis of yet unknown cues (Fritts et al., 1994). One explanation of the high incidence of unusual bites is that snakes in environments with extremely scarce prey may be motivated (futilely) to feed on small children and small body parts of adults (Fritts et al., 1994).

Motivation for Feeding Brown Treesnakes respond to chemical cues involving blood and birth products (Chiszar et al., 1993). Attacks on newborn puppies, rabbits, and other domestic animals are common in Guam (R. Donner, D. V. M., pers. comm., 1989), and indeed the benefits are obvious for snakes that prey on litters of newborn or young mammals. Neonatal mammals often occur in groups, are relatively unlikely to escape, and are less likely to actively defend themselves against predation. In 11 of 147 cases of snakes biting sleeping humans in Guam, the victim (usually an infant but all under four years old) was being constricted by the snake when discovered by a family member (Fritts et al., 1994; and Fritts, unpubl. data). Other factors such as the high frequency of multiple bites (evidence of prolonged chewing) and the selection of the smallest infants support this hypothesis (Fritts et al., 1994).

Examples of Serious Brown Treesnake Bites Eleven cases of snakebite in Guam have involved serious signs and symptoms, although no fatalities are documented. All the victims were in bed when bitten.

118

Fritts & McCoid

Most bites occurred between dusk and dawn. In 2 of 11 cases, the infant bitten was sleeping between parents or was with older siblings who were not bitten. All the children with serious signs were less than a year old (7 days-10 months; X = 2.9 months). Estimates of the size of the snake were approximations, but snakes associated with serious bites were larger (1.17 m vs. 0.82 m) than snakes involved in bites of other children less than six months of age for which no serious symptoms were reported. Relative to Brown Treesnakes in Australia and the rest of native range, Guam snakes tend to be somewhat larger (Rodda et al., this volume, Chap. 2), possibly increasing the severity of bites (see “Amount and Composition of Venom” later in this chapter) and the predatory pressure on large snake prey. Snakes larger than 1 m are predisposed to seek endothermic prey (Savidge, 1988), and such prey are now relatively scarce on Guam except near villages, farms, and human population centers (Savidge, 1988; Fritts and McCoid, 1991; Rodda et al., this volume, Chap. 2). In the 11 cases discussed here, large, hungry snakes may have responded to the small children with feeding behaviors, injecting more Duvernoys secretion than that likely to be injected in defensive bites, and producing bites warranting medical attention. The severe symptoms did not become evident until one to six hours after the bite. In 10 of 11 cases, the victims were hospitalized, and 9 victims received ventilation or intubation to assist breathing.

PARALLELS BETWEEN BOIGA AND BUNGARUS EPIDEMIOLOGY Kraits (Bungarus) are not closely related to catsnakes (Boiga). Like other elapids, kraits have fixed front fangs and highly toxic venom. Although they eat a variety of vertebrates, they are generally thought to be specialized for ophiophagy (snake eating; Mao, 1970; De Silva, 1992). They are morphologically dissimilar from catsnakes as well, generally possessing heads only slightly larger than their necks, banded coloration, and a flattened or triangular body shape. Like catsnakes, kraits are nocturnal. Several similarities exist between the patterns of snakebites seen for Brown Treesnakes in Guam and those reported for kraits (Bungarus spp.) in Sri Lanka and the Indian subcontinent (De Silva, 1987). In areas where both types of snakes occur, some chance exists for bites by Bungarus and Boiga to be confused.

Invasion of Houses In Guam, 79% of the 284 persons responding to a questionnaire reported seeing snakes in or near their homes. Approximately 80% of all treated snakebites in Guam occur while the victim is sleeping indoors. In Sri Lanka, nearly all (about 90%) bites involving kraits take place inside homes while the victim is sleeping (De Silva, 1980, 1992). In one sample of 19 people bitten by kraits, all the victims were sleeping indoors (De Silva, 1981). Similarly, bites by Bungarus in India

Boiga Bites on Guam

119

almost always occur in homes while the victim is sleeping (Saha and Hati, 1983; Hati et al., 1988). In one sample from India, 15 of 17 bites (88%) involved vic¬ tims sleeping in their homes (Hati et al., 1988). In one large sample of venomous snakebites on Taiwan, a krait (Bungarus multicinctus) accounted for 18% of bites for which the snake was identified, and overall (all taxa) 29% of bites occurred inside residences (Sawai and Tseng, 1969).

Motivation for Bites Snakes of both genera are known to strike and hold onto the victim rather than striking and releasing, as most snakes do when biting as a defensive behavior. Kraits reportedly hold on when they bite humans and in some cases must be pried off the victim (De Silva, 1992). Of 22 kraits brought in with snakebite victims, 9 had one broken fang (De Silva, 1992), perhaps from holding onto a struggling victim. One sample of kraits involved in snakebites showed a preponderance (91%) of male snakes, most of which had empty stomachs (De Silva, 1992). These circumstances suggest that kraits that bite humans in nondefensive situations are attempting to feed. In the daytime, kraits are known to be relatively mild-man¬ nered, nonaggressive snakes (De Silva, 1992).

Parts of Body Bitten Bungarus (De Silva, 1981; Hati et al., 1988) and Boiga bites are similar in the parts of the body bitten. Most cobra and viper bites occur when people contact the snakes while walking or working outdoors; 75% of bites by these snakes in Sri Lanka were on toes, feet, and lower legs, but less than 1% were on the head or trunk (De Silva, 1981). In contrast, the head and trunk were the sites of 22% and 14% of Bungarus and Boiga bites, respectively (Fig. 6.1). In one sample of 27 krait bites, 48% were on the head and trunk, 48% on the upper arm, and 4% on the thigh (De Silva, 1992). This pattern is also discernible for Boiga in Guam, where most Boiga bites involve sleeping victims, with up to 21% of the bites on the head and trunk. Active victims in Guam had few bites on these parts of the body and showed a higher incidence of bites on the feet and lower extremities (Fig. 6.2). Differences between the body sites bitten in Sri Lanka and Guam may reflect cultural and economic differences between Guam and Sri Lanka. The higher incidence of Boiga bites on feet and lower extremities in Sri Lanka may reflect a rural population with less access to protective footgear (De Silva, 1981). Relative to Guamanians, Sri Lankans probably have fewer air-conditioned homes and therefore more often sleep with their lower extremities uncovered. Bites on hands and fingers in active victims may stem from placing the hand on or near a snake, but the equally numerous bites on the hands of sleeping victims are more likely

120

Fritts & McCoid 90 80

(/>

0 70

Upper Extremity ■ Boiga spp.

Head/Trunk/Thigh

Lower Extremity

MBungarus spp. □ Cobras/Vipers

Figure 6.1 Parts of body where snakebites were recorded by De Silva (1981) in Anuradhapura District, Sri Lanka. Sample sizes are Boiga spp., 52; Bungarus spp., 53; and cobras/vipers, 248.

80

■ Active Guam ^ Asleep Guam □ Sri Lanka Figure 6.2 Parts of body where Boiga spp. snakebites were recorded in Guam (data from Fritts et al., 1994, and this study) and in Sri Lanka (data summarized from De Silva, 1976a). Samples sizes are active victims, Guam, 19; sleeping victims, Guam, 39; and all victims, Sri Lanka, 52.

Boiga Bites on Guam

121

V) 1.3 m total length) not only produce more Duvernoy’s secretion than other rear-fanged snakes and smaller conspecifics, but also possess a more toxic secretion (Vest et al., 1991; Weinstein et al., 1991,1993; Chiszar et ah, 1992). These studies also sug¬ gested the presence of a neurotoxic component in the venom, perhaps account¬ ing for the clinical observations of drooping eyelids (ptosis) and poor muscle tone in several small infants after bites (Fritts et ah, 1994). Even though the reports available to us do not enable us to verify that neurotoxicity occurred, it is notable that several reports and physicians in Guam independently mentioned signs attributable to a neurotoxic effect. The presence of visible signs of neurotoxic envenomation is another factor that could contribute in Sri Lanka and elsewhere to bites by Boiga being attributed to Bungarus spp., the latter being well known for having neurotoxic venoms (De Silva, 1992). In some cases, biologists assume that fatal bites attributed to Boiga (especially B. forsteni) actually involved kraits or cobras (De Silva, 1976a). The severity of elapid and viperid bites is well documented, and the lack of comparable infor¬ mation for colubrids may contribute to the common attribution of all serious bites to Elapidae or Viperidae. In tropical regions with diverse snake faunas, the percentage of cases treated in medical facilities for snakebite by an unknown species varies widely: Sri Lanka, 44% (De Silva, 1972), 82% (De Silva and Hewage, 1987); Malaysia, 26% (Singh, 1980); various portions of Southeast Asia, 60-90% (Sawai et ah, 1972); and China, 20% (Yu et al., 1989).

Bites by Boiga in Sri Lanka Of 519 fatal snakebite cases reviewed by De Silva (1976a), 15% were attributed to snakes other than elapids and viper ids (most to Boiga forsteni), but De Silva assumed that some of these fatalities were due to bites by kraits and vipers that were misidentified. Boiga forsteni is one of the most feared of all snakes in Sri Lanka, but based on personal experience with one or more bites by this species, De Silva (1976b; De Silva and AJoysius, 1983) considered it to have a mild venom. It is probable that De Silva was speaking from experience with defensive bites, but the bites on the snakebite victims could have involved feeding bites, in which larger quantities of venom were injected, causing more severe symptoms. Of 393 snakebites reported in Anuradhapura District of Sri Lanka (De Silva, 1981), 86 resulted in deaths (approximately 18 fatalities per 100,000 inhabitants). In this data set, about equal numbers of bites were attributed to Boiga and to the noto¬ riously dangerous kraits, Bungarus spp. (53 vs. 56, or 14.0% vs. 14.9% of all bites, respectively); however, although 33% of all fatalities occurred from bites attrib¬ uted to kraits, no fatalities were linked to Boiga in this study. Three species of Boiga are common near and in dwellings in the region: B. trigonata, B. ceylonenesis, and

124

Fritts & McCoid

B. forsteni. The frequency with which people seek treatment for bites attributed to Boiga and the remarkable fear they have of these snakes may justify increased scrutiny of the identifications assigned to snakes involved in snakebite cases in Sri Lanka and other areas where Boiga occurs. Several aspects of the snakebite cases attributed to Boiga in Sri Lanka closely match those seen in B. irregularis bites in Guam. The preponderance of bites on the head and trunk (Fig. 6.3) and the snakes’ tendency to enter homes and attack sleeping victims are suggestive of animals motivated to feed on hopelessly large human prey. In light of the difficulty in definitively attributing snakebites to par¬ ticular snake species, it may be appropriate to treat all Boiga envenomations as potentially serious.

Victims Most at Risk from Boiga Bites Based on the available information, special attention may be warranted when one or more of the following risk factors is present following bites by the Brown Treesnake or other Boiga species: L 2. 3. 4.

Young children (1.3m) Evidence of chewing or multiple bites

5. Victim displays signs of: drooping eyelids, poor muscle tone or coordi¬ nation, depressed heart rate, respiratory distress, prominent swelling and discoloration beyond the actual site of the bite The symptoms and signs displayed by victims bitten by the Brown Treesnake exhibit considerable variation. We suggest that medical personnel use the above guidelines to deal with cases most likely to experience serious signs and symp¬ toms. We also suggest that reptile specialists and medical personnel studying the epidemiological patterns in snakebite in Southeast Asia and other areas where species of the genus Boiga occur attempt to gather more precise data on circum¬ stances, symptoms, and severity of bites involving these snakes.

SUMMARY The Brown Treesnake, Boiga irregularis, is a rear-fanged (opisthoglyphic) colubrid snake whose venom is not considered life-threatening in its native range. On Guam, however, this introduced snake is exceptionally abundant, and potentially atal bites to small children occur relatively frequently. The recognition that this snake can be responsible for serious bites is aided by the absence of other snakes on Guam; all Guam snakebites can be attributed to B. irregularis. A peculiar char¬ acteristic of serious Brown Treesnake bites is that they occur primarily to children s eepmg in bed at night. Several lines of evidence suggest that Brown Treesnakes

Boiga Bites on Guam

125

bite sleeping children while seeking prey rather than as a defense against active human contact (as is characteristic of most snakebites). Such offensive bites are extremely rare, but they are known from the elapid genus Bungarus. Epi¬ demiological parallels between the bites of Bungarus and Boiga extend to the age and behavior of victims, the site of envenomation, and the behavior of the snakes.

ACKNOWLEDGMENTS We thank the staffs of the Guam Memorial Hospital and Naval Hospital for shar¬ ing information. We are also indebted to Dr. Robert Haddock and Lisa Close for assistance in gathering information of selected cases. We are grateful to Sherman Minton and David Hardy for information they generously shared on snakebite and medical phenomena. Portions of this work were supported by the U.S. Fish and Wildlife Service and the U.S. Department of the Interior’s Technical Assis¬ tance Program. We thank Carolyn Imamura, Phil DeLongchamps, Ernest Kosaka, and Linda Wolfe for their continued support of our snake studies.

LITERATURE CITED Chiszar, D., T. M. Dunn, and H. M. Smith. 1993. Response of Brown Tree Snakes

(Boiga irregularis) to human blood. J. Chem. Ecol. 19:91-96. Chiszar, D., S. A. Weinstein, and H. M. Smith. 1992. Liquid and dry venom yields from Brown Tree Snakes, Boiga irregularis (Merrem). In P. D. Strimple and J. L. Strimple, eds., Contributions in Herpetology, pp. 11-13. Cincinnati, Ohio: Greater Cincinnati Herpetological Society. Cogger, H. G. 1992. Reptiles and Amphibians of Australia, 5th ed. Ithaca: Cornell Univ. Press. de Silva, A. 1972. Identification of venomous snakes of Sri Lanka. Loris 12:288-293. -. 1976a. The pattern of snake bite in Sri Lanka. Snake 8:43—51. -. 1976b. Venomous snakes of Sri Lanka. Snake 8:31-42. -. 1980. Snakebites and antivenom treatment in Sri Lanka. Snake 12:134-137. -. 1981. Snakebites in Anuradhapura District. Snake 13:117-130. -. 1983. Moderately and Mildly Venomous Snakes of Sri Lanka. Colombo: Ceylon Medical Society. -. 1987. Some epidemiological and clinical aspects of Bungarus caeruleus bite. Proc. Kandy Soc. Med. 10:113-115. -. 1990. Colour Guide to the Snakes of Sri Lanka. Portishead, England: R 8c A Publications. -. 1992. Bungarus caeruleus: Its ecology and bite in Sri Lanka. In P. Gopalakrishnakone and C. K. Tan, eds., Recent Advances in Toxicology Research, vol. 1, pp. 746-760. Singapore: National Univ. Singapore. de Silva, A., and D. J. Aloysius. 1983. Moderately and mildly venomous snakes of Sri Lanka. Ceylon Med. J. 28:122-123.

126

Fritts & McCoid

A., and P. Hewage. 1987. Snake bite in Mahaweli System Proc. Kandy Soc. Med. 10:115.

de Silva,

H

area: Gal Newa.

H. 1992. Encyclopedia of Australian Animals. Reptiles. Sydney: Angus & Robertson.

Ehmann,

T. H. 1988. The Brown Tree Snake, Boiga irregularis, a Threat to Pacific Islands. U.S. Fish Wildl. Serv., Biol. Rep. 88 (31).

Fritts,

T. H., and M. J. McCoid. 1991. Predation by the Brown Tree Snake, Boiga irregularis, on poultry and other domesticated animals in Guam. Snake 23-7580.

Fritts,

M. J. McCoid, and R. F. Haddock. 1990. Risks to infants on Guam from bites of the Brown Tree Snake (Boiga irregularis). Am. J. Trop. Med Hye 42:607-611.

Fritts, T. H.,

. 1994. Symptoms and circumstances associated with bites by the Brown Tree Snake (Colubridae: Boiga irregularis) on Guam. J. Herpetol. 28:27-33. Hati, A. K., S. G. Saha, D. Banerjee, S. Banerjee, and D. Panda. 1988. Clinical features of poisoning by common kraits and treatment with polyvalent antivenin. Snake 20:140-143. N. 1980. Consecutive bites on three teen-age girls by a single lecherous viper. Toxicon 18:163-164.

Herz,

J. R. 1964. The Snakes of Australia, rev. ed. Sydney: Halstead Press. Mao, S.-H. 1970. Food of the common venomous snakes of Taiwan. Herpetoloeica 26:45-48. Kinghorn,

McCoy, M. 1980. Reptiles of the Solomon Islands. Handbook No. 7. Wau, Papua New Guinea: Wau Ecology Institute. F. 1983. The snakes of the western province. Wildl. Papua New Guinea 82:32-33.

Parker,

G. H., T. H. Fritts, and P. J. Conry. 1992. Origin and population growth of the Brown Tree Snake, Boiga irregularis, on Guam. Pac. Sci. 46:46-57. Saha, S. G., and A. K. Hati. 1983. A longitudinal study on snakebites in a subsidiary health centre in west Bengal. Snake 15:86-90. Rodda,

J. A. 1988. Food habits of Boiga irregularis, an introduced predator on Guam J. Herpetol. 22:275-282.

Savidge,

Sawai,

Y., K.

Ibrahim,

Koba,

T.

T.

Okonogi, S. Mishima,

Y.

Kawamura,

H.

Chinzei,

A. B.

Bin

Devaraj, S. Phong-Aksara, C. Puranananda, E. S. Salafranca,

J. S. Sumpaico, C.-S. Tseng, J. F. Taylor, C.-S. Wu, and T.-P. Kuo. 1972. An epidemiological study of snake bites in the Southeast Asia. Japan. J Exp Med 42:283-307. Sawai,

Y.,

and

C. S.

Tseng.

1969. Snakebites on Taiwan. Snake 1:9-18.

I. 1980. Ecology of land snakes and epidemiology of snake bites in Malaysia Snake 12:37-44.

Singh, K.

D. K., S. P. Mackessy, and K. V. Kardong. 1991. The unique Duvernoy’s secretion of the Brown Tree Snake (Boiga irregularis). Toxicon 29:532-535. Weinstein, S. A., D. Chiszar, R. C. Bell, and F. A. Smith. 1991. Fethal potency and fractionation of Duvernoy’s secretion from the Brown Tree Snake, Boiga irregularis. Toxicon 29:401-407. Vest,

Boiga Bites on Guam

127

S. A., B. G. Stiles, M. J. McCoid, L. A. Smith, and K. V. Kardong. 1993. Variation of lethal potencies and acetylcholine receptor binding activity of Duvernoy’s secretions from the Brown Tree Snake, Boiga irregularis Merrem. J. Nat.

Weinstein,

Toxins 2:187-198. Yu, P., A. Tang, J. Liang, Q. Huang, and F. Mo. 1989. An investigation on Wuzhou’s snake bite epidemiology during the 1973—1984 period. In M. Matsui, T. Hikida, and R. C. Goris, eds., Current Herpetology in East Asia, pp. 489-492. Kyoto: Herpetological Society of Japan.

7 Venom Delivery by the Brown Treesnake [Boiga irregularis) and the Habu (Trimeresurus flavoviridis) Kenneth V. Kardong

The pharmacology, especially the toxicity, of many specific snake venoms is I today well known. Consequently, most laymen might be surprised to learn that it remains difficult to assess the human health risks posed by venomous snakes. The reason is that snakebite involves a complex organism (a snake) that injects a complex suite of chemicals (venom) into another complex organism (a human) (Minton, 1987). As a result, the outcomes of snakebite can be extremely different from case to case, ranging from painless annoyance to trauma and death within an hour. Even when the same species of snake is evaluated, outcomes of envenomation can be quite variable. Estimates differ, but as many as half of the bites of humans by venomous snakes can be dry bites, in which no appreciable amount of venom is injected (Parrish, 1959; Reid, 1970; Russell, 1980). The small Arizona or Sonora Coralsnake (Micruroides euryxanthus) possesses a very toxic venom but has never caused a known human fatality (Minton, 1987). In large measure, the great variations in snakebite outcomes are the result of different efficiencies of the venom delivery system itself. How well a complex organism (snake) succeeds in delivering a complex secretion (venom) into another complex organism (human) is determined by the design and performance of its venom delivery system. The Habu and the Brown Treesnake represent two basic venom delivery systems found in advanced snakes. In this chapter I will compare and contrast the venom delivery systems of these two snakes, with the purpose of evaluating the human health risks from snakebites.

GENERAL VENOM SYSTEMS Snakes possess a variety of oral glands that release their products into the mouth (Kochva, 1978). Among the advanced snakes, the best studied oral glands are the venom glands located in the temporal region behind the eye in members of the families Elapidae and Viperidae. The venoms are suites of chemicals that function primarily to immobilize and kill prey rapidly, and to initiate digestion (e.s. Russell, 1980; Tu, 1982; Rosenberg et al., 1985; Chippaux et ah, 1991). Other com¬ ponents of the venom may facilitate or enhance these primary activities (Mackessy, 1988). Secondarily, venoms used in defense may be life-threatening for

128

Venom Delivery by Brown Treesnake and Habu

129

humans or may cause permanent dysfunction of the stricken limb (Russell, 1980; Tu, 1982; Snyder, 1991). Some species of the family Colubridae produce oral secretions that exhibit mild (Grogan, 1974; Vest, 1981a, 1981b; Hayes and Hayes, 1985; Rosenberg et ah, 1985) to alarming toxicity (McKinstry, 1983; Sakai et ah, 1984; Fukushima, 1986; Kikuchi et ah, 1987), with reports of human deaths following bites by some colubrid species (Mittleman and Goris, 1974; Ogawa and Sawai, 1986; Minton, 1990). Yet colubrid snakes lack a venom gland. Instead, many possess a Duvernoy’s gland (Taub, 1966, 1967; Weinstein and Kardong, 1994), which is evolutionarily homologous to the venom glands of elapids and viperids (Kochva, 1965; Kochva and Wollberg, 1970), but surprisingly distinct from a true venom gland (Kardong and Lavin-Murcio, 1993). Because the venom delivery systems of the Habu (viperid) and the Brown Treesnake (colubrid) are distinct, they pose different kinds of health risks.

VENOM DELIVERY SYSTEMS Duvernoy's Gland The paired Duvernoy’s gland of the Brown Treesnake is an encapsulated, branched tubular gland located in the temporal region of the skull. The secretory epithe¬ lium of the gland releases the complex product, Duvernoy’s secretion, into a very small storage area within the gland, so there is a relatively small extracellular reser¬ voir of secretion (Zalisko and Kardong, 1992). Duvernoy s gland does not empty into a hollow fang. Instead, it empties by a single duct to a cuff of oral epithelium that encircles the grooved posterior maxillary teeth (Fig. 7.1a; Kardong and LavinMurcio, 1993).

Venom Gland The paired venom gland of the Habu is a thickly encapsulated, branched tubular gland located in the temporal region of the skull. The thin secretory epithelium releases the complex venom into an extensive storage area derived from the luminal spaces within the gland, so a relatively large and ready extracellular reservoir of venom is available. As with most viperids (Kochva, 1978), the Habu venom gland includes a main venom gland from which departs the single main venom duct, which passes through the accessory venom gland to emerge as the secondary venom duct, which connects to the base of the hollow fang (Fig. 7.1b).

Discharge of Secretion Two general systems of venom discharge have been described in snakes: lowpressure systems and high-pressure systems (Kardong and Lavin-Murcio, 1993). The Brown Treesnake uses a low-pressure system, in which release of Duvernoy’s

130

Kardong

Figure 7.1 Duvernoys gland versus venom gland, (a) The Duvernoy’s gland (Dv) is shown in cutaway view revealing the small extracellular reservoir within the duct system draining the gland lobules to a single main duct (based on the Brown Treesnake). (b) The venom gland includes a main venom gland (vg) with considerable storage space, a main venom duct (pd), accessory gland (ag), and secondary venom duct (sd) opening to the base of a hollow fang (f) (based on the Northern Pacific Rattlesnake [Crotalus viridis oreganus], a viperid). Abbreviations: ag, accessory gland; c, capsule of gland fibrous connective tissue; eg, compressor glandulae jaw muscle; Dv, Duvernoys gland; f, fang; lu, lumen; mx, maxilla; pd, primary duct; sd, secondary duct; se, secretory epithelium; slg, supralabial gland; vg, main venom gland.

secretion is often protracted. No striated muscles directly insert on the Duvernoy’s gland (Fig. 7.2a), and no large, ready reservoir of secretion is available; the duct from Duvernoys gland opens adjacent to the teeth but is not tightly applied to them, and the teeth bear an open groove (Kardong and Young, 1991). The venom system of the Brown Treesnake cannot generate or sustain a high pressure head to inject secretion during a bite. The Habu uses a high-pressure system, in which release of venom occurs in a sudden pulse delivered during a relatively high pressure surge. Striated jaw mus¬ cles insert directly into the thick capsule of the main venom gland (Fig. 7.2b), placing them in a position to generate a sudden, high internal pressure on the venom reservoir within the lumen of the main gland. During the strike, the venom duct forms a tight couple with the base of the fang to prevent loss of pressure and to convey the discharged venom through the fang and deep into the tissues of prey or victim as the fangs penetrate the integument.

Venom Delivery by Brown Treesnake and Habu

131

Boiga

Trimeresurus

Figure 7.2 Striated muscles associated with oral glands, (a) Boiga irregularis. The Duver¬ noy’s gland (shaded) lies behind the eye and superficial to the adductor superficialis (as) muscle. Although coursing along the medial face of Duvernoy’s gland, the adductor superficialis does not insert on the gland, therefore leaving the gland with no direct jaw muscle attachments, (b) Trimeresurus flavoviridis. The venom gland (shaded) receives the insertion of the compressor glandulae (eg), whose action is to squeeze the venom gland, raise intraglandular pressure, and produce the sudden pulse of venom during a bite.

The actual delivery of venom is also quite different in the two species. In the Brown Treesnake, 3.3-3.6 mg of Duvernoy’s secretion may be delivered to mice during a feeding strike (Hayes et al., 1993), compared with 13.0-22.5 mg of venom injected by the Habu (Hokama, 1978; Hokama et al., 1981). Further, only about 55% of this Duvernoy’s secretion actually reaches the viscera; the remainder remains on or embedded within the skin. However, in the rattlesnakes and elapids studied, up to 90% of the venom reaches the viscera (Hayes et al., 1992, 1995). Comparable figures are not available for the Habu, but given its long fangs, the Habu is likely able to deliver most of its venom deep into tissues and not leave much behind on or in the integument (Hokama, 1978; Hokama et al., 1981).

Strike Behavior During a predatory strike, the Brown Treesnake launches the anterior part of its body at the prey, making first contact with its open mouth. Usually this is quickly

132

Kardong

followed by constriction of the prey, especially if the prey is relatively large (Chiszar, 1991). At that moment, Duvernoy’s secretion may tranquilize the prey or even result in protracted death. However, if deprived of the secretion from Duvernoy’s gland during prey capture, a Brown Treesnake nevertheless kills the prey just as quickly as when Duvernoy’s secretion is available (Rochelle and Kar¬ dong, 1993). Therefore, the Brown Treesnake depends primarily on mechanical (constriction) and not chemical (venom) means to kill prey. The grooved teeth that receive Duvernoy’s secretion are located at the posterior end of the long maxillary bone, and thus reside at the back of the mouth, an inconvenient kine¬ matic position (Kardong, 1979) from which to deliver secretion during the feed¬ ing strike. Once the prey is dead, the snake releases its hold, investigates, and usually begins to swallow the prey head-first, walking its jaws over the prey (e.g., Kardong, 1986a). During the swallowing process, the posterior maxillary teeth surrounded by the swollen cuff of oral epithelium can be seen to engage the surface of the prey as the jaws advance reciprocally during swallowing (Fig. 7.3). This suggests that additional delivery of Duvernoy’s secretion may occur during swallowing, and that the secretion participates in a yet undefined way in aiding subsequent digestion of prey (Kardong, 1982; Jansen, 1983; Rodriguez-Robles and Thomas, 1992). During a defensive strike, the Brown Treesnake launches its anterior body by straightening the body curves generally in the direction of the threat. The snake may keep its jaws closed and “butt’’ the individual, open its jaws to bite and release the individual, or bite and hold on.

Figure 7.3 Outline of right, advancing side of the jaws of the Brown Treesnake during swal¬ lowing of rodent prey. Note enlarged cuff of oral epithelium (solid arrow) receiving Duvernoy’s secretion and associated with the grooved posterior maxillary tooth.

Venom Delivery by Brown Treesnake and Habu

133

Unlike the Brown Treesnake, the Habu depends on chemical means to kill its prey. During predatory strikes, the Habu either strikes and holds or strikes and releases rodent prey, but it never constricts prey. If it releases the prey, it is likely to do so quickly after the strike, often within less than a second. The prey may then scamper a short distance, but soon becomes immobile and dies (Hayashi et al., 1982). The Habu then recovers the prey and begins swallowing it head-first. Defensive Habu strikes may involve up to two-thirds of the body length and deliver up to 0.5 ml of whole venom (Anon, 1985). Humans bitten defensively by Habu are mostly adults who inadvertently tread on the snake while going about their work (Araki and Tomihara, 1989).

VENOMS Duvernoy's Secretion The liquid Duvernoy’s secretion contains 15% dry matter, with the protein con¬ tent of the dry fraction ranging between 52% and 100%. A myotoxic component is present in the secretion, but hemorrhagic activity is absent (in rats). The murine i.p. LD50 ranges from 10.5 mg/kg to 34.1 mg/kg. Proteolytic activity is relatively low, and the toxicity is proportional to proteolytic activity. The composition and activity of the secretion are quite variable, even within the same individual examined on different occasions; generally, secretion from larger snakes is more toxic than that from small snakes. Lethal activity occurs in at least 2 of 16 peaks resolved by FPLC. One of these peaks includes a neurotoxic component, and small snakes seem to have higher concentrations of the neurotoxin in their Duvernoy’s secretion (Vest et al., 1991; Weinstein et al., 1991, 1993; Chiszar et al., 1992).

Habu Venom Habu venom includes toxic principles that are responsible for the lethal action of the venom as well as causing severe damage to the envenomated soft tissues (Ohsaka et al., 1960). The damage is severe and long lasting. If treatment is delayed, residual symptoms may result (Sawai et al., 1962). Significant hemorrhaghic activity is associated with the venom (Omori-Satoh and Ohsaka, 1970; Takahashi and Ohsaka, 1970a). Several hemorrhagic principles are involved (Takahashi and Ohsaka, 1970b). The pharmacological properties of Habu venom are directly correlated with actual clinical manifestations of Habu bites (e.g., Sawai et al., 1962, this volume, Chap. 5).

CONCLUSION The Duvernoy’s secretion of the Brown Treesnake exhibits some ontogenetic changes, becoming more toxic in larger snakes. A neurotoxic component is pre¬ sent, but apparently in low concentration. The secretion may be quite variable,

134

Kardong

even in the same individual on different occasions. Most clinical differences in reactions to the Brown Treesnake bites appear to be related to different success or inclination of the snakes in delivering venom. The factors affecting the efficiency of snake venom delivery systems are not well understood. The fact that up to 10% of the venom of viperid and elapid snakes may remain outside the prey (Morrison et al., 1982, 1983; Hayes et al., 1992) suggests inefficient venom delivery during feeding strikes. Defensive strikes by viperid and elapid snakes may result in no effective delivery of venom. The reasons for such delivery inefficiencies, especially during defensive strikes, are unknown, but may include disruption of normal jaw mechanics resulting from poor or unfavorable fang placement (Kardong, 1986b). In the Brown Treesnake, the grooved teeth that could deliver the secretion of Duvernoys gland are located on the posterior end of the maxilla, at the back of the mouth. Therefore, unlike teeth involved immediately in prey capture, which are located at the front of the mouth (Kardong, 1979, 1980), the grooved teeth of the Brown Treesnake must be moved forward by kinematic displacement after the initial strike to subsequently engage them into prey or victim. The inconvenient location of the grooved teeth may in part account for the fact that almost 45% of Duvernoys secretion remains in or on the skin of the prey (Hayes et al., 1993). However, during subsequent swallowing, the posterior location of these grooved teeth, like those of many colubrids (Kardong, 1986a), makes their engagement with the prey more likely (Kardong, 1980). Unlike most viperid and elapid snakes, a significant part of the Brown Treesnake’s venom delivery may occur not during the strike but during swallowing. This raises the possibility that health risks to human bite victims increase if the Brown Treesnake engages in swallowing behavior that gives greater opportunity for engagement of the posterior maxillary teeth that deliver venom. The clinical symptoms associated with Brown Treesnake bites show evidence of myotoxic effects, but reports of hemorrhage are inconsistent with the phar¬ macological properties of Duvernoys secretion. Further, although pharmacolog¬ ical analysis documented the presence of a neurotoxin, clinical symptoms from bites do not show evidence of neurotoxic involvement (except see Fritts et al., 1990). The inefficiencies of the low-pressure venom system of the Brown Treesnake likely contribute to the irregular medical consequences of envenomations by this snake. This is one more reminder of the variable outcomes of snakebite, which involves a complex organism, a complex secretion, and another complex organism.

LITERATURE CITED Anon.

1985. Handbook for the Control of Habu, or Venomous Snakes in the

Ryukyu Islands. Okinawa International Centre, Japan International Cooperation Agency.

Venom Delivery by Brown Treesnake and Habu Araki,

Y.,

and

Y.

Tomihara.

135

1989. Snake bites in Okinawa Prefecture. In M. Matsui,

T. Kikida, and R. C. Goris, eds., Current Herpetology in East Asia, pp. 482-488. Kyoto: Herpetological Society of Japan. Chippaux, J. P., V. Williams, and J. White.

1991.

Snake venom variability:

Methods of study, results and interpretation. Toxicon 29:1270-1303. Chiszar,

D. 1991. The behavior of the Brown Tree Snake: A study in applied com¬

parative psychology. In D. A. Dewsbury, ed., Contemporary Issues in Comparative Psychology, pp. 101-123. Sunderland, Mass.: Sinauer. Chiszar,

D., S. A. Weinstein,

and

H. M.

Smith.

1992. Liquid and dry venom yields

from Brown Tree Snakes, Boiga irregularis (Merrem). In

P.

D. Strimple and J. L.

Strimple, eds., Contributions in Herpetology, pp. 11-13. Cincinnati, Ohio: Greater Cincinnati Herpetological Society. Fritts, T. H., M. J. McCoid, and R. L.

Haddock.

1990. Risks to infants on Guam

from bites of the Brown Tree Snake, Boiga irregularis. Am. J. Trop. Med. Hyg. 42:607-611. Fukushima, H. 1986. Clinical aspects of bite by Yamakagashi, Rhabdophis tigrinus [in Japanese]. J. Kagoshima Soc. Int. Med. 18:60-85. Grogan, W. L. Jr. 1974. Effects of accidental envenomation from the saliva of the Eastern Hognose Snake, Heterodon platyrhinos. Herpetologica 30:248-249. Hayashi,

Y., H.

Kihara, and

H.

Tanaka.

1982. Immobilizing effect of venom of

Habu (Trimeresurus flavoviridis) on small rodents. Snake 14:35-39. Hayes,

W. K.,

and

R.

Hayes.

1985. Human envenomation from the bite of the

Eastern Garter Snake, Thamnophis s. sirtalis (Serpentes: Colubridae). Toxicon 23:719-721. Hayes, W. K., I. I.

Kaiser, and

D.

Duvall.

1992. The mass of venom expended by

Prairie Rattlesnakes when feeding on rodent prey. In J. C. Campbell and E. D. Brodie Jr., eds., Biology of Pitvipers, pp. 383-388. Tyler, Tex.: Selva. Hayes, W. K., P. Lavin-Murcio, and K. V. Kardong. 1993. Delivery of Duvernoy’s secretion into prey by the Brown Tree Snake, Boiga irregularis (Serpentes. Colubridae). Toxicon 31:881-887. _. 1995. Northern Pacific Rattlesnakes (Crotalus viridis oreganus) meter venom when feeding on prey of different sizes. Copeia 1995:337-343. Hokama, Z. 1978. Study on experimental envenomation by the Habu (Trimeresurus flavoviridis). Snake 10:107-113. Hokama, Z., T. Kamura, and M. Nozaki.

1981. Study on experimental envenoma¬

tion by Sakishima-habu (Trimeresurus elegans). Snake 13:27-31. Jansen, D. W. 1983. A possible function of the secretion of Duvernoy’s gland. Copeia 1983:262-264. Kardong, K. V. 1979. “Protovipers” and the evolution of snake fangs. Evolution 33:433-443. -. 1980. Evolutionary patterns in advanced snakes. Am. Zool. 20:269—282. -. 1982. The evolution of the venom apparatus in snakes from colubrids to viperids and elapids. Mem. Inst. Butantan 46:105—118. -. 1986a. Kinematics of swallowing in the Yellow Rat Snake, Elaphe obsoleta quadrivittata: A reappraisal. Japan. J. Herpetol. 11:96—109.

136

Kardong

-. 1986b. The predatory strike of the rattlesnake: When things go amiss. Copeia 1986:816-820. Kardong,

K. V.,

P. A.

and

Lavin-Murcio.

1993. Venom delivery of snakes as high-

pressure and low-pressure systems. Copeia 1993:644-650. Kardong,

K. V.,

and

B. A.

Young.

1991. Fangs and snakes: Flow do open grooves

inject venom into enclosed spaces? Am. Zool. 31:51 A. Kikuchi,

H.,

T. Takamura,

M.

Ishii, T. Ichihara,

Y.

Kawamura, and

Y.

Sawai.

1987. Study on the effectiveness of the Yamakagashi (Rhabdophis tigrinus) antivenom. Snake 19:84-86. E. 1965. The development of the venom gland in the opisthoglyph snake

Kochva,

Telescopus fallax with remarks on Thamnophis sirtalis (Colubridae, Reptilia). Copeia 1965:147-154. -. 1978. Oral glands of the Reptilia. In C. Gans and K. A. Gans, eds., Biology of the Reptilia, vol. 8, pp. 43-161. New York: Academic Press. E.,

Kochva,

and

M.

Wollberg.

1970. The salivary glands of Aparallactinae

(Colubridae) and the venom glands of Elaps (Elapidae) in relation to the taxonomic status of the genus. J. Linn. Soc. (Zool.) 49:217-224. Mackessy,

S. P. 1988. Venom ontogeny in the Pacific rattlesnakes Crotalus viridis

helleri and C. v. oreganus. Copeia 1988:92-101. D. M. 1983. Morphologic evidence of toxic saliva in colubrid snakes. A

McKinstry,

checklist of world genera. Herpetol. Rev. 14:12-15. Minton,

S. A. 1987. Poisonous snakes and snakebite in the U.S.: A brief review.

Northwest Sci. 61:130-137. -. 1990. Venomous bites by nonvenomous snakes: An annotated bibliography of colubrid envenomation. J. Wildl. Med. 1:119-127. Mittleman,

M. B.,

and

R. C.

Goris.

1974. Envenomation from the bite of

the Japanese colubrid snake Rhabdophis tigrinus (Boie). Flerpetologica 30:113119. Morrison,

J. J., J. H.

Pearn,

N. T.

Charles, and

A. R.

Coulter.

1983. Further

studies on the mass of venom injected by elapid snakes. Toxicon 21:279284. Morrison,

J. J., J. H.

Pearn, and

A. R.

Coulter.

1982. The mass of venom injected

by two Elapidae: The Taipan (Oxyuranus scutellatus) and the Australian Tiger Snake (Notechis scutatus). Toxicon 20:739-745. Ogawa,

H.,

and

Y.

Sawai.

1986. Fatal bite of Yamakagashi (Rhabdophis tigrinus).

Snake 18:53-54. Ohsaka,

A., H.

Ikezawa,

H.

Kondo,

S.

Kondo, and

N.

Uchida.

1960. Hemorrhagic

activities of Habu snake venom, and their relations to lethal toxicity, proteolytic activities and other pathological activities. Br. J. Exp. Pathol. 41:478-486. Omori-Satoh,

T.,

and

A.

Ohsaka.

1970. Purification and some properties of

hemorrhagic principle I in the venom of Trimeresurus flavoviridis. Biochim. Biophys. Acta 207:432-444. Parish,

H. M. 1959. Poisonous snakebites resulting in lack of venom poisoning. Va.

Med. Mon. 86:396.

Venom Delivery by Brown Treesnake and Habu Reid,

137

H. A. 1970. The principles of snakebite treatment. Clin. Toxicol. 3:472-482.

Rochelle,

M.,

and

K. V.

1993. Constriction vs. envenomation in prey

Kardong.

capture by the Brown Tree Snake Boiga irregularis (Squamata: Colubridae). Herpetologica 49:297-300. Rodriguez-Robles, J. A., and R. Thomas. 1992. Venom function in the Puerto Rican Racer, Alsophis portoricensis (Serpentes: Colubridae). Copeia 1992:62-68. Rosenberg,

H. I., A.

Bdolah, and

E.

Kochva.

1985. Lethal factors and enzymes in

the secretion from Duvernoy’s gland of three colubrid snakes. J. Exp. Zool. 233:5-14. Russell, F. E. 1980. Snake Venom Poisoning. Philadelphia: J. B. Lippincott. Sakai,

A., M.

Honma, and

Y.

1984. Study on the toxicity of venoms extracted

Sawai.

from Duvernoy’s gland of certain Asian colubrid snakes. Snake 16:16-20. Sawai,

Y. M.,

Makino,

I.,

Tateno,

T.

Okonogi, and

S.

Mitsuhashi.

1962. Studies

on the improvement of treatment of Habu snake (Trimeresurus flavoviridis) bite. 3. Clinical analysis and medical treatment of Habu snake bite on the Amami Islands. Japan. J. Exp. Med. 32:117-138. Snyder, C. C. 1991. Snakebite treatment by plastic surgery applications. In A. Tu, ed., Handbook of Natural Toxins. Vol. 5, Reptile Venoms and Toxins, pp. 645-686. New York: Marcel Dekker. Takahashi, T., and A.

Ohsaka.

1970a. Purification and characterization of a

proteinase in the venom of Trimeresurus flavoviridis: Complete separation of the enzyme from hemorrhagic activity. Biochim. Biophys. Acta 198:293-307. -. 1970b. Purification and some properties of two hemorrhagic principles (HR2a and HR2b) in the venom of Trimeresurus flavoviridis; complete separation of the principles from proteolytic activity. Biochim. Biophys. Acta 207:65—75. Taub, A. M. 1966. Ophidian cephalic glands. J. Morphol. 118:529-542.

-. 1967. Comparative histological studies on Duvernoy’s gland of colubrid snakes. Bull. Am. Mus. Nat. Hist. 138:1-50. Tu, A. T., ed. 1982. Rattlesnake Venoms: Their Actions and Treatment. New York: Marcel Dekker. Vest, D. K. 1981a. Envenomation following the bite of a Wandering Garter Snake (Thamnophis elegans vagrans). Clin. Toxicol. 18:573-579. -. 1981b. The toxic Duvernoy’s secretion of the Wandering Garter Snake, Thamnophis elegans vagrans. Toxicon 19:831-839. Vest, D. K., S. P. Mackessy, and K. V. Kardong. 1991. The unique Duvernoy’s secretion of the Brown Tree Snake (Boiga irregularis). Toxicon 29:532-535. Weinstein,

S. A., D.

Chiszar, R. C. Bell, and

L. A.

Smith.

1991. Lethal potency

and fractionation of Duvernoy’s secretion from the Brown Tree Snake Boiga irregularis. Toxicon 29:401-407. Weinstein, S. A., and K. V. Kardong. 1994. Properties of Duvernoy’s secretions from opisthoglyphous and aglyphous colubrid snakes: A critical review. Toxicon 32:1161-1185. Weinstein, S. A., B. G.

Stiles,

M. J.

McCoid,

L. A.

Smith, and

K. V.

Kardong.

1993. Variation and lethal potencies and acetylcholine receptor binding activity of

138

Kardong Duvernoy’s secretions from the Brown Tree Snake, Boiga irregularis. Copeia

1992:791-799. Zalisko, E. J., and

K.

V.

Kardong.

1992. Histology and histochemistry of the

Duvernoy’s gland of the Brown Tree Snake, Boiga irregularis. Copeia 1992:791799.

8 Factors Affecting Annual Incidence of Habu Bites, and How Residents Develop and Transfer Cognition of High-Risk Sites Hiroshi Tanaka Yoshihiro Hayashi Satoshi Nakamura

O

n Amamioshima and Tokunoshima, both in the Amami Islands, 200-300 people were bitten by Habu (Trimeresurus flavoviridis) each year between

1954 and

1976. During this period, the number of bites decreased on

Amamioshima and increased remarkably on Tokunoshima. Such dramatic shifts in the epidemiology of Habu bites had not been observed previously (Sawai et al., 1967). To understand what caused these changes, scientists initiated an epidemi¬ ological study of factors suspected to affect the number of people bitten by Habu (Tanaka et al., 1979). For many years, residents had been notified of areas where the risk of being bitten by Habu was high, but this public health campaign yielded no decrease in the number of people struck by Habu. To analyze why the number of bites went up in spite of the education campaign, Nakamura et al. (1987) initiated a study of how Tokunoshima residents learn about high-risk sites and how they transfer that knowledge to other people.

EPIDEMIOLOGICAL BACKGROUND Until the 1950s, more people were bitten by Habu on Amamioshima than on Tokunoshima. The area of Amamioshima is 81,908 ha, and in 1954 the total human population of Amamioshima was 100,370; the population of Tokunoshima, an island of 24,811 ha, was 50,892 at that time. In 1955, 208 people were bitten by Habu on Amamioshima and 120 people were bitten on Tokunoshima. The number of Habu bites per 100,000 people was similar on both islands, and it was simply accepted that fewer people were bitten on Tokunoshima because its human and snake populations were smaller. During the late 1950s, however, the annual rate of Habu bites on Amamioshima began to decrease while the rate increased on Tokunoshima. In 1976, the annual incidence of Habu bites was 61 bites per 100,000 people (51/84,184) on Amamioshima and 429 bites per 100,000 people (151/35,188) on Tokunoshima (Sawai and Kawamura, 1980). Furthermore, while the bite rate on Tokunoshima increased, the human popula¬ tion decreased by more than 15,000 people.

139

140

Tanaka, Hayashi, & Nakamura

Table 8.1 Correlation coefficients between number of snakebites and predictor variables for 10 villages on Amamioshima and Tokunoshima during 1976.

Total bites

No. of farmers

Area cultivated per farmer (ha)

Total area of farmland (ha)

x2

X3

X4 = X,/X3

X5

0.934*

0.087

0.948*

0.746

0.943*

0.927

0.057

0.942*

0.743

0.941*

0.918

0.172

0.927

0.727

0.910

Area under cultivation (ha)

Total human population

x,

(Yi) Outdoor bites (Y2) Indoor bites

(Y3) Note: Correlation coefficients between Xj and X3, between X! and X5) and between X3 and X5 were 0.965, 0.986, and 0.973, respectively. * Significant coefficient at P < 0.05.

An epidemiological study was undertaken to determine what factors accounted for the changes in numbers of Habu bites reported on Amamioshima and Tokunoshima. There were 7 villages on Amamioshima and 3 on Tokunoshima. The numbers of Habu bites and various social and environmental factors were summarized for the 10 villages. The Habu bites (Yi) occurring on Amamioshima and Tokunoshima from 1972 through 1976 were further categorized as outdoor cases (Y2) and indoor cases (Y3). Social and environmental factors suspected of affecting rates of Habu bites were area (ha) of farmland under cultivation (XT), total human population in 1976 (X2), number of farmers in 1976 (X3), area cultivated per farmer (X4 = Xx/X3)y and total area (ha) registered as farmland (X5). Correlation coefficients were calculated to determine the strength of relationships between the number of Habu bites (Yly Y2y Y3) and each of the suspected factors (Xb X2, X3, X4, X5). Several of the correlation coefficients were high (Table 8.1). In particular, the total number of Habu bites (Yi) was positively correlated with area of cultivated farmland (XJ, total area registered as farmland (X5), and number of farmers (X3). The number of outdoor cases of Habu bites was also positively correlated with X3 and X^ Interestingly, the number of Habu bites was not closely correlated with total human population. Xj, X3, and X5 were positively correlated with each other. Because the analyses described above represented only one year (1976), we con¬ ducted further analyses to determine whether or not the observed correlations were consistent over the long run. We used the data from three years (1955, 1965, and 1975), covering a 20-year span, and developed a multiple regression model to describe the relationship between various factors and the total number of Habu bites (Yx) per village per year:

Annual Incidence of Habu Bites

141

Yi = 0.0485 + 0.02213(X!) + 0.000798(X3) where X! is the area of cultivated farmland and X3 is the number of farmers. We also calculated correlation coefficients between the total number of Habu bites (Yj) and each of the two independent factors, X! and X3, for the 10 villages of Amamioshima and Tokunoshima. Again, the number of Habu bites was positively correlated with the area of cultivated farmland (Table 8.2). On Amamioshima, the number of farmers decreased over the 20-year span, and the number of Habu bites went down accordingly. On Tokunoshima, however, the number of Habu bites increased even as the number of farmers decreased. An important difference in the events taking place on the two islands may account for this discrepancy. Between 1955 and 1975, 45% of Amamioshima’s farmland fell out of cultivation, and the numbers of farmers and Habu bites dropped by 67% and 61%, respectively. These data are con¬ sistent with the original conclusion that the area of cultivated farmland and the number of farmers are positively correlated with the number of Habu bites. During the same period on Tokunoshima, the area of farmland under cultiva¬ tion and the number of Habu bites increased by 14% and 37%, respectively, but the number of farmers decreased by 43%. Apparently, the major factor directly influencing the number of Habu bites is the area of farmland under cultivation. The relationship between area of farmland under cultivation and the number of Habu bites can be explained by Habu densities. Habu depend on small rodents, especially Rattus spp., for approximately 83% of their diet (Mishima, 1966). Because cultivated farmland attracts many small rodents, it also attracts many Habu. Greater densities of Habu have been observed in cultivated fields and adjacent shrublands than in abandoned crop fields. Therefore, keeping farmland under active cultivation, particularly when sugarcane is the crop, can result in enhanced populations of Habu.

Table 8.2 Changes (represented by data taken at 10-year intervals) in numbers of Habu bites, area of farm fields, and numbers of farmers in seven villages on Amamioshima and three villages on Tokunoshima, 1955-1975. Area cultivated (ha)

No. of snakebites (T)

(A)

_

No. of farmers (X3)

Island

1955

1965

1975

1955

1965

1975

1955

1965

1975

Amamioshima

208

104

81

5017

4470

2752

65,315

44,398

21,291

Tokunoshima

120

132

189

5053

5895

5856

45,065

36,020

25,632

Note: Correlation coefficients between Xx and Y\ and between X3 and Yx were 0.8682 and 0.7436, respectively.

142

Tanaka, Hayashi, & Nakamura

ESTABLISHING AND TRANSFERRING COGNITION OF SITES WHERE THE RISK OF HABU BITE IS HIGH For many years, residents of the Amami Islands were exposed to education cam¬ paigns designed to teach them how to avoid being bitten by Habu. Between the circadian patterns of human activities and the Habu’s habits (Sawai et al., 1967), however, it was difficult to prevent people from being bitten. Flabu are completely nocturnal snakes (Tanaka et al., 1967; Tanaka, 1973; Tanaka and Wada, 1977) that generally withdraw to daytime refugia (e.g., dense, shady vegetation in agricul¬ tural fields), where they remain inactive until night. Most Habu bites occur during daylight hours, however, when farmers are active, and in the evening, when both humans and Habu are active. Typically, a farmer at work in the field unknowingly comes too close to a Habu hidden in its daytime refugium and is bitten. People can learn to recognize and avoid places where the risk of being bitten is high. It is important that they convey that knowledge correctly to other people as well. To understand how this process takes place, Nakamura et al. (1987) initi¬ ated a psychological study in a 640 X 640 m area of Village San on Tokunoshima. The area was divided into a grid of 256 squares that measured 40 X 40 m. The sites where people had been bitten by Habu were identified and plotted on a grid map of the area (Fig. 8.1), and then questionnaires were distributed to junior high school students (ages 12-14) and their parents living in the study area. The ques¬ tionnaires asked children to identify areas where they had sighted Habu, and asked adults and children to identify areas they perceived as being high-risk areas for being bitten by Habu. Because the respondents were not equally familiar with all parts of the study area, the sample size of individuals contributing risk estimates for a given sector varied. Vegetation density, measured on a scale of 0-3, was determined for each of the 256 squares by examining aerial photos of the study area. For each square in the grid, survey item frequencies (e.g., number of Habu sighted in a given area and fraction of respondents who perceived a given area as a high-risk area) were calculated and then smoothed out by calculating moving averages for each square (the square in question was averaged with the two, three, or four squares surrounding it). The averaged values for each square were then classified into one of four levels. We used shading to illustrate the observed levels on grid maps that represent the geographical distribution of responses to each survey item (Figs. 8.1-8.4). The distribution of the sites adults perceived as being high-risk areas (Fig. 8.2) most closely matched the distribution of Habu bites (Fig. 8.1). In contrast, the distribution of sites perceived by children as being high-risk areas (Fig. 8.3) dif¬ fered markedly from the distribution of Habu bites. Instead, the areas where chil¬ dren had observed Habu were the areas they perceived as being high-risk areas (Fig. 8.4). The degree of vegetation coverage did not seem correlated with any of the distributions shown in Figures 8.1-8.4.

'

Figure 8.1 Geographical distribution of sites

. ... ...

where Habu bites occurred in a 640 X 640 m area of Village San, Tokunoshima, 1978-1982. Density of shading indicates the number of bites in each 40 X 40 m sector after smoothing by calculation of moving

0 - 0.03

average two times. The

0.10-0.20

dashed line represents

0.04 - 0.09

>0.20

■

the perimeter of the residential area.

Figure 8.2 Geographical distribution of 40 X 40 m sites perceived by 11 adults as being high risk for Habu bites within the regions represented in Figure 8.1. Density of shading represents the fraction of respondents who identified the sector as being high risk, smoothed as in Figure 8.1.

Figure 8.3 Geographical distribution of 40 X 40 m sites perceived by 48 junior high school children as being high risk for Habu bites (1982). See Figure 8.2 for computation of values.

Figure 8.4 Geographical distribution of sites where Habu were observed (1978-1982) by 48 junior high school children as reported in interviews conducted in 1982. Density of shading indicates the fraction of respondents reporting a sighting, as in Figure 8.2.

Annual Incidence of Habu Bites

145

To verify the conclusions we drew through visual comparisons of the figures, we subjected the data to correlation tests (Table 8.3). The correlation between sites of Habu bites and adults’ cognition of high-risk areas (B) was high (0.45) com¬ pared with the correlation between bite sites and children’s cognition of high-risk areas (C) (0.26). On the other hand, there was a high correlation (0.64) between areas children perceived as high-risk areas (C) and areas where children had observed Habu (A). The difference between adults’ and children’s perceptions of what constitutes a high-risk area may be due to the fact that the two groups use different sites. Adults are active in a greater variety of sites than children, whose activities and perceptions of high-risk areas are restricted primarily to the roads along which they travel to and from school. Some of the areas used by children are indeed high-risk places, accounting for the upper left correlation in Table 8.3, but many other areas are also high risk, explaining the adults’ more accurate per¬ ceptions. The distribution of high-risk areas did not follow a simple association with vegetated areas: vegetation density was not significantly correlated with any other surveyed items, for either children or adults (Table 8.3). Because the distribution of sites perceived by adults as being high-risk areas closely matched the actual distribution of sites where bites had occurred, it could be assumed that adults develop their perceptions through accumulating infor¬ mation about where Habu bites occur. However, the minimal overlap between adult and child cognition of high-risk areas indicates that adults are not trans¬ ferring this knowledge to their children. Instead, children seem to have developed their perceptions of high-risk areas through their own observations of Habu. The apparent lack of information transfer from adults to children reflects a recent social trend: increasingly, children are gaining their knowledge and learning about cultural traditions from nonfamilial sources. Further analyses of the data showed differences in areas perceived as high risk by children who had and had not observed Habu previously. For children who

Table 8.3 Correlations between sites where Habu bites occurred, sites where Habu were sighted by children, sites perceived by adults and children as being at high risk for Habu bites, and vegetation density in 256 40 x 40 m squares on a 640 x 640 m grid, San village, Tokunoshima. Factor (A) Habu seen

(B) Adult

(C) Child

by children

cognition

cognition

—

—

—

—

—

0.1815

—

0.1431

0.0452

Factor

Bite sites

A (Habu seen by children)

0.4098

B (Adult cognition)

0.4533*

0.4205

C (Child cognition)

0.2585

0.6427*

D (Vegetation density)

0.0859

* Correlation is comparatively high.

-0.0258

146

Tanaka, Hayashi, & Nakamura

had observed Habu previously, the areas perceived as high risk encompassed more than the area where Habu had actually been sighted. High-risk perceptions were far more circumscribed in children who had not observed Habu. Essentially, the correlations (Table 8.3) led us to the same conclusions that we drew from the grid maps (Figs. 8.1-8.4). We believe the grid illustrations and the correlation matrices provided equally correct results, but that the grids provided more detailed information by virtue of their three-dimensional nature.

LITERATURE CITED Mishima,

S. 1966. Studies on the poisonous snake, Habu, Trimeresurus flavoviridis

flavoviridis. 1. Food habit of Trimeresurus flavoviridis flavoviridis on the Amami Islands [in Japanese with English summary]. Japan. J. Sanit. Zool. 17:1-21. Nakamura,

S.,

H. Hashimoto, T. Tanaka, and H. Tanaka.

1987. Establishment,

justification and transfer of dwellers’ cognition of high risk areas for bites by the crotalid snake Habu, Trimeresurus flavoviridis, as analyzed by the mesh method. Snake 19:111-119. Sawai, Y., and

Y.

Kawamura.

1980. Snakebites on the Amami Islands of Japan in 1976

[in Japanese with English summary]. Snake 12:1-7. Sawai,

Y., Y.

Kawamura,

I.

Ebisawa,

T.

Okonogi,

Z.

Hokama, and

M.

Yamakawa.

1967. Studies on the improvement of treatment of Habu (Trimeresurus flavoviridis) bites. Habu bites on the Amami and Ryukyu Islands in 1964. Japan. J. Exp. Med. 37:51-59. Tanaka,

H. 1973. Activity and behavior of Habu (Trimeresurus flavoviridis). Snake

5:116-132. Tanaka,

H., Y.

Hayashi, and

Y.

Wada.

1979. Epidemiological studies of factors

correlated with snakebite by Habu, Trimeresurus flavoviridis, on the Amami Islands [in Japanese with English summary]. Snake 11:79-83, 130-131. Tanaka,

H., S.

Mishima, and

Y.

Abe.

1967. Studies on the behavior of Trimeresurus

flavoviridis, a venomous snake, on Amamioshima Island in regard to speed of movement, nocturnal activity and sensitivity to infra-red radiation. Bull. Tokyo Med. Dent. Univ. 14:79-104. Tanaka,

H., and Y. Wada. 1977. Venomous snakes. In M. Sasa, H. Takahashi, R. Kano,

and H. Tanaka, eds., Animals of Medical Importance in the Nansei Islands in Japan, pp. 29-71. Tokyo: Shinjuku Shobo.

Part III

BEHAVIORAL AND SENSORY BIOLOGY

Sensory biology is fundamentally concerned with receptor systems and the stimuli they detect, particularly as animals go about their normal activities, such as predation, migration or other movements, reproduction, and avoiding enemies. Sensory biologists also study the specialized behaviors through which animals gather information from their environments. Tongue flicking by snakes is an example: the tongue gathers molecules from the environment and transports them to sensory organs in the roof of the mouth (vomeronasal or¬ gans or organs of Jacobson). As is true for other areas of biological research, sensory biologists use many methods, and many different questions have inspired laboratory and field studies. Most of the work presented here was aimed at the identification of stimuli that can be used to attract snakes to traps or to repel them from sensi¬ tive areas such as human domiciles, cargo containers, ships, and aircraft. Hirosawa and Takami (Chap. 9) describe the sensory tissues of Habu—particularly the pit organs (sensitive to infrared radiation), the vomeronasal organs, and the eyes—revealing structural organi¬ zation similar to that of other pitvipers. This suggests that Habu should respond to the same range of stimuli that are known to guide the behaviors of other pitvipers, especially rattlesnakes. Since a substantial volume of data exists on sensation and perception in rattlesnakes, it seems likely that some of this information will be applicable to Habu. Nishimura (Chap. 10) and Niwa, Hattori, Kihara, Sato, and Murata (Chap. 11) focus on chemical cues in an effort to identify Habu attractants and repellents. Chiszar, Dunn, and Smith (Chap. 12) do the same for Brown Treesnakes, concentrating on commer¬ cially available preparations of chemical cues. These three chapters reveal that a sequence of research efforts is needed, beginning with the identification of chemical components of prey odors and con¬ tinuing with behavioral bioassays that assess the effectiveness of chemical cues under laboratory conditions. Finally, all the investiga¬ tors agree that successful laboratory tests must be followed by field tests because stimuli effective in the laboratory may not be equally functional in nature.

147

148

Partlll Mason (Chap. 13), working with Brown Treesnakes, explores other chemicals that might constitute attractants, namely, the sexual pheromones produced by females and the aggregation pheromones produced by both sexes. In other snake species, pheromones are known to attract individuals over considerable distances, suggesting that Brown Treesnake pheromones might constitute a useful bait. Efforts along these lines have been tried with Habu, but the results so far have been disappointing. Mason suggests, however, that re¬ search with viperid pheromones is still in its infancy, and there is no background of physiological and behavioral data on which Habu researchers can capitalize. Therefore, it is too early to conclude that pheromones constitute a blind alley for Habu control; rather, Mason encourages investigators to focus on the development of basic laboratory conditions that routinely lead to reproduction in Habu. Once these are identified, it will become possible to study pheromones and other reproductive phenomena with greater prob¬ ability of success. Much the same recommendations are made for Brown Treesnake researchers. The best attractants so far known for Habu and Brown Treesnakes are live rodents. Both species seem to rely on their chemical senses to locate these prey, although it is probable that visual cues play an important role as well. Habu likely also make use of infrared cues produced by mice, but this factor has not yet been studied in detail. Effective artificial attractants will probably com¬ bine food-derived chemical cues with other stimuli (e.g., visual, thermal, vibratory), instead of using only cues from one modality. This is suggested by Nishimura in Chapter 10 and by Rodda and Chiszar (1993).

LITERATURE CITED Rodda,

G. H.,

and

D.

Chiszar.

1993. Are cotton swab presentations a

sufficient model for ophidian prey recognition? Poster presented at the meeting of the American Society of Ichthyologists and Herpetol¬ ogists, Austin, Tex., 27 May-2 June (see pp. 265-266 of the Program and Abstracts).

9 Histology of the Habu's Sensory Organs Kazushige Hirosawa Shigeru Takami

H

uman fatalities resulting from Habu (Trimeresurus flavoviridis) bites continue to occur on some Japanese islands. If we are to further diminish the occurrence of fatal Habu bites, there must first be a better understanding of the Habu’s sensory systems. Until recently, the histology of the Habu’s sensory organs had not been described in detail. To that end, this chapter summarizes the recent histological studies of the Habu’s pit membrane (Hirosawa, 1980), vomero¬ nasal organ (Jacobson’s organ; Takami and Hirosawa, 1987, 1990), and retina (Hirosawa, 1978).

METHODS The tissues we examined were taken from Habu captured on Amamioshima. To fix Habu tissues, we injected a perfusate solution (2% formaldehyde and 1.25% glutaraldehyde in a 0.1 M phosphate buffer) into the Habu’s ascending aorta. Dis¬ sected specimens were further fixed with 1% osmium solution, whereupon spec¬ imens were prepared for observation by light and electron microscopy. The pit membrane tissues were embedded in epon, cut 1 pm thick, and stained with toluidine blue solution. The posterior halves of Habu eyeballs were immersed in aldehyde fixatives and postfixed with osmium tetroxide. Araldite-embedded specimens were cut for light and electron microscopy.

RESULTS AND DISCUSSION The Pit Membrane The pit membrane in the Habu pit organ is about 15pm thick (Fig. 9.1). Myeli¬ nated nerve fibers from the trigeminal nerve enter the pit membrane and swell into palmate structures on demyelination (Fig. 9.2). Demyelinated fibers branch repeatedly, and their terminals contain many mitochondria. The terminal por¬ tions are not surrounded by Schwann cell processes (Fig. 9.3). In the extracellular spaces of the pit membrane, there are many small vesicles (30-60 nm in diameter) that associate closely with the nerve terminal membranes. The branches of each palmate structure correspond with a discrete unit area detected by using electrophysiological methods. No “unit structure,” however, was revealed through electron microscopy.

149

150

Hirosawa & Takami

© Figure 9.1 A light microscopic cross-sectional view of the Habu pit membrane. The arrow indicates the exterior surface; the star indicates the myelinated portion of the nerve fibers.

Figure 9.2 A surface view (photograph) of the nerve fibers and their palmate structures (arrow) in the pit membrane of the Habu. Specimens were silver-impregnated and wholemounted on a glass slide.

Histology of Habu Sensory Organs

151

Figure 9.3 An electron micrograph of the Habu pit membrane showing layers from the outer, cornified layer to the inner, con¬ nective tissue layer. Note the myelinated nerve fibers and nude nerve fiber terminal with numerous mitochondria.

Intraepithelial free nerve endings exist in the outer epithelial layer. The nerve bundles from the trigeminal branches contain some unmyelinated fibers, which come in close contact with vascular elements in the pit membrane.

The Vomeronasal Organ: Observations by Light Microscopy Light microscopic observations of the Habu vomeronasal organ (VNO) revealed that the organ is encased in the os vomeris and the vomeronasal cartilage on either side of the nasal septum. The VNO is divided into two parts: a dome-shaped portion, mainly composed of sensory epithelium (SE); and a hemispherical mush¬ room body, which is covered with a ciliated epithelium. Between the mushroom body and the SE there is a narrow lumen (Fig. 9.4), which leads to the oral cavity via the vomeronasal duct. The SE consists of a sup¬ porting cell layer that faces the lumen, and an underlying sensory cell layer. Distal processes of the sensory cells pass through the cytoplasmic meshwork of the supporting cells and terminate as small swellings. As reported in other snake species, many of the sensory cells form columns (Fig. 9.5). Along the anteroposterior axis of the VNO, the columns reach their

152

Hirosawa & Takami

Figure 9.4 A scanning electron micrograph of the vomeronasal organ of the Habu. L indicates the lumen between the mushroom body (MB) and sensory portion (SU and C) of the organ. The sensory portion is clearly separated into two layers, a supporting cell layer (SU) and a layer of sensory columns (C). Scale bar = 200 pm.

maximum heights (>300 pm) in the middle of the dorsal and median portions and then gradually decrease toward the lateral portions. Each column contains undifferentiated basal cells, sensory cells, and satellite cells; the sensory cells are arrayed in a stringlike fashion. Each sensory cell contains a round nucleus and prominent lipofuscin granules (Fig. 9.6); cells situated in the neck regions of the columns contain more lipofuscin granules. Vasqular plexi occur around the necks and bottoms of the columns. Aggregates of heterogeneous cells were observed in the anterolateral region of the dome. A number of pigment cells were seen in the epineurium and the perivascular regions of the dorsal and lateral portions of the VNO. Goblet cells were present in the lateral boundary between the dome and the mushroom body.

The Vomeronasal Organ: Observations by Transmission and Scanning Electron Microscopy Electron microscopy of the Habu vomeronasal sensory epithelium reaffirmed the occurrence of a layer of superficial supporting cells and an underlying layer of sensory cells arranged in columns. The supporting cell layer consists of both sup-

Histology of Habu Sensory Organs

153

Figure 9.5 An enlarged view of the scanning electron micrograph of the sensory portion of the Habu vomeronasal organ. LS, luminal surface; SU, support¬ ing cell layer; C, col¬ umn structure of the sensory cells. The arrow indicates an outlet of vomeronasal nerve fibers. Scale bar = 50 pm.

porting cells and dendrites of the underlying sensory neurons. The apical regions of sensory cell dendrites contain numerous microtubules, many elongated mito¬ chondria, centrioles, and electron-dense bodies. The sensory cell dendrites terminate as dendritic knobs, from which microvilli project into the vomeronasal lumen. Smooth vesicles are abundant in the den¬ dritic terminals and their vicinity. The supporting cells, which also bear microvilli, contain large, electron-opaque granules and dense vesicles near their free surfaces. Cytoplasmic extensions of the supporting cells form a meshwork that separates the dendrites from each other in the vicinity of the luminal surface. The mesh¬ work becomes obliterated in the infranuclear region of each supporting cell. Bipolar-shaped sensory cells with lightly stained round nuclei contain the characteristic cell organelles of neurons and are thought to be sensory neurons. These cells are characterized by well-developed lamellae of rough endoplasmic reticulum. The perikarya of cells located in the apical region of the cell columns tend to contain larger amounts of smooth endoplasmic reticulum and lipofuscin granules than the perikarya of cells located in lower regions.

154

Hirosawa & Takami

Figure 9.6 A light micrograph of the sen¬ sory portion of the Habu vomeronasal organ. Many small bulges are visible at the luminal surface (L) of the supporting layer. Clear separation of the supporting cell layer (SU) and the sensory cell columns (C) is demonstrated. N, vomeronasal nerve bundle. Scale bar = 50 pm.

Undifferentiated cells are found in the basal region of the sensory cell columns. Satellite cells form the framework of the columns and are also found among neu¬ ronal elements.

The Retina Habu eyeballs have no vascular conus, and the retina is avascular. The thickness of the retina averages about 200 pm, with the central area being thicker than the peripheral region (Fig. 9.7). There is no central fovea. The majority of photoreceptor cells are rods, with the number of cone cells being about one-seventh of that of rod cells. The cone cells are distributed among the rod cells (Fig. 9.7). The disk membranes of the rod cells have no invagina¬ tions. Two types of cone cells are distinguishable: double cones and single cones. The single cones and the principal cone of the double cones are similar in struc¬ ture, but the accessory cone of the double cones has a “paranuclear body” in the perinuclear region.

Histology of Habu Sensory Organs

155

Figure 9.7 A light microscopic view of the Habu retina. The essential cellular architecture is similar to that of other vertebrate retinae. The retinal pigment epithelium (f) contains large lipid droplets. The outer segments (o) of rods and cones are long, and the parabo¬ loids of the cones (thin arrows) are distinguishable among the slender paraboloids of the rods. The thick arrow indicates the outer limiting membrane. The outer granular layer (g) is relatively thick (4-5 nuclei), and the inner granular layer (q) is 6-7 nuclei thick. The open arrow indicates where synaptic terminals are localized; n, inner plexiform layer, h, ganglion layer. Scale bar = 50 pm.

The synaptic ribbons in the cone spherules are larger than those in the rod pedicles. The latter are small, particularly in the winter. The thickness of the outer granular layer is about 15 pm, and that of the inner granular layer is approximately 30 pm. The ganglion cells in the ganglion cell layer are sparse. Our study revealed nothing unusual in the sensory tissues of Habu. For example, according to the most recent review of retinal stiucture in reptiles (Peterson, 1992), our findings for Habu are consistent with findings for other nocturnal snakes. Whereas the Habu exhibits a relatively primitive duplex retina, this is seen in other snake species as well. Likewise, the structure and neurophys¬ iology of Habu pit organs as described in Hirosawa (1980) were considered typical for pitvipers (Crotalinae) in a review by Molenaar (1992). Although the

156

Hirosawa & Takami

development of the vomeronasal organs varies considerably among reptiles, these organs are well developed in most snakes. The Habu is no exception, and data from Takami and Hirosawa (1987, 1990) were taken as typical in an extensive review of nasal chemical senses by Halpern (1992). Accordingly, we hypothesize that Habu will respond to more or less the same stimuli that other pitvipers utilize during such vital activities as foraging, detecting enemies, and detect¬ ing potential mates. Evidence along these lines is now accumulating; Waters and Burghardt (1994) revealed a fundamental similarity between the predatory behavior of Habu and that of rattlesnakes. Habu, like rattlesnakes, exhibit strikeinduced chemosensory searching (SICS) in which stimulation associated with a successful predatory strike triggers a high rate of tongue flicking coupled with searching movements, especially of the snakes head (see also Waters et al., 1996). This finding may have practical application in efforts to control Habu, as it can be predicted that Habu will respond more strongly to chemical cues after pre¬ datory strikes than when no strike has occurred or when no prey item has been detected. Attempts to identify chemical attractants might therefore be enhanced by studying Habu responses to putative attractants during periods of SICS, when the snakes are actively searching for chemical information. More generally, we suspect that many findings regarding the sensory biology of rattlesnakes and other pitvipers will likely apply to Habu (e.g., see Molenaar’s discussion of the biolog¬ ical role of the pit organs, 1992), providing numerous potential opportunities for research on control techniques.

LITERATURE CITED Halpern, M. 1992. Nasal chemical senses in reptiles: Structure and function. In C.

Gans and D. Crews, eds., Biology of the Reptilia, vol. 18, pp. 423-523. Chicago: Univ. Chicago Press. Hirosawa, K. 1978. 1. Report of Fundamental Research Division. 1-6: A microscopic

observation on the retina of Habu, Trimeresurus flavoviridis [in Japanese with English summary]. Snake 10:20-23, 95. -. 1980. Electron microscopic observations on the pit organ of a crotaline snake Trimeresurus flavoviridis. Arch. Histol. Jap. 43:65-77. Molenaar, G. J. 1992. Anatomy and physiology of infrared sensitivity of snakes. In

C. Gans and P. S. Ulinski, eds., Biology of the Reptilia, vol. 17, pp. 367-453. Chicago: Univ. Chicago Press. Peterson, E. H. 1992. Retinal structure. In C. Gans and P. S. Ulinski, eds., Biology of

the Reptilia, vol. 17, pp. 1-135. Chicago: Univ. Chicago Press. Takami, S., and K. Hirosawa.

1987. Light microscopic observations of the

vomeronasal organ of Habu, Trimeresurus flavoviridis. Japan. J. Exp. Med. 57:163-174. -. 1990. Electron microscopic observations on the vomeronasal sensory epithe¬ lium of a crotaline snake, Trimeresurus flavoviridis. J. Morphol. 205:45-61.

Histology of Habu Sensory Organs Waters,

R. M.,

and

G. M.

Burghardt.

157

1994. Strike-induced chemosensory search¬

ing in the Habu (Trimeresurus flavoviridis). In M. Maeda, ed., Reports of Ecolo¬ gical Researches to Diminish Bites of Habu (Trimeresurus flavoviridis) in Okinawa Prefecture, vol. 17, pp. 29-34. Ozato, Okinawa, Japan: Habu Study Sect., Okinawa Prefectural Inst, of Health Env. Waters, R. M., D. Chiszar, and G. M. Burghardt. 1996. Strike-induced chemosen¬ sory searching in the Habu (Trimeresurus flavoviridis), an Asian pit viper. J. Herpetol. 30:147-151.

10 Repellents and Use of Prey Items for Delivering Toxicants for Control of Habu (Trimeresurus flavoviridis) Masahiko Nishimura

T

he Habu, a deadly viper, is greatly feared in Japan, and the nonnative Brown Treesnake has been the cause of mass extinctions of wildlife on Guam. Two

approaches have been used to decrease population densities and the likelihood of human contact with these snakes: (1) trapping and killing the snakes; and (2) repelling them from areas of human activity, especially homes and agricultural fields. In this review, I focus on studies of methods for administering a toxicant used to kill Habu directly; and on various repellents meant to deter or eject Habu from cavities, homes, and agricultural fields.

METHODS FOR CAPTURING AND KILLING SNAKES Traps with live rodents as attractants have been effective for catching Habu (Trimeresurus flavoviridis) on the Amami and Okinawa Islands of Japan, and Brown Treesnakes (Boiga irregularis) on Guam. Other methods used to decrease local densities of Habu include hand capture, which people have been doing for a long time, and trapping Habu with drift-net traps, a relatively new method. Other methods that have been tested for effectiveness in capturing or killing Habu include adhesive traps lined with a dermal toxicant (Kihara and Yamashita, 1979) and infecting Habu with pathogenic amoebae (Nakasone et al., 1979; Ishii and Sawai, this volume, Chap. 31); these two methods have yet to be applied in the field.

Potassium Chloride as an Oral Toxicant for Killing Habu One gram of KC1 (LD50 = 1.14 g/kg) was enough to kill a Habu within a day when administered orally (Katsuren et al., 1979; Kuwae and Nakata, 1980, 1982), and captive Habu died after eating chicks fitted with capsules (attached to the chicks leg) containing 1 g of KC1 (Fig. 10.1; Shiroma et al., 1980). Habu in the field also died after eating chicks fitted with capsules (Nishimura et al., 1982). An oral toxicant used for killing snakes must be harmless to other animals, and Katsuren et al. (1979) showed that a capsule of KC1 attached to a chick was harmless to several domestic and wild mammals. In addition, prey carrying poison intended

158

Repellents and Toxicants for Control of Habu

159

Figure 10.1 Chick with a capsule of KC1 (a chemical deadly to Habu) attached to its leg {top); imitation prey soaked with mouse odor (bottom).

for Habu must live long enough for Habu to discover and eat them. Unfortunately, most of the chicks in the field study were killed by cats and mongooses within a day of release, and chicks that were not killed died from other causes within sev¬ eral days (Nishimura et al., 1982). Prey with a potential for greater longevity would be necessary for killing Habu with this method.

160

Nishimura

Artificial Foods (Dummies) A possible technique for administering KC1 to Habu without the problems of live prey is to use imitation prey (dummies) to mimic the small mammals preyed on by Habu (Mishima, 1966; Nishimura et aL, 1991). Araki (1984) used dummies made of sponge covered with cloth, and several Habu swallowed them. Observa¬ tions of Habu having difficulty grasping and swallowing the original dummies led researchers to make the dummies smaller and heavier (Fig. 10.1; Nishimura, 1990a, 1991a). Habu did not vomit the dummies containing a KC1 capsule and died within several days of eating them. The key to enticing a Habu to eat a dummy is to apply scent to the dummy. Habu swallowed dummies covered with mouse skin as frequently as they swal¬ lowed real mice (Nishimura and Nohara, 1991). Dummies rubbed with the back skin (i.e., dermal odor vs. scent gland odor) of mice, rats, or rabbits, or with a man’s face (Nishimura, 1989b; Nishimura and Nohara, 1990), were swallowed less frequently, and dummies rubbed with materials extracted from mouse skin with water or organic solvent (Nishimura and Nohara, 1991; Nishimura et al., 1992) were swallowed even less often than those rubbed with skin. Habu did not swal¬ low any of the dummies rubbed with mouse excrement (Nishimura, 1989b) or with any of several artificial lipids (Nishimura and Nohara, 1990). Both furry and nude mice were swallowed with high frequency after Habu first used vomeronasal cues (i.e., extended the tongue a short distance) to examine them (Nishimura, 1992a). Apparently, mouse hair is not the stimulus that causes Habu to swallow preylike material.

Tests of Habu Repellents In Okinawajima, a variety of materials, including sulfur, have been incorrectly believed to be effective for repelling Habu. Shiroma (1984, 1985) screened several volatile materials for their effectiveness by setting them under a hiding box (refugium) for Habu; only a few materials, paint thinner among them, proved to be effective at keeping the snakes from using the hiding place. Secretions from the scent glands of snakes have strong odors, and Mason (1992) suggested such odors might serve to alarm other snakes. Habu, however, did not respond any differently to hiding boxes (Nishimura, 1988b) or T-mazes (Nishimura and Arakaki, 1989a) that were either treated or not treated with secretions from Habu scent glands. Other products we tested repelled Habu kept in an outdoor enclosure. The snakes avoided hiding cavities whose substrate was treated with creosote soap, kerosene, light oil, or heavy oil (Fig. 10.2; Nishimura, 1988b; Nishimura and Arakaki, 1989b). One of the heavy oils tested was effective for four years (Nishimura, 1995).

Repellents and Toxicants for Control of Habu

161

Figure 10.2 Hiding cavity (without a cover) whose substrate has been treated with a Habu repellent {top). Spray repellent is used to eject Habu from a cinder block hole {bottom).

Sprays for Ejecting Habu from Refugia and Fields My section receives frequent requests from residents of Okinawajima to eject Habu from their hiding places, including holes and tomb cavities, and several chemicals were screened for effectiveness in ejecting Habu from such places. More

162

Nishimura

than half the snakes tested in the laboratory were ejected from a cinder block tunnel (80 cm or 160 cm long) within 10 minutes after xylene (injected with a pump spray bottle) or kerosene-based insecticides (commercial sprays) were sprayed into the tunnel (Fig. 10.2; Nishimura and Akamine, 1988). In the field, however, these chemicals did not eject snakes from cavities of stone walls in which several snakes might have been resting (Nishimura and Kamura, 1992).

Difficulties in Studying and Applying Chemical Repellents Screenings and evaluations of chemicals for repelling Habu are difficult to conduct and interpret. The snake’s large size requires that testing devices (e.g., the Y-maze or arena described in Nishimura et al., 1988) and rearing cages be large. Therefore, sample sizes are usually small. Also, most materials are tested on snakes in captivity, and the standard amount of time allowed for snakes to habituate to testing devices (one to four days) might be too short for snakes to resume normal behavior, thus biasing results. It will be necessary to test prey odors and repellents in the field before their effectiveness can be judged. The studies reviewed above were designed to provide direct evaluations of techniques or chemicals intended to kill or repel snakes. Hence, dependent vari¬ ables included swallowing of poison vehicles or avoidance of normally preferred hiding places. Other basic methods, such as counting tongue flicks (Burghardt, 1967) or observing snakes trailing prey (Noble, 1937; Brock, 1980; Ford and Low, 1984) could have been used, but these provide only indirect tests. Information gained from these methods can be used, however, to benefit our research. For example, behavioral observations led Nishimura (1988a) to conclude that Habu might not strike prey in response to chemical cues, as is also the case for other viperid snakes (Chiszar et al., 1982); therefore, studying odors meant to induce Habu to strike poisoned prey is probably a waste of time. In another study, Habu were observed tracing the surface of dead mice with frequent flicks and short extensions of their tongues; such behaviors can be quantified to measure the attractiveness of scents applied to dummies (Nishimura, 1991a). Anecdotal information may also prove useful. For instance, Takao Kamura (pers. comm., 1985) reported that a Habu was ejected from a tomb when an insecticide was sprayed into the tomb, and a farmer reported that Habu are sometimes ejected from an agricultural field when the field is sprayed with a dilute solution of creosote soap. Ultimately, however, we need to focus on direct methods for developing Habu controls, and all methods should be tested in the field. We must also evaluate biological materials with potential for repelling snakes (e.g., the odor of an ophiophagous snake). Observing the reactions of one snake to another snake, or to a snake odor (within and between sexes and species) applied to the substrate (see Nishimura, 1989a, 1991b), might lead us to useful repellents or attractants. In this kind of study, researchers should pay particular attention to key behaviors such as tongue flicking, lip rubbing, or pushing the

Repellents and Toxicants for Control of Habu

163

snout against substrate as a snake is examining a trail or a chemical deposit (Golan et al., 1982). Chemical “fences” intended to deter Habu from entering large areas (i.e., entire villages, agricultural fields) may be difficult or impossible to develop, but we can use inexpensive net fencing to exclude Habu from these areas (Nishimura, this volume, Chap. 22). To keep Habu out of specific cavities, however, repellents may be the more practical alternative, and there is a high probability that sitespecific repellents will be effective. Because Habu need humid cavities in which to lay their eggs, an added advantage of treating cavities with repellents could be local decreases in Habu breeding (Yoshida et al., 1980; Nishimura, unpubl. data). An important concern with respect to the application of repellents is environ¬ mental pollution. Only small amounts of chemical material are necessary to eject Habu from a given cavity, but more chemical materials must be applied to keep Habu from using those cavities in the future. If we apply heavy oil or creosote soap directly onto a soil or limestone substrate, not only does pollution occur, but the repelling effect is quickly diminished as the chemical soaks into the substrate. One way to resolve these problems is to apply the chemicals to artificial absorbent substrates that can be removed later. Chipboard has been used for this purpose, and repellents applied to it remained effective for a long time (Nishimura, 1990b). The board was difficult to use in narrow, complex cavities, however, and was not effective in preventing heavy oils from soaking into the ground. There are some resins that can be sprayed into cavities, where they harden and form an artificial substrate capable of containing heavy oil, creosote soap, and other repellents (Nishimura, 1992b). Research designed to discover and develop products for killing Habu and for excluding or ejecting them from target areas must continue. Past studies have yielded valuable information that should be used to guide future research.

LITERATURE CITED Y. 1984. Studies on “imitation food” of Habu, Trimeresurus flavoviridis [in Japanese]. In M. Hara, ed., Reports of Ecological Researches to Diminish Bites of Habu (Trimeresurus flavoviridis) in Okinawa Prefecture, vol. 7, pp. 17-20. Ozato, Okinawa, Japan: Habu Study Sect., Okinawa Prefectural Inst, ot Health

Araki,

and Env. Brock, O. G. 1980. Predatory behavior of Eastern Diamondback Rattlesnakes (Crotalus adamanteus): Field enclosure and Y-maze laboratory studies emphasiz¬ ing prey trailing. Ph.D. diss., Florida State Univ., Tallahassee. Burghardt, G. M. 1967. Chemical-cue preferences of inexperienced snakes: Comparative aspects. Science 157:718-721. Chiszar, D., C. Andren, G. Nilson, B. O’Connell, J. S. Mestas Jr., H. M. Smith, and C. W. Radcliffe. 1982. Strike-induced chemosensory searching in Old World vipers and New World pit vipers. Anim. Learn. Behav. 10:121—125.

164

Nishimura N. B., and J. R. Low Jr. 1984. Sex pheromone source location by garter snakes: A mechanism for detection of direction in nonvolatile trails. J. Chem. Ecol.

Ford,

10:1193-1199. Golan, L., C. Radcliffe, T. Miller, B. O’Connell, and D. Chiszar. 1982. Trailing behavior in Prairie Rattlesnakes (Crotalus viridis). J. Herpetol. 16:287293. Katsuren, S., N. Kuwae, and K. Nakata. 1979. Efficiency of KC1 to kill Habu,

Trimeresurus flavoviridis [in Japanese]. In S. Iha, ed., Reports of Ecological

Researches to Diminish Bites of Habu (Trimeresurus flavoviridis) in Okinawa Prefecture, vol. 2, pp. 164-167. Ozato, Okinawa, Japan: Habu Study Sect., Okinawa Prefectural Inst, of Health and Env. Kihara, H., and H. Yamashita. 1979. Development of a new type of trap with adhesive seat containing pesticides [in Japanese with English summary]. Snake 11:6-10, 119. Kuwae, N., and F.

1980. Effects of the snake killing chemical KC1 on Habu, Trimeresurus flavoviridis [in Japanese]. In S. Iha, ed., Reports of Ecological Researches to Diminish Bites of Habu (Trimeresurus flavoviridis) in Okinawa Prefecture, vol. 3, pp. 75-87. Ozato, Okinawa, Japan: Habu Study Sect., Okinawa Prefectural Inst, of Health and Env. -. 1982. Physiological changes of Habu, Trimeresurus flavoviridis, killed by KC1 [in Japanese]. In S. Iha, ed., Reports of Ecological Researches to Diminish Bites of Habu (Trimeresurus flavoviridis) in Okinawa Prefecture, vol. 5, pp. 61-71. Ozato, Okinawa, Japan: Habu Study Sect., Okinawa Prefectural Inst, of Health and Env. Mason, R. T. 1992. Reptilian pheromones. In C. Gans and D. Crews, eds., Biology of the Reptilia, vol. 18, pp. 114-228. Chicago: Univ. Chicago Press. Mishima, S. 1966. Studies of the poisonous “Habu” Trimeresurus flavoviridis flavoviridis 1. Food habit of Trimeresurus flavoviridis flavoviridis on the Amami Islands [in Japanese with English summary]. Japan. J. Sanit. Zool. 17:1-21. Nakasone, T., K. Tokumura, and T. Kamura. 1979. Experimental oral inoculation of Entamoeba invadens in Sakishima-habu, Trimeresurus elegans [in Japanese]. Ann. Rep. Okinawa Prefectural Inst. Public Health 13:98-102. Nishimura, M. 1988a. Observations of the predatory behavior of Habu, Trimere¬ surus flavoviridis—preliminary direct observations and video tape recordings [in Japanese]. In C. Yoshida, ed., Reports of Ecological Researches to Diminish Bites of Habu (Trimeresurus flavoviridis) in Okinawa Prefecture, vol. 11, pp. 27-32. Ozato, Okinawa, Japan: Habu Study Sect., Okinawa Prefectural Inst, of Health and Env. -. 1988b. Tests of repellents of Habu, Trimeresurus flavoviridis, 2—selecting hiding boxes [in Japanese], In C. Yoshida, ed., Reports of Ecological Researches to Diminish Bites of Habu (Trimeresurus flavoviridis) in Okinawa Prefecture, vol. 11, pp. 12-17. Ozato, Okinawa, Japan: Habu Study Sect., Okinawa Prefectural Inst, of Health and Env. -. 1989a. Testing repellents and attractants through observation of behaviors of Habu, Trimeresurus flavoviridis, 1—preliminary observations of behaviors between individuals [in Japanese]. In C. Yoshida, ed., Reports of Ecological Researches to Diminish Bites of Habu (Trimeresurus flavoviridis) in Okinawa Prefecture, vol. 12, Nakata.

Repellents and Toxicants for Control of Habu

165

pp. 13-20. Ozato, Okinawa, Japan: Habu Study Sect., Okinawa Prefectural Inst, of Health and Env. -. 1989b. Tests feeding imitation food to Habu, Trimeresurus flavoviridis—food with smell of body surface or excrement of mice [in Japanese]. In C. Yoshida, ed., Reports of Ecological Researches to Diminish Bites of Habu (Trimeresurus flavoviridis) in Okinawa Prefecture, vol. 12, pp. 39-42. Ozato, Okinawa, Japan: Habu Study Sect., Okinawa Prefectural Inst, of Health and Env. -. 1990a. Behaviors of Habu, Trimeresurus flavoviridis, to imitation food 1 [in Japanese]. In C. Yoshida, ed., Reports of Ecological Researches to Diminish Bites of Habu {Trimeresurus flavoviridis) in Okinawa Prefecture, vol. 13, pp. 61-67. Ozato, Okinawa, Japan: Habu Study Sect., Okinawa Prefectural Inst, of Health and Env. -. 1990b. Tests of absorbents for repellents of Habu, Trimeresurus flavoviridis, 2— results through the previous year [in Japanese]. In C. Yoshida, ed., Reports of Ecological Researches to Diminish Bites of Habu {Trimeresurus flavoviridis) in Okinawa Prefecture, vol. 13, pp. 35-41. Ozato, Okinawa, Japan: Habu Study Sect., Okinawa Prefectural Inst, of Health and Env. -. 1991a. Behaviors of Habu, Trimeresurus flavoviridis, to imitation food 2—three methods to check the responses of Habu to mouse or imitation food [in Japanese]. In C. Yoshida, ed., Reports of Ecological Researches to Diminish Bites of Habu {Trimeresurus flavoviridis) in Okinawa Prefecture, vol. 14, pp. 33—45. Ozato, Okinawa, Japan: Habu Study Sect., Okinawa Prefectural Inst, of Health and Env. -. 1991b. Tests of repellents and attractants through observations of behaviors of Habu, Trimeresurus flavoviridis, 3—responses of Habu to bedding with Habu odor [in Japanese]. In C. Yoshida, ed., Reports of Ecological Researches to Diminish Bites of Habu {Trimeresurus flavoviridis) in Okinawa Prefecture, vol. 14, pp. 9-10. Ozato, Okinawa, Japan: Habu Study Sect., Okinawa Prefectural Inst, of Health and Env. -. 1992a. Behaviors of Habu, Trimeresurus flavoviridis, to imitation food 3 responses of Habu to nude mice, etc. [in Japanese]. In C. Yoshida, ed., Reports of Ecological Researches to Diminish Bites of Habu (Trimeresurus flavoviridis) in Okinawa Prefecture, vol. 15, pp. 29—33. Ozato, Okinawa, Japan: Habu Study Sect., Okinawa Prefectural Inst, of Health and Env. -. 1992b. Tests of repellents of Habu, Trimeresurus flavoviridis, 6—selection of hiding boxes [in Japanese]. In C. Yoshida, ed., Reports of Ecological Researches to Diminish Bites of Habu {Trimeresurus flavoviridis) in Okinawa Prefecture, vol. 15, pp. 11-16. Ozato, Okinawa, Japan: Habu Study Sect., Okinawa Prefectural Inst, ot Health and Env. -. 1995. Tests of repellents of Habu, Trimeresurus flavoviridis, 10—selection of hiding boxes and inhibition of crawling [in Japanese]. In Y. Oshiro, ed., Reports of Ecological Researches to Diminish Bites of Habu {Trimeresurus flavoviridis) in Okinawa Prefecture, vol. 18, pp. 21-38. Ozato, Okinawa, Japan: Habu Study Sect., Okinawa Prefectural Inst, of Health and Env. Nishimura, M., and H. Akamine. 1988. Tests of repellents of Habu, Trimeresurus flavoviridis, 1—ejecting from a tunnel [in Japanese]. In C. Yoshida, ed., Reports of Ecological Researches to Diminish Bites of Habu {Trimeresurus flavoviridis) in

166

Nishimura Okinawa Prefecture, vol. 11, pp. 6-11. Ozato, Okinawa, Japan: Habu Study Sect., Okinawa Prefectural Inst, of Health and Env.

Nishimura,

M., H.

Akamine, and

S. Arakaki. 1988. Trial devices to test attractants

and repellents of Habu, Trimeresurus flavoviridis, 1—enclosures for observation and Y-maze [in Japanese]. In C. Yoshida, ed., Reports of Ecological Researches to Diminish Bites of Habu (Trimeresurus flavoviridis) in Okinawa Prefecture, vol. 11, pp. 18-21. Ozato, Okinawa, Japan: Habu Study Sect., Okinawa Prefectural Inst, of Health and Env. Nishimura,

M.,

and

S.

Arakaki.

1989a. T-maze tests of response to chemical cues

by Habu, Trimeresurus flavoviridis, I—preliminary tests of scent glands of Habu and other materials [in Japanese]. In C. Yoshida, ed., Reports of Ecological Researches to Diminish Bites of Habu (Trimeresurus flavoviridis) in Okinawa Prefecture, vol. 12, pp. 7-12. Ozato, Okinawa, Japan: Habu Study Sect., Okinawa Prefectural Inst, of Health and Env. -. 1989b. Tests of repellents of Habu, Trimeresurus flavoviridis, 3—selection of hiding boxes [in Japanese]. In C. Yoshida, ed., Reports of Ecological Researches to Diminish Bites of Habu (Trimeresurus flavoviridis) in Okinawa Prefecture, vol. 12, pp. 21-28. Ozato, Okinawa, Japan: Habu Study Sect., Okinawa Prefectural Inst, of Health and Env. Nishimura,

M., Y.

Araki,

S.

Higa, and

S.

Nishie.

1982. Experiment using chicks

carrying KC1 capsules (field study) [in Japanese]. In S. Iha, ed., Reports of Ecological Researches to Diminish Bites of Habu (Trimeresurus flavoviridis) in Okinawa Prefecture, vol. 5, pp. 73-80. Ozato, Okinawa, Japan: Habu Study Sect., Okinawa Prefectural Inst, of Health and Env. Nishimura,

M., Y.

Araki,

H.

Ueda, and

Y.

Kawashima.

1991. Frequencies of prey

items of Habu, Trimeresurus flavoviridis (Viperidae), in the Okinawa Islands 1. Snake 23:81-83. Nishimura,

M.,

and

T.

1992. Tests of repellents of Habu, Trimeresurus

Kamura.

flavoviridis, 7—tests to eject Habu from stone wall [in Japanese]. In C. Yoshida, ed., Reports of Ecological Researches to Diminish Bites of Habu (Trimeresurus flavoviridis) in Okinawa Prefecture, vol. 15, pp. 21-23. Ozato, Okinawa, Japan: Habu Study Sect., Okinawa Prefectural Inst, of Health and Env. Nishimura,

M.,

and

Y.

Nohara.

1990. Tests to feed imitation food to Habu,

Trimeresurus flavoviridis, 2—feeding in cages [in Japanese]. In C. Yoshida, ed., Reports of Ecological Researches to Diminish Bites of Habu (Trimeresurus flavoviridis) in Okinawa Prefecture, vol. 13, pp. 51-60. Ozato, Okinawa, Japan: Habu Study Sect., Okinawa Prefectural Inst, of Health and Env. -. 1991. Tests to feed imitation food to Habu, Trimeresurus flavoviridis, 3— feeding in cages and in arena [in Japanese]. In C. Yoshida, ed., Reports of Ecological Researches to Diminish Bites of Habu (Trimeresurus flavoviridis) in Okinawa Prefecture, vol. 14, pp. 25-32. Ozato, Okinawa, Japan: Habu Study Sect., Okinawa Prefectural Inst, of Health and Env. Nishimura,

M., Y.

Nohara, and

I.

Moriyama.

1992. Tests feeding imitation food

to Habu, Trimeresurus flavoviridis, 4—feeding in cages [in Japanese]. In C. Yoshida, ed., Reports of Ecological Researches to Diminish Bites of Habu (Trimeresurus

Repellents and Toxicants for Control of Habu

167

flavoviridis) in Okinawa Prefecture, vol. 15, pp. 25-28. Ozato, Okinawa, Japan: Habu Study Sect., Okinawa Prefectural Inst, of Health and Env. Noble, G.

K. 1937. The sense organs involved in the courtship of Storeria, Thamnophis

and other snakes. Bull. Am. Mus. Nat. Hist. 73:673-725. Shiroma, H. 1984. The box device to check the effect of repellents of Habu, Trimeresurus flavoviridis [in Japanese]. In M. Hara, ed., Reports of Ecological Researches to Diminish Bites of Habu (Trimeresurus flavoviridis) in Okinawa Prefecture, vol. 7, pp. 23-29. Ozato, Okinawa, Japan: Habu Study Sect., Okinawa Prefectural Inst, of Health and Env. -. 1985. Test of chemicals to control Habu, Trimeresurus flavoviridis, in the box device [in Japanese]. In C. Yoshida, ed., Reports of Ecological Researches to Diminish Bites of Habu {Trimeresurus flavoviridis) in Okinawa Prefecture, vol. 8, pp. 5-9. Ozato, Okinawa, Japan: Habu Study Sect., Okinawa Prefectural Inst, of Health and Env. Shiroma, H., S. Katsuren, Y.

Kinoshita, and

M.

Sasaoka.

1980. Extermination of

Habu, Trimeresurus flavoviridis, with the snake killing chemical KC1—outdoor experiment 1 [in Japanese]. In S. Iha, ed., Reports of Ecological Researches to Diminish Bites of Habu (Trimeresurus flavoviridis) in Okinawa Prefecture, vol. 3, pp. 67-74. Ozato, Okinawa, Japan: Habu Study Sect., Okinawa Prefectural Inst, of Health and Env. Yoshida, C., T. Kamura,

and

Y.

Araki.

1980. Study to inhibit Habu, Trimeresurus

flavoviridis, from breeding [in Japanese]. In S. Iha, ed., Reports of Ecological Researches to Diminish Bites of Habu {Trimeresurus flavoviridis) in Okinawa Prefecture, vol. 3, pp. 88-123. Ozato, Okinawa, Japan: Habu Study Sect., Okinawa Prefectural Inst, of Health and Env.

Collection and Analysis of Airborne Rat Odors Takashi Niwa Shosaku Hattori Hiroshi Kihara Toshiya Sato SlZUAKI MuRATA

M

any snake behaviors are elicited or guided by chemical cues. The chemicals are detected by the snake’s two well-developed olfactory systems: the main olfac¬

tory system and the vomeronasal system. Many published studies describe the importance of chemical cues to snakes engaged in searching for and trailing prey items. The main prey item taken by Habu inhabiting the Amami Islands is the black rat (Rattus tanezumi, formerly Rattus rattus Auct. in part; Kumanezumi in Japanese), and Habu have been caught in traps supplied with black rats as attractants; laboratory rats (Rattus norvegicus var. albinus) also serve fairly well as attractants for Habu. Using live prey for attractants in the field is impractical, how¬ ever, whereas rat odor could be an inexpensive and practical alternative. With this goal in mind, our study was designed to test methods of collecting large amounts of rat odors efficiently, and to analyze and identify the odors collected.

MATERIALS AND METHODS Collection of Rat Odor Samples A water aspirator was used to move filtered air through the headspace of a rat cage. Airborne odors from the rats and associated materials were subsequently collected as the air flowed into a cold trap that was set in a — 50°C bath of dry ice-ethanol (Fig. 11.1). The resulting ice film that formed in the cold trap was melted at room temperature. Odors were extracted by shaking the melted liquid with a small amount of hexane and were concentrated by placing the extract in a 0°C bath under low pressure. The concentrated odors were then analyzed with gas chromatography (GC) and gas chromatography-mass spectrometry (GC-MS). A large-scale collection process was needed to produce enough rat odor for use in the field. Therefore, we devised a connected system of two or three containers, each containing several rat cages (for a total of 20 rats per container; Fig. 11.2). The many rats in this system created so much humidity that condensation formed

168

Collection and Analysis of Airborne Rat Odors

169

dry ice-ethanol Figure 11.1 Device for small-scale collection of airborne volatiles from rats and associated materials.

Figure 11.2 Device for large-scale collection of airborne volatiles from rats and associated materials.

in the aspirating apparatus. In addition, a large ice block formed in the cold trap, closing off the flow of air through the system. Therefore, we devised a cooler that circulated 0°C water around the aspirated air, causing much of the moisture in the aspirated air to condense and fall into a condensate trap before the air was moved to the cold trap (Fig. 11.2). Even with this pretreatment, we also had to

170

Niwa, Hattori, Kihara, Sato, & Murata

increase the flow of air to 10-151/min to keep the rats healthy. To ensure collec¬ tion of the more volatile, smaller, and more hydrophobic odors, we directed the air flow through a hexane trap after it flowed through the cold trap. Both traps were set in the same — 50°C bath (Fig. 11.2). We analyzed the material collected from the water condensate, the cold trap, and the hexane trap. Rats were main¬ tained in the chambers overnight to generate sufficient odor material, necessi¬ tating provisions for sanitation. Initially, rat cages were lined with a bedding of wood shavings. Later, we eliminated the bedding materials (which produced their own odors) and used wire net cage floors, under which we applied sodium azide, a bactericide. The large-scale collection process was designed to produce large volumes of odor for field testing. To analyze small quantities of airborne odors directly, TenaxGC tubes (packed with the porous polymer 2,6-diphenyl-p-phenylene oxide) were set in the rat containers. A minipump was used to circulate headspace air through this Tenax-GC tube for 17 hours. We used two methods to analyze the odors col¬ lected. First, odors were thermally desorbed from the Tenax tubes into hexane and then subjected to GC or GC-MS analysis. In the second method, a flush sampler (FLS-3, Shimazu Co., Japan) was used to introduce thermally desorbed odors directly from the Tenax tubes to GC columns for analysis. Figure 11.3 illustrates an improved version of the device for large-scale collec¬ tion of rat odors. In this model, humidity is decreased by placing a waterabsorbent material (Sanwet 1M-1000, Sanyo Chemicals, Japan) under the wire net cage floors. To improve the delivery of air, filtered air was aspirated into the first in a series of rat cages; between cages and after the last cage, the air flowed through a cooler-condensate trap apparatus (Fig. 11.3). Finally, the air flowed through a cold trap ( — 40 to — 60°C). As a result of these improvements, we were able to

rat cage

cold trap

Figure 11.3 Improved device for large-scale collection of airborne volatiles from rats and associated materials.

Collection and Analysis of Airborne Rat Odors

171

reduce the rate of air flow to one-tenth of the former system (i.e., 11/min) with¬ out harming the rats. Also, this slowdown in the rate of air flow resulted in more odor compounds being trapped by minipump-Tenax bypasses. The bypasses were set in three locations within the device: one next to the cooler of the last cage, one next to the cold trap, and one between the last silica gel tube and the water aspi¬ rator. Odors collected by Tenax tubes were subjected to GC or GC-MS analyses as mentioned above.

GC and GC-MS Analysis GC analyses were performed with a Shimazu GC-9AM gas chromatograph (Shimazu Co., Japan), in which we used either a packed column (3 m) with 5% PEG-20M or a fused silica capillary column (25 m X 0.24 mm) that was wall-coated with Ulbon HR-3000A. GC-MS analyses were performed with a double-focusing mass spectrometer system (JMA-DX300/JMA-3500, Japan Electron Optics Labo¬ ratory, Japan), in which we used a fused silica capillary column (30 m X 0.32 mm) wall-coated with PEG-20M (MS condition: electron impact, IV = 70 eV).

RESULTS GC packed-column analyses of extracts collected by the first method (small-scale cold trap) yielded several peaks, mainly sesquiterpene alcohols (mol. wt. = 222). We verified the results against known GC retention times for sesquiterpene alco¬ hols and with mass spectra for sesquiterpene alcohols reported in the literature. GC-MS analyses with the capillary column yielded 13 sesquiterpene alcohol peaks (Table 11.1). Interestingly, peaks 7 and 11 did not match any reported in the lit¬ erature; however, their mass spectra were similar to those reported for epicubebol, which is found in a species of marine plant (Phaeophyta), Dictyopteris divaricata (ezoyahazu in Japanese; Suzuki et al., 1981). One of us (Murata) attempted to synthesize cubebols according to methods described by Tanaka et al. (1972). The cubebol fraction of the synthesized sample yielded three isomer peaks, and gas chromatograms showed that the retention times of the first two peaks of the cubebol fraction matched the first two sesquiterpene alcohol peaks obtained from the rat odor sample (peaks 7 and 11 in Table 11.1). Sato obtained the mass spectrum for cubebol in his laboratory by using cubebol that he separated from cubeb oil (the essential oil extracted from Piper cubeba). Sato’s spectrum matched one of those found in Murata s sample. Therefore, it may be concluded that peak 11 of the rat odor sample is cubebol and that peak 7 is a cubebol isomer (but it is not epicubebol because it cannot be pro¬ duced through this synthetic procedure). The packed-column GC analysis of odors extracted from the first aqueous con¬ densate of the large-scale device (Fig. 11.2) showed the presence mainly of sesquiterpene alcohols. The hexane trap caught many more compounds, most

172

Niwa, Hattori, Kihara, Sato, & Murata

Table 11.1 Mass spectral data of sesquiterpene alcohol peaks in GC-MS analysis of the cold trap of rat odor. Base Peak

M+

peak,

no.

m/z

m/z

7

222

43

(-) 11

222

161

(4.4) 12

222

121

(23.7) 13

222

161

(1.1) 14

222

119

(0.7) 15

222

59

(-) 16

222

41

(15.3) 17

222

95

(2.6) 20

222

161

(-) 22

222

95

(5.1) 23

222

161

(-) 24

222

59

(-) 25

222

59

(-) 26

222 (5.7)

95

Compound name

Main fragment peaks, m/z (%) 207

161

119

105

81

71

cubebol

(69.3)

(75.9)

(40.5)

(52.6)

(36.5)

(86.9)

isomer

207

121

119

105

81

43

cubebol

(64.7)

(41.6)

(40.5)

(60.3)

(42.7)

(83.9)

204

108

107

93

81

41

(5.4)

(46.0)

(39.0)

(22.6)

(91.6)

(20.1)

204

179

119

105

81

41

(33.8)

(69.1)

(39.0)

(41.2)

(34.2)

(35.3)

204

161

105

(28.5)

(56.9)

(51.8)

204

189

(6.6)

55

41

(35.6)

(64.7)

(40.9)

161

107

93

81

(74.6)

(38.0)

(35.0)

(64.7)

(40.9)

164

149

119

95

69

43

(66.8)

(74.5)

(56.2)

(75.5)

(75.2)

(96.3)

151

150

81

69

43

41

(60.6)

(63.1)

(39.1)

(35.8)

(36.9)

(39.1)

204

105

95

81

43

41

(31.0)

(28.5)

(21.5)

(29.9)

(32.4)

(27.4)

204

161

121

71

43

41

(35.8)

(44.5)

(61.7)

(35.0)

(90.5)

(40.9)

204

119

105

95

93

43

(27.0)

(58.3)

(46.2)

(37.3)

(39.2)

(40.6)

204

189

161

149

107

96

(49.0)

(50.6)

(47.2)

(65.7)

(39.3)

(38.5)

164

149

123

108

93

41

(20.6)

(35.5)

(18.4)

(22.0)

(20.0)

(25.0)

204

161

121

93

81

43

(34.2)

(37.1)

(76.2)

(31.6)

(33.6)

(62.9)

82

a-bisabolol

cubenol

epicubenol

a-elemol

unknown

a-cedrol

T-cadinol

T-muurolol

torreyol

a-eudesmol

(3-eudesmol

a-cadinol

Abbreviations: M+: molecular ion; m/z: mass-to-charge ratio; m/z (%): mass-to-charge ratio intensities relative to base peak.

likely the smaller molecules. GC-MS PEG-20M capillary analysis of the hexane trap sample yielded primarily aliphatic hydrocarbons, C12-C2o> and a group of sesquiterpene hydrocarbons (Table 11.2). Only two sesquiterpene alcohols were identified in this analysis: cubebol and its isomer, both of which are tricyclic and apparently more volatile than other sesquiterpene alcohols of the first aqueous condensate (Table 11.2).

Collection and Analysis of Airborne Rat Odors

173

Table 11.2 Sesquiterpenes and its alcohols found in hexane trap of rat odor. Peak no.

M+ m/z

Base Peak, m/z

Main fragment peaks, m/z (%)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18

204 204 204 204 204 204 204 204 204 204 204 204 204 204 204 204 222 222

105 119 161 161 93 119 161 93 161 161 105 161 119 102 132 159 207 207

119 (90), 161 (80) 105 (96), 161 (75) —

69 (90), 93 (60) 81 (80), 133 (60) 123 (75), 121 (55) 105 (60) 80 (40) 119 (40), 105 (40)

Compound name a-cubebene a-copaene (3-cubebene caryophyllene isomer caryophyllene widdrene —

a-humulene y-muurolene —

—

161 (65) 134 (60) 105 (65), 161 (60)

a-muurolene 6-cadinene a-cedrene —

—

—

—

161 (70), 43 (61) 161 (90), 43 (70)

ar-curcumene calamenene cubebol isomer cubebol

Note: For explanations of M+, m/z, and (%), see bottom of Table 11.1.

The Tenax tubes yielded sesquiterpene alcohols, sesquiterpene hydrocarbons, and more volatile smaller compounds. A higher ratio of sesquiterpene hydro¬ carbons to sesquiterpene alcohols was obtained from rat cages lined with a bed¬ ding of wood shavings. The results shown in Table 11.2 are from the wire-netfloor cages. Few compounds were caught by the last Tenax tube (the one set between the last silica gel tube and the water aspirator) in the improved version of the largescale odor collection device (Fig. 11.3); however, more than 70 compounds were identified from the first two tubes using thermal desorption into hexane and con¬ centration. The concentrates showed almost the same pattern in samples from each tube (Table 11.3): aliphatic compound*—12 straight-chain hydrocarbons, 7 alicyclic hydrocarbons, 12 oxygen compounds, 3 sulfur compounds, 1 halogen compound, and 1 heterocyclic compound; avomatic compounds

13 hydrocar¬

bons, 6 oxygen compounds, 1 halogen compound; and perhaps 6 contaminants.

DISCUSSION Analyses of headspace air collected from rat cages with cold traps revealed the presence of several sesquiterpene alcohols that were either from the rats them¬ selves or from their environment. This phenomenon was also reported by Niwa et al. (1985, 1986, 1987, 1988). In 1982, Wheeler et al. reported finding (E)farnesol, farnesol hydrate, and farnesol dihydrate in temporal gland secretions

174

Niwa, Hattori, Kihara, Sato, & Murata

Table 11.3 Airborne odors trapped by Tenax tube set in the improved device for large scale odor collection. Peak

M+

Base peak,

Main fragment peaks,

no.

m/z

m/z

m/z (%)

Compound name

a. Aliphatic compounds 1. Straight-chain hydrocarbons (mass chromat.)

n-decane, C10H22

41

57, 70, 55, 69

1-decene, C10H20

57

43, 71, 85,41

n-undecane, CnH24

(mass chromat.)

1-undecene, CnH22

57

43, 71, 85,41

n-dodecane, C12H26

168

70

56, 55, 69,41, 83

1-dodecene, C12H24

37

184

57

43,71,85,41

n-tridecane, Ci3H28

44

198

57

43,71,85,41

n-tetradecane, Ci4H30

47

212

57

43,71,85,41

n-pentadecane, C15H32

53

226

57

43, 71, 85, 99

n-hexadecane, C16H34

54

224

83

69, 55, 111

1-hexadecene, C16H32

58

240

57

43, 71, 85,41

n-heptadecane, Ci7H36

5

142

6

140

12

156

16

154

25

170

29

—

2. Alicyclic hydrocarbons 15

138

95

69, 67, 82,81

camphane, C10H18

50

204

69

161, 189

caryophyllene, C15H24

56

204

161

105

y-muurolene, C15H24

59

204

105

161

a-murrolene, C15H24

61

204

161

134, 105,91, 119, 189

6-cadinene, C15H24

63

204

119

—

a-cedrene, C15H24

64

202

159

—

calamenene, Ci5H24

3. Oxygen compounds 1

112

42

55, 43, 69, 84

2,4-dimethylcyclopentanone

7

100

44

56, 72

n-hexanal

9

114

71

2-methyl-3-hexanone

13

114

43 ?

15*

114

43

56, 71

2-heptanone

18

114

58

43, 71

2, 4-dimethylpentanal

28

128

43

72, 99, 29, 71, 57

3-octanone

35

126

43

41, 108, 69, 55

6-methyl-5-hepten-2-one

36

118

57

45, 41, 29, 87, 75, 100

2-n-butoxyethanol

40

154

57

43, 72, 29, 99, 85

5,7-dimethyl-7-octen-3-one

41

142

57

43, 56,41,55, 70, 98

n-nonanal

43

154

57

43,41,70, 83, 98, 112

5-methylene-4-nonanone

(mass chromat.)

dimethyl disulphide

45, 47

methyl-ethyl disulphide

79, 111,45,47

dimethyl trisulphide

43,41,42, 55, 69, 84

2-chlorohexane

108

benzothiazole

—

4-heptanone?

4. Sulfur compounds 4

94

23

108

61

34

126

126

—

5. Halogen compounds 33

120

56

6. Heterocyclic compound 65

135

135

Collection and Analysis of Airborne Rat Odors

175

Table 11.3 Continued. Peak

M+

Base peak,

no.

m/z

m/z

Main fragment peaks, m/z (%)

Compound name

b. Aromatic compounds 1. Hydrocarbons 3

92

91

65

toluene

8

106

91

105 (5%)

ethylbenzene

10

106

91

105 (12%)

p-xylene

11

106

91

105 (25%)

m-xylene

14

106

91

105 (20%)

20

120

105

119 (0.7%)

o-xylene 1 -methyl-4-ethylbenzene

21

120

105

119 (12%)

1,3,5-trimethylbenzene

24

120

105

119 (0.0%)

1 -methyl-2-ethylbenzene

26

120

105

119 (22%)

1,2,4-trimethylbenzene

30

134

105

119

sec-butylbenzene

32

120

105

119 (24%)

1,2,3-trimethylbenzene

52

128

128

62

142

142

—

naphthalene

141, 115

methyl naphthalene (?-)

104

phenylacetaldehyde

2. Oxygen compounds 17

120

91

42

106

106

105, 77, 51

benzaldehyde

48

162

105

77, 120

acetophenone

67

94

94

66, 65, 39

phenol

72

108

107

77, 79, 90

p-cresol

73

108

108

107, 77, 79, 90

m-cresol

148, 111

p-dichlorobenzene

3. Halogen compounds 146 146 38

c. Contaminants 1. Silicon polymer 2

(222)

?

207 (mass chromat.)

19

(370)

73

355,267

[ —OSi (CH3)2] — [74] X 3 [74] X 5

39

(444)

73

341,429, 147

[74] X 6

46

(518)

73

281, 147,415

[74] X 7

57

(592)

73

355, 147, 221

[74] X 8

2. Antioxidant 70

220

205

—

(2,6-di-ter-butyl-4-methylphenol)

Note: For explanations of M+, m/z, and (%), see bottom of Table 11.1. * Minor component.

(TGS) collected from mature African Elephants (Loxodonta africana). These were the first sesquiterpene alcohols found in mammal materials. Rasmussen et al. (1990) detected (E)-farnesol in TGS collected from bull Asian Elephants (Elephas maximus). Rasmussen et al. also reported that the components of TGS are sub¬ ject to quantitative changes during a musth episode. The first sesquiterpene hydrocarbons discovered in the secretions of mammal glands were (E)-[3- and (E,E)-ot-farnesene, which were found in the preputial

176

Niwa, Hattori, Kihara, Sato, & Murata

glands of male mice (Niwa and Ninomiya, 1991). In tests of social hierarchies among mice, Harvey et al. (1989) reported finding a- and (3-farnesene in the urine of dominant male mice one week after dominance was established. If the urine was collected directly from the dominant male’s bladder, however, neither of the two sesquiterpenes was detected. The first terpenoids discovered in the external glandular secretions of mam¬ mals came from the supracaudal gland of the Red Fox (Vulpes vulpes), and may be involved in chemical communication (Albone, 1975, 1977). One of the ter¬ penoids from the fox’s gland, dihydroactinidiolide (a monoterpenoid), is a minor component among many terpenoid compounds occurring in the plant Actinidia polygama (matatabi in Japanese). Matatabi is famous for its strong effects on the behavior of felids, and this minor component also has a weak effect on felids, but the effect of this dihydroactinidiolide on the behavior of the Red Fox is unknown. Rats are not yet known to have special skin glands that secrete sesquiterpenes or sesquiterpene alcohols, and the sesquiterpenes we detected may have been secreted by sebum glands of the skin. They also may have been present in sweat, feces, and urine. Asakawa et al. (1981) reported that sesquiterpenes given to rab¬ bits by mouth were excreted in urine as sesquiterpene alcohols, which may have been produced by detoxification in the liver (Asakawa et al., 1981; Ishida et al., 1982). As mentioned above, the sesquiterpene alcohols found in TGS of elephants, the sesquiterpene hydrocarbons found in the preputial glands of mice, and the monoterpenoids found in the supracaudal glands of Red Foxes are expected to perform specific roles in intra- or interspecific chemical communication, but their modes of action remain unknown. We do know that behavioral episodes may result in changes in the chemical components of glandular secretions, and the changes may be directly or inversely related to the behaviors. We obtained more than 70 volatile and small components from rat headspace air by using the Tenax-GC tube for collection (Table 11.3). In Table 11.4, we com¬ pare these results with those obtained from the urine of House Mouse (Mus musculus) by Schwende et al. (1986), wolf (Canis lupus) by Raymer et al. (1986), and Red Fox by Jorgenson et al. (1978); from the fecal pellets of rabbit (Oryctolagus cuniculus) by Goodrich et al. (1981); and from the TGS of African Elephants by Rasmussen (1988). It is evident that many of these compounds occur in more than one species: 2 compounds are found in five species, 3 compounds are found in four species, 5 compounds are found in three species, and 26 compounds are found in two species. Besides the compounds found in the TGS from elephants, it may be noteworthy that there are some terpenoids found in the species listed: a monoterpene, linalool, and a sesquiterpene, caryophyllene, in mouse urine; a sesquiterpene, (3-gurjenene, in the fecal pellets of rabbits; and a monoterpenoid, geranylaceton, in the urine of Red Foxes. We have made significant progress in identifying the chemical components of rat odors, and this work is continuing; however, we have yet to determine what

"O

v©

C

£ f-H lb

1

1 f3 —I

g

00 .3

r

-i->

£ S

« 8, S! E

CN £ 3 B

(N

in (N

NO CN

hv CN

CU G O oo u

00 (N

cn

179

temporal gland

Elephant6

vo CN

0> G O CN

I

G

ID

in Ui

in

Pho X* cu

VO

Pi

OO CO

VO (N

VO

00 CN O



o

c o

CO

13

>~ r-H

3.25, a = 0.005. 1 Available from Manufacturing Systems, Inc., P.O. Box 816, Belle Fourche, SD 57717. 2 Available from M8cM Fur Co., Box 15, Bridgewater, SD 57319. 3 Available from most tackle shops in the United States. 4 Provided by U.S. Department of Agriculture, National Wildlife Research Center.

material, every snake in the experiment rubbed its jaws vigorously against the sub¬ strate as if attempting to remove the compound. This behavior did not occur in the control condition (y2 = 10.0, df = 1, P < 0.05; see also Dunn and Chiszar, 1993). The rubbing behavior indicates that the material was aversive. We think it may be important that the snakes showed no tongue flick response, as this sug¬ gests that elevated tongue flicking is associated with positive appetitive stimuli or neutral ambiguous cues, but not necessarily with noxious chemicals (see also Chiszar et al., 1978). If this assertion is generally true, then the interpretation of data such as those in Table 12.1 is considerably simplified. Finally, the experiments are listed in Table 12.1 in the chronological order in which they were conducted, from July 1991 to February 1993. Positive results were

Chemical Attractants for Brown Treesnakes

191

obtained early in the series of experiments (July 1991), in the middle of the series (October 1991), and late (May 1992). Hence, there is no evidence that progressive sequence effects such as sensitization or habituation occurred during our studies.

EXPERIMENT 2 The apparent aversiveness of synthetic monkey pheromone inspired an additional study. In experiment 2 we compared the response of B. irregularis to rodent prey that had been rubbed with either tap water (control), synthetic monkey pheromone, or several of the other substances listed in Table 12.1 (Russ Carmens Coon Lure #2, Bobcat Chunk Bait, synthetic fermented egg, Russ Carmen’s Hudson Seal Muskrat, James Mast #5, and Russ Carmen’s Triple Threat). Most of the commercial baits produced positive responses in experiment 1. Hence, it was predicted that these compounds should produce no aversive reactions and might even potentiate feeding responses. Synthetic monkey pheromone, on the other hand, should interfere with predatory attack if B. irregularis attends to chemical as well as visual cues before striking prey.

Method Snakes were presented with euthanized rodent prey (adult mice, Mus musculus, or weanling rats, Rattus norvegicus; all 25-45 g) suspended by the tail from for¬ ceps. These were the usual prey for our snakes, and all the snakes had been accepting such items weekly for at least two years prior to this study. The snakes had been deprived of food for one week before the present tests. Just before being presented to snakes, the rodents were lightly rubbed with a cotton-tipped applicator containing either tap water, synthetic monkey pheromone, or the other substances mentioned above. Hence, the rodents fur contained about the same amount of substance as would an applicator. Tap water was applied to eight trial prey items, and synthetic monkey pheromone was applied to six. Synthetic fermented egg was applied to two trials, Coon Lure to three, Bobcat Chunk Bait to four, Hudson Seal Muskrat to three, James Mast #5 to three, and Russ Carmen’s Triple Threat to four trials. For each trial, a snake was selected randomly, and no snake was used twice unless a month had passed between the first and second trials. Our dependent variable was latency (seconds) to strike the prey item. If the prey was not struck within 60 seconds, the trial was terminated and the snake was assigned a score of 60.

Results and Discussion Table 12.2 presents the mean strike latency for each condition, and it is clear that the conditions differed (F = 3.98; df = 2, 31; P < 0.05), with the synthetic

192

Chiszar, Dunn, & Smith

Table 12.2 Mean latency period for Brown Treesnakes (Boiga irregularis) to strike euthanized rodent prey with various chemicals rubbed on their fur. H20 (control) Mean latency to strike prey (sec) Standard error of the mean

Six substances from Table 12.1 that generated positive

Synthetic monkey

tongue flick responses_pheromone

28.3

51.6

9.4

8.3

monkey pheromone having the longest latency. Tukey’s HSD post hoc test revealed that the control mean was not significantly different from the mean of the second column of Table 12.2, but both of these means were significantly lower than the mean for synthetic monkey pheromone. Indeed, five of the six snakes in the latter condition did not strike within 60 seconds, though all of them approached the prey rapidly. Two of these snakes commenced jaw rubbing even though they never touched the prey. The snake that struck swallowed its prey after an unusually long period (35 minutes), probably because this snake spent considerable time rubbing its jaws while holding the prey. Incidentally, all of these snakes were offered control prey within 5 minutes of the synthetic monkey pheromone test, and all struck with a mean latency of 6.8 seconds. Hence, the long latencies in the test trials were not a result of diminished hunger.

GENERAL DISCUSSION Synthetic monkey pheromone should be studied in greater detail because it holds promise as a repellent (Nishimura and Kamura, 1993). Furthermore, although B. irregularis is a visually guided predator that usually resorts to chemical cues only when visual cues are unavailable (Chiszar et al., 1988b), experiment 2 revealed that the strong aversive qualities of synthetic monkey pheromone were detected by five of six snakes during their visually guided approach to prey, and the approach was terminated as a consequence. Hence, B. irregularis clearly can make use of chemical information during the brief (one to three second) period involved in approaching a prey item that has a normal visual appearance. That the presumed attractants (Table 12.2, column 2) did not result in significantly shorter strike latencies than control rodents might be disconcerting. It should be recognized, however, that rodents are the preferred prey of large B. irregularis, and our specimens were well accustomed to eating them. Hence, the snakes were already responding quickly in the control condition and it would be very difficult for any factor to produce still shorter latencies against this so-called floor effect. In larger cages where snakes could not see the initial introduction of prey, it is probable that the attractant chemicals would exert an effect on forag¬ ing behavior. Such tests should be conducted as a preliminary to field studies.

Chemical Attractants for Brown Treesnakes

193

Earlier we pointed out that the interpretation of the data in Table 12.1 is sim¬ plified if we assume that elevated rates of tongue flicking are associated with pos¬ itive appetitive stimuli rather than with noxious stimuli. Indeed, this is true for virtually all tongue flick data arising from experiments in which prey-derived chemical cues were presented on cotton-tipped applicators or in other ways. Only when predatory strikes also occur can we be certain that high rates of tongue flick¬ ing are associated with appetitive motivation. Since experiment 1 did not permit predatory strikes at cotton-tipped applicators, we must consider whether the sim¬ plifying assumption is tenable. Two facts suggest that it is. First, the noxious syn¬ thetic monkey pheromone did not produce an elevated tongue flick response in experiment 1 although this material thwarted predatory attacks in experiment 2. Second, the presumed attractants that elevated the rate of tongue flicking in experiment 1 did not thwart predation in experiment 2. Therefore, there is no reason to suppose that the materials that induced lingual investigation were aversive to B. irregularis. Finally, it is impossible to predict how Habu (Trimeresurus flavoviridis) will respond to any of the substances mentioned in this chapter. Initial results with other pitvipers (Crotalus viridis) are quite promising, however, and suggest that at least some of these substances hold promise as Habu attractants (Chiszar et al., 1993b).

ACKNOWLEDGMENTS The authors thank the U.S. Department of Interior, Fish and Wildlife Service for Award 14-16-0009-87-959. We also thank the University of Colorado (Depart¬ ment of Psychology) Animal Care Facility for providing rodent prey and for other forms of logistical support. Albert Petkus, D.V.M., provided valuable advice about the acquisition and deployment of most of the compounds listed in Table 12.1. Dr. Peter Savarie, Denver Wildlife Research Center, kindly provided an ample supply of synthetic monkey pheromone. Cameron Byall gathered the data for syn¬ thetic monkey pheromone reported in experiment 1, and he presented a prelim¬ inary account of his findings at a local scientific meeting (Byall et al., 1993). Eric Goldberg gathered the data for the 7/9/91 entry in Table 12.1.

LITERATURE CITED Burghardt,

G. M. 1970. Chemical perception in reptiles. In J. W. Johnston Jr., D. G.

Moulton, and A. Turk, eds., Advances in Chemoreception I. Communication by Chemical Signals, pp. 241-308. New York: Appleton-Century-Crofts. -. 1980. Behavioral and stimulus correlates of vomeronasal functioning in reptiles. Feeding, grouping, sex and tongue use. In D. Miiller-Schwarze and R. M. Silverstein, eds., Chemical Signals: Vertebrates and Aquatic Invertebrates, pp. 275-301. New York: Plenum Press.

194

Chiszar, Dunn, & Smith

Byall,

C., H. M.

Smith, and

D.

Chiszar.

1993. Response of Brown Tree Snakes

(.Boiga irregularis) to synthetic monkey pheromone. J. Colo.-Wyo. Acad. Sci. 25:28. D., T. M.

Chiszar,

Dunn, and

H. M.

Smith.

1993a. Response of Brown Tree Snakes

(Boiga irregularis) to human blood. J. Chem. Ecol. 19:91-96. -. In prep. Response of Brown Tree Snakes to serum and cellular residue derived from rodent and lagomorph blood. Chiszar, D., K. Fox, and H. M. Smith. 1992. Stimulus control of predatory behavior in the Brown Tree Snake (Boiga irregularis) IV: Effect of mammalian blood. Behav. Neural Biol. 57:167-169. Chiszar, D., G. Hobika, and H. M. Smith. 1993b. Prairie Rattlesnakes (Crotalus viridis) respond to rodent blood with chemosensory searching. Brain Behav. Evol. 41:229-233. Chiszar, D., K.

Kandler,

R.

Lee, and

H. M.

Smith.

1988b. Stimulus control of

predatory attack in the Brown Tree Snake (Boiga irregularis). 2. Use of chemical cues during foraging. Amphibia-Reptilia 9:77-88. D., K.

Chiszar,

Kandler, and

H. M.

Smith.

1988a. Stimulus control of predatory

attack in the Brown Tree Snake (Boiga irregularis) 1. Effects of visual cues arising from prey. Snake 20:151-155. Chiszar, D., K. M. Scudder, L.

Knight, and

H. M.

Smith.

1978. Exploratory

behavior in Prairie Rattlesnakes (Crotalus viridis) and Water Moccasins (Agkistrodon piscivorus). Psychol. Rev. 28:363-368. Cooper,

W. E. Jr.,

and

G. M.

Burghardt.

1990. A comparative analysis of scoring

methods for chemical discrimination of prey by squamate reptiles. J. Chem. Ecol. 16:45-66. Dunn, T. M.,

and

D.

Chiszar.

1993. Grooming in the Brown Tree Snake (Boiga

irregularis). Bull. Psychon. Soc. 31:299-300. Fritts,

T. H. 1988. The Brown Tree Snake, Boiga irregularis, a Threat to Pacific

Islands. U.S. Fish Wildl. Serv., Biol. Rep. 88(31). Halpern,

M. 1992. Nasal chemical senses in reptiles: Structure and function. In C.

Gans and D. Crews, eds., Biology of the Reptilia, vol. 18, pp. 423-523. New York: Academic Press. Halpern, M., and J. L.

Kubie.

1980. Chemical access to the vomeronasal organs of

garter snakes. Physiol. Behav. 24:367-371. Hayashi,

Y., H.

Kihara,

H.

Tanaka, and

M.

Kurosawa.

1984. Evaluation of a bait

trap for Habu, the venomous snake, Trimeresurus flavoviridis. Japan. J. Exp. Med. 54:171-175. Kirk,

R. E. 1982. Experimental Design. Belmont, Calif.: Brooks/Cole.

Nishimura,

M.,

and

T.

Kamura.

1993. Tests of repellents of Habu, Trimeresurus

flavoviridis 9—tests to eject Habu from stone wall 2 [in Japanese]. In M. Yamakawa, ed., Reports of Ecological Researches to Diminish Bites of Habu (Trimeresurus flavoviridis) in Okinawa Prefecture, vol. 16, pp. 21-24. Ozato, Okinawa, Japan: Habu Study Sect., Okinawa Prefectural Inst, of Health and Env. Nishimura,

M., Y.

Nohara, and

I.

Moriyama.

1993. Tests to feed imitation food

to Habu, Trimeresurus flavoviridis 5—feeding in cages and observation of snake behaviors. In M. Yamakawa, ed., Reports of Ecological Researches to Diminish Bites

Chemical Attractants for Brown Treesnakes

195

of Habu (Trimeresurus flavoviridis) in Okinawa Prefecture, vol. 16, pp. 25 30. Ozato, Okinawa, Japan: Habu Study Sect., Okinawa Prefectural Inst, of Health and Env. Rodda,

G. H.,

and

D.

Chiszar.

1993. Are cotton swab presentations a sufficient

model for ophidian prey recognition? Poster presented at the meeting of the American Society of Ichthyologists and Herpetologists, Austin, Tex., 27 May-2 June (see pp. 265-266 of Program and Abstracts). Rodda, G. H., and T. H. Fritts. 1992. The impact of the introduction of the Brown Tree Snake, Boiga irregularis, on Guam’s lizards. J. Herpetol. 26:166-174. Rodda,

G. H., T. H.

Fritts, and

P. J.

Conry.

1992. Origin and population growth of

the Brown Tree Snake, Boiga irregularis, on Guam. Pac. Sci. 46:46-57. Shiroma,

H. 1993. Tests of attractants of Habu, Trimeresurus flavoviridis V [in

Japanese]. In M. Yamakawa, ed., Reports of Ecological Researches to Diminish Bites of Habu (Trimeresurus flavoviridis) in Okinawa Prefecture, vol. 16, pp. 1—7. Ozato, Okinawa, Japan: Habu Study Sect., Okinawa Prefectural Inst, of Health and Env. Siegel,

S. 1956. Nonparametric Statistics for the Behavioral Sciences. New York:

McGraw-Hill.

13 Integrated Pest Management: The Case for Pheromonal Control of Habu (Trimeresurus flavoviridis) and Brown Treesnakes (Boiga irregularis) Robert

T. Mason

T

he study of chemical communication, semiochemicals (chemicals with signal function), and pheromones is recognized as a small but critical area of rep¬

tilian social behavior. Indeed, reptiles in general, and snakes in particular, are arguably the most sensitive of all vertebrates to chemical cues in the environment. Chemical cues are very efficient energetically in that they are cheap to produce, they relay messages long after the producer is gone, and they work both in the dark and over very great distances. These behavior-modifying chemicals may be of value to vertebrate pest managers interested in inducing a variety of specific responses such as trailing, aggregation, mate selection, courtship, and repellency. Great expectations exist because pheromones are ecologically friendly, chemically safe, and efficient in terms of cost of production and the small quantities usually needed to produce the desired effect (Shani, 1991). The three main pest management uses of vertebrate pheromones are mass trapping, disruption of communication, and repellency. Pheromones used for attraction (mass trapping) and mating disruption have proven to be quite effec¬ tive in selected examples of insect pest management (Ridgway et al., 1990). These approaches also seem to hold the most promise for snake management. The use of semiochemicals for repellency is less well developed.

GENERAL CONSIDERATIONS Attraction and mating disruption appear at this time to hold the most promise for biological control because behavioral responses are elicited at very low pheromone concentrations, and pheromones produce highly specific behaviors and have potential for immediate use in pest management. Pheromones may indirectly kill snakes by luring them into traps, to toxicants, or to pathogens; or they may alter normal reproductive or aggregation behavior. In some cases, such as mating behavior, pheromones may affect only one sex; however, aggregation and trailing pheromones affect both sexes. No cases are known in which

196

Pheromonal Control of Habu and Brown Treesnakes

197

pheromones leave toxic residues or directly affect species other than the one that produces them. A novel strategy that will not be discussed in regard to integrated pest man¬ agement of reptiles warrants mention here: the use of pheromones in conjunc¬ tion with naturally occurring pathogens. For example, one might use pheromones to attract snakes to pathogen-contaminated “traps,” from which contaminated snakes would then be allowed to escape. During aggregations or mating, the pathogen would be passed on to additional individuals, which in turn would pass it on to others. This strategy necessitates detailed knowledge of potential conta¬ gious pathogens in reptiles (Cooper and Jackson, 1981; Frye, 1981; Hofl et al., 1984). Research into potential viral pathogens of Brown Treesnakes is currently under way (D. Nichols, pers. comm., 1997).

ATTRACTANT PARADIGMS Development of attractants usually involves five steps (modified from Roelofs, 1979): 1. Chemical characterization of the chemical(s) and documentation of their potency by laboratory and field tests. The relative attractancy of a defined mixture may be determined by comparison with live snake— attractant traps or by the percentage caught from a known population. 2. Trap catch should be maximized by determining the best release rate, component ratio, and trap design. The blends used in efficacy tests should be well described, including the chemical purity and compo¬ nent ratios; type of dispenser; chemical load and release rates; type of trap, including size, shape, and method of snake killing or contain¬ ment; and placement in the field. 3. Field tests prior to large-scale efficacy tests should determine the opti¬ mum trap spacing and density for the specified population levels. 4. Toxicants, sterilants, or pathogens may be combined with attractants. If this is to be done, it is necessary to define an acceptable delivery method for the combination and to provide data on the number of bait centers needed to effect a significant population reduction. 5. Large-scale efficacy tests should be of a duration at least equal to that of the mating season. The plot size should be large enough that immi¬ gration from surrounding, untreated areas does not mask the test results.

MATING DISRUPTION Another prospect for the control of pest snakes with pheromones is by use of a technique known as “male confusion. If males locate females primarily through

198

Mason

pheromone plumes or trails, and if the environment can be permeated with artificial pheromone—masking natural pheromone trails, saturating the males receptors, or inducing habituation—the number of males locating females may be greatly reduced, and this decrease in mating activity may result in lower pop¬ ulation levels in following generations (Birch and Haynes, 1982). To demonstrate mating disruption, small-scale test plots are established in which traps supplied with sexually attractive females or synthetic pheromones are surrounded by trails of synthetic pheromone. Alternatively, sexually attractive, unmated females are placed in the center of a plot to determine if males can find and mate with them. It is crucial that pheromone trails lead to the female’s loca¬ tion. Experimental plots are paired with control plots, each of which contains the control trap or unmated female but has no surrounding pheromone trails. There should be a significant reduction in the number of males that reach the traps with synthetic pheromones or those with live females as attractants in those trials in which the trap is surrounded by decoy pheromone trails. Multigeneration popu¬ lation monitoring is needed to determine if mating disruption has any demo¬ graphic effect.

SNAKE BEHAVIORS MEDIATED BY PHEROMONES Studies of pheromones in the Habu (Trimeresurus flavoviridis) and the Brown Treesnake (Boiga irregularis) are extensions of earlier research conducted on a North American colubrid, the gartersnake (Thamnophis sirtalis), and the Swedish adder (Vipera berus). G. K. Noble (1937) was the first investigator to study the pheromones that female snakes use to attract males. He discovered that the source of the sex pheromones is the dorsal skin of the female, and not the cloacal gland secretions as others had proposed (Baumann, 1929). That the skin is the source of sex pheromones in many snakes has been known for some time by zoo workers (Radcliffe and Murphy, 1983). Indeed, this was confirmed by a series of experiments in which skin tubes from male and female gartersnakes were used to induce courtship from males (Gillingham and Dickinson, 1980). During the breeding season, male gartersnakes and male adders initiate courtship behavior in response to a pheromone on the female’s dorsum (Nilson, 1980; Andren, 1982, 1986). Rapid tongue flicking delivers these cues to the male’s vomeronasal organ. The vomeronasal organ is the sole mediator of the reception and perception of pheromone cues in those snakes studied to date (Kubie et al., 1978; Kubie and Halpern, 1979; Halpern and Kubie, 1980; Andren, 1982). Treatment with exogenous estrogen causes intact, reproductively inactive, and ovariectomized female Thamnophis s. sirtalis to become attractive to and stim¬ ulate courtship from sexually active males (Crews, 1976). Thus, it appears that ovarian steroids positively affect the expression of sexual attractivity in female gartersnakes. Estrogen treatment in conjunction with shedding appears to be the most effective means of eliciting pheromone production (Kubie et al., 1978).

Pheromonal Control of Habu and Brown Treesnakes

199

To characterize the sex pheromones of the gartersnake, Garstka and Crews (1981, 1986) used plasma from estrogen-injected females and males to induce courtship from sexually active males, but they were unable to identify the com¬ pound responsible for the effect. The sex attractiveness pheromone of the Red-sided Gartersnake, Thamnophis sirtalis parietalis, has now been isolated, identified, and synthesized (Mason et al., 1989, 1990, 1993). The pheromone, a novel series of long-chain saturated and monounsaturated methyl ketones, elicited courtship behavior from sexually active males in the field. To date, this is the only reptile pheromone identified. Work on the reproductive behavior and pheromonal communication of the Habu has been conducted primarily by Masahiko Nishimura and his colleagues at the Habu Study Section, Okinawa Prefectural Institute of Health and Environ¬ ment. A first report of the mating behavior of the Habu describes the male and female being intertwined (Nishimura et al., 1983), a behavior seen in many snakes during mating. Subsequent reports of mating Habu describe similar behaviors (Nishimura, 1990a, 1991b). Like the gartersnake and the adder, the male Habu also appears to tongue-flick chemical cues off the female’s dorsum (Nishimura, 1989, 1990a, 1991b). Chin rubbing and rapid tongue flicking by males are indi¬ cative of courtship. To test the hypothesis that estrogen injection causes female attractiveness in Habu as in gartersnakes, Nishimura and Kamura (1984) injected female Habu with a range of concentrations of estradiol, from 1% to 10 times the concentra¬ tion (53 pg estradiol benzoate/lOOg BW) reported by Garstka and Crews (1981). In contrast to the results with gartersnakes, however, protein and lipid extracts of estrogen-treated female Habu skin failed to elicit courtship behavior from sexu¬ ally active males. In a follow-up study (Nishimura and Kamura, 1985), 13 females were treated with estradiol benzoate, each receiving injections ranging from 10% to 10 times the standard concentration of 53pg/100g BW. Eight males received standard daily testosterone injections of 667pg/100g BW in order to induce courtship behavior. Extracts of proteins and lipids were derived from the females’ blood, skin, fat bodies, liver, and ovarian follicles. All of these extracts failed to elicit courtship behavior from males. Nishimura and Kamura (1984) performed the same experiments Noble (1937) used to demonstrate that the source of the female sex attractiveness pheromone is the dorsal skin surface. The dorsal surface of females was rubbed on glass plates and wooden and glass models. These skin lipid trails were sufficient in Noble s experiments to initiate chin rubbing, tongue flicking, and sexual behavior fiom males, but Nishimura and Kamura observed no courtship behavior in response to the rubbings. The most recent research on gartersnake pheromones has concentrated on the skin lipids as the source of sex attractiveness. As gartersnakes and Brown Treesnakes are both members of the family Colubridae, it was postulated that their sex pheromones might be related chemically as well. Hexane extracts of female

200

Mason

gartersnakes were fractionated and yielded one active fraction that contained a novel series of saturated and monounsaturated methyl ketones ranging in molecular mass from 422 to 534 daltons (Mason et al., 1990). Using the same experimental paradigm, Murata et al. (1991) isolated from Brown Treesnakes a similar series of saturated and monounsaturated methyl ketones. The main components, however, were a novel additional series of methyl ketodienes. Ketodienes have never been detected in gartersnake skin lipids. Behavioral trials with Brown Treesnakes using ketodienes have been ambiguous. The major problem is that reliable sexual behavior is difficult to induce in Brown Treesnakes in captivity. Similar investigations of the skin lipids of Habu are now being planned.

TRAILING AND AGGREGATION BEHAVIOR Studies of trailing behavior in snakes have focused primarily on three behaviors: detection and location of conspecifics during the breeding season, aggregation, and migration to winter hibernacula. Only the first two are relevant to this dis¬ cussion. At least 10 species in five families of snakes are known to utilize pheromone trails (see Ford, 1986; Mason, 1992, for review). Most reports of trail¬ ing in snakes were recorded during the breeding season. It is generally the case that male snakes preferentially follow female trails. Females in general tend not to trail males or females during the breeding season. Tests with Habu using female trails laid down on glass plates failed to elicit trailing behavior from males (Nishimura and Kamura, 1984). Such experi¬ ments have not been attempted with Brown Treesnakes, but trailing should be studied, given the number of reports in the literature of its existence in other colubrids. Another major behavior mediated by pheromones in snakes is aggregation. Gravid female snakes have been reported to aggregate, presumably to gain some benefit in gestating offspring or eggs (see reviews in Gillingham, 1987; Gregory et al., 1987). It is not clear what proximate factors are responsible for aggregation in the field. In the laboratory, temperature, humidity, and stress have been demon¬ strated to be directly related to aggregation (see Mason, 1992, for review); how¬ ever, whether or not these are major factors operating in nature has not been tested. In most investigations of aggregation in snakes, filter paper containing chemical cues left by conspecifics is placed under shelters. In a number of exper¬ imental procedures using several different species of snakes, investigators have repeatedly found that when snakes are offered a choice, they choose to occupy shelters with conditioned bedding or conspecifics over shelters without them. That this tendency to aggregate is based on chemical cues left on the substrate has been demonstrated in many studies (see Mason, 1992, for review). The source of the chemical cues differs among species, but the two most effective groups of com¬ pounds are skin lipids and cloacal gland secretions.

Pheromonal Control of Habu and Brown Treesnakes

201

Aggregation behavior has been described in both the Habu and the Brown Treesnake. Aggregations of Habu have been observed in the field (Nishimura et al., 1983); similarly, an aggregation of 14 Brown Treesnakes was reported in a hollow tree on Guadalcanal (Pendleton, 1947). Aggregation behavior in the Brown Treesnake has not yet been investigated, but aggregation behavior has been examined in the Habu. A preliminary attempt to look for aggregation behavior in the field utilized four females housed in three traps (Nishimura, 1983). The traps were observed for 49 days, but no snakes were captured. The investigators attempted to elicit aggregation with bedding previ¬ ously used by Habu; the responses were unremarkable (Nishimura, 1990b). Males and females did not respond strongly to the bedding of conspecifics, although males and females tended to associate. In a follow-up study, Nishimura (1991a) found that male and female Habu exhibited no unusual behaviors in response to the bedding of conspecifics; however, male Habu seemed to exhibit chin-rubbing behavior or at least intense chemosensory investigation of bedding from a recently mated female. This implies that male Habu are able to discriminate aggregation cues of sexually attractive females from those of males.

REPELLENTS: CLOACAL GLAND SECRETIONS All snakes have paired cloacal scent glands; these are located in the tail dorsal to the hemipenes in males and in the corresponding position in females. The glands are holocrine in nature and produce primarily lipids (Oldak, 1976). The function of the glands has been the subject of much controversy over the past century. It may be that their secretions are used to repel or alarm potential predators. Snakes often emit cloacal scent gland secretions when they are handled, espe¬ cially if they are stressed or agitated. Many authors have proposed that the secre¬ tions are a defense mechanism (see Mason, 1992, for review). Indeed, many snakes will smear the secretion of the cloacal glands on a predator by writhing the body. The malodorous excretion can be expelled rapidly by a muscular sphincter, sug¬ gesting a defensive or alarm function for the glands. Some authors have suggested that cloacal gland secretions of disturbed snakes can alarm conspecifics. Animals exposed to conspecific scent gland secretions reacted more intensely to stressful conditions than did animals that were not primed by the scent gland secretions (Graves and Duvall, 1988). A common response of a snake to these scent gland secretions is to flee. It would seem beneficial for conspecifics to be able to interpret these semiochemicals as an indicator of imminent and proximate danger. The reactions of Habu to cloacal scent gland secretions were tested in a T-maze (Nishimura and Arakaki, 1989; Nishimura and Tome, 1990), with negative results. 4o date, no behavioral studies have been conducted on the repellency of Brown Treesnake cloacal gland secre¬ tions. Studies on the chemical constituents that constitute the cloacal gland secretions have been conducted (Mason, unpubl. data).

202

Mason

CONCLUSION The role of pheromonal agents in integrated pest management strategies as applied to reptilian pests is still in its infancy. The four behaviors that seem to hold the most promise for manipulation with pheromones are sexual behavior, trailing behavior, aggregation behavior, and alarm or repellent behavior. Much more work is needed before pheromones can be employed in integrated pest man¬ agement. As humankind becomes more reluctant to add synthetic pesticides to the environment, however, the value of pheromone technology will grow. Only by integrating our knowledge of the basic biology, behavior, reproduction, and chemical ecology of ophidian pest species into an integrated pest management strategy will meaningful and significant advances be made toward controlling snake populations.

ACKNOWLEDGMENTS I thank Drs. Y. Sawai and T. Fritts for organizing the joint congress that made the exchange of ideas reflected in this chapter possible in the first place. Y. Hayashi and G. Rodda greatly facilitated the congress proceedings and this volume. The Habu Study Section, Okinawa Prefectural Institute of Public Health and Envi¬ ronment, and the Japan Snake Institute are also gratefully acknowledged for their hospitality in hosting the congress. I especially thank Masahiko Nishimura for sharing so much of his data and ideas on the use of pheromones in snakes. This work was supported by U.S. Fish and Wildlife Service Cooperative Work Agree¬ ment 141600091577 and NSF INT 9114567.

LITERATURE CITED Andren, C. 1982. The role of the vomeronasal organs in the reproductive behavior

of the adder, Vipera berus. Copeia 1982:148-157. -. 1986. Courtship, mating and agonistic behaviour in a free-living population of adders, Vipera berus (L.). Amphibia-Reptilia 7:353-383. Baumann, F. 1929. Experimente liber den Geruchsinn und den Beuteerwerb der Viper (Vipera aspis). Zeit. vergl. Physiol. 10:36-119.

Birch, M. C., and K. F.

Haynes. 1982. Insect Pheromones. London: Edward Arnold.

Cooper, J. E., and O. F. Jackson. 1981. Diseases of the Reptilia. London: Academic

Press. Crews, D. 1976. Hormonal control of male courtship behavior and female attrac-

tivity in the garter snake (Thamnophis sirtalis parietalis). Horm. Behav. 7:451460. Ford, N. B. 1986. The role of pheromone trails in the sociobiology of snakes. In D. Duvall, D. Muller-Schwarze, and R. M. Silverstein, eds., Chemical Signals in Verte¬ brates. Vol. 4, Ecology, Evolution, and Comparative Biology, pp. 261-278. New York: Plenum Press.

Pheromonal Control of Habu and Brown Treesnakes

203

F. L. 1981. Biomedical and Surgical Aspects of Captive Reptile Husbandry. Edwardsville, Kans.: Veterinary Medicine Publishing. Garstka, W., and D. Crews. 1981. Female sex pheromone in the skin and circula¬

Frye,

tion of a garter snake. Science 214:681-683. -. 1986. Pheromones and reproduction in garter snakes. In D. Duvall, D. MiillerSchwarze, and R. M. Silverstein, eds., Chemical Signals in Vertebrates. Vol. 4, Ecol¬ ogy, Evolution, and Comparative Biology, pp. 243-260. New York: Plenum Press. Gillingham, J. C. 1987. Social behavior. In R. A. Seigel, J. T. Collins, and S. S. Novak, eds., Snakes: Ecology and Evolutionary Biology, pp. 184-209. New York: Macmillan. J. C., and J. A. Dickinson. 1980. Postural orientation during courtship in the Eastern Garter Snake, Thamnophis s. sirtalis. Behav. Neural Biol. 28:211—217. Graves, B. M., and D. Duvall. 1988. Evidence of an alarm pheromone from the cloacal sacs of Prairie Rattlesnakes. Southwest. Nat. 33:339-345. Gregory, P. T., J. M. Macartney, and K. W. Larsen. 1987. Spatial patterns and movements. In R. A. Seigel, J. T. Collins, and S. S. Novak, eds., Snakes: Ecology and Evolutionary Biology, pp. 366—395. New York: Macmillan. Halpern, M., and J. L. Kubie. 1980. Chemical access to the vomeronasal organ of

Gillingham,

the garter snake. Physiol. Behav. 24:367-371. Hoff, G. L., F. L. Frye, and E. R. Jacobson. 1984. Diseases of Amphibians and Rep¬ tiles. New York: Plenum Press. Kubie, J. L., J. Cohen, and M. Halpern. 1978. Shedding enhances the sexual attractiveness of oestradiol treated garter snakes and their untreated penmates. Anim. Behav. 26:562-570. Kubie, J. L., and M. Halpern. 1979. The chemical senses involved in garter snake prey trailing. J. Comp. Physiol. Psychol. 93:648-667. Mason, R. T. 1992. Reptilian pheromones. In C. Gans and D. Crews, eds., Biology of the Reptilia, vol. 18, pp. 114-228. Chicago: Univ. Chicago Press. _. 1993. Chemical ecology of the Red-sided Garter Snake, Thamnophis sirtalis parietalis. Brain Behav. Evol. 41:261-268.

R. T., H. M. Fales, T. H. Jones, L. K. Pannell, J. W. Chinn Jr., and D. Crews. 1989. Sex pheromones in snakes. Science 245:290-293. Mason, R. T., T. H. Jones, H. M. Fales, L. K. Pannell, and D. Crews. 1990. Char¬ acterization, synthesis, and behavioral response to sex pheromones in garter snakes.

Mason,

J. Chem. Ecol. 16:27-36. Murata, Y., T. H. Jones, L. K. Pannell, H. Yeh, H. M. Fales, and R. T. Mason. 1991. New ketodienes from the integumental lipids of the Guam Brown Tree Snake, Boiga irregularis. J. Nat. Prod. 54:233-240.

G. 1980. Male reproductive cycle of the European adder, Vipera berus, and its relation to annual activity periods. Copeia 1980:729-737. Nishimura, M. 1983. A trial to attract males of Habu, Trimeresurus flavoviridis, with a female [in Japanese]. In S. Iha, ed., Reports of Ecological Researches to Diminish Bites of Habu (Trimeresurus flavoviridis) in Okinawa Prefecture, vol. 6, pp. 67-70. Ozato, Okinawa, Japan: Habu Study Sect., Okinawa Prefectural Inst, ot

Nilson,

Health and Env.

204

Mason

_. 1989. Tests of repellents and attractants through observations of behaviors of Habu, Trimeresurusflavoviridis, 1—preliminary observations of behaviors between individuals [in Japanese]. In C. Yoshida, ed., Reports of Ecological Researches to Diminish Bites of Habu (Trimeresurus flavoviridis) in Okinawa Prefecture, vol. 12, pp. 13-20. Ozato, Okinawa, Japan: Habu Study Sect., Okinawa Prefectural Inst, of Health and Env. -. 1990a. Observations of social behaviors of snakes in the Ryukyu Islands 1 [in Japanese with English table]. Biol. Mag. Okinawa 27:47—51. -. 1990b. Tests of repellents and attractants through observations of behaviors of Habu, Trimeresurus flavoviridis, 2—responses to bedding with Habu odor [in Japanese]. In C. Yoshida, ed., Reports of Ecological Researches to Diminish Bites of Habu (Trimeresurus flavoviridis) in Okinawa Prefecture, vol. 13, pp. 9-20. Ozato, Okinawa, Japan: Habu Study Sect., Okinawa Prefectural Inst, of Health and Env. _. 1991a. Tests of repellents and attractants through observations of behaviors of Habu, Trimeresurus flavoviridis, 3—responses of Habu to bedding with Habu odor [in Japanese]. In C. Yoshida, ed., Reports of Ecological Researches to Diminish Bites of Habu (Trimeresurus flavoviridis) in Okinawa Prefecture, vol. 14, pp. 9—10. Ozato, Okinawa, Japan: Habu Study Sect., Okinawa Prefectural Inst, of Health and Env. -. 1991b. Tests of repellents and attractants through observations of behaviors of Habu, Trimeresurus flavoviridis, 4—behaviors of Habu to other snakes [in Japanese]. In C. Yoshida, ed., Reports of Ecological Researches to Diminish Bites of Habu (Trimeresurus flavoviridis) in Okinawa Prefecture, vol. 14, pp. 11-18. Ozato, Okinawa, Japan: Habu Study Sect., Okinawa Prefectural Inst, of Health and Env. Nishimura, M., and S. Arakaki. 1989. T-maze tests of the effects of chemical cues on Habu, Trimeresurus flavoviridis, I—preliminary tests of scent glands of Habu and other materials [in Japanese]. In C. Yoshida, ed., Reports of Ecological Researches to Diminish Bites of Habu (Trimeresurus flavoviridis) in Okinawa Pre¬ fecture, vol. 12, pp. 7-12. Ozato, Okinawa, Japan: Habu Study Sect., Okinawa Pre¬ fectural Inst, of Health and Env. Nishimura, M., and T. Kamura. 1984. Studies to detect female sex pheromone of Habu, Trimeresurus flavoviridis [in Japanese]. In M. Hara, ed., Reports of Ecolo¬ gical Researches to Diminish Bites of Habu (Trimeresurus flavoviridis) in Okinawa Prefecture, vol. 7, pp. 37—43. Ozato, Okinawa, Japan: Habu Study Sect., Okinawa Prefectural Inst, of Health and Env. -. 1985. Studies to detect female sex pheromone of Habu, Trimeresurus flavoviridis, 2 [in Japanese]. In C. Yoshida, ed., Reports of Ecological Researches to Diminish Bites of Habu (Trimeresurus flavoviridis) in Okinawa Prefecture, vol. 8, pp. 21—28. Ozato, Okinawa, Japan: Habu Study Sect., Okinawa Prefectural Inst, of Health and Env. Nishimura, M., T. Otani, and E. Nakamoto. 1983. Mating and combat dance of the Habu, Trimeresurus flavoviridis. Japan. J. Herpetol. 10:42-46. Nishimura, M., and N. Tome. 1990. Tests by T-maze on chemical cues to Habu, Trimeresurus flavoviridis, 2—responses to the contents of scent glands of Habu [in Japanese]. In C. Yoshida, ed., Reports of Ecological Researches to Diminish Bites of Habu (Trimeresurus flavoviridis) in Okinawa Prefecture, vol. 13, pp. 5-7. Ozato, Okinawa, Japan: Habu Study Sect., Okinawa Prefectural Inst, of Health and Env.

Pheromonal Control of Habu and Brown Treesnakes Noble,

G.

205

K. 1937. The sense organs involved in the courtship of Storeria,

Thamnophis, and other snakes. Bull. Am. Mus. Nat. Hist. 73:673-725. Oldak,

P. D. 1976. Comparison of the scent gland secretion lipids of twenty-five

snakes: Implications for biochemical systematics. Copeia 1976:320-326. Pendleton,

R. C. 1947. A snake “den” tree on Guadalcanal Island. Herpetologica

3:189-190. Radcliffe, C. W.,

and

J. B.

Murphy.

1983. Precopulatory and related behaviors in

captive crotalids and other reptiles: Suggestions for future investigation. Inti. Zoo Yearb. 1983:163-166. Ridgway, R. L., R. M. Silverstein,

and

M. N.

Inscoe,

eds. 1990. Behavior Modify¬

ing Chemicals for Insect Management. New York: Marcel Dekker. Roelofs,

W. L., ed. 1979. Establishing Efficacy of Sex Attractants and Disruptants for

Insect Control. Lawrence, Kans.: Entomology Society of America. Shani,

A. 1991. Will pheromones be the next generation of pesticides? J. Chem. Educ.

59:579.

, V

v.

-- ,

/*

?

'

TibiM 9 ||I|H

Part IV

POPULATION BIOLOGY

A crucial but rarely observed event in the history of any population is the moment when it began. Although the origin of the Habu population is lost in the recesses of geologic time, the Brown Treesnake population on Guam was established in this century, and new populations may be starting elsewhere at any moment. Unfortu¬ nately, we will probably never know the details of how any popu¬ lation began, but we can obtain clues from the records of snakes that are intercepted outside their known range. In this section’s first chapter, Fritts, McCoid, and Gomez (Chap. 14) piece together a re¬ markably large data set about Brown Treesnakes that have arrived on new islands and their means of transport. The alacrity with which Brown Treesnakes enter cargo containers and vehicles is a characteristic that sharply distinguishes them from Habu and many other snakes. Snakes are normally cryptic and secretive, which makes it ex¬ tremely difficult for humans to assess their population densities, movements, and other characteristics. Thus it is not surprising that snake demographers rely on radiotelemetry as a tool. Radiotelemet¬ ric data have been incorporated into the three other chapters in this section. Tanaka, Wada, Hayashi, and Ikeda (Chap. 15) describe the Habu’s relatively small-scale movements, and Tanaka, Hayashi, and Wada (Chap. 16) explain the removal methods they use for estimat¬ ing population density in this species. In particular, they introduce the Tanaka catch-curve computation method, a technique that may have broad applicability to reptiles. They also provide a useful con¬ ceptual framework for population enumeration studies by segregat¬ ing the problems of perceptual bias and availability bias. Rodda, McCoid, Fritts, and Campbell (Chap. 17) describe the origin of the Guam population of the Brown Treesnake, the densities and popu¬ lation changes that have occurred there, and the factors that appear to limit populations of the snake. The authors of Chapters 16 and 17 use very different techniques for counting, but both groups have achieved unique success in the always challenging enumeration of snake numbers. Their research gives unprecedented insight into the demography of tropical and semitropical snakes.

207

.

14 Dispersal of Snakes to Extralimital Islands: Incidents of the Brown Treesnake (Boiga irregularis) Dispersing to Islands in Ships and Aircraft Thomas Michael

H. J.

Fritts

McCoid

Douglas M. Gomez

E

xotic snakes pose an enormous ecological threat to islands, and it is imperative that they be eliminated from interisland transport. Our present

understanding of unintentional translocation by ship or aircraft is necessarily based on the few cases in which snakes were fortuitously discovered while they were in the process of dispersal or colonization. Information from each such incident is precious, and it is highly desirable that all details be conveyed to responsible researchers, however discomfiting such information might be to those involved. In this chapter we compile all the information available for Brown Treesnakes, Boiga irregularis, as of 1993 plus three additional incidents from 1994, including incidents noted in Fritts, 1987, 1988; and McCoid and Stinson, 1991. Extralimital dispersal of Brown Treesnake to sites such as Spain in 1994 and Corpus Christi, Texas, in 1993 is not included in this discussion of island records.

SOURCES OF DATA The documentation of extralimital occurrences (i.e., outside the native range) is complicated by the widely diverging sources of information and data quality. Ideally, the animals are captured and preserved, and eyewitnesses are promptly interviewed by knowledgeable authorities. More often, the snakes are killed and destroyed before an identification can be definitively verified. Most tenuous aie cases in which snakes are sighted but escape capture. In evaluating reports, we endeavor to (1) establish some confidence that the informant actually saw what was reported; (2) realize the inaccuracy of verbal descriptions of coloi, size, and other characteristics; and (3) evaluate the circumstances in which the incident oc¬ curred, as a source of supplemental information. The Brown Treesnake is an extremely slender snake. Consequently, humans,

209

210

Fritts, McCoid, & Gomez

Year Figure 14.1 Putative Brown Treesnake sightings on islands outside of Guam or the snake’s native range. Documented occurrences of Brown Treesnakes are labeled with island names. For example, in 1991 there were seven reported extralimital sightings, of which five were confirmed to be Brown Treesnakes. Note that the time scale is not constant for sightings before 1986.

and especially inexperienced observers, normally underestimate its length (Rodda and Fritts, unpubl. data). This tendency to underestimate the length of slender snakes is evident in data obtained from electrical line crews (Guam Power Authority data), from snakebite victims (Fritts et al., 1994), and probably from informants involved in snake sightings in extralimital situations.

OAHU, HAWAII Documentation exists for 7 Brown Treesnakes that have reached the state of Hawaii since 1981 (Fig. 14.1). Each identification was confirmed by a qualified herpetologist. Many other species of snakes have been captured in Hawaii, and there are numerous reports of snake sightings on Oahu and other Hawaiian Islands for which the species was not identified. According to records of the Hawaii Department of Agriculture, more than 150 snakes were discovered in Hawaii during the 20-year period from 1969 through 1988; 21 snakes were discovered in 1987. The circumstances associated with the Brown Treesnakes in Oahu, Hawaii, are listed below:

Brown Treesnake Dispersal in Ships and Aircraft

211

1. April 1981, Customs area of Honolulu International Airport, Honolulu. No additional information exists on the circumstances in which this snake was discovered; the snake may have been alive when found. The specimen is preserved in the Hawaii Plant Quarantine Collection. 2. 5 July 1981, Barbers Point Naval Air Station. A poorly preserved Brown Treesnake (650 mm snout-vent length [SVL]) that had not been previ¬ ously identified was found in the collections of the Hawaii Department of Agriculture’s Plant Quarantine Office. A note in the jar indicated that the snake had been found near an aircraft hangar at Barbers Point Naval Air Station by Kenneth Brown. 3. 5 May 1986, Hickam Air Force Base, Honolulu. A live Brown Treesnake (about 450 mm in length) was discovered and killed by an aircraft maintenance worker near an aircraft parked in a transit aircraft area. The specimen was found later near a stream behind Building 3242 by Mr. Mato, an entomologist, 15th Civil Engineering Squadron, on the base. 4. 4 October 1989, Hickam Air Force Base, Honolulu. A subadult Brown Treesnake (about 875 mm in length) was apparently dead when first found by ground maintenance personnel under a C-141 military air¬ craft (flight 084) that had recently arrived from Guam. No evidence of injury existed, and the snake was assumed to have died of exposure during the flight or from hyperthermia once the aircraft was on the ground. 5. 2 September 1991, Hickam Air Force Base, Honolulu. A Brown Treesnake was found at 2245 near the second row of tires on a U.S. Air Force C-5 (flight 5002) in transit to Travis Air Force Base from Andersen Air Force Base in Guam while it was being refueled. The plane had landed 20 minutes earlier, at 2225, on 2 September 1991 on the same runway as the Continental flight mentioned in item 6 below. The snake was alive but stunned, moving slowly; it remained under the aircraft after being dislodged from the tire (1-C-l tire) by ground personnel (Staff Sergeant Ben Acoba). Information obtained through Hawaii Department of Agriculture indicates that the C-5 had been serviced and then parked for several days on Guam at the extreme northern end of Andersen Air Force Base because the aircraft contained dangerous cargo. This nonstandard parking area is close to large tracts of second-growth forest and the cliff line, where snakes are abundant. 6. 3 September 1991 (within 8 hours of the previous incident), Honolulu International Airport and Hickam Air Force Base (jointly used flight line), Honolulu. A snake (about 1 m in length) with a smashed head was found on a taxiway (Alpha 2-3; 45 m from Kilo Intersection) near the main terminal building at 0725. It is not known whether it was

212

Fritts, McCoid, & Gomez dead or injured when it fell from an aircraft, or was subsequently run over by an aircraft or ground vehicle. A Continental Airlines flight from Guam (CO 006) had used that taxiway at 0630 that morning, and at least one C-5 U.S. Air Force plane had also landed from Guam several hours earlier that night. See comments under previous incident. 7. 20 December 1994, Schofield Barracks, Wahaiwa Area. A subadult Brown Treesnake (about 740 mm in length) was captured alive after being discovered in a military warehouse containing materials recently returned to Oahu from Guam. The specimen apparently had arrived nine days before in equipment that was transported by truck from Hickam Air Force Base to the army base where the snake was discovered. Other reports exist for Hawaii, but none can be specifically associated with the

Brown Treesnake. One incident involved a law enforcement agent, employed by the Federal Bureau of Investigation, who killed a small brown snake that was found under his open suitcase a few hours after he had landed in Honolulu from Guam. The snake was discarded, and the report came to our knowledge several months after the incident. One of us (THF) interviewed the observer and could not eliminate the possibility that the snake killed was a small Brown Treesnake that had entered the suitcase in Guam, rather than a blind snake (Ramphotyphlops braminus), which is smaller, darker, and has a less pronounced head. The latter

species occurs on both Guam and Oahu. Hawaiian newspapers have published several reports of other Brown Treesnake sightings that probably involved some other species instead.

MARIANA ISLANDS Saipan Several sightings of snakes have occurred on Saipan, but only in a few cases has the snake been captured and killed, allowing positive identification. The proximity of Saipan to Guam and the amount of air and ship traffic between them create a sub¬ stantial opportunity for Brown Treesnakes to arrive on Saipan, however, and this must be considered when interpreting sightings of unidentified snakes. Since a snake sighting in 1986 and a workshop on snake containment organized later that same year, a heightened awareness has existed in the Northern Marianas Division of Fish and Wildlife. Until multilingual public education campaigns about the Brown Treesnake were initiated in 1991 and 1992, however, many people, especi¬ ally non-English-speakers, were unaware of the importance of reporting snake sightings. Consequently, reports to Fish and Wildlife personnel were delayed. Such delays, coupled with language difficulties and observational deficiencies, have severely handicapped efforts to intercept snakes and document occurrences.

Brown Treesnake Dispersal in Ships and Aircraft 1. 16 June 1986, commercial seaport, Charley Dock. A moderately large snake was observed at night on the dock as it crawled out of a pile of used wooden utility poles that had recently arrived from Guam on the ship M.V. Turnon. The snake moved into the main warehouse on Charley Dock, and the frightened observer left the area immediately. A night watchman at the port was informed, but he did not report the sighting until the following day. An extensive search of the ware¬ house and dock area was coordinated by the Division of Fish and Wildlife, with the cooperation of other agencies and private compa¬ nies, but the snake was not found. 2. 1986, Middle (W-2) Road, near Saipan Hospital. A motorist reported seeing a snake cross the busy road at night. This area is directly inland from the Smiling Cove Area and near a forested area contiguous with the port and a storage area for cargo containers. 3. July 1987, commercial seaport container storage area. An executive of Sealand shipping company reported seeing a snake resembling a Brown Treesnake crawl from the corner channel of a shipping con¬ tainer as the container was being moved within the port area. The incident was not reported until several months later because the observer lived on Guam and did not realize that the snake was not a common animal on Saipan. The information on this sighting was obtained by MJM, who contacted the man for other reasons. 4. 14 May 1990, Saipan Airport. A Brown Treesnake was discovered dead at the airport inside an air cargo container belonging to Conti¬ nental Air Micronesia. The snake was in relatively fresh condition. Subsequently, an airline employee on Guam was overheard to claim responsibility for deliberately placing the snake in the container as a joke, knowing it would cause a stir on Saipan. 5. 15 November 1990, Saipan Airport. Before dawn, a departing pas¬ senger reported seeing a snake move off the concrete tarmac of an outdoor patio onto a partially illuminated grassy area. The patio was adjacent to an aircraft parking area and a baggage handling area. An airport employee was informed, and while the employee left to seek help in capturing the snake, the animal moved across the grass and disappeared into a hole at the base of a large ornamental boulder. The area was searched thoroughly by Fish and Wildlife personnel during the day and the following night for several hours, but the snake was not found. 6. 8 March 1991, Sadog Tasi Bridge on road to Lower Base, near com¬ mercial seaport. A nearly mature male Brown Treesnake was found dead on a busy road between two patches of mangrove forest. The snake had apparently been killed by automobile traffic. Based on the girth of the specimen relative to its length, it appeared to be well fed

213

214

Fritts, McCoid, & Gomez and heavier than most individuals seen on Guam. This specimen was deposited in the U.S. National Museum of Natural History. Extensive trapping and nighttime searches in the area did not result in additional captures. 7. 2 April 1991, Tanapag Village, 2 miles north of seaport, near elementary school. A motorist reported seeing in the lights of her car at night a large snake cross the highway toward a densely vegetated area across from the school yard. No snake was captured. 8. 17 April 1991, Koblerville, about 1 mile west of the Saipan Airport. A police officer reported seeing a snake at night. No other information is available. 9. December 1991, near apartments in Chalan Kanoa. A resident originally from the Philippines, and familiar with snakes in his native country, reported killing a greenish snake about 300 mm in length which he found while clearing vegetation in a residential area. The snake was discarded, and the incident was not reported for several months. 10. 14 March 1992, Lourdes (As Teo). An individual killed a snake found in a guava tree (Psidium guajava) at around 1300. The snake was reportedly brown, 0.9 m in length, and about 2 cm in girth. The snake’s body was wrapped in a newspaper and disposed of in tall grass behind the man’s house, but a search of the area two weeks later produced only the newspaper stained with blood. 11. 20 March 1992, Smiling Cove Marina, south of the commercial sea¬ port. At 0030, a snake was seen in front of a car as the snake crossed the road from the beach. The driver stopped and watched it slither into the grass on the other side of the road. The snake was described as 1.2 m long with a girth of 5 cm and brown or black in color. 12. 2 April 1992, wooded area near Smiling Cove Marina, south of the commercial seaport. At 1000, while walking through the area to check snake traps, one of us (DMG) observed a long, thin tail presumed to be a snake disappear into a dried thicket of vegetation before it could be captured. Based on the movement, color, and size of the tail, the animal was considered to be a snake. This sighting occurred about 50 m from the area where sighting no. 10 occurred 12 days earlier. 13. 21 May 1992, Papago. A motorist saw a snake in the middle of the road at 2400 and passed over it without hitting it. He stopped and reversed to examine the snake again, but it had already fled. The snake was described as black, 1 m long, and 7 cm in girth. 14. 23 May 1992, near elementary school, Tanapag Village. A motorist driving at 2300 saw a snake (estimated to be 450 mm total length)

Brown Treesnake Dispersal in Ships and Aircraft

215

cross a road onto the school grounds. The perimeter of the school grounds and the area across the road from the school had extensive woody vegetation adequate to harbor the Brown Treesnake. 15-16. 29 September 1992, near elementary school, Tanapag Village. Two sightings occurred within 20 m of each other and within a one-hour interval around midnight. Both observers, in automobiles, saw a snake cross toward a densely vegetated area across from the school yard, and independently reported their observations to Fish and Wildlife personnel. 17. 30 September 1992, road between Baker and Charley Docks, com¬ mercial seaport. A moderate-sized green snake was observed crawling across a dirt road at 1745 by a deckhand assigned to a tug stationed at Baker Dock. The report was received by Fish and Wildlife eight days later, and the observer was interviewed by DMG and THF at the site. The observer is from the Philippines and had no doubt that he saw a snake crawl from the strip of woody vegetation on one side of the road into tall grass and tangantangan woods on the other side. The man was frightened of snakes and several days later mentioned the incident to his supervisor, who relayed the information to Fish and Wildlife. 18. 4 January 1993, Marianas High School, Susupe. A visitor from Guam, who was aware of the attempt to keep snakes off Saipan, reported trying to run over a snake as it passed in front of his car on the beach road at 0300. News of the incident did not reach Fish and Wildlife personnel until several weeks later. 19. 23 March 1993, Nauru Building, Susupe. A security guard at a gar¬ ment factory reported finding a brown snake, about 500 mm long, after a dog forced it out from under cover. Four cargo containers originating in Guam had been opened and unloaded in the garment factory compound 24 hours before the sighting. 20. 11 May 1993, Papago. A snake was sighted just a few hundred meters from the spot where a sighting had occurred 21 May 1992. 21. 18 March 1994, Saipan Airport. A juvenile Brown Treesnake was cap¬ tured alive on a perimeter fence between the main terminal and the commuter terminal in the early morning hours. There have been other reports of snakes on Saipan, but we suspect these were not Brown Treesnakes. One snake sighting resulted in the capture of a live gravid Dendrelaphis caudolineatus, a diurnal snake native to Southeast Asia and adjacent archipelagos, from under an automobile in a residential area. Reports and rumors known to be erroneous are not included here.

216

Fritts, McCoid, & Gomez

Tinian A single incident involving snakes has been reported for Tinian. On 2 February 1990, one or more snakes were sighted in a shipping container on the ship M.V. Celeste from Guam while it was docked at Tinian. Careful examination of old fishing nets and other materials in the container resulted in the discovery of two blind snakes (Ramphotyphlops braminus),b\xi not the larger individual resembling the Brown Treesnake that had originally been reported by the deckhand on the ship. The container was in extremely poor condition, with numerous rusted spots facilitating escape of the snake onto the deck of the ship or into other cargo during the confusion of the search. The ship continued on to Saipan after cargo for Tinian was offloaded.

Rota Two incidents involving snakes have been reported for Rota. One, a snake alleged to have been killed and destroyed on a ship as cargo was unloaded, later was discounted as probably fictitious by Fish and Wildlife personnel who investigated the report. Another incident involved two snakes found in the same cargo container. 1-2. 22 October 1991, seaport, Rota. A dead Brown Treesnake (825 mm SVL) was found in a cargo container carrying shoring jacks (and other construction materials) and palletized food products that had been shipped from Guam. More careful inspection of the shoring jacks resulted in the discovery of another dead Brown Treesnake (637 mm SVL) inside the hollow tubes of the jacks. Visual inspection of all pipes in the shipment produced no other snakes, although some pipes were occluded, preventing adequate visual inspection. The materials had been moved from a site on northern Guam to the Guam seaport before being shipped to Rota. Presum¬ ably the snakes died of dehydration or high temperatures as a result of being in the container during hot weather.

Cocos Island, Guam A large Brown Treesnake was killed by a backhoe operator on Cocos in early July 1989. An independent sighting of a smaller snake resembling a Brown Treesnake near the hotel facilities on the island in July 1988 suggests that snakes may have colonized the island relatively recently. The island was damaged severely by major typhoons in 1988,1990, and 1992. The vegetative cover of the island was markedly altered by these storms and by construction associated with replacing tourist facilities. The status of snakes on this island, which is about 2 km from the southern tip of Guam, is unknown at present.

Brown Treesnake Dispersal in Ships and Aircraft

217

ELSEWHERE IN THE PACIFIC Kwajalein, Republic of the Marshall Islands A Brown Treesnake crawled out of the landing gear of a C-131 cargo plane at a Kwajalein Atoll army installation associated with the Kwajalein Missile Test Range on 21 April 1979. The specimen was captured, killed, and ultimately sent by U.S. Air Force personnel to the Bishop Museum (no. 8305), where it was identified as a Brown Treesnake in 1988. The person who captured the snake was interviewed in Kwajalein by THF several years later.

Pohnpei, Federated States of Micronesia Reports of snake sightings on Pohnpei, Chuuk, and Kosrae have been received and investigated by us in cooperation with local authorities. None could be shown to involve a Brown Treesnake until a Brown Treesnake was found dead on 3 November 1994 on top of a cargo container being offloaded on Pohnpei from a civilian ship that had arrived from Guam with intermediate stops in Palau, Yap, and Chuuk. A live Wolf Snake (Lycodon aulicus) was found in some Philippine lumber, and a diurnal racer (Dendrelaphis) was found dead near a school. Two other undocumented sightings involved snakes that were not identified. Reports of sightings of snakes on Chuuk and Kosrae could not be substantiated.

Wake Island A Brown Treesnake presumed to have arrived in the large volume of air traffic from the Solomon Islands or Manus (Admiralty Islands) through Guam and Saipan after World War II was reported by Bryan (1949, 1959). The snake was found alive in a tree on 28 February 1949 by the airport engineer, C. Morgan Holmes. It was captured, killed, and shipped by air to the Bishop Museum in Honolulu, Hawaii, where it was identified by C. H. Edmonson as a cat snake, probably the common and widespread Boiga irregularis, found throughout Indonesia, New Guinea, the Bismarck Archipelago and the Solomon Islands” (Bryan, 1949). A specimen of a Gopher Snake native to North America (Pituophis) in the Bishop Museum is labeled as originating from Wake, and Bryan’s earlier, more detailed account (Bryan, 1949) appears to describe a Pituophis rather than a Boiga (“its large scales were creamy-ivory in color, and scattered across its back was a pattern of dark brown scales, suggesting that of a rattler ). Whether the Wake specimen was confused with a specimen from somewhere else or was indeed a Pituophis rather than a Boiga cannot be determined. The description of the specimen suggests a Pituophis, but these data could come from Bryan rather than from the original collector or Edmonson, who identified the snake. Bryan mentioned another specimen of what was probably the same species collected on Los Negros Island, close to Manus, Admiralty Islands, presented to the Bishop

218

Fritts, McCoid, & Gomez

Museum by Lieutenant W. M. Wagner Jr. The earlier specimen was not available to Edmonson for comparison because it was on loan to the Chicago Natural His¬ tory Museum. The fact that Bryan (via Edmonson) was so specific in his reports that the snake was a Brown Treesnake that arrived from the South Pacific in mil¬ itary air traffic suggests that other incidents involving the Brown Treesnake may have been known. It is unlikely that the identity of the snake reported by Bryan can ever be determined beyond question. It is also inconceivable that anyone fa¬ miliar with snakes could confuse these two types of snakes, and this suggests that some major transposition of specimens occurred between the identification by Edmonson and the report by Bryan. Irrespective of this question, it is relevant to point out that Bryans report of a Boiga irregularis arriving at a central Pacific island in military traffic precedes the discovery of Boiga in Guam and is the first (albeit poorly documented) report of this species as a wayward dis¬ perser. Had the species already infested Guam and become known as a common stowaway in military cargo, it would be more likely that any snake would be assumed to be a Brown Treesnake, especially by observers with little knowledge of snakes.

Okinawajima, Ryukyu Islands Biologists at the University of the Ryukyus contacted us in September 1992 about the possibility that the Brown Treesnake might accidentally become established on Okinawajima. They provided photographs of an apparent adult Brown Treesnake that had been captured on the flight line at Kadena Air Force Base in central Okinawajima by Master Sergeant John D. Jackson. The snake had arrived alive and was given to the Okinawa Prefectural Department of Health, Depart¬ ment of Habu, where it was photographed before being sent to the Okinawa Zoo. The zoo kept the snake for about a year. All five photographs are clearly identifiable as a Brown Treesnake. Follow-up interviews with security and civil engineering personnel at Kadena Air Force Base confirmed that the incident had occurred, but few details were available. Attempts to locate the security personnel who found the snake and transferred it to Okinawan counterparts were unsuc¬ cessful. This incident was subsequently reported by Katsuren et al. (1996) to have occurred in 1990 and to have involved a snake 1.5 m snout-vent length. No other incidents can be substantiated, but the Okinawan biologists we spoke with believe others have occurred.

DIEGO GARCIA ATOLL, INDIAN OCEAN Information provided by U.S. Navy pest control personnel documented the dis¬ covery of a Brown Treesnake on board a civilian supply ship contracted to the U.S. Navy as it came to anchor off Diego Garcia Atoll. The specimen was identified and disposed of by military personnel. Diego Garcia is an important U.S. military

Brown Treesnake Dispersal in Ships and Aircraft

219

facility in the Indian Ocean, and the Brown Treesnake may have arrived in naval cargo transhipped through Guam.

OTHER EXAMPLES OF SNAKE INTRODUCTIONS Examples of the dispersal of snakes as a result of human traffic among continents and islands are common; nearly every museum collection in the United States has examples of boa constrictors and other tropical snakes that were found in bananas at the supermarket, at wholesale distributor warehouses, or in seaports and air ports. However, the number of introduced populations of snakes (i.e., where snakes have established reproductive populations) is quite low relative to the num¬ bers of birds, mammals, insects, and other animals that have colonized new areas as a result of human-aided dispersal. Humans’ deep-seated disdain for snakes may have furnished some protection against accidental transport to new areas, especially when ship and air traffic involved slower movements, smaller amounts of cargo, more human labor in offloading, etc. Current levels of traffic and containerized shipping of cargo may greatly increase the risk of future intioductions of snakes. The Wolf Snake, Lycodon aulicus, colonized Christmas Island in the Indian Ocean in the 1980s (Fritts, 1993). There is evidence suggesting that this species was unintentionally dispersed to the Indonesian, Philippine, and Mascarene Islands areas by humans within the last 200 years (Leviton, 1965). The species was first detected on Christmas Island in 1987, but by 1990 sightings were common¬ place and the snake was widespread on the island. The extent of ecological damage likely to result from this snake’s establishment on a formerly snake-free island cannot be estimated yet, but 5 of 10 endemic reptile species and several endemic birds are likely to be prey for this snake. A possibility exists that the venomous Eastern Brownsnake (Pseudonaja textilis) was introduced into New Guinea by World War II traffic from its native range in Eastern Australia (Slater, 1968). Nothing is known about the effect this intro¬ duction has had on the relatively diverse fauna of northern Papua New Guinea.

MODE OF DISPERSAL All incidents of Brown Treesnakes arriving on Oahu were associated with aircraft or airports. Five snakes appear to have arrived in military air traffic (Air Force and Navy); one snake was found in the Customs area of Honolulu International Air¬ port and presumably originated from commercial air traffic; one specimen was found on a taxiway used by both civilian and military aircraft. Five of the seven Brown Treesnakes were at the Hickam Air Force Base/Honolulu International Air¬ port Facility. The two exceptions were at Barbers Point Naval Air Station and Schofield Army Barracks. The snakes reported on Okinawajima and Kwajalein were also linked to military air traffic.

220

Fritts, McCoid, & Gomez Seven of 21 sightings of snakes on Saipan were in the vicinity of the seaport,

a focal point for all commercial ship traffic to the island. Three snake sightings, one of which appears to have been a prank, occurred at the Saipan Airport; an additional sighting was 1 mile west of the airport. All other sightings and reports came from areas of the island away from seaport and airport facilities. The repeated reports from the seaport (7) and Tanapag Village (4) indicate that snakes may be established, or are likely to become established, in these areas of Saipan. There is little military traffic to Saipan, and all the incidents were attributable to civilian traffic. The incidents on Tinian and Rota were similarly associated with commercial ship and barge traffic. Tinian has a large unmanned military use area, to which various military units are sent for maneuvers and special training exercises, but none of the snake sightings was associated with this sporadic military traffic. The Brown Treesnake killed on a ship at anchor near Diego Garcia Atoll is the only record of a snake reaching an extralimital island as a result of military ship traffic; however, Brown Treesnakes were discovered on board the subtender USS Proteus while it was anchored in the military port at Guam.

FACTORS CONTRIBUTING TO THE DISPERSAL OF SNAKES FROM GUAM The possibility that the Brown Treesnake will disperse to other islands from Guam is heightened by a number of factors. The snake is a successful colonizer, as evi¬ denced by its presence on numerous islands with varying ecological conditions in its native range. It tolerates second-growth habitats and is successful in main¬ taining high population levels in close contact with people. Its extreme abundance on Guam, even near urban and developed areas with partially depleted bird and mammal populations, causes it to enter transportation facilities in search of prey. Once in ports and cargo facilities, the desire to hide during the day leads snakes to enter vehicles, crates, and other materials that are commonly moved by air and sea to other islands. Guam is a major transportation center for both civilian and military traffic in the central Pacific region. Brown Treesnakes have been discovered in association with sea and air traffic from Guam on numerous extralimital islands, and unconfirmed reports exist for other islands. The detection of snakes in bulky cargo shipments by visual inspection is difficult, and it is likely that many snakes have traveled in cargo in the past without being noticed or reported.

ASSESSING THE RISK The repeated discovery of snakes in and around seaports, as has occurred in Saipan; and in airport facilities, as has occurred in Honolulu and at other sites where cargo is stored and used (Saipan), contributes to the likelihood that breed-

Brown Treesnake Dispersal in Ships and Aircraft

221

ing populations of this pest species will become established. As a good colonist of islands and a species probably capable of long-term sperm storage (Rodda et al., this volume, Chap. 2), the Brown Treesnake may be able to colonize with a minimal number of individuals. Abundant prey, good hiding sites, and sui-table habitats in immediate proximity to cargo handling areas contri¬ bute to the survival of individuals arriving in cargo. With repeated arrivals, a viable reproductive population could build up. These same factors will poten¬ tially contribute to survival of progeny and a rapid buildup of population numbers. The most critical decision in intercepting snakes on islands will be assessing when to interpret the discovery of a snake as an indication of a potential colo¬ nization rather than an isolated event. Deciding whether to treat a sighting as an isolated event or as an indication of an incipient population will help to deter¬ mine how much effort to spend in dealing with the incident, the size of the area to be searched, and the relative duration of control activities. Several questions are relevant to determining the relative threat: 1. Does sufficient habitat or other structural complexity exist to suggest that other snakes could go undetected in the area in the absence of a deliberate program to detect them? 2. Are prey sufficiently abundant and is the habitat appropriate to facili¬ tate the survival, growth, and reproduction of snakes in the vicinity? 3. Is there evidence of repeated sightings at the same site or in the same general vicinity? 4. Do the nutritional condition, intestinal contents, and circumstances of a captured snake suggest that it may have been present for some time before being discovered? 5. Does the size, reproductive state, or number ol snakes suggest a repro¬ ductive population? 6. Is the extent of human activity in the area at night sufficient to ensure opportunistic discovery of any snakes, or will deliberate nocturnal detection efforts be necessary? Juveniles and subadults (i.e., 500-900 mm SVL) are the most common age classes of Brown Treesnakes in Guam, and thus these are the sizes most likely to arrive on other islands. The discovery of adults and/or hatchlings should be given special consideration as an additional indication of the existence of a reproductive population. There are few situations on Saipan where the answers to the questions above would not suggest high risk, but the level of effort must depend on the actual cir¬ cumstances. The discovery of a snake in a ship container as the container is being offloaded might justify an intensive search of that container and any others from the same ship, but a discovery in a container that has been stored in the port might

222

Fritts, McCoid, & Gomez

indicate that control efforts be directed at all containers in the vicinity and surrounding vegetation. Considering the extreme case of the discovery of a snake in Saipan on a road in close proximity to two forest patches, a more comprehen¬ sive effort would be justified. The history of increasing snake sightings on Saipan has prompted efforts to detect resident snakes using visual searches and trapping in areas where snakes are most likely to be present. A position for a herpetologist to coordinate snake interdiction was created in the Division of Fish and Wildlife in February 1989. To date, no snakes have been captured, and only one snake (2 April 1992) has been seen during visual searches. Each week Saipan receives about 20 shipping containers (20 and 40 ft.) of material from or transhipped through Guam. An infestation rate for snakes in cargo containers of only 1% would bring 10 snakes to Saipan each year, and based on the density of snakes in Guam, such an estimate is not unrealistic. An infesta¬ tion of only 0.1% of containers would place the island at risk of snake coloniza¬ tion. Although interdiction or eradication of incipient snake colonies is expensive and frustrating, the costs are likely to be much smaller than the expenses incurred through permanent loss of native and domestic animals, compromised electric power systems, and the human health hazards associated with the Brown Treesnake. In addition, each new colonization creates a potential source area for further colonizations. For example, all islands with transportation links to Honolulu will face a much greater risk if the Brown Treesnake becomes established on Oahu.

LITERATURE CITED Bryan,

E. H. Jr. 1949. Snakes in paradise: One found in Wake. Honolulu Advertiser,

6 March 1949. -. 1959. Notes on the geography and natural history of Wake Island. Atoll Res. Bull. 66:1-22. Fritts,

T. H. 1987. Movements of snakes via cargo in the Pacific region. ’Elepaio

47:17-18. -. 1988. The Brown Tree Snake, Boiga irregularis, a Threat to Pacific Islands. U.S. Fish Wildl. Serv., Biol. Rep. 88(31). -. 1993. The Common Wolf Snake, Lycodon aulicus capucinus, a recent colonist of Christmas Island in the Indian Ocean. Wildl. Res. 20:261-266. Fritts,

T. H., M. J.

McCoid, and

R. L.

Haddock.

1994. Symptoms and circum¬

stances associated with bites by the Brown Tree Snake (Colubridae: Boiga irregu¬ laris) on Guam. J. Herpetol. 28:27-33. Katsuren,

S., M.

Nishimura, and

T.

Kamura.

1996. Snakes collected in non-native

areas in the Okinawa Islands, Ryukyu Archipelago. Biol. Mag. Okinawa 34:1-7. Leviton,

A. E. 1965. Contributions to a review of Philippine snakes, VIII. The snakes

of the genus Lycodon H. Boie. Philippine J. Sci. 94:117-140.

Brown Treesnake Dispersal in Ships and Aircraft McCoid,

M. J.,

and D. W. Stinson.

223

1991. Recent snake sightings in the Mariana

Islands. ’Elepaio 51:36-37. Slater, K. R. 1968. A Guide to the Dangerous Snakes of Papua. Port Moresby: V. P. Bloink.

15 Movements of Habu, as Observed by Radio Tracking in the Field Hiroshi Tanaka Yoshitake Wada Yoshihiro Hayashi Kenji Ikeda

W

e initiated radiotelemetry studies of Habu (Trimeresurus flavoviridis) to determine the locomotory activity cycle of the snake. Most Habu bites occur during the day and early evening, the times of greatest human activity. Professional Habu hunters, however, say that Habu move at night. Movements of Habu in a naturalistic enclosure (Tanaka et al., 1967) were exclusively nocturnal.

MATERIALS AND METHODS Radiotelemetry of terrestrial snakes is technically challenging because radiation of the signal is limited by the proximity of the transmitting antenna to the ground. We chose to use an ultra-high-frequency (UHF) 600-700 MHz transmitter to allow extra precision in location because it was vital that these highly venomous snakes be unequivocally located for recapture at the end of our observations. A disadvantage of UHF is that transmissions are easily reflected by boulders and vegetation, which block the signal and create confusing reflections. All equipment was fabricated by Ikeda and his associates using commercially available UHF components. The equipment was modified and improved repeatedly. An ideal transmitter should be small, light, and easy to attach to the snake. Our initial attempts were not entirely satisfactory. The first transmitter contained three transistors, one for transmitting 600 MHz and two for generating the 2 KHz oscillation that was converted to an audible signal on reception (Fig. 15.1; Tanaka and Wada, 1977). All circuits were enclosed in a small copper case measuring 7 X 10X13 mm (after waterproofing), and a 75 mm length of piano wire was attached for use as an antenna. In addition to the 1.5 g transmitter, the final package included two 1.3 V mercury batteries in a waterproof module weighing 12.5 g (placement on snakes, which averaged 460 g, gave a transmitter/snake weight ratio of 2.7%). Waterproofing was necessary to protect the snake from electrical shock. This model had a range of about 50 m in heavy vegetation and an expected battery life of one to two weeks.

224

Radio-Tracking Habu in the Field

225

(-2.6V)

C : 200PF O: 500PF L : 0.3 0 enameled wire 4 times wound at 1.5 0

Rb: 47kQ Re: 68 Q Rb:1MQ Rf: 1M 0 Rc: 200k Q

Ti: 2SC288A T2: 2SC183 lx 2SC183

Figure 15.1 Circuit diagram of original transmitter, 5 X 6 X 10 mm.

The transmitter was implanted subcutaneously, the battery module was implanted intraperitoneally, and the vertical whip antenna protruded middorsally. The flexible antenna bent on contact with vegetation but maintained a vertical position at other times. The substantial weight of this package and the require¬ ment for surgical implantation created doubts about the naturalness of the behavior monitored with this equipment. The initial design was subsequently modified (Ikeda et al., 1978, 1979) to ex¬ tend the reception range to 500 m, improve battery life to four weeks, reduce the transmitter/snake weight ratio (to about 1.7%), and facilitate attachment to the snake. Improved models used pulsed transmission (30bps) and four 2.8 V lithium cells that could be attached to the snake with tape or glue (Fig. 15.2). The earliest receiving unit, a combination of television (VHF) and AM and FM radio receivers, measured 70 X 115 X 155 mm, and weighed 1140 g including the six dry cells used for power. The signal was audibly monitored through head¬ phones as well as through a signal-strength lamp and meter (Fig. 15.3). The pri¬ mary antenna system consisted of a 17 + 22-element dual Yagi antenna mounted with a slight downward angle on a 3.6 m mast. A hand-held 14 element single Yagi antenna was used for approaching the snake on foot. Before we used this equipment on free-ranging snakes in inhabited areas, we radiotracked two snakes under more controlled conditions. In May 1970, a Habu was tracked in a naturalistic enclosure, and in June 1970 a Habu was released and tracked for a single night on an uninhabited islet. The successful recapture of this

226

Tanaka, Wada, Hayashi, & Ikeda

snake reassured residents of inhabited study areas that free-ranging Habu could be radiotracked without escape. Subsequently, snakes were radiotracked in agricultural areas on Amamioshima in the Amami Islands, Japan. The sampling of radiotracked snakes was as follows: 1 snake for three days (1971), 2 snakes for three days (1972), 3 snakes for three days (1973), 12 snakes for three days (1977), and 5 snakes for two to seven days (1978). To facilitate accurate mapping of movements, the snakes were not released at their place of capture. No more than five days elapsed between time of capture and release. All radiotelemetered snakes were adults. The snakes were tracked in July and August, well after the mating season but overlapping with the period of oviposition (late July-early August).

10cm Antenna Vinyl coated wires

L epoxy

T 14mm

\

JL % :

/ fj1 !( 1

transmitter r Li 2.8V 1i

N

30mm

epoxy

5 mm

Li Li Li

2.8V 2.8V 2.8V

T

14mm

;-L

30mm

Th.:5mm Figure 15.2 Configuration of the pulsed transmitter package. Li, lithium battery.

Figure 15.3 Components of the receiving unit.

Radio-Tracking Habu in the Field

227

RESULTS The snake tracked in the naturalistic enclosure moved only at night, during several movement bouts of about 30 minutes each (Tanaka and Wada, 1977). During the nine hours the snake was monitored on the uninhabited islet, it moved in a localized area and showed long periods of inactivity, with two episodes of greater movement resulting in nearly straight movements of 10 m and 30 m (Tanaka and Wada, 1977). The snakes released in agricultural areas (Wada et al., 1979) typically began their movements 30—60 minutes after sunset and continued moving episodically until dawn. The time of the final move was inconsistent, however, and snakes occasionally continued to move until 1100, especially on cloudy or rainy days or if the refugium selected at dawn proved insufficiently shady. Nocturnal movements were interrupted by rest periods of several hours. When moving, a typical speed was lm/min or slower, with 3m/min the maximum recorded. Movement on the first and second nights after release was greater than on subsequent nights. Movements often followed the outlines of natural features such as field edges, hills, or watercourses. The area covered by a snake in a typi-

bered in chronological order.

228

Tanaka, Wada, Hayashi, & Ikeda

cal night had a diameter of about 30 m. Daytime refugia were usually shady, moist places. Most snakes rested during the day in the same refugium they had used the previous day. To this extent, the Habu has a home range (Fig. 15.4, Table 15.1). The Habu’s habit of remaining in or returning to a previously used refugium meant that the net daily distance moved was normally zero. As a measure of cen¬ tral tendency therefore, mean net daily movement distance is a poor measure and we did not compute it. When net movement did occur, however, we noted that Habu that were radiotracked without surgical implantation were more active. Active males covered greater distances than did active females. Males tended to move for one to two hours, alternating with one hour rest periods. When Habu reached a new site after moving for a long distance, they remained relatively seden¬ tary for several days afterward.

Table 15.1 Time course of snake movements mapped in Fig. 15.4. Time

Locality__Action

14 August 1971 Start of tracking

1700 1700-1930

1

1940

2

1945

3

1955

4

2005-2050

5

2100-2240

6

2250

7

2310

8

2335

9

No movement

No movement Slow intermittent movement around 6

15 August 1971 0005

12

0010

13

0015

14

0020

15

0025-0140

16

0150

17

0200

18

0210-0320

19

0330

20

0335

21

0350

22

0400

23

0410

24

0420

25

0439

26

0450-0530

27

0540

28

No movement

Slow intermittent movement around 19

Slow intermittent movement around 27 Slow movement around 10, 21, 22, and 23 along ditch

0600-1000 1100-

♦

3

No movement

Radio-Tracking Habu in the Field

229

DISCUSSION The Habu is similar to other “nocturnal” snakes in exhibiting primarily noctur¬ nal movement, with some shifts of daytime refugium in the morning, especially if disturbed by high temperatures (Rodda et ah, this volume, Chap. 2). Short pe¬ riods of movement (e.g., 30 minutes) were interrupted by long periods (several hours) of immobility, as is typical of a sit-and-wait forager (Henderson, 1974; Shine, 1980). Movement, when it did occur, was slow, rarely exceeding 1 m/min (Montgomery and Rand, 1978; Greene and Santana, 1983). Compared with the Brown Treesnake, the Habu is relatively sedentary in the range of its movements, with a relatively small home range and recurrent use of known refugia (Greene and Santana, 1983; cf. Rodda et al., this volume, Chap. 2). Long-term radio-tracking studies that would allow us to establish the seasonal range of the Habu have not yet been conducted.

LITERATURE CITED Greene, H. W., and M. A. Santana. 1983. Field studies of hunting behavior by

Bushmasters. Am. Zool. 23:897. Henderson, R. W. 1974. Aspects of the ecology of the Neotropical vine snake Oxybelis aeneus (Wagler). Herpetologica 30:19-24. Ikeda, K., N. Iwai, Y. Wada, and Y. Hayashi. 1979. Technical development of the radiotracking system for studies of Habu movement [in Japanese with English summary]. Snake 11:29-31, 122-123. Ikeda, K., Y. Wada, Y. Hayashi, N. Iwai, H. Kihara, Y. Noboru, and H. Yamashita. 1978. The wireless tracking of movement of Habu, Trimeresurus flavoviridis [in Japanese with English summary]. Snake 10:26-39, 96. Montgomery, G. G., and A. S. Rand. 1978. Movements, body temperature and hunting strategy of a Boa constrictor. Copeia 1978:532—533. Shine, R. 1980. Ecology of the Australian Death Adder, Acanthophis antarcticus (Elapidae): Evidence for convergence with the Viperidae. Herpetologica 36:281289. Tanaka, H. 1973. Activity and behavior of Habu, Trimeresurus flavoviridis [in

Japanese with English summary]. Snake 5:116-132. Tanaka, H., S. Mishima, and Y. Abe. 1967. Studies on the behavior of Trimeresurus flavoviridis (Hallowell, 1860), a venomous snake, on Amami Oshima Island in regard to speed of movement, nocturnal activity and sensitivity to infra-red radiation. Bull. Tokyo Med. Dent. Univ. 14:79—104. Tanaka H., and Y. Wada. 1977. Venomous snakes. In M. Sasa, H. Takahashi, R. Kano, and H. Tanaka, eds., Animals of Medical Importance in the Nansei Islands in Japan, pp. 29-71. Tokyo: Shinjuku Shobo. Wada, Y., K. Ikeda, H. Kihara, Y. Hayashi, N. Iwai, Y. Noboru, and H. Yamashita. 1979. Tracing of movement of Habu, Trimeresurus flavoviridis, by

radio tracking [in Japanese with English summary]. Snake 11:32—36, 123.

16 Population Density of Habu on the Amami Islands, as Estimated by Removal Methods Hiroshi Tanaka Yoshihiro Hayashi Yoshitake Wada

T

he traditional method for estimating animal abundance involves marking and releasing individuals for subsequent capture. Using known populations in

naturalistic outdoor enclosures, we tested the validity of the mark-recapture method for the Habu, Trimeresurus flavoviridis. Because the Habu is highly ven¬ omous, however, this approach was deemed undesirable for use in populated areas because people might be envenomated by animals that could have been removed from the wild but were released for population studies. Therefore Tanaka et al. (1971) explored the use of removal methods (Zippin, 1958). Using extrapolations from the declining capture rate obtained with repeated one hour samples, Tanaka et al. (1971) developed the “catch-curve” calculation to correct for the perception bias of visual searches. That is, the method estimates the number of snakes that could have been seen and captured with unlimited effort in a night, but does not estimate the total population, which includes animals that were unavailable because they remained underground, etc. We estimated the proportion of un¬ available snakes using the difference between a known population of snakes and the number estimated to be present using the perception bias correction. Using both perception and availability correction factors, we then surveyed 33 popula¬ tions of Habu in the Amami Islands of Japan.

METHODS AND METHODOLOGICAL RESULTS The accuracy of mark-recapture methods was evaluated using an enclosure into which 100 snakes had been released (Hayashi and Tanaka, 1980). The enclosure was provided with natural vegetation and rock shelters. Professional Habu hunters searched the enclosure up to eight times to obtain the recapture data, which were analyzed using Chapmans (1951) modification of the Lincoln-Peterson index, Baileys (1951) modification, and the Paloheimo method (1963). The best estimators from the Chapman and Bailey two-capture markrecapture methods were 83.0 and 80.5, respectively. The Paloheimo computation can be applied to multiple recaptures. On the basis of four and eight capture

230

Population Density of Habu on the Amami Islands

231

events, we obtained estimates of 98.6 and 103.5, respectively. Although the Paloheimo computation provided the most accurate estimator, the confidence limits of all methods covered the real value, and the double-catch estimates were less laborious to obtain. Removal methods protect people from accidentally coming into contact with a previously caught snake. We applied this method to the decline in capture rate that occurred during four one-hour area-constrained searches of five study areas of about lha (Tanaka et al., 1971). We used four professional Habu searchers seeking snakes from around 2200 to 0200, a time of maximal snake activity. We computed our population estimate from the following relationship: nt = p(N — ]Tni)

where n, is the number of snakes captured during the zth collection, is the cumulative number of snakes that have been captured through the ith occasion, p is the slope of the regression of n, on 2n,, and N is the estimated total

population. N can be computed from the above equation, and it is also the Xintercept of the n, on Xn, regression. This method was applied to a variety of discrete plots on Amamioshima and Tokunoshima in the Amami Islands, Japan. The area of each plot was calculated from aerial photos. In our five calibration trials (four cropland, one wildland riparian zone), an average of 47% (p = 0.47) of the uncaught extrapolated total was captured during each one-hour time block. Therefore, an average of about 28% of the snakes remained undiscovered after two one-hour searches; 1 — (1 — p) — 0.72, when p = 0.47. Alternatively, one can multiply a two-pass capture total by 1.39 ( = 1/0.72) to estimate the total number of snakes that would have been found had all visible snakes been captured. We used this multiplier to correct for perception bias in our two-pass visual searches. Although the above perception correction indicates the number of snakes that would have been captured if snake searching had continued for an entire night, it does not correct for the fraction of snakes that were hidden and therefore unavailable at the time of sampling. We estimated the correction factor for this “availability” bias using two enclosure trials (Hayashi and Tanaka, 1979). In each, we compared the known snake total with the number estimated using the perception correction applied to multiple searches. The difference was assumed to be due to availability bias. In the first trial, 50 snakes were released in an enclosure and three searches of the enclosure yielded a estimate of 21.7 snakes after incorporating the perception correction. In the second trial, the respective values were 30 snakes and 11.7 estimated. Taken together, these values provided us with a correction factor of (50 + 30)/(21.7 + 11.7) — 2.4. This is the multiplier that was used to compensate for availability bias in population estimates. Because of the limited opportunities to estimate the availability bias parameter, confidence intervals for the population total were not estimated.

232

Tanaka, Hayashi, & Wada

POPULATION RESULTS AND DISCUSSION We retrospectively applied the perception and availability bias corrections to 40 population samples collected from 33 sites in the Amami Islands, 1970-1977 (Tables 16.1, 16.2). The maximum population was 19.6 Habu/ha at Kaitsu B on Amamioshima (Table 16.2), an agricultural area surrounding a pond. This maximum exceeds that for other vipers reported in a recent review (Parker and Plummer, 1987; Rodda et al., this volume, Chap. 17). Other cultivated areas ex¬ hibited more typical densities, however, with values ranging from 0 to 12/ha, and an average of 4.4 and 5.1/ha on Tokunoshima and Amamioshima, respectively. According to professional Habu searchers, Habu are most common near streams. Our searches of riparian zones yielded means of 2.6 and 3.6/ha for the two islands, densities that were lower than but did not differ statistically from those obtained on farmland (F = 0.82; df= 2, 32; P — 0.37). Two residential areas searched had estimated snake populations of 2.1 and 2.8/ha, although the unique difficulties of finding snakes in residential areas probably resulted in underestimation (indicated in the tables by a plus following the estimated density). The residential areas that were searched were adjacent to agricultural fields and probably had true snake densities similar to the agricul¬ tural areas.

Table 16.1 Estimated Habu population density for various sites on Tokunoshima, 1970-1977. Area Habitat Farm

Habu/ha

Habu/ha

Habitat

(1970-1971)

(1977)

mean

Site

(ha)

Samples

Ikema A

0.8

4

12

—

Ikema B

0.7

4

10.2

—

Inokawa

1.4

2

—

2.4

Inokawa

1.3

2

—

7.7

Ketokina

1.2

2

—

0

Ketoku

0.9

2

—

0

0.7

1

San

1.6

2

—

0

Todoroki

0.7

2

—

0

7.3 +

—

4.4 +

All farm Residential

Inokawa

1.7

1

2.8 +

—

2.8 +

All residential Stream

Ketokina

1.6

2

—

4.1

Ketoku

1.4

2

—

3.1

Matsubar

1.2

2

—

0

Oohara

0.9

1

1.7

2

5.7 + —

—

0 2.6+

All stream Totals

7.6

1.7

3.7

Notes: See text for computational methods. Farms are cultivated fields; streams are riparian zones in uncultivated areas; one residential area was sampled.

Population Density of Habu on the Amami Islands

233

Although we did not find a statistically significant difference in the snake den¬ sities in the two islands sampled (F = 1.27; df = 2, 34; P = 0.27), there was a significant overall decline in Habu densities during the study (F — 14.78; df= 2, 34; P = 0.0000). We do not have an explanation for this decline. We made a crude estimate of the total population of Habu in the Amami Islands in the period 1970-1973 by multiplying the average perception-biascorrected population density value (2.8/ha) for the habitats sampled in 1970-1973 by the availability multiplier (2.4) and extrapolating the result to the total land area of the Amami Islands (31,560 ha). The estimated total snake population was 214,000 (Tanaka, 1973). This value should not be used rigorously, as the sampling was neither randomized nor stratified, the multipliers have not been validated,

Table 16.2 Population density estimates for various sites on Amamioshima, 1970-1977. Area Habitat Farm

Habu/ha

Habu/ha

(1970-

(1972-

Habu/ha

Habitat

1973)

(1977)

mean

(ha)

Samples

1971)

Kaitsu A

1.2

4

6.1

—

—

Kaitsu B

1.1

4

19.6

—

—

1.1

—

6.1

—

—

—

6

—

—

3.1

—

2.6

—

Site

Katetsu A

1.3

2 2 2 2

Katetsu B

1.0

2

—

3.3

—

Kawauchi

0.7

2

—

4.6

—

Kuba A

2.0

—

8.4

—

Kuba B Kudadon A

1.5

—

6.5

—

0.7

2 2 2

—

—

Kudadon B

2.1

2

—

—

Nakakachi

1.9

2

—

Sokaru A

1.8

2

—

5.6

—

1.8

—

—

0

—

9.5

—

0.7

2 2 2

—

—

9.6

Uragami A

1.7

2

—

2

—

Uragami B

0.7

2

—

4.8

—

1.1 1.1

Sokaru B

0.7

0

0 0 —

5.1 +

All farm Residential

Ashikebu

4.5

1

2.1 +

—

—

2.1 +

All residential Stream

Adachi

1.9

Haazaki

0.8

Nishikomi

1.2

Sanbachi

1.4

2 4 2 2

Yadori

0.9

2

0

—

9

—

—

2.9

—

2.4

—

3.6 3.6

All stream Totals Note: See Table 16.1 for additional details.

9.2 +

4.9

2.8

4.1

234

Tanaka, Hayashi, & Wada

and the confidence limits on the total are unknown. However, it gives an order-of-magnitude estimate of the number of Habu that were subject to control measures.

SUMMARY Population estimates are notoriously difficult to obtain for snakes, but we found that removal methods could be used to estimate Habu population densities. The segregation of undersampling difficulties into perceptual bias and availability bias seems appropriate for many snake populations that include individuals that may be underground or otherwise inaccessible to searchers. Even at times ot maximal snake activity, only about 42% of the population was accessible to searchers. In addition, we found that only about 47% of the accessible Habu were detected by searchers combing a 1 ha area for one hour. These correction factors make it possible to estimate approximate Habu population densities from simple timeconstrained visual searches. Using these, we were unable to confirm the assertion of hunters that the Habu prefers riparian areas; it appears to be equally dense in cultivated cropland. The average Habu densities we calculated (2-5/ha) are similar to those reported for other large vipers.

LITERATURE CITED Bailey,

N. T. J. 1951. On estimating the size of mobile populations from recapture

data. Biometrika 38:293-306. Chapman, D. G. 1951. Some properties of the hypergeometric distribution with applications to zoological sample censuses. Univ. Calif. Publ. Stat. 1:131160. Hayashi,

Y.,

and

H.

Tanaka.

1979. On the methods for the population estimation

of Habu (Trimeresurus flavoviridis) [in Japanese with English summary]. Snake 11:72-77, 129. -. 1980. Accuracy of methods for population estimation of Habu analyzed in the open enclosure [in Japanese]. In Y. Sawai and H. Tanaka, eds., Report on the Studies to Control the Habu (Trimmeresurus flavoviridis) on the Amami Islands in 1980, pp. 79-84. Kagoshima, Japan: Kagoshima Prefecture. Paloheimo,

J. E. 1963. Estimation of catchabilities and population sizes of lobsters.

J. Fish. Res. Board Can. 20:59-88. Parker, W. S., and M. V. Plummer. 1987. Population ecology. In R. A. Siegel, J. T. Collins, and S. S. Novak, eds., Snakes: Ecology and Evolutionary Biology, pp. 253-301. New York: Macmillan. Tanaka, H. 1973. Activity and behavior of Habu, Trimeresurus flavoviridis [in Japanese with English summary]. Snake 5:116-132. Tanaka,

H., and Y. Wada. 1977. Venomous snakes. In M. Sasa, H. Takahashi, R. Kano,

and H. Tanaka, eds., Animals of Medical Importance in the Nansei Islands in Japan, pp. 29-71. Tokyo: Shinjuku Shobo.

Population Density of Habu on the Amami Islands Tanaka,

H.,

Y.

Wada, Y.

Matshushita.

1971.

A

Oguma,

M.

Sasa, Y.

Noboru,

T.

Ono,

and

235 N.

method of population estimation for the Habu, Trimere-

surus flavoviridis [in Japanese with English summary]. Snake 3:9-13. Zippin, C.

1958. The removal method of population estimation. J. Wildl. Manage.

22:82-90.

17 Population Trends and Limiting Factors in Boiga irregularis Gordon

H.

Rodda

Michael J. McCoid Thomas

H.

Fritts

Earl W. Campbell

III

A

n incipient animal population is likely to undergo an initial period of expo¬ nential growth, followed by a period with a less predictable outcome (Gray et

al, 1986; Groves and Burdon, 1986; Mooney and Drake, 1986; Drake et al., 1989; Hengeveld, 1989). If ecological feedback loops are short and the introduced species can coexist with the resident species, the invader’s numbers may approach carrying capacity slowly and then stabilize, with no radical changes in community structure. On the other hand, if feedback loops are long or coexistence is not possible, a period of chaotic changes in the community may ensue. Entire popu¬ lations may be extirpated, with each loss of a member species further disturbing the community’s equilibrium. In the case of the introduced Brown Treesnake (Boiga irregularis) on Guam, we are now witnessing a period of extreme insta¬ bility. Documentation of the community changes may help us understand the ecological forces that direct and constrain the responses of predator and prey. Understanding these forces is necessary for wise management, both on Guam and on any other islands the snake may reach. After the degree of the snake’s impact on Guam’s avifauna was recognized in the mid-1980s, systematic demographic studies of snake populations were initiated. The size structure of a sampled population on Guam suggested high recruitment and dense populations in recently colonized areas (Fritts and Scott, 1985). Throughout Guam, the Brown Treesnake population showed sexual size dimorphism, differences in population composition, and variation in densities between urban and rural sites and between northern and southern areas (Savidge, 1991). Body size distributions in Australia were similar in shape to those on Guam, but suggested smaller maximal sizes (Shine, 1991a). The evidence of how the snake arrived on Guam and some of the changes in population density through 1990 were summarized by Rodda et al. (1992). The population sampling tech¬ niques used and their limitations were reviewed by Rodda and Fritts (1992a). In this chapter we review the available data on the densities and composition of the Guam snake population, compare the vertebrate community on Guam with those on snake-free oceanic islands and six localities in the snake’s native range,

236

Population Trends in Boiga irregularis

237

and discuss population limiting factors that may guide control of extralimital populations.

MATERIALS AND METHODS The techniques used for the analysis of historical trends (before 1985) in Brown Treesnake communities are given in Rodda et al., 1992. Native birds are no longer an important food for Brown Treesnakes on Guam (Savidge, 1988). Thus, to analyze the recent community dynamics affecting the Brown Treesnake, we have concentrated on quantifying populations of the snake and its mammalian and reptilian predators, competitors, and currently available prey. We quantified nocturnal species by time-constrained (>1500 person-hours) or distanceconstrained (>600km) visual searches (Rodda and Fritts, 1992a). We augmented the visual counts with snap-trap sampling of rats (>1200 trap-nights). Diurnal lizards were sampled using adhesive traps (>31,000 trap-hours; Rodda et al., 1993). In addition to using visual surveys for relative counts of Brown Treesnakes, we estimated absolute population sizes from six mark-recapture studies (Rodda and Fritts, 1992a). Not all techniques were applied to all sites. We used one or more techniques on at least 61 sites on Guam, 67 sites on snake-free islands of the Marianas, and 35 sites in the Brown Treesnake’s native range (Papua New Guinea and the Solomon Islands). Snakes that were to be released were first held to a tape and stretched with light pressure until they could be measured when momentarily relaxed. They were sexed using a very thin sexing probe (see Jordan and Rodda, 1994, for validation of technique). Snakes not released were measured without stretching (relaxed after death by anesthesia), weighed, and then sexed. In paired measurements of 22 snakes measured both before and after death, the lengths taken from live specimens averaged 104.5% (SD = 1.4%) of the lengths taken from dead but unstretched snakes. The mathematical analysis of mark-recapture data is well developed (Otis et al., 1978; White et al., 1982; Pollock et al., 1990; Lebreton et al., 1992), but the available methods are ill-suited to the demographics of snakes (Parker and Plummer, 1987). Under most conditions, snakes tend to be rarely seen and difficult to capture, and capture numbers often exhibit high coefficients of vari¬ ability. In the following discussion, N represents population size and p expresses the percentage of the population caught on a given capture occasion. In our experience, snake populations that are compact enough for study generally have N < 50 and p < 0.10. White et al. (1982: 165), argued that “if N < 100 and p < 0.35, no capture-recapture experiments will provide unbiased and precise estimates.” An additional concern is that the sampled area is often contiguous with areas of nonsampled snakes (the population is open ). Because the movements of individual snakes are poorly understood, it is often difficult to discern the boundary between sampled and nonsampled areas.

238

Rodda, McCoid, Fritts, & Campbell Estimating snake densities is a two-sided problem. One must establish the

number of snakes in a population (abundance) as well as estimate the size of the area from which the sampled snakes were drawn (abundance/area = density). We arrived at two possible solutions to the problem of estimating abundance. The computer program

surge

(Lebreton et al., 1992) has excellent algorithms for

estimating p using open population models, although population size. If

surge

surge

does not estimate

fails to find temporal variability in p, it is probably

reasonable to assume that the population size in a study area is in dynamic equilibrium and that the average capture probability (p) is well estimated by surge.

For the same period of time, the mean number of captures, fi, is tabulated

(not estimated). N, the estimated total population size, is given by the relation¬ ship N = nip (Pollock et al., 1990). This approach produces results that are con¬ sistent with the trend of the population estimations of closed population models (program

capture,

White et al., 1982) over intervals of 20,15,10, and 5 days pro¬

jected to the X-intercept. The advantage of using surge is that the full amount of information in the 20+ day sample can be utilized, yielding greater precision. The problem of estimating the size of the area sampled is harder to solve. The capture

program has an algorithm for estimating the size of the area sampled,

but this closed-population model is inappropriate for the rapid population turnover of Brown Treesnakes, and the algorithm relies on small subsets of the traps for abundance estimates of subsets of the population. In our experience, the confidence intervals for these small subsets are too large for usable extrapolation with snakes. Instead, we estimated the sampling area by comparing the capture rates of central versus peripheral traps (Fig. 17.1). We assumed the average central trap captured snakes from its sector of the trapping grid, but the peri¬ pheral traps had more captures because their sample included areas beyond the nominal grid boundary (i.e., the outer area for which they were not competing with other traps for captures). If the trap spacing is 15m (as in Fig. 17.1), the average central trap samples 225 m2. One can estimate the size of the peripheral area sampled by multiplying the product of this number and the number of peripheral traps by the ratio of average captures for peripheral traps/average captures of central traps. An equivalent approach is to assume that all traps sample 225 m2 and reduce the estimated total population size by the fraction of captures represented by excess captures in the peripheral traps. In our studies, only a modest percentage (10-20%) of total captures were “excess” captures of the peripheral traps (Fig. 17.1).

RESULTS AND DISCUSSION Population Density The Brown Treesnake population on Guam originated in the immediate post-Wo rid War II period. Population growth and expansion were continuous but not necessarily steady from 1950 to 1980 (Rodda et al., 1992). When snake pop-

Population Trends in Boiga irregularis

239

Figure 17.1 Row and column totals for Orote Point trap grid, 1991. Dashed lines show means for interior traps and give a graphical indication of the excess

captures in the

peripheral traps. The procedure for computation of means for central versus peripheral traps is described in the text.

ulations were first quantified in 1985, an especially dense population of snakes at a site in northern Guam was estimated at 80-120 snakes/ha. This probably rep¬ resented the initial irruption peak for snakes at that northern site. Rodda et al. (1992) used

capture,

which assumes a closed population, to compute this den¬

sity, perhaps overstating the real population size. Using improved open popula¬ tion analysis we recomputed the estimations from the raw data and found that a modest downward revision is warranted. Although insufficient data exist to put firm confidence intervals on the original estimates, subsequent measurements of comparable areas suggest that 50—100 snakes/ha were probably present in selected localities at the peak of the initial irruption. This range is still high for large snakes away from water or dens (Fig. 17.2). With the large numbers of captures (145-345 in 1.5 months) made possible with mouse-attractant snake traps (Rodda et al., this volume, Chap. 20) and the improved computational methods described above, we estimate recent densities of snakes at Orote Point, Guam, at 49/ha (95% confidence limits 36-73) in 1990, and 37/ha (27-54) in 1991, and 24/ha (13-49) at a Northwest Field site in 1992. The Orote site is near the location of Guam’s initial colonization and has prob¬ ably had snakes for several decades. The lower density at the Northwest Field site is consistent with the declining population in northern Guam noted by Rodda et al. (1992). It may reflect a failure of the prey community to recover from the high snake population of 1985. The existence of a dense snake population at Orote, in an area long colonized, suggests that Brown Treesnakes will continue to occur, at least episodically, at high densities. In addition to these absolute population estimates, we have visual sighting lates for about 60 sites on Guam. These indicate that populations in many areas change dramatically from year to year, often with no obvious general trend. Adjacent areas sometimes follow contrary trajectories. Areas of the same forest type often do not exhibit similar sighting rates. These inconsistencies suggest that factors other than

240

Rodda, McCoid, Fritts, & Campbell 9

ou

Eryx tataricus Coluber constrictor Elaphe dione Elaphe obsoleta Elaphe vulpina

91 a: 1

|

co

I

co

Z)

Heterodon nasicus Heterodon platyrhinos

Lampropeltis caUigaster

—I

o o

Lampropeltis triangulum Masticophis taeniatus Pituophis melanoleucus Taphrometron lineolatum Boiga irregularis Agkistrodon con tort rix

[]

Agkistrodon halys Crotalus cerastes

co

^ U Crotalus horridus LU cl [] Crotalus viridis > 1 | Trimeresurus flavoviridis I 1 Vipera berus

Figure 17.2 Maximum reported densities of large snakes away from water or dens

Viperus ursinii “i-1

0

20

40

60

Snakes/ha

80

100

120

{non-Boiga values from Parker and Plummer, 1987).

habitat structure are important in determining the abundance of the snake. We are collecting data to test the hypothesis that prey abundances and an area’s history of predator-prey interactions are the key determinants of population density. The densities of snakes continue to be higher than the maximum reported densities for all avian prey species at their most favorable sites (26.1 birds/ha: Engbring and Ramsey, 1984). To date, the only directly observed predation on a wild native bird by a Brown Treesnake was a snake with a snout-vent length (SVL) of 785 mm consuming an egg of a Mariana Crow {Corvus kubaryi), Guam’s largest passerine. Thus, all but the youngest hatchling snakes should be capable of eating eggs of the smaller birds (Chiszar, 1990). When given the opportunity, Brown Treesnakes can sustain a dietary intake of at least 40% of their body mass per month (Collins and Rodda, 1994). Based on the weights of snakes captured in our surveys, a high-density Brown Treesnake population on Guam has an aggregate mass of 3-5 kg/ha. Thus the snakes in a hectare could potentially eat 1.2-2.0 kg prey/month, or 14-24 kg/year. Unfortunately, average weights are not available for many of Guam’s birds. Micronesian Honeyeaters {Myzomela saffordi) averaged

Population Trends in Boiga irregularis

241

13.3 g (Jenkins, 1983); at the other extreme are Guam Rails (Rallus owstoni), males of which averaged 241 g (Jenkins, 1979). We interpolated among these, guided by the birds’ known lengths (Jenkins, 1983) and the masses of North American birds of similar lengths (Dunning, 1984). The estimated mass of each species multiplied by the number of each species at the maximum reported density yields a total bird mass of around 0.8 kg/ha. Therefore, a dense population of Brown Treesnakes on Guam has the capacity to consume annually about 18—30 times the biomass of adult birds that were present under the most favorable conditions. /

Size Structure Savidge (1991) showed that large Brown Treesnakes on Guam in the early 1980s were more conspicuous in urban areas and in southern Guam. Southern Guam has extensive savanna habitat, and rats are relatively abundant there (Barbehenn, 1974; Savidge, 1986). Urban areas tend to have more commensal endotherms, especially rats and chickens, and snakes are more likely to be found in association with these. Compared with forests, where the native endotherms have been extir¬ pated, savannas and urban areas have a relatively rich food base lor large snakes. Thus it is not surprising that large snakes are more common in these areas (Fig. 17.3; Savidge, 1991). Large snakes are overwhelmingly male (Fig. 17.3; McCoid, 1990; Savidge, 1991). The evolutionary and proximate causes of the large sexual size dimorphism pre¬ sent in the Guam population are not evident. Sexual size dimorphism may occur because one sex terminates growth earlier than the other, one sex grows more rapidly than the other, or one sex enjoys lower mortality than the other. We have no evidence with regard to the duration of growth in Brown Treesnakes. Rodda et al. (this volume, Chap. 2) cited evidence for a less variable but not higher growth rate in maturing males. Some evidence for greater female mortality comes from comparing the size distributions of captive and wild-caught snakes (Fig. 17.3). The captive males exhibit sizes in the range of wild conspecifics, whereas the largest captive females greatly exceed the size of the largest wild females. Captive females differ from their wild counterparts in having ample food but, in this case, no opportunity to breed. We have no evidence to refute the possibility that the denial of breeding opportunities redirects growth into somatic channels in captive females only, but breeding activities may also result in highei energy stiess and therefore greater mortality in the wild females. Some support for this position comes from the work of Jordan and Rodda (1994), who found that recent samples of adults from Guam were more male biased in sex ratio than earlier samples and that adult females had sharply deteriorated in their mass-to-length ratio, suggest¬ ing physiological stress. The extirpation of native endotherms, the primary food of adult Brown Treesnakes (Greene, 1989), may have produced a lood shortage more acutely felt by reproductive females. The Brown Treesnakes Nichols (in litt., 1992) examined on Guam were unusually low in fat reserves, suggesting food stress.

Rodda, McCoid, Fritts, & Campbell

242

Savanna/urban

20

□ captive □ wild

>»10

O

0

o

cr 0 10

20

I

I

I

I

500

1000

1500

2000

Figure 17.3 Size histogram of snakes from Guam and captivity (captive values from D. Chiszar, pers. comm., 1993). Savanna and urban areas are grouped because they have a relatively rich supply of endotherm

Snout-Vent Length (mm)

prey, especially rats and domestic fowl.

Sexual size dimorphism is not unique to Brown Treesnakes on Guam, how¬ ever. We found it to be present in specimens from Papua New Guinea and the Solomon Islands, and Shine (1991a) recorded it for Australia. A leading evolu¬ tionary argument for male-biased sexual size dimorphism in snakes is that large males enjoy a mating advantage (Shine, 1978). This would occur if females pre¬ ferred to mate with larger males, or if larger males were better able to exclude smaller males from breeding opportunities. The latter is believed to occur in a variety of venomous snakes, and in those species “combat dances” between males are observed relatively often. We have observed several thousand Brown Treesnakes, but we have not seen behavior suggestive of “combat dances.” Nor has this behavior been reported in other parts of the range. A wildlife conservation officer on Guam reported a brief encounter between two male snakes that involved the larger snake pushing downward on the smaller snake (McCoid, pers. comm., 1989), but this observation is more notable for its uniqueness than for its conclusiveness. We have found no evidence for intermale avoidance in small

Population Trends in Boiga irregularis

243

aggregations of snakes found in traps or natural crevices (Rodda et al., this volume, Chap. 2). The other leading hypothesis for the evolution of sexual size dimorphism is that of feeding niche specialization between the sexes (Camilleri and Shine, 1990). Dietary studies in Australia (Shine, 1991a), Guam (Savidge, 1988), and nonAustralia parts of the native range (Greene, 1989) emphasize the breadth rather than the specialization of Boiga diets. Shine (1991a) explicitly rejected a sexual dif¬ ference in diet. If the fairly extreme sexual size dimorphism of Brown Treesnakes has an evolutionary basis, the selective forces responsible have yet to be identified.

Sex Ratio Skewed sex ratios may be the result of differential mortality or differential catchability. Collections of juvenile Brown Treesnakes rarely differ significantly from a 1:1 ratio, but samples of adults are usually male biased in Guam (Savidge, 1988) and Australia (Shine, 1991a). This could be due to greater male catchability, greater male survivorship, different growth trajectories, or a combination of these. As Rodda et al. noted in Chapter 2 of this volume, gravid females are rarely col¬ lected. We have too few data to determine if the paucity of gravid females is responsible for all of the difference between the sexes in capture frequencies. It could account for some of the sex ratio differences between sites (Fig. 17.4), but a small shift among areas in the fraction of adult females that are gravid would not likely produce the radical changes in sex ratio among sites shown in Figure 17.4. Note that the sex ratio differences among sites are more pronounced than are those among habitats (Fig. 17.4). When sites are pooled (e.g., Andersen Air Force Base, North Guam, and habitat comparisons in Fig. 17.4), sex ratio differ¬ ences tend to be muted and stable over time. The significant differences tend to occur in single sites rather than broad areas. Savidge (1991) suggested greater mo¬ bility of males to explain a preponderance of males in an area more recently col¬ onized. Movements might also account for some ot the other local sex ratio dif¬ ferences, but direct documentation and evidence for the proximate causes of the movements are lacking.

Mature Fraction The average size at maturity appears to vary among different parts of the Brown Treesnakes range (Greene, 1989; Shine, 1991a; Rodda et al., this volume, Chap. 2). We based our estimate of mature fraction on size distributions relative to the applicable local criterion for size at maturity (Rodda et al., this volume, Chap. 2). Mature snakes are relatively uncommon on Guam (Fig. 17.5). The mature frac¬ tion is low not only in contrast to samples from the native range (Fig. 17.5), but also in relation to other nonvenomous snakes (Fig. 17.6). The contrast is sharper

244

Rodda, McCoid, Fritts, & Campbell ii Females □ Males

Percentage of Sample

100

*

Spatial variation

80-

60-

40-

20

-

0-

Harmon Naval Air Station Andersen AFB NCTAMS

Tarague

NW Field Yigo

Percentage of Sample

100 i

H Females □ Males

Temporal variation

80

60

40-

20

-

0

1990

1991

< 1988

1992

1989-1990

North Guam

Orote Point

Percentage of Sample

100

H Females □ Males

Habitats

80-

60-

40-

20

-

0-*

Trees

Ground

1990-1991

Nonurban

Urban

1980-1985

Nonurban

Urban

1988-1990

Figure 17.4 Sex ratios from a variety of sites in Guam. Asterisks indicate samples whose counts deviate significantly from 1:1 (a < 0.05; G test with Williams correction). (North Guam data are from Jordan and Rodda, 1994; 1980-1985 data are from Savidge, 1991.)

Population Trends in Boiga irregularis

245

70 -I

Native Range

Guam

Figure 17.5 Percentage of mature Brown Treesnakes in samples collected from Guam and the native range. (Australia data from Shine, 1991; non-Australia data from Greene, 1989, U.S. National Museum [USNM] data are mostly from the islands north of Papua New Guinea [Admiralty and Bismarck Archipelagos].)

than that suggested by Figure 17.6, however, as low mature fractions are associ¬ ated with high-fecundity species (Fig. 17.7), and B. irregularis, is a species with low fecundity. The anomalous position of current Brown Treesnake populations on Guam may be due to an unusually large number of small snakes, a dearth of large snakes, or both. In the following section we present some data suggesting that food for juvenile snakes is more abundant on Guam than it is in the native range. The greater abundance of food might enhance juvenile survivorship and produce relatively high numbers of juveniles on Guam. In contrast, the recent extirpation of most native endotherms appears to have reduced the prey supply for large snakes on Guam, perhaps increasing adult mortality and reducing the abundance of mature snakes.

Limiting Factors It is an article of faith among many policy makers that the high abundance of Brown Treesnakes on Guam is due to the absence of some predator that suppresses populations of the snake in its native range. Although Boiga has competitors throughout its native range, we have been unable to identify any predators in the

Percentage Mature Figure 17.6 Percentages of mature snakes in samples of nonvenomous snakes reported by Parker and Plummer (1987). Each species is represented once, by the average of the avail¬ able data. The arrows show values for single venues of B. irregularis.

e—

♦

•

Figure 17.7 Maturity in relation to fecundity. (Clutch or litter size averages from Seigel and Ford, 1987.)

Population Trends in Boiga irregularis

247

eastern end of its range. On Malaita, Solomon Islands, for example, there are no carnivorous mammals, no snake-eating birds, and, with one exception, no reptiles that eat Brown Treesnake adults or eggs. The one exception, Varanus indicus, is also present on Guam, where it apparently does not hold down snake numbers. Humans on Malaita do not eat Brown Treesnakes. The vegetation on Malaita (a mix of gardens and old- and second-growth forests) is similar to that on Guam and seems ideal for the snake. Yet, our average sighting rate for Brown Treesnakes on Malaita was 0.10/h, about 5% of the average for the same searchers on Guam. From this and other examples we conclude that neither predation nor habitat structure is the primary factor limiting populations of Brown Treesnakes in many areas. Shine (1991b) reached a similar conclusion for Australian elapids. On one occasion we saw a Brown Treesnake attempt to eat a conspecific. In this case the snake being eaten was larger; consummation of the feeding attempt was interrupted and may not have been possible. There are frequent opportuni¬ ties for Brown Treesnakes to eat smaller conspecifics, and the absence of other reports (through direct observation or analysis of the contents of the 1500+ stom¬ achs that have been examined) suggests that cannibalism is not an important eco¬ logical pressure on Brown Treesnake populations. Are competitors limiting the snake? Competition may be either direct (inter¬ ference competition) or indirect (depletion of shared resources). We know of no evidence to suggest that any organism directly interferes with foraging Brown Treesnakes, but there is evidence that prey availability is lower in the native range than it is on Guam or nearby snake-free islands (Fig. 17.8). This is consistent with a role for either competition or simple food shortage (i.e., not attributable to com¬ petitors). Figure 17.8 shows the abundances of the primary foods for small and medium-sized Boiga. Our data on endothermic prey were too sparse for a com¬ parable analysis of food for large snakes. With larger prey species, the vulnera¬ bility of the prey to the snake is probably as important as the number of prey pre¬ sent. In the absence of coevolutionary experience with snakes, the birds of Guam were probably highly vulnerable to the snake (Savidge, 1986). Adult endotherms in the native range do not appear to be nearly as vulnerable. We constructed a stepwise multiple regression model to explore the indepen¬ dent contributions of the abundances of frogs, skinks, geckos, and snakes other than Boiga to the abundances of Boiga that we documented in 11 sites (6 native range, 5 Guam). Day-active lizard abundances were expressed in captures per trap-day (adhesive traps). The other rates were sightings per hour. All variables were natural log(n +1) transformed. Using several different selection criteria, the best model was one that omitted frog abundances but included the other vari¬ ables (Table 17.1). The best model explained 83% of the variance, indicating that any omitted variables were relatively unimportant (R — 0.91; F — 24.27; df — 3, 7; p = 0.0004). Note that a negative parameter value is associated with abundances of non-Boiga snakes; this is consistent with indirect competition for food (prob¬ ably reflecting the endothermic prey we were unable to quantify directly). In the

248

Rodda, McCoid, Fritts, & Campbell 20

-|

Nocturnal lizards

Native Range

Marianas

Figure 17.8 Relative abundances of diurnal (mostly skinks) and nocturnal (mostly geckos) lizards at selected sites in Papua New Guinea (PNG), Solomon Islands, and the Mariana Islands. Islands lacking snakes are represented with dark bars.

Population Trends in Boiga irregularis

249

Table 17.1 Multiple regression model for the dependent variable Boiga sightings per hour. Predictor variable Skinks Other snakes Geckos

Parameter

Standard

Type II sum

estimate

error

of squares

F

0.1996

0.078

0.2373

6.69

0.036

-1.7264

0.676

0.2316

6.53

0.038

0.5457

0.147

0.4827

13.61

Probability

0.0078

absence of information on nonfood variables, the model suggests that prey abun¬ dance is the most important ecological variable limiting the abundance of the Brown Treesnake. Using the multiple regression equation, we estimated suitabilities for Brown Treesnakes of various sites in the native range and Marianas (Fig. 17.9). The com¬ posite environmental suitability index indicates that Guam and Mariana Island sites have a high suitability for the Brown Treesnake. The success of the Brown Treesnake on Guam probably derives from the abundance of food there. Based on this estimation of environmental suitability, we predict the snake will do well on Saipan and other presently snake-free islands of the Marianas if it should become established there. The data that we used to quantify the abundances of snake food on Guam (for Fig. 17.9) were collected in 1988-1990, after the demise of most of the native forest vertebrates. However, Guam sites broadly overlap the suitability ranges of the presently snake-free sites of the Marianas, where the native vertebrates are rel¬ atively intact. This persisting high suitability of Guam sites is attributable to the success on Guam of introduced prey species, especially the house gecko Hemidactylus frenatus and the terrestrial skink Carlia cf. fusca. Other important food items are introduced birds, especially chickens (Gallus gallus), francolins (Francolinus francolinus), drongos (Dicrurus macrocercus), sparrows (Passer montanus), pigeons (Columba livid), turtle doves (Streptopelia bitorcpiata), rats (Rattus tanezumi and R. norvegicus), and native lizards (Emoia caeruleocauda-, Lepidodactylus lugubris, and Gehyra mutilata). The native lizards that have survived are very small and have high reproductive rates (McCoid, 1989; Rodda and Fritts, 1992b). The introduction and high populations of rats on Guam before the arrival of Boiga and the irruptions of shrews (Suncus murinus) after 1952 and skinks (Carlia cf. fusca) after 1960 undoubtedly accelerated the snakes population expansion. The succession of these and other introductions must be considered in evaluating the snake populations that developed on Guam. Were it not for the highly successful introduced prey species, Guam would probably not now have a dense population of Brown Treesnakes (McCoid, this volume, Chap. 37).

250

Rodda, McCoid, Fritts, & Campbell

Wau, PNG Madang, PNG

Native Range

Karkar I. Manus I. Guadal. I. Malaita I.

Guam NavCAMS Orote 1988 Orote 1990 NWF Road Potts Jet.

Marianas

Environmental Suitability Index Figure 17.9 Environmental suitability of various sites in Papua New Guinea, the Solomon Islands, and the Mariana Islands. The index is based on the factors and weightings obtained from the multiple regression model in Table 17.1. Islands lacking snakes are represented with dark bars.

Population Trends in Boiga irregularis

251

MANAGEMENT RECOMMENDATIONS In the early 1990s, Brown Treesnakes continued to attain high densities in areas long colonized on Guam. Snakes will probably continue to exert significant pre¬ dation pressure in some areas after the initial irruption has passed. The adverse impacts of high snake densities (Rodda et al., this volume, Chap. 2; Fritts and Chiszar, this volume, Chap. 4; Fritts and McCoid, this volume, Chap. 6) and frequent opportunities for accidental transport to other islands (Fritts et al., this volume, Chap. 14) are likely to continue. The changing sex ratios, mature fraction, and size distributions indicate a pop¬ ulation that is experiencing unusual local population changes. These aspects should be explored further for the management opportunities they may present. Adult females seem to be especially vulnerable to stress. If confirmed, and if augmentation of the stress were practical, it might be possible to bring about significant reductions in the number of reproductive females or their reproduc¬ tive output. Basic information is lacking regarding snake age, conditions necessary for reproduction, location of eggs, average clutch size on Guam, frequency of repro¬ duction, and survivorship schedules. Documentation of these demographic parameters is a necessary precursor of intelligent management planning. The introduction of snake predators will not replicate the limiting factors that constrain Brown Treesnake populations in the snake’s native range. Other rationales for biological control agents should be evaluated in light of other island experiences (Fiowarth, this volume, Chap. 32). Care should be taken to ensure that measures taken to make an environment inhospitable to Brown Treesnakes do not inadvertently create conditions fa¬ vorable to prey species. For example, the use of bright lights to repel the nocturnal snake could incidentally increase the abundance of geckos (Petren et al., 1993), thereby indirectly increasing the attractiveness of the site for foraging snakes. Control of the introduced species that are prey for the Brown Treesnake (pigeons, sparrows, rats, Carlia cf. fusca) should be considered in sites where it is practical (ports, airports, homes). Reducing prey abundance would be likely to limit snake numbers, although more study is needed to determine what size prey-reduced zone would be needed to redirect a significant number of Brown Treesnakes to other areas. Study of the unintended consequences of such manipulations may also be warranted.

ACKNOWLEDGMENTS Much of our work was made possible by funding provided by the U.S. Depart¬ ment of Interior’s Office of Territorial and International Affairs. We thank the U.S. Department of Defense for financial support and access to base forests, our many

252

Rodda, McCoid, Fritts, & Campbell

friends for assistance with surveys and trap monitoring; and R. Rondeau, S. Corn, and J. Oldemeyer for suggesting improvements to the manuscript.

LITERATURE CITED Barbehenn, K. R. 1974. Estimating density and home range size with removal grids:

The rodents and shrews of Guam. Acta Theriol. 19:191-234. Camilleri, C., and R. Shine. 1990. Sexual dimorphism and dietary divergence:

Differences in trophic morphology between male and female snakes. Copeia 1990:649-658. Chiszar, D. 1990. Behavior of the Brown Tree Snake, Boiga irregularis: A study in

applied comparative psychology. In D. Dewsbury, ed., Contemporary Issues in Comparative Psychology, pp. 101-123. Sunderland, Mass.: Sinauer. Collins, E. P., and G. H. Rodda. 1994. Bone layers associated with ecdysis in

laboratory-reared Boiga irregularis (Colubridae). J. Herpetol. 28:378-381. Drake, J. A., H. A. Mooney, F. Di Castri, R. H. Groves, F. J. Kruger, M. Rejmanek, and M. Williamson. 1989. Biological Invasions: A Global Perspective. Chichester,

England: John Wiley 8c Sons. Dunning, J. B. Jr. 1984. Body Weights of 686 Species of North American Birds. West.

Bird Banding Assoc. Monogr. No. 1. Cave Creek, Ariz. Engbring, J., and F. L. Ramsey. 1984. Distribution and Abundance of the Forest

Birds of Guam: Results of a 1981 Survey. U.S. Fish Wildl. Serv., FWS/OBS-84/20. Fritts, T. H., and N. J. Scott Jr. 1985. The Brown Tree Snake on Guam: Studies of

its ecology, control, and threats to other islands. Report to U.S. Fish and Wildlife Service, Region 1, Portland, Ore. Gray, A. J., M. J. Crawley, and P. J. Edwards. 1986. Colonization, Succession and

Stability. Oxford: Blackwell. Greene, H. W. 1989. Ecological, evolutionary, and conservation implications of

feeding biology in Old World cat snakes, genus Boiga (Colubridae). Proc. Calif. Acad. Sci. 46:193-207. Groves, R. H., and J. J. Burdon. 1986. Ecology of Biological Invasions. Cambridge:

Cambridge Univ. Press. Hengeveld, R. 1989. Dynamics of Biological Invasions. London: Chapman 8c Hall. Jenkins, J. M. 1979. Natural history of the Guam Rail. Condor 81:404-408.

-. 1983. The Native Forest Birds of Guam. AOU Ornithol. Monogr. 31. Jordan, M. A., and G. H. Rodda. 1994. Identification of sex in Boiga irregularis:

Implications for population dynamics in Guam. J. Herpetol. 28:381-384. Lebreton, J.-D., K. P. Burnham, J. Clobert, and D. R. Anderson. 1992. Modeling

survival and testing biological hypotheses using marked animals: A unified approach with case studies. Ecol. Monogr. 62:67-118. McCoid, M. J. 1989. Job progress report, research project segment, biology of the

Brown Tree Snake. In C. F. Aguon, G. J. Wiles, and L. L. Mariano, eds., Annual Report Fiscal Year 1989, pp. 157-201. Agana, Guam: Guam Division of Aquatic and Wildlife Resources.

Population Trends in Boiga irregularis

253

_. 1990. Job progress report, research project segment. In R. B. Anderson, G. J. Wiles, and L. L. Mariano, eds., 1990 Annual Report, pp. 178-199. Agana, Guam: Guam Division of Aquatic and Wildlife Resources. Mooney, H. A., and J. A. Drake. 1986. Ecology of Biological Invasions of North America and Hawaii. New York: Springer-Verlag. Otis, D. L., K. P. Burnham, G. C. White, and D. R.

Anderson.

1978. Statistical

inference from capture data on closed animal populations. Wildl. Monogr. 62:1-135. Parker, W. S.,

and

M. V.

Plummer.

1987. Population ecology. In R. A. Siegel,

J. T. Collins, and S. S. Novak, eds., Snakes: Ecology and Evolutionary Biology, pp. 253-301. New York: Macmillan. Petren, K., D. T. Bolger, and T. J.

Case.

1993. Mechanisms in the competitive

success of an invading sexual gecko over an asexual native. Science 259.354—358. K. H., J. D.

Pollock,

Nichols,

C.

Brownie, and

J. E.

Hines.

1990. Statistical

inference for capture-recapture experiments. Wildl. Monogr. 107:1-97. Rodda, G.

H.,

and T.

H.

Fritts.

1992a. Sampling techniques for an arboreal snake,

Boiga irregularis. Micronesica 25:23-40. _1992b. The impact of the introduction of the Brown Tree Snake, Boiga irregu¬ laris:, on Guam’s lizards. J. Herpetol. 26:166-174. Rodda, G. H., T. H. Fritts, and P. J. Conry. 1992. Origin and population growth of the Brown Tree Snake, Boiga irregularis, on Guam. Pac. Sci. 46.46—57. Rodda,

G. H., M. J.

McCoid, and

T. H.

Fritts.

1993. Adhesive trapping II. Herpetol.

Rev. 24:99-100. Savidge, J. A. 1986. The role of disease and predation in the decline of Guam’s avifauna. Ph.D. disst., Univ. Illinois, Urbana-Champaign. _. 1988. Food habits of Boiga irregularis, an introduced predator on Guam. J. Herpetol. 22:275—282. __ 1991. Population characteristics of the introduced Brown Tree Snake (Boiga irregularis) on Guam. Biotropica 23:294—300. Seigel, R. A., and N. B. Ford. 1987. Reproductive ecology. In R. A. Seigel, J. T. Collins, and S. S. Novak, eds., Snakes: Ecology and Evolutionary Biology, pp. 210-252. New York: Macmillan. Shine, R. 1978. Growth rates and sexual maturation in six species of Australian elapid snakes. Herpetologica 34:73-79. _# 1991a. Strangers in a strange land: Ecology of Australian colubrid snakes. Copeia 1991:120-131. _. 1991b. Australian Snakes: A Natural History. Ithaca: Cornell Univ. Press. White,

G. C., D. R.

Anderson,

K. P.

Burnham, and

D. L.

Otis.

1982. Capture-

Recapture and Removal Methods for Sampling Closed Populations. Los Alamos, N.M.: Los Alamos National Laboratory.

Part V

CAPTURE AND DETECTION

2/*> Perhaps the greatest contribution of applied herpetologists has been in the area of techniques. Whereas academic biologists tend to use the best available technique, wildlife managers often must develop new techniques to accomplish their goals. In this section we explore the refinement of techniques for trapping snakes, detecting snakes with dogs, and excluding snakes with barriers. The methodological improvements outlined here will no doubt be of value to academic herpetologists who need to sample populations efficiently and with minimal bias; however, the thrust of the present work is to make possible the resolution of conflicts between snakes, native species, and humans in high priority areas. Four chapters of this section (Chaps. 18-19, 24-25) describe projects undertaken to trap and remove Habu from circumscribed localities in Japan. Habu researchers have the world’s bestdeveloped program for the scientific testing of snake traps, but many of their results have been unavailable in English until now. Rodda, Fritts, Clark, Gotte, and Chiszar (Chap. 20) describe devel¬ opment of traps for the Brown Treesnake. Whereas snake traps have traditionally been thought of as a means to snag the occasional snake, the latest techniques suggest that significant fractions of a snake population can now be trapped on demand. Chapters 21-23 discuss the use of barrier fences. Nishimura (Chap. 22) and Hayashi, Sawai, Tanaka, and Mishima (Chap. 23) describe electrical and mechanical barriers developed for excluding Habu from villages and agricultural fields; Campbell (Chap. 21) de¬ scribes electromechanical barriers used for excluding Brown Treesnakes from transportation sites and wildlife conservation areas. The last two chapters describe the use of trained dogs for detect¬ ing Habu (Shiroma and Ukuta, Chap. 26) and Brown Treesnakes (Imamura, Chap. 27). Although herpetologists in the past rarely used dogs to obtain specimens (cf. Dugan, 1982), experiences with these snake detector dogs show that their extraordinary olfactory sensitivity can provide a perceptual window that no human or ma¬ chine can match. These techniques, pioneered by these authors, hold great promise not only for management of Habu and Brown Treesnakes, but also for application to smaller-scale problems caused by other species

255

256

Port V such as ratsnakes and rattlesnakes around electrical substations and research installations in continental areas.

LITERATURE CITED Dugan, B. A. 1982. The mating behavior of the Green Iguana, Iguana

iguana. In G. M. Burghardt, and A. S. Rand, eds., Iguanas of the World: Their Behavior, Ecology, and Conservation, pp. 320-341. Park Ridge, N.J.: Noyes.

18 Development of the Box Trap for Habu Shosaku Hattori

T

he traditional Japanese rat trap is a live trap with a conical entrance made of flexible metal rods. Forty years ago it was observed that Habu (Trimeresurus

flavoviridis) occasionally entered, and were themselves captured, in rat traps that had captured live rats. This discovery inspired the use of snake traps with flexible cone entrances that used live rats as attractants. In this chapter I summarize developments of the Habu box trap with regard to material of the trap body, entrance type, attractant species, and placement of traps.

MATERIAL OF THE TRAP BODY The first tests of materials used to make Habu traps were conducted by Kihara et al. (1978). They compared traps made of wood (type A), metal mesh (type B), and plastic pipe (type C) in field tests in the Amami Islands of Japan (Fig. 18.1). Type A.

The wood traps had dimensions of 400 (L) X 300 (W) X 200 (H) mm.

Each end of the trap had a 75-mm-diameter conical entrance made of thin bamboo rods pointing inward. A Habu could easily spread the converging bam¬ boo to enter, but it was deflected by the closed apex of the cone if it attempted to exit the box. Within the box was a rodent housed in a wire mesh cage. Food and water for the rodent were replenished weekly. Type B.

The metal mesh traps were similar to the wood traps but had a body of

8-mm-diameter open metal mesh, and a single entrance. Snakes over one year old could not escape from the wire mesh traps. Type C. The third design was a modification of the Japanese eel trap. The main body of the trap was a polyvinyl chloride (PVC) pipe (55mm inside diameter) 600 mm long with a conical entrance at one end. At the opposite end of the pipe was a rodent chamber set in a T pipe fitting (Fig. 18.1). Traps of all three designs were placed in habitats where Habu densities were high—fields of sugarcane, rice, or vegetable crops. The wood traps used mice (Mus musculus) or wild rats (Rattus tanezumi) as attractants; the eel traps were supplied with mice; and the mesh traps were supplied with wild rats, laboratory mice, or laboratory rats (Rattus norvegicus). During these trials (Table 18.1), three species

257

258

Hattori

Figure 18.1 Three types of Habu traps tested for effectiveness, each made of different materials: type A, wooden board; type B, wire mesh; type C, polyvinyl chloride (PVC) pipe (modified version of Japanese eel trap).

of snake were captured: the Habu, Hime-habu (Trimeresurus (= Ovophis) okinavensis), and Akamata (Dinodon semicarinatus). The wood traps supplied with wild rat attractants captured all but one of the Habu (mean capture success = 1.3%); the one exception was a Habu captured in a mesh trap supplied with a mouse (mean capture success = 0.07%). Maintenance of the rodents was easiest in the wood traps; rats in mesh traps died of exposure to cold when rain was blown through the mesh, and a buildup of heat killed many mice in the pipe traps.

Development of the Box Trap for Habu

259

Table 18.1 Number of snakes captured in three different types of traps, June-November 1977. No. snakes captured Period

Trap type

Attractant

T.f

T.o.

D.s.

75

0

0

0

Trap-days

Mesh

9-13 June

rat

Mesh

5 July-28 Sept.

mouse

1204

0

0

0

Pipe

5 July-28 Sept.

mouse

994

0

0

0

Wood

25 July-26 Sept.

mouse

640

0

0

0

Wood

1 Oct.-19 Nov.

wild rat

376

5

1

0

Wood

1 Oct-19 Nov.

mouse

376

0

0

3

Mesh

1 Oct.-19 Nov.

wild rat

190

0

0

0

Mesh

1 Oct-19 Nov.

mouse

190

1

0

0

Source: Kihara et al., 1978. Note: T.fi = Trimeresurus flavoviridis, T.o. — T. okinavensis, D.s. = Dinidon semicarinatus.

Cone Figure 18.2 Structures of two one-way entrances.

Most of the traps subsequently used in the Amami Islands have had trap bodies of either solid wood or solid polypropylene board.

TRAP ENTRANCES Many types of trap entrances were tested on captive snakes (Hayashi et al., 1979, 1984). Two types of entrances (on box traps) were tested in the field, a cone and a flap (Fig. 18.2). The cone entrance was a flexible cone of thin, inward-pointing bamboo rods. The base of the cone was 75 mm in diameter, and the length of cone was 150 mm. The flap entrance was a one-way swinging door, hinged at the top to permit ingress but not egress. The trap door was made of sheet metal (12 cm X 12 cm, 40g). In the laboratory trial (conducted in a 7X13m room), traps with flap entrances did capture Habu. In field trials, however, no snakes were captured in traps with the flap entrances (Table 18.2). Therefore, we now use traps with two conical entrances (polyethylene has been substituted for bamboo in the entrance cones). The trap bodies are made with polypropylene boards instead of wood, and the underside of the trap body is cut away in the middle to allow easy insertion and removal of the attractant cage (Fig. 18.3).

260

Hattori

Table 18.2 Number of Habu captured using different entrance types on traps, May-June 1978. No. Habu Trap type

Rate

Period_Attractant_Trap-days_captured_(%)

Cone

8 May-13 June

rat

922

8

0.87

Flap

8 May-13 June

rat

925

0

0

Cone

8 May-13 June

mouse

799

4

0.5

Flap

8 May-13 June

mouse

757

0

0

Source: Hayashi et al., 1979.

Figure 18.3 The box trap used to capture Habu.

ATTRACTANT SPECIES Live rodents were superior to chemical attractants for attracting Habu to traps. Our tests of live rodents used the three species obtained most easily: wild rats, laboratory rats, and laboratory mice (Table 18.3). There were two paired tests for comparing attractants: test A, laboratory rats versus wild rats, and test B, labora¬ tory rats versus laboratory mice. In test A, using wild rats as attractants resulted in significantly more snake captures (Table 18.3); the capture rate (1.29%) was six times greater than when laboratory rats were used (0.20%). In test B, however, laboratory rats yielded a higher capture rate (0.69%) and surpassed that of labo¬ ratory mice (0.13%). The two tests of laboratory rat traps occurred at the same site and time; there¬ fore, the capture rate difference between trials was probably not due to differences in snake density. The capture rate difference between trials may indicate that the traps were close enough together that there was “competition” between neigh¬ boring traps. Once captured, a snake is unavailable for capture by nearby traps. Suppose that a snakes home range includes one relatively attractive trap and one less attractive trap. The snake is more likely to be captured by the better trap and therefore will be unavailable for capture by the poorer trap. In this case the capture rate of the poorer trap will be lower than it would have been if the poor trap had been competing with a yet poorer trap, or if the traps had been far enough apart that the traps were not competing at all (Hayashi et al., 1979). Accordingly, the capture rate of laboratory rat traps in test A shown in Table 18.3

Development of the Box Trap for Habu

261

Table 18.3 Number of Habu captured in traps with rodent attractants, May-June 1978. Test A B

Period 14 June-5 Sept. 8 May-5 Sept.

Attractant

Trap-

No. Habu

Rate

days

captured

(%)

rat

2446

5

0.20

wild rat

2402

31

1.29

rat

2455

17

0.69

mouse

2284

3

0.13

Probability

0.0010 0.0204

Source: Hayashi et al., 1979. Notes: Rat = laboratory rat, wild rat = wild black rat, mouse = laboratory mouse. Fisher s exact method was used to calculate probabilities.

may have been depressed as a result of being paired with the more effective wild rat traps. Few immature snakes were captured in traps using rodent attractants (Hayashi et al., 1979). Most of the captured snakes were adults greater than 1100 mm snoutvent length, snakes estimated to be at least three years old. Immature Habu are relatively numerous, but they were not captured in our box traps. Three hypotheses might account for this trap failure. (1) small Habu have such a small home range that they rarely encounter a trap, (2) small Habu are able to escape from the traps, and (3) small Habu are not attracted to the relatively large prey that were used in the traps. Mishima (1966) reported that the chief prey of adult Habu was wild rats (82%), while Habu of less than 1000 mm SVL ate predomi¬ nantly lizards (49%), as well as Watases Shrew (Crocidura horsfieldii, (28%) and frogs (21%). Some Habu, especially juveniles, climb trees (Koba, 1971; Tanaka, 1973), and Habu are believed to eat sleeping birds on Tokunoshima. For these reasons, we attempted to expand the range of snakes caught by using alternate attractants, in this case quail (Coturnix) and shrews (Crocidura horsfieldii). The shrews were not successful, although they were tested for fewer trap-days than were the other attractants (Table 18.4). The capture success rate with quail was numerically lower than with laboratory rats, but the difference was not significant (Table 18.4). All the snakes caught with rat and quail attractants were adults (300-700 g body weight); thus the challenge of trapping juvenile Habu has not been met. Wild rats have consistently been the best species attracting Habu into traps.

TRAP PLACEMENT Box traps were set at intervals greater than 30 m in a small grove ol trees sur¬ rounded by a sugarcane field and in the woods or brush surrounding the sugar¬ cane field. The grove proved to be the best location for capture. Our Habu trap was used on a large scale (27.5 ha) for controlling snakes around the village of Tete on Tokunoshima (Table 18.5). Traps were set both inside and

262

Hattori

Table 18.4 Number of Habu captured using birds (quail, Coturnix coturnix) and insectivores (Watase's Shrew, Crocidura horsfieldii) as attractants.

9 May-22 Aug.

Rate (%)

quail

2100

4

0.19

laboratory rat

2100

10

0.48

252

0

0.00

1260

4

0.32

shrew

11 Sept.-23 Oct.

No. Habu captured

Trap-days

Attractant

Period

laboratory rat Source: Hayashi et al., 1984.

Table 18.5 Number of Habu captured in box traps within and outside the snake barrier fence that surrounds Tete village on Tokunoshima, May-October 1980. Location (ha)

Trap-days

No. Habu captured

Rate (%)

13 May-19 Oct.

inside (13.7)

23,550

91

0.39

15 May-19 Oct.

outside (13.8)

23,250

36

0.15

Period

Source: Tanaka et al., 1987.

outside a snake barrier fence that enclosed the village (Tanaka et al., 1987). In combination with other control actions, the traps contributed to a 10-fold reduction in snakebites.

ACKNOWLEDGMENTS This project was supported by the Japanese Bureau of Land Development and the Prefecture of Kagoshima.

LITERATURE CITED Hayashi, Y.,

H.

Kihara,

H.

Tanaka, and

M.

Kurosawa.

1984. Evaluation of a bait

trap for Habu, the venomous snake, Trimeresurus flavoviridis. Japan. J. Exp. Med. 54:171-175. Hayashi, Y., H. Kihara, H. Tanaka, Y. Noboru, H. Yamashita, and T. Minami.

1979. Studies on bait attractancy to Habu (Trimeresurus flavoviridis) in the field on Tokunoshima Island [in Japanese with English summary]. Snake 11:45-53, 126-127. Kihara, H., Y.

Hayashi, and

I.

Wakisaka.

1978. Studies on Habu attractants using

traps [in Japanese with English summary]. Snake 10:46-55, 98. Koba, K.

1971. Natural history of the Habu, Trimeresurus flavoviridis (Hallowell) [in

Japanese with English summary]. The Snake 3:75-96. Mishima,

S. 1966. Studies of the poisonous snake “Habu,” Trimeresurus flavoviridis

flavoviridis 1. Food habit of Trimeresurus flavoviridis flavoviridis on the Amami Islands [in Japanese with English summary]. Japan. J. Sanit. Zool. 17:1-21.

Development of the Box Trap for Habu Tanaka,

263

H. 1973. Activity and behavior of Habu, Trimeresurus flavoviridis [in

Japanese with English summary]. Snake 5:116-132. Tanaka,

H., Y.

Hayashi,

Kurosawa, and

Y.

H.

Sawai.

Kihara, S. Hattori, S. Mishima,

Y.

Wada, M.

1987. Population control of Habu, Trimeresurus

flavoviridis, the venomous snake, studied on Tokunoshima Island by a research group FY 1980 to 1983. Snake 19:26-40.

19 Trap Capture of Habu (Trimeresurus flavoviridis) with Odor Extracted from Rats Shosaku Hattori Yoshihisa Noboru Hiroshi Kihara Yoshihiro Hayashi

B

ox traps are widely used for reducing populations of the venomous Habu, Trimeresurus flavoviridis (Hattori, this volume, Chap. 18). To date, the most

successful attractants for Habu traps have been live birds or rodents, especially the wild rat (Rattus tanezumi). For example, in the most recent trapping series on Amamioshima, capture rates with live rats varied from 0.46% (village of Sani, 1992; 7980 trap-nights) and 0.56% (Katoku, 1992; 7980 trap-nights) to 0.95% (Kudadon, 1991; 1890 trap-nights). These capture rates are high in rela¬ tion to those obtained in most snake-trapping studies (Rodda et al., this volume, Chap. 20). The primary drawback of these box traps is their requirement of a live animal as attractant. Trapping with live bait is inconvenient for several reasons: food and water for the rat must be replenished weekly; traps must be placed in the shade or the rat will overheat from the sun; there is a risk that the rat will escape; and there are animal welfare concerns about the use of live animals. To eliminate the use of live attractants, we conducted four experiments to determine if Habu could be attracted to the odor of rats: 1. In the field feasibility study we compared trap capture success between traps in which the rat was visible to approaching snakes and those in which the rat was hidden in an opaque box. 2. In the laboratory feasibility study we observed the behavior of Habu exposed to rats or the odor extracted from rats. 3. In the box trap study we compared the capture success of traps sup¬ plied with either live rats or the odor extracted from rats. 4. In the net trap study we compared capture success among traps that were empty or supplied with either live rats or the odor extracted from rats.

264

Trap Capture of Habu with Rat Odor

265

MATERIALS AND METHODS Field Feasibility Study The field feasibility study compared trap capture success between conventional box traps with live rats (Hattori, this volume, Chap. 18) that differed only with regard to the opacity of the rat chamber. The “visible” traps used conventional mesh rat chambers; the “not-visible” traps had the rats in vented but opaque metal boxes. Each group was set for 2100 trap nights on Tokunoshima during September-November 1985.

Laboratory Feasibility Study In the laboratory feasibility study, captive Habu were exposed to either a live rat or air extracted from rat cages using the headspace and cold trap methods (Niwa et al., this volume, Chap. 11). The extracts were mixed before introduction into the olfactometer. The snakes were observed through a slit in the olfactometer (Fig. 19.1). In this study, only qualitative observations were made of the snakes.

Box Trap Experiment In this experiment, capture rates were compared between box traps supplied with either live rats or odor extracts. Each group was set for 5754 trap-nights on Tokunoshima from May 1988 to November 1989. The extract was obtained by the same method as used in the laboratory feasibility study, but cages with a larger number of rats (40-60) were needed to obtain sufficient material for release during the long field trials.

Observation

Box

Figure 19.1 Arrangement of olfactometer used for laboratory feasibility study.

266

Hattori, Noboru, Kihara, & Hayashi

Net Trap Experiment The net trap experiment included three conditions: blank control, odor extract, and live rat. These were tested in a net trap (Fig. 19.2). To be caught by a net trap, a snake must push its head under a loose overhang of polyethylene netting that is suspended by fiberglass poles and whose draped inner edge is anchored to the ground at intervals. Once a snake’s entire body is inside the net curtain, it has no means to lift the anchored edge of the netting to escape. Net traps (1.8 X 1.5 X 0.7 m) are much larger that box traps (0.4 X 0.3 X 0.2 m) and must be set in forested areas where weed growth is suppressed by reduced light levels. The live rat and odor extract conditions were tested over four years (1988-1991) on Tokunoshima (9422 trap-nights for each condition), and the blank control was tested during May-November 1988 on Tokunoshima (1190 trap-nights).

RESULTS AND DISCUSSION Field Feasibility Study There was no significant difference (8 vs. 10) in the number of snakes captured in traps supplied with “visible” or “not-visible” rats. We believe the snakes could have detected the rats by odor, vibration, or their calls or noises. Snakes lack middle ear bones, however, and are thought to be insensitive to airborne sounds (Smith, 1960). Therefore, the success of the “not-visible” traps seems attributable to some combination of odors or vibration.

Figure 19.2 The net trap for Habu. Polyethylene net is supported by fiberglass poles. The outside dimensions of the trap are 1.8 (W) X 1.5 (L) X 0.7 (H) m.

Trap Capture of Habu with Rat Odor

267

Laboratory Feasibility Study No qualitative difference was observed between Habu exposed to live rats or odor extracts. Both groups exhibited head orientation, lateral movements of the head and neck, and tongue flicking in the presence of the cue. All three behaviors were absent in the absence of a cue (blank control). This result suggests that odor cues alone are sufficient to release prey-searching behavior in Habu.

Box Trap Experiment No Habu were captured in the traps emitting only odor cues, whereas 31 Habu were captured in traps housing live rats. This suggests that Habu can discriminate between traps with live rats and those emitting only odor extracts. We do not know if the snakes’ initial orientation to the odor cues was present, as in the lab¬ oratory feasibility study, but the snakes did not progress to the stage of entering the box traps if live rats were absent.

Net Trap Experiment Some Habu entered net traps that emitted only odor cues, although the number of captures was lower than for the traps supplied with live rats (5 vs. 36). The smaller sample of blank controls yielded no captures, suggesting that Habu rec¬ ognize that a source of rat odors is of interest, but it is not as strong an attractant as is a live rat. However, the capture rate of traps with only odor cues was too low to be of practical value. The odor of rats is a very complex mixture of compounds (Niwa et al., this volume, Chap. 11), and it is possible that our cue was not of the optimal mix. Should it be found that the key components are difficult to obtain in useful quantities, it may be very challenging to develop a practical artificial attractant. Our present goal is to develop a more accurate method of assaying Habu attractants in captivity.

ACKNOWLEDGMENTS This work was supported by the National Land Agency, Japan, and the Kagoshima Prefectural Government.

LITERATURE CITED Smith,

H.

M.

Winston.

1960. Evolution of Chordate Structure. New York: Holt, Rinehart &

20 A State-of-the-Art Trap for the Brown Treesnake Gordon

H.

Rodda

Thomas

H.

Fritts

Craig S. Clark Steve W. Gotte David Chiszar

T

raps can be used to reduce the density of snakes in high-priority sites such as

homes, ports, airports, and wildlife areas. To be of practical value, however, traps must capture a high percentage of the resident snakes and must be easy to use. Between 1985 and 1993 we greatly improved trap convenience and effectiveness. Savidge (1986, 1987, 1991) used traps to quantify snake densities and document Brown Treesnake (Boiga irregularis) predation on trapped birds. Fritts and his co¬ workers (Fritts and Scott, 1985; Fritts et al., 1989) tested several trap designs and demonstrated the effectiveness of an inanimate attractant. Rodda et al. (1992b) showed the practicality of using live geckos as lures, discovered that large num¬ bers of snakes were escaping from open-funnel traps, and showed that selected design elements (olfactory guide ropes, soft flaps, double funnel entrances) were not successful. Fritts and McCoid (unpubl. data) demonstrated that snap traps baited with mammal attractants were relatively ineffective (less than 0.0002 cap¬ tures per trap-night). More recently, we showed (Rodda et al., 1992a) that high capture rates are possible with live mice as attractants, opaque chambers are effective, and flap entrances can be successful. Rodda and Fritts (1991) argued that increased trap effectiveness made it practical to use traps for operational snake control. Rodda and Fritts (1992) discussed the merits of traps and other capture methods for scientific sampling of Brown Treesnake populations. To date, we have used 49 snake trap designs for a total of more than 24,000 trap-nights of tests. In this chapter we describe the state-of-the-art trap, review the experiments that led to the currently favored design, and outline unresolved facets of Brown Treesnake trap design.

DEFINITIONS The capture rate of a trap is the number of captures per trapping event, often given as a percentage. In the studies described below, each trapping event is a

268

State-of-the-Art Trap for the Brown Treesnake

269

single trap for a single night, or one trap-night. Thus a 2% capture rate represents two snakes captured from 100 traps set for one night (or from one trap set for 100 nights, etc.). As a result of multiple captures, we sometimes obtain capture rates greater than 100%. Traps were checked each morning; however, we have seen snakes enter and escape from a trap before the morning trap check. The frequency of escapes can be estimated using a crushable object in the trap (we used a piece of aluminum foil shaped into a cylinder). Brown Treesnakes larger than about 1 m snout-vent length (SVL) invariably crush the foil cylinder while moving around in the trap. Smaller snakes usually crush the foil cylinder. Thus the presence of crushed foil in a snakeless trap (a “crush”) indicates at least one escape. The presence of crushed foil in a trap that holds a snake at the standard trap check time is not counted as a crush; it is a capture. The escape rate is the number of crushes as a percentage of the corresponding trap entries (entries = crushes + captures). The retention rate is the complement of the escape rate (e.g., if the escape rate is 40%, the retention rate is 60%). The entrance rate is entries per trap-night. Because of escapes, the capture rate is smaller than the corresponding entrance rate. The cap¬ ture rate (captures/trap-night) equals the product of entrance rate ([crushes + captures]/trap-night) and retention rate (1 - [crushes/{crushes + captures}]). Entrance rate is a useful statistic for comparing the potency of attractants. Re¬ tention rates are useful for documenting the efficacies of features that prevent or discourage snakes from escaping. Capture rate is a widely used statistic for com¬ paring overall trap effectiveness. To compare capture rates, however, trapping events should be of the same duration, unless escape rates are negligible and all captures are independent (assumptions rarely satisfied). Capture rate is a statistic often used to compare trapping results from different sites. More captures are expected in areas with more snakes, however, and a highly effective trap might have 2% success in an area with few snakes while a poor trap might have 7% success in an area of high snake density. Thus a better measure for comparing trap effectiveness is p, the estimated average probability that a given snake will be caught during a capture event; p is a density-corrected expression of capture rate. If an array of traps exhibits a p of 25%, about one-fourth of all snakes will be captured by that array in a single event. Thus p is useful for com¬ paring trap array efficacies. However, all measures of capture success, including p, are sensitive to the interaction between the environment and trap design. For ex¬ ample, traps that work well in areas with few prey may be less effective (have a lower p) in areas of high prey abundance. Closely spaced traps will catch a greater proportion of the resident snakes (higher p) than will widely spaced traps, but the capture rate (captures/trap-night) will often be lower for closely spaced traps. Trap effectiveness (p or capture rate) depends on context. A disadvantage of p is that the abundance of snakes must be known to estimate it. Snake abundance is not known for most trap sites, and when available, the population size estimates often have extremely wide confidence intervals.

270

Rodda, Fritts, Clark, Gotte, & Chiszar The two values used to compare overall trap efficacies, p and capture rate, are

easy to confuse with each other because they are similar in magnitude and both are often given as percentages. The capture rate and p are equal only when the population of snakes being sampled coincidentally numbers 100; p can only be used to describe the yield from an array of traps; capture rate can be computed for arrays or single traps.

SUMMARY OF METHODS Readers should consult the papers cited above for full methodological details. Our successful Brown Treesnake traps are “minnow” traps: mesh cylinders about 50 cm long X 20-30 cm diameter with inward-pointing funnel ends. We captured a few snakes using adhesive traps, but our capture rates with these traps were too low for control use in field situations. The minnow traps had live prey or prey compounds as attractants. Mice were protected by metal cages; they could not be eaten by snakes. Most traps were hung 1-2 m high, spaced 15 m apart in a square array in a forest. Traps were checked each morning, at which time captured snakes were marked and released at their place of capture. Attractant animals were re¬ placed or refreshed as needed. We monitored the traps for periods ranging from 20 to 49 days. Because it is impossible to directly compare trap efficacies from dif¬ ferent sites and times, we prefer to compare design permutations that are tested simultaneously in a randomized Latin square array. Most comparisons had three dimensions, each with two states, for a total of eight permutations. The use of balanced designs allows us to base statistical comparisons on numbers of entries or captures rather than rate values. We use rates to compare nonsimultaneous comparisons.

SELECTED RESULTS Trap Design Our currently favored trap, a modified commercial minnow trap, has a body of 6 mm galvanized steel mesh. In a simultaneous comparison, commercial minnow traps (Cuba Specialty Co., Fillmore, N.Y.) outperformed hand-made windowscreen traps (Fritts, 1988). Minnow traps caught more snakes (211 vs. 164; G = 5.906, df= 1, P — 0.015), and more snakes entered minnow traps (305 vs. 223; G = 12.78, df= 1, P — 0.0003). There was no evidence that the escape rate differed between trap types (94 of 305 vs. 59 of 223; G = 1.19, df= 1, P = 0.27). Commercial traps are of uniform quality and are easier to use. Window-screen traps are more easily damaged by nontarget animals, especially crabs. Both types are useful, however: commercial minnow traps capture only about 25% more snakes than window-screen traps, and window-screen traps may be preferred on islands where commercial traps are difficult to acquire. Bulk window screening is easy to transport to remote sites for assembly of traps in situ.

State-of-the-Art Trap for the Brown Treesnake

271

Solid plastic traps performed very poorly in a matched comparison with window-screen traps (capture rates 0.09% vs. 0.96%) and should not be used without additional testing. This result differs from that experienced by Habu (Trimeresurus flavoviridis), researchers, who routinely use solid traps (Hayashi et al., 1979, 1984b; Tanaka et al., 1987; Hattori, this volume, Chap. 18), although some solid designs work poorly (Kihara et ah, 1978). The arboreal Brown Treesnake may be less inclined than the terrestrial Habu to enter a burrowlike solid trap. The designs we currently favor are traps with either a plastic root (to protect the attractant) or a surrounding cylinder of black plastic sheeting forming a sleeve. Unlike the solid plastic traps, which performed poorly, our plastic-sleeved mesh traps are not completely opaque, and they allow air flow through the mesh ends. We hypothesized that darkened traps might provide an appealing refugium for Brown Treesnakes. In addition, Lankford (1989) showed that Brown Treesnakes were more likely to enter chambers to investigate chemical cues if the contents of the chambers were not visible. We compared the capture rates of traps surrounded by black plastic with identical traps with only the top half of the trap covered by black plastic. The number of snakes entering the sleeved traps was insignificantly greater (277 vs. 251; G = 1.28, df= 1, P = 0.26), although there was a slight but significant reduction in the escape rate with sleeves (66.5% vs. 75.1%; G = 4.68, df= i, p = 0.030) and therefore a significantly higher number of captures (208 vs. 167; G = 4.49, df= 1, P = 0.034). This supports the use of a sleeve in situa¬ tions for which the escape rate is a consideration. The sleeve may be undesirable on flap traps, which have a negligible escape rate, because the sleeves trap heat and may place additional thermal stress on captive and attractant animals. Inside each trap we place a hide, or refugium tube, a 200-300 mm length of 50 mm (inside) diameter black plastic pipe. These may function by withholding visual information in the manner noted by Lankford (1989). In a simultaneous comparison, more snakes entered traps with hide tubes than entered traps lack¬ ing hide tubes (283 vs. 234; G — 4.65, df= 1, P — 0.03). In that test, with openfunnel traps, fewer snakes escaped from the traps with hide tubes; thus the cap¬ ture totals differed more (215 vs. 130; G = 21.16, df— 1, P < 0.0001). The escape rate difference may be due to the snakes being less motivated to escape when they have ready access to tubes that give them shelter from bright light, drying wind, and rising daytime temperatures. However, the significant improvement in number of entries into traps having hide tubes suggests that tubes should be used even in traps from which escape is not possible. To our knowledge, refugium tubes have not been used to enhance capture success in other snake species.

Trap Placement Our current trap is placed 1—2 m high in forested areas. We have not tested traps in urban or agricultural areas. In one test we placed traps at the top of a chain-

272

Rodda, Frills, Clark, Gotte, & Chiszar

link fence known to be regularly used by Brown Treesnakes. The capture rate was very low (0.19%), for reasons that have not been determined. When we compared breast-high placement of traps in forested areas with ground-level traps, we had few captures in either location (average capture rate 0.08%). Although the breasthigh traps captured seven snakes and the ground-level traps captured only one, this difference was not significant with such a small sample size (G = 2.75, df = 1, P = 0.097). Damage from nontarget animals such as crabs was much greater in the ground-level traps, leading us to discontinue their use. We did not use drift fences or features other than attractants to concentrate snakes near the traps (cf. Imler, 1945; Campbell and Christman, 1982). Drift fences have been widely used for capturing terrestrial snakes, but most Brown Treesnakes in Guam move above ground level. Furthermore, the soil in Guam does not allow the routine placement of drift fences. Inspired by Chiszar et al. (1988), who found that captive Brown Treesnakes used odor trails to locate rodent nests, we attempted to lead snakes to traps by draping bird-litter-scented ropes through the forest toward the entrances of traps. The traps with the olfac¬ tory guide ropes captured fewer, not more, snakes (15 vs. 24). We discontinued the use of these ropes, although we believe the concept is valid and will work if an appropriate chemical cue can be discovered. We found it judicious to minimize disturbances by limiting visits to trap areas. The traps that were monitored immediately after construction of adjacent trails exhibited three to five days of below-average trap success, whereas traps that were monitored after leaving the forest undisturbed for the preceding five days exhib¬ ited normal entrance rates from the first day. Trapping success in the middle of the trapping period was usually reduced on the nights when we concurrently con¬ ducted visual searches of trap areas. We are still testing the effects of disturbance. Our working hypothesis is that trap captures are higher when trap areas are not disturbed. The most effective traps are those placed in areas known to produce high trap yields. This recommendation has little practical value, but it does point up an unresolved problem in the placement of traps. There is a poorly understood in¬ teraction between trap design and the environment. For example, identical geckoattractant traps had capture rates of 2.9% and 0.14% at two sites on Guam (Orote Point and Northwest Field, respectively). Although snakes were estimated to be about 60% more abundant at Orote, this does not account for the 2086% greater trap success there. Identical traps with mouse attractants had capture rates of 25.3% at Orote in 1990, 18.3% at Orote in 1991, and 3.3% at a different North¬ west Field site in 1992. Again, differences in snake density account for only part of the difference in capture rates. Preliminary data indicate that the structure of the forest and the abundance of natural prey may influence snake trap success. Seasonal changes in capture success need to be evaluated. All interactions between trap design and environment merit further testing.

State-of-the-Art Trap for the Brown Treesnake

273

The distance between traps depends on the reasons for trapping. If the goal is to maximize p (the percentage of the population trapped), the traps should be set as close together as is practical. If the per-trap capture rate is to be maximized, traps should be spaced more than 15 m apart. This criterion is based on several lines of evidence indicating that traps placed 15 m apart (or less, presumably) are in some sense competing for the same snakes (Rodda et al., 1992a, for Brown Treesnakes; for Habu, see Hayashi et al, 1979; Hattori, this volume, Chap. 18). For example, mouse-attractant traps at Orote exhibited the following capture rates: 18.3% for traps spaced at a trap density of 44/ha in 1991, 25.3% for traps spaced at 22/ha in 1990, and 60.1% for traps spaced at 5/ha in 1992. Snake densities did not differ significantly among these venues. The shape of the mathematical function describing capture rate by trap spac¬ ing has not been established for any species-environment combination. The func¬ tion will depend on trap design elements such as attractant type, because the sampling radius of a trap is partially dependent on the attraction radius of the attractant. For example, a trap could capture snakes within 100 m of the trap during a given night. The natural movements of the snakes might account for 90 m of this sampling area, while snakes that travel to within 10 m of the trap might be attracted to the prey. Two traps placed 50 m apart in such a situation have the potential to capture the same snake in the same night. Once caught, a snake is unavailable to neighboring traps. If the traps are checked weekly, the sampling radius of such a trap might be 500 m, of which 490 m might be due to the snakes weekly movements. Thus the degree of competition among traps will depend on the frequency of trap checks. If capture rate is to be maximized, traps that are checked infrequently should be relatively widely spaced. A practical problem arises when traps are used to reduce snake densities near an object such as a freight container or a nest tree of an endangered bird (Aguon et al., this volume, Chap. 38). If more snakes are attracted toward traps than are actually captured, placing the trap near the nest or container could increase the chances that a snake approaching a trap would find the nest or seek refuge in the container. The problem can be viewed as a matter of determining the degree to which the sampling radius of a trap is the result of the attraction radius of the trap or the natural movements of the snake. The natural movements of the snake cannot be controlled, but the traps should be placed so that their attraction zones draw snakes away from protected objects. We have not identified a way to piecisely quantify the limits of the trap attraction zone.

Attractants We used live mice as attractants. In similar but not identical situations, mouseattractant traps exhibited capture rates of around 24% compared with 6% for live quail, 3% for live geckos, and 1% for bird litter. No direct comparison between

274

Rodda, Fritts, Clark, Gotte, & Chiszar

endothermic prey has been conducted because live quail are no longer readily available on Guam. In a simultaneous comparison between live mice and live geckos, the mouse-attractant traps had not only more entries (465 vs. 52; G = 379.3, df= 1, P < 0.0001), but also a lower escape rate (30% vs. 60%). Thus the capture rates varied 15-fold (324 vs. 21; G = 314.6, df= 1, P < 0.0001). Work in progress with inanimate attractants indicates that blood, a commercial catfish bait, and a commercial snake bait did not differ among themselves in attracting snakes. Each exhibited approximately l/20th the capture rate of traps supplied with live mice in a simultaneous comparison. Despite this limited success, we will continue to test inanimate attractants, because the discovery of a durable inani¬ mate attractant would greatly improve the convenience of snake trapping. Habu researchers have reached similar conclusions: rodents are a highly effective attractant (e.g., Hayashi et al., 1979), but inanimate attractants are a high research priority (Kihara et ah, 1978; Niwa et ah, this volume, Chap. 11; Hattori et ah, this volume, Chap. 18). In the trap, the live mouse is housed in a rectangular chamber of 3 mm metal mesh closed with a rigid but porous metal cap (Fig. 20.1) and furnished with a slice of potato for moisture and a mix of grains for food. We replenish these every five days.

Entrances The state-of-the-art trap has a device for keeping the snakes in the trap. Arboreal snakes tend to spread their weight over a large number of supports; thus we have not experimented with a treadle trap, as we judge that the‘sensitivity required to detect the partial weight of a small snake (total mass < 10 g) would probably result in a trap that triggered prematurely in response to movements of the trap in the wind or movements of the attractant animals or trap motion induced by other animals, especially crabs, climbing on the outside of the trap. Furthermore, treadle traps are usually limited to one capture before they must be reset; our traps frequently have multiple captures per night. We experimented with long entrance funnels to reduce the escape rate; but in a matched comparison, the long funnels had a higher escape rate (38% vs. 32%). One successful device for forcing a snake to remain in a trap is a glueboard. We used the paper glueboards sold for capturing household mice (Victor Holdfast traps, Woodstream Corp., Fititz, Pa.). In a simultaneous test of traps in which the glueboard was either flat in the trap or rolled inside the hide tube, the rolled glueboards caught almost twice as many snakes (34 vs. 20), although the small sample precluded statistical significance (G = 3.67, df= 1, P = 0.055). Sur¬ prisingly, a larger proportion of snakes were not stuck to the glueboard in the traps with the exposed glueboards (7 of 20 vs. 2 of 34; G = 5.32, df — 1, P = 0.021). Thus traps with glueboards inside tubes not only catch more snakes, but such glueboards are also more likely to adhere to their victims. When open-fun-

State-of-the-Art Trap for the Brown Treesnake

275

View from inside trap

Figure 20.1 Designs for mouse chambers and entrance flaps. The mouse chamber body is constructed of 3-mm galvanized steel mesh made into a tube with a double-roll seam. The tube is squared and cut at the corners on one end to create four end flaps, which are lolded to close one end of the box. The chamber cap is made from perforated sheet metal stamped into a shallow open box slightly larger than the chamber body. The chamber body is slipped inside the cap, and the two are held together by a heavy rubber band. The flap housing is a plumbing fitting, a 2 X 1.5 inch flush bushing of ABS plastic. It is drilled to accommodate the hinge pin and the bolt above the hinge pin. The bolt functions only to fill the space above the hinge that a very small snake could escape through. A tilted flap is shown; a vertical flap has a smaller door that hangs vertically against the inner lip of the housing. The flap is made from either 6-mm black plastic mesh or galvanized steel mesh, and is attached to the hinge pin with small metal rings (bent wire or jewelry rings). When using wire mesh flaps it may be necessary to fold over protruding wires so they do not bind on the housing. The hinge pin may be either a straightened chrome-plated paper clip or stainless steel wire. The hinge pin is longer than the width of the housing, with the excess length bent sharply around the outside of the housing. The housing is forced into the 55 mm opening of a crawfish-style minnow trap (Cuba Specialty Co., Fillmore, N.Y., or equivalent) from the inside. The flange on the housing makes it impossible for a snake to push the housing out; the trap ends squeeze the housing between the screw head-hinge pin ends and the flange, making it difficult for the housing to be pushed in (animals are not likely to push inward, except against the flap, which opens easily).

276

Rodda, Fritts, Clark, Gotte, & Chiszar

nel (i.e., flapless) traps were supplied with glueboards, the estimated escape rate dropped from 38 to 4%. A disadvantage of glueboards is that they must be replaced periodically. The frequency with which glueboards must be replaced has not been established, but the accumulation of dust and moisture on the glue surface degrades the adhesive properties of glueboards on a scale of days. Glueboards cost $0.50—1 each. Kihara and Yamashita (1979) tested the ability of glueboards to administer dermal toxi¬ cants to Habu. Knight (1986) reported high capture rates for glueboards used to capture snakes in and under buildings. To our knowledge, glueboards have not been used previously to retain snakes in enclosure traps. In the 1991 experiment described above, flapless traps had an average escape rate of 38%. That experiment used live mice as attractant and included hide tubes in the traps. Both features reduced escape rates. In a recent experiment with traps possessing hide tubes and mouse attractant, however, the escape rate was 66%. Traps lacking these features sometimes have average escape rates in excess of 80%. Thus it is possible to substantially improve the capture rate by adding a flap, glueboard, or other device to force snakes to remain in traps. In our experiments all flap designs had escape rates of 2-4%, and we found no difference in escape rate among the flap designs tested. A disadvantage of some flap designs is that snakes may be so reluctant to enter a flap trap that the number of captures is lower than that for open-funnel traps, despite the greater escape rate from open-funnel traps. These flaps reduce cap¬ ture rates. Such a result occurred in our tests of clear acrylic flaps, soft plastic screening flaps, double funnels (the outer entrance was open, the inner entrance had a soft, springy flap), several tests of metal 6-mm-mesh flaps, and our first test of plastic 6-mm-mesh flaps. Given that the latter two were both constructed from 6 mm mesh, they were surprisingly different in their capture totals in the first matched comparison (53 vs. 4 captures; G = 53.9, df = 1, P < 0.0001). A later experiment showed that there is a three-way interaction between flap material, the angle of the flap, and whether or not the flap is painted. A full factorial array of the eight permutations from the above three factors, plus an open-funnel con¬ trol condition, produced three clusters of capture rates. Four permutations con¬ stituted the low-effectiveness cluster: the two configurations of plastic flaps that were painted (the black plastic was painted silver to be the same color as the unpainted metal flaps) and the two metal flaps that were vertical (i.e., flush with the surrounding housing; Fig. 20.1). The open-funnel traps were intermediate in capture rate. The high-effectiveness cluster had capture rates about 1.7 times those of the open funnels (G = 4.23, df — 1, P = 0.04) and 2.5—2.8 times the capture rates of the poor cluster (G = 16.29, df = 1, P < 0.001). The high-effectiveness cluster consisted of the tilted metal flaps (paint did not matter with metal flaps) and the unpainted black plastic flaps (angle did not matter with plastic flaps). The state-of-the-art trap has one of the four better flap designs. We are puzzled by the absence of an obvious explanation for the variation in capture yields among flap designs. The best flap designs that we have tested

State-of-the-Art Trap for the Brown Treesnake

111

have a lower entry rate than open-funnel traps, but a higher capture rate. There¬ fore, an opportunity exists to discover a flap design that combines the high entrance rate of an open-funnel trap with the low escape rate of a flap trap. Flaps and other entrance obstructions have been widely used in snake traps (Dargan and Stickel, 1949; Fitch, 1951; Vogt and Hine, 1982; see Appendix), but only Habu researchers have systematically tested flaps for their effect on capture rate (Kihara et al., 1978; Hayashi et al., 1979; Hattori, this volume, Chap. 18). In light of our Brown Treesnake results, experiments that compare a single flap design with an open funnel are not sufficient to reveal the full range of efficacies that flaps may have.

Sampling Considerations Unfortunately, our current trap designs preferentially capture medium- and large¬ sized snakes. Habu traps have the same problem (Hayashi et al., 1984a; Shiroma and Araki, 1986; Shiroma, 1989; Hattori, this volume, Chap. 18). In all Brown Treesnake trap experiments conducted to date, the snakes captured by hand in the vicinity of the traps exhibited a wider range of sizes than those captured by traps. In some comparisons, a few hand-captured snakes were larger than the trap cap¬ tives, but most of the difference is attributable to the paucity of small snakes cap¬ tured by traps (Fig. 20.2). Traps that use geckos as attractants catch smaller snakes than traps that use mice (Fig. 20.3), but we have been unable to extend the range

High 75%-ile Mean Median 25%-ile Low Outlier >1.5 IQR beyond median

400

N=18

85

I

I

hand

trap

1990

23

1i hand

176

1

trap

1991

68

20

1 trap

1 hand

1992

Figure 20.2 Size distribution of snakes caught by trap or hand at three venues. The sample sizes are numbers of different snakes.

278

Rodda, Fritts, Clark, Gotte, & Chiszar High 75%-ile Mean Median 25%-ile Low Outlier > 1.5 IQR beyond median

Snout-Vent Length (mm)

1300

1200

-

1100

-

1000

-

900

800

Figure 20.3 The size

700

distribution of snakes caught in mouse- or 600

N=

ip

21 I

Mice

Geckos

Bait

gecko-attractant traps at Orote Point, Guam, in 1990. Sample sizes as in Figure 20.2.

of sizes caught by simultaneously including in traps both gecko and mouse attractants (a nonsignificant contrary result was obtained). When snakes of various sizes were placed in empty open-funnel traps, the smaller snakes were the ones more likely to escape (Fig. 20.4); however, we observed a size discrepancy between trap- and hand-caught snakes even when using traps with negligible escape rates (flaps or glueboards; Fig. 20.2, 1992 se¬ ries). Thus, it appears that small snakes are less likely to enter traps. This limits but does not nullify the utility of snake traps for control purposes. We have as¬ signed a high priority to research to overcome this problem, but the means of solving it are not apparent.

THE UTILITY OF TRAPPING AS A CONTROL TECHNIQUE The traps we currently favor cost about US$6-12 each for the trap and mouse chamber. There is no commercial supplier for the entrance flaps or mouse cham¬ ber (Fig. 20.1). The labor involved in preparing these depends on the quantity

State-of-the-Art Trap for the Brown Treesnake

(7)

279

(5)

Figure 20.4 The escape rates of snakes of various sizes following evening confinement in unbaited open-funnel screening traps. The traps were checked at the standard time the fol¬ lowing morning. The sample sizes are in parentheses at the top of each bar.

desired; materials are $0.50-1 per flap. Prices and availability vary locally for mice. Once established, a colony of 100 mice requires about one hour per day of maintenance. The cost of establishing a trapping area depends on the amount of clearing of vegetation necessary for access to the traps. We spent around 50 person-hours constructing a kilometer of access trails. The labor cost of setting the traps and the labor cost of monitoring the traps depends on the spacing be¬ tween traps. It takes only a minute or two to hang a prepared trap; it takes a few seconds to check an empty trap that is not damaged. In addition to time spent processing captives, the major time costs while monitoring are replenishing mouse food and traveling between traps. Replacing mouse food takes about 1

I

280

Rodda, Frills, Clark, Gotte, & Chiszar

minute per trap; traveling between traps takes about 1.5 minutes per 100 m in easy terrain. The number of traps that can be monitored on a routine basis depends on the spacing between traps, the amount of damage the traps sustain, and the frequency of trap checks. We judge that for traps spaced at 25 m intervals, a team of two full-time persons should be able to monitor 300—600 readily accessible traps weekly. The crucial statistic for assessing the value of trapping for reducing snake populations is p. We have not been able to determine the cause of variation among sites in p. Seasonal variation is likely to play a role. The worst site tested to date exhibited an average p of about 7%, but we have obtained p values as high as 28% with the same traps. The higher rate should make possible the very rapid reduction of snake populations in bounded areas (Rodda et al., this volume, Chap. 39). In some situations, such as when manpower or traps are limited, it may be desirable to maximize capture rate rather than p. The discovery of a potential new population, such as on Saipan, calls for a high capture rate to delimit the incipi¬ ent population. Very high capture rates were obtained with widely spaced traps. For example, the 20 mouse-attractant traps at Orote Point in 1992 achieved an average capture rate over 41 days of 60%. The three best traps in this array aver¬ aged 117, 107, and 102%.

A Brief Review of Snake Trapping How do the Brown Treesnake capture rates compare with those of other snake¬ trapping experiments? No values for p are available. We compiled literature cap¬ ture rates for 322 species-trap-type permutations (Appendix). Studies using only pitfall traps were not included. About one-third of the available values are for Habu. Typical Habu studies differed in character from the other studies with regard to their use of drift fences (not used), attractant (used), and exit barriers (used). Enclosure traps were used in all cases: snap, treadle, and adhesive traps were not successful at capturing wild snakes. Drift fences were used in 43% (91 of 214) of the non-Habu studies, but in only 3% (3 of 108) of the Habu studies. To our knowledge, the effect of drift fences on snake capture has not been quantified; an opportunity exists to make this com¬ parison starting with the substantial body of data on Habu trapping in the absence of drift fences. Only 2% (2 of 108) of Habu trap studies omitted attractants, whereas 49% (105 of 214) of the other studies used none. This may account for the generally higher success rate obtained in the Habu studies (Fig. 20.5). Several varieties of rodents have been used as attractants (Mus were used in 107 of 109 non-Habu trap experiments and 87 of 106 Habu trap arrays using attractants), but there is little experience with other types of attractant for any species other than the Brown Treesnake.

State-of-the-Art Trap for the Brown Treesnake

0.0015 0.011 0.082 0.0041 0.030 0.22

0.61

4.5 1.6

281

33 12

90

Capture Rate (%) ■ Habu E3 Other snakes Figure 20.5 The distribution of capture rates in a sample of papers that describe snake trapping (Brown Treesnake studies excluded). Studies of Trimeresurus flavoviridis are shown in solid bars. The average capture rates for three Brown Treesnake studies from Orote Point, Guam, are shown by the arrows at above right (“gecko” and “mouse” desig¬ nate the attractant used in those studies; the numbers indicate the year). The abscissa is logarithmic, in increments of 1.0loge. Studies that obtained zero captures do not appear on this logarithmic distribution.

Note that the values in Figure 20.5 have been natural log transformed to nor¬ malize the distribution; the range of values in Figure 20.5 covers more than five orders of magnitude. Studies also varied about 5000-fold in the amount ot sam¬ pling effort they represented. The median Habu study reported on about 2260 trap-days; the median of the other species was 6400 trap-days. The median cap¬ ture rate of the other studies was 0.017%, about l/16th the comparable value for Habu studies (0.268%). This suggests that Habu traps are relatively effective and that opportunities exist for significantly improving the yields from most nonHabu snake-trapping projects. None of the Habu studies used open funnels, whereas 48% (103 of 214) of the other studies used open or frayed open funnels ( metal whiskers ). A great number of entrance types are in use, especially for the non-Habu studies. In light of the significance associated with entrance design in the Brown Treesnake stud¬ ies, there appears to be a need for validation of the entrance designs used in most other snake studies.

282

Rodda, Fritts, Clark, Gotte, & Chiszar Most North American snake traps had a body of mesh, although traps that

were partially or wholly solid were used in 92% (99 of 108) of the Habu studies. Only 56% (119 of 214) of the other studies used such traps; most of these were projects conducted incidental to Habu studies. The appropriateness of a material probably varies from species to species, but it is notable that Okinawa-style Habu traps utilize mesh sides, whereas Amami-style Habu traps are completely solid. In matched comparisons of solid and mesh-sided traps for the Habu, the mesh-sided traps proved superior in some cases and inferior in others, but the Okinawa-style traps have not been compared directly with the Amami-style traps (Kihara et al., 1978; Shiroma and Akamine, 1987; Shiroma and Arakaki, 1989; Shiroma, 1990; Shiroma and Nohara, 1991; Nishimura, 1992). Compared with both Habu and other species trap experiments, the Brown Treesnake capture rates are relatively high (Fig. 20.5 arrows). Most published cap¬ ture rate values greater than 1% were obtained in special situations (Appendix). For example, the 11.6% capture rate for Diudophis punctutus must be considered in light of the extremely high densities (719-1849/ha) reported for this dimi¬ nutive snake (Appendix). The 28.45% capture rate reported for Entechinus (Cyclophiops) semicarinatus is an aggregate of all captures along a 90 m fence trap, and is therefore not comparable with the enclosure traps typically used. For ordinary enclosure traps, the highest capture rates for large snake species were around 2.5% for Habu and 6% for Pituophis melanoleucus. These indicate that the 25-60% capture rates of Brown Treesnakes are exceptional.

ACKNOWLEDGMENTS We thank Michael McCoid (Guam Division of Aquatic and Wildlife Resources) for providing mice and laboratory facilities; the U.S. Department of the Interior’s Office of Territorial and International Affairs for support; the U.S. Department of Defense for financial support and access to base forests; Renee Rondeau, Brian Smith, Lisa Close, Matt Reid, and Todd Mabee for assistance monitoring traps; and R. Rondeau, Earl W. Campbell III, and Ken Dodd for suggesting improve¬ ments to the manuscript.

LITERATURE CITED Campbell, H. S., and S. R Christman. 1982. Field techniques for herpetofaunal

community analysis. In N. J. Scott Ir., ed., Herpetological Communities, pp. 193-200. U.S. Fish Wildl. Serv. Wildl. Res. Rep. 13. Chiszar, D., K. Kandler, R. Lee, and H. M. Smith. 1988. Stimulus control of preda¬ tory attack in the Brown Tree Snake (Boiga irregularis). 2. Use of chemical cues during foraging. Amphibia-Reptilia 9:77-88. Dargan, L. M., and W. H. Stickel. 1949. An experiment with snake trapping. Copeia 1949:264-268.

State-of-the-Art Trap for the Brown Treesnake

283

H. S. 1951. A simplified type of funnel trap for reptiles. Herpetologica 7:77-80. Fritts, T. H. 1988. The Brown Tree Snake, Boiga irregularis, a Threat to Pacific Islands. U.S. Fish Wildl. Serv., Biol. Rep. 88(31). Fritts, T. H., and N. J. Scott Jr. 1985. The Brown Tree Snake on Guam: Studies of its ecology, control, and threats to other islands. Report to U.S. Fish and Wildlife

Fitch,

Service, Region 1, Portland, Ore. Fritts, T. H., N. J. Scott Jr., and B. E. Smith. 1989. Trapping Boiga irregularis on Guam using bird odors. J. Herpetol. 23:189-192. Hayashi, Y., S. Hattori, and H. Tanaka. 1984a. The electric net trap for venomous snake, Flabu, Trimeresurus flavoviridis. Japan. J. Exp. Med. 54:207-210. Hayashi, Y., H. Kihara, H. Tanaka, and M. Kurosawa. 1984b. Evaluation of a bait trap for Habu, the venomous snake, Trimeresurus flavoviridis. Japan. J. Exp. Med. 54:171-175. Hayashi, Y., H. Kihara, H. Tanaka, Y. Noboru, H. Yamashita, and T. Minami. 1979. Studies on attractancy of Habu (Trimeresurus flavoviridis) in the field on Tokunoshima Island [in Japanese with English summary]. Snake 11:45-53, 126-127. Imler, R. H. 1945. Bullsnakes and their control on a Nebraska wildlife refuge. J. Wildl. Manage. 9:265-273. Kihara, H., Y. Hayashi, and I. Wakisaka. 1978. Studies on attractants to Habu using traps [in Japanese with English summary]. Snake 10.46—55, 98. Kihara, H., and H. Yamashita. 1979. Development of a new type trap with adhe¬ sive seat containing pesticides [in Japanese with English summary]. Snake 11:6-10, 119Knight,

J. E. 1986. A humane method for removing snakes from dwellings. Wildl.

Soc. Bull. 14:301-303. Lankford, J. D. 1989. Stimulus control of foraging in Brown Tree Snakes {Boiga irregularis). J. Colo.-Wyo. Acad. Sci. 21:12 (abstr.). Nishimura, M. 1992. Trapped numbers of Trimeresurus flavoviridis and Dinodon semicarinatus—analyses of the published data [in Japanese with English sum¬ mary]. Biol. Mag. Okinawa 30:15—23. Rodda, G. H., and T. H. Fritts. 1991. The practicality of snake elimination from small bounded plots. Report on file with National Ecology Reseach Center, Washington, D.C. _. 1992. Sampling techniques for an arboreal snake, Boiga irregularis. Micronesica 25:23-40. Rodda, G. H., T. H. Fritts, C. S. Clark, S. W. Gotte, E. W. Campbell III, and M. J. McCoid. 1992a. Trap technology and population dynamics 1990-1992. Report on file with National Ecology Research Center, Washington, D.C. Rodda, G. H., R. J. Rondeau, T. H. Fritts, and O. E. Maughan. 1992b. Trapping the arboreal snake Boiga irregularis. Amphibia-Reptilia 13:47-56. Savidge, J. A. 1986. The role of disease and predation in the decline of Guams avifauna. Ph.D. diss. Univ. Illinois, Urbana-Champaign. _. 1991. Population characteristics of the introduced Brown Tree Snake (Boiga irregularis) on Guam. Biotropica 23:294—300.

284

Rodda, Fritts, Clark, Gotte, & Chiszar

Shiroma,

H. 1989. On the population estimate of the Habu, Trimeresurus flavoviridis,

by removal method with traps. In M. Matsui, T. Hikida, and R. C. Goris, eds., Cur¬ rent Herpetology in East Asia, pp. 384-392. Kyoto: Herpetological Society of Japan. -. 1990. Catching Habu, Trimeresurus flavoviridis, by traps [in Japanese]. In C. Yoshida, ed., Reports of Ecological Researches to Diminish Bites of Habu

(Trimeresurus flavoviridis) in Okinawa Prefecture, vol. 13, pp. 105-107. Ozato, Okinawa, Japan: Habu Study Sect., Okinawa Prefectural Inst, of Health and Env.

Shiroma, H., and H. Akamine. 1987. Some improvements of Habu (Trimeresurus flavoviridis) trap 2 [in Japanese]. In C. Yoshida, ed., Reports of Ecological Researches to Diminish Bites of Habu (Trimeresurus flavoviridis) in Okinawa Prefecture, vol. 10, pp. 49-52. Ozato, Okinawa, Japan: Habu Study Sect., Okinawa Prefectural Inst, of Health and Env. Shiroma,

H., and S. Arakaki. 1989a. Comparing the trapping efficiencies of two

types of Habu (Trimeresurus flavoviridis) trap [in Japanese]. In C. Yoshida, ed., Reports of Ecological Researches to Diminish Bites of Habu (Trimeresurus

flavoviridis) in Okinawa Prefecture, vol. 12, pp. lb-11. Ozato, Okinawa, Japan: Habu Study Sect., Okinawa Prefectural Inst, of Health and Env. Shiroma,

H., and Y. Araki. 1986. An estimation of density of Habu, Trimeresurus

flavoviridis, in a forest of Okinawa Island [in Japanese with English summary]. Biol. Mag. Okinawa 24:43-48. Shiroma,

H., and Y. Nohara. 1991. Catching Habu, Trimeresurus flavoviridis, by

traps II [in Japanese]. In C. Yoshida, ed., Reports of Ecological Researches to Diminish Bites of Habu (Trimeresurus flavoviridis) in Okinawa Prefecture, vol. 14, pp. 97-100. Ozato, Okinawa, Japan: Habu Study Sect., Okinawa Prefectural Inst, of Health and Env. Tanaka, H.,

Y. Hayashi, H. Kihara, S. Hattori, S. Mishima, Y. Wada, M.

Kurosawa, and

Y. Sawai. 1987. Population control of Habu, Trimeresurus

flavoviridis, the venomous snake, studied on Tokunoshima Island by a research group FY 1980 to 1983. Snake 19:26-40. Vogt,

R. C., and R. L. Hine. 1982. Evaluation of techniques for assessment of

amphibian and reptile populations in Wisconsin. In N. J. Scott Jr., ed., Herpeto¬ logical Communities, pp. 201-217. U.S. Fish Wildl. Serv., Wildl. Res. Rep. 13.

p S HH £ C/5

o w I—I

ffi C/5


73

£

6 *3 ^ 6 05 -G

G cn

in

05

cO

"G

5

C/5

C/5 Lh

c3 4-,

73 -G

4-4


H

VO CO

CN

Ov N5"

^-i

CN

in

00 o

o

d

(V. O o o

IV. O o

^3

^

(U

< cn D

CJ

vo o o d

(N

-a x>

Cl Cl Cl Cl o O

CM

CL Cl Cl OOO

N

p

C

CJ '>L

1—1

ON rf

OO

OO LT> (N

ON

r^ 0 rs

m

u

X

03

cd

» E CD

in x

in

x

+->

u

u • —< C/5

bO

3

*3 3

"c

T3 ^3 ■pCX Cx CX Cx cx ex cx ex ex ex Oh Oh Q o 'o O £ C/5 l/> o o o o o o o o o o O J-. Vh 1-1 l-H Xr

VO vO 0 &o 00 O O oT o3 o3 o3 r* c n ri rj •G •G ^H _G _£ _£

o

o o

5' 03' 03 03 03 G

G

G

G

£

03 G

(4 ct Kf Vh o3 -G -G -G rG G G G G -H H-H l-H o3 ca ca 03 03 £ £ £ ca ca 03 03 ca o3 ca 03 G G G G G G G C

CO "G G r2 CO

7272727272727272

OOOOOOOO G G G G o3 o3 o3 Oh Oh Oh Oh OJ 03 03

G G o3 03 Oh Oh Oh Oh t 03 o3 03 o3 G

03

G 03

CO

c ca C ^2 • iH

o3

o3

o3

G

G

G

c c c

2 2 2

2 2

o

C/3

C/3

CO

C/5

CO

C/5

d "d d coo cd r3 o3 M

L—H

H-H

c3 d cd

III 0 0 0

12 72 2

O O O G

G G o3 Oh Oh Oh 03 o3 03

o H->

D

ca £

• ^H

hO co O c o hX e2 c ca

Oh

ca

(SJ

fli

(V)

m

m

m

m

(SJ (N rc (N

(D

« « w



r«J

CL 03 cd 03 H H CL X W X 03 03 o < G 03 CQ hG G X

13 33

6

^

« | o

X,

X

Sc

5 C/3

S C/5

O

G ©

c/5 ©

2

•-

02

Lh

©

XG

R

.a G

P-1

2

T3 G cd

cw HH q

w OS

-

^

p

G

• f—H

§ ON

Q

cO

C/3

s

z X Si
1-H

^ O +H _r m "P NO G I -H

53

I NO cn »“H

S r '

G 03

cu

^

e

3

7T

.-a

2

2

g

Ph

cO

o

o» > O

c

■P C c0 P5

g

p © p 3

^ p

c0

%

T m

Vi

|o

3

Ctf X

G co CP co

03 c/5 03 G co CP cO

os o 03 +C pH CO N

oo

C0

I G

.L• T-* P O

X 2

X

c

o

vi

03

o

■d 3J

^ r ^ XJ C l-h

G co Cp

Xi

y

cC ^ - >'

'P' P +-> C/3

03 N

CU

r LH

Cp

cO

I G

cS

"G G O u '

2

co CP CP « co ^

l

"

_T

03

_G

K

^

2

0

a L. a

t3

cp

nd G o3 cy>

G

-X C/5 cO Lh

'O J-H 4-* X G CP O CJ 03 Lh Lh 4-* • ^ 2 O z cn -rf g W in in vo vo - *r* G -S GO ^ 3 § .3 3* Lh DO ^ CD 2 GCD 3 co Si ca

cd 1 £ GG ^3

CD 3 3 C/5 Eh u G

Ch cd Oh cd

So\ UN

^ JhI CJ

^

* S I

u 3 *^7 C/5 -1 G « W O

-a

c Ed 3 S £ L*-h ° 4-> c CD

3 L. 3 CO 2 CL

S

^ x o H X ° -7


tLH G O W 3 CO

c/3 05

3 co

Si

03

G
5 Q Z
o c l_l • x«» CX X o £ X 3

£ 3 C/5 Q

*-x § G co

5 00 • ^ CG

ON ON

e 3 of 3 > V5 03 05 l-i | 05 G G £ 5—* 05 ^ • , —1

3 (A