EVOLUTION, ECOLOGY, CONSERVATION, AND MANAGEMENT OF HAWAIIAN BIRDS: A VANISHING AVIFAUNA IRONTISPIICI: Hawai'i '0'0 (1o1< nbilis), largesl of the '0'0's and best-known by the indigenous inhab- itants of Hawai'i. Formerly widespread and ½onon on the island of Hawai'i, it disappeared early in the 20  century, with Ihe last known specimen from Ka'0 Crater on 13 May 1902 (Kepler el ai. in press). Painting by H. Douglas Pram Jr. EVOLUTION, ECOLOGY, CONSERVATION, AND MANAGEMENT OF HAWAIIAN BIRDS: A VANISHING AVIFAUNA J. Michael Scott, Sheila Conant, and Charles van Riper, III, Editors Studies in Avian Biology No. 22 A PUBLICATION OF THE COOPER ORNITHOLOGICAL SOCIETY Cover photograph of 'Anianiau (Hemignathus parvus) foraging on kanawao (Broussaisia arguta) by Jack Jeffrey. STUDIES IN AVIAN BIOLOGY Edited by John T Rotenberry Department of Biology University of California Riverside, CA 92521 Studies in Avian Biology is a series of works too long for The Condor, published at irregular intervals by the Cooper Ornithological Society. Manu- scripts for consideration should be submitted to the editor. Style and format should follow those of previous issues. Price $29.00 for softcover and $48.50 for hardcover including postage and handling. All orders cash in advance; make checks payable to Cooper Orni- thological Society. Send orders to Cooper Ornithological Society, % Western Foundation of Vertebrate Zoology, 439 Calle San Pablo, Camarillo, CA 93012. ISBN: 1-891276-25-5 (cloth) ISBN: 1-891276-18-2 (paper) Library of Congress Control Number: 2001 131292 Printed at Allen Press, Inc., Lawrence, Kansas 66044 Issued: 16 March 2001 Revised: 8 March 2002 Copyright ¸ by the Cooper Ornithological Society 2001 CONTENTS DEDICATION ..................................................... viii LIST OF AUTHORS ............................................... ix INTRODUCTION .................................................. ........... J. Michael Scott, Sheila Conant, and Charles van Riper, III 1 HISTORICAL PERSPECTIVES Introduction ... Charles van Riper, III, Sheila Conant, and J. Michael Scott 14 How many bird species in Hawai'i and the Central Pacific before first con- tact? ................................ John Curnutt and Stuart Pimm 15 Patterns of success among introduced birds in the Hawaiian Islands ...... ............. Michael P. Moulton, Karl E. Miller, and Eric A. Tillman 31 SYSTEMATICS Introduction ........................................ Helen E James 48 Molecular systematics and biogeography of the Hawaiian avifauna ....... .......................... Robert C. Fleischer and Carl E. Mcintosh 51 Evolutionary relationships and conservation of the Hawaiian anatids ..... ................................................ Judith M. Rhymer 61 The interplay of species concepts, taxonomy, and conservation: lessons from the Hawaiian avifauna .......... H. Douglas Pratt and Thane K. Pratt 68 Why the Hawai'i Creeper is an Oreomystis: what phenotypic characters re- veal about the phylogeny of Hawaiian honeycreepers H. Douglas Pratt 81 Phylogenetic placement of the Po'ouli, Melamprosops phaeosoma, based on mitochondrial DNA sequence and osteological characters ............. ................. Robert C. Fleischer, Cheryl L. Tarr, Helen E James, Beth Slikas, and Carl E. Mcintosh 98 STATUS AND TRENDS Introduction ............... J. Michael Scott and Charles van Riper, III 106 The status and population trends of the Newell's Shearwater on Kaua'i: insights from modeling .... David G. Ainley, Richard Podolsky, Leah Deforest, Gregory Spencer, and Nadav Nur 108 Migration of Northern Pintail across the Pacific with reference to the Ha- waiian Islands ......... Miklos D. E Udvardy and Andrew Engilis, Jr. 124 The Hawai'i rare bird search 1994-1996 .............................. ..................... Michelle H. Reynolds and Thomas J. Snetsinger 133 Status and distribution of the Po'ouli in the Hanaw¾ Natural Area Reserve between December 1995 and June 1997 .............. Paul E. Baker 144 ECOLOGY Introduction ......................................... Sheila Conant 152 Drepanidine movements in relation to food availability in subalpine wood- land on Mauna Kea, Hawai'i ...................................... ............... Steven C. Hess, Paul C. Banko, Michelle H. Reynolds, Gregory J. Brenner, Leona P. Laniawe, and James D. Jacobi 154 Breeding productivity and survival of the endangered Hawai'i Creeper in a wet forest refuge on Mauna Kea, Hawai'i ........................... ............... Bethany L. Woodworth, Jay T. Nelson, Erik J. Tweed, Steven G. Fancy, Michael P. Moore, Emily B. Cohen, and Mark S. Collins 164 Significance of old-growth forest to the Hawai'i ',kepa ................ ................................................ Leonard A. Freed 173 Demographic comparisons between high and low density populations of Hawai'i 'kepa .................................... Patrick J. Hart 185 Breeding characteristics of the 'Akohekohe on east Maui ............... .......................... Ellen M. VanGelder and Thomas B. Smith 194 'kohekohe response to flower availability: seasonal abundance, foraging, breeding, and molt ............................................... ...................... Kim E. Berlin, John C. Simon, Thane K. Pratt, James R. Kowalsky, and Jeff S. Hatfield 202 Age-related diet differences in two nectar-feeding Drepanidines: the ',kohe- kohe and the 'Apapane ........................... John H. Carothers 213 LIMITING FACTORS Introduction ............... J. Michael Scott and Charles van Riper, III 220 Limiting factors affecting Hawaiian native birds ....................... ......................... Charles van Riper, III, and J. Michael Scott 221 Habitat use and limiting factors in a population of Hawaiian Dark-rumped Petrels on Mauna Loa, Hawai'i .................................... ........... Darcy Hu, Catherine Glidden, Jill S. Lippert, Lena Schnell, James S. MacIvor, and Julian Meisler 234 Interaction between the Hawaiian Dark-rumped Petrel and the Argentine ant in Haleakalg National Park, Maui, Hawai'i .......................... ... Paul D. Krushelnycky, Cathleen S. N. Hodges, Arthur C. Medeiros, and Lloyd L. Loope 243 Distribution and potential impacts of avian poxlike lesions in 'Elepaio at Hakalau Forest National Wildlife Refuge ......... Eric A. VanderWerf 247 Immunogenetics and resistance to avian malaria in Hawaiian honeycreepers (Drepanidinae) ................... Susan I. Jarvi, Carter T Atkinson, and Robert C. Fleischer 254 Changes in native and introduced bird populations on O'ahu: infectious dis- eases and species replacement ..................................... ................ Cherie Shehata, Leonard Freed, and Rebecca L. Cann 264 What caused the population decline of the Bridled White-eye on Rota, Mar- iana Islands? ............. Steven G. Fancy and Thomas J. Snetsinger 274 The evolution of passefine life histories on oceanic islands, and its impli- cations for the dynamics of population decline and recovery .......... ........................................... Bertram G. Murray, Jr. 281 Newly emergent and future threats of alien species to Pacific birds and ecosystems .... Lloyd L. Loope, Francis G. Howarth, Frederick Kraus, and Thane K. Pratt 291 RECOVERY AND MANAGEMENT Introduction ...................... J. Michael Scott and Sheila Conant 306 Effects of predator control on the survival and breeding success of the en- dangered Hawaiian Dark-rumped Petrel ............................. ............. Cathleen S. Natividad Hodges and Ronald J. Nagata, Sr. 308 Foraging behavior and temporal use of grasslands by NSnS: implications for management ................. Friederike Woog and Jeffrey M. Black 319 An ecosystem-based management approach to enhancing endangered water- bird habitat on a military base ........................ Diane Drigot 329 Why isn't the Nihoa Millerbird extinct? ............................... ................................... Sheila Conant and Marie Morin 338 Reintroduction and translocation of 'Oma'o: a comparison of methods .... ............ Steven G. Fancy, Jay T. Nelson, Peter Harrity, Jope Kuhn, Marla Kuhn, Cyndi Kuehler, and Jon G. Giffin 347 Restoration techniques for Hawaiian forest birds: collection of eggs, artificial incubation and hand-rearing of chicks, and release to the wild ......... .......... Cyndi Kuehler, Alan Lieberman, Peter Harrity, Marla Kuhn, Jope Kuhn, Barbara Mcllraith, and John Turner 354 Conservation status and recovery strategies for endemic Hawaiian birds .. ................. Paul C. Banko, Reginald E. David, James D. Jacobi, and Winston E. Banko 359 Evaluating the cost of saving native Hawaiian birds .................... ............................................ William W. M. Steiner 377 LITERATURE CITED .............................................. 384 DEDICATION This STUDIES IN AVIAN BIOLOGY volume is dedicated to Dean Amadon, Paul H. Baldwin, and David Woodside, colleagues and friends who laid the foundation for the recent renaissance of studies of the endemic birds of Hawai'i and a link with ornithologists of the late 19 th century. It is because many of the researchers in Hawai'i, and those in particular who have contributed to this book, have anchored their scientific premises and hypotheses on the contributions of these three men, that we dedicate this STUDIES IN AvIAN BIOLOGY to them. Dean Amadon was stationed with the U.S. Army in Hawai'i in 1944 and 1945, spending most of his time on the island of O'ahu, and two months on the Big Island as well. His interest in Hawaiian honeycreepers had been aroused earlier while he was at the American Museum of Natural History working with the ornithological collections of Lord Walter Rothschild. In Hawai'i, Amadon worked with Bishop Museum collections and got into the field to observe birds whenever he was free from his military duties. After the war he returned to academia to earn his doctorate at Cornell University. His dissertation, eventually published as The Hawaiian Honeycreepers (Amadon 1950), became a classic work on the systematics of the honeycreepers. It was the first thorough revision of the group based on Mayifs "modem synthesis" of evolutionary theory. While working on the Big Island, Amadon had been assisted by Paul Baldwin, whose research focused on life history and ecology of the honeycreepers. Paul H. Baldwin was one of the true pioneers of Hawaiian ornithology. During the 1930s while Paul was working on his master's of science (on ocean crabs) at the University of Hawai'i, he was selected biologist for the Civilian Conservation Corps, stationed at Hawai'i Volcanoes National Park. It was at this position that Paul began collecting the first quantitative behavioral information on the Hawaiian avifauna. Following World War II, he enrolled at the University of California at Berkeley to complete his PhD. Coupling information that he had collected at Volcanoes National Park during the 1930s with intensive fieldwork in 1948-1949, Paul completed the first intensive behavioral work on banded Hawaiian honeycreepers. His study quantified for the first time physi- ological cycles, population movement patterns, avian diets, and evolutionary patterns in Hawaiian birds. He correlated these data with environmental factors (particularly climate), forest structure, and resource availability. Paul Baldwin's 1953 paper, Annual cycle, environment and evolution in the Hawaiian honeycreepers (Aves: Drepaniidae), still stands as a milestone in Hawaiian ornithol- ogy. Paul's contributions to Hawai'i extend far beyond his 1953 work, with seminal papers on the Nene, a number on introduced birds (e.g., the Red-billed Leiothrix), economic impacts of the in- troduced mongoose, and impacts of cattle grazing on the native forests. David Woodside was 15 years old when he began assisting George C. Munro in the field. Munro later published Birds of Hawaii (1944), which included the first comprehensive survey of the dis- tribution of Hawaiian forest birds since the turn of the century. Woodside has worked with virtually every well-known ornithologist and agency that has engaged in research on Hawaiian birds, and has probably seen more Hawaiian birds and visited more haunts of Hawaiian birds than any living person. He was employed as a wildlife biologist for the Territory and later the State of Hawai'i for many years. After retiring from the state wildlife agency, he began working for the refuge branch of the U.S. Fish and Wildlife Service in 1980, where he continues to work today. Dave joined the Hawaii Audubon Society as a charter member when he was 15, and has contributed his time and expertise to studies and conservation of Hawaiian birds for a lifetime. Although he has witnessed the extinction of many Hawaiian birds, he is among the fortunate few living souls who have seen such birds as the O'ahu 'Alauahio, 'O'fi, Kama'o, and Kaua'i 'O'O. LIST OF AUTHORS DAVID G. AINLEY H. T. Harvey and Associates 3150 Almaden Expressway, Suite 145 San Jose, CA 95118 CARTER T. ATKINSON U.S. Geological Survey Pacific Island Ecosystems Research Center P.O. Box 218 Hawaii National Park, HI 96718 PAUL E. BACER U.S. Geological Survey Pacific Island Ecosystems Science Center P.O. Box 44 Hawaii National Park, HI 96718 (Present address: 8 Raglan Court, Silloth Cumbria, CA5 4BW, UK) PAUL C. BANKO U.S. Geological Survey Pacific Island Ecosystems Research Center Kilauea Field Station, P.O. Box 44 Hawaii National Park, HI 96718 WINSTON E. BANKO U.S. Geological Survey Pacific Island Ecosystems Research Center Kilauea Field Station Hawaii National Park, HI 96718 (Present address: 332 Redwood Place, College Place, WA 99324) KIM E. BERLIN U.S. Geological Survey Pacific Island Ecosystems Research Center P.O. Box 44 Hawaii National Park, HI 96718 JEFFREY M. BLACK Department of Wildlife Humboldt State University Arcata, CA 95521-8299 GREGORY J. BRENNER U.S. Geological Survey Pacific Island Ecosystems Research Center P.O. Box 44 Hawaii National Park, HI 96718 REBECCA L. CANN Department of Genetics & Molecular Biology John A. Bums School of Medicine University of Hawaii at Manoa Honolulu, HI 96822 JOHN H. CAROTHERS Museum of Vertebrate Zoology and Department of Zoology University of California Be/keley, CA 94720 EMILY B. C(HEN U.S. Geological Survey Pacific Island Ecosystems Research Center P.O. Box 44 Building 344 Hawaii National Park, HI 96718 MARK S. COLLINS U.S. Geological Survey Pacific Island Ecosystems Research Center P.O. Box 44, Bu.ilding 344 Hawaii National Park, HI 96718 SHEILA CONANT Department of Zoology University of Hawaii at Manoa 2538 McCarthy Mall Honolulu, HI 96822 JOHN CURNUTT Department of Ecology and Evolutionary Biology University of Tennessee Knoxville, TN 37919 REGINALD E. DAVID Rang Productions EO. Box 1371 Kailua-Kona, HI 96745 LEAH DEFOREST EO. Box 6122 Hilo, HI 96720 DIANE DRIGOT Environmental Department Marine Corps Base Hawaii MCBH Kaneohe Bay, HI 96863-3002 ANDREW ENGILIS, JR. Ducks Unlimited, Inc. 3074 Gold Canal Drive Rancho Cordova, CA 95670 (Present address: Department of Wildlife, Fish, and Conservation Biology University of California Davis, CA 95616) STEVEN G. FANCY U.S. Geological Survey Pacific Island Ecosystems Research Center P.O. Box 44, Building 344 Hawaii National Park, HI 96718 ROBERT C. FLEISCHER Molecular Genetics Laboratory National Zoological Park Smithsonian Institution Washington, DC 20008 LEONARD A. FREED Department of Zoology University of Hawaii at Manoa Honolulu, HI 96822 JON G. GIFFIN Hawaii Dept. Land and Natural Resources Division of Forestry and Wildlife P.O. Box 4849 Hilo, HI 96720 CATHERINE GLIDDEN Hawaii Volcanoes National Park P.O. Box 52 Hawaii National Park, HI 96718-0052 PETER HARRITY The Peregrine Fund Keauhou' Bird Conservation Center P.O. Box 39 Volcan,o, HI 96785 PATRICK J. HART . Department of Zoology University of Hawaii at Manoa Honolulu, HI 96822 JEFF S. HATFIELD U.S. Geological Survey Patuxent Wildlife Research Center Laurel, MD 20708-4017 STEVEN C. HESS U.S. Geological Survey Pacific Island Ecosystems Research Center P.O. Box 44 Hawaii National Park, HI 96718 CATHLEEN S. NATIVIDAD HODGES Haleakala National Park Resources Management Division P.O. Box 369 Makawao, Maui, HI 96768 FRANCIS G. HOWARTH Department of Natural Sciences Bernice P. Bishop Museum P.O. Box 19000-A Honolulu, HI 96819 DARCY HU Hawaii Volcanoes National Park P.O. Box 52 Hawaii National Park, HI 96718-0052 JAMES D. JACOBI U.S. Geological Survey Pacific Island Ecosystems Research Center P.O. Box 44 Hawaii National Park, HI 96718 HELEN E JAMES Department of Vertebrate Zoology National Museum of Natural History Smithsonian Institution Washington, DC 20560 SUSAN I. JARVl Molecular Genetics Laboratory National Zoological Park Smithsonian Institution 3001 Connecticut Ave. NW Washington, DC 20008 (Present address: Department of Biology University of Hawaii-Hilo 200 W. Kawili St. Hilo, HI 96720) JAMES R. KOWALSKY U.S. Geological Survey Pacific Island Ecosystems Research Center P.O. Box 44 Hawaii National Park, HI 96718 FRED KRAUS Department of Land and Natural Resources Division of Forestry and Wildlife 1151 Punchbowl Street Honolulu, HI 96813 PAUL D. KRUSHELNYCKY U.S. Geological Survey Haleakala National Park Field Station Box 39 Makawao, HI 96768 CYNDI KUEHLER The Peregrine Fund Keauhou Bird Conservation Center P.O. Box 39 Volcano, HI 96785 JOPE KUHN The Peregrine Fund Keauhou Bird Conservation Center P.O. Box 39 Volcano, HI 96785 MARLA KUHN The Peregrine Fund Keauhou Bird Conservation Center P.O. Box 39 Volcano, HI 96785 LEONA P. LANIAWE U.S. Geological Survey Pacific Island Ecosystems Research Center P.O. Box 44 Hawaii National Park, HI 96718 ALAN LIEBERMAN The Peregrine Fund Keauhou Bird Conservation Center P.O. Box 39 Volcano, HI 96785 JILL S. LIPPERT Hawaii Volcanoes National Park P.O. Box 52 Hawaii National Park, HI 96718-0052 LLOYD L. LOOPE U.S. Geological Survey Pacific Island Ecosystems Center Haleakala Field Station P.O. Box 369 Makawao, Maui, HI 96768 JAMES S. MACIVOR 1207 Tedford Way Oklahoma City, OK 73116 BARBARA MCILRAITH The Peregrine Fund Keauhou Bird Conservation Center P.O. Box 39 Volcano, HI 96785 CARL E. MCINTOSH Molecular Genetics Laboratory National Zoological Park Smithsonian Institution Washington, DC 20008 ARTHUR C. MEDE1ROS U.S. Geological Survey Haleakala National Park Field Station Box 39 Makawao, HI 96768 JULIAN MEISLER P.O. Box 851 Burlington, VT 05402 KARL E. MILLER Department of Wildlife Ecology & Conservation P.O. Box 110430 University of Florida Gainesville, FL 32611-0430 MICHAEL P. MOORE U.S. Geological Survey Pacific Island Ecosystems Research Center P.O. Box 44, Building 344 Hawaii National Park, HI 96718 MARIE MORIN Department of Zoology University of Hawaii at Manoa 2538 McCarthy Mall Honolulu, HI 96822 MICHAEL P. MOULTON Department of Wildlife Ecology & Conservation P.O. Box 110430 University of Florida Gainesville, FL 32611-0430 BERTRAM G. MURRAY, JR. Graduate Program in Ecology and Evolution 80 Nichol Avenue Rutgers Uniqersity New Brunswick, NJ 08901-2882 RONALD J. NAGATA, gR. Haleakala National Park Resources Management Division P.O. Box 369 Makawao, Maui, HI 96768 JAY T. NELSON U.S. Geological Survey Pacific Island Ecosystems Research Center P.O. Box 44, Building 344 Hawaii National Park, HI 96718 NADAV NUR Point Reyes Bird Observatory Stinson Beach, CA 94970 STUART PIMM Department of Ecology and Evolutionary Biology University of Tennessee Knoxville, TN 37919 RICHARD PODOLSKY Avian Systems Fort Lee, NJ 07024 H. DOUGLAS PRATT Museum of Natural Science Louisiana State University Baton Rouge, LA 70893 THANE m. PRATt U.S. Geological Survey Pacific Island Ecosystems Research Center P.O. Box 44 Hawaii National Park, HI 96718 MICHELLE H. REYNOLDS U.S. Geological Survey Pacific Island Ecosystems Research Center P.O. Box 44 Hawaii National Park, HI 96718 JUDITH M. RHYMER Department of Wildlife Ecology University of Maine Orono, ME 04469 LENA SCHNELL US Army, CDR, USAG-HI-PTA Attn: Evironmental Office APO AP 96556-5703 J. MICHAEL SCOTI' U.S. Geological Survey IDCFWRU College of Natural Resources University of Idaho Moscow, ID 83844-1141 CHERIE SHEHATA Department of Genetics and Molecular Biology John A. Burns School of Medicine University of Hawaii at Manoa Honolulu, HI 96822 JOHN C. SIMON U.S. Geological Survey Pacific Island Ecosystems Research Center P.O. Box 44 Hawaii National Park, HI 96718 BETH SLIKAS Molecular Genetics Laboratory National Zoological Park Smithsonian Institution Washington, DC 20008 THOMAS B. SMITH Department of Biology San Francisco State University 1600 Holloway Ave. San Francisco, CA 94132 (Present address: Center for Population Biology University of California Davis, CA 95616) THOMAS J. SNETSINGER U.S. Geological Survey Pacific Island Ecosystems Research Center P.O. Box 1319, Kaua'i Forest Bird Project Kekaha, HI 96752 GREGORY SPENCER P.O. Box 6122 Hilo, HI 96720 WILLIAM W. M. STEINER U.S. Geological Survey Pacific Island Ecosystems Research Center 3190 Maile Way Honolulu, HI 96822 CHERYL L. TARR Department of Biology and Institute of Molecular Evolutionary Genetics 208 Erwin W. Mueller Laboratory Pennsylvania State University University Park, PA 16802 ERIC A. TILLMAN Department of Wildlife Ecology & Conservation EO. Box 110430 University of Florida Gainesville, FL 32611-0430 JOHN TURNER The Peregrine Fund Keauhou Bird Conservation Center EO. Box 39 Volcano, HI 96785 Etak J. TWEED U.S. Geological Survey Pacific Island Ecosystems Research Center EO. Box 44, Building 344 Hawaii National Park, HI 96718 MIKLOS D.E UDVARDY (DECEASED) California State University, Sacramento EPac A. VANDERWERF University of Hawaii at Manoa Department of Zoology Edmonson Hall, 2538 The Mall Honolulu, HI 96822 ELLEN M. VANGELDER U.S. Geological Survey Halemkala National Park Box 369 Mmkawao, HI 96768 CHARLES VAN RIPER III U.S. Geological Survey Forest and Rangeland Ecosystem Science Center Colorado Plateau Field Station P.O. Box 5614 Northern Arizona University Flagstaff, AZ 86011 BETHANY L. WOODWORTH U.S. Geological Survey Pacific Island Ecosystems Research Center EO. Box 44, Building 344 Hawaii National Park, HI 96718 FRIEDERIKE WOOG Libanonstr. 66 70184 Stuttgart Germany Studies in Avian Biology No. 22:1-12, 2001. INTRODUCTION J. MICHAEL SCOTT, SHEILA CONANT, AND CHARLES VAN RIPER, III Hawai'i, a string of high and low islands stretching 1,900 km across the Central Pacific, has long captured the imagination of ornitholo- gists. The Hawaiian Islands are the most isolated archipelago in the world, and as a result, were one of the last places on the planet to be popu- lated (Fig. 1). The islands range from 25 million year-old Kure, at the extreme northwest end of the archipelago, to Hawai'i, the largest, south- ernmost, and the youngest island at less than 1 million years old (Fig. 2; Stearns 1966, Carson and Clague 1995). The climate varies dramati- cally from arid, tropical seashores receiving less than 26 cm (10 in) of precipitation on the lee- ward slopes of the main islands, to the windward peaks of the Alaka'i Swamp on Kaua'i, where it is not uncommon for torrential rains to drop 52 cm (20 in) in a day, or to record 1,152 cm (450 in) in a single year. The tropical lowland areas contrast dramatically with the high alti- tude, alpine ecosystems, and stone deserts, where it freezes every night. The landscape is as varied as it is dynamic. The tropical environ- ments at sea level contrast dramatically with the snow capped peaks of Mauna Loa and Mauna Kea, which reach more than 4,000 m in height above sea level and more than 9,000 m from their base in the ocean from which they were born (Stearns 1966, Carson and Clague 1995). The Hawaiian archipelago is extremely dynam- ic, with Loihi Seamount, an incipient island, presently going through the birthing process at a depth of 950 meters 30 km off the southern coast of Hawai'i (Carson and Clague 1995). Polynesians first reached the Hawaiian archi- pelago about 500 AD, and Europeans not until Captain James Cook's third voyage of discovery in 1778. With a little imagination and use of early voyagers' and naturalists' notes, one can create in the mind's eye a pre-Polynesian Ha- wai'i (Rothschild 1893-1900, Henshaw 1902a; Kirch 1982a, 1985). In these presettlement is- lands, millions of seabirds nested not only on offshore islets, isolated cliff faces, and barren subalpine areas where they are found today, but on the beaches and in adjacent forests, bringing tons of nitrates and phosphates from the sea. The transport of nutrients from marine environments by birds has significant impact on terrestrial en- vironments, resulting in increased plant growth and increases in those species that depend on plants for habitat and food (Polls and Hurd 1996, Ryan and Watkins 1989; Anderson and Polis 1998, 1999). As one moved inland, nu- merous species of geese, including ten that we know were flightless, grazed in the open grass- lands. The forests must have been alive with various species of Hawaiian honeyeaters, honey- creepers, owls, and hawks, flightless species (such as rail and ibis), and a variety of large- billed finches. The dawn song chorus of this ghost avifauna will never again be heard, but one can dream. Captain Cook's third voyage of discovery did not contribute greatly to our ornithological knowledge of the islands. Only 11 species and subspecies were ,described based on specimens collected during Cook's voyage, all from Kaua'i and Hawai'i (Medway 1981). The first compre- hensive characterizations of Hawaiian birds were the almost simultaneous publications by Rothschild (1893-1900) and Wilson and Evans (1890-1899). These detailed descriptions of Ha- waiian birds were augmented by the careful doc- umentation of the natural history and ecology of these birds by Henshaw (1902a,b) and Perkins (1893, 1901, 1903). These works established a foundation from which all current Hawaiian or- nithology is measured. In this monograph, we hope to provide another milestone of informa- tion on the avifauna of the Hawaiian Islands and the surrounding Pacific area, from which during the next century ornithologists might measure future changes in this avifauna. And most cer- tainly there will be changes. Historical changes to the Hawaiian avifauna started early, and only 100 years after Cook's exploration of the islands there were reports of species that had apparently gone extinct (Perkins 1903). At the turn of the century, R. C. L. Per- kins (as cited in Munro 1944:69) wrote: "When I first arrived in Kona, the Great Ohia trees, at an elevation of 2,500 feet, were a mass of bloom and each of them was literally alive with hordes of Crimson 'Apapane and Scarlet 'I'iwi; while continually crossing from the top of one great tree to another, the ''6 could be seen on the wing sometimes six or eight at a time .... The 'Amakihi was nu- merous in the same trees but less conspicuous and occasionally one of the long billed Hem- ignathus. Feeding on the fruit of the Ieie could be seen the Hawaiian Crow commonly and the 'O'fl in great abundance. The picture of this noisy, active, and often quarrelsome assembly 2 STUDIES IN AVIAN BIOLOGY NO. 22 FIGURE 1. The Hawaiian archipelago and other major islands in the Pacific Ocean. Kure Atol o .Midway Islands  Pearl and Hermes Reef Lisianski Island Laysan ß Island Maro Reef Gardner Pinnacles Tern Island La Perouse Pinnacle . French Frigate Shoal., ß Necker Island , Nihoa Kaua'i Ni'ihau  Kaula-  Lana'i '==- Maul Kaho'olawe '-- J Hawai'i  The Hawaiian Islands 100 200 300 Statute Miles FIGURE 2. The Hawaiian Islandsß Detailed maps of individual islands can be found in the chapters that follow, INTRODUCTION--Scott et al. 3 of birds, many of them brilliant colors, was one never to be forgotten. After the flowering of the Ohia was over, the great gathering nat- urally dispersed, but even then the bird pop- ulation was very great." By 1930 however, things had changed greatly when Munro (1944:68) stated: "Since civilization came to the Hawaiian Is- lands, the experience of the native perching birds has been tragic. My conclusions after the survey (1936-1937) were that 25 species have a fair chance of survival, while 30 species were gone or likely to become extinct." Today native birds are almost absent from the remaining lowland forests of Kona. In their place is an eclectic group of alien species, the result of a large number of planned and un- planned releases (see Moulton et al. this vol- ume). Today, only the 'Elepaio (Chasiempis sandwichensis), Hawai'i 'Amakihi (Hemigna- thus virens), 'I'iwi (Vestiaria coccinea), and 'Apapane (Himatione sanguinea) can be seen re- liably, and these not in all areas. The large-billed finches, honeyeaters (species once ubiquitous), 'O'O (Psittirostra psittacea), and 'Oma'o (Myadestes obscurus) are gone, while the 'ke- pa (Loxops coccineus), 'Akiap61'au (Hemig- nathus munroi), and Hawai'i Creeper (Oreomys- tis bairdi) occur in vanishingly small numbers in fewer than five isolated pockets of native for- est. At this writing, the number of free-flying 'Alal (Corvus hawaiiensis) can be counted on one hand. The true magnitude of these losses would, however, not be known until the pioneering re- search of the husband-and-wife team of Storrs Olson and Helen James (James and Olson 1991, Olson and James 1991). They documented the extinction of at least 50% of the Hawaiian avi- fauna prior to the first use of the Linnean System to describe a Hawaiian species. One hundred nine endemic species are known to have oc- curred in the Hawaiian Islands, 35 of which (32%) are still extant. Nineteen additional taxa were extant in the 18 th century, and 55 (50%) are known only from the fossil and subfbssil rec- ord (Table 1). Reasons for losses of many Hawaiian bird species have been well documented, including the destruction of habitat (Cuddihy and Stone 1990) and taking of birds (van Riper and van Riper 1982, Banko et al. this volume, Hu et al. this volume, van Riper and Scott this volume), predatory mammals (Tomich 1969, Kramer 1971, Atkinson 1977), introduced birds (Schwartz and Schwartz 1949, Lewin 1971, Lewin and Lewin 1984, Mountainspring and TABLE 1. BIRDS KNOWN FROM FOSSIL RECORDS OR KNOWN TO BREED IN THE HAWAIIAN ISLANDS Group dan- Popula- Species known ger- tions Spe- ed __ Histor- cies spe- 51 Fossil ic extant cies <50 500 Seabirds 1 22 22 2 2 1 Herons 0 1 0 0 0 0 Ibises 2 0 0 0 0 0 Waterfowl 10-11 3 3 3 0 0 Hawks 2 1 I 1 0 0 Rails 10 4 2 2 0 0 Stilts 0 I 1 1 0 0 Owls 4 1 1 0 0 0 Crows 3 1 1 1 1 0 Honeyeaters 0 6 1 1 1 0 Oldworld Flycatch- ers 0 1 1 0 0 0 Oldworld Warblers 0 1 1 1 0 0 Hawaiian Thrushes 0 6 2 1 0 1 Honeycreepers 13 31 20 9 4 3 Notes: Information on fossil birds includes only those records assigned species status (James and Olson 1991, Olson and James 1991 ). Additional species are being described. Historical status is based on several sources (Scott et al. 1989, Stone 1989, and Pyle 1997). In the last 11 years three species have become extinct: Kaua'i '6'6 (Moho broccatus), Kama'o (Myadestes myadestinus), and Oloma'o (Myadestes lanaiensis) based on the standard of extinct until proven extant (Diamond 1987). The 'l'iwi (Vestiaria coccinea) is declining in numbers and is disappearing froIll areas formerly occupied. The numbers of two other species have de- creased to less than 50 individuals. Species with less than 50 and 500 censused individuals are provided as indicators of jeopardy. The effective population size lbr these species is unknown but likely to be one-half to one-quarter censused population size (Primack 1993). For the 29 species listed by the U.S. Fish and Wildlife Service as endangered, 8 continue to decline, 6 are of unknown status, and 15 are stable in numbers (USFWS 1996a). Scott 1985), and diseases (Warner 1968, van Riper et al. 1986, Atkinson et al. 1993a,b,c, 1995; Jarvi et al. this volume, Shehata et al. this volume). The combined effect of these losses has been summarized in papers by Scott et al. (1986), van Riper and van Riper (1985), Ralph and van Riper (1985), Freed et al. (1993), and van Riper and Scott (this volume). While many species have succumbed to ex- tinction (Table 1), major steps have been taken recently to save Hawai'i's endangered species. The U.S. Fish and Wildlife Service has estab- lished two national wildlife refuges (Hakalau Forest and Kona Forest National Wildlife Ref- uges) on the island of Hawai'i with a primary objective of protecting endangered forest birds. Combined, these preserves total nearly 16,194 ha (40,000 acres). The National Park Service has eliminated goats (Capra hircus) and sheep (Ovis aries) from Hawai'i Volcanoes and Haleakala National Parks. In addition, large acreages are now pig- (Sus scrofa) free in that park. Similar- 4 STUDIES IN AVIAN BIOLOGY TABLE 2. CHECKLIST OF THE BIRDS OF HAWAII NO. 22 Symbols for status R - Resident native species; normal does not leave islands: Re - Resident, endemic species, not extinct; Rx - Resident, endemic species, presumed extinct; Res - Resident; indigenous species, subspecies is endemic; Hawaiian; Ri - Resident; indigenous species, Hawaiian form is not endemic. A - Alien introduced species; resident; normally does not leave the islands: Al - Alien; long established and breeding since before 1940; An - Alien, new introduced since 1950; apparently established; Ax - Alien; formerly long established and breeding for more than 25 years, but now no longer present in Hawaii. E (or T) immediately preceding the genus name designates a species or subspecies currently listed as Endangered (or Threatened) on the Federal List of Endangered species. B - Breeding species in Hawaii, native, most individuals leave Hawaii when not breeding: Bo - Breeder, species breeds only in Hawaii; Bes - Breeder, species also breeds elsewhere; Hawaiian subspecies breeds only in Hawaii; Bi - Breeder, Hawaiian form also breeds elsewhere. V - Visitor species, breeds elsewhere, occurs in Hawaii when not breeding: Vc - Visitor, common migrant to Hawaii; Vr - Visitor, regular migrant to Hawaii in small numbers; Vo - Visitor, occasional to frequent migrant to Hawaii; Vs - Visitor, accidental straggler to Hawaii; Vd = Visitor, accidental straggler to Hawaii, recorded in Hawaii only as dead remains. Common name Scientific name Status GREBES PODICIPEDIDAE Pied-billed Grebe Podilymbus podiceps Ri Horned Grebe Podiceps auritus Vs Red-necked Grebe Podiceps grisegena Vs Eared Grebe Podiceps nigricollis Vs ALBATROSSES DIOMEDEIDAE Laysan Albatross Phoebastria immutabilis Bi Black-footed Albatross Phoebastria nigripes Bi Short-tailed Albatross E-Phoebastria albatrus Vo PETRELS, SHEARWATERS PROCELLARIIDAE Northern Fulmar Fulmarus glacialis Vo Kermadec Petrel Pterodroma neglecta Vs Herald Petrel Pterodroma arminjoniana Vs Murphy's Petrel Pterodroma ultima Vs Mottled Petrel Pterodroma inexpectata Vo Juan Fernandez Petrel Pterodroma externa Vo (Hawaiian Petrel)--Dark-rumped Petrel E-Pterodroma phaeopygia Res sandwichensis White-necked Petrel Pterodroma cervicalis Vo Bonin Petrel Pterodroma hypoleuca Bi Black-winged Petrel Pterodroma nigripennis Vo Cook's Petrel Pterodroma cookii Vs Stejneger's Petrel Pterodroma longirostris Vd Bulwer's Petrel Bulweria bulwerii Bi Jouanin's Petrel Bulweria fallax Vs Streaked Shearwater Calonectris leucomelas Vs Flesh-footed Shearwater Puffinus carneipes Vo Wedge-tailed Shearwater Puffinus pacificus B i chlororhynchus (New Zealand Shearwater)Buller's Puffinus bulleri Vs Shearwater Sooty Shearwater Short-tailed Shearwater Christmas Shearwater (Newell's Shearwater) Townsend's Shearwater Little Shearwater STORM-PETRELS Wilson's Storm-Petrel Fork-tailed Storm-Petrel Leach's Storm-Petrel (Hawaiian or Harcourt's Storm-Petrel)-- Band-rumped Storm-Petrel (Sooty Storm-Petrel)--Tristram's Storm-Pe- trel TROPICBIRDS White-tailed Tropicbird Red-billed Tropicbird Red-tailed Tropicbird Puffinus griseus Vr Puffinus tenuirostris Vo Puffinus nativitatis Bi T-Puffinus auricularis newelli Be Puffinus assimilis Vs HYDROBATIDAE Oceanires oceanicus Vs Oceanodroma furcata Vs Oceanodroma leucorhoa Vr ' Oceanodroma castro Bi Oceanodroma tristrami Bi PHAETHONTIDAE Phaethon lepturus Ri Phaethon aethereus Vs Phaethon rubricauda Bi rothschiMi INTRODUCTION--Scott et al. 5 TABLE 2. CONTINUED Conlmon name Scientific nanle Status BOOBIES SULIDAE (Blue-faced Booby)--Masked Booby Sula dactylatra personata Ri Brown Booby Sula leucogaster plotus Ri Red-footed Booby Sula sula rubripes Ri CORMORANTS PHALACROCORACIDAE Pelagic Cormorant Phalacrocorax pelagicus Vs FRIGATEBIRDS FREGATIDA E Great Frigatebird Fregata minor palmerstoni Ri Lesser Frigatebird Frigata ariel Vs HERONS, EGRETS ARDEIDAE Great Blue Heron Ardea herodias Vs Great Egret Ardea alba Vs Snowy Egret Egretta thula Vs Little Blue Heron Egretta caerulea Vo Cattle Egret Bubulcus ibis An (Green-backed Heron)--Green Heron Butorides virescens Vs Black-crowned Night-Heron Nycticorax nycticorax hoactli Ri IBISES THRESKIORNITHIDAE White-faced Ibis Plegadis chihi Vs GEESE, DUCKS ANATIDAE Fulvous Whistling-Duck Dendrocygna bicolor Ri (White-fronted Goose)--Greater White- Anser albifrons Vs fronted Goose Emperor Goose Chen canagica Vo Snow Goose Chen caerulescens Vs Canada Goose Branta canadensis Vo (Nn)--Hawaiian Goose E-Branta sandvicensis Re Brant Branta bernicla Vo (Whistling Swan)--Tundra Swan Cygnus columbianus Vs Gadwall Anas strepera Vs (European Wigeon)--Eurasian Wigeon Anas penelope Vs American Wigeon Anas americana Vr Mallard Arias platyrhynchos A1, Vo (Koloa)--Hawaiian Duck E-Arias wyvilliana Re Laysan Duck E-Arias laysanensis Re Blue-winged Teal Arias discors Vo Cinnamon Teal Arias cyanoptera Vs Northern Shoveler Arias clypeata Vc Northern Pintail Anas acuta Vc Garganey Arias querquedula Vo Green-winged Teal Arias crecca Vr Canvasback Aythya valisineria Vs Redhead Aythya americana Vs Common Pochard Aythya ferina Vs Ring-necked Duck Aythya collaris Vo Tufted Duck Aythya fuligula Vs Greater Scaup Aythya marila Vo Lesser Scaup Aythya affinis Vr Harlequin Duck Histrionicus histrionicus Vs Surf Scorer Melanitta perspicillata Vs Black Scoter Melanitta nigra Vs Long-tailed Duck Clangula hyemalis Vs Bufflehead Bucephala albeola Vo Common Goldeneye Bucephala clangula Vs Hooded Merganser Lophodytes cucullatus Vs Common Merganser Mergus merganser Vs Red-breasted Merganser Mergus serratot Vs Ruddy Duck Oxyura jamaicensis Vs HAWKS, EAGLES ACCIPITRIDAE Osprey Pandion haliaetus Vo Black Kite Milvus migrans Vs Steller's Sea-Eagle Haliaeetus pelagicus Vs 6 TABLE 2. CONTINUED STUDIES IN AVIAN BIOLOGY NO. 22 Common name Scientific nalne Status Northern Harrier Circus cyaneus Vs Gray Frog-Hawk Accipiter soloensis Vs ('Io)--Hawaiian Hawk E-Buteo solitarius Re Rough-legged Hawk Buteo lagopus Vs Golden Eagle Aquila chrysaetos Vs FALCONS FALCONIDAE Merlin Falco columbarius Vs Peregrine Falcon E-Falco peregrinus Vo FRANCOLINS, OLD WORLD QUAIL, TURKEY PHASIANIDAE Chukar Alectoris chukar A1 Gray Francolin Francolinus pondicerianus An Black Francolin Francolinus?ancolinus An Erckel's Francolin Francolinus erckelii An Japanese Quail Coturnix japonica A1 Red Junglefowl Gallus gallus A1 Kalij Pheasant (Green Pheasant, Common Pheasant)-- Lophura leucomelanos An Ring-necked Pheasant Phasianus colchicus A1 Common Peafowl Pavo cristatus A1 Wild Turkey Meleagris gallopavo A1 NEW WORLD QUAIL ODONTOPHORIDAE California Quail Callipepla californica A1 Gambel's Quail Callipepla gambelii A1 RAILS, GALLINULES, COOTS RALLIDAE Laysan Rail Porzana palmeri Rx Hawaiian Rail Porzana sandwichensis Rx (Hawaiian Gallinule)--Common E-Gallinula chloropus Res Moorhen sandvicensis (American Coot)--Hawaiian Coot E-Fulica alai Res American Coot Fulica americana Vs CRANES GR UIDAE Sandhill Crane Grus canadensis Vs PLOVERS CHARADRIIDAE (Gray Plover)--Black-bellied Plover Pluvialis squatarola Vr (Lesser or American Golden-Plover)-- Pluvialis Mva Vc Pacific Golden-Plover Mongolian Plover Charadrius mongolus Vs Common Ringed Plover Charadrius hiaticula Vs Semipalmated Plover Charadrius semipalmatus Vo Killdeer Charadrius vociferus Vs Eurasian Dotterel Charadrius morinellus Vs STILTS RECURVIROSTRIDAE (Hawaiian Stilt)--Black-necked Stilt E-Himantopus mexicanus knud- Res seni SANDPIPERS, WADERS SCOLOPACIDAE Greater Yellowlegs Tringa melanoleuca Vs Lesser Yellowlegs Tringa flavipes Vr Wood Sandpiper Tringa glareola Vs Solitary Sandpiper Tringa solitaria Vs Willet Catoptrophorus semipalmatus Vs Wandering Tattler Heteroscelus incanus Vc (Siberian Tattler, Polynesian Tattler)-- Heteroscelus brevipes Vs Gray-tailed Tattler Spotted Sandpiper Actiris macularia Vs Whimbrel Numenius phaeopus Vs Bristle-thighed Curlew Numenius tahitiensis Vr Far Eastern Curlew Numenius madagascariensis Vs Hudsonian Godwit Limosa haemastica Vs Bar-tailed Godwit Limosa lapponica Vo Marbled Godwit Limosa kdoa Vs Ruddy Tumstone Arenaria interpres Vc Red Knot Calidris canutus Vs Sanderling Calidris alba Vo INTRODUCTION--Scott et al. 7 TABLE 2. CONTINUED Common name Scientific name Status Semipalmated Sandpiper Calidris pusilla Vs Western Sandpiper Calidris mauri Vo Red-necked Stint Calidris ruficollis Vs Little Stint Calidris minuta Vs Long-toed Stint Calidris subminuta Vs Least Sandpiper Calidris minutilla Vo Baird's Sandpiper Calidris bairdii Vs Pectoral Sandpiper Calidris melanotos Vr Sharp-tailed Sandpiper Calidris acuminata Vr Dunlin Calidris alpina Vr Curlew Sandpiper Calidrisferruginea Vs Buff-breasted Sandpiper Tryngites subruficollis Vs Ruff Philomachus pugnax Vo Short-billed Dowitcher Limnodromus griseus Vo Long-billed Dowitcher Limnodromus scolopaceus Vr Common Snipe Gallinago gallinago Vo Pin-tailed Snipe Gallinago stenura Vs Wilson's Phalarope Phalaropus tricolor Vo Red-necked Phalarope Phalaropus lobatus Vs Red Phalarope Phalaropus fulicaria Vs JAEGERS, GULLS, TERNS, NODDIES LARIDAE South Polar Skua Stercorarius maccormicki Vs Pomarine Jaeger Stercorarius pomarinus Vr Parasitic Jaeger Stercorarius parasiticus Vs Long-tailed Jaeger Stercorarius longicaudus Vs Laughing Gull Larus atricilla Vo Franklin's Gull Larus pipixcan Vs Black-headed Gull Larus ridibundus Vs Bonaparte's Gull Larus philadelphia Vo Mew Gull Larus canus Vs Ring-billed Gull Larus delawarensis Vo California Gull Larus califbrnicus Vs Herring Gull Larus argentatus Vo Slaty-backed Gull Larus schistisagus Vs Western Gull Larus occidentalis Vs Glaucous-winged Gull Larus glaucescens Vo Glaucous Gull Larus hyperboreus Vs Black-legged Kittiwake Rissa tridactyla Vs Gull-billed Tern Sterna nilotica Vs Caspian Tern Sterna caaTia Vs Great Crested Tern Sterna bergii Vs Sandwich Tern Sterna sandvicensis Vs Common Tern Sterna hitundo Vs Arctic Tern Sterna paradisaea Vo Little Tern Sterna albifrons Vs Least Tern Sterna antillarum Vo Gray-backed Tern Sterna lunata Bi Sooty Tern Sterna fuscata oahuensis Bi Whiskered Tern Chlidonias hybridus Vs Black Tern Chlidonias niger Vs (Common Noddy)--Brown Noddy Anous stolidus pileatus Ri (Hawaiian Noddy, White-capped Noddy)-- Anous minutus melanogenys Res Black Noddy Blue-gray Noddy Procelsterna cerulea saxarilis Ri (Common Fairy-Tern, Fairy Tern)-- Gygis alba rothschildi Ri White Tern AUKLETS, PUFFINS ALCIDAE Cassin's Auklet Ptychoramphus aleuticus Vs Parakeet Auklet Aethia psittacula Vd Horned Puffin Fratercula corniculata Vs Tufted Puffin Fratercula cirrhata Vd 8 TABLE 2. CONTINUED STUDIES IN AVIAN BIOLOGY NO. 22 Common name Scientific name Status SANDGROUSE Chestnut-bellied Sandgrouse DOVES Rock Dove (Chinese Dove, Lace-necked Dove)- Spotted Dove (Barred Dove)--Zebra Dove Mourning Dove PARAKEETS Rose-ringed Parakeet CUCKOOS Common Cuckoo Yellow-billed Cuckoo BARN OWLS Barn Owl TYPICAL OWLS (Hawaiian Owl)--Short-earned Owl NIGHTHAWKS Common Nighthawk SWIFTLETS (Uniform, Island or Gray SwiftleO--Guam Switflet KINGFISHERS Belted Kingfisher HONEYEATERS 'O'o'a'a--Kaua'i 'O 'O O'ahu 'O'o (Moloka'i 'O'o)--Bishop's 'O'O Hawai'i 'O'O Kioea CROWS ('Alala)--Hawaiian Crow MONARCH FLYCATCHERS 'Elepaio { Kaua'i 'Elepaio } -- { O'ahu 'Elepaio }-- { Hawai 'i 'Elepaio }- LARKS (Eurasian Skylark)--Sky Lark SWALLOWS Barn Swallow TITS (Japanese Tit, Yamagara)--Varied Tit BULBULS Red-vented Bulbul Red-whiskered Bulbul OLD WORLD WARBLERS (Uguisu)--Japanese Bush-Warbler Millerbird { Laysan Millerbird }-- {Nihoa Millerbird}- THRUSHES, SOLITAIRES (Shama Thmsh)--White-rumped Shama (Large Kaua'i Thrush)--Kg. ma'o (O'ahu Thrush)--'maui Oloma'o {(Moloka'i Thrush)Moloka'i Olo- ma'o}-- PTEROCLIDIDAE Pterocles exustus An COLUMBIDAE Columba livia A1 Streptopelia chinensis AI Geopelia striata AI Zenaida macroura An PSITTA CIDAE Psittacula krameri An CUCULIDAE Cuculus canorus Vs Coccyzus americanus Vs TYTONIDAE Tyto alba An STRIGIDAE Asio fiammeus sandwichensis Res CAPRIMULGIDAE Chordeiles minor Vs APODIDAE Aerodramus bartschi An ALCEDINIDAE Ceryle alcyon Vs MELIPHA GIDAE E-Moho braccatus Re Moho apicalis Rx Moho bishopi Rx Moho nobdis Rx Chaetoptila angustipluma Rx CORVIDAE E-Corvus hawaiiensis Re MONARCHIDAE Chasiempis sandwichensis C. s. sclateri Re C. s. ibidis Re C. s. sandwichensis, ridgwayi, Re bryani ALA UDIDAE Alauda arvensis AI, Vs HIRUNDINIDAE Hirundo rustica Vs PARIDAE Parus varius Ax PYCNONOTIDAE Pycnonotus cafer An Pycnonotus jocosus An SYLVIIDAE Cettia diphone A1 Acrocephalus familiaris A. f familiaris Rx E-A. f kingi Re TURDIDAE Copsychus malabaricus AI E-Myadestes myadestinus Re Myadestes woahensis Rx Myadestes lanaiensis E-M.l. rutha Re INTRODUCTION--Scott et al. 9 TABLE 2. CONTINUED Conmon name Scientific name Status { (Lgna'i Thrush)--Lgna'i Oloma'o}-- (Hawai'i Thrush)--'0ma'o (Small Kaua'i Thrush)--Puaiohi BABBLERS Greater Necklaced Laughing-thrush Gray-sided Laughing-thrush (Melodious Laughing-thrush, Chinese Thrush)--Hwamei (Pekin Nightingale, Japanese Hill-robin)- Red-billed Leiothrix WHITE-EYES (Mejiro)--Japanese White-eye MOCKINGBIRDS Northern Mockingbird STARLINGS, MYNAS European Starling Common Myna PIPITS Olive-backed Pipit Red-throated Pipit American Pipit EMBERIZIDS Yellow-faced Grassquit Saffron Finch (Brazilian Cardinal)Red-crested Cardinal Yellow-billed Cardinal Savannah Sparrow Snow Bunting CARDINALS (American or Kentucky Cardinal)-- Northern Cardinal MEADOWLARKS, GRACKLES Western Meadowlark Great-tailed Grackle FINCHES CARDUELINE FINCHES (Linnet)--House Finch Common Redpoll (Green Singing-Finch)--Yellow- fronted Canary (Canary)--Common Canary HAWAIIAN HONEYCREEPERS FINCH-BILLED HONEYCREEPERS Laysan Finch Nihoa Finch Lgna'i Hookbill Palila Lesser Koa-Finch Greater Koa-Finch (Grosbeak Finch)--Kona Grosbeak Maui Parrotbill SLENDERBILLED HONEYCREEPERS Hawai'i 'Amakihi { Hawai'i 'Amakihi }-- { Maui 'Amakihi }-- O'ahu 'Amakihi Kaua'i 'Amakihi M. l. lanaiensis Rx Myadestes obscurus Re E-Myadestes palmeri Re T1MALllDAE Garrulax pectoralis A1 Garrulax caerulatus A1 Garrulax canorus A1 Leiothrix lutea A1 ZOSTEROP1DAE Zosterops japonicus A1 MIMIDAE Mimus polyglottos AI STURN1DAE Sturnus vulgaris Vs Acridotheres tristis A1 MOTA C1LL1DAE Antbus hodgsoni Vs Antbus cervinus Vs Antbus rubescerts Vs EMBER1Z1DAE Tiaris olivacea An Sicalis fiaveola An Paroaria coronata A1 Paroaria capitata A1 Passerculus sandwichensis Vs Plectrophenax nivalis Vs CARD1NAL1DAE Cardinalis cardinalis A1 1CTER1DAE Sturnella neglecta A1 Quiscalus mexicanus Vs FRINGILLIDAE CARDUELINAE (subfamily) Carpodacus mexicanus A1 Carduelis fiammea Vs Serinus mozambicus An Serinus canaria A1 DREPANIDINAE (subfamily) PS1TT1ROSTR1N1 (tribe) E- Telespiza cantans Re E- Telespiza ultima Re E-Psittirostra psittacea Re Dysmorodrepanis munroi Rx E-Loxioides bailleui Re Rhodacanthis fiaviceps Rx Rhodacanthis palmeri Rx Chloridops kona Rx E-Pseudonestor xanthoph3;s Re HEM1GNATH1N1 (tribe) Hemignathus virens H. v. virens Re H. v. wilsoni Re Hemignathus fiavus Re Hemignathus kauaiensis Re 10 TABLE 2. CONTINUED STUDIES IN AVIAN BIOLOGY NO. 22 Common name Scientific name Status (Lesser 'Amakihi)--'Anianiau (Green Solitaire)--Greater 'Amakihi Lesser 'Akialoa Greater 'Akialoa { Kaua'i 'Akialoa }-- { O 'ahu 'Akialoa }-- {Lana'i 'Akialoa}-- Nukupu'u {Kaua'i Nukupu'u}-- {O'ahu Nukupu'u }-- {Maui Nukupu'u }-- 'Akiap61g'au (Kaua'i Creeper)--'Akikiki (Olive Green Creeper)--Hawai'i Creeper (O'ahu Creeper)--O'ahu 'Alauahio (Moloka'i Creeper)--Kgkgwahie (Maui Creeper)--Maui 'Alauahio { Maui 'Alauahio }-- { Lana'i 'Alauahio }-- (Kaua'i kepa)--'Akeke'e 'kepa { O'ahu ',kepa }-- {Maui ',kepa }-- { Hawai'i 'kepa }- RED AND BLACK HONEYCREEPERS Ula-'ai-hawane 'I'iwi Hawai'i Mamo (Perkins Mamo)--Black Mamo (Crested Honeycreeper)--'kohekohe 'Apapane {Laysan Honeycreeper }- { 'Apapane }-- Po'ouli OLD WORLD SPARROWS (English Sparrow)--House Sparrow WAXBILLS, MANNIKINS Red-cheeked Cordonbleu Lavender Waxbill Orange-cheeked Waxbill (Red-eared Waxbill)--Black-rumped Waxbill Common Waxbill (Strawberry Finch, Red Munia)--Red Avadavat African Silverbill (Ricebird, Spotted Munia)--Nutmeg Mannikin Tricolored Munia Chestnut Munia Java Sparrow Hemignathus parvus Re Hemignathus sagittirostris Rx Hemignathus obscurus Rx Hemignathus ellisianus H. e. procerus Rx H. e. ellisianus Rx H. e. lanaiensis Rx Hemignathus lucidus E- H. l. hanapepe Re H. I. lucidus Rx E-H. 1. qffinis Re E-Hemignathus munroi Re Oreomystis bairdi Re E-Oreomystis maria Re E-Paroreomyza maculata Re Paroreomyza fiammea Rx Paroreomyza montana P.m. newtoni Re P.m. montana Rx Loxops caeruleirostris Re Loxops coccineus L. c. wolstenholmei Rx E-L. c. ochraceus Re E-L. c. coccineus Re DREPANIDINI (tribe) Ciridops anna Rx Vestiaria coccinea Re Drepanis pac(fica Rx Drepanis funerea Rx E-Palmeria dolei Re Himatione sanguinea H. s. freethii Rx H. s. sanguinea Re E-Melamprosops phaeosoma Re PASSERIDAE Passer domesticux AI ESTRILDIDAE Uraeginthus bengalus An Estrilda caerulescens An Estrilda melpoda An Estrilda troglodytes An Estrilda astrild An Amandava amandava A1 Lonchura cantans An Lonchura punctulata A1 Lonchura malacca A1 Lonchura atricapilla An Padda oryivora An Note,: This table is modified from Robert Pyle's 1997 checklist of the birds of Hawaii. In all cases we have deferred to the American Ornithologist Union's 1998 Checklist of North American birds and the 42nd Supplement to the Checklist (AOU 2000) for common and scientific names and sequence of families and species, We have added macrons. diacritical marks, and glottal stops to all common names as indicated by Pyle (1997). Subspecies of resident species known to occur in the islands are indicated in brackets. Common names in parentheses are those commonly used in Hawai'i but not accepted by the AOU Check-list. INTRODUCTIONsScott et al. 11 ly, The Nature Conservancy has pursued an ag- gressive control program for alien species that threaten the viability of native species popula- tions and the ecological integrity of native Ha- waiian ecosystems and established several large biological reserves. While the Sierra Club, Na- tive Plant Society, Hawaii Audubon Society, and a number of state and federal agencies have all taken actions on behalf of Hawai'i's native flora and fauna, despite their efforts and extensive re- search efforts in the last 25 years (Banko et al. this volume, Steiner this volume), populations and species of native birds continue to be lost. Nonnative birds species comprise a large part of the current avifauna (Table 2). If there is to be any hope of retaining even a majority of the currently endangered and threat- ened native Hawaiian species, more aggressive efforts are needed to seriously reduce agents known to be detrimental to native species (Smith 1985, 1989; Cuddihy and Stone 1990, Stone 1989, Banko et al. this volume, Scott and van Riper this volume). Despite widespread docu- mentation of the impact of feral cats (Felis ca- tus) on birds (Eberhard 1954, van Aarde 1978, Jehl and Parkes 1982, Tomkins 1985, Churcher and Lawton 1987, van Reusenburg and Bester 1988, Bloomer and Besler 1992, Seto and Co- nant 1996, Athens 1997, Radunzel et al. 1997), there are currently no cat control programs in place for passerine species and only limited ef- forts on behalf of seabirds (Hodges and Nagata this volume). Likewise, while the impact of rats (Rattus spp.) on Hawai'i's avifauna has yet to be fully documented, Atkinson's (1977) corre- lational study was suggestive, as was the extinc- tion of five populations of native birds on Big South Cape Island in New Zealand shortly after the arrival of the roof rat (Rattus rattus; Atkin- son 1985). Studies in New Zealand (Atkinson and Bell 1973) and elsewhere have shown the strong positive response of native species when nonnative rats are eliminated (Radunzel et al. 1997). In spite of this evidence, predator control programs are rare and are not being implement- ed over areas large enough to elicit a population response by native species. The elimination of rats from Midway Island is an exception (R. Shallenberger, pers. comm.). In the absence of management activities to control or eliminate known causes of mortality to Hawaiian avifauna over areas comparable to the size of the distributional area of the threats, individuals will die, populations will be lost, and species will continue to go extinct. For some threats (e.g., predators, ungulates), known con- trol techniques (e.g., Taylor and Katahira 1988, Katahira et al. 1993) only need be applied at a scale that is meaningful (the distributional area of a population or species). For others, such as avian malaria and avian pox, new techniques such as genetic engineering of disease resistant birds and introduction of sterile male mosquitoes must be developed and applied. A first step to buy time and simultaneously to restore populations of other endemic Hawaiian species (plants and invertebrates) would be to restore the composition and structure of higher elevation xeric and mesic forest habitats on Maui and Hawai'i by eliminating alien animals and plants (e.g., rats, cats, ungulates, and foun- tain grass) from these areas. These recovered and restored habitats would act as refugia from avian diseases so prevalent at lower elevations. The idea for this book came during informal discussions at the 67 th annual meeting of the Cooper Ornithological Society in Hilo, Hawai'i, in April 1997. During that meeting there were 47 presentations on natural history, ecology and taxonomy of Hawaiian birds. We invited select- ed authors of those presentations to submit manuscripts for consideration in a peer reviewed book on the birds of HawaiT To fill gaps in topics covered we solicited eight additional manuscripts. There was a high degree of redun- dancy in references cited among authors. Be- cause of this we chose to create a combined lit- erature cited. Common and scientific names of birds follow the 7 th edition of the American Ornithologists Union Check-list (AOU 1998). Quentin Tom- ich's Mammals in Hawaii (Tomich 1986) was our reference for mammal names. For flowering plants we relied on Manual of the Flowering Plants of Hawai'i (Wagner et al. 1990 a,b). "Pronunciation of Hawaiian names is aided by the use of a reversed apostrophe ('), to indicate the glottal stop, a stopping of sound, as between the vowell sounds in oh-oh in English; and by macrons over vowels--a, , , o, fi--which de- note long stress. An asterisk preceding a place name indicates that pronunciation is uncertain" (Armstrong 1983:231). The orthography follows the revised and enlarged Hawaiian Dictionary (Pukui and Elbert 1986). For place names we followed the revised and enlarged Place Names of Hawaii (Pukui et al. 1976). When names could not be located there the spelling in the Atlas of Hawaii (Armstrong 1983) was followed. This monograph includes 35 papers, most of which were presented at the 67 th meeting of the Cooper Ornithological Society in Hilo, Hawai'i, in April 1997. Each paper has been peer re- viewed by the editors and at least one outside reviewer. We have grouped the 35 chapters in 12 STUDIES IN AVIAN BIOLOGY NO. 22 this book into six sections, each introduced with a historical review. Taken together, they report on the state of our knowledge concerning the Hawaiian avifauna at the end of the 20 th century. Hopefully, this synthesis volume will assist in some small way to help preserve the unique avi- fauna of Hawai'i and the Pacific islands so that future generations will be able to observe and hear some of the incredible sights and sounds that we have been privileged to experience dur- ing our short 'tour of duty' researching one of the most unique avifaunas on this planet. ACKNOWLEDGMENTS We thank Robert Pyle for permission to use a mod- ified version of his Check-List of Hawaiian Birds and for his comments on drafts of Table 2. Sue McMurray tracked manuscripts and correspondence to author's queries. Sue McMurray and Andrea Reese completed the onerous task of combining references from indi- vidual papers into a single combined Literature Cited. Steve Mosher found a second home in the library as he checked references against the original publications. Lenny Freed was instrumental in launching the idea for publishing manuscripts from the Hilo meeting of the Cooper Ornithological Society as an integrated monograph in STUDIES 1N AVlAN BIOLOGY and provided valuable comments on manuscripts. Melissa Madsen read all manuscripts for grammar and adherence to STUDIES 1N AVlAN BIOLOGY format, consistency with place names of Hawa'i, and correct usage of glottal stops, macrons, and diacritical marks in the spelling of Hawaiian words. Kathy Merk's unfailing commitment to completing this book was a huge morale booster; she tracked manuscripts, corresponded with authors, and made edits on manuscripts as needed. John Roten- berry was the epitome of what a professional editor should be; his insightful comments, rigorous attention to detail, and manner of conveying need for change made him a pleasure to work with. H. Douglas Pratt graciously provided a painting of the Hawai'i '0'8 for the frontespiece of this book, as well as the line draw- ings that precede each section. We thank Patrick Ching for the numerous drawings of native Hawaiian birds scattered throughout the text. Jack Jeffrey kindly pro- vided the photograph of an 'Anianiau feeding on a kanawao that graces the cover. Funds for publication of this book and administrative support were provided by the U.S. Geological Survey, Idaho Cooperative Fish and Wildlife Research Unit, and the Department of Fish and Wildlife, University of Idaho, Moscow, Ida- ho. To all these individuals a special mahalo nui loa (deep thanks) for all that you have done. Historical Perspectives Studies in Avian Biology No. 22:14, 2001. HISTORICAL PERSPECTIVES--INTRODUCTION CHARLES VAN RIPER, III, SHEILA CONANT, AND J. MICHAEL SCOTT The record of Hawai'i's avifauna is one of change; a change that is reflected in steadily di- minishing numbers of species and abundance (Pratt 1994). Our historical perspectives provide insights into how many species there were and some documentation of their distribution, but only minor insights into their abundance, with size, shape, and bill forms allowing vague in- ferences concerning niches occupied and re- sources exploited. Nothing is known of clutch sizes, population characteristics, or ecological interactions of extinct species. For these reasons, more than 50% of Hawai'i's bird species will always be a ghost avifauna. The history of ornithological exploration in Hawai'i is a legacy of missed opportunities, with the first extensive surveys of the avifauna com- ing 100 years after the discovery of the islands by Europeans in 1778 (Olson and James 1994a). Historically, recorded species are but a small fraction of what occurred in the islands prior to European colonization. Some species were sim- ply overlooked; the Po'ouli (Melarnprosops phaeosorna) was not discovered until 1972 (Cas- ey and Jacobi 1974). Olson and James (1991, James and Olson 1991) nearly doubled the known number of endemic species based on their descriptions of new species from fossil and subfossil remains. New discoveries of fossil spe- cies continue today. In the first chapter of this volume, Curnutt and Pimm estimate that the Pacific avifauna was composed of nearly 1,500 species, of which ap- proximately 240 survive. For example, they es- timate that there were 12 species of rails endem- ic to the Hawaiian Islands, versus the 7 currently described (Olson and James 1991; Table 2). In the second chapter, Michael Moulton and his co- authors document the introduction of 140 spe- cies in 14 different orders and ask, "Why do some introduced species succeed and others fail?" 14 Studies in Avian Biology No. 22:15-30, 2001. HOW MANY BIRD SPECIES IN HAWAI'I AND THE CENTRAL PACIFIC BEFORE FIRST CONTACT? JOHN CURNUTT AND STUART PIMM Abstract. Since European settlement, extinctions of Pacific island birds have been widespread and well documented. Subfossil evidence indicates that the Polynesians caused extinctions of an even greater magnitude. Estimating the prehuman Pacific avifauna is difficult because the existing fossil record is inevitably incomplete. We use the theoretical framework of island biogeography to make estimates of the numbers of endemic rails, parrots, pigeons and doves that existed in the Pacific before human contact. We formulate two sets of estimates for each taxon by assuming that: (1) endemism is defined as a distribution limited to a single island, and (2) endemism is a distribution limited to a single-island group. These two assumptions lead to different results (884 compared with 242 endemic species). We refine our predictions by applying topographical and disturbance parameters. Our best estimate is that 332 endemic species of the three taxa once existed in the Pacific, of which 210 are not accounted for in the paleontological and historical data. Applying this ratio of known to missing species for all landbirds, we estimate the original Pacific avifauna to be composed of less than 1,500 species, of which approximately 230 survive. Our estimate of the original Pacific avifauna falls be- tween two earlier conflicting predictions (800 and much greater than 2,000). Our predictions of the number of species missing on each type of island are testable. Our results can be used to focus research efforts on islands that are more likely to have held species of interest. Furthermore, our results can be interpreted to predict the risk of future extinctions that may result from habitat loss or rising sea levels. Key Words: biogeography; doves; extinctions; Pacific Islands; parrots; pigeons; rails; sea level; tsu- namis. The Hawaiian Islands form one of the largest and most diverse archipelagoes in the Pacific. As a group, they lead the world in numbers of his- torically extinct and currently endangered spe- cies of birds (King 1985). This dismal legacy, however, did not befall the Hawaiian Islands alone. Untold bird extinctions doubtlessly oc- curred across the Pacific over the four millennia since humans first set sail there. What was the magnitude of the loss of bird species in the Pa- cific? "The Pacific" denies an easy definition. De- fined in the context of human settlement over the last 4,000 years, we will consider 41 island groups (Fig. 1). They span the Hawaiian Islands in the northeast, west to the Marianas and Palau, southwest through Vanuatu, south to New Zea- land and east to Easter Island. Pratt et al.'s (1987) field guide covers all but Vanuata (for which see Bregulla 1992), New Zealand (see Falla et al. 1983), and Easter (which has no ex- tant landbirds). There are roughly 240 extant native species of landbirds in this region (Falla et al. 1983, Pratt et al. 1987, Bregulla 1992). The largest families are Pachycephalidae (whistlers; 40 spp.), Columbidae (pigeons and doves; 34 spp.), Muscicapidae (Old World flycatchers; 28 spp.), Rallidae (rails; 21 spp.), Psittacidae (parrots; 19 spp.), and Fringillidae (Hawaiian honeycreepers; 19 spp.). To the above number of species we must add those that we know once existed but are now known only through historical records and fos- sils. Among the islands of the Pacific, the many vertebrate extinctions that occurred since the sixteenth century subsequent to the arrival of European explorers are well documented. For example, Diamond (1984) reported that, since 1600, Micronesia and Polynesia suffered rough- ly 100 bird species extinctions. The forces re- sponsible for the loss of these species were the same as those that operate today, primarily hab- itat loss and the introduction of exotic species (Steadman 1997a,b). A much greater extinction event preceded the arrival of Europeans and was concurrent with the first human contact (Stead- man 1997a,b). Beginning about 4,000 years ago with Melanesia and Micronesia and ending about 1,500 years ago with Hawai'i, Easter Is- land, and New Zealand, humanity brought the last habitable places on Earth under its domain (Rouse 1986). European explorers found well-developed, ag- ricultural-based societies on all of the larger Pa- cific islands. It is not known how many of the smaller, less suitable islands were visited only temporarily by the wandering islanders (Oliver 1961). Habitat loss and exotic species (including dogs and pigs) doubtlessly caused the extinction of many species of endemic birds on the per- manently settled islands. Even on smaller unin- habited islands endemic species, many of them flightless rails that had evolved in the absence 15 16 STUDIES IN AVIAN BIOLOGY NO. 22 14. "24 2O FIGURE 1. x,6o ' The islands of the Pacific. Numbers refer to 34 ß tie, 40' island groups referred to in the text and listed in Table 1. of terrestrial predators, could have been har- vested to extinction by temporary human occu- pants. We have evidence of these unrecorded extinc- tion events in the fossil record (Olson and James 1982a, 1991; Milberg and Tyrberg 1993). Ar- cheological efforts in Hawai'i by Olson and James (1982a, 1991; James and Olson 1991) and throughout the rest of the Pacific (Balouet and Olson 1987; Steadman 1991, 1992, 1993, 1997a,b; Kirch et al. 1995), have uncovered a large number of avian fossils that were depos- ited concurrently with early human occupation of the islands. Not all islands have been searched, and even if they were, it is unlikely that all extinct species would be found. Thus, the total number of extant and extinct species identified to date is an underestimate of the di- versity of the prehuman Pacific avifauna. An exact count of the number of landbird spe- cies known only as fossils is difficult to tally because they are not clearly enumerated in some published accounts. The Hawaiian Islands held 62 fossil species (James and Olson 1991, Olson and James 1991) and New Zealand held 44 spe- cies (Steadman 1995). The other islands of the Pacific that have been searched held something less than 100 additional species (Steadman 1995). Thus, roughly 200 species of Pacific landbirds are known only from the fossil record. Summing the number of extant, historically extinct, and prehistorically extinct (fossil) spe- cies, there are 540 known species of landbirds in the Pacific. This number is too low because the fossil record is incomplete. An accurate es- timate of the prehuman Pacific avifauna depends on an accurate estimate of the "missing" fossil species. Pimm et al. (1994) estimated the prehuman number of Pacific island landbirds by applying sampling analyses to fossil data. Briefly, given the number of species known only by fossils, those known by modern observations (i.e., those that still survive and those extinct since Euro- pean colonization), and those known by both fossils and modern observations one can deduce the number of "missing" species from an island. Applying this method to data on the landbirds HOW MANY BIRD SPECIES--Curnutt and Pimm 17 of the tropical Pacific (including New Caledo- nia), Pimm et al. (1994) deduced that the num- ber of known fossil species (ca. 200) is only half of the actual number of species that disappeared before European colonization. Pimm et al. (1994) estimated the original avifauna to include nearly 800 species of landbirds. Excluding data from New Caledonia and including data l¾om New Zealand, to fit the boundaries to the current study, does not appreciably change these esti- mates. A much higher estimate of the original Pacific avifauna was proposed by Steadman (1995, 1997). On finding fossil evidence of up to three or four now extinct species of flightless rails on islands he investigated, Steadman (1995, 1997a,b) suggested that the 800 major islands of the Pacific held more than 2,000 species of this taxon and lower numbers of other taxa--all driven to extinction as a result of first human contact. Steadman's (1995) approach set the question of original avifauna in the context of island biogeography. In this paper we apply a robust theoretical framework, island biogeography theory (Mac- Arthur and Wilson 1967a), to the Pacific islands to determine the number of islands that could have held endemic species of rails (Rallidae), pigeons and doves (Columbiformes), and parrots (Psittaciformes). We chose these taxa because they are well represented in the fossil record. Thus, we do not estimate the entire prehuman landbird fauna; instead our results can indicate the magnitude of the loss of bird diversity that has occurred since first human contact. We in- clude in our analyses all named islands of New Zealand, Micronesia, central and eastern Mela- nesia, and Polynesia that experienced first hu- man contact no earlier than 4,000 years before present (Rouse 1986). Unlike Steadman (1995, 1997), we incorporate data on habitat diversity, changing sea levels during the Holocene, and tsunamis. Each of these factors influences the effective size of islands for landbirds. Put sim- ply, MacArthur and Wilson's (1967a) theory of island biogeography predicts more species on larger islands and those close to a source of im- migrants, and fewer species on small or isolated islands. We perform two distance analyses: dis- tance-from-source, as proposed by MacArthur and Wilson (1967a); and, distance between is- lands--isolated islands are more likely to pro- duce species endemic to one island than those that have very near neighbors (Mayr 1963). By applying reasonable assumptions to this ques- tion, we hope to develop a more accurate esti- mate of the prehuman Pacific avifauna than has been produced to date. We first identify those islands of the Pacific that have the potential to maintain populations of landbirds. We then extrapolate the numbers of endemic rails, pigeons, and parrots that could have existed on all of these islands by applying the known maximum of each taxon recorded on different island sizes and types. In lhct, we cal- culate two estimates of the number of endemic species by using two definitions of endemism. We then refine our estimates by considering eco- logical and environmental characteristics. IDENTIFYING THE BIRD ISLANDS We do not expect all islands of the Pacific to hold birds. Some islands are too small to support viable populations of landbirds. Some islands may also fall outside of the known range of the taxa we are investigating. These limitations to bird distribution are diagrammed in Figure 2 Our first task, then, is to estimate how many islands there are in the Pacific, and which of these could support a population of landbirds. How MANY ISLANDS No one knows how many islands there are in the Pacific Ocean. Estimates range from 30,000 to less than half of that number (Bryan 1963). The distribution of island sizes is fractal--that is, as one looks at the Pacific at finer scales, one finds more islands in a characteristic way. Thus, most islands are very small. We limited our data to named islands. We obtained gazetteer data (latitude, longitude, name) from the U.S. De- fense Mapping Agency's (DMA) database avail- able on the internet. This search yielded 3,463 islands. We assigned each island to an island group according to an arbitrary grouping scheme. Ob- vious archipelagos were identified as groups (e.g., the Gilbert Islands), as were single islands not obviously associated with an archipelago (e.g., Rapa). The result was 41 island groups (Table 1; Fig. 1). As described below, we first grouped islands that are very close to each other. Our primary reason for this was to add small islets to the larger islands that they surround and to unite many "islands" that occur as parts of individual atolls. Second, we determined which islands are too far from a source of immigrants for each taxon. Finally, we determined the size and topography of each island. ISLANDS AND ISLETS If two islands were near enough to each other to allow a species to move between them, then neither would produce an endemic species (Ricklefs and Schluter 1993). But how close is close enough? No data exist on this subject for birds in the Pacific. We know that the limiting distances between islands are surely taxon spe- 18 STUDIES IN AVIAN BIOLOGY NO. 22 Species' distribution limit Region of possible endemism Lack of isolation Island size FIGURE 2. Theoretical framework for endemism of Pacific island birds. Islands that are too small to maintain persistent populations will not produce endemic species (lower size limit), nor will larger islands subject to inhibitory disturbance regimes (effective lower size limit). Some islands are close enough to allow genetic exchange between populations and will not produce endemic species (lack of isolation), while others lay outside of the distribution of some taxa (species' distribution limit). cific and this, in turn, is affected by the mode and propensity of movement exhibited by each taxon. For the three taxa we consider in this pa- per, rails have a higher wing load (ratio of weight to wing area) than pigeons or parrots (Raynet 1988). Thus, it would take relatively more energy for a rail to fly a fixed distance than it would a pigeon. Left free to speculate, we chose a minimum distance equivalent to 0.1 ø of latitude or longitude ( 11 km at the equator) as sufficient for allowing isolation of breeding pop- ulations. We chose this distance primarily for ease of calculation, but also we feel that such a distance would provide an adequate barrier to movement for rails--the most stationary taxon because of its propensity to quickly evolve to- ward flightlessness (Trewick 1997). We summed the sizes of all islands that were closer than 0.1 ø of latitude or longitude to each other. This grouping scheme reduced our data to 788 island sets. Hereafter, we refer to island sets as "islands." WHICH ISLANDS ARE TOO FAR Landbirds are not distributed evenly across is- lands. Just as islands that are too close will pro- hibit divergence; islands that are too distant from a source population may not be colonized at a rate sufficient to allow persistence (Ricklefs and Schluter 1993). We tested for the effect of distance-from- source on the distribution of each of our three taxa with multiple regressions. All of the taxa we consider in this paper have their origins in the Old World (rails: Ripley 1977; pigeons: Goodwin 1983; parrots: Forshaw 1977). We used Map © (Apple Computers, Inc.) software to determine distances between geographic centers of island groups and the following (geologically) continental source areas: Australia (Brisbane), Papau New Guinea (New Britain), Philippines (Manila), and Taiwan (Taipei). Since island size is the most effective predictor of species diver- sity (MacArthur and Wilson 1967a), we per- formed stepwise multiple linear regression of the number of species on total area of each island group, then added distance. We repeated this process for each of the distances generated from the four sources listed above. At best, these multiple regressions only weak- ly explained the variation in species numbers with distance (R < 0.2) and were only signifi- cant for parrots and pigeons (P < 0.05). For this analysis, it is better for the data to speak for themselves. Figure 3 shows the distribution of rails, parrots and pigeons among the 41 island groups of the Pacific. Rails arc found throughout the region, reaching the most remote groups in- cluding Hawai'i and Easter Island. Paradoxical- ly, rails, for which even the largest ocean is not HOW MANY BIRD SPECIES--Curnutt and Pimm 19 TABLE 1. ISLAND GROUPS OF THE PACIFIC OCEAN IN- CLUDED IN OUR ANALYSES No. of Group island Topogra- Group number sets Area (km 2) phy Melanesia Vanuatu 1 38 11,400 H Fiji Islands 2 74 1,860 H Micronesia Palau 3 8 447 H Yap 4 2 175 H Chuuk 5 21 230 L Mariana Islands 6 13 910 H Pohnpei 7 2 360 H Kosrae 8 1 100 H Marshall Islands 9 28 255 L Gilbert Islands 10 18 290 L Nauru 11 2 36 L Polynesia NW Hawai'i 12 2 8 L Hawai'i 13 9 16,700 H Wake 14 1 230 L Johnson Atoll 15 1 2 L Howland 16 1 10 L North Line Islands 17 7 745 L Phoenix 18 5 37 L Tuvalu 19 6 27 L Rotuma 20 1 49 H Wallis and Futuna 21 2 275 H Samoa 22 8 3,500 H Tokelau Islands 23 3 13 L North Cook Islands 24 5 10 L Tonga Islands 25 18 563 H Niue 26 1 258 L South Cook Islands 27 9 234 H South Line Islands 28 2 8 L Marquesas Islands 29 11 1,062 H Society Islands 30 10 1,710 H Tuamotu Arch. 31 11 248 L Gambier 32 6 21 L Pitcairn Islands 33 2 8.5 H Rapa 34 1 40 H Tabuai Islands 35 4 120 H Easter Island 41 1 170 H Kermadec Islands 36 2 34 H Norfolk 37 1 37 H Lord Howe 38 1 10 L New Zealand 39 33 267,800 H Chatham Islands 40 4 1,085 H "Group Number" refers to numbers shown on Figure 1. Island sets are named Islands that are within 0.1 ø latitude and longitude of each other. We include only those sets with combined areas of >150 HA. Topogra- phy is either high relief (H) or low-relief (L). large enough to prohibit colonization, can quick- ly evolve to flightlessness (Diamond 1991). The distribution of pigeons has apparently been lim- ited by the vast expanses of ocean that isolate Hawai'i and Easter Island, for neither has ap- parently held this taxon. For Easter Island, the nearest island to have ever held a pigeon is Pit- cairn (1,600 km), and for Hawai'i, it is the North Cook Islands (3,500 km). Parrots have been found on Easter but not the Hawaiian Islands (nearest island with parrots-Marquesas, 3,800 km distant). For our analyses, therefore, we consider all islands of suitable size as potential sites for rail colonization; all but the Hawaiian and North- west Hawaiian Islands for parrots; and, all but the Hawaiian groups and Easter Island for pi- geons. SIZES OF ISLANDS The final parameters we consider in determin- ing which island sets could maintain populations of landbirds are size and topography. We ob- tained data on the sizes of islands from various sources in the literature and from direct mea- surements from maps (ranging in scale from 1: 10,000 to 1:300,000). Some islands listed in the DMA database were not found on maps (or re- ferred to in any literature we searched), thus, we have no data on their sizes. However, we are confident that we have size estimates for all of the major islands (i.e., > 2 km 2) and for many lesser islands, and those with missing data are from the smallest size classes. Our confidence lies in the fact that island sizes fall within a class of negative exponential distributions known as Zipf-Mandelbrot (Fairthorne 1969). For the is- lands for which we have data, we plotted the size distributions on log-log axes. The Zipf- Mandelbrot distribution predicts a straight line for this graph (Fig. 4), and we can interpret de- viations from the linear fit as "missing" islands. By extending the linear fit below 1 km 2 to our smallest recorded island size (10 ha), we predict that about 800 islands are missing from our is- land size data set. While landbirds do occur on very small is- lands in the Pacific, these are members of sat- ellite populations of larger nearby islands. For example, the Antipodes Island Parakeet (Cy- anoramphus unicolor) is found in low numbers on Archway Island (6 ha)--the smallest of the Antipodes Islands (Taylor 1985). The species is also found on the 54-ha Bollons Island, which is much less than 1 km from Archway Island. The greatest part of this species' population, however, is on the 20 km 2 Antipodes Island-- about 1 km from Bollohs. The loss of the An- tipodes Island population would probably lead to the eventual extinction of this species. It would not make ecological sense to identify Archway Island as one suitable for sustaining a population of parrots. Similarly, we can safely ignore the existence of the 800 "missing" is- lands in our data because they are too small to hold endemic species of landbirds. The smallest Pacific island known to hold an endemic rail is Wake Island, 6.5 km 2 and home 20 STUDIES IN AVEAN BIOLOGY NO. 22 'R ' R,P,D ' RD' .p  . R,P % R,P,D ;tO' 40' FIGURE 3. The distribution of rails (R), parrots (P), and pigeons and doves (D) among the Pacific islands. Numbers correspond to group names in Table 1 and indicate island groups that hold none of the three taxa mentioned above. to Rallus wakensis. The smallest island to hold an endemic pigeon is 28 km 2 Maketea (Tuamotu Archipelago), home to Ptilinopus chalcurus; and the smallest island to hold an endemic parrot is Norfolk Island (33.7 km 2) where remains of Nestor produetus have been recovered. These minima may not be actual; all islands have not been sampled. We performed a Monte Carlo simulation (Efron and Tibshirani 1993) to predict the minimum size of an island that should support an endemic species from the ob- served distribution of island sizes with endemic species. Using data on island sizes, we randomly selected a number of islands equivalent to the number that we knew held endemic species of each taxon. For example, 23 islands held at least one endemic species of rail. We randomly se- lected 23 islands from the entire set of 834 and recorded the minimum size of this subset. We then calculated the mean minimum value of 100 repetitions. By repeating this process with in- creasing cutoff values applied to the entire data set, we determined the lower 95% confidence limit within which our known minimum island size fell (Fig. 5). Some islands have held more than one en- demic species of a taxon. For parrots and pi- geons there were one and two islands, respec- tively. For these taxa we could not perform the above described simulation to determine the minimum island sizes for two or more species-- the sample size is too small. For rails, however, of which 10 islands held more than one endemic species, we could estimate the minimum island size for two species by applying the simulation (with a sample size of 10). To determine which islands could have held more than two species of rail (or more than one species of parrot or pigeon), we assumed that the smallest island for which we had data was the actual minimum. TYPES OF ISLANDS Our measure of habitat diversity was very coarse. We described islands as "high-relief" or HOW MANY BIRD SPECIES--Curnutt and Pimm 21 0.5- 0 i   4 5 6 Log Size (ha) FIGURE 4. The relationship between island sizes and their frequency. The linear fit was calculated after excluding the two smallest size classes (open circles) and the three largest size classes (not shown). The area within the triangle represents islands with size data missing from our data set, assuming island sizes ex- hibit a Zipf-Mandelbrot distribution. "low-relief." High-relief islands were those de- scribed in the literature as volcanic, hilly, or mountainous or whose representation on maps included hachures. Low-relief islands were all of those described as atolls or were lacking ha- chures on maps that normally include such data. High-relief islands are rich in habitat diversity compared to low-relief islands (Adler 1992). We apply the same topography to entire groups by summing the areas of all islands within groups and defining them as high relief if > 50% of the total area is attributed to high-relief islands. EXTRAPOLATING ENDEMICS To estimate the potential number of endemic species that each taxon held, we determined the known maximum number of endemics (living and fossil) on islands of different sizes and to- pographies throughout the Pacific. After esti- mating the size of the smallest islands which we would expect to find endemics on, we used these numbers to predict the maximum numbers of en- demic species with reference to the distribution of island sizes and topographies within each is- land group (Fig. 6). We tallied the number of known endemics and the number of predicted endemics across taxa for each island group then 5000- 4500 4000 3500 3000 2500 2000  15002 1000 500. 0; ....... o Pigeons (B) ........ ,.- ........ Rails (C) ........  ........ Parrots (A) Predicted Minimum size (ha) FIGURE 5. Results of a simulation whereby we randomly selected a number of islands equivalent to the number occupied by endemic species of each taxon. The x-axis represents the lowest value in the data set for each simulation, the y-axis is the 95% lower confidence limit of the mean of 100 repetitions. A, B, and C represent the actual minimum sizes for parrots, pigeons, and rails, respectively. The vertical lines intercept the x-axis at the smallest island size we would expect to find endemics of the respective species. 22 STUDIES IN AVIAN BIOLOGY NO. 22 -4 ß Low-relief islands [] High-relief islands / (12) - 3 )8) - 2 J (1) .................................. (1) J (4) (4,) 1 Log Size (ha) FIGURE 6. The size distribution of islands of the Hawai'i group classified as high relief and low relief. The solid line indicates the maximum number of endemic rails found on all high-relief islands in the Pacific while the dashed line indicates maxima for low-relief islands. We multiplied the maxima for each size class by the number of islands in each size class to predict the number of endemic rails that could have existed in each island group. Numbers in parentheses indicate the number of rails expected for each island size X the number of islands. calculated the proportion of missing endemic species. Our use of maxima reflects the potential lack of fossil data on some islands. For example, well-searched Mangaia of the South Cook Is- lands group held four endemic rails. Tofua of the Tonga Islands, with a similar size and topogra- phy, revealed none. For our estimates we assume that Tofua held four endemic rails. This may be incorrect; to paraphrase Montaigne, speciation is not so often the result of great design as of chance. There may never have been endemic rails on Tofua simply because no rails have sur- vived there long enough to speciate. Since the true number of prehistoric endemics cannot be known, we must be content with es- timating this number by setting realistic limits based on the available data. Of the four factors we consider as aftkcting endemism, we have data on absolute lower island size and distance from source. Data do not exist for two other fac- tors-eftkctive lower island size (disturbance ef- fects) and the minimum distance between is- lands needed to produce endemism (dispersal ef- fects). Thus, we are left with the familiar quan- dary of decreasing our certainty as we increase the number of parameters. We address the prob- lem of prehistoric disturbance on a group by group basis later. Our approach to effective dis- tance between islands is as follows. As noted earlier, we grouped all islands 11 km or closer to each other into sets. While an 11 km expanse of ocean may prohibit the move- ment of a flightless rail, it may have less effect on a strong-flying pigeon. We could further group our islands by diftkrent distances for each taxon, but this would be a series of educated guesses at best. Instead, we approach this prob- lem by determining the maximum number of en- demics that we know to occur in each island group. For example, the Red-bellied Fruit Dove (Ptilinopus greyii) is found on 28 islands of the Vanuatu group (total area of 11,000 km2). Thus, it does not fit our definition of a single-island endemic. It is, however, found only in the Va- nuatu group, so it does exhibit a form of endem- ism. In Vanuatu, this species is found on both low- and high-relief islands. We conclude then that any island group that is dominated by high- relief islands and has a combined area the size of the Vanuatu group would hold an endemic pigeon. We, therefore, produce two estimates for each taxon--the number of endemics at single islands and the number of endemics at island groups. The true number of endemic rails, pigeons, and parrots that have existed in the Pacific probably falls somewhere between these two values. THE BIRDS We chose rails, parrots, and pigeons for our analyses because they are well represented in the fossil record. We reviewed all available litera- ture on the distribution of extant, historically ex- HOW MANY BIRD SPECIES--Curnutt and Pimm 23 TABLE 2. AN ESTIMATE OF THE NUMBER OF RAIL SPE- CIES 1N THE PACIFIC BEFORE HUMAN COLONIZATION Num- Predict- bet of cd total Number spe- number of is cies/is- of spe- Island size and topography lands land cies <600 ha, high and low relief 578 0 0 600-1000 ha, high and low relief 44 1 44 <1000 ha, low relief 61 I 61 1000-6400 ha, high relief 86 2 172 <6400 ha, high relief 65 4 260 Total 834 537 Maximum numbers of species are gleaned from the data for each size/ topography of island. The predicted number of species is the product of maxima and the number of islands. tinct, and subfossil species of these taxa in the Pacific. We assigned each species to all islands on which it was known to occur. RAILS Single-island endemics We catalogued 55 species of rails known to have occurred in the Pacific. Of these, only five (all extant) are not restricted to either single-is- land sets or single-island groups. Two-thirds of the species are known only from fossil data and 65% are endemic to one island. Endemic rails are found on only 13 of the 41 island groups The results of our simulation show that the smallest island with an endemic rail (6.5 km 2) falls within a distribution that has a lower 95% confidence limit of 6 km 2. Both high- and low- relief islands have held single endemic species of rails, thus, we expect that all 256 islands that are larger than 6 km 2 held at least one species. Ten islands, all high relief, held more than one endemic species. The smallest of these was Lord Howe Island (10 km2), which held two species, followed by Mangaia (64 kin2), which held four. Since four species of endemic rails is the max- imum we encountered, we apply this value to all larger islands. Table 2 and Figure 6 illustrate our method of prediction of the number of rail spe- cies for the entire Pacific and specifically for the Hawaiian Island group. We performed the same analysis on each is- land group and estimated that approximately 537 endemic rail species existed in the Pacific, of which 482 are not accounted for by a living or fossil species. Over one-third (36%) of the miss- ing endemics are attributed to only two groups--Vanuatu (94) and Fiji (86). Whereas 13 groups hold no endemics nor are expected to, 14 others hold none but should. Of the remaining 13 groups, 11 hold fewer endemics than ex- pected, and two (Wake Island and Lord Howe Island) hold the number of endemics we predict (one and two, respectively). Island-group endemics Eleven of the 55 species of rails in the Pacific are endemic to groups of islands. The occur- rence of the Wake Island Rail (Rallus wakensis) on Wake Island, an island group in itself, insures the expectation of at least one endemic rail on all low-relief groups except Johnston Atoll, which is too small. For groups with high-relief islands, the maximum number of endemics rang- es from two for groups as small as 10 km 2 (Lord Howe) to 12 for groups larger than 16,700 km 2 (Hawai'i). Summing over all groups, we expect 143 endemic rails in the Pacific based on our island group analysis. PARROTS Single-island endemics Of the 24 species of parrots we catalogued, 9 are endemic to single islands. The majority of these (5) are found in the southwest Pacific. No low-relief islands hold endemic parrots. Norfolk Island (33.7 km 2) represents the smallest island to hold an endemic parrot (Nestor produetus). We estimated that the lower size limit of islands that would support endemic parrots is 28.5 km 2. Excluding the Hawaiian islands and Easter Is- land, there are 110 high-relief islands of 28.5 km 2 or greater. The only island with more than one species of endemic parrot is the largest in our data set--South Island, New Zealand (149,000 km2). Thus, we attribute three species to this island only, for a total of 94 species ([91 islands * 1 species] + [1 island * 3 species]). Island-group endemics In contrast to the rails, a large proportion of parrot species (30%) in the Pacific show endem- ism to single groups of islands. The smallest group to hold an endemic is Norfolk (34 km2), home to Nestor produetus. We apply this value of one endemic to 18 of the 22 island groups that contain high-relief islands. We predicted two endemic parrot species to Vanuatu and Fiji. New Zealand held four endemics. The total number of endemic parrots we expect from our analyses of island groups is a mere 29 species. PIGEONS AND DOVES Single-island endemics We catalogued 43 species of pigeons and doves in the Pacific. Only nine of these are en- demic to single islands. Of these, five are known only from fossil remains and are identified only to genus. Huahine of the Society Islands held the highest number of endemics with three of 24 STUDIES IN AVIAN BIOLOGY NO. 22 the unknown species (Ducula sp., Gallicolumba sp., and Ptilinopus sp.). Henderson Island of the Pitcairn group held two endemics--the extant Henderson Island Fruit Dove (Ptilinopus insu- laris) and a fossil Gallicolumba sp. The remain- ing four endemics were found on Rapa (Rapa Fruit Dove, Ptilinopus huttoni), Mangaia of the South Cook Islands (Gallicolumba sp.), Makatea of the Tuamotu Archipelago (Makatea Fruit Dove, Ptilinopus chalcurus), and Espiritu Santo of the Vanuatu group (Santa Cruz Ground Dove, Gallicolumba sanctaecrucis). The smallest island to hold an endemic was Makatea of the Tuamotu Archipelago. Makatea is 28 km 2 and low relieœ We estimate that the small- est island likely to hold an endemic pigeon or dove would be 20.7 km 2. Islands with more than one endemic are Henderson (36 km 2) with two species and Huahine (75.5 km 2) with three--both of these islands are high relief. Again, excluding Easter Is- land and the Hawaiian groups, our estimate of the total number of enderrfics is thus: (53 islands * 1 species) + (25 islands * 2 species) + (50 islands ß 3 species) = 253 species. Island-group endemics Just as we saw that a greater proportion of parrots showed endemism to groups of islands than the less mobile rails, a full 51% of the pi- geons and doves are restricted to single-island groups compared to 30% for parrots. Thus, there appears to be a positive relationship between flight ability and area over which endemism ex- tends. Vanuatu held the most species (6) of pigeons and doves that were restricted to an island group, and the Marianas held the next highest number (5). These, and the other large groups of islands (Chuuk, Fiji, New Zealand, the Soci- ety Islands, and Tonga) account for 39 of the total 64 species of island-group endemic pigeons and doves. Unlike parrots, endemic pigeons and doves are also found on large low-relief groups. Two species are restricted to the Tuamotu Ar- chipelago, a fact that leads us to predict the same number of species on the Marshall Islands. THE ESTIMATED NUMBER OF ENDEMICS Our exercise produced two sets of estimates of the number of endemic species in each of three taxa. For estimates based on single-island endemism, we predict 537 species of rails, 94 species of parrots, and 253 species of pigeons and doves for a total of 884. We can account for only 57 single-island endemic species of the three taxa as either fossil, extinct or extant. Es- timates based on island-group endemism yield 145 species of rails, 29 species of parrots, and 64 species of pigeons and doves (Fig. 7). We can account for 40 of these as fossil, extinct, or extant. Thus, we predict that the total number of endemic species of these taxa that once occurred in the Pacific falls between 242 and 884. TESTING THE MODELS: KNOWN VERSUS ESTIMATED ENDEMISM We may now investigate factors that would refine our predictions. Which estimates better re- flect the known distribution of endemic rails, parrots, and pigeons and doves in the Pacific-- those derived from single-island endemics or those from island-group endemics? To answer this question, we compare our predicted values with the known distribution of endemic birds. We calculated two indices of the proportion of total missing endemics (all taxa combined) per island group, one for each of our definitions of endemism. We added I to all values of the total number of endemics known to exist and to the totals predicted from our two definitions of endemism. We did this so that we could calcu- late proportions (number of known endemics/ number of predicted endemics) without having zero values in either the numerator or denomi- nator. We arcsine transformed the proportions to make the distribution normal and ranked the re- sults. We then compared the ranks by perform- ing a linear regression of single-island endemic ranks on island-group endemic ranks (Fig. 8). Not surprisingly, the linear fit was significant (F = 18.37, P < 0.01). The slope was less than unity (b = 0.56) suggesting that when the pre- dicted number of endemics corresponds with the actual number of island-group endemics, the sin- gle-island prediction is low and vice versa. We tested for the influence of the number of islands in each group on both of our predictions. Neither set of predictions correlates with this parameter (r < 0.2 for both). Identifying each group as high relief (50% of total area is high relief) or low relief reveals the pattern responsible for the disparity between the two sets of ranks (Fig. 9). Predictions correspond best with known endem- ism for low-relief groups when endemism is de- fined as a single-island distribution. Conversely, for high-relief groups, predictions based on group endemism correspond best with the num- ber of known endemics. We believe there are ecological reasons for this. Groups of low-relief islands tend to have smaller islands than high-relief groups (ANO- VA: F = 4.21, P = 0.04). For low-relief groups, an individual island approach to endemism would successfully identify those few large is- lands in the group that could support a large population of birds. In contrast, predictions based on group endemism would lead to over- estimates because the area across each group is HOW MANY BIRD SPECIES--Curnutt and Pimm 1000 25 800' 600' [] Parrots [] Pigeons [] Rails ß ', 400' E 200' 4--' A B C D FIGURE 7. Total predicted numbers of endemic rails, parrots, and pigeons in the prehistoric Pacific under four sets of assumptions: (A) endemic species are those that occur on only one island; (B) endemic species are those that occur within single-island groups; (C) low-relief island groups produce endemic species at single islands and high-relief island groups produce endemics at island groups; and, (D) the same as (C) with modifications driven by patterns of disturbance (sea-level change and tsunamis). summed. Conversely, the assumption of single- island endemism for the larger islands of high- relief groups ignores factors that potentially lim- it the size of bird communities. In his analysis of the assembly of the fruit-pigeon guild in New Ranked prop. of group endemism FIGURE 8. Ranked proportions of predicted num- bers of endemic birds (rails, parrots, and pigeons com- bined) over known numbers of endemics. Values on the y-axis were generated using the assumption of sin- gle-island endemism while those on the x-axis were generated with the assumption of island-group endem- ism. Guinea, Diamond (1975) showed that the entire species pool is never found in one locality. Some species never occurred together and some sets of species excluded particular species. This ef- fect is primarily due to competition between species with closely related niches. Another eco- logical factor that over inflates the estimates for high-relief islands stems from our grouping across taxa. Some high-relief islands may pro- vide habitat for each of the three taxa we dis- cuss, but it may be unreasonable to assume that all of them do. We can now refine our original estimates of endemism by calculating the totals for each tax- on separately for low- and high-relief island groups using the appropriate assumptions of en- demism (low-relief and single-island endemism; high-relief and island-group endemism). This yields 206 species of rails, 38 species of parrots, and 101 species of pigeons and doves (Fig. 7). These sum to 345 species across taxa. WHERE THE ENDEMICS ARE AND WHERE THEY ARE NOT Five island groups (Johnson Atoll, Howland, South Line, Gambier, and North Cook) are all low relief. They have no endemic species, nor are expected to under the assumption of single- island endemism. Our interpretation of the re- 26 STUDIES IN AVIAN BIOLOGY NO. 22 10' -10 ÷ -2O -3O ß oo o o ß o o ß o o o. o ß ß o ß ß ß o ß ß ß 0 0 ß 0 ß 0 ß ß High-relief Low-relief 0 10 20 30 40 50 Ranked proportion of group endemism FIGURE 9. Residuals of the linear relationship of predicted endemic species under single-island endemism versus island-group endemism (Fig. 8) with each island group defined as either high or low relief. suits for the remaining 36 island groups depends on two assumptions. First, the maximum number of endemics recorded represents the actual max- imum of each taxon that could occur on each type of island and, second, the recorded maxima on each island size/topography are applicable to all islands in each class. There is a chance that the first assumption is incorrect. Continued ex- cavation of subfossil remains may well produce more species of birds, even on islands that are already well represented with endemics. The second assumption ignores differences in the history of islands across the Pacific. While there is little we can do to refine our predictions in light of the uncertainty of the first assumption, we can investigate the history of the Pacific is- lands to uncover patterns of species numbers on island groups. The name "Pacific" belies this ocean's vio- lent history. Natural disturbance of the Pacific islands can be a potentially limiting factor in speciation among birds. Stoddard and Walsh (1992) list five environmental factors that influ- ence island ecosystems: vulcanicity and earth- quakes, sea-level change, tsunamis, rainfall pat- terns, and hurricanes. We investigate two of these: sea-level change and tsunamis. We chose these factors because they operate at regional scales, their effects are unambiguous, and they occur across a temporal scale that is consistent with evolutionary time. SEA-LEVEL CHANGE A number of studies concerning sea-level change in the Pacific over the last 10,000 years have been reported in the literature (Ota et al. 1988, Pirazzoli and Montaggioni 1988, Yonek- ura et al. 1988, Pirazzoli 1991). Throughout the Pacific, sea level was much lower 10,000 years before present (BP) than any time since. At that time, global sea levels were rising rapidly with the melting of the glacial ice sheets. Indeed, the massive infusion of water into the oceans led to regions of hydroisostasy (depression of the ocean floor by water loading) and consequent elevated sea levels (Pirazzoli 1991). Thus, from 6,000 BP to as late as 1,200 BP some island groups had sea levels significantly higher than at present. During the last glacial maximum (18,000 BP), when sea levels were nearly 150 m lower than today, all islands of the Pacific were larger. For example, the Fiji group currently has a com- bined area of 18,600 km 2, whereas at 18,000 BP its area was over 35,000 km 2 (Gibbons and Clu- nie 1986). With rising sea level there would have been a loss of area and habitat. Thus, many island groups probably held more endemic spe- cies in the distant past than they did even in prehistoric times. Isostatic effects have been re- corded for French Polynesia, the South and North Cook Islands, and the Marquesas Islands HOW MANY BIRD SPECIES--Curnutt and Pimm 27 (Pirazzoli and Montaggioni 1988, Yonekura et al. 1988, Stoddard and Walsh 1992; Table 2). As late as 1,200 BP, these groups exhibited less sur- face area than today--with groups such as the Tuamotu Archipelago disappearing almost com- pletely (Gibbons and Clunie 1986). This scenario raises two important consider- ations for our estimates of the prehuman avifau- na. First, the decrease in area of many large is- lands that began at 18,000 BP would have caused a decrease in the number of bird species. This decrease may not have been contemporary with the decrease in area. Diamond (1972) showed that the reduction of one large island, the D'Entrecasteaux Shelf, into a number of small fragments should have led to a reduction of the number of bird species to a new equilib- rium. However, he suggests that the time to reach the new equilibrium is dependent on the size of the new island. Thus, there could be a lag time (of several thousands of years in the above case) before the actual species numbers reflect the restraints of the size of the new island. We are not aware of any studies similar to Di- amond's (1972) that address the islands included in our analyses. We will assume that the avifau- na of the islands was at equilibrium at 4,000 BE In doing so, we risk underestimating the number of species on all islands but those affected by the above mentioned isostatic effect; for these islands, our estimates would be to high. The second consideration regarding sea level and endemism is the effect of elevated sea levels on low-relief islands. The low-relief island groups of Gainbier, North Cook, and the Tua- motu Archipelago were affected by isostatic sea levels (Table 3). Of these, only Tuamotu is ex- pected to have single-island endemics. We pre- dict six species of rails and six species of pi- geons-one pigeon exists (Ptilinopus chalcu- rus). Of this group's 60 islands, only five are greater than 30 km 2. Apparently, this species was able to survive the elevated sea level of 6,000-1,200 BP among these islands. The Fiji group is dominated by large high-relief islands but also holds a large number of surrounding low-relief islands. This group experienced sea levels nearly 2 m higher than present as late as 2,500 BP (Gibbons and Clunie 1986). Endem- ism would have been improbable in these is- lands up to that time because of the lower extent of the area. We predict that eight species of pi- geons and 16 species of rails could have inhab- ited these low islands--none are known to have existed there. We removed the low-relief islands from the total area and calculated the number of endemic species we would expect on Fiji based on group endemism. This had no effect on our predictions. The size of Fiji's high-relief islands TABLE 3. ISLAND GROUPS FOR WHICH PUBLISHED DATA EXIST ON MEAN SEA LEVELS (RELATIVE TO PRESENT; IN METERS) AT THREE PERIODS OF THE HOLOCENE (FROM PIRAZZOLI 1991); MAXIMUM SEA LEVEL AND TIME OF OC- CURRENCE (OTA ET AL. 1988, PIRAZZOLI AND MONTAG- GION! 1988, YONEKURA ET AL. 1988, PIRAZZOLI 1991); AND MAXIMUM TSUNAMI RUN-UP HEIGHT (NATIONAL GEO- LOGIC DATA CENTER) Group Mean relive sea level Years before present x 10 3 Maxi- Maximum mum 10 5 2.5 sea level run-up Melanesia Vanuatu Fiji Micronesia Palau Yap Chuuk Marianas Pohnpei MarshalIs Gilbert Polynesia Hawai 'i North Line Tuvalu Samoa Norda Cook Tonga South Cook Marquesas Society Tuamotu Gainbier Pitcairn Rapa Tabuai Kermadec Norfolk New Zea- land Chatham + 1 0 2 (2500) -40 -2 - 1 +4.5 +2.4 -40 -5 -2 +2.4 -3 +2.4 >-15 0 0 +0.6 5 -2 -17 -1 +1 >-20 +0.5 +1 > 20 +0.9 +0.9 0 5.9 0 1.9 1.9 0 16.8 0 1.9 1 (1500) 0 0 1.7 (3400) 0 1 (1500) 9 1 (1500) 3.4 1 (1200) 2.3 1 (1500) 0 1 (1500) 1.8 1 (1500) 0 12 0 0 5.9 0 0 are near the maximum for the Pacific, and the removal of the low-relief islands did not lead to a change of the maximum number of species expected. Johnson et al. (1996), investigating the evo- lution of cichlid fish, reported the most rapid vertebrate speciation known--on the order of 3,000 years. Thus, high sea levels up to 1,200 BP must have reduced bird speciation on some Pacific islands. The eflbct of our sea-level anal- yses on our predictions results in the removal of five species of pigeons and six species of rails from our total. TSUNAMIS Tsunamis are a series of high-energy waves propagated by a major displacement of earth un- 28 STUDIES IN AVIAN BIOLOGY NO. 22 FIGURE 10. Areas affected by tsunamis (shaded) and the direction of tsunamis arrows) in the Pacific from 1900 to 1983 as reported in the Worldwide Tsunami Database (Lockridge and Smith 1984). der the sea. They can have devastating effects on islands. For example, in the early morning hours of I April 1946 an earthquake in the Aleu- tian Islands, Alaska, caused a tsunami. Within minutes a manned lighthouse on Unimak Island had been obliterated with all hands lost. Four and a half hours later and over 3,000 km away the same tsunami hit the Hawaiian Islands. Reaching a maximum run-up height of nearly 17 m, it smashed into the Island of Hawai'i taking another 241 lives. This same series of waves caused casualties and property damage in Cali- fornia and as far south as central Chile (Lock- ridge and Smith 1984, Myles 1985). Tsunamis of this magnitude are frequent with 14 occurrences in the Pacific Basin from 1900 to 1983 (Lockridge and Smith 1984). As with sea-level change, the effect of tsunamis on is- lands is variable. Islands without surrounding submarine shelves are more susceptible to re- motely generated tsunamis because there is little to absorb the energy of the waves before they make contact. Topography and elevation above sea level are also obvious factors in determining the effect of tsunamis on islands. We accessed the Worldwide Tsunami Data- base, compiled by the National Geologic Data Center (http ://j ulius.ngdc.noaa.gov/seg/hazard/ tsudb.html), for recorded occurrences of tsuna- mis within our study site. Uninhabited islands are not well represented in the data set. For each occurrence we noted the location of the tsunami, its maximum run-up height, and its point of or- igin. We then classified our island groups as ei- ther susceptible to tsunamis or unaffected (Table 3). The earliest recorded tsunami in our study area occurred in 1843. Since then over 130 tsu- namis have been recorded. The Hawaiian Is- lands have seen the most tsunamis, a result of their central location relative to areas of seismic activity around the Pacific Rim and the lack of any energy-absorbing shelves around the group. Figure 10 shows regions affected by tsunamis and, when known, the direction traveled by tsu- namis from their point sources. HOW MANY BiRD SPECIES--Curnutt and Pimm 29 We have data on tsunamis for 21 of our 41 island groups. Ten of these, however, have max- imum recorded run-up heights of zero. That is, tsunami events do not noticeably affect these groups. Many of these fortunate island groups are low relief, including the extensive Marshall Islands. Ten of the remaining eleven groups are high relief and have experienced run-up heights from less than 2.0 to 16.8 m. The sole low-relief group affected by tsunamis is the Tuamotu Ar- chipelago with a maximum run-up of 2.3 m. The disturbance caused by tsunamis on high- relief islands is primarily limited to coastal ar- eas, below the altitudinal distribution of most of the species we are concerned with. The effect of tsunamis on the fauna of the Tuamotu Archipel- ago, however, could be devastating. Most of the islands of this group are only a few meters in elevation, and the combined effect of higher sea level during the mid- and late-Holocene with tsunamis helps explain why this group has fewer endemics than we predict based on its size and topography. Finally, the Tonga group experi- enced a maximum run-up height of 4.0 to 6.0 m. This group is dominated by high-relief is- lands; however, 193 kul 2 of its total 563 km 2 consists of low-relief islands. Assuming tsuna- mis were frequent and devastating enough to prevent endemism on these low islands, we can calculate a refined estimate of the number of en- demics for this group by excluding all low-relief islands. This exercise results in the loss of one species of rail and one species of pigeon, leaving 35 rails, 12 pigeons, and 4 parrots attributed to the Tonga group. Combining the effects of sea-level change and tsunamis, we can refine our previous estimate of predicted endemic species in the Pacific as fol- lows: 199 endemic rails, 38 endemic parrots, and 95 endemic pigeons and doves (Fig. 7). PROBLEM GROUPS Even after incorporating the above adjust- ments to our predicted numbers of species, ac- tual species account for less than half of the pre- dicted numbers for 13 of the 23 high-relief groups. Six groups (Rotuma, Tabuai, Wallis and Futuna, Yap, Tonga, and Kermadec) have no ac- tual island-group endemics although we predict from two to five species for these groups. For low-relief groups, 10 (Nauru, Northwest Ha- wai'i, Tokelau, Tuvalu, Gilbert, Niue, Phoenix, Chuuk, Marshall Islands, North Line Islands) have no actual single-island endemics although we predict from 1 to 20 species for these groups. In all, we predicted 210 species of rails, pigeons, and parrots that are not accounted for as either fossil, extinct, or extant. DISCUSSION We estimate that there were approximately 330 species of rails, pigeons, and parrots on the islands of the Pacific before human colonization began 4,000 years ago. Approximately one-third of these species are accounted for as either ex- tant, historically extinct, or as fossils. Pimm et al. (1994), who looked for all landbirds, pre- dicted that the fossil record was only half com- plete and that the original avifauna was about 800 species. In reviewing the fossil, historical, and current data, we could account for only one- third of the estimated number of species in the taxa we looked at. We should therefore apply a three-fold correction to the total number of known landbirds (540) and conclude that the en- tire Pacific landbird fauna was comprised of 1,620 or so species before human colonization. This simple multiplication, however, ignores dif- ferences in extinction rates between taxa. Stead- man (1997a,b) suggested that flightless rails suf- fered a greater proportion of extinctions than any other taxon of birds. If so, an estimate of 1,500 species would be too high. In comparing our results to Steadman's (1995) estimates, we must limit our consideration to rails--the only taxon that Steadman makes a quantitative estimate of. We estimate that the prehuman Pacific held about 200 species of rails, of which 21 are extant. Steadman's (1995) esti- mate (2,000+ species of rails) is an order of magnitude greater than ours. Like Steadman, we based our analyses on the roughly 800 larger islands of the Pacific. However, where Steadman simply multiplied a maximum number of rails per island by the number of islands, we incor- porated into our analyses statistical probabilities and geographical, topographical and environ- mental data. Thus, we believe that Steadman's (1995) estimate of the prehuman avifauna is too high. More fieldwork will inevitably bring new data to light. The discovery of more fossil species will potentially alter our estimates because of the multiplicative nature of our analyses. The dis- covery of one new fossil rail on a small island could conceivably add 800 to our current esti- mate of 200. This would still be half as much as the highest proposed number of rails (Stead- man 1995). Currently, we suggest that the pre- human avifauna consisted of more than 800 and less than 1,500 species of landbirds. Further re- search (as outlined below) is needed to refine our estimates and to conserve the remaining spe- cies of the Pacific islands. CONSERVATION CONCERNS The loss to extinction of even our lowest pre- dicted number of endemic species is disturbing. 30 STUDIES IN AVIAN BIOLOGY NO. 22 Much more disturbing is the potential effect of this prehistoric loss on the biodiversity of the future. Habitat loss and the introduction of ex- otic species have had profound negative effects on endemic Pacific landbirds (Atkinson 1985, Pimm 1987). For rails, some of the progenitors of the clan of now extinct endemics may have themselves become extinct and anthropogenic disturbance on many islands may make recolon- ization by extant rails impossible. Thus, even for a rapidly speciating taxon like flightless rails, the potential for diversity has been greatly dimin- ished. Another conservation concern for Pacific landbirds is the rise of global sea levels. Al- though predictions of the rate of sea-level rise are rife with uncertainty, it is clear that global warming and subsequent rises in sea level will occur for centuries into the future (Hutter et al. 1990). Even with a moderate estimate of 4 to 6 cm per decade (Hutter et al. 1990, Wigley and Raper 1993) many low-relief islands will be in- undated within the next few centuries. FUTURE RESEARCH OPPORTUNITIES Our predictions of the prehistoric Pacific is- land avifauna are testable. Using our results, re- searchers can focus excavation efforts on those islands that we predict will hold fossils of the greatest number of extinct species. Thus, we provide our analyses and results as a guide for continued work in this area of biodiversity. We conclude with the following suggestions for fur- ther study: Where to look for subfossil birds We predict that the greatest number of extinct landbirds existed on high-relief islands of at least I km 2 in size. The greatest part of the "missing" rails are from Fiji and Vanuatu. These areas should be surveyed intensely for subfossil remains. Searches should, perhaps, also include island shelves that are currently submerged. Gibbons and Clunie (1986) make a strong argument for extending archeological ex- cavations to these areas because they were ex- posed and possibly colonized during the human expansion into the Pacific. Analyze the loss of potential species richness A thorough understanding of the phylogenetic relationship between the landbird species of the Pacific would serve to identify the mechanisms of speciation and the ancestral species that most contribute to the potential diversity of each tax- on. A molecular genetic analysis and mapping of the relationship of these species may also un- cover phylogenetic differences in speciation rates, dispersal, and habitat utilization. Predict the ef/kcts of rising sea level on current bird diversity We have described the effect of area and to- pography on bird species diversity. Currently, models are available that predict changes in sea level both globally and regionally (Wigley and Raper 1993). The application of sea-level change projections to Pacific islands would re- sult in predicted size distributions of islands, to which our approach can be applied. This will allow us to predict the expected loss of bird spe- cies in the Pacific in the coming century. These analyses, coupled with more traditional efforts (e.g., Franklin and Steadman 1991) could also be used to map a survival strategy for Pacific biodiversity in light of the threat of future sea- level rise. ACKNOWLEDGMENTS We thank M. Moulton, D. Steadman, K. Reese, R. Walker, S. Conant, and one anonymous reviewer for their comments. SLP thanks the Pew Fellowship in Conservation and the Environment for support. Studies in Avian Biology No. 22:31-46, 2001. PATTERNS OF SUCCESS AMONG INTRODUCED BIRDS IN THE HAWAIIAN ISLANDS MICHAEL P. MOULTON, KARL E. MILLER, AND ERIC A. TILLMAN Abstract. At least 140 species of 14 different orders of birds have been introduced to the six main Hawaiian Islands. The introduced species came from six continents and the introductions were carried out by a variety of agents including state and local governments, private citizens, and the acclimati- zation society known as the Hui Manu. The introductions mostly occurred during the early to mid- twentieth century. Most (79%) of the intentional introductions were of species from three orders: Galliformes, Columbiformes, and Passeriformes. Introduction success rates were significantly greater for passeriforms than for either columbiforms or galliforms, although the reasons for this are unknown. In predicting the fate of future introductions, only the columbiforms showed an "all-or-none" pattern of introduction history. Successful species had larger native geographic ranges than did unsuccessful species, which supports the hypothesis that range size is correlated with the ability to adapt to a new environment. Finally, in a partial test of the introduction effort hypothesis we found that galliforms successfully introduced to the island of Hawai'i were introduced in significantly larger numbers than unsuccessful species. Key Words: doves; game birds; introduced species; introduction effort; introduction success; native range size; perching birds; pigeons. Numerous species of birds from six continents have been introduced to the Hawaiian Islands (Caum 1933, Berger 1981, Long 1981, Pratt et al. 1987). These species were introduced by a variety of groups for a variety of reasons. As noted by Berger (1981), the first avian introduction came with early Polynesians who brought the Red Jun- glefowl (Gallus gallus) for food. Since that time, a number of private citizens have brought species to Hawai'i (e.g., Caum 1933). Some of these in- troductions were made inadvertently as individual birds escaped captivity (e.g., Melodious Laughing- thrush or Hwamei, Garrulax canorus, on O'ahu), whereas others were intentionally released for aes- thetic reasons or even as an attempt at biological control (Caum 1933). There also have been inten- sive efforts both by private citizens (e.g., Lewin 1971) as well as state and county agencies (Schwartz and Schwartz 1949; Walker 1966, 1967) to establish populations of various game birds for recreational hunting. In the early to mid-twentieth century, the acclimatization society known as the Hui Manu actively introduced several species to various islands (Caum 1933, Berger 1981). Regardless of their source, a central question in any study of introduced birds is "Why do some species succeed and others fail?" In sev- eral papers we and our colleagues have argued that competition has played an influential role in determining the outcomes of passerine species' introductions in Hawai'i (Moulton and Pimm 1983, 1986a, 1987; Moulton 1985, 1993; Moun- tainspring and Scott 1985; Moulton et al. 1990; Moulton and Lockwood 1992). These arguments are based on three main findings. First, intro- ductions tend to be less successful when more species of introduced birds are already present (Moulton 1993; Moulton and Pimm 1983, 1986a). Second, there is a pattern of limiting similarity among congeneric pairs of introduced birds: differences in bill length are significantly greater in pairs that coexist than in pairs of spe- cies that were not able to coexist (Moulton 1985). And third, successful introduced passer- ines show a pattern of morphological overdis- persion (Moulton and Pimm 1987, Moulton and Lockwood 1992); i.e., successful species are morphologically more different from each other than expected by chance. Although these three patterns are consistent with predictions from competition theory, other explanations for patterns in introduction out- comes have been advanced. These include intro- duction history of a species (Simberloff and Boecklen 1991) and introduction effort (e.g., Pimm 1991, Veltman et al. 1996). The idea that introduction history can predict future introduction outcomes is appealing in its simplicity. The concept comes from Simberloff and Boecklen (1991) who argued that whenever and wherever a given species is introduced, it tends to either always succeed or always fail. This leads to an "all-or-none" pattern in the dis- tribution of birds introduced onto a series of is- lands: some species being successful on "all" the islands in the series and others being suc- cessful on "none" of the islands. If introduced birds actually follow this pattern, then predicting the outcome of future introductions would be greatly simplified. Moulton (1993) and Moulton and Sanderson (1997), however, argued that the all-or-none pattern reported by Simberloff and Boecklen (1991) for passerine birds was pri- marily an artifact of sample size. 31 32 STUDIES IN AVIAN BIOLOGY NO. 22 Another factor that might influence the out- come of introductions is the effort invested in the introduction process. Griffith et al. (1989) found that introduction effort along with habitat quality were associated with introduction out- come. Similarly, Pimm (1991 ) studied introduc- tions of seven game bird species (all of which had been successfully introduced somewhere in the world) in the western United States and found that there was a very high (360/424 = 85%) failure rate. Pimm's analysis indicated that the failure rate was particularly high when fewer than 75 individuals were released. More recent- ly, studies of introduced birds in New Zealand (Veltman et al. 1996, Duncan 1997, Green 1997) have concluded that introduction effort is the most influential variable in determining which species succeed. In each of the three studies, the authors reported that successful species were in- troduced in larger numbers and more frequently than were unsuccessful species. Several authors have reported a positive re- lationship between the size of the native geo- graphic range of a species and its average abun- dance (e.g., Bock and Ricklefs 1983, Brown 1984). If widespread species tend to be ecolog- ically more generalized than species with narrow distributions, we would predict that successful introduced species would tend to be those that have larger native ranges. Many analyses of avian introduction success in Hawai'i have focused on passerine birds (e.g., Moulton and Pimm 1983, 1986a,b, 1987; Wil- liams 1987, Moulton and Lockwood 1992), yet passerines represent fewer than half the total number of birds that have been introduced to the Hawaiian Islands (Berger 1981, Long 1981). Our objectives in this paper were to examine patterns of success for introduced species in Ha- wai'i across three taxonomic orders of birds: Galliformes, Columbiformes, and Passeriformes. Specifically, are the success rates of nonpasser- ine birds different from those of the passerines? Second, is an all-or-none pattern evident in the nonpasserine orders? Third, is native range size greater for successful introduced species than for unsuccessful introduced species in passerines and nonpasserines? And, fourth, does introduc- tion effort play a role in determining the success of introduced birds in Hawai'i? METHODS AND MATERIALS We used Caum (1933), Schwartz and Schwartz (1949), Munro (1960), Walker (1966, 1967), Lewin ( 1971), Berger ( 1981 ), Long ( 1981 ), Lever (1987), and Pratt et al. (1987) to compile lists of nonindigenous birds introduced to the Hawaiian Islands. In compiling our lists we attempted to ascertain not only the current status of each species but also the date of first intro- duction. In our analyses we considered species to be successful if they were present on an island in 1990. We considered species to be unsuccessful if there were no recorded observations after 1990. Scientific names of 140 species introduced in the Hawaiian Islands are provided in Appendix 1. Scientific names of intro- duced species not included in our statistical analyses are provided in Appendix 2. In order to determine success rates for the species in the different orders, we considered a species to be suc- cessful if it succeeded on any island, and unsuccessful only if it failed on every island on which it was released. By this approach, even if a species fails on all but one island, we believe that environmental conditions in the archipelago overall were potentially suitable for establish- ment and that perhaps differences in the mechanics of the release or interactions with other species might have oc- curred on islands where the species failed. We compared introduction success rates across orders with a chi-square test of equal proportions. We used range maps in Long (1981) to estimate native range size for all introduced species, except Garrulax caerulatus and Callipepla douglasii, which were not included by Long. We used a grid method similar to the methods of Moulton and Pimm (1986b). We placed a small acetate grid over the native range map in Long (1981) and counted the number of squares that were intersected. Each square represented approximately 259,000 km 2. In earlier analyses of na- tive range size of introduced passerines in Hawai'i (Moulton and Pimm 1986b), Uraeginthus angolensis and U. cyanocephala were omitted because of concern about the potential confusion with young U. bengalus in the field. However, we included all three Uraegin- thus species in this analysis because Berger (1981) re- ported each was seen and identified in the wild. We used Mann-Whitney tests for all our range size comparisons because data were not normally distrib- uted. We compared native geographic range sizes of successful versus failed introductions, both within and across orders. RESULTS At least 140 species of nonindigenous birds from 14 orders have been released in the Ha- waiian Islands (Table 1). Our results differ from earlier totals of 162 species (Long 1981) and 170 species (Berger 1981) for two reasons. First, those authors followed a somewhat different tax- onomy. For example, Berger (1981) listed the Green Pheasant (Phasianus versicolor) as being a distinct species, whereas we followed Sibley and Monroe (1990) and treated it as being con- specific with the Ring-necked Pheasant (Phasi- anus colchicus). Second, at least among the pas- serines, we have excluded several species in- cluded by Long and Berger on grounds that sim- ply too few individuals (i.e., < 5) were released. Simberloff and Boecklen (1991) list 14 of these species in their Appendix B, although based on Berger (1981 ) we included the two Uraeginthus species (U. angolensis and U. cyanocephala). Although a great diversity of species has been released into the Hawaiian Islands, for the most INTRODUCED BIRDS--Moulton, Miller, and Tillman 33 TABLE 1. SPECIES OF BIRDS INTRODUCEO TO THE HA- WAIIAN ISLANDS (CAUM 1933, BEI.GER 1981, LONG 1981) Number of Order species Tinamiformes 1 Pelecanifonnes 1 Ciconiifonnes 3 Falconiformes 1 Galliformes 40 Turniformes 1 Gruifonnes 1 Charadriiformes 2 Anserifonnes 4 Columbiformes 18 Psittacifonnes 14 Strigifonnes 1 Apodifonnes 1 Pas seritbnnes 52 part three orders accounted for the bulk of the introductions. These are the game birds (Galli- formes), pigeons and doves (Columbiformes), and perching birds (Passeriformes). These spe- cies represent 110 introductions (Appendices 3- 5). Berger (1981) lists 14 species of a fourth order, Psittaciformes. However, according to Berger (1981), 13 of these species were acci- dental introductions. Moreover, Pratt et al. (1987) considered only one species of this order (Psittacula krameri) to be successful in Hawai'i. Thus, we restricted our tests to the three orders for which there was evidence for intentional in- troductions: Galliformes, Columbiformes, and Passeriformes. HISTORICAL PERSPECTIVE In order to develop a historical perspective on the phenomenon of introductions for the galli- forms, columbiforms, and passeriforms, we cat- egorized introductions by time period (Fig. 1). Historical peaks in the number of introductions were evident for each order. For galliforms, the number of species' intro- ductions increased steadily from 1901 until the early 1960s and then declined to zero. There has not been an introduction of a new species of galliform into the Hawaiian Islands since 1965 (Francolinus adsperus). For columbiforms, the peak occurred in the 1920s. Indeed, there have been only two introductions (Zenaida asiatica in 1961 and Zenaida macroura in 1962) of species from this order since 1960. The passeriforms also appear to show a decline in the number of introductions after the 1960s (Fig. 1). Closer in- spection reveals an even sharper decline in the frequency of introductions, with only one new passerine species introduced since 1980 (Estril- da astrild in 1981). The remaining nine species were all present on other islands in the archi- pelago prior to 1975 and possibly arrived onto new islands via interisland colonization. SUCCESS RATES Success rates differed significantly among or- ders (X 2 = 14.59, df = 2, P < 0.005). Among 25 pre-1876 1876-1900 1901-25 1926-50 Year 1951-75 post-1975 ß Galliformes []Columbiformes [] Passeriformes ] FIGURE 1. Chronology of species introductions to the Hawaiian Islands. 34 STUDIES IN AVIAN BIOLOGY NO. 22 0.8 0.6 0.4 0.2 o Hawaii Maui Lanai Molokai Oahu Kauai Island [] Galliformes [] Columbiformes [] Passeriformes FIGURE 2. Success rates (number of successful introductions/total number of introductions) per order across the six main Hawaiian Islands. passetines, 33 of 52 (64%) species have been successful on at least one island (Appendix 3). The success rates for galliform and columbiform species were not nearly so high. Only 12 of 40 (30%) introduced galliform species (Appendix 4) and 4 of 18 (22%) introduced columbiform species (Appendix 5) have been successful on at least one island. Within islands the success rates also were variable (Fig. 2). For passetines, Moloka'i and Lana'i shared the highest rates of success at 1.00 (13/13 for Moloka'i and 11/11 for LUna'i). Lana'i also had the highest success rate for gal- liforms (9/15, 0.60), whereas Moloka'i had the highest rate for columbiforms (3/4, 0.75; Fig. 2; Table 2). Although it is tempting to compare rates among islands across the different orders, results of any tests would be misleading because of the high potential for nonindependence. For example, with respect to passetines, only seven species were introduced to islands other than TABLE 2. SUCCESS RATES (NUMBER OF SUCCESSFUL INTRODUCTIONS/ToTAL NUMBER OF INTRODUCTIONS) PER ORDER ACROSS THE SIX MAIN HAWAIIAN ISLANDS Island Galliformes Columbiformes Passedformes Hawai'i 0.32 0.56 0.80 Maui 0.45 0.27 0.84 Lana'i 0.53 0.43 1.00 Moloka'i 0.53 0.60 1.00 O'ahu 0.26 0.33 0.60 Kaua'i 0.40 0.375 0.75 O'ahu (five to Kaua'i and two to Hawai'i). For galliforms, only O'ahu and Hawai'i have any unique species. ALL-OR-NONE PATTERNS The hallmark of an all-or-none distributional pattern of introduced birds on islands would be presence of few, if any, mixed species. Mixed species are those that are successful on some islands and unsuccessul on others (Simberloff and Boecklen 1991). In principle, species re- leased onto one island could show a mixed out- come if they spread to another island and then fail on one of the two islands. In practice this is very difficult to detect, because those species with the ability to spread to other islands could do so repeatedly giving the impression that they were established on the second island even if they were actually not able to survive there. This would be an example of what Brown and Ko- dric-Brown (1977) have termed a "rescue ef- fect." With this in mind we believe that analyses for all-or-none patterns should be limited to those species that were physically introduced to more than one island. In their analysis of introduced Hawaiian birds, Simberloff and Boecklen (1991) reported that among 19 introduced columbiform species, only one (Pterocles exustus) showed a mixed out- come, having succeeded on Hawai'i, and failed on Moloka'i and Kaua'i. However, Sibley and Monroe (1990) placed this species in the order Ciconiiformes. If this species is excluded, 18 columbiform species remain, 11 of which were INTRODUCED BIRDS--Moulton, Miller, and Tillman 35 30 25 ß - 2O ta 15 o 5 10 5 0 I IUnsuccessful []Successful <10 10-20 20-30 30-40 40-50 50-75 Size of Native Range >75 FIGURE 3. Size of native geographic range for unsuccessful versus successful introduced species. Range size measured in number of 259,000 km 2 map blocks (see Methods and Materials). introduced onto more than one island. The ob- servation of no mixed species out of eleven pos- sible is evidence for an all-or-none pattern. Among the Galliformes, 23 species were re- leased onto two or more islands. At least seven of 23 (30%) have had mixed outcomes: Calli- pepla californica, C. gambelii, Alectoris chukar, Coturnix japonica, Gallus gallus, Pavo crista- tus, Meleagris gallopavo (Appendix 4). Thus, there is not an all-or-none pattern among the in- troduced game birds. Moulton and Sanderson (1997) and Moulton (1993) argued that mixed species tended to be those introduced onto more islands. With this in mind we compared the median numbers of is- lands of introduction for always unsuccessful, always successful, and mixed species. Medians differed significantly (H = 8.95, P = 0.012), with the highest median recorded for mixed spe- cies (6.0), and medians of 3.0 for always unsuc- cessful species and 6.0 for always successful species. As a further test we combined species that were always successful with those that were always unsuccessful and compared the com- TABLE 3. RESULTS OF NATIVE RANGE SIZE COMPARI- SONS BETWEEN SUCCESSFUL (S) AND UNSUCCESSFUL (F) INTRODUCED SPECIES (MEAN NUMBER OF 259,000 KM 2 GRID SQUARES) Order S F P Galliformes 22.2 16.6 0.099 Columbiformes 36.1 23.3 0.123 Passeriformes 33.0 17.5 0.015 bined median with that of the mixed species. These medians also differed significantly (H = 5.23, P = 0.022). RANGE SIZE We estimated native range size tbr 108 intro- duced species. Range size was significantly larger in successful species than in unsuccessful species (approximate X2 = 10.95, df = 1, P < 0.001; Fig. 3; Table 3). Within orders, range size differences were significant for passefines (P = 0.015) and marginally significant in game birds (P = 0.099). However median range size did not diftr signifi- canfly between successful and unsuccessful col- umbiforms (P = 0.123). In all three orders, the successful species had larger median native range sizes than did unsuccessful species. INTRODUCTION EFFORT Data are available for only a partial test of the influence of introduction eftbrt on introduction success. Lewin (1971) provided numbers of indi- viduals released for 26 Galliformes on the island of Hawai'i (Table 4). Most of the data were de- rived from private releases by the owners of the Pu'u Wa'awa'a Ranch, but in some instances data from releases made by state agencies were includ- ed. We excluded Coturnix japonica because it al- ready was successful, apparently having colonized the island froln Maui and/or Lana'i (Schwartz and Schwartz 1949), and there were no further releases by the state or the owners of the ranch. The me- dian number of individuals introduced was 179 for successful galliform species (N = 9) and 14 for unsuccessful galliform species (N = 17). Medians 36 STUDIES IN AVIAN BIOLOGY NO. 22 TABLE 4. NUMBER OF INDIVIDUALS OF SPECIES OF GAME BIRDS, PIGEONS, AND DOVES RELEASED ON HAWAI'I (LEwIN 1971) Number Species released Status Colinus virginianus a Oreortyx pictus Callipepla squamata Callipepla californica Callipepla gambelii Callipepla douglasii Ammoperdix griseogularis Cyrtonyx montezumae Alectoris chukar Alectoris barbara Francolinus francolinus Francolinus pintadeanus Francolinus pondicerianus Francolinus adsperus Francolinus icterorhynchus Francolinus clappertoni Francolinus erckelii Francolinus leucoscepus Coturnix chinensis Bambusicola thoracica Lophura leucomelanos Gallus sonneratii Phasianus colchicus Syrmaticus reevesii Pavo cristatus Meleagris gallopavo Zenaida macroura Zenaida asiatica Streptopelia risoria (= decaocto ?) Streptopelia chinensis Geopelia striata 108 F 88 F 14 F 412 S 546 F ll3 F 20 F 8 F 110 S 104 F 226 S 10 F 214 S 4 F 9 F 10 F 179 S 27 F 8 F 12 F 67 S 14 F 244 S 180 F 2 S ll5 S 168 S 40 F ll F 8 S 18 S a See Appendix 1 for common names. were significantly different in a Kruskal-Wallis test (H = 5.25, P = 0.02). Data for the columbiforms appear to be equal- ly compelling, although we have not tested this group since there were just seven species intro- duced and two of these already were established on Hawai'i at the time of the introductions by the Pu'u Wa'awa'a Ranch (Lewin 1971). DISCUSSION The introduction process in the Hawaiian Is- lands has been highly nonrandom with respect to phylogeny. Thus 10 of the 14 orders are repre- sented by five or fewer species. The three orders that are represented by more species are those that have been the focus of intentional introductions. Thus, most galliforms were likely introduced to enhance prospects for recreational hunting, and most columbiforms were introduced for recrea- tional hunting or lbr aesthetic reasons. Passefines were introduced for a variety of reasons, including biological control and aesthetic reasons, as well as accidental releases of cage birds. The phenomenon of avian introductions, at least for the three orders we have focused on here, ap- pears to be historical, with most introduction ef- forts having come to a close. There have been no columbiform or galliform introductions to the Ha- waiian Islands in more than 30 years. Moreover, no new passefine species have been introduced to the islands since 1981. This is not to say that there will not be future introductions from these, or oth- er, taxa. Indeed, there have been recent sightings of various parrot species since 1990. For example, Pyle (1994) reported that 10 to 15 Nanday Para- keets (Nandayus henday) were seen on the island of Hawai'i. In terms of success rates, we found that pas- serine species had a significantly higher overall success rate than either of the nonpasserine or- ders. The reasons for this are unclear, but the pattern is highly significant. It is possible to ex- plain some of this result via the propagule size hypothesis. We found a significant relationship between propagule size (i.e., introduction eflbrt) and the success rates of galliforms introduced to the island of Hawai'i. Caum (1933) also noted that several columbiform species apparently were introduced in very small numbers. How- ever, it remains to be shown that passerines were systematically released in larger numbers. The simplest potential predictor of the out- come of species' introductions is introduction history (Simberloff and Boecklen 1991). If in- troduction history alone were an adequate pre- dictor of introduction outcomes we should have detected clear all-or-none patterns within the or- ders we analyzed. Moulton (1993) and Moulton and Sanderson (1997) argued that the all-or-none patterns reported for passerines introduced to the Hawaiian Islands and elsewhere may be due to sampling artifact. When we extended the anal- ysis here to include the columbiforms and gal- liforms, only the columbiforms show any evi- dence for such a pattern. Thus, we found little evidence to support the notion that introduction history is an adequate predictor of future intro- duction outcomes. Our analyses suggested that one consistent predictor of introduction success was size of na- tive geographic range. In all three orders we ob- served that successfully introduced species had larger native ranges than unsuccessful species. These results are consistent with the hypothesis that species with larger ranges are ecologically more generalized (Brown 1984) and hence better able to adapt to a new environment. In a partial test of the introduction eftbrt hy- pothesis, we found that galliforms introduced suc- cessfully to Hawai'i were introduced in larger numbers than were unsuccessful species. However, it should be noted that some species were suc- cessful with initial releases of as few as two in- dividuals; e.g., a single pair of Peafowl (Pavo cris- tams) released on the Pu'u Wa'awa'a Ranch in 1909 led to the successful establishment of the INTRODUCED BIRDS--Moulton, Miller, and Tillman 37 species on Hawai'i (Lewin 1971). Also, for 6 of the 15 unsuccessful species, >85 individuals were released (Colinus virginianus, Callipepla dougla- sii, Callipepla gambelii, Syrmaticus reevesii, Or- eotyx pictug Alectoris barbara; Table 4). We do not know if successful game birds on islands other than Hawai'i were introduced in higher numbers than were unsuccessful species. Because data are lacking for passeriform and columbiforrn species, a thorough test of the introduction effort hypoth- esis was not possible. ACKNOWLEDGMENTS We thank J. M. Scott, S. Conant, R. Walker, B. Den- nis, and an anonymous reviewer for their comments on earlier versions of the manuscript. Florida Agricul- tural Experiment Station JS #R-07766. A. van Doorn assisted with review of the literature of psittaciform introductions. MPM wishes to thank C. J. Ralph, C. P. Ralph, and A. C. Ziegler for their kindness and hos- pitality during fieldwork in Hawai'i. APPENDIX 1. SCIENTIFIC AND COMMON NAMES OF 142 SPECIES INTRODUCED TO THE HAWAIIAN ISLANDS (NOMEN- CLATURE FOLLOWS SIBLEY AND MONROE 1990) Scientific name Common name Acridotheres tristis Agapornis roseicapillis Alauda arvensis Alectoris barbara Alectoris chukar Amandava amandava Amazona ochrocephala Amazona viridigenalis Ammoperdix griseogularis Anas discors Anas platyrhynchos Ara macao Bambusicola thoracica Brotegeris jugularis Bubulcus ibis Cacatua galerita Cacatua moluccensis Callipepla californica Callipepla douglasii Callipepla gambelii Callipepla squamata Caloenas nicobarica Cardinalis cardinalis Carpodacus mexicanus Cettia diphone Chalcophaps indica Chrysolophus amherstiae Chrysolophus pictus Colinus virginianus Collocalia vanikorensis Columba livia Copsychus malabaricus Copsychus saularis Coturnix chinensis Coturnix japonica Coturnix pectoralis Crax rubra Cyanoptila cyanomelana Cygnus olor Cyrtonyx montezumae Eclectus roratus Eolophus roseicapilla Erithacus akahige Erithacus komadori Estrilda astriM Estrilda caerulescens Estrilda melpoda Common Myna Rosy-faced Lovebird Skylark Barbary Partridge Chukar Red Avadavat Yellow-crowned Parrot Red-crowned Parrot See-see Partridge Blue-winged Teal Mallard Scarlet Macaw Chinese Bamboo-Partridge Orange-chinned Parakeet Cattle Egret Sulphur-crested Cockatoo Salmon-crested Cockatoo California Quail Elegant Quail Gambel's Quail Scaled Quail Nicobar Pigeon Northern Cardinal House Finch Japanese Bush-Warbler Emerald Dove Lady Amherst Pheasant Golden Pheasant Northern Bobwhite Uniform Swiftlet Rock Pigeon White-rumped Shama Oriental Magpie-Robin Blue-breasted Quail Japanese Quail Stubble Quail Great Currasow Blue-and-White Flycatcher Mute Swan Montezuma Quail Eclectus Parrot Galah Japanese Robin Ryukyu Robin Common Waxbill Lavendar Waxbill Orange-cheeked Waxbill 38 APPENDIX 1. CONTINUED. STUDIES IN AVIAN BIOLOGY NO. 22 Scientific name Common name Estrilda troglodytes Falco (rusticolus ?) Francolinus ada7verus Francolinus clappertoni Francolinus erckelii Francolinus francolinus Francolinus icterorhynchus Francolinus leucosepus Francolinus pintadeanus Francolinus pondicerianus Gallicolumba luzonica Gallus gallus Gallus sonneratii Garrulax albogularis Garrulax caerulatus Garrulax canorus Garrulax chinensis Garrulax pectoralis Geopelia cuneata Geopelia humeralis Geopelia striata Geophaps lophotes Geophaps plumikra Geophaps smithii Geotrygon montana Gracula religiosa Grallina cyanoleuca Lagonosticta senegala Larus novaehollandiae Larus occidentalis Leiothrix lutea Leptotila verreauxi Leucosarcia melanoleuca Lonchura cantans Lonchura malacca Lonchura oryzivora Lonchura punctulata Lophura leucomelanos Lophura nycthemera Melanocorypha mongolica Meleagris gallopavo Melopsittacus undulatus Mimus polyglottos Myiopsitta monachus Nandayus henday Neochen jubata Nothoprocta perdicaria Numida meleagris Oreortyx pictus Ortalis cinereiceps Paroaria capitata Paroaria coronata Paroaria dominicana Parus varius Passer domesticus Passerina ciris Passerina cyanea Passerina leclancherii Pavo cristatus Penelope purpurascens Perdix perdix Phalacrocorax carbo Phaps chalcoptera Phasianus colchicus Phoenicopterus ruber Black-rumped Waxbill Gyrfalcon? Red-billed Francolin ClappertoWs Francolin Erckel's Francolin Black Francolin Heuglin's Francolin Yellow-necked Spurfowl Chinese Francolin Grey Francolin Luzon Bleeding-Heart Red Junglefowl Grey Junglefowl White-throated Laughingthrush Grey-sided Laughingthrush Hwamei Black-throated Laughingthrush Greater Necklaced Laughingthrush Diamond Dove Bar-shouldered Dove Zebra Dove Crested Pigeon Spinifex Pigeon Partridge Pigeon Ruddy Quail-Dove Hill Myna Magpie-Lark Red-billed Firefinch Silver Gull Western Gull Red-billed Leiothrix White-tipped Dove Wonga Pigeon African Silverbill Black-headed Munia Java Sparrow Scaly-breasted Munia Kalij Pheasant Silver Pheasant Mongolian Lark Wild Turkey Budgerigar Northern Mockingbird Monk Parakeet Nanday Parakeet Orinoco Goose Chilean Tinamou Helmeted Guineafowl Mountain Quail Grey-headed Chachalaca Yellow-billed Cardinal Red-crested Cardinal Red-cowled Cardinal Varied Tit House Sparrow Painted Bunting Indigo Bunting Orange-breasted Bunting Common Peafowl Crested Guan Grey Partridge Great Cormorant Common Bronzewing Ring-necked Pheasant Greater Flamingo INTRODUCED BIRDS Moulton, Miller, and Tillman APPENDIX 1. CONTINUED. 39 Scientific name Common name Platycercus adscitus Porphyrio porphyrio Psittacula krameri Pterocles exustus Pycnonotus cafer Pycnonotus jocosus Rhipidura leucophrys Rollulus rouloul Serinus leucopygius Serinus mozambicus Sicalis fiaveola Streptopelia chinensis Streptopelia decaocto Sturnella loyca Sturnella neglecta Syrmaticus reevesii Syrmaticus soemmerringii Tiaris olivacea Turnix varia Tympanuchis cupido Tympanuchus phasianellus Tyro alba Uraeginthus angolensis Uraeginthus bengalus Uraeginthus cyanocephala Urocissa erythrorhyncha Vidua macroura Zenaida asiatica Zenaida macroura Zosterops japonicus Pale-headed Rosella Purple Swamphen Rose-ringed Parakeet Chestnut-bellied Sandgrouse Red-vented Bulbul Red-whiskered Bulbul Willie-Wagtail Crested Partridge White-romped Seedeater Yellow-fronted Canary Saffron Finch Spotted Dove Eurasian Collared-Dove Long-tailed Meadowlark Western Meadowlark Reeve's Pheasant Copper Pheasant Yellow-faced Grassquit Painted Buttonquail Greater Prairie Chicken Sharp-tailed Grouse Barn Owl Blue-breasted Cordonbleu Red-cheeked Cordonbleu Blue-capped Cordonbleu Red-billed Blue Magpie Pin-tailed Wydah White-winged Dove Mourning Dove Japanese White-Eye APPENDIX 2. LIST OF 31 SPECIES FROM I 1 ORDERS NOT INCLUDED IN STATISTICAL ANALYSES. WITHIN EACH CELL, THE FIRST LINE INDICATES DATE OF FIRST INTRODUCTION (OR FIRST REFERENCE TO INTRODUCTION) AND STATUS (S SUCCESSFUL; F = FAILED); THE SECOND LINE INDICATES MODE OF INTRODUCTION (1 = PRIVATE; 2 = STATE OR COUNTY AGENCY; 3 = UNKNOWN, INCLUDES ESCAPE FROM CAPTIVITY; 4 = POLYNESIANS; 5 = HuI MANU); AND THE THIRD LINE INDICATES REFERENCE Species O'ahu Kaua'i Maul Hawai'i Moloka'i Lna'i Nothoprocta perdicaria 1966 F 2 1 Phalacrocorax carbo Phoenicopterus ruber Bubulcus ibis Pterocles exustus Falco (rusticolus?) a 1929 F 1 1 1959 S 1959 S 1959 S 1959 S 1 1 1 1 7 7 7 7 1961 F 1961 S 2 2 5,11 5,11 1929 F 1 1 1959 S 1 7 1961F 2 5,11 1890s F 1 1 1959 S 1 7 40 APPENDIX 2. CONTINUED. STUDIES IN AVIAN BIOLOGY NO. 22 Species O'ahu Kaua'i Maul Hawai'i Moloka'i Lana'i Turnix varia Poyphyrio porphyrio 1933 F 3 1 Larus novaehollandiae 1924 F 3 1 Larus occidentalis 1933 F 3 1 Cygnus olor Neochen jubata 1922 F 2 1 Anas platyrhynchos b Anas discors c 1932 F 2 1 Tyto alba 1959 S 2 7 Collocalia vanikorensis 1962 S 2 7 Brotegeris jugularis 1933 F 3 1 Cacatua galerita 1933 F 3 1 Cacatua roseicapilla 1933 F 3 1 Cacatua moluccensis 1981 F 3 7 Ara macao 1933 F 3 1 Melopsittachus undulatus 1933 F 3 1 Psittacula krameri 1933 S 3 1,7 Nandayus nenday 1981 F 3 7 Myiopsitta monachus 1970 F 3 7 Amazona viridigenalis 1971 U 3 7 1922 F 1922 F 2 2 I I 1933 F 3 1 1928 F 2 1 1920 F 1 1 1955 S 1 6 1981S 1981S 3 3 7,10 7,10 1981U 3 13 1959 S 1959 S 1958 S 1959 S 2 2 2 2 7 7 7 7,11 INTRODUCED BIRDS--Moulton, Miller, and Tillman APPENDIX 2. CONTINUED. 41 Species O'ahu Kaua'i Maul Hawai'i Moloka'i Lfina'i Amazona ochrocephala 1969 F 3 7 Eclectus roratus 1981 F 3 7 Agapornis roseicapillis 1973 F 3 7 Platycercus adscitus Urocissa erythrorhyncha 1966 F 1 7 1877 F 1 1 References: I - Caum 1933:2 Schwartz and Schwartz 1949:3 - Munro 1960; 4 Walker 1966:5 Walker 1967:6 Lewin 1971; 7 - Berger 1981; 8 - Moulton and Pimm 1983:9 - Scott et al. 1986:10 - Pratt et al. 1987; I 1 - Simberloff and Boecklen 1991; 12 - Moulton 1993; 13 - Pyle 1994:14 - Wunz 1992. a Caum (1933) listed F. rusticolus only as a tentative identification. b May have interbred with natural migrants, as well as fetal individuals. c Species identity uncertain. Caum (1933) stated the species is Querquedula discors (Blue-winged Teal, Anas discors); however, he also reported that the individuals came from Australia where the Blue-winged Teal does not occur. APPENDIX 3. INTRODUCED PASSERINES ON SIX MAIN HAWAIIAN ISLANDS (SEE APPENDIX 2 FOR EXPLANATION OF TERMS) Species O'ahu Kaua'i Maui Hawai'i Moloka'i LSna'i Acridotheres tristis 1872 S 1883 S 1 3 12 8 Alauda arvensis 1867 S 1870 F 3 1 12 1,8 Amandava amandava 1900 S 3 12 Cardinalis cardinalis 1929 S 1929 S 3,5 1 1,8 1,8 Carpodacus mexicanus 1870 S 1886 S 3 3 12 8 Cettia diphone 1929 S 1988 S 1,2 3 1 11 Copsychus malabaricus 1939 S 1931 S 5 1 11 1 Copsychus saularis 1932 F 1922 S 5 1 1 1,12 Cyanoptila cyanomelana 1929 F 2,5 1,8 Erithacus akahige 1929 F 2 1 Erithacus komadori 1931 F 3 8 Estrilda astrild 1981 S 3 12 1883 S 1883 S 3 3 8 8 1886 S 1902 S 3 3 8 8 1987 S 1987 S 3 3 11 11 1949 S 1929 S 3 2 8 1 1886 S 1886 S 3 3 8 8 1980 S 3 11 1937 F 5 8 1883 S 1883 S 3 3 8 8 1917 S 1917 S 3 3 8 8 1951S 1957 S 3 3 8 8 1886 S 1886 S 3 3 8 8 1979 S 1980 S 3 3 11 11 42 APPENDIX 3. CONTINUED. STUDIES IN AVIAN BIOLOGY NO. 22 Species O'ahu Kaua'i Maui Hawai'i Moloka'i Lfina'i Estrilda caerulescens 1965 S 1978 S 3 3 12 11 Estrilda melpoda 1965 S 1989 S 3 3 12 MPM Estrilda troglodytes 1965 F 1975 S 3 3 12 11 Garrulax albogularis 1919 F 1 1 Garrulax caerulatus 1947 S 3 8 Garrulax canorus 1900 S 1918 S 1902 S 1909 S 1909 S 3 1 1 1 1 1,8 1,8 1,8 1,8 1,8 Garrulax chinensis 1931 F 1 1 Garrulax pectoralis 1962 S 3 11 Gracula religiosa 1960 S 3 11 Grallina cyanoleuca 1922 F 1922 F 2 2 1 1 Lagonosticta senegala 1965 F 3 11 Leiothrix lutea 1928 S 1918 S 1928 S 1928 S 1928 S 2 1 2 2 2 1 1 1 1 1 Lonchura cantans 1984 S 1984 S 1978 S 1972 S 1981 S 3 3 3 3 3 11 11 11 11 11 Lonchura malacca 1936 S 1976 S 3 3 8 11 Lonchura oryzivora 1964 S 1983 S 1986 S 1981 S 3 3 3 3 12 11 11 11 Lonchura punctulata 1883 S 1883 S 1883 S 1883 S 1883 S 3 3 3 3 3 8,12 8 8 8 8 Melanocorypha mongolica 1914 F 1 8 Mimus polyglottos 1931 S 1946 S 1933 S 1959 S 1951 S 5 3 5 3 3 1,8 8 1,8 8 8 Paroaria capitata 1973 S 3 11 Paroaria coronata 1928 S 1928 S 1960 S 1976 S 1963 S 1,5 3 3 3 3 1,11 8,11 11 11 11 1979 S 3 11 1883 S 3 8 1970 S 3 11 1976 S 3 11 INTRODUCED BIRDS--Moulton, Miller, and Tillman APPENDIX 3. CONTINUED. 43 Species O'ahu Kaua'i Maui Hawai'i Moloka'i Lg. na'i Paroaria dominicana 1931 F 5 1 Parus varius 1928 F 1890 F 1928 F 1928 F 2 1 2 2 1,8 1,8 1,8 1,8 Pasxer domesticus 1871 S 1917 S 1917 S 1917 S 1917 S 1917 S 3 3 3 3 3 3 1,8 8 8 8 8 8 Passerina ciris 1937 F 5 8 Passerina cyanea 1934 F 1937 F 3 5 8 8 Passerina leclancherii 1941 F 1941 F 5 5 8 8,11 Pycnonotus cafer 1966 S 3 11 Pycnonotus jocosus 1965 S 3 11 Rhipidura leucophrys 1926 F 2 1,8 Serinus leucopygius 1965 F 3 11 Serinus mozambicus 1964 S 1977 S 3 1 11 11 Sicalis fiaveola 1965 S 1966 S 3 3 11 11 Sturnella Ioyca 1931 F 1 1 Sturnella neglecta 1931 F 1931 S 1934 F 2 1 3 8 1,11 3 Tiaris olivacea 1974 S 3 11 Uraeginthus angolensis 1965 F 3 7 Uraeginthus bengalus 1965 F 1973 S 3 3 11 11 Uraeginthus cyanocephala 1969 F 3 12 Vidua macroura 1962 F 3 12 Zosteropsjaponicus 1929 S 1929 S 1938 S 1937 S 1938 S 1938 S 2,5 5 3 5 3 3 1,11 1,11 8 8 8 8 44 APPENDIX 4. OF TERMS) STUDIES IN AVIAN BIOLOGY NO. 22 INTRODUCED GAME BIRDS ON THE SIX MAIN HAWAIIAN ISLANDS (SEE APPENDIX 2 FOR EXPLAINATION Species O'ahu Kaua'i Maui Hawai'i Moloka'i Lana'i Crax rubra 1928 F 2 1 Penelope purpurascens 1928 F 2 1 Ortalis cinereiceps 1928 F 2 1 Numida meleagris 1928 F 1874 F 1928 F 1928 F 1908 F 1914 F 1 1 1 1 I 1 1,10 1,10 1,10 1,10 1,10 1,10 Colinus virginianus 1906 F 1906 F 1906 F 1906 F 1906 F 1906 F 2 2 2 I 2 2 1,4 1,4 1,4 4 1,4 1,4 Oreortyx pictus 1929 F 1929 F 2 2 1 1 Callipepla squamata 1961 F 2 6 Callipepla californica 1855 F 1855 S 1855 S 1855 S 1855 S 1855 S 3 3 3 3 3 3 1,10 1,10 1,10 1,10 1,10 1,10 Callipepla gambelii 1958 F 1958 F 1958 F 1958 S 1958 S 2 2 2 1,2 2 4,10 4,10 4,10 6,10 4,10 Callipepla douglasii 1959 F 1 6 Tympanuchus cupido 1895 a F 1933 a,b F 1 1 1 1 Tympanuchus phasianellus 1932 F 2 1 Cyrtonyx montezumae 1961 F l 6 Ammoperdix griseogularis 1959 F 1 6 Alectoris chukar 1923 F 1957 S 1957 S 1949 S 1923 S 1923 S 2 2 2 2 3 3 1,10 4,10 4,10 5,10 4,10 4,10 Alectoris barbara 1961 F 1959 F 1961 F 1959 F 2 1,2 2 2 4,10 4,6 4,10 4,10 Francolinus francolinus 1959 S 1959 S 1959 S 1959 S 2 2 1,2 2 9 9 6 9 Francolinus pintadeanus 1962 F 1 6 Francolinus pondicerianus 1958 S 1958 S 1958 S 1959 S 1958 S 1958 S 2 2 2 I 2 2 4,5,10 4,5,10 4,5,10 6 4,5,10 5 Francolinus adsperus 1965 F 1 6 Francolinus icterorhyn- 1961 F chus 1 6 INTRODUCED BIRDS--Moulton, Miller, and Tillman APPENDIX 4. CONTINUED. 45 Species O'ahu Kaua'i Maui Hawai'i Moloka'i Lfina'i Francolinus clappertoni 1961 F 1 6 Francolinus erckelii 1957 S 1957 S 1957 S 1958 S 1957 S 1957 S 2 2 2 1,2 2 2 5,10 5,10 5,10 6 5,10 5,10 Francolinus leucosepus 1959 F 2 6 Perdix perdix 1910 F 1926 F 1929 F 1 1 2 1 1 1 Coturnix chinensis 1922 F 1910 F 1922 F 1922 F 1922 F 2 1 2 2 2 1 1 1 1 1 Coturnix pectoralis 1922 F 1922 F 2 2 3 3 Coturnixjaponica 1921 F 1921 S 1921 S 1921 S 1921 S 1921 S 3 3 2 3 3 2 2,10 2,10 1,10 2,10 2,10 1,10 Rollulus rouloul 1924 F 2 1 Bambusicola thoracica 1959 F 1961 F 2 1 4,5,10 6 Lophura leucomelanos 1962 S 1 6,10 Lophura nycthemera 1932 F 1870 F 2 1 1 1 Gallus gallus PH c S PH S PH F PH F PH F PH F 4 4 4 4 4 4 2,10 2,10 2,10 2,10 2,10 2,10 Gallus sonnerati 1962 F 1 6 Phasianus colchicus 1865 S 1865 S 1865 S 1865 S 1865 S 1865 S 1 1 1 1 1 1 1,10 1,10 1,10 1,10 1,10 1,10 Syrmaticus reevesii 1960 F 1960 F 1960 F 1959 F 1960 F 1960 F 2 2 2 1 2 2 4,10 4,10 4,10 6 4,10 4,10 Syrmaticus soemmerringii 1907 F 1907 F 1907 F 2 2 2 1 I 1 Chrysolophus pictus 1932 F 1870 F 2 1 1 1 Chrysolophus amherstiae 1932 F 2 1 Pavo cristatus 1860 S 1860 F 1860 S 1928 S 1860 F 1860 F 1 1 1 1 1 1 1,10 1,10 1,10 1,10 1,10 1,10 Meleagris gallopavo 1815 F 1815 F 1815 S 1815 S 1815 S 1815 S 1 1 1 I I 1 1,10 1,10 1,10 1,10 1,14 1,10 May have been Tympanuchus phasianellus (Caum 1933). Based on "indefinite reports" (Caum 1933). Prehistoric introduction. 46 APPENDIX 5. TERMS ) STUDIES IN AVIAN BIOLOGY NO. 22 INTRODUCED COLUMBIDS ON SIX MAIN HAWAIIAN ISLANDS (SEE APPENDIX 2 FOR EXPLANATION OF Species O'ahu Kaua'i Moloka'i Hawai 'i Moloka'i LSna'i Caloenas nicobarica 1928 F 1922 F 2 2 1 1 Chalcophaps indica 1924 F 2 1 Columba livia 1796 S 1796 S 1796 S 1796 S 1796 S 1796 S 3 3 3 3 3 3 1 1 1 1 1 1 Gallicolumba luzonica 1929 F 1 1 Geopelia cuneata 1928 F 1929 F 2 2 1 1 Geopelia humeralis 1992 F 1922 F 1928 F 2 1 2 1 1 1 Geopelia striata 1922 S 1922 S 1922 S 1922 S 1922 S 1922 S 2 2 2 2 2 2 1 1 1 1 1 1 Geophaps lophotes 1922 F 1922 F 1922 F 1922 F 2 2 2 2 1 1 1 1 Geophaps plumifera 1922 F 1922 F 2 2 1 1 Geophaps smithii 1992 F 1922 F 2 2 1 1 Geotrygon montana 1933 F 3 3 Leptotila verreauxi 1933 F 3 3 Leucosarcia melanoleuca 1922 F 1922 F 2 x, 2 1 x 1 Phaps chalcoptera 1922 F 2 1 Streptopelia chinensis 1879 S 1890 S 1890 S 1890 S 1890 S 1890 S 3 3 3 3 3 3 1 8 8 8 8 8 Streptopelia decaocto 1928 F 1920 F 1928 F 1 1 2 1 1 1 Zenaida asiatica 1961 F 2 6 Zenaida macroura 1962 S 1 9 Systematics Studies in Avian Biology No. 22:48-50, 2001. SYSTEMATICS--INTRODUCTION HELEN E JAMES Unreachable to amphibians, reptiles, and most land mammals, the Hawaiian Archipelago has been colonized naturally only by the most vagile of vertebrates. The native terrestrial vertebrates of the islands consist entirely of birds and a cou- ple of species of bats. Indeed, the islands are so remote from other complex terrestrial ecosys- tems that even birds have difficulty establishing themselves. The native birds that dwell as year- round residents in Hawai'i's terrestrial and wet- land habitats can be traced to as few as 20 col- onizing species (James 1991). These successful colonists speciated and evolved in the islands to give rise to an avifauna with over 100 resident species. Sadly, many ex- traordinary species are extinct and known only through fossil remains. The fossil species in- clude large flightless waterfowl, flightless wood- land ibises, many flightless rails, a variety of raptors, three or four large crows, and diverse species of Hawaiian honeycreepers or drepani- dines (Olson and James 1991, James and Olson 1991). Despite these losses, a host of remarkable endemic species survived in the islands long enough to be studied and appreciated by orni- thologists. Most of the survivors are passerinc forest birds, including many species in the adap- tive radiation of drepanidines. Besides passer- ines, the only birds that escaped early extinction are a hawk, an owl that is probably a recent colonist, the Hawaiian Goose (Branta sandvi- censis), and a variety of smaller waterbirds (in- cluding some that had moved into terrestrial habitats). The Hawaiian Islands are one of the world's hottest of hot spots for the extinction of birds. Twenty-four endemic species of birds have be- come extinct there since 1778, and another eight are either recently extinct or imminently threat- ened (these figures vary slightly according to the taxonomy followed; Pratt 1994). In addition, the thirty-five fossil species that have been de- scribed and approximately twenty that are cur- rently waiting to be described are thought to have disappeared mainly in the prehistoric pe- riod of human settlement (Olson and James 1991, James and Olson 1991). The causes of the decline and extinction of so many birds include habitat degradation and loss, introduced patho- gens such as avian malaria and poxvirus, and introduced predators such as the small Indian mongoose (Herpesres auropunctatus). Hawai'i's avifauna has garnered considerable 48 attention from ecologists and evolutionary biol- ogists. The extreme geographic isolation of the resident birds, the clear-cut barriers to dispersal within the archipelago (water gaps between is- lands), and the roughly linear progression of is- land ages (the islands to the northwest being old- er than those to the southeast), provide a rela- tively simple setting where the processes that underlie modem biogeographic patterns may be relatively accessible to inference. Classic papers on Hawaiian birds have addressed such topics as the allopatric model of speciation (Amadon 1950), character displacement (Hock 1970), dy- namic equilibrium theory in island biogeography (Juvik and Austring 1979), and the processes un- derlying macroevolutionary change (Amadon 1950, Hock 1970, 1979). The basic information relied upon in these studies is the systematics and distribution of Hawai'i's endemic birds. Formal study of the systematics and distri- bution of Hawai'i's birds began in the late eigh- teenth century, when the specimens collected on Captain James Cook's third voyage (in 1778- 1779) reached England. The century that fol- lowed saw the steady addition of new species from Hawai'i, as subsequent voyages returned to western ports with specimens, and later, var- ious foreigners took up residence in the islands and made their own collections (Olson and James 1991, 1994a). The lure of discovery fi- nally inspired a period of intense exploration of the islands aimed specifically at collecting and describing the native birds and other endemic organisms. Between 1887 and 1902, the islands' birds were thoroughly sampled by Scott Wilson, R. C. L. Perkins, and especially by Lord Walter Rothschild's collectors Henry Palmer, G. C. Munro, and E. Wolstenholme, followed shortly by H. W. Henshaw (Olson and James 1994a). These efforts lead to three comprehensive pub- lications (Rothschild 1893-1900, Wilson and Evans 1890-1899, Henshaw 1902a, and Perkins 1903). Decades after this age of exploration and dis- covery, papers on the systematics and evolution of Hawaiian birds began to appear with regular- ity again. Miller (1937) studied anatomical ad- aptations for terrestriality in the Hawaiian Goose, while most other authors focused on the adaptive radiation of drepanidines (e.g., Amadon (1950) on eclectic systematics and speciation, Richards and Hock (1973) on functional anato- my, Raikow (1977) on myology, Sibley and SYSTEMATICS Helen F. James 49 Ahlquist (1982) on DNA-DNA hybridization, Johnson et al. (1989) on protein electrophoresis, Tarr and Fleischer (1993, 1995) on mitochon- drial DNA). Also, beginning in the 1970s, fossil birds were being found in Hawai'i with surpris- ing frequency (e.g., Olson and Wetmore 1976, Olson and James 1982b, 1984, James et al. 1987, Olson and James 1991, James and Olson 1991). As the papers in this volume attest, more ef- fort is now focused on the systematics of Ha- waiian birds than at any time since the 1890s. This coincides with a renaissance in phyloge- netic research, spurred by advances in methods of analysis and by the technological revolution in molecular genetics. Hawaiian birds attract ex- tra attention because of the urgency of studying species threatened with extinction, and the need to place the new fossil species in an evolution- ary context. The most active program in molecular genet- ics of Hawaiian birds is that of Robert C. Fleischer and his collaborators. A long-term goal of this program is to study the evolutionary genetics of each endemic lineage of Hawaiian birds. Fortunately, even the extinct fossil lin- eages can be studied, through amplification and sequencing of DNA fragments f¾om fossil bones (Cooper et al. 1996, Paxinos 1998, Sorenson et al. 1999). By including appropriate outgroups and assuming a molecular clock based in part on earlier Hawaiian drepanidine research (Fleischer et al. 1998), Fleischer and Mcintosh (this vol- ume) are able to estimate the length of time that each lineage has been present in the islands. Their paper offers a glimpse of the types of questions we can answer with molecular genet- ics that we could only speculate about before, and also hints at the large number of molecular genetic studies of Hawaiian birds that are cur- rently in progress. The value of genetics and systematics to con- servation of endangered species is exemplified by Judith Rhymer's contribution on the endan- gered Hawaiian Duck (Anas wyvilliana) and Laysan Duck (Anas laysanesis). Using a battery of molecular genetic techniques, Rhymer ad- dresses several pressing questions that will af- fect the management plans for these two species. First, she shows that the Hawaiian and Laysan Ducks have separate evolutionary histories and certainly merit species rather than subspecies status. She also cites anecdotal evidence that Laysan Ducks rarely hybridize in captivity. Combined with her previous collaborative re- search showing that the former range of the Lay- san Duck included the main Hawaiian islands (Cooper et al. 1996), this lays the groundwork for possible reintroduction of the Laysan Duck in the main islands. Rhymer also develops mo- lecular markers that can be used to monitor the extent of hybridization between Hawaiian Ducks and introduced Mallards. Such hybridization threatens the survival of Hawaiian Ducks on O'ahu but, so far, not on Kaua'i. The informa- tion and genetic tools provided by Rhymer will be indispensable in formulating management plans for these rare species. Phylogenetic analysis can also contribute to conservation planning by providing a way to as- sess the phylogenetic "distinctiveness" of threatened species. The number of threatened species is disproportionate to the funding that is available to help them, forcing managers to make hard decisions about which species to fo- cus upon. One objective of such decisions is to preserve evolutionary diversity. It is consequent- ly useful to know to what degree a particular threatened species differs from its surviving rel- atives. The study by Fleischer et al. in this vol- ume assesses the evolutionary relationships and phylogenetic distinctiveness of an endangered drepanidine, the Po'ouli (Melamprosops phaeo- soma), using genetic and osteological data. Both datasets place the Po'ouli within the clade of drepanidines. However, an index of distinctive- ness applied to both datasets also indicates that the Po'ouli is very different from other living drepanidines, both genetically and morphologi- cally. Fleischer et al. conclude that saving the Po'ouli from imminent extinction would be well worth the effort from this perspective. Douglas Pratt, who contributes a cladistic analysis of the drepanidine radiation based on eclectic phenotypic characters, recommends that no changes be made to his taxonomy in the light of molecular genetic data, which he regards as preliminary, inconsistent, and in the case of mi- tochondrial DNA sequences, perhaps giving a false signal due to hybridization (although there are no confirmed hybrids among the drepanidi- nes). Where his results conflict with my disser- tation research on drepanidine osteology (James 1998), he describes my work as perhaps based on superficial resemblances and illustrative of the weaknesses of "single character or single- complex analyses." My results are remarkably congruent with Raikow's (1977) early cladistic analysis of myology and external anatomy, but Pratt also considers Raikow's character analysis to be vague where it conflicts with his own re- sults. I can only urge readers to consult the orig- inal sources and form their own opinions. Two corrections should be made here, how- ever. Pratt (p. 88, this volume) implies that my tree topologies bring together unrelated species with similar bill shapes in the red-and-black plumaged group and the green plumaged group. 50 STUDIES IN AVIAN BIOLOGY NO. 22 Actually, my analysis (James 1998) recognized the red-and-black birds as a clade, including the full range of bill morphologies from "finch- like" to long and sickled. None of the green birds with parallel bill morphologies joined this clade. Also, whereas Pratt states that James and Olson (1991) previously suggested lumping Loxioides and Chloridops, we actually wrote that future research may justify merging Loxioi- des with Telespiaa. The contribution on species concepts by Pratt and Pratt is very much in the tradition of Pratt's dissertation (Pratt 1979), an eclectic assessment of alpha taxonomy with emphasis on vocaliza- tions, plumages, and behavior as potential iso- lating mechanisms. Many allopatric populations of island birds were long ago demoted to sub- species by Ernst Mayr and others who embraced his biological species concept. For example, in his dissertation, which was supervised by Mayr, Amadon (1950) applied the biological species concept to the drepanidines and came up with many fewer species than were recognized by the late 19 th century authorities (see Pratt 1979). However, Pratt and Pratt argue that Amadon and Mayr often erred in applying their own species concept, or simply lacked information that would have kept them from lureping. Properly applied, they feel that the biological species con- cept would elevate most of Amadon's allopatric subspecies to full species status. Although they stress potential isolating mechanisms in their evaluations, their way of applying the biological and phylogenetic species concepts result in very similar taxonomic lists. While the debate over species concepts continues, non-taxonomists can take comfort in knowing that, with the growth of knowledge about Hawaiian birds, the choice of species concept now appears to have little ef- fect on the species-level taxa that are recog- nized. This is an exciting time for evolutionary and biogeographic studies of HawaiTs avifauna. The abundance of fossils enables us to study morphological change through time, calculate rates of species turnover and extinction using data with real time depth, and gain insight into the former ranges and habitat preferences of en- dangered species. With ancient DNA we can identify fossil species, place them on phyloge- netic trees, and even study their population ge- netics over long stretches of time. Because the genetic divergences between isolated island pop- ulations cannot be older than the islands them- selves, multiple local calibrations of the mini- mum rates of DNA sequence change are possi- ble in Hawai'i. Putting aside differences of opin- ion on whether genetic or phenotypic data are best for phylogenetic analysis (see Pratt, this volume, and Fleischer and Mcintosh, this vol- ume), phylogenetic hypotheses can be strength- ened and insights into character evolution can be gained through comparison of data and re- sults from these two types of studies. The con- fluence of knowledge from these various sources is leading to a much improved picture of change in Hawai'i's avifauna through time. The growth of information from genetics, phylogenetics, and paleontology is contributing not only to basic knowledge, but in important ways to conserva- tion management as well. Studies in Avian Biology No. 22:51-60, 2001. MOLECULAR SYSTEMATICS AND BIOGEOGRAPHY OF THE HAWAIIAN AVIFAUNA ROBERT C. FLEISCHER AND CARL E. MCINTOSH Abstract. The Hawaiian avifauna is exceptional for its high proportion of endemic taxa, its spectac- ular adaptive radiations, and its level of human induced extinction. Little has been known about the phylogenetic relationships, geographical origins, and timing of colonization of individual avian lin- eages until recently. Here we review the results of molecular studies that address these topics. Mo- lecular data (mostly mitochondrial DNA sequences) are available for 14 of the 21 or more lineages of Hawaiian birds. We briefly review results of phylogenetic analyses of these data for lineages that have experienced major and minor radiations, and for single differentiated species and probable recent colonists. When possible, we determine the mainland species that are genetically most closely related. We find evidence that roughly half of the >21 lineages colonized from North America; not even a quarter appear to have come from South Pacific Islands. Our data also provide little evidence that Hawaiian bird lineages predate the formation of the current set of main islands (i.e., >5 Ma), as has been found for Hawaiian Drosophila and lobeliads. Key Words: adaptive radiation; biogeography; Hawaiian avifauna; mitochondrial DNA; molecular systematics. In 1943 Ernst Mayr published a short paper in The Condor summarizing his hypotheses about the geographic origins and closest living rela- tives of each known lineage in the Hawaiian avi- fauna. Mayr (1943) concluded that half of 14 hypothesized colonizations were of American origin and only two lineages arose from Poly- nesia. Therefore, although Hawai'i is considered part of the "Polynesian Region" because most of its biota and its human inhabitants had Pol- ynesian ancestors, in terms of its birds Hawai'i is in the Nearctic Region. Since Mayr's paper, other authors have posited similar systematic hy- potheses and biogeographic scenarios based on morphological, ecological, and distributional data (e.g., Amadon 1950, Pratt 1979, Berger 1981). Paleontology has offered only minor res- olution of the relationships of ancestral lineages or the timing of speciation events; although there is an excellent Holocene fossil record in Hawai'i (Olson and James 1982a, 1991; James and Olson 1991), the pre-Holocene record is ex- tremely limited (though one excellent fauna dates to >0.12 Ma ago; James 1987). In recent years, molecular methods have prov- en extremely useful for inferring evolutionary relationships among taxa and the relative time frames during which taxa evolved (Avise 1994, Hillis et al. 1996). Inference from molecular data may be the best available way to reconstruct phylogenetic relationships and determine geo- graphical origins and evolutionary time frames for Hawaiian taxa. In part this is because mor- phological or behavioral changes are often adap- tive responses subject to natural or sexual selec- tion (i.e., as part of the process of adaptive ra- diation), and they do not usually show constancy in their rates of change. Thus they can poten- tially mislead on issues of common ancestry via homoplasy. DNA sequences, on the other hand, while obviously not evolving in a perfect clock- like fashion (see below), do change over time, and evolve more continuously than morphology. Also, with the exception of a relatively few non- synonymous changes within protein sequences, they generally evolve via mutation and drift (Nei 1987, Avise 1994), and are not as subject to ho- moplasy via convergence or stasis as are mor- phological or other characters. Thus major adap- tive shifts in, for example, the bills of Hawaiian honeycreepers, may occur within some lineages (e.g., to thin and decurved in the nectarivorous 'I'iwi, Vestiaria coccinea), while not in others (e.g., conical and finchlike in the Laysan Finch, Telespiza cantans), in spite of an identical amount of time since evolving from their puta- tively "finch-billed" common ancestor. There are methods for detecting symplesiomorphic versus synapomorphic characters in phylogenet- ic analysis, but the higher variance in rates of change of morphological characters remains a problem for phylogenetic reconstruction (Hillis et al. 1996). While there have been significant molecular investigations of particular Hawaiian plant and invertebrate taxa (especially Drosophila; e.g., Hunt and Carson 1983, DeSalle and Hunt 1987, DeSalle 1992), few molecular studies detailing evolutionary histories of the Hawaiian avifauna have been made until recently (e.g., Tarr and Fleischer 1993, 1995; Feldman 1994, Cooper et al. 1996; Fleischer et al. 1998, 2000, this vol- ume; Paxinos 1998, Sorenson et al. 1999, Fleischer et al. in press, Rhymer this volume; C. Tarr, E. Paxinos, B. Slikas, H. James, S. Olson, A. Cooper, and R. Fleischer, unpubl. data). 51 52 STUDIES IN AVIAN BIOLOGY TABLE 1. THE ELEMENTS OF THE HAWAIIAN AVIFAUNA NO. 22 Taxon Family No. of species a Geographic origin b Comments c Non-passeriformes: Ibises Plataleidae -->2 N.A. minor radiation, flight- less, Apteribis? Night Heron Ardeidae 1 N.A. recent colonist, Nyctico- rax nycticorax Moa-nalos Anatidae >4 W. Hemisphere minor radiation?, 3 flightless duck gen- era? True Geese Anatidae ->3 N.A. minor radiation, Bran- ta, ?e Moderu Ducks Anatidae 2 N.A. and Asia I _+ diftrentiated, 1 re- cent colonist, Arias e Porzana Rails Rallidae ->12 Pacific/unknown major radiation?, ->2 colonizations? Large rallids Rallidae 2 N.A.? recent colonists?, coot and moorhen Black-necked Stilt Recurvirostridae 1 N.A. recent colonist, Himan- topus knudseni e Eagle Acciptridae I Asia recent colonist, Haliaee- ms leucophrys? Buteo Acciptridae 1 N.A. differentiated, Buteo so- litarius e Harder Acciptridae 1 Unknown differentiated, Circus dossenus? Long-legged Owls Strigidae 4 Unknown minor radiation, Grailis- trix spp. 47 Short-eared Owl Strigidae l Unknown recent colonist, Asio fiammeus sandwichen- sis Passeriformes: Crows Corvidae ->4 Unknown minor radiation?, Corvus spp., 3?, I e Millerbird Sylviidae I South Pacific differentiated, Acroce- phalus familiaris e 'Elepaio Myiagridae -> 1 Australasia differentiated, Chasiem- pis sandwichensis Thrushes Muscicapidae 5 W. Hemisphere minor radiation, Myad- estes spp., 37, 1 e Honeyeaters Meliphagidae ->6 South Pacific minor radiation, Moho spp., Chaetoptila, all? Honeycreepers Fringillidae >50 Asia or N.A.? major radiation, drepani- dines, most? or e >21 lineages 13 families ->102 species a Number of species within each lineage/family, based on James and Olson (1991), Olson and James (1991), and H. James (pers. comm.). b N.A. - North America; W - West. c ? denotes at least SOllle Illembers extinct; e denotes at least some illelllbers endangered. Components of the Hawaiian avifauna vary greatly in the degrees to which they have spe- ciated and become modified morphologically and ecologically (Table 1). For example, the Ha- waiian drepanidines (Hawaiian finches or hon- eycreepers) have evolved incredible morpholog- ical, ecological, and behavioral diversity across more than 50 species and are one of the most often cited cases of adaptive radiation (Roths- child 1893-1900, Perkins 1903, Amadon 1950, Raikow 1977, Freed et al. 1987a, James and Ol- son 1991, Tarr and Fleischer 1995, Fleischer et al. 1998). Several species of extinct, large, flightless waterfowl (moa-nalos) show extreme morphological modification in their apparent shift into a ratite/grazing mammal/tortoise niche (Olson and James 1991; Sorenson et al. 1999). Other avian lineages have not speciated and have changed morphologically little or not at all from putative mainland relatives (e.g., Black- crowned Night Heron, Nycticorax nycticorax hoactli; Short-eared Owl or Pueo, Asiofiammeus sandwichensis). Is this variance in levels of spe- ciation and phenotypic differentiation related merely to the lengths of time that lineages have been evolving in the islands (Simon 1987, Car- HAWAIIAN BIRD MOLECULAR SYSTEMATICS--FIeischer and Mclntosh 53 I I I I  : Main Kaual I I .......... .i.11.i.i.i'11..i!...,.i._....o...11.b:i.i..}...: .... . (SAmy) Oa I Ts,w: ,65,,w. (3.7 my) ¾"  - [ e Maui-Nui w -nk (1.9 my) :: VaCL? e ......... Hawaii  ................................... :  .43 my) Main Hawaiian Islands (K-At ages) FIGURE 1. Map of the main Hawaiian Islands (plus inset map of main and leeward Hawaiian Islands). Ages of the oldest rocks from the main islands based on K-Ar dating are noted. Maui-Nui is composed of the islands of Maul, Lana'i, Kaho'olawe, and Moloka'i, all of which were connected until about 0.34).4 Ma ago and again during more recent periods of low sea level. son and Clague 1995)? Or are there other factors that have promoted stasis in some lineages and change in others, regardless of length of time in the islands? As noted above, the fossil record provides little resolution of this question. Thus, estimates of the age of separation from ancestors outside of the Hawaiian Archipelago, or the age of a radiation within the islands, can only be inferred from molecular data. The Hawaiian Islands and its avifauna are ex- tremely isolated from continental and other Pa- cific island avifaunas. This is likely the primary reason for the relatively low number of indepen- dent taxonomic avian lineages that occur in the islands (Mayr 1943, Pratt 1979). While the total number of such lineages has been increased (and continues to increase) from recent fossil findings (Olson and James 1982a, 1991; James and Olson 1991), the islands still appear to have far fewer independent avian lineages than one might ex- pect for a tropical archipelago of this size and topographic diversity, and there may be addi- tional factors involved that limit the primary di- versity of the avifauna. Here we summarize molecular and other data relevant to systematics and biogeography of the Hawaiian aviafauna. We first provide a brief overview of the geological history of the Ha- waiian Archipelago and its utility for calibrating rates of molecular evolution (Tarr and Fleischer 1993, Fleischer et al. 1998). We then consider the origins and phylogenetic histories of each lineage within the avifauna, addressing exten- sive and minor radiations, well-differentiated single species, and undifferentiated (and likely recent) colonists. We also apply a molecular clock approach to obtain rough estimates of the maximum period of time that a lineage could have existed in the Hawaiian Islands. GEOLOGICAL HISTORY AND THE CALIBRATION OF MOLECULAR EVOLUTIONARY RATES The Hawaiian Islands have an unusual geo- logical history (Clague and Dalrymple 1987, Walker 1990, Carson and Clague 1995; Fig. 1). They form as the Pacific Plate drifts northwest over a "hot spot" where magma extrudes from the earth's mantle through the crust to build huge shield volcanos (often to >4 km above sea level). The extreme weight of a new island, combined with the cooling of the crust as it moves away from the hot spot, causes a rela- tively rapid subsidence in island elevation and area. Subsidence continues slowly beyond this point, as does erosion, and islands shrink to be- come small coral and sand atolls and ultimately undersea mounts (Fig. 1). The Hawaiian Islands are ordered by age in a linear pattern, with the oldest main island in the 54 STUDIES IN AVIAN BIOLOGY NO. 22 northwest (Kaua'i at 5.1 Ma) and the youngest in the southeast (Hawai'i at 0.43 Ma; Fig. 1). This volcanic conveyor belt provides an excep- tional system for evolutionary studies, as it sets up a temporal framework that can be used to estimate the timing of evolutionary events and rates of evolution. The age of an island is the maximum age for a population inhabiting the is- land. These ages can be used to calibrate rates of molecular change if phylogenies reveal that the pattern of cladogenesis parallels the timing of island formation, and if populations colonize near to the time of island emergence (Bishop and Hunt 1988, Tarr and Fleischer 1993, Givnish et al. 1995, Fleischer et al. 1998). We used this rationale to calibrate part of the mitochondrial cytochrome b (cyt b) gene in Ha- waiian drepanidines (Fleischer et al. 1998). The overall rate of cyt b divergence, corrected for minor saturation, transition bias, rate variation among sites, and potential lineage sorting is 1.6% sequence divergence/Ma. This value is similar to a rate we estimated for overall restric- tion site divergence in mitochondrial DNA (mtDNA) in drepanidines (--2%/Ma; Tarr and Fleischer 1993). Note that rates calibrated using this approach are based on a time period of di- vergence up to only about 4 Ma. Recently, Moore et al. (in press) showed through simula- tion modeling that cyt b sequence divergence is accurate as a predictor of time of divergence only to about 5 Ma (i.e., about 10% overall se- quence divergence). Predictions of dates older than 5 Ma are generally underestimated. Nonlin- earity of sequence divergence due to saturation and rate variation among sites appears to be- come problematic above about 10% overall se- quence divergence for birds (Krajewski and King 1996, Randi 1996, Moore and DeFilippis 1997). Thus the drepanidine or other cyt b rates are not likely to be applicable to events that hap- pened appreciably earlier than 5 Ma, and caution must be exercised when making predictions or calibrations from cyt b sequence divergences over 10%. Our drepanidine rates (Trot and Fleischer 1993, Fleischer et al. 1998) are within the range of estimates for avian and mammalian taxa based on calibrations derived from relatively re- cent fossil evidence of cladogenesis. This is true for both restfiction fragment length polymor- phisms (RFLPs) in total mtDNA and sequence divergence in the cyt b gene. Examples of avian rates include RFLP variation in geese at --2%/ Ma (Shields and Wilson 1987); cyt b sequences in partridges versus Gallus at 2.0%/Ma (Randi 1996; however, Arbogast and Slowinski [1998], corrected the divergences using an HKY [Has- egawa et al. 1985] model with a F-correction to obtain a rate of about 5.0%/Ma); RFLP variation in New World quail at 2.0%/Ma (reported in Klicka and Zink 1997); woodpecker cyt b at 2.0%/Ma (Moore et al. in press); cyt b in cranes at 0.7%/Ma for Balearicines versus Gruines (old split) and up to 1.7%/Ma for comparisons within the Gruines (Krajewski and King 1996); and cyt b in albatross at 0.65%/Ma (Nunn et al. 1996, recalculated for total sequence change in Klicka and Zink 1997). In the crane and albatross stud- ies the slower rates could be caused by the lon- ger generation times in these species, or perhaps by reduced metabolic rates in these larger-bod- ied taxa (Martin and Palumbi 1993, Rand 1994, Bromham et al. 1996, Nunn and Stanley 1998). Alternatively, the difference may relate to the fossil dates used for calibration: for both studies these dates are older than 10 Ma, whereas for all but the partridge/Gallus comparison (Randi 1996) the dates are before 5 Ma. Both studies attempt to correct for saturation (Krajewski and King 1996, Nunn et al. 1996), but may severely underestimate divergence (Arbogast and Slow- inski 1998). This could be considered an inverse prediction of the findings of Moore et al. (in press): using dates older than 5 Ma to calibrate may result in an underestimate of the rate. Sup- porting this is a negative correlation between di- vergence times and divergence rates (Spearman rho - -0.51, P - 0.042) from Table 2 of Martin and Palumbi (1993). Avian rates are similar to most mtDNA/cyt b rates calculated for mammal taxa (e.g., --2%/Ma; Brown et al. 1979, Irwin et al. 1991, Stanley et al. 1994, Janacek et al. 1996). In general, then, calibrated rates of mtDNA protein coding sequence divergence in birds and mammals do not appear to vary greatly from about 2%/Ma. Most rate variation appears to be correlated with variation in body size and its correlates (i.e., metabolic rate, generation time; Martin and Palumbi 1993, Rand 1994), although some of the variation may be due to differing selective constraints on proteins in different lin- eages or to fluctuations in population size (Ohta 1976). In summary, with the exception of the very rapidly evolving control region (which in some sections may be evolving an order of mag- nitude faster than the average for mtDNA; e.g., Quinn 1992), most avian and mammalian rate calibrations based on corrected mtDNA diver- gence and dates before 5 Ma ago reveal rates at about, or above, 2% divergence/Ma. Based on the rather detailed rationale described above we feel that mtDNA (RFLP or cyt b) sequence di- vergence between a Hawaiian taxon and its clos- est non-Hawaiian relatives that is below about 10% would indicate an origin near the time of or after the formation of the island of Kaua'i. HAWAIIAN BIRD MOLECULAR SYSTEMATICS--Fleischer and Mclntosh 55 ORIGINS AND EVOLUTION OF THE HAWAIIAN AVIFAUNA There were more than 102 species of native breeding land- or waterbirds (i.e., non-seabirds) in the Hawaiian Islands (Table 1; constructed from James and Olson 1991, Olson and James 1991; and H. James, pers. comm.). These 102 species sort into six songbird families (Passeri- formes) and seven non-songbird families (Table 1). Some families have a relatively large number of species (i.e., >4) and, in some cases, it is fairly clear that each group of species in a family represents an in situ radiation from a single col- onization (e.g., drepanidines, thrushes). it is clear that in some families (e.g., anatids, rallids) there has been more than a single colonization event, while for others (e.g., corvids, meliphag- ids) it is difficult to determine how many inde- pendent colonization events have occurred. Avian biologists working in the islands have been fortunate to have an excellent Holocene fossil record (Olson and James 1982a, 1991; James and Olson 1991). Without this record, we would be missing a tremendous amount of in- formation about distributions, phylogeny, bio- geography, and ecology of these birds. Even so, additional fossil taxa continue to be discovered and, thus, our knowledge remains incomplete. The advent of genetic studies employing the polymerase chain reaction (PCR) has opened a new and exciting avenue for study of these fos- sils. Our laboratory has had considerable success amplifying mtDNA sequences from these sub- fossil remains. Here we summarize what has been learned about the evolution of Hawaiian birds from phylogenetic analyses of mtDNA se- quences from a number of extinct and extant taxa. EXTENSIVE RADIATIONS The drepanidines (Hawaiian finches or hon- eycreepers) are by far the most speciose group in Hawai'i, with 33 species known from histor- ical collections and more than 17 known from subfossil remains (totaling over 50 species; James and Olson 1991; H. James, pers. comm.). The drepanidine radiation is remarkable for its extreme morphological, ecological, and behav- ioral diversity (Rothschild 1893-1900, Perkins 1903, Amadon 1950, Baldwin 1953, Raikow 1977, Pratt 1979, Freed et al. 1987a, James and Olson 1991). However, major adaptive shifts ap- pear to have modified many characters tradition- ally used for phylogenetic reconstruction, while others less subject to selection have been con- served and provide little or no phylogenetic in- formation. The somewhat chimeric associations of morphological traits in the group have even led to the suggestion that the drepanidines are not monophyletic (Pratt 1992a,b). Molecular data may prove especially useful for assessing evolutionary relationships in this group, and they do support a cardueline ancestry and, thus far, monophyly of the drepanidines (Fleischer et al. 1998; Fig. 2c). Molecular data may also be effective in esti- mating a time frame for the drepanidine radia- tion. The radiation of the drepanidines would seem quite deep based on their relative degree of phenotypic diversity. Molecular evolutionary rate estimates based on DNA-DNA hybridiza- tion data (Sibley and Ahlquist 1982) are in sup- port of this prediction with an estimated split of drepanidines from a cardueline outgroup of about 15-20 Ma. Molecular rate estimates from both allozyme (Johnson et al. 1989, Fleischer et al. 1998) and mtDNA data (Tarr and Fleischer 1993, 1995; Fleischer et al. 1998), however, strongly contradict the results of Sibley and Ahlquist (1982) and suggest a basal split that began about 4 Ma ago and a separation from a mainland cardueline ancestor (not necessarily the closest outgroup; Fig. 2c) of <5-6 Ma ago. These mtDNA results are based on several in- ternal rate calibrations estimated as outlined above for cyt b. Sibley and Ahlquist's (1982) results may be biased by their use of continental biogeographic points in their calibration (Quinn et al. 1991) or by use of too distant outgroups for comparison. No other avian radiation in Hawai'i is so di- verse in morphology or number of lineages as the drepanidines. Extinct flightless rails, classi- fied as Porzana (Olson and James 1991), in- cluded perhaps more than 12 species, with as many as three species on each major island. Un- til recently it has not been clear whether these species comprise a single highly radiated clade, or represent a number of independent coloniza- tions from mainland or other Pacific island sources. Molecular phylogenetic analyses (B. Slikas, S. Olson, R. Fleischer, unpubl. data) in- dicate that each of the two historically collected Porzana species resulted from independent col- onizations. For Porzana palrneri the Kimura 2- parameter corrected distance (Kimura 1980; dis- tance and SE calculated in MEGA, Kumar et al. 1993) for 197 base pairs (bp) of ATPase8 was 2.1 ñ 1.1% distant from its closest non-Hawai- ian Poraana relative. For P. sandwichensis the ATPase8 Kimura 2-parameter corrected distance was 5.9 +- 1.8% to its closest non-Hawaiian Por- aana relative. Molecular analyses of Poraana taxa known only from subfossil remains are un- derway. 56 STUDIES IN AVIAN BIOLOGY NO. 22 Old World Ibises (9 spp.) Spoonball r-Glossy Ibis It -white-faced Ibis (Piegaols) I I  Scarlet b s ,, 1 i__Whitelbls (Eudoc. mus) L- -Lanai Flightless Ibis .-----New World Ibises (5 spp ) Outgroups b. I---Hawaiian Hawk, B. solitarius 71J i Galapagos Hawk, B. galapagoensis (Buteo) -Swainsoh'S Hawk, B. swainsoni H L--Short-tailed Hawk, B. brachyurus 9t -- Red-tailed Hawk, B. jamaicensis 94 L Old World Buteos (5 spp.)  I "Woodland" Buteo I - Other Buteo/Parabuteo L-- Buteogallus urlbltlnga ,-J69  Whlte-browed Rosefinch (Carpodacus thura) ,- [ -I'-'---'-.:._1 ............. House F,nch (C ........... ) Carduelini ' I I  Green Honeycreeper (Chlorophanes splza) '  --I  I Scarlet-rumped Tanager r Ll. prnthnmer Tanager (P'rangaru ra) (Ramphøcetusp .......... ) J , j Prothonotary Warbler (Prolonoh'r,a C trea) Emberizinae J , _ .b__. ____ Brown-headed Cowbird (Mololhrus ater) Buff-barred Warbler (Phylloscopus pulcher) : I Townsend's Sohtaire '-L Ruf.-throated Sohtaire .... [____"'1____ Omao Myadestes -b____- Swainso n's Thrush b Veery Catharus ............. d6y %'ru;R' '(''''i e ! Dabbltng Ducks ß  ............ _S_o_u ?_ A_ _m _e r c_a_n_ D_ u_ c_k_s ....... --- Thambetochen [ iL__ __'-___ __'TT2_ U _P?_i_o_c_h_e_n_ ...... _M_o_a  L Diving Ducks and relabves I I Shelducks & Shelgeese t "True" Geese & Swans New World Jays "OId World" Jays (+ Corvus rnonedula) Corvus corax - Raven - Eurasia Corvus corax - Raven - NA Corvus hawaiiensis- Alala Corvus corone- Hooded Crow orvus brachyrhynchos - Amer, Crow1 orvus brachyrhynchos - Amer. Crow2 us kubaryi- Mariana Crow ß Corvus tnstis - Grey Crow FIGURE 2. Abbreviated phylogenetic reconstructions for six Hawaiian taxa. a. Summarized maximum parsi- mony tree based on 407 nucleotide sites of 12s ribosomal RNA (A. Cooper, S. Olson, H. James, R. Fleischer, unpubl. data). b. Summarized parsimony phylogram based on preliminary analysis of over 1500 bp of mtDNA sequence (ATPase8, ND2, cyt b, and COI) in Buteo and related taxa (R. Fleischer, P. Cordero, C. Mcintosh, I. Jones, and A. Helbig, unpublished). c. Summary of relationships of outgroups and drepanidines based on par- simony analysis of 675 bp of cyt b sequence. d. Parsimony phylogram constructed from 700 bp of cyt b sequence from two Myadestes and three Catharus taxa with '0ma'o and Turdus outgroup. e. Parsimony tree of two moa- nalo genera and a wide sampling of other waterfowl taxa showing two moa-nalo genera to be sister taxa and related to dabbling ducks. Tree is summarized from Sorenson et al. (1999), and based on over 1200 bp of mtDNA sequence. f. Parsimony phylogram showing summary of jay relationships to Corvus and a sampling of Corvus taxa based on 1008 bp of cyt b. The 'Alal is most closely related to the Common Raven. MINOR RADIATIONS Seven other Hawaiian avian groups have un- dergone what appear to be minor radiations, each with fewer than six species (Table 1). These include thrushes (genus Myadestes), hon- eyeaters (genera Moho and Chaetoptila), a lin- eage of owls (genus Grallistrix), several crows (genus Corvus), flightless ibises (genus Apteri- bis), and two waterfowl (Anatidae) lineages: true geese (genus Branta) and the highly modi- fied dabbling duck relatives called "moa-nalos" (genera Chelychelynechen, Ptaiochen, and Thambetochen). The five species of thrushes were placed orig- inally in their own genus, Phaeornis, but were considered aligned with solitaires (Myadestes; Stejneger 1887, Amadon 1950), robins (Turdus) or nightingale-thrushes (Catharus; Ripley 1962). Most of the morphological and other evidence (e.g., Kepler and Kepler 1983) clearly favors placement of thrushes in Myadestes (Pratt 1982). We analyzed variation in about 700 bp of the cyt b gene of mtDNA (C. Mcintosh and R. Fleischer, unpubl. data), for the Hawai'i Thrush (or 'Oma'o, M. obscurus), three Catharus, two American Myadestes and a Turdus species, along with outgrou_p taxa. The resulting trees clearly place the 'Oma'o within the Myadestes clade, regardless of the tree building algorithm (i.e., maximum parsimony, Fig. 2d; maximum likelihood or minimum evolution). We could not resolve with certainty using this data set whether the '0ma'o is more closely related to M. geni- barbis, a Caribbean solitaire, or M. townsendi of western North America. The Kimura 2-parame- ter corrected distance between the '0ma'o and the solitaires is 6.7% for the 700 bp. HAWAIIAN BIRD MOLECULAR SYSTEMATICS--FIeischer and Mcintosh 57 The meliphagid genera Chaetoptila (Kioea; 2 spp.) and Moho (the '(7)'6s; 4 spp.) may repre- sent independent colonizations from south Pa- cific meliphagids (Perkins 1903), although Mayr (1943) considers both genera derived from a sin- gle colonist. One species of Moho occurs on each of Kaua'i, O'ahu, Maul Nui (Maui, Lana'i, Moloka'i, and Kaho'olawe), and Hawai'i, and this well-differentiated lineage (Pratt 1979) may provide an opportunity to estimate a rate cali- bration. The closest sister groups for the Ha- waiian meliphagids are unknown, with some au- thors suggesting Gymnomyza of Fiji and Samoa (e.g., Mayr 1943) and others favoring Foulehaio of Samoa or the New Zealand tui's (Prosthe- madera; e.g., Munro 1944, Pratt 1979). Molec- ular studies are underway to address the origin and monophyly of the Hawaiian forms and the possibility of a rate calibration from the four Moro species. A calibration could be used to estimate the date of separation from the most recent common ancestor. This date is important because we estimate from our drepanidine cali- brations that nectarivorous drepanidines evolved only 2-3 Ma ago, while Givnish et al. (1995) used a calibration of chloroplast DNA restriction fragment variation to estimate that bird-pollinat- ed flowering lobeliads (genus Cyanea) evolved 8-17 Ma ago. Thus it is highly unlikely that dre- panidines "coevolved" with these plants in the islands (as was suggested by Givnish et al. 1995). The meliphagids are the only other known native, obligate nectarivores in the is- lands and, if they are older, could be the coe- volved taxon. At least four crows (Corvus) occurred in the islands (James and Olson 1991; H. James, pets. comm.). Three of these are known only from subfossils; two of which have been described and the fourth is the highly endangered Hawai- ian Crow (Corvus hawaiiensis), hereafter re- ferred to as 'Alala. It is unclear at present wheth- er these represent a single colonization and sub- sequent radiation, or multiple colonizations by the same or different ancestral taxa (James and Olson 1991). Preliminary phylogenetic analyses of the 'AlMa and seven other Corvus taxa indi- cate that it is more closely related to the Com- mon Raven (Corvus corax) than to more typical crows, including two South Pacific island crows (R. Fleischer and C. Mcintosh, unpubl. data; Fig. 2f). The Kimura 2-parameter corrected se- quence divergence for 1,008 bp of cyt b between 'AlMa and North American Common Raven is about 8.4 + 1.0%. Subfossil bones and owl pellets are all that remain of four species of long-legged owls (Grallistrix) that apparently were morphologi- cally adapted to feeding on birds. While no DNA analyses have yet been made on this group, it appears likely that they represent the results of a single colonization and subsequent minor radiation. At least four lineages of waterfowl have col- onized the Hawaiian Islands. Of these, only two, the moa-nalos (Olson and James 1991, Sorenson et al. 1999) and the modern geese (Branta; Ol- son and James 1991, Paxinos 1998; E. Paxinos et al. unpubl. data), have speciated beyond a sin- gle endemic species. All of the moa-nalos evolved to very large size, flightlessness, and highly modified cranial morphology. They have become convergent in morphology to rafites in terms of postcranial morphology, and one spe- cies in particular has converged to tortoise-like cranial morphology. Like the moas of New Zea- land (Darwin 1859), the moa-nalos occupied a grazing mammal or tortoise niche (Olson and James 1991). One genus and species (Chelyche- lynechen quassus, the Turtlejawed Goose) is re- stricted to Kaua'i and one (Ptaiochen) to Maui, but Thambetochen is found on both Maui Nui and O'ahu, suggesting the genus may have orig- inated on O'ahu and later walked across the Pen- guin Bank land bridge (Fig. 1) to Moloka'i. No moa-nalo is known from the young island of Ha- wai'i (but see below). Olson and James (1991) suggested that the moa-nalos were related to either dabbling ducks or shelducks (tadornines) on the basis of skeletal characters, primarily the presence and shape of their syringeal bullae. Livezey (1996) tentatively concluded from a cladistic analysis of morphol- ogy that the moa-nalos were sister to a "true" geese and swan clade, and not to anatids. Mi- tochrondrial DNA analyses for two of the three genera (Thambetochen and Ptaiochen; Sorenson et al. 1999) have provided a phylogenetic hy- pothesis and estimates of minimum genetic di- vergence from anatid outgroups. The two genera form a well-supported clade that is itself sister to the "dabbling" ducks, although perhaps somewhat more similar to several South Amer- ican Anas or Anas relatives than to North Amer- ican dabblers (Fig. 2e). Molecular data do not support a close relationship with either tadorni- nes or true geese. The distance between the moa-nalos and their closest anatid outgroup, based on 1,009 mtDNA sites, is 6.9 _+ 0.5%. The Nn or Hawaiian Goose (B. sandvicen- sis) is the only extant representative of what ap- pears to be a minor radiation of Branta in the islands (Olson and James 1991, Paxinos 1998; E. Paxinos et al., unpubl data.). Nene are clearly derived from Canada Geese (B. canadensis; Quinn et al. 1991), and distances based on mtDNA restriction fragment and cyt b sequence data suggest that the two taxa shared a common 58 STUDIES IN AVIAN BIOLOGY NO. 22 ancestor sometime within the past 1 Ma (Quinn et al. 1991). At least two, and probably more than three additional Branta species existed in the islands (Olson and James 1991, Paxinos 1998; E. Paxinos et al., unpubl. data). One of these, the "very large Hawai'i goose" is the largest land vertebrate known from Hawai'i and is restricted in distribution to the island of Ha- wai'i (Giffin 1993). The species is highly mod- ified morphologically with a massive body, short, stout wings (it was flightless, but may have used its wings for fighting; S. Olson, pets. comm.); and cranially quite similar to the moa- nalos. In fact, it appears to be a superb example of convergent evolution to the moa-nalos. Mi- tochrondrial DNA sequence analyses (Paxinos 1998) strongly support placement of the very large Hawaiian goose Branta and also indicate a sister taxon relationship with the Nene and its close, larger relative, B. hylobadistes. Two species of ibis (Apteribis) have been de- scribed from subfossil material (Olson and Wet- more 1976, Olson and James 1991). Apteribis had stouter legs and shorter wings than other ibises and were flightless. The two or more spe- cies were limited to Maul Nui, and the discon- nection of Maul, Lana'i, and Moloka'i 0.3-0.4 Ma ago may have initiated the speciation event(s). Analyses of mitochondrial 12S ribo- somal DNA sequences of Apteribis and 21 other ibis species (Fig. 2a; A. Cooper, S. Olson, H. James and R. Fleischer, unpubl. data) indicate that the closest sister taxon to Apteribis is the New World White Ibis (Eudocimus albus). The Kimura 2-parameter pairwise distance between the two taxa for 407 bp of 12S rRNA sequence is 3.2 _+ 1.0%. SINGLE DIFFERENTIATED SPECIES Two raptors, a duck, and two songbirds rep- resent single differentiated species. These taxa apparently colonized the islands and differenti- ated considerably from their ancestors but did not undergo subsequent speciation. The two rap- tors are the endangered Hawaiian Hawk or 'Io (Buteo solitarius) and an extinct accipiter-like harrier (Circus dossenus). The 'Io is currently restricted to the island of Hawai'i but has been found in fossil form on other islands (Olson and James 1991; S. Olson, pers. comm.). Like many other species of Buteo, the 'Io exhibits a light and a dark color morph. Preliminary phyloge- netic analyses of more than 1,500 bp of mtDNA sequence in 18 species of Buteo (R. Fleischer, E Cordero, C. Mcintosh, I. Jones, and A. Helbig, unpubl. data) provides weak support for a clade containing the 'Io, the North American Short- tailed Hawk (Buteo brachyurus; to which it is least divergent; Fig. 2b), the North American Swainson's Hawk (Buteo swainsoni; as suggest- ed by Mayr 1943), and the endemic Galpagos Hawk (Buteo galapagoensis). The 'Io does not have a close relationship with any Old World Buteo we assessed. The Kimura 2-parameter (Kimura 1980) corrected sequence divergence from Buteo brachyurus is only 1.4 + 0.8% for part of cyt b. We have no molecular data for the extinct and highly modified Circus. The Laysan Duck (Arias laysanensis) is a rel- atively differentiated, small duck whose very small and vulnerable wild population inhabits only the tiny leeward island of Laysan. It has been consistently classified as either a subspe- cies of the Hawaiian Duck (Arias wyvilliana), hereafter referred to as Koloa, or of the Mallard (Arias platyrhynchos) on the basis of morphol- ogy and allozyme data (see Amadon 1950, Liv- ezey 1991, Browne et al. 1993). Recent DNA analyses (Cooper et al. 1996; J. Rhymer, unpubl. data), however, have strongly countered the above scenarios, indicating instead that the Lay- san Duck is differentiated from the Koloa and Mallard and may be more closely aligned with the South Pacific Black Duck (Arias supercilio- sa) clade. The Koloa, on the other hand, does cluster closely with the North American Mallard or Mottled Duck (Arias fulvigula) clades. Anal- yses of mitochondrial control region sequences of subfossil bones (Cooper et al. 1996) have also revealed that the Laysan Duck occurred in the main Hawaiian Islands well into the period of Polynesian settlement, and in forested habitats and higher elevations (> 1,500 m) not considered typical for a dabbling duck. The level of mito- chondrial control region sequence divergence between the Laysan Duck and its closest out- group taxon is about 10%; overall mtDNA di- vergence is lower than this (J. Rhymer, unpubl. data). The fourth "nonradiating" species, the 'Ele- paio (Chasiempis sandwichensis), is polytypic at the subspecies level and occurs on the islands of Kaua'i, O'ahu, and Hawai'i (enigmatically, no fossils have been found of this species on Maul Nui; James and Olson 1991). The 'Elepaio is likely related to Polynesian flycatchers in the ge- nus Monarcha (Mayr 1943, Amadon 1950) and is one of the few species for which differentiated subspecies have been identified on a single small island (Hawai'i; Pratt 1980). Molecular analyses of each island subspecies may, however, reveal differentiation sufficient to elevate them to spe- cies level. PROBABLE RECENT COLONIZATIONS Several taxa show little phenotypic diver- gence from mainland outgroups, suggestive of a very recent colonization (Table 1). These in- HAWAIIAN BIRD MOLECULAR SYSTEMATICS--Fleischer and Mcintosh 59 clude the Black-necked Stilt (Himantopus mex- icanus knudseni), Hawaiian Coot (Fulica alai), Common Moorhen (Gallinula chloropus sand- vicensis), Koloa, Black-crowned Night Heron (Nycticorax nycticorax hoactli), an eagle (Hal- iaeetus), and the Short-eared Owl. Of these, only the Black-crowned Night Heron is not currently considered to be distinct from mainland forms at the subspecies or species levels, but the Short- eared Owl, in spite of its subspecific designation, is thought to be a post-Polynesian colonist (Ol- son and James 1991). The Common Moorhen, Hawaiian Coot, Black-crowned Night Heron, and Short-eared Owl are extremely similar morphologically to outgroup relatives (Amadon 1950), but no DNA data currently exist with which to assess the age of their splits. As noted above, the Koloa is a very close relative of the Mottled Duck and Mallard (<3% mitochondrial control region di- vergence; Cooper et al. 1996). The endemic sub- species of the Black-necked Stilt differs from North American Black-necked Stilts (H. m. mex- icanus) by only about 1.5 + 0.6% sequence di- vergence in 447 bp of mtDNA control region (R. Fleischer et al., unpubl. data). The North Amer- ican Black-necked Stilts are considered to be the closest mainland relatives on the basis of mor- phology. Cyt b and 12S rRNA sequences from a subfossil bone of the extinct eagle (Haliaeetus sp.; Fleischer et al. 2000) are not different from the Old World White-tailed Eagle (H. albicilla), and the two species differ by 1.5% for the ATP- ase8 gene. Skeletal characteristics could not dif- ferentiate the Hawaiian eagle bones from either White-tailed Eagle or Bald Eagle (H. leucoce- phalus; Olson and James 1991). Thus, for at least three of these seven taxa the supposition of a recent split from a mainland ancestor and re- cent arrival in the islands is supported by the molecular data. SUMMARY: GEOGRAPHIC ORIGINS AND TEMPORAL FRAMEWORK Above we summarize recent molecular sys- tematic studies of the Hawaiian avifauna. We use these data to infer, if possible, the closest living relatives and the geographic origins of the Hawaiian taxa we sampled. Our biogeographic analyses indicate (Table 1) that at least 9 or 10 of the --> 21 independent lineages appear to be of North American or at least Western Hemi- sphere origin, 4 appear to be of South Pacific or Australasian origin, 2 or 3 are of Asian origin, and 5 are of currently unknown geographic or- igin. Thus Mayr's (1943) conclusion that about half the Hawaiian avifauna is of American origin is still supported by our molecular data. We found a relatively low level of molecular divergence between the Hawaiian taxa and their closest non-Hawaiian (mostly mainland) rela- tives (i.e., from zero to 10.3% sequence diver- gence for 14 lineages). Based on these results, none of these Hawaiian lineages split from mainland ancestors earlier than about 6.4 Ma. In fact, most of our estimates, although rough and lacking meaninglhl standard errors, fall well within the period of formation of the current set of main islands (i.e., Kaua'i at 5.1 Ma and later, Fig. 1). Only the drepanidines (10.3%), the corvids (8.4%), and perhaps the moa-nalos (6.9%) and the thrushes (6.7%) have Kimura 2- parameter sequence divergences from mainland relatives that suggest colonization prior to even the formation of O'ahu (3.7 Ma), and in each of these cases we may not have obtained sequence for the closest mainland outgroup (which we may not have sampled or it might be extinct). The overall picture suggests that while native Hawaiian Drosophila (Beverley and Wilson 1985, Thomas and Hunt 1991, DeSalle 1992, Russo et al. 1995) and lobeliads (Givnish et al. 1995) may have colonized the archipelago well before the formation of Kaua'i, thus far we have little evidence that any bird lineages have done SO. These findings lead us to consider factors be- yond simple isolation by distance and the an- thropogenically induced Holocene extinction that may help to explain Hawai'i's low primary avian diversity. First, the unique geology of the islands (Carson and Clague 1995) results in a situation in which individual islands have a lim- ited "lifespan" (--5-7 Ma) as a high island. Lin- eages that have colonized older islands, but for some reason cannot succeed onto younger is- lands, will be ultimately lost as their island dis- appears into the sea (this may be especially true for forms that have evolved to be flightless). There may be reduced chance for taxonomic di- versity to build up over long evolutionary peri- ods relative to archipelagos with longer surviv- ing islands. Secondarily, what secondary enrich- ment of avifaunal lineages by speciation that does occur in the islands may allow "niches" to be filled (perhaps by now locally adapted taxa) such that they are no longer available for occu- pation by new (and not locally adapted) colo- nists from elsewhere. Thus, primary diversity could be reduced by competitive exclusion. Continued paleontological research in the is- lands combined with studies of DNA sequence variation should help us to address these hy- potheses. We hope these new fossils and se- quences will continue to shed light on the sys- tematics, biogeography, and timescale of avian evolution on the Hawaiian conveyor belt. 60 STUDIES IN AVIAN BIOLOGY NO. 22 ACKNOWLEDGMENTS We thank C. Tarr, S. Olson, H. James. E. Paxinos, A. Cooper, B. Slikas, J. Rhymer, B. Arbogast, S. Co- nant, M. Sorenson, T. Quinn, A. Helbig, I. Jones, A. Driskell, and T. Pratt for information concerning and discussion of many of the topics covered in this paper, and C. Tart, B. Slikas, J. M. Scott, reviewer #1, and especially H. James for comments on an earlier draft of the manuscript. Samples for many of our analyses were provided by museum or field collections of tis- sues and we gratefully acknowledge the cooperation of C. Kishinami and A. Allison (B. P. Bishop Museum), S. Conant (University of Hawai'i), E Sheldon (Loui- siana State University), M. Robbins and B. Slikas (Academy of Natural Sciences-Philadelphia), P. Bruner (Brigham Young University-Hawai'i), E. Bermingham (Smithsonian Tropical Research Institute), S. Olson, P. Angle, and M. Braun (U.S. National Museum), R. Cann (University of Hawai'i), S. Rowher and S. Ed- wards (Burke Museum), and C. Cicero (Museum of Vertebrate Zoology-Berkeley). We greatly appreciate permission from Dave Swofford to use PAUP* (a gem of a program). Funding for many of the results pre- sented above was provided by the Smithsonian Insti- tution Scholarly Studies Program, Friends of the Na- tional Zoo, U.S. National Science Foundation, U.S. Fish and Wildlife Service, the National Geographic Society, and the Biological Resources Division of the U.S. Geological Survey. Studies in Avian Biology No. 22:61-67, 2001. EVOLUTIONARY RELATIONSHIPS CONSERVATION OF THE HAWAIIAN ANATIDS JUDITH M. RHYMER Abstract. The Hawaiian Duck or Koloa Maoli (Arias wyvilliana), hereafter retErred to as Koloa, and Laysan Duck (A. laysanensis) are two endangered species of waterfowl in the mallard complex that are endemic to the Hawaiian lslands. These nonmigratory, nondimorphic species were thought to be derived from stray migratory, sexually dimorphic common Mallards (A. platyrhynchos), that subse- quently lost the dimorphic plumage character. Laysan Ducks currently occur only on the tiny island of Laysan, while Koloa are found on O'ahu, Hawai'i, and, primarily, Kaua'i. Recent ancient DNA analysis shows that subfossil bones in deposits on the Big lsland, Hawai'i, belong to the extant Laysan Duck. Similar fossils have been found on many of the major Hawaiian Islands, indicating that the species was formerly more widespread. Because of extensive hybridization between introduced Mal- lards and Koloa and the superficial morphological similarity between the Hawaiian taxa, their taxo- nomic status and phylogenetic relationships have been controversial. The perception that they may be subspecies of the Mallard, or even conspecific, has influenced their recovery programs. Molecular analyses indicate that Koloa and Mallard are distinct but very closely related species, whereas the Laysan Duck is very distinct from either. Some of the nondimorphic species in the mallard complex, such as the Laysan Duck, may have evolved from a nondimorphic ancestor rather than the common Mallard. Repeated bottlenecks, inbreeding, and small population size have likely contributed to a loss of genetic variation in the Laysan Duck, but it is now possible to plan a captive breeding program to preserve remaining variation for possible reintroduction of the species to other previously occupied Hawaiian Islands. Hybridization with Mallards is one of the factors contributing to the decline of Koloa on O'ahu and Hawai'i. The Kaua'i population represents a stronghold for the species, but thorough census data and basic information on the ecology of Koloa on Kaua'i, essential for devel- oping a specific conservation plan, are not available. Key Words: Arias laysanensis; Arias platyrhynchos; Arias wyvilliana; ancient DNA; hybridization; molecular phylogeny; reduced genetic variation; species limits. The Laysan Duck (Anas laysanensis) and Ha- waiian Duck or Koloa Maoli (A. wyvilliana), hereafter referred to as Koloa, are endangered species of waterbirds endemic to the Hawaiian Islands. Laysan Ducks are restricted to the tiny 370 ha island of Laysan in the northwestern Ha- waiian chain. They survived a severe bottleneck in the early part of the century, as their popu- lation was estimated to have plummeted to fewer than 10 individuals by 1911 (Moulton and Wel- let 1984). This precipitous population decline was caused by overhunting and by habitat de- struction by introduced rabbits. A ban on hunt- ing plus extermination of the rabbits allowed numbers of Laysan Duck to rebound to about 500 birds over the next few decades, but a se- vere drought in 1993 reduced the population to fewer than 150 individuals (Cooper et al. 1996), an indication of the extreme vulnerability of this species. Harsh environmental conditions on Lay- san Island likely represent less than optimal hab- itat for the Laysan Duck. Recent analysis of DNA isolated from late Holocene subfossils, found in lava tubes in for- ested habitats at elevations as high as 1,800 m on Hawai'i, indicates that they are Laysan Duck (Fig. 1), an indication that the species was once found elsewhere in the Hawaiian Islands (Coo- per et al. 1996). Similar subfossil anatid speci- mens found on O'ahu, Kaua'i, and Moloka'i suggest that the range of Laysan Ducks was once more widespread. This situation is not unique: remains of over 30 other, now extinct, passerine and nonpasserine avian species of late Holocene age have also been found on Kaua'i, O'ahu, Moloka'i, Maul, and Hawai'i (James and Olson 1991, Olson and James 1991). These pre- historic avian extinctions are attributed primarily to predation by Polynesians and introduced predators and to habitat destruction (Olson and James 1991). It may be possible to reintroduce Laysan Ducks to other islands, provided preda- tors such as rats, mongoose, feral cats, and dogs are controlled and wetland and upland nesting habitats are protected. Koloa once occurred on all the major islands in the lower Hawaiian chain except Lfina'i and Kaho'olawe (Griffin et al. 1989). The only sub- stantial population is now found on the island of Kaua'i, in montane areas and on the Hanalei Na- tional Wildlife Refuge. There are a few birds on O'ahu, but hybridization with introduced com- mon Mallards (A. platyrhynchos) is a serious problem there (Browne et al. 1993). The total population of Koloa has been roughly estimated at 2,500 birds, (2,000 on Kaua'i-Ni'ihau, 300 on O'ahu, 25 on Maui and 200 on Hawai'i; Engilis and Pratt 1993), but in reality, there are few 61 62 STUDIES IN AVIAN BIOLOGY NO. 22 lOO A. laysanensis Subfossil 1 Subfossil 2 63 99 A. platyrhynchos 2 I I  A. wyvilliana1 73  A. wyvilliana 2 A. platyrhynchos I A. sparsa I I I I I 0.12 0.09 0.06 0.03 0 Kimura's 2- Parameter Corrected Distance FIGURE 1. Neighbor-joining tree obtained with MECA, based on Kimura's 2-parameter corrected distances using mtDNA control region sequences (after Cooper et al. 1996). Bootstrap values are shown. Sequences from Holocene subfossils are compared to those of extant Hawaiian anatids. good data from which to estimate their current population size. Surveys do not cover montane streams and wetlands where most birds reside. In fact, little is known about their breeding ecol- ogy, reproductive success, movements, and an- nual habitat requirements. Specific conservation action has been limited, except for sporadic re- leases of captive-reared birds on O'ahu, Maul, and Hawai'i, which have had marginal success. Current recovery plans call for wetland protec- tion and management and removal of the threat of hybridization (USFWS 1985). Management will include water level control, predator con- trol, minimizing disturbance, improved census techniques, and monitoring of contaminants and avian disease. The Laysan Duck and Koloa are thought by some to be derived from perhaps two waves of stray migratory Mallards that became isolated on the Hawaiian Islands and subsequently lost the Mallard's sexually dimorphic plumage (Wel- ler 1980). They represent 2 of 14 closely related, nonmigratory, sexually nondimorphic species and subspecies in the worldwide mallard com- plex of waterfowl. The taxonomic status of many species in this complex has been contro- versial (e.g., Johnsgard 1961, Palmer 1976, Young and Rhymer 1998), and the specific sta- tus of Laysan Ducks and Koloa are no exception (Weller 1980). Detailed morphological analysis of the genus Anas by Livezey (1991) placed wy- villiana and laysanensis as sister species within a northern hemisphere mallard clade, not supris- ing given their close geographic distribution and small body size; they are about one-half to two- thirds the size of common Mallards. Livezey (1991) considered them to be lull species, as did Berger (1972) and the American Ornithologists' Union (AOU 1983). In other studies, their status has variously been described as (1) both Laysan Duck and Koloa as subspecies of the Mallard (Delacour and Mayr 1945, Johnsgard 1978, Wel- ler 1980); (2) Koloa as a subspecies of Mallard, but Laysan Duck as a full species (Ripley 1960); (3) Koloa as a full species with Laysan Duck as a subspecies of Koloa (Brock 1951b, Griffin et al. 1989); and (4) Laysan Duck as a full species that evolved from Koloa (Warner 1963). Three issues have been raised that have im- portant implications for conservation of the Lay- san Duck and Koloa: (1) recognition of species limits--are the Laysan Duck and Koloa distinct species from one another and from the common Mallard and, therefore, more worthy of protec- tion? (2) hybridization with introduced spe- cies-what is the extent of hybridization with introduced Mallards and is it a possible threat to the species' integrity of Koloa? (3) loss of ge- netic variation--have small population size and population bottlenecks led to a loss of genetic variation in Koloa and Laysan Duck? These is- sues are addressed using molecular genetic anal- yses of mitochondrial and nuclear DNA. METHODS MOLECULAR ANALYSIS As part of a larger study of phylogenetic relation- ships in the mallard complex of species, blood and/or muscle or heart tissue samples were collected from common Mallard (North America, N = 28; Europe, N HAWAIIAN ANATID CONSERVATIONsRhymer 63 20); Koloa (Kaua'i, N = 19), Laysan Duck (foun- ders of captive flock, Smithsonian Conservation Re- search Center, N 15), African Black Duck (A. sa), the nondimorphic sister species to the mallard complex (Cooper et al. 1996, Johnson and Sorenson 1998, J. Rhymer, unpubl. data; captive flock, Wildlife Preservation Trust, N = 1); and Green-winged Teal (A. crecca; N = 2) as an outgroup (Johnson and Sorenson 1998). DNA was isolated from each sample using stan- dard procedures (Rhymer et al. 1994). Mitochondrial DNA (mtDNA ) The two most variable domains of the mitochondrial control region (631 base pairs, bp, from the 5' and 3' regions) were amplified using primers developed for waterfowl (Cooper et al. 1996). DNA sequencing was done on an ABI automated sequencer (model 373A) and sequences were aligned using Geneworks © (lntelliGenetics, Inc.) and by eye. Single-copy nuclear DNA (scnDNA) Five lg of DNA from each individual were digested with l0 enzymes that recognize six-base sequences. Fragments in digested samples were separated on 0.7%-1.2% agarose gels and transfered to nylon mem- branes (MSI Magnagraph) via Southern (1975) blot- ting. One arian oncogene, v-myc (Alitalo et al. 1983) and five anonymous single-copy nuclear DNA (scn- DNA) clones were used as probes, for a total of 30 probe/enzyme combinations. Anonymous scnDNA clones were obtained using standard procedures (Quinn and White 1987a, Parsons et al. 1993). Two hundred ng of probe were labeled with 32p for each hybridization, and membranes were then exposed to Kodak XAR film for 24-72 hours. Amplified fragment length polymorphisms (AFLP) The amplified tragment length polymorphisms (AFLP) technique is based on the detection of genomic restriction fragments by polymerase chain reaction (PCR) amplification, which produces fing.erprints with- out prior sequence knowledge (Vos et al. 1995). Pro- tocols provided with the AFLP © Analysis System I and AFLP Starter Primer Kit (GibcoBRL) were fol- lowed. Briefly, this includes an initial restriction di- gestion of 150 ng genomic DNA with EcoR I and Mse I, followed by ligation of EcoR I and Mse I adapters, amplification of the restriction fragments, labeling of an EcoR I primer with [3,3P]ATP, reamplification with the labeled EcoR I primer and an Mse I primer, and separation of labeled, amplified fragments on a 6.0% denaturing polyacrylamide sequencing gel. Primers used were EcoR I (AAG) with Mse I (CAG), and EcoR I (AA) with Ma'e I (CAA). DNA ,fingerprinting with minisatellites. Five lg DNA were digested with Hae III, fragments were separated on agarose gels, and were then trans- fenced to nylon membranes via Southern blotting using standard procedures (Loew and Fleischer 1996). Mem- branes were hybridized with 32p labeled Jeffrey's 33.15 minisatellite probe and exposed to Kodak XRP-1 x- ray film for 24 hours. STATISTICAL ANALYSES Phylogenetic relationships using mitochondrial DNA control region sequences were estimated using maximum parsimony (PAUP 3.1.1; Swofford 1993) and the neighbor-joining algorithm (Saitou and Nei 1987) with Kimura's 2-parameter model (MEGA 1.01; Kumar et al. 1993). One thousand bootstrap replica- tions were performed to estimate robustness of tree topologies and decay indices (the number of additional steps in the shortest tree(s) without a given node) were also calculated (Bremer 1988). For AFLPs, alleles at polymorphic loci were scored as 1 (present) or 0 (ab- sent), and the resulting data matrix was also analyzed using maximum parsimony. A strict consensus of most parsimonious trees was calculated. For scnDNA data, genetic distances were estimated for each pair of species according to Nei's (1987) method for unmapped fragment data, using the anal- ysis package RESTSITE (vl.1; Nei and Miller 1990), which allows for the inclusion of multiple individuals of each taxon analyzed with several probe/enzyme combinations. Relationships among species were esti- mated using the neighbor-joining method. Data were not available for an outgroup for either AFLP or scnDNA analyses. Two methods were used to estimate genetic diver- sity within species. First, band-sharing coefficients were calculated from Jeffrey's 33.15 minisatellite DNA data, comparing unrelated individuals of Mal- lards (N = 5), Koloa (N = 5), and Laysan Ducks (N 5) on the same gel. Second, proportion of polymor- phic loci (P) were calculated for each species using AFLPs, as the number of loci at which the most com- mon allele had a fYequency of less than 0.95 divided by the total number of individuals in the sample. RESULTS RECOGNITION OF SPECIES LIMITS Phylogenetic analysis of mtDNA control re- gion sequences indicate that there are two di- vergent lineages of common Mallards in the world (Figs. I and 2), one that has a Holarctic distribution (Mallard 1) and one that is appar- ently found only in North America (Mallard 2; Young and Rhymer 1998). The Koloa is very closely related to the Mallard, particularly lin- eage 2 (as are the other North American non- dimorphic mallard species, the Mottled Duck, A. fulvigula, and American Black Duck, A. rubri- pes; J. Rhymer, unpubl. data). Divergence of Laysan Duck from the Koloa/Mallard clade is well supported (Fig. 2). The occurrence of two divergent mtDNA Mallard lineages suggests either retention of an ancestral polymorphism or hybridization among taxa in North America. One of the problems that can arise from analysis of maternally inherited mtDNA is the possibility that the gene tree is not congruent with the species phylogeny (Arise et al. 1990). This possibility prompted analysis of biparentally inherited nuclear DNA molecular markers (scnDNA and AFLPs) to determine if 64 STUDIES IN AVIAN BIOLOGY NO. 22 lOO 14 -- A. platyrhynchos I 99 I A. platyrhynchos 2 8 A. wyvilliana --A. laysanens A. sparsa A. crecca FIGURE 2. Single most parsimonious tree obtained with PAUP 3.1.1, relating mitochondrial control region sequences for A. platyrhynchos, A. wyvilliana, A. lay- sanensis and A. sparsa (length = 127, CI excluding uninformative characters 0.73, RI = 0.70). Tree rooted with A. crecca as an outgroup. Branch lengths are proportional to the number of inferred changes along each branch. Decay indices are shown below each node; bootstrap values are shown above relationships among species suggested by the mtDNA results would be upheld. Only one lin- eage of Mallards was found at the nuclear level and both nuclear DNA datasets support the very close relationship between the common Mallard and the Koloa (Figs. 3 and 4). The divergence between Laysan Duck and Koloa was also high- ly repeatable, regardless of the nuclear DNA method employed. HYBRIDIZATION WITH INTRODUCED SPECIES Based on morphology, many of the Koloa-like birds on O'ahu appear to be hybrids. However, hybrid individuals are increasingly difficult to identify morphologically after more than one or two generations of backcrossing to one of the parental species (e.g., Rhymer et al. 1994). Mo- lecular methods provide an unambiguous as- sessment of the extent of hybridization and in- trogression between species. One putative Ko- loa/Mallard hybrid has been analyzed with mtDNA and nuclear markers so far. This indi- vidual was phenotypically similar to Koloa but possessed a Mallard 2 mitochondrial haplotype. Analysis of nuclear DNA using AFLPs indicates that the hybrid is indistinguishable from Koloa (Fig. 4). These data suggest that the hybrid in- dividual was not an F1 but a backcross into the Koloa. Further, because mtDNA is inherited only from the female parent, whereas AFLP loci are biparentally inherited, these data also indi- cate that the initial cross involved a female Mal- lard hybridizing with a male Koloa. Loss OF GENETIC VARIATION There is considerably less mtDNA haplotype diversity in both Hawaiian and Laysan Ducks than in the Mallard. Five to ten haplotypes (with minor changes) have been found in each of the two Mallard lineages (Avise et al. 1990; J. Rhy- mer, unpubl. data), whereas only two haplotypes are found in the Koloa and one in Laysan Duck (Fig. 1). Analyses of minisatellite DNA and AFLPs also indicate an apparent loss of varia- tion in Laysan Ducks (Table 1). Average num- bers of scorable bands for the Jeffrey's 33.15 probe were similar for Mallards and Koloa but much reduced in Laysan Ducks. Similarly, band- sharing coefficients for Mallards and Koloa are within the range (0.2-0.5) for unrelated individ- uals in outbred avian populations (Haig and Av- ise 1996), while those for unrelated Laysan A. platyrhynchos I A. wyvilliana A. laysanensis i I I I I I I  0.007 0.006 0.005 0.004 0.003 0.002 0.001 0 Genetic Distance (Nei 1987) FIGURE 3. Neighbor-joining tree obtained with RESTSITE 1.1, based on scnDNA data using Nei's (1987) distances for unmapped restriction sites for A. platyrhynchos, A. wyvilliana, and A. laysanensis HAWAIIAN ANATID CONSERVATION--Rhymer 65 wyvilliana 2 wyvilliana 2 wyvllliana 2 wyvilliana 1 wyvilliana x platyrhynchos 2 wyvllliana 1 wyvilliana 2 wyvilllana 1 wyvilliana 1 wyvilliana 1 wyvilliana 1 wyvllliana 2 wyvilllana 2 platyrhynchos 1 platyrhynchos 2 laysanansis laysanensia laysanansis laysanensis laysanensis laysanensls laysanensis FIGURE 4. Strict consensus of 53 most parsimonious trees relating variation in AFLP loci of the two Mallard mtDNA lineages (platyrhyncos 1 and 2), Koloa (wyvilliana 1 and 2 refer to mtDNA haplotypes), and Laysan Duck (laysanensis). Several individuals are included to illustrate variation within and among taxa. One putative Koloa x Mallard hybrid with a Mallard 2 mtDNA haplotype clusters with Koloa, suggesting a backcross indi- vidual Ducks were extremely high (> 0.8; Table 1). On two AFLP gels, the proportion of polymorphic loci (P; corrected for sample size) in Laysan Ducks was only about one-tenth that of Mallards with Koloa intermediate to the other two species (Table l). TABLE 1. COMPARISON AMONG SPECIES OF NUMBER OF SCORABLE BANDS (+ SE) AND BAND-SHARING COEFFI- CIENTS (S), BASED ON JEFFREY'S 33.15 MINISATELLITE PROBE AND PROPORTION OF POLYMORPHIC LOCI (P), BASED ON AFLP DATA Minisatellite DNA AFLP a /4 scorable bands N (-+ SE) s N P Mallard 5 42.8 _+ 0.60 0.22-0.40 4 0.097 Koloa 5 43.4 _+ 0.62 0.30-0.51 16 0.049 Laysan Duck 5 34.6 + 0.26 0.82-0.90 10 0.014  104 of 401 scorable AFLP bands were variable. DISCUSSION Genetic analyses of both Laysan and Koloa provide insights into problems with systematics, hybridization, and loss of genetic variation of these endangered species that have important consequences for their conservation. Small body size of the two Hawaiian species places them together in recent systematic treatments based on detailed morphological analyses (Livezey 1991, 1993), but some plumage characters of the Koloa are more similar to Mallard than to Lay- san Duck. Adding to the taxonomic confusion, however, are results of a recent allozyme study that showed a deep split between Mallard and Koloa (an order of magnitude greater than all anatid and most other avian congeneric genetic distances previously observed; Arise and Aquadro 1982), but virtually no differences be- tween Laysan Duck and Koloa (Browne et al. 1993). In contrast, the analyses of mtDNA and nuclear DNA in this study show that the diver- 66 STUDIES IN AVIAN BIOLOGY NO. 22 gence of Laysan Duck from a Koloa/Mallard clade is robust, whereas the Koloa and Mallard are very closely related (with only a few spe- cies-specific diagnostic markers). It is possible that the anomalous protein results stem from the analyses of different tissue types in different samples, which could artificially inflate esti- mates of divergence among taxa. Using a clock calibration of about 8% se- quence evolution per million years (calculated for the more variable 5' end of the mitochondrial control region by Sorenson and Fleischer 1996), it is estimated that the Koloa may have diverged from the North American lineage of Mallards (Mallard 2) as recently as 130,000 years ago, but from the Holarctic lineage (Mallard 1) as long ago as 0.8 million years ago (Ma). Divergence of the Laysan Duck from both common Mallard lineages, as well as from the Koloa, also appears to be on the order of 0.8 Ma. The evolution of these species from A. sparsa (about 1.7 Ma) is well supported even when all 14 species and subspecies in the mallard complex are included in the analysis (J. Rhymer, unpubl. data). In ad- dition, some species in the mallard complex, such as the Laysan Duck, may well have evolved from the nondimorphic ancestor rather than the common Mallard (Fig. 2; see also John- son and Sorenson 1998, Young and Rhymer 1998). Confusion over the taxonomy and evolution- ary history of these species has been compound- ed by the propensity of introduced Mallards to hybridize whenever possible with some of the nondimorphic species in the mallard complex, e.g., Grey Duck (A. superciliosa) in New Zea- land (Rhymer et al. 1994), Black Duck and Mot- tled Duck in North America (Johnsgard 1967, Mazourek and Gray 1994), and the former Mex- ican Duck (A. diazi; Hubbard 1977). Extensive hybridization with introduced species can lead to a kind of genetic extinction of rare native flo- ra and fauna (Rhymer and Simberloff 1996). As a result, the specific status of the taxa involved can be called into question, with important con- sequences for the protection of some endangered species (Meffe and Carroll 1994, Avise and Hamrick 1996). Nevertheless, current thinking does not consider the retained ability to inter- breed as sufficient evidence to preclude specific status and protection (O'Brien and Mayr 1991, Rhymer and Simberloff 1996). Hybridization between Koloa and introduced Mallards on O'ahu has been so extensive that this population is no longer considered to have pure Koloa. Removal of the threat of hybridiza- tion is an essential component for the species recovery (USFWS 1985). As an aside, the con- tention that hybridization between Mallards and closely related nondimorphic species occurs pri- marily because females of the nondimorphic species are more attracted to the colorful Mal- lard male was not upheld in a detailed study of New Zealand Grey Ducks and introduced Mal- lards (Rhymer et al. 1994), and the same appears true for Koloa. Only one known Koloa x Mal- lard hybrid has been analyzed so far and this individual resulted from a Mallard female x Ko- 1oa cross. More importantly, there is a popula- tion of Koloa on Kaua'i that is largely unaffect- ed by hybridization, so far. Knowledge of the potential threat and the availability of diagnostic molecular markers can now help to monitor in- cursion of hybridization on this island. Apart from guarding against hybridization, detailed studies of Koloa ecology are of the utmost im- portance in understanding its population dynam- ics. It is the Kaua'i population that will provide a stronghold for the Koloa, so it is suprising that little is known about the ecology of this endan- gered species. Captive breeding programs and/ or translocations are a final resort. It is better to understand the species' ecology in planning the prevention of further declines. The Laysan Duck is in an even more precar- ious situation with fewer than 150 individuals surviving the drought of 1993 (Cooper et al. 1996). In this case, a captive breeding program seems warranted. Results of mitochondrial and nuclear DNA analyses indicate that repeated bottlenecks, inbreeding, and/or low population numbers have probably contributed to a loss of genetic variation in this species. Only one mi- tochondrial haplotype remains and the number of minisatellite DNA bands and polymorphic loci (using AFLPs) is reduced compared to that found for either the Koloa or the Mallard. High levels of band sharing among apparently unre- lated individuals suggest a history of inbreeding, similar to those observed in another species of endangered Hawaiian waterfowl, the Nn (Branta sandvicensis; Rave et al. 1994). Al- though few empirical data are available showing a direct link between loss of genetic variation (as indicated by molecular markers) and fitness (Lynch 1996), it is generally understood that adaptive evolutionary change is the primary means of responding to selective challenges (i.e., genetic variation is important for isolated species to adapt to environmental perturbations). All indications are that the beleaguered Laysan Duck does not adapt well to the harsh environ- mental conditions on Laysan Island. A captive program should be undertaken to reintroduce the Laysan Duck to other islands, provided preda- tors are controlled and the habitat protected. It is now possible to plan a captive breeding pro- gram to maximize maintenance of the remaining HAWAIIAN ANATID CONSERVATION--Rhymer 67 genetic variation in this species (e.g., Haig et al. 1990). We now know what the conservation issues are for the endangered Koloa and Laysan Duck and genetic considerations provide one starting point for developing comprehensive strategies to ensure their protection. ACKNOWLEDGMENTS I want to thank R. Fleischer for his encouragement over the years to work on the Hawaiian anatids. DNA fingerprints, using Jeffrey's 33.15 probe, were done in his lab at the National Zoological Park, while I was supported on a Smithsonian Institution Visiting Sci- entist Fellowship through the Migratory Bird Program. Thanks to A.J. Jeffrey for the use of his 33.15 mini- satellite probe. Also thanks to H. James and S. Olson for thought provoking discussions of evolution in Ha- waiian waterfowl. Special thanks to D. Heckel for his encouragement, for allowing me the use of his lab at Clemson University for mtDNA sequencing and AFLP analysis, and for support through an NSF EPSCoR grant to University of South Carolina. I also want to thank L. Gahan for her expert advice on AFLP tech- nology and E. Beedle for running sequences. The scnDNA RFLP analysis was done as part of a larger project on the North American mallard species com- plex at the Smithsonian Institufion's Laboratory of Molecular Systematics under the direction of M. Braun, while I was supported on a Smithsonian Insti- tution Molecular Evolution Postdoctoral Fellowship. I owe a deep debt of gratitude to the late Trish Sawaya for her many hours of enthusiastic and patient tutelage in lab techniques during that time. ,( Studies in Avian Biology No. 22:68-80, 2001. THE INTERPLAY OF SPECIES CONCEPTS, TAXONOMY, AND CONSERVATION: LESSONS FROM THE HAWAIIAN AVIFAUNA H. DOUGLAS PRATT AND THANE K. PRATT Abstract. The Hawaiian Islands, with their unique geological history and geographic position, pro- vide an excellent natural laboratory in which to evaluate currently competing biological (BSC) and phylogenetic (PSC) concepts of the species. Although the BSC as historically applied in archipelagic situations is shown to be flawed in producing overlumped polytypic species, it nevertheless remains the preferable concept for most practical purposes. A review of the taxonomic history and species limits in Hawaiian birds under both concepts reveals that, when properly applied, the BSC yields a species total remarkably close to that produced under the PSC, contrary to what many proponents of the latter have supposed. We propose that the widespread adoption of the PSC for conservation purposes is potentially harmful. The PSC trivializes the species taxon and introduces new problems of deciding when a population becomes diagnosable, the possibility that species could appear and disappear in a reticulate fashion, and the likelihood that genetically diagnosable but phenotypically identical, and therefore not field identifiable, populations could be ranked as species. All of these problems negatively impact such things as constructing credible and politically defensible lists of endangered species, the prioritization of limited conservation resources, and the gathering of field data. We contend the BSC is arguably a more rational concept that better supports the activities of both scientific and nonprofessional observers. Biological species limits in oceanic archipelagoes worldwide need to be reevaluated using modern concepts and technologies before rational conservation decisions can be made. Key Words: avian conservation; biological species; endangered species; Hawaiian Islands; phylo- genetic species; polytypic species; species limits. Avian systematists have recently joined in a great debate over the definition of species. The long-accepted biological species concept (BSC) of Mayr (1942a) has been challenged by a new one from the field of phylogenetic systematics, usually called the phylogenetic species concept (PSC). As defined by Cracraft (1983), a phylo- genetic species is a population or cluster of in- dividuals "diagnosably different from other such clusters, and within which there is a parental pattern of ancestry and descent." Because diag- nosability can be established by "any feature or set of features, ranging from single fixed nucle- otide substitutions to major phenotypic (but ge- netically based) features" (Zink and McKitrick 1995), the PSC would elevate virtually all iso- lated subspecies to species and add many more based on small populations with one or more distinctive traits. Zink and McKitrick (1995) and Zink (1997) summarized the debate and argue in favor of the PSC, whereas Mayr (1992), with recent support from Snow (1997) and Collar (1997), defended the BSC. For popular over- views of the controversy, see Myers (1988) and Sibley (1997). Many might regard this debate as purely ac- ademic. Recently, however, some conservation- ists have suggested that the PSC would better serve their purposes than the BSC, showing that such esoteric pursuits do, indeed, have relevance in the "real world." Hazevoet (1996) has even charged that the BSC "promotes the extinction of endemic birds," by classifying many distinc- tive island forms as subspecies. Because conser- vation efforts often focus only on "full" species (Collar et al. 1994), there is some validity to Hazevoet's claim. In this review, we use the avi- fauna of the Hawaiian Islands to demonstrate: (1) that proper application of the BSC in archi- pelagic situations can produce a species list much closer to one based on the PSC than has been previously appreciated; (2) that the BSC itself is sound and that the many problems with it cited by some conservationists and systema- tists arise from misapplication of the concept rather than weaknesses of it; and (3) that the PSC suffers from its own problems in practice such that a shift to it could be worse for con- servation than maintaining the BSC (Collar 1997). Because Hawai'i is the most isolated oceanic archipelago, with numerous large and ecologi- cally varied islands, it has long been regarded as a superb natural laboratory for the study of evo- lution and biogeography. With the possible ex- ception of Galfipagos's birds, Hawai'i's is the best studied of any insular avifauna, and repre- sents a much later stage of evolution than that of Darwin's younger islands, with a much higher level of endemism. Unfortunately, the Hawaiian Islands have also suffered considerably more ecological degradation (for a review, see Pratt 1994; Van Riper and Scott this volume) than the Galfipagos and have more extinct and endan- 68 SPECIES CONCEPTS IN HAWAIIAN BIRDSPratt and Pratt 69 gered birds than any comparable region. They also have the largest component of introduced species of any modern avifauna (Long 1981), but we will show that even alien birds can teach evolutionary lessons on islands. Thus Hawai'i's birds provide all the necessary ingredients for evaluating the relationship of the competing spe- cies concepts to each other and to conservation. They further provide an important counterpoint to Hazevoet's (1995) use of the Cape Verde Is- lands avifauna as evidence of the need to aban- don the BSC. AVIAN TAXONOMY IN HAWAi'I Most recognizable forms of Hawaiian birds were first described as separate species under the Linnean typological or morphological species concept. Even some forms no longer regarded as subspecies were so described (e.g., the three populations of Hemignathus virens wilsoni; Wil- son and Evans 1890-1899). All authors of the "classical period" of Hawaiian bird research (Wilson and Evans 1890-1899, Rothschild 1893-1900, Bryan 1901, Henshaw 1902a, Per- kins 1903) used a morphological species con- cept, although all were evolutionists. Perkins's (1903) "family tree" of the Hawaiian honey- creepers, an endemic taxon variously ranked as the Drepanididae, Drepanidinae, or Drepanidini, was the first phylogenetic treatment of any Ha- waiian birds. After the flurry of ornithological research in the islands around the turn of the twentieth century, a period of neglect ensued, with only a few scattered notes and papers on Hawaiian birds appearing over the next four de- cades, and avian taxonomy remained static. Elsewhere during this quiescent period, sys- tematists, with ornithologists prominent among them, were formulating the "modern synthesis" that culminated in Mayr's (1942a) classical def- inition of the biological species that has been memorized by generations of biologists. The BSC is operational, rather than morphological, and is based on the ability or inability of popu- lations to interbreed freely. It introduced the concept of polytypic species (comprising several subspecies) for clusters of morphological "spe- cies" that could or would interbreed in nature. It thereby created the vexing problem of how to classify distinctive isolated (allopatric) forms whose ability or willingness to interbreed cannot be objectively demonstrated. Mayr (1942a) sug- gested the use of "potential isolating mecha- nisms" to gain inferences as to what might hap- pen during a hypothetical future period of con- tact. He also suggested that systematists look to the degree of diflrence between related sym- pattic species as a guideline to evaluate allopat- tic forms in a given group. We will show that properly applied, these precepts lead to species lists that can be corroborated by other procedu- res, such as phylogenetic analyses and genetic studies. However, early practitioners of the BSC too often ignored their own fundamental guide- lines and engaged in hasty lumping of vaguely similar forms. One wag has dubbed the period "Lumparama." In many cases, no reasons other than general similarity and geographic separa- tion were ever stated for lumping closely related forms previously considered separate species (see numerous examples in Mayr and Short 1970). It was taxonomy by decree. Virtually all mid-century authors treated geo- graphically replacing island populations the same as such populations on continents, even when differences were striking and consistent. However, subsequent genetic studies (e.g., Boag 1988) showed that island colonization is a unique phenomenon that differs fundamentally from the kind of isolation that results from hab- itat fragmentation, glacial cycles, and other con- tinental phenomena. Diamond (1977) showed that speciation differed on islands as compared to continents, but his study suflred from the state of knowledge of the time in that several assumptions he made about Hawai'i in particular (e.g., that intraisland subspeciation has not oc- curred on islands smaller than New Zealand, but see Pratt 1980; that the Hawaiian Crow, Corvus hawaiiensis, represents a single colonization with no subsequent intra-archipelagal dispersal, but see Olson and James 1982b) have been shown to be false. The failure to appreciate the different character of insular allopatry was a ba- sic misunderstanding that contributed to over- lumping many island taxa. The problem was exacerbated by Mayr's (1942a, 1969) clearly stated belief that allopatric populations of uncertain status should be con- sidered subspecies. The "when in doubt, lump" precept may be appropriate for closely related isolates on continents (Snow 1997), but we will show that for traditional studies of archipelagic speciation, exactly the opposite bias ("when in doubt, split") is more likely to result in a species list that will stand up to independent corrobo- ration. Indeed, every recent study of strikingly marked insular "subspecies" of which we are aware has revealed potential behavioral or eco- logical isolating mechanisms to support recog- nition of the forms as separate biological spe- cies. Although Mayr (1942a) introduced the concept of the superspecies for strongly diflr- entiated allopatric species (allospecies), he stat- ed that (p. 170): "It would be an abuse of this concept if an author were to call every polytypic species, composed of insular and thus well- marked subspecies, a superspecies." Again, it 70 STUDIES IN AVIAN BIOLOGY NO. 22 now appears that the real abuse of the superspe- cies concept is its under use in insular situations. Subsequently, Sibley and Monroe (1990) modi- fied the Mayrian definitions for the BSC and rec- ognized many well-marked island "subspecies" as allospecies. Even Mayr himself (E. Mayr and J. Diamond, unpubl. data) has elevated many of his earlier (Mayr 1945) subspecies to allospe- cies. The first review of the Hawaiian avifauna to apply the "modern synthesis" was that of Bryan and Greenway (1944), who combined many geographically replacing morphological species. Areadon (1950) carried the process further, lumping many strikingly differentiated island forms into large polytypic species (his work dealt mainly with the honeycreepers, but he re- viewed the other land and freshwater species in an appendix). His classification exemplifies mid- century evolutionary thinking. For example, Amadon (1950) considered plumage color rela- tively unimportant as an isolating mechanism, despite the fact that birds are highly visual or- ganisms. The de-emphasis of coloration as a guideline to species limits was undoubtedly in- fluenced by numerous hybridization studies dur- ing the period that lumped such different-look- ing continental forms as the three North Amer- ican flickers (Colaptes spp.; Short 1965), the various "dark-eyed" juncos (Junco spp.; Mayr 1942b), "Black-crested" and Tufted titmice (Baeolophus spp.; Dixon 1955), Australian mag- pies (Gymnorhina spp.), silvereyes (Zosterops spp.), and many others (reviewed by Ford 1987), the "Northern" orioles (Icterus spp.; Sibley and Short 1964), Black-headed and Rose-breasted grosbeaks (Pheucticus spp.; West 1962), Eastern and Spotted towhees (Pipilo spp.; Sibley and West 1959), and numerous others. Some of these studies have withstood subsequent scrutiny, but many have not. The trend of the era led to lump- ing of such other taxa as Glossy and White- faced ibises (Plegadis spp.; Palmer 1962), Pa- learctic and Nearctic Green-winged Teal (Anas spp.; Delacour and Mayr 1945), "Black-shoul- dered" kites (Elanus spp.; Parkes 1958), the three "yellow-bellied" sapsuckers (Sphyrapicus spp.; Howell 1952), and Holarctic rosy-finches (Leucosticte spp.; Mayr 1927, French 1959), based solely on inference rather than actual stud- ies. Most of the latter lumpings have subse- quently been shown to be erroneous or ill-ad- vised. We will show that, among Hawaiian birds, behavioral and genetic studies virtually al- ways support the premise that those that look different, are different. Interestingly, although Amadon (1950) was applying the BSC, his work largely ignored the relatively little biological data available at the time and was based almost entirely on museum skins. But his study was state-of-the-art, and we should not be surprised that some of his polytypic "species" have sub- sequently been shown to be amalgams of several biological species (see section on 'Alauahios be- low). Amadon's (1950) classification of Hawai- ian birds remained the standard for three de- cades. The 1970s saw a renaissance in ornithological field studies in Hawai'i. Many observers, in- cluding the authors, confronted by overwhelm- ing potential isolating mechanisms among many very strikingly marked "subspecies," began to question Amadon's (1950) taxonomy. H. Doug- las Pratt conducted a complete review of avail- able data from a variety of lines of inquiry and combined it with new information on vocaliza- tions (Pratt 1996b), foraging behavior, nesting habits, and ecology to produce the first complete taxonomic revision of the endemic avifauna (Pratt 1979) since Amadon (1950). First appear- ing in a dissertation, his classification was the basis of that published by Berger (1981), who did not accept all of Pratt's splits at the species level. Berger's (and hence most of Pratt's) tax- onomy was then adopted by the American Or- nithologists' Union (AOU) Check-list (AOU 1983), which has been followed by most sub- sequent authors. Pratt et al. (1987) adopted all of Pratt's (1979) species limits, and in a series of papers expanding on his dissertation, Pratt (1982, 1987, 1989, 1992b) defended them, and all were eventually adopted by the AOU (1985, 1991, 1993, 1995). Shortly after H. Douglas Pratt's work became widely known, another new classification ap- peared in the form of a review of recently dis- covered subfossil Hawaiian bird remains (Olson and James 1982b). As further discoveries came to light, these authors revised their classification and presented an updated version in tabular form (Olson and James 199l). Their arrangement of genera differs irreconcilably (Conant et al. 1998, Pratt this volume) with that of Pratt (1979) and the AOU Check-list (AOU 1998) as revised, but at the species level the two classifications differ only slightly and could eventually agree totally. In a footnote, Olson and James (1991) expressed the view that "distinctive, allopatric, insular forms" are best regarded as species. Their spe- cies-level taxonomy is thus the closest yet to ap- plication of the PSC to the Hawaiian avifauna. During the 1970s, the first systematic studies of Hawaiian birds using the new technique of cladistics appeared. Raikow's (1977, 1986) an- atomical studies produced the first cladistic phy- logeny of Hawaiian honeycreepers (Pratt [1979] was influenced by this technique, but his first classification was not strictly cladistic). Since SPECIES CONCEPTS IN HAWAIIAN BIRDSPratt and Pratt 71 then, virtually all analyses of Hawaiian bird evo- lution have been cladistic. Until recently, cladis- tic methods did not affect decisions at the spe- cies level, but the PSC is itself an outgrowth of cladistic thinking (Cracraft 1983, Zink 1997). The recent split of the O'ahu 'Amakihi (Hern- ignathus fiavus; see below) was based solely on a reconstruction of phylogenetic history through the study of mitochondrial DNA and shows that some decisions by proponents of the BSC come surprisingly close to PSC reasoning. Among Ha- waiian birds, genetic studies at the molecular level have usually supported species limits de- termined by more traditional methods and are an important independent corroboration of them (Johnson et al. 1989; Tarr and Fleischer 1993, 1995; Fleischer et al. 1998). Indeed, many recent splits were not accepted until biochemical data supported them, but such data are not, in the operational sense of the BSC, biological (Green- wood 1997). Rather, biochemical systematists may base their decisions on the Mayrian tech- nique of comparing degrees of difference, in this case genetic, between allopatric forms and those between related sympatric ones, or on measure- ments of the length of time allopatric popula- tions have been evolving independently. Thus they implicitly subscribe to the BSC but deal with data that are outside the realm of traditional isolating mechanisms. THE SPECIES OF HAWAIIAN BIRDS The following is a review of all historically known Hawaiian land and freshwater birds and one nesting seabird whose species limits have been controversial. It shows that a near consen- sus on species limits has developed during the past decade. All lines of inquiry have contrib- uted to it, and the result is a species list, based on the BSC, that differs little from one based on the PSC. It also suggests that in practice, appli- cation of the PSC is not as simple as it first appears. HAWAIIAN PETREL The Hawaiian petrel breeds in barren alpine zones of the Hawaiian Islands, with the main colony near the summit of Haleakal on Maul. The birds' range at sea is poorly documented, but they are believed to remain in the central Pacific near Hawai'i year-round (Pratt et al. 1987). From the earliest days of its discovery, the similarity of the Hawaiian Petrel to the Dark- rumped Petrel (Pterodrorna phaeopygia) of the Galfipagos was obvious, and virtually all tax- onomists regarded it as an allopatric subspecies P. p. sandwichensis. With the advent of tech- nology that allowed detailed vocal comparisons of the two populations, differences in voice be- came apparent. Tomkins and Milne (1991) sug- gested that these differences were sufficient to be regarded as isolating mechanisms between species, and Sibley and Monroe (1993) recog- nized the Hawaiian Petrel (P. sandwichehsis) as distinct. This case demonstrates a longstanding and increasing appreciation among BSC propo- nents of vocalizations as isolating mechanisms. Recently, strong genetic divergence of the two petrels was demonstrated using allozyme elec- trophoresis (Browne et al. 1997) and as yet un- published mtDNA studies (G. Nunn fide R. Fleischer, pers. comm) had similar results. Be- cause of their genetic diagnosability and geo- graphic separation, the two forms would clearly qualify as phylogenetic species. ENDEMIC DUCKS The Hawaiian Islands have two endemic ducks that are apparent derivatives of the Mal- lard (Arias platyrhynchos). The form wyvilliana (Hawaiian Duck, hereafter referred to as Koloa) is known historically from the main islands, whereas laysanensis was historically restricted to Laysan. Both endemics were originally de- scribed as separate species, but Bryan and Greenway (I 944), Munro (1944), and Amadon (1950) considered them conspecific but distinct I¾om the Mallard. Delacour and Mayr (1945) lumped them all. For the next two decades most authors (e.g., Brock 1951 a, Bailey 1956, Warner 1963) followed the former taxonomy, but Ripley (1960) advocated species status for the Laysan Duck while keeping the Koloa a subspecies of Mallard. Alternatively, Berger (1972) consid- ered both endemics fiall species, whereas Weller (1980) again lumped both with the Mallard. Vir- tually all of these varied treatments resulted from subjective treatment of morphological characters with little consideration given to some rather obvious potential isolating mecha- nisms. For example, Mallards and their relatives are notorious hybridizers, especially in captivity. Yet Ripley (1960) indicated that captive Laysan Ducks failed to hybridize with Koloa when they had the opportunity. In a recent survey of wa- terfowl collections worldwide, only three of 46 collections holding Laysan Ducks reported that laysanensis hybridized with another duck spe- cies (M. Reynolds, pers. comm.). Ripley (1960) further described numerous ecological peculiar- ities of the Laysan Duck, but based his taxonom- ic reasoning solely on morphological characters such as distinctive downy plumage. For the Ko- loa, Pratt (1979) pointed out that migratory ducks form pair bonds on the wintering grounds, a fact overlooked by previous treatments of this complex. Koloa breed year-round (Swedberg 1967) and form pairs within sight of occasional 72 STUDIES IN AVIAN BIOLOGY NO. 22 wild Mallards. Swedberg (1967) further states that even on small ponds the local ducks tend to avoid wintering migrants, another obvious be- havioral isolating mechanism. The near total ge- netic swamping of Koloa by domestic Mallards on O'ahu (Browne et al. 1993) does not negate the inference gained from earlier, more natural situations. Species status for the two endemic ducks is now also supported by both laboratory and paleontological studies. Browne et al. (1993), using allozyme electrophoresis, pro- posed that A. wyvilliana and A. laysanensis are sister taxa, separate from A. platyrhynchos. Dis- covery of subfossil remains of what appeared to be laysansensis on the main Hawaiian Islands (Olson and Ziegler 1995) suggested prehistoric sympatry with wyvilliana. Sequencing of mt- DNA from the subfossil bones (Cooper et al. 1996, Cooper 1997, Rhymer this volume) indi- cated that they were close to the Laysan Duck but not the Koloa, strongly suggesting former sympatry. Rhymer's (this volume) results differ from those of Browne et al. (1993) in showing a close Mallard/Koloa relationship, with the Laysan Duck very distinct genetically. Whatever their phylogeny, these three forms appear to be good species under virtually any species con- cept. HAWAIIAN COOT All authors after Bryan and Greenway (1944) considered the Hawaiian Coot a subspecies of the American Coot (Fulica americana) until Pratt (1987) showed that its differences were of the same degree as those of other allospecies of the worldwide coot superspecies, and involved characters important in species recognition. He suggested it be classified as F. alai as originally described, and was followed by Sibley and Mon- roe (1990), Olson and James (1991), and the AOU (1993). Because it has consistent diagnos- tic characters that distinguish it from other coots, the Hawaiian Coot is also a phylogenetic spe- cies. HAWAIIAN STILT Like the coot, the endemic stilt of the Ha- waiian Islands has been regarded by most au- thors as a subspecies of its North American counterpart, the Black-necked Stilt (Himantopus mexicanus). It is behaviorally quite similar but has many distinctive plumage features (Pratt et al. 1987) as well as adaptations to the unique Hawaiian environment. Mayr and Short (1970) recognized eight species of stilt in the superspe- cies H. himantopus, including the Hawaiian H. knudseni, rather than engage in "partial dubious lumping with insufficient knowledge." They stated that some forms "will undoubtedly prove conspecific," and virtually no one followed their split. Olson and James (1991), without com- ment, ranked the Hawaiian Stilt as a full species. In light of what we now know about discrete plumage differences as indicators of relationship among island birds, that decision was probably sound. Under the PSC, the Hawaiian Stilt would unquestionably be a separate species because of its diagnostic plumage differences, and now mo- lecular data (Fleischer and Mcintosh this vol- ume) show large genetic divergence as well. It likely is a valid biological species. HAWAIIAN SOLITAIRES The relationship of the Hawaiian thrushes (Turdinae) to the American solitaires (Myad- estes) was hypothesized by the earliest research- ers (Stejneger 1887, 1889) but was not generally accepted until Pratt (1982) reviewed and ampli- fied the evidence supporting it. This classifica- tion has subsequently been corroborated by new osteological comparisons (Olson 1996) and ge- netic studies (Fleischer and Mcintosh this vol- ume). The various forms exhibit only slight vari- ation in plumage, but differ strongly in bill mor- phology and vocalizations. They might all have been considered conspecific except for the fact that two of them are sympatric on Kaua'i. The smaller of those, the Puaiohi (M. palmeri), has always been considered a separate species, but mid-century workers regarded all the others as conspecific. Pratt (1982) documented the vocal differences mentioned by early researchers and showed by playback experiments that these were effective isolating mechanisms, at least between the Kgma'o (M. myadestinus) of Kaua'i and the 'Oma'o (M. obscurus) of Hawai'i. The status of the then rare (and probably now extinct) form of Moloka'i (see Reynolds and Snetsinger this volume) and the extinct forms of O'ahu and Lgna'i had to be assessed by inference. Pratt (1982) recognized the Oloma'o (M. lanaiensis) as a species on the basis of its reportedly dis- tinctive song. He found that the named subspe- cies on Lgna'i (nominate) and Moloka'i (M. lan- aiensis rutha) could not be differentiated on the basis of plumage, but maintained the subspecies because of a reported difference in vocal behav- ior. Munro (1944) reported that the Moloka'i bird sang and the Lgna'i one did not. This dif- ference disappeared, however, when Oloma'o on Moloka'i fell silent as they became rare and thinly distributed (pers. obs. based on reports of various field workers). Although the bird has been observed, its song has not been heard for decades. Because the O'ahu specimens had been lost, Pratt (1982) only tentatively recognized the ',maui (M. woahensis) as an additional species. Following rediscovery of the two known speci- SPECIES CONCEPTS IN HAWAIIAN BIRDS--Pratt and Pratt 73 mens of the latter, Olson (1996) re-evaluated the O'ahu form and considered it a subspecies of M. lanaiensis pending comparison with subfossil remains from Maui (which lost its solitaire be- fore the arrival of ornithologists). He emended the name to M. lanaiensis woahensis. If this ar- rangement stands up to further scrutiny, it will represent a pattern of speciation unique in the Hawaiian Islands. Whether the three populations of Oloma'o are phylogenetic species is difficult to say, given our limited knowledge of them, but the O'ahu form has a stronger claim to status under the PSC than the other two because of its slightly different coloration and longer period of isolation (Moloka'i and Lna'i were joined with Maui to form Maui Nui during the last glacia- tion). 'ELEPA1OS Hawai'i's monarchine flycatchers comprise the endemic genus Chasiempis and are distrib- uted on Kaua'i (sclateri), O'ahu (ibidis; former- ly gayi but see Olson 1989), and Hawai'i (sand- wichensis), but are enigmatically absent from the Maui Nui cluster. The three island forms are strikingly different in coloration, but their voic- es, ecology, and general behavior are rather sim- ilar. Also, sandwichensis exhibits considerable intraisland variation and has three named forms (nominate, ridgwayi, and bryani) with zones of intergradation (Pratt 1980). The three major forms were first lumped by Bryan and Greenway (1944), and until very recently no one had chal- lenged that classification. Pratt (1980) regarded them as megasubspecies (Amadon and Short 1976) to emphasize the two different levels of differentiation. Reflecting their previously stated beliefs about distinctive island forms, Olson and James (1991) recognized three species without elaboration, and Olson (1996) maintained that classification. Conant et al. (1998) were the first to document behavioral and ecological differ- ences among 'elepaios. They showed that the obvious and diagnostic plumage differences are reinforced by other, more subtle potential isolat- ing mechanisms. Conant et al. (1998) recom- mended biological species status for the Kaua'i 'Elepaio (C. sclateri), O'ahu 'Elepaio (C. ibi- dis), and Hawai'i 'Elepaio (C. sandwichensis), and we endorse their conclusion. Whether the three subspecies of the Hawai'i 'Elepaio would be considered phylogenetic spe- cies is problematical because some of their ob- served intergradation may be primary and clinal rather than secondary (Pratt 1980). The three forms were presumably in constant genetic con- tact in the recent past, but because of habitat destruction the very distinctive Mauna Kea pop- ulation (C. sandwichensis bryani) is now an iso- late (Scott et al. 1986) with distinctive ecology as well as plumage. Preliminary studies of one zone of intergradation between C. sandwichensis bryani and C. sandwichensis ridgwayi on the southeastern flank of the mountain have found evidence of secondary contact with possibly some assortative mating (E. VanderWerf, pets. comm.). Thus C. sandwichensis bryani may be in the very earliest stage of speciation by the BSC. In a PSC view, none of the three intrais- land variants would be recognized taxonomical- ly while they remained in genetic contact, but presumably "C. bryani" is now a phylogenetic species. MILLERBIRDS The only Old World warblers (Sylviinae) na- tive to the Hawaiian Islands are restricted to the Northwestern Hawaiian Islands. The extinct Laysan Millerbird (Acrocephalusfamiliaris) and the endangered Nihoa Millerbird (A. kingi) have long been considered conspecific, but Olson and Ziegler (1995) split them without elaboration. Biological support for such a split is presented by Morin et al. (1997), although they maintained the single species. Certainly the differences be- tween them are of the same degree as those ex- isting between other Pacific island Acrocephalus (Pratt et al. 1987), and separate species status is probably warranted. They clearly are phyloge- netic species. The question has more than aca- demic significance because of recent proposals to introduce Nihoa Millerbirds to Laysan (M.P. Morin and S. L. Conant, pers. comms.). If they are a different species from the original Laysan bird, the proposal should perhaps be reconsid- ered. DREPANIDINE FINCHES The finches of Laysan and Nihoa present an instructive example of differing appearance as an indicator of biological isolating mechanisms. They differ strikingly in overall size as well as relative size of bill. Plumages are similar but di- agnostically different with females more diver- gent than males. Amadon (1950) and other mid- century authors regarded them as conspecific, but Banks and Laybourne (1977) split them after reporting very different molt and maturational sequences. Other authors (e.g., Ely and Clapp 1973, Clappet al. 1977) reported differences in nesting behavior, and Pratt (1979, 1996a) de- scribed vocal differences. Because of the many potential isolating mechanisms, all recent au- thors have recognized both Laysan Finch (Te- lespiza cantans) and Nihoa Finch (T. ultima) as both biological and phylogenetic species. Re- cently, proof of biological species status was re- ported by James and Olson (1991), who found 74 STUDIES IN AVIAN BIOLOGY NO. 22 fossil remains of both species together on Mo- loka'i. Genetically, the two differ to the degree expected between pairs of closely related but bi- ologically distinct species (Fleischer et al. 1998). These finches are one of many examples in which plumage differences that were dismissed by mid-century workers accurately predicted bi- ological species status. 'AMAKIHIS This is a group of small, black-lored olive green birds with down curved, short bills. The extinct Greater 'Amakihi (Hemignathus sagitti- rostris) of the island of Hawai'i had a longer, straighter bill and was probably a close relative, although some authors place it in the monotypic genus Viridonia. The 'Anianiau (H. parvus), a Kaua'i endemic, has a shorter, straighter bill and rather different coloration and, despite its occa- sional designation as "Lesser 'Amakihi," is probably not very closely related. Conant et al. (1998) reevaluated the morphological data and placed the 'Anianiau in the monotypic genus Magumma, and Fleischer et al. (1998) found ge- netic evidence to support their treatment. Both of these birds appear to have influenced the evo- lution of "typical" 'amakihis by character dis- placement: the Hawai'i form H. virens virens has the shortest bill in the complex, whereas the Kaua'i 'Amakihi (H. kauaiensis) has the longest (Pratt 1979). Mid-century authors regarded all typical 'amakihis as conspecific but almost al- ways noted the much larger bill of the Kaua'i bird. The bill is both longer and heavier with virtually no overlap in measurements with any other form (Conant et al. 1998). The larger bill results in different feeding behavior and general ecology. Vocalizations of the Kaua'i 'Amakihi are also distinctive (Pratt et al. 1987, Pratt 1996b). Nevertheless, Berger (1981), and the AOU (1983), failed to follow Pratt's (1979) split. After biochemical data (Johnson et al. 1989, Tarr and Fleischer 1993) corroborated Pratt's findings, the split was accepted (AOU 1995), although the Check-list Committee cited no "traditional" data in support of the change. Conant et al. (1998) summarized the numerous potential isolating mechanisms of the Kaua'i 'Amakihi. Surprisingly, Tarr and Fleischer's (1993) anal- ysis of restriction-site variation in mtDNA showed that the O'ahu 'Amakihi (H. chloris), which had never been considered a biological species in modern times, was genetically distant from the morphologically, ecologically, and vo- cally similar 'amakihis of Maul Nui and Ha- wai'i. Furthermore, their evidence indicated that the O'ahu birds were the sister taxon to H. kauaiensis and therefore could not be conspe- cific those of Maui Nui and Hawai'i. On this basis, the AOU (1995) accorded the O'ahu 'Amakihi species status. Then Fleischer et al. (1998) altered their earlier branching-sequence hypothesis as the result of a new analysis in- volving sequencing of mtDNA. They now be- lieve the O'ahu taxon is, after all, sister taxon to the Maui/Hawai'i forms. Still, the genetic dis- tance between the O'ahu 'Amakihi and its sister taxa is of the same order of magnitude as that between the Kaua'i and Maui/Hawai'i forms, so the species status of the O'ahu 'Amakihi is val- id. This example shows why caution is dictated in making taxonomic innovations based solely on a single genetic study. The AOU (1995) de- cision, though now upheld for different reasons, could easily have proven incorrect and may have been premature. Interestingly, the only clue that the O'ahu bird might be a separate species prior to the DNA studies was its distinctive plumage. Again, the character considered least important by mid-cen- tury workers was, in fact, the most telling. Male O'ahu 'Amakihi are more yellow below and more strikingly two-toned than other 'amakihis, with the typical pale eyebrow reduced to a small supraloral spot. Females are even more distinc- tive in being much less yellow or olive than oth- ers and especially in retaining as adults the pale wingbars seen in juveniles of all forms. O'ahu 'Amakihi can be distinguished from those of other islands with virtually 100% accuracy on plumage characters alone. Vocal differences, such as a higher pitched song (Pratt 1996b), may also exist but have not been adequately investi- gated. The same cannot be said of the remaining two forms. Separate names were originally proposed for the populations on Moloka'i, Lgna'i, and Maui, but both Amadon (1950) and Pratt (1979) found them inseparable. As a group they differ on average from Hawai'i birds in coloration and bill length (longer), but overlap is so broad that only extreme individuals could be diagnosed on characters alone (Pratt 1979). Thus they form a biological subspecies H. virens wilsoni. Whether practitioners of the PSC would consider this form a species is unclear because despite their obviously divergent histories, they are not com- pletely diagnosable on phenotypic characters. 'AKIALOAS 'Akialoas look like giant 'amakihis with ex- tremely long bills. All forms are extinct, making biological assessment difficult. Forms are known historically from Kaua'i, O'ahu, Lgna'i, and Ha- wai'i, but those from the central islands are known only from a handful of specimens. Their classification has produced a nomenclatural SPECIES CONCEPTS IN HAWAIIAN BIRDS--Pratt and Pratt 75 tangle (Olson and James 1995), and their sys- tematics is as yet unsettled. Most authors (e.g., Berger 1981, AOU 1983, Pratt et al. 1987, Sib- ley and Monroe 1990) follow Pratt (1979) in placing 'akialoas in a large genus Hemignathus defined on the basis of a suite of synapomor- phies (Conant et al. 1998) in coloration, plum- age sequence, and degree of sexual dimorphism, bill shape, and vocalizations, but Olson and James (1995) segregate them in their own genus Akialoa. (For a defense of "greater" Hemigna- thus, see Conant et al. 1998). At the species lev- el, the situation is historically complicated. Bry- an and Greenway (1944) lumped all forms, but Amadon (1950) recognized two species on the basis of the strikingly diflrent relative bill lengths of the Kaua'i and Hawai'i forms. Having seen only two immature specimens of the Lna'i form and none of the O'ahu one, he included both with the shorter-billed Hawai'i birds as Hemignathus obscurus and separated the Kaua'i 'Akialoa (H. procerus emended to H. stejnegeri by Olson and James 1995). Pratt (1979) and Ol- son and James (1982b) showed that the Lna'i and O'ahu 'akialoas were actually closer to the Kaua'i 'Akialoa in bill length, and lumped all forms again. The AOU (1983), however, main- tained Amadon's (1950) split. Pratt et al. (1987: 302) reviewed the situation and pointed out that if two species are recognized, the line of sepa- ration had to go between Lna'i and Hawai'i with resultant nomenclatural changes. They sug- gested the names Lesser 'Akialoa (H. obscurus) for the Hawai'i bird and Greater 'Akialoa (H. ellisianus) for the other three forms; the AOU (1997) eventually adopted this two-species clas- sification. But the situation is complicated by recent pa- leontological data. James and Olson (1991) de- scribed a second species of 'akialoa, H. upupi- rostris, from Kaua'i and O'ahu that was sym- pattic with the historically known forms. Addi- tionally Olson and James (1995) reported two sympatric prehistoric 'akialoas from Maui and a larger species sympatric with the Lesser 'Aki- aloa on Hawai'i, all as yet undescribed. Because the relationships of these forms are unresolved, Olson and James (1991, 1995) recommend rec- ognition of all described forms as species: Ha- wai'i 'Akialoa (Akialoa = Hemignathus obscu- ra), Maui Nui 'Akialoa (A. lanaiensis), O'ahu 'Akialoa (A. ellisiana), Kaua'i 'Akialoa (A. ste- jnegeri), and Hoopoe-billed 'Akialoa (A. upu- pirostris). Interestingly, plumage variation among the historically known forms is of the same degree as that in several other groups or pairs of species (e.g., 'amakihis, Hawaiian soli- taires, O'ahu and Maui 'alauahios) and is non- clinal (for illustrations of all forms, see Pratt in press). This case, perhaps more than any other, shows the folly of the old "if in doubt, lump" dictum. Obviously, 'akialoas cannot all be con- specific no matter what their interrelationships turn out to be. Presumably, the five species de- limited by Olson and James (1995) can be con- sidered phylogenetic as well as biological. NUKUPU'US The three island forms of Nukupu'u and the 'Akiap6l'au comprise another group of hon- eycreepers with long, hooked bills. Each was de- scribed in the 1800s as a separate species: Hem- ignathus lucidus from O'ahu, H. hanapepe from Kaua'i, H. affinis from Maui, and H. wilsoni from Hawai'i. Bryan and Greenway (1944) combined all four, but Amadon (1950) separated the 'Akiapl'au because of its unique straight, rather than decurved, lower mandible. This tax- onomy was supported by Olson and James' (1994) morphological studies and discovery of a specimen of Nukupu'u supposedly from Ha- wai'i (Olson and James 1994), indicating pos- sible sympatry. Thus the Nukupu'u and 'Akia- p61'au cannot even constitute a superspecies. Since Amadon's (1950) work, systematists have ignored the nukupu'u complex, and the AOU (1983) considered the Kaua'i and Maui forms as subspecies of H. lucidus. With all three taxa ex- tinct or nearly so, their classification must de- pend on careful study of the fewer than 100 specimens scattered among a dozen museums from Honolulu to Berlin. Ongoing studies by T. K. Pratt and J. K. Lepson (pers. comm.) reveal that measurements and coloration consistently, and in some cases strikingly, distinguish the three nukupu'us from each other. The PSC would certainly consider them three species, but it is likely that by the criteria of the BSC the same outcome would be reached. Fleischer et al. (1998) identified the nukupu'us as a good test case for seeking a match between genetic diver- gence and sequence of colonizing new islands as they emerge down the Hawaiian chain. 'ALAUAHIOS These small warblerlike birds of the genus Paroreomyza are confined to the central islands of O'ahu and the Maui Nui complex. Despite extreme interisland color variation that ranged from brilliant scarlet to dull gray, Amadon (1950) considered the four named forms of Pa- roreomyza conspecific with the two species of Oreomystis from Kaua'i and Hawai'i. Certainly, the inclusion of the brilliant scarlet Kkwahie (P. flammea) of Moloka'i, with yellow and green birds from O'ahu, Maul, and Lna'i, should have been a red flag indicating the exis- tence of more than one species. But Amadon 76 STUDIES IN AVIAN BIOLOGY NO. 22 (1950:166) stated that "variation from yellow to red is obviously accomplished readily and need not be considered as necessarily indicating spe- cific difference." Lumping the Moloka'i, Lana'i, and Maui forms meant that their striking differ- ences had to have evolved since the breakup of Maui Nui, a period we now know to have been as little as 10,000 years. In fairness, we should point out that such geological information was unavailable in the period in which Amadon (1950) worked. Pratt (1979) hypothesized that the fact that the Kakawahie was the largest Pa- roreomyza and had the heaviest bill, and the Maui/Lna'i form was the smallest with the smallest bill suggested character displacement during a period of sympatry on Maui Nui. Olson and James (1982b) found paleontological evi- dence of such sympatry and agreed that Paro- reomyza had to comprise more than one species. The other two Maui Nui forms, known histori- cally from Lna'i (montana) and Maui (newto- ni), are very similar, differing only in that the Lgna'i birds are slightly brighter dorsally. No one since Bryan and Greenway (1944) has ever suggested that they are other than a single bio- logical species, the Maui 'Alauahio (P. mon- tana), but whether they qualify as phylogenetic species is problematical. The slight but consis- tent color differences they exhibit, rather than evolving in 10,000 years, may represent frag- ments of a former interisland cline, such as that shown by 'elepaios on Hawai'i (Pratt 1980), in which paler birds inhabited the lower and drier parts of Maui Nui and darker ones the rain for- ests of Haleakalg. The relatively few specimens from west Maul do appear somewhat interme- diate in dorsal coloration. Questions such as at what point the fragments of a f)rmer cline be- come phylogenetic species show that the PSC is not free of subjective judgments (Collar 1997, Snow 1997). The O'ahu 'Alauahio (P. macula- ta), now possibly extinct, was considered con- specific with the Maui/Lgna'i bird by Olson and James (1982b), but later (James and Olson 1991 ) they joined other authors in separating it. Its bill is intermediate between those of P. fiammea and P. montana but the coloration of both males and females is clearly different and diagnostic (Pratt et al. 1987). '/KEPA Representatives of the drepanidine "cross- bills" (Loxops) were known from Kaua'i, O'ahu, Maui, and Hawai'i, with a distinct taxon on each island. Most forms are small birds with yellow or gray bills, the males red or orange- yellow, the females gray-green, and neither sex with any bold black patterning or other mark- ings. The Kaua'i form is so distinctive that it was at first placed in its own genus Chrysomi- tridops (Wilson 1890). It is larger, with a pro- portionally larger blue bill. Both sexes are pat- terned in yellow and green with a prominent dark mask and pale forehead, although males are brighter than females. Bryan and Greenway (1944) recognized two species of Loxops: L. ca- eruleirostris ('Akeke'e) for the Kaua'i form, and L. coccineus ('kepa) for the O'ahu, Maui, and Hawai'i forms. Despite the striking plumage dif- ferences, which he did not consider great, Ama- don (1950) believed it "by no means improbable that they all would interbreed freely were their ranges to overlap" and considered them all con- specific. Pratt (1979) showed that the plumage and bill differences were paralleled by others in vocalizations, but his recommendation of a re- turn to Bryan and Greenway's classification was not adopted by Berger (1981). Thus the AOU (1983) maintained Amadon's single species of 'fkkepa. Further research by Pratt (1989) and others (summarized by Lepson and Freed 1997, Lepson and Pratt 1997) revealed fundamental differences in nest construction and ecology. As a result, the AOU (1991) finally recognized the 'Akeke'e as a separate species. This is yet an- other case in which plumage differences pre- dicted potential isolating mechanisms in other aspects of the birds' biology. The status of the three named forms of 'ke- pa is less clear because the O'ahu form (wol- stenholmei) is extinct and known from only a few specimens, and the Maui one (ochracea) is very rare if not extinct and was never common in historical times. Males of each fbrm can be distinguished with near 100% accuracy on col- oration alone, but females are more difficult to identify visually. Whether the color differences are sufficient isolating mechanisms, in the ab- sence of other data, for recognition of O'ahu and Maui 'akepas as biological species is moot (Pratt 1989), and their status as phylogenetic species is likewise unclear. Perhaps biochemical data, as yet unavailable, will reveal clearer differences. 'APAPANES The 'Apapane (Himatione sanguinea) is found in montane forests throughout the main Hawaiian Islands with no geographic variation. A now extinct related form on low, unforested Laysan was long regarded as a subspecies, but Olson and James (1982b, 1991) regarded it as a species (H. freethi) without comment. Schlanger and Gillett (1976) had considered the Laysan Honeycreeper a relict of the days when Laysan was a high island, but Olson and Ziegler (1995) believed it to be a colonizer from the main is- lands that has speciated on Laysan. With dis- tinctive coloration (orangish rather than bright SPECIES CONCEPTS IN HAWAIIAN BIRDS--Pratt and Pratt 77 crimson body feathering, dingy pale brown rath- er than white undertail coverts), a shorter bill, and distinctive cranial osteology, it is unques- tionably a species under the PSC. Olson and Ziegler (1995) split it on the basis of unspecified osteological differences. Overlooked in most discussions are several obvious potential isolat- ing mechanisms of the Laysan Honeycreeper: distinctive song and song phenology (Rothschild 1893-1900); different feeding behavior (includ- ing often walking on the ground to forage among flowers; Fisher 1903); different nest placement and structure (Schauinsland 1899, Bailey 1956); and, most obviously, totally dif- ferent habitat. A previously unreported anatom- ical difference, noticed by H. Douglas Pratt in preparing illustrations (Pratt in press) is that the Laysan bird has differently shaped tips to its pri- maries, lacking or possessing in very reduced form the truncation that produces the 'Apapane's wing noise. It now appears highly unlikely that these birds, adapted to two different worlds, could successfully interbreed, much less do so freely. Although Fancy and Ralph (1997) con- sidered it a subspecies, future authors, including Pratt (in press), will likely split it, bringing the BSC and PSC into agreement on 'apapanes. SUMMARY A wealth of new morphological, behavioral, ecological, and genetic data have dramatically changed the systematics and taxonomy of Ha- waiian birds. For example, a comparison of Amadon's (1950) classification of Hawaiian honeycreepers with the one we outline above shows that for 40 named taxa, the number of biological species (if all that have been proposed are accepted) swells from 23 to between 34 and 38, the final figure depending upon the classifi- cation of 'fikepas and nukupu'us. Correspond- ingly, the number of taxa designated as subspe- cies has dwindled from 17 to 6 or as few as 2! These two poorly differentiated taxa (Maui Nui 'amakihi and Lfina'i 'alauahio) amount to small pickings indeed over which to debate the BSC versus PSC. The status of the 25 undifferentiat- ed, and therefore unnamed, island populations ('O'fi [Psittirostra psittacea], 'Apapane, and 'I'iwi on six islands, the three Maul Nui 'ama- kihis, and 'kohekohe [Palrneria dolei] on Mo- loka'i and Maui) does not change. Likewise, the 19 named populations of songbirds that are not honeycreepers have increased from 10 to 15 spe- cies, with one subspecies sunk, one in dispute, and one subspecies of Hawai'i 'Elepaio inter- grading clinally with the nominate race, and an- other isolated but with limited and, as yet, little understood secondary contact. Why is interisland endemism at the species level so striking in Hawai'i? The answer lies partly in the geographical setting: the Hawaiian Archipelago comprises moderately large islands with relatively few offshore islets and atolls in- habitable by landbirds. Distances between main island groups average 58 kin, a formidable crossing for most sedentary songbirds. Birds newly colonizing one island from another could become quickly isolated genetically by weight of numbers. Because the pool of potential im- migrants on neighboring islands is much smaller than would be the case if the source area were a continent or much larger island, conspecifics would arrive infrequently, and in low numbers they would enter a resident population number- ing in the hundreds of thousands at least. Thus, adaptation to local conditions would proceed al- most immediately without significant genetic in- put from ancestral populations, and evolution of endemic forms could proceed rapidly (Freed et al. 1987a). Grant (1994) found that Hawaiian native finches exhibit less variability in bill measure- ments than Galfipagos finches and attributed the difference to greater specialization in feeding habits, greater genetic distance among species, and near absence of hybridization. All of these comparisons relate to the very different geologic history of Hawai'i (Fleischer et al. 1998) as compared to the Galfipagos, a tighter cluster of islands of relatively much younger age (Grant 1986). Species saturation was achieved in both archipelagos primarily by adaptive radiation of descendants of very few successful transoceanic colonizations (Diamond 1977, Juvik and Austr- ing 1979), but levels of differentiation fit each unique situation. Because Hawaiian bird popu- lations become genetically isolated virtually from the start, they can quickly evolve differ- ences in plumage and voice, both of which are effective isolating mechanisms. Thus they soon become both biological and phylogenetic spe- cies, with only a brief period of intermediacy. The most straightforward case of this has been proposed by Fleischer et al. (1998), who provide genetic data indicating that the four 'amakihis originated from interisland colonizations that followed shortly after emergence of new islands in a conveyor-belt fashion as the archipelago moved across a mid-ocean "hot spot." Nevertheless, interisland colonizations in Ha- wai'i obviously proceeded in both directions to produce the species-rich faunas of each island as well as the several examples of intra-archipelag- ic double invasions (Myadestes on Kaua'i, Pa- roreomyza on Maui Nui, 'akialoas on several lands, etc.). Also, some Hawaiian birds are widespread in the islands with no detectable in- tedsland variation. The three Maul Nui 'amakih- 78 STUDIES IN AVIAN BIOLOGY NO. 22 is and the two populations of 'kohekohe are fragments that were panmictic during recent pe- riods of lower sea level, but other undifferen- tiated populations belong to species that disperse widely with relatively frequent intra- and inter- island movements. Despite huge historical pop- ulations and the widest geographic range possi- ble, two of those, the 'Apapane and 'I'iwi, are among the least genetically diverse of honey- creepers (Tart and Fleischer 1995, Jarvi et al. this volume). Both may have suffered recent se- vere genetic bottlenecks then expanded their populations and ranges, and the recently extinct 'O'fi, which has not been investigated geneti- cally, probably exhibited the same pattern. Ab- sence of interisland variability in five species of Hawaiian waterfowl reflects large scale interis- land movements and reproductively cohesive populations, as confirmed by banding studies (Engilis and Pratt 1993). The fact that far-rang- ing species move in both directions shows that not all speciation in Hawai'i has resulted from Fleischer et al.'s (1998) conveyor belt. Virtually all oceanic island avifaunas, though always de- pauperate in number of species as compared to continental areas, have very high levels of en- demism (Stattersfield et al. 1998). Isolated, geo- logically old archipelagos with large interisland distances, such as the Marianas, Carolines, Tua- motus, Marquesas, and many others, can be ex- pected to exhibit species-level endemism com- parable to that of Hawai'i as their avifaunas are re-examined for the presence of potential isolat- ing mechanisms. Reflecting upon the history of avian system- atics and taxonomy in Hawai'i, we repeatedly see that coloration, long regarded as relatively insignificant in determining species limits, may be the first and most reliable indicator. Consis- tent, unique vocalizations or discretely different bill size or shape also virtually always corre- spond to interspecific boundaries. In every case in which species limits determined on these bas- es have been tested by biochemical or paleon- tological data, decisions based on an enlightened use of traditional phenotypic investigations have been upheld. Far from being single characters that identify species, appearance and vocaliza- tions predict where other more subtle isolating mechanisms exist. Because the Hawaiian Islands could well be regarded as the quintessential oce- anic archipelago, the lesson is clear: island birds that look or sound different are very unlikely to be conspecific. Allopatric populations that have only average rather than diagnostic differences are little diverged genetically and can be rec- ognized as subspecies. The old prejudice that similar allopatric populations should be classed as subspecies until proven otherwise has not withstood the test of actual practice on oceanic islands, and the underlying assumptions that produced it must now be questioned or discard- ed, at least for insular taxa. Properly applied to island endemics, the BSC produces species lim- its comparable to those of the PSC, and further allows for the recognition of subspecies, a cat- egory the PSC would essentially eliminate (Snow 1997, Zink 1997). Because the taxonomy of island birds elsewhere in the tropical Pacific is still based largely on studies done in the first half of the century, we can anticipate a major increase in the number of biological species rec- ognized in the region when the data are re-eval- uated with the insights gained from the Hawai- ian experience. However, we caution future workers not to follow their predecessors in mak- ing taxonomic changes based solely on infer- ence. SPECIES CONCEPTS AND CONSERVATION Our paper began with, and was largely prompted by, the conflict between the BSC and PSC as debated by Hazevoet (1996) and Collar (1996). Because of the Hawaiian Islands' ex- tremes of location and geologic history, their birds define the issue better than any other iso- lated insular avifauna. However, the outcome is unexpected: most diagnosable, allopatric taxa can be argued to be biological species on the criteria that they either (1) are not sibling spe- cies, or (2) were formerly reproductively isolat- ed in sympatry but now live apart in contracted, relictual ranges, or (3) are genetically and mor- phologically distinct to a degree similar to re- lated biological species living in sympatry. A few recognizable taxa do not qualify by these criteria, but we question whether these are either truly diagnosable (e.g., Maui 'Amakihi) or evo- lutionary units (three subspecies of Hawai'i 'Elepaio). Changing views of biological species limits in Hawai'i has had surprisingly little impact on the course of conservation efforts because the U.S. Endangered Species Act of 1973 does not focus on, nor limit endangered status to, full species only. No named Hawaiian taxon deserving in- creased protection was omitted from the list be- cause of its designation as a subspecies. Al- though undiagnosable and unnamed populations were not considered federally, a few were in- cluded in an otherwise parallel list of popula- tions protected by the state of Hawai'i. Actual recovery efforts have been less encompassing, however, and reflect the need to engage in triage. Faced with a depressing list of 32 endangered birds, 13 of them on the brink of extinction, state and federal agencies focused their limited per- SPECIES CONCEPTS IN HAWAIIAN BIRDS--Pratt and Pratt 79 sonnel and funding on managing tractable spe- cies such as NCn½ (Branta sandvicensis), Koloa, Laysan Duck, Newell's Shearwater (Puffinus au- ricularis newelli), and Hawaiian Crow, or 'AI- ala. Beginning in the 1980s, recovery efforts be- gan to focus on restoration and protection of habitat, to the benefit of entire bird communities. In the mid-1990s, special programs were initi- ated for two more endangered birds, the Puaiohi and Po'ouli (Melamprosops phaeosoma). The fact that these projects were funded, and not one to restore the Hawai'i 'Amakihi on Moloka'i, shows that even with a program that focuses on populations, conservationists' attentions in Ha- wai'i as well as worldwide (Collar 1997) are in- evitably closely tied to the species concept. On what few phenotypic or genetic characters should one describe a phylogenetic species? Re- cent introductions of the endangered Laysan Finch, with subsequent rapid evolution in bill size (Conant 1988a), present proponents of the PSC with some yet-to-be resolved issues. For example, do diagnosable populations that evolved through founder effects and local ad- aptation in only two decades qualify as phylo- genetic species? If not, at what point would they? Further, if introductions result in the cre- ation of populations that are diagnosably distinct (Conant 1988b), and therefore are "new" phy- logenetic species, how can this technique con- tribute to the conservation of the parent popu- lation? A related situation is that some intro- duced birds in Hawai'i, such as House Sparrows (Passer domesticus, Johnston and Selander 1964), may already be phenotypically diagnos- able. Conservationists are unlikely to regard such introduced populations as endemic phylo- genetic species. As Fleischer (1998) has pro- posed, artificially fragmented populations of en- dangered species in Hawai'i could become diag- nosable at the molecular level through genetic drift and presumably therefore qualify as phy- logenetic species. Recovery actions cannot save endangered species when new "species" are created from recently fragmented or introduced populations. A second problem with the PSC is the pos- sibility that species can appear and then disap- pear in a reticulate fashion (Zink 1997) because their delimitation does not require genetic iso- lation. Consider again the example of the 'Ele- paio on Mauna Kea. Because its range is almost exactly congruent with that of the endangered Palila (Loxioides bailleui), it will be strongly af- fected by efforts to restore habitat for that spe- cies. If plans to connect the upper forests of Mauna Kea (the range of Chasiempis sandwich- ensis bryani) with the rain forests of Hakalau Forest National Wildlife Refuge (where C. s. ridgwayi occurs) succeed, broad contact be- tween two now isolated forms of 'Elepaio, each a potential phylogenic species, could be re-es- tablished, resulting in extensive interbreeding. As our Hawaiian examples show, automatic splitting of all populations with diagnosable dif- ferences (Cracraft 1997) under the PSC is not as simple in practice as it sounds (Collar 1997) and could undermine the use of such time-honored and successful management techniques as rein- troduction and habitat restoration. We agree with Collar (1997) that the PSC would trivialize the species concept and severely stretch limited re- sources without providing any rational basis for formulating conservation priorities. Even when, as in the United States, conser- vation authorities are enlightened about the sometimes arbitrary way that species limits are applied and protect endangered populations of whatever status, alpha taxonomy is still far more than just an academic exercise. Much more is involved in the conservation of island birds than just the decision as to which ones are officially listed as endangered. Often, the only information available on birds of remote islands comes from recreational birders, who seek out endemic spe- cies and generally ignore those that are "just subspecies" (see for example Pratt 1990, Wauer 1990a,b). One can argue the rationality of that mindset, but no one can deny that in the eyes of recreational birders, conservationists, and the general public, species status has almost magical properties. It is quite possible that many island species worldwide could become endangered or extinct without anyone noticing because birders ignored forms ornithologists called subspecies. Witness the case of the Island Scrub-jay (Aphel- ocoma insularis) endemic to Santa Cruz Island off California. Few birders were even aware of its existence before it was recognized as a spe- cies, but almost immediately afterwards, a small industry developed for the sole purpose of en- abling people to see the bird (Atwood and Col- lins 1997). Had this been an endangered species, we believe the increased population monitoring would have contributed data valuable to the bird's recovery. An example of the latter phe- nomenon is the case of Bicknell's Thrush (Ca- tharus bicknelli). No one voiced concern about its conservation status until it was elevated to species status (Thurston 1998). Attention f¾om birders may be important even before a bird is listed as endangered. For ex- ample, the O'ahu 'Elepaio was rarely sought out except on Christmas Bird Counts because, as a subspecies, it did not score differently with bird- ers. Thus, its sudden population crash in the past two decades (Pratt 1994) went largely unnot- iced. Now that it is a candidate for species status 80 STUDIES IN AVIAN BIOLOGY NO. 22 (Conant et al. 1998) as well as for listing as an endangered species (Conant 1995), birders have become more interested (Pratt 1993), and re- search on the species has resumed (VanderWerf et al. 1997, VanderWerf 1998a). Now, young visitors to Honolulu's Hawai'i Nature Center have their own species of 'Elepaio and 'amakihi on which to focus their local pride and interest. The fact that ecotourists would visit a locality for the sole reason of observing an endemic bird, or that school children take pride in and learn about their local avian specialty, increases public awareness and interest, especially in small countries with limited resources (Wille 1991) in whose hands the fate of many species ultimately lies. Collar (1997) cited the numerous valuable contributions of recreational birders to taxono- my through their worldwide travel, tape record- ing, photography, and note taking on breeding biology and general natural history and behav- ior. Janzen et al. (1993) even refer to birders as "parataxonomists" in recognition of their con- tributions. We support Collar's (1997) observa- tion that birders are today the ornithologist's most important ally in clarifying species limits and conservation status of birds, and managers of parks and reserves should encourage and fa- cilitate birding rather than discourage it as has all too often been the case in some Hawaiian reserves (Pratt 1993, pers. obs.). Conservationists and the general public need a rational and observable basis for species rec- ognition. By increasing the number of trivial look-alike "species" to a bewildering and over- whelming degree, adoption of the PSC could de- stroy scientific credibility with governmental of- . ficials and the general public who have little in- terest in or knowledge of the subtleties of tax- onomic philosophy. The BSC makes intuitive sense through its use of observable isolating mechanisms and the subspecies category for in- termediate stages, and provides a credible basis for conservation strategies. Although Hazevoet (1996) may be correct that "taxonomic neglect" promotes extinction of island birds, his proposed solution of switching to the PSC will actually increase such neglect by augmenting the taxo- nomic workload, providing a confused taxono- my for conservation practices (Collar 1997), and recognizing "species" that defy common sense. Besides, his main goal (increasing the number of recognized species on islands) can be accom- plished within the BSC without all of the dis- advantages of the PSC. Proper application of the BSC, including a long overdue review of the taxonomic status of island taxa worldwide, will do far more for avian conservation than adop- tion of the phylogenetic species concept. ACKNOWLEDGMENTS We thank J. M. Scott for the initial inspiration for this paper, and R. C. Fleischer for thought provoking comments and insights from his unpublished data. S. Conant, M. Morin, and M. Reynolds kept us apprised of current conservation efforts in the Northwestern Ha- waiian Islands. J. K. Lepson and E. VanderWerf shared their unpublished museum and field observations with us. Colleagues J. V. Remsen and F. Sheldon contrib- uted thoughts to our formulation of the species concept debate and commented on the manuscript. Studies in Arian Biology No. 22:81-97, 2001. WHY THE HAWAI'I CREEPER IS AN OREOMYSTIS: WHAT PHENOTYPIC CHARACTERS REVEAL ABOUT THE PHYLOGENY OF HAWAIIAN HONEYCREEPERS H. DOUGLAS PRATT Abstract. A Phylogenetic Analysis Using Parsimony (PAUP) of 39 phenotypic characters of myol- ogy, osteology, tongue morphology, bill morphology, plumage and coloration, behavior, and ecology produced a tree that strongly supports, with a few exceptions, current American Ornithologists' Union classification of Hawaiian honeycreepers (Drepanidinae). These results are compared with those from three different biochemical and genetics laboratories and those of a cranial osteology study. The honeycreepers, including the aberrant genera Melamprosops and Paroreomyza, are shown to be mono- phyletic and a subgroup of the Fringillidae. The Maui Parrotbill Pseudonestor xanthophrys is related to thin-billed taxa rather than to the drepanidine finches. The genus Hemignathus, the present limits of which have been widely challenged, is shown to be strongly supported by a large suite of characters, except that the parrotbill may belong in it and the 'Anianiau (H. parvus) should be removed from it and placed in its own genus Magumma. Hemignathus can be divided into four or five subgenera. The generic pairs Chloridops/Loxioides, Himatione/Palmeria, and Vestiaria/Drepanis can justifiably be lumped as Loxioides, Himatione, and Drepanis respectively. The genera Paroreomyza and Oreomystis are not closely related, and the latter includes the Hawai'i Creeper (O. mana). Synapomorphies of the two species of Oreomystis include: lack of adult sexual dimorphism; lack of wing-bars; distinctive juvenal plumages; bill shape and coloration; foraging behavior; flocking behavior; juvenal begging calls; and a simple, narrow, nontubular tongue unique among honeycreepers. Hypothesized relation- ships of the Hawai'i Creeper with '.kepas (Loxops) based on mtDNA studies, or to 'amakihis (H. virens and relatives) based on osteology, are incompatible with hypotheses based on a wide range of other characters. Key Words: Drepanidinae; Hawai'i Creeper; Hawaiian honeycreepers; Hemignathus; Magumma; Or- eomystis; Pseudonestor. The classification of the Hawaiian honeycreep- ers (Drepanidinae) has been controversial since the American Ornithologists' Union (AOU 1983) abandoned the longstanding classification of Amadon (1950) in favor of a new one based on Berger's (1981) use of my revision (Pratt 1979). This classification has been followed in most general references since, including Scott et al. (1986), Pratt et al. (1987), Sibley and Monroe (1990), and the AOU (1983, 1991, 1998), but its use has not been without criticism. Amadon (1986) felt that "the genera of the Hawaiian honeycreepers have been bandied about in rather cavalier fashion," and Olson and James (1995) bemoaned the wide acceptance of my classifi- cation "among non-taxonomists without any consideration having been given to its merits." Olson and James (1982) introduced a different classification, based largely on osteological stud- ies, that has evolved in subsequent works (James and Olson 1991; Olson and James 1991, 1988, 1995), but has not as yet been widely adopted. The two schools have come to agreement on several points, and the remaining differences in- volve primarily the limits of the genera Loxops and Hemignathus and the placement of the Ha- wai'i Creeper (Oreomystis maria of AOU 1998 or Loxops maria of James and Olson 1991) and 'Akikiki or Kaua'i Creeper (O. bairdi). James (1998) conducted a phylogenetic analysis of cra- nial osteology, the first study to include all taxa, both historical and subfossil. Her phylogeny (for historically known taxa only) is presented by Fleischer et al. (this volume). Recently, various allozyme (Johnson et al. 1989, Fleischer et al. 1998) and mtDNA studies (Tarr and Fleischer 1993, 1995; Feldman 1997; Fleischer et al. 1998; Fleischer et al. this volume) have sug- gested patterns of relationship that challenge both AOU (1998) and James and Olson's (1991) taxonomy. Because genetic technologies are still advancing, hypotheses of relationships based on them must be considered tentative. Each suc- ceeding study seems to change the picture, the various methods show little concordance in their results, and the various laboratories do not agree even when performing essentially the same anal- yses. To their credit, the authors of these studies have been very conservative in recommending taxonomic changes. Molecular studies virtually never mention phenotypic characters, the tradi- tional tools of systematists, because they con- sider such "adaptive" characters too subject to the vagaries of natural selection to be evolution- arily informative (R. Fleischer, pers. comm.). Also, no genetic study of Hawaiian honeycreep- ers has addressed the possibility that past hy- bridization could have a profound effect on per- 81 82 STUDIES IN AVIAN BIOLOGY NO. 22 ceived patterns of divergence, although hybrid- ization has been shown to have played a major role in the adaptive radiation of the similar-aged Darwin's finches (Grant 1994). Although DNA studies may ultimately answer all phylogenetic questions, I agree with Raikow (1986) that con- cordance testing with more traditional methods is still the only reasonable way to evaluate their hypotheses. In this volume, Fleischer et al. do exactly that by using data from mtDNA along with phenotypic osteological characters to assess the phylogenetic placement of the Po'ouli (Me- larnprosops phaeosorna). In the two decades since my first effort (Pratt 1979), many new pos- sible phenotypic synapomorphies have been dis- covered and others re-evaluated. Clearly now is the time to provide a cladistic analysis of this eclectic mix of traditional phenotypic characters, so that meaningful comparisons with genetic studies can be made. METHODS Scientific names used herein are those of the AOU (1998) unless otherwise noted. I conducted phyloge- netic analyses of 39 characters (Table 1) derived from studies of myology, osteology, tongue morphology, bill morphology, plumage and coloration, behavior, and ecology using PAUP* (Swofford 1999) and MacClade 3.01 (Maddison and Maddison 1992). Table 2 shows the data matrix. The first 3 characters were segregated to simplify some manipulations done with them. The 26 taxa include the chaffinches (Fringilli- nae) and cardueline finches (Carduelinae) as out- groups. Groups are coded as possessing a character if any included species does so. Question marks indicate gaps in the data. I have liberally used vernacular names for three reasons: 1) to be as taxonomically noncom- mittal as possible in entering the data; 2) to make my trees directly comparable to others presented in this volume that also use Hawaiian names; and, most im- portantly, 3) because these are the only available names that have remained unambiguous for two cen- turies. Phenotypic data are admittedly subject to some ma- nipulation by the investigator because characters can be described in various ways. Thus the coding of sev- eral characters requires explanation. in Character 21, for example, long sickle-shaped bills are found among 'akialoas (Hemignathux spp.) and in the 'l'iwi (Ves- tiaria coccinea) and mamos (Drepanis spp.), but they differ between the two groups in the nature of the bony support (Baldwin 1953). By combining two features, Character 21 codes this character without introducing known homoplasy. Tongue shape (Character 15) and bill shapes (Characters 19 22) could have been ap- proached several different ways, but I found that qual- itative descriptions worked better than quantitative ones. I also did not order these characters because whether they represent transtbrmational series is un- certain. Character 26 ('amakihi coloration) represents a suite of possibly synapomorphic characters that ap- pear to have evolved in tandem. 'Amakihi coloration includes: 1) plumage olive green dorsally; 2) under- parts yellow to olive green, paler than dorsum; 3) lores narrowly dark gray or black; 4) bill dark gray to black, usually with bluish base to mandible; 5) females and juvenals like males but less yellow; and 6) juvenals with at least faint wingbars. These characters must be grouped because they are not independent of one an- other. I applied similar techniques and the same data set (plus other characters) in a different analysis that will be explained under the discussion of the Hawai'i Creeper below. RESULTS With all characters at the same weight, I con- ducted a heuristic search that yielded a total of 390 equally parsimonious trees. From those, a 50% majority rule consensus tree (Fig. la) was computed that had a length (L) of 130 steps, a consistency index (CI) of 0.546, and a retention index (RI) of 0.720. The numbers on the lines indicate the percentage of trees that possess the branch shown. The result produced some appar- ent anomalies. Although the two 'alauahios (Pa- roreomyza montana and P. rnaculata) and the Kgkgwahie (P. fiarnrnea) stand apart as I pre- dicted (Pratt 1992b), the Po'ouli remains imbed- ded in the largest clade even though it also lacks the "defining characters" (Pratt 1992a), Char- acters 1-3 in Table 1, that presumably cause Pa- roreomyza to segregate in the tree. The differ- ence for the Po'ouli is that it possesses an inter- orbital septum (Characters 11-12) like those of other Hawaiian honeycreepers (Zusi 1978; James and Olson 1991; Fleischer et al., this vol- ume). Such a topology requires that the "defin- ing" characters be secondarily lost in Melam- prosops. This hypothesis lacks credibility be- cause: 1) only these three among the 46 char- acters are virtually exclusive to Hawaiian honeycreepers as compared with all other pas- serines; 2) they probably represent gene com- plexes rather than single loci; and 3) they were favored by natural selection in the Hawaiian en- vironment and retained in most of the drepani- dine taxa, so it is difficult to discern how a re- versal would be advantageous. If, as hypothe- sized by Pratt (1992a), drepanidine odor is a de- fense against predation, then for a lineage to lose it and have to compensate for the loss by the redevelopment of energy-taxing predator mob- bing behavior (which dreps with the odor also lack), is certainly counterintuitive if not unpar- simonious. Similarly, the loss of lingual wings (or conversely the development of a squared-off base to the tongue) seems unlikely to have oc- curred more than once among the honeycreepers because it has happened only one other time (among sunbirds) in the entire passerine order. A strict consensus tree of the same data set (Fig. lb; L = 125, CI = 0.576, RI = 0.715) collapsed HAWAI'I CREEPER--Pratt 83 many of the nodes and revealed a lack of reso- lution among most taxa (but note that the Hem- ignathus/Pseudonestor clade, discussed below, survives, as do pairings of mamos and 'I'iwi, Palila and Kona Grosbeak, and the two creep- ers). Consequently, I conducted a second analysis giving Characters 1-3 a weight of 2, with all others remaining weighted at 1. This run pro- duced 150 equally parsimonious trees. The ma- jority-rule consensus tree (Fig. lc; L = 136, CI = 0.551, RI = 0.723) has a much more intuitive- ly satisfying topology and is also more consis- tent with the findings of Fleischer et al. (this volume) and Pratt (1992a) with regard to Me- lamprosops. Furthermore, its topology is so ro- bust that most of it survives in a strict consensus tree (Fig. ld; L = 125, CI = 0.576, RI = 0.715). These consensus trees support a number of hypotheses, some of which have taxonomic im- plications: 1) the Hawaiian honeycreepers, in- cluding Melamprosops and Paroreomyza, are monophyletic; 2) Melamprosops and Paroreo- myza independently diverged from the "main line" of drepanidine evolution very early, before the "defining characters" of Pratt (1992a, b) evolved; 3) the drepanidine finches form a clade that does not include the 'O'fi (Psittirostra psit- tacea), Lina'i Hookbill (Dysmorodrepanis mun- roi), or the Maui Parrotbill; 4) the genera Chlor- idops and Loxioides are sister taxa, as suggested by James and Olson (1991); 5) the 'amakihis, 'akialoas, and "heterobills" form a clade that corresponds to the currently recognized genus Hemignathus (AOU 1998) except that 6) the 'Anianiau (H. parvus) is not included in it, as suggested by Conant et al. (1998); 7) Pseudo- nestor may be a Hemignathus; it is more closely related to the thin-billed taxa than to the drepan- idine finches as suggested very early by Perkins (1903) and later by Bock (1970) and Pratt (1979) but not accepted by the AOU (1983); 8) the remaining honeycreepers may divide into two clades along the traditional "red" vs. "green" lines; 9) several of the "red" genera are closely related and possibly warrant merger; 10) Paroreomyz. a is not closely related to Or- eomystis; which 11) includes the Hawai'i Creep- er. Several of these require further comment. DISCUSSION DREPANID1NE FINCHES Amadon (1950) placed all the drepanidine finches (except the hookbill, which he regarded as an aberrant specimen) in the genus Psittiros- tra rather than recognizing the five genera pre- viously named, most of which at the time would have been monotypic. This arrangement also re- flected his hypothesis that these birds' finchlike characters were secondarily derived from a thin- billed ancestor. Greenway (1968) split the genus into Psittirostra for the 'O'fi and Loxioides for the rest, and Banks and Laybourne (1977) ad- vocated re-establishment of the original five genera, primarily on the basis that Amadon's Psittirostra was morphologically too broad, and breaking it up reflected degrees of phenotypic divergence comparable to those among various mainland finch genera. With a cardueline ances- try fairly well established, Amadon's large Psit- tirostra also appeared to represent a paraphyletic assemblage based on plesiomorphies (Pratt 1979). Olson and James (1982b) maintained Amadon's Psittirostra but recognized five sub- genera. Later (James and Olson 1991), they rec- ognized all five genera, several of which by then had gained new members described from pre- historic remains, and added several new finch- like genera. Although my phylogeny would sup- port Greenway's (1968) classification, I would caution against making any sweeping taxonomic changes at this time. This study included rela- tively few characters that could differentiate the finch genera, so the apparent monophyly of the group could easily be an artifact. Any changes, with the possible exception of the merger of Chloridops and Loxioides suggested by both this study and James and Olson (1991 ), should await publication of James's (1998) dissertation, new fossil discoveries, and ongoing studies based on ancient DNA extracted and amplified from pre- historic remains (R. L. Fleischer, pers. comm.). MAUl PARROTBILL Not only does the parrotbill cluster with the thin-billed taxa contra previous classifications (Raikow 1977, AOU 1983), but it may belong in the genus Hemignathus. Once the conflation of its huge but fundamentally different bill with the large bill of the 'O'fi (Raikow 1977) is elim- inated, the similarities of the parrotbill to the hemignathines, especially the 'Akiap61'au (H. munroi), are overwhelming. Synapomorphies are as varied as a modified jaw muscle (Zusi 1989) and juvenile call notes (pers. obs.). Inter- estingly, the mtDNA phylogeny of Fleischer et al. (1998, this volume) also supports a close 'Ak- iap61'au/parrotbill relationship, although not necessarily the current composition of Hemig- nathus (see below). The parrotbill's tongue (Character 16) is unique among the honeycreep- ers, elongated with lateral and terminal projec- tions. It looks very much like a drepanidine tu- bular tongue that has simply been unrolled, and can easily be seen as derived from a tubular an- cestor. However, osteological studies (James 1998, Fleischer et al. this volume) group the par- 84 STUDIES IN AVIAN BIOLOGY TABLE 1. CHARACTER STATES FOR PAUP* ANALYSES OF HAWAIIAN HONEYCREEPERS NO. 22 Characters used in Figure 1 Defining characters of Hawaiian honeycreepers (Pratt 1992a): 1. Drepanidine odor 0. Absent 1. Present 2. Proximal end of tongue 0. With prominent "lingual wings." 1. Squared off, with no large backward projections. 3. Mobbing behavior 0. Present 1. Absent Anatomy: 4. Pattern of insertion of the 3 branches of M. flexor digitorum longus (Raikow 1978) 0. ABB 1. ABA 5. Condition of M. peroneus brevis tibial head (from Raikow 1978) 0. Absent 1. Present 6. Condition of M. pterygoideus retractor (Zusi 1989) 0. Not enlarged 1. Highly enlarged 7. Tibial head of the shank muscle M. peroneus brevis (Raikow 1977, 1978) 0. Absent 1. Present 8. Coracoidal head of the upper forelimb muscle M. deltoideus minor (Raikow 1977) 0. Absent 1. Present *9. Condition of M. plantaris (Raikow 1977) 0. Present 1. Absent 2. Variable within taxon. 10. Solid bony palate (Sushkin 1929, Amadon 1950) 0. Absent 1. Present 11. Interorbital septum thickness (Zusi 1978) 0. Thin, single-walled 1. Thick, double-walled 2. Thick, double-walled but with thin area in center 12. Fenestration of interorbital septurn (Richards & Bock 1973, Zusi 1978) 0. Large fenestrae 1. Solid 2. Small fenestrae or none (variable) 13. Floor of cranial fenestra in profile (Zusi 1978) 0. With hump or upward protrusion 1. Flat 14. Palatine process of the premaxilla (Bock 1960, Richards & Bock 1973) 0. Present 1. Absent ( fused) with lateral flange at anterior end 2. Absent (= fused) with reduced lateral flange Tongue adaptations: '15. Overall shape 0. "Nontubular, fleshy above, corneous below and caudolaterally" with "a rounded tip edged with small papillae" (James et al. 1989). 1. As above but "far less fleshy, more slender" (Gadow 1899). 2. Straight and shallowly troughlike (Richards and Bock 1973). 3. Thin, tubular for half or more of length. 4. Fleshy but narrow, with spoonlike tip (Bock 1978). HAWAI'I CREEPER Pratt 85 TABLE 1. CONTINUED Characters used in Figure I '16. Tongue margins 0. Smooth, not raised dorsad (Gadow 1899, Gardner 1925, Clark 1912, Amadon 1950, Raikow 1977, James et al. 1989). 1. Slightly raised, with short lateral and terminal laciniae at distal end (Gadow 1899, Richards and Bock 1973). 2. Slightly raised, with long lateral and terminal laciniae (Rothschild 1893-1900). 3. Strongly raised and curved inwards progressively toward tip, lateral laciniae interlaced distally (Gadow 1899, Raikow 1977). '17. Seed-cup modifications 0. Mixed within taxon. 1. No specialization for seeds 2. Seed-cup tip (Gadow 1899, Amadon 1950) Bill morphology (mostly pers. obs.).' 18. Nasal Operculum (Raikow 1977, James et al. 1989) 0. Not expanded downward 1. Partially developed 2. Expanded downward to nearly cover nostril 19. Finchlike bill shape 0. Finchlike 1. Finchlike but elongated (i.e. tanager-like) 2. Not finchlike *20. Unique morphologies 0. Bill shape represented elsewhere among passerines 1. Heavy, hooked maxilla 2. Heavy, parrotlike bill 3. Slightly crossed bill tips 4. "Heterobill" morphology '21. Sickle-shaped bills 0. Not sickle-shaped 1. Sickle-shaped, thin 2. Sickle-shaped, thick 22. lnflation of bill 0. Bill not inflated 1. Bill highly inflated, subglobose *23. Profile of gonys 0. Strongly convex 1. Slightly convex 2. Straight to slightly concave 3. Strongly concave Plumage and Coloration (pers. obs.) 24. Sparrow-like streaking 0. Present at least in juveniles 1. Never present *25. Juvenal plumage 0. No age-related plumage variation 1. Juvenile distinct but patterned like adult female 2. Juvenile patterned differently from either adult 26. Presence of "amakihi coloration" (see text for details): 0. Not present 1. Present 2. Present with secondary modifications 3. Present with loss of distinctive female and juvenile plumages 27. Purring or cooing wing note in flight 0. No 1. Yes 86 TABLE 1. CONTINUED STUDIES IN AVIAN BIOLOGY NO. 22 Characters used in Figure 1 28. Primaries with truncate tips 0. No 1. Yes 29. Plumage texture 0. Soft, non-shiny 1. Shiny or hardened *30. Predominant plumage colors 0. Yellow-green, yellow, or red 1. Black, red, and/or yellow 2. Brown and black 3. Dull green or gray 4. Variable in group Behavior and ecology '31. Song quality 0. Canarylike (Perkins 1903, Pratt 1996a) 1. Dissonant whistles, bell-like and mechanical sounds (Perkins 1903, Bryan 1908, Pratt 1996a) 2. Lively, quiet chittering (Engilis 1990, Kepler et al. 1996) 3. Lively whistles interspersed with call-like notes (Pratt 1992b, Pratt 1996a) 4. Song of simple trills or warbles (Perkins 1903, Henshaw 1902, Pratt 1996a) *32. Song complexity (Newton 1973; Pratt 1979, 1996) 0. Complex 1. Mixed complex and simple 2. Simple *33. Distinct juvenal call beyond fiedging 0. Absent or unrecognized I. Rapid juvenal begging calls in flocks (Scott et al. 1979; Fig. 2) 2. Evenly spaced "sound beacon" from solitary chick (BNA; pers. ohs.) 34. Whisper songs (Pratt 1979, 1996a, b) 0. No whisper song 1. Whisper songs similar to primary songs. 2. Whisper songs distinct from primary songs. 35. Nest sanitation 0. Absent at some point in nesting cycle (Newton 1973; van Riper 1980a; Pletschet and Kelly 1990; Morin 1992a, b; BNA) 1. Throughout nesting cycle. *36. Primary adult diet (Perkins 1903, Berger 1981, BNA) 0. Seeds 1. Soft fruits 2. Nectar 3. Mixed 4. Invertebrates 37. Nest construction roles (Newton 1976, Morin 1992b, BNA) 0. Construction by female only. 1. Construction mainly by female with limited help from male. 2. Construction by both sexes. 38. Size of territory (Newton 1976, BNA) 0. Large texTitories. 1. Small territories in immediate area of nest. *39. Display flights over breeding area (Newton 1976, Morin 1992a, BNA) 0. Absent 1. Present HAWAI'I CREEPER--Pratt 87 TABLE 1. CONTINUED Characters used in Figure 1 *40. Presence of red in plumage 0. Yes 1. No '41. Bill color 0. Pale throughout (may have darker tip) 1. Pale with dark culmen 2. Brown or gray with pale base 3. Black with bluish base to mandible 4. All black *42. Attenuation of bill 0. None 1. Slight 2. Moderate 3. Pronounced 4. Extreme *43. Presence of yellow in plumage (adult male) 0. Yellow head only 1. No yellow (or very little) 2. Yellow underlying entire plumage, nowhere bright 3. Yellow throughout plumage, with bright areas 4. Nearly all yellow. *44. Black or gray feathering in face 0. None 1. Broad, not confined to lores 2. Confined to lores *45. Presence of wing bars 0. Never present 1. Faint in juveniles, absent in adults 2. Present in juveniles only 3. Present in some adults *46. Color pattern of crown and supraloral area 0. Uniformly colored 1. Indistinct pale eyebrow 2. Bold, distinct eye stripe 3. Contrasting crown and forehead 4. Pale supraloral fleck Notes: All characters ordered except those with asterisks. Citations for every data point not given. Summaries are cited where useful. The abbreviation BNA refers to the Birds of North America series of the American Ornithologists' Union (Baird 1994; Fancy and Ralph 1997, 1998; Lepson 1997, Lepson and Freed 1997, Lepson and Pratt 1997, Pratt et al. 1997, Sixnon et al. 1997, Lindsey ct al. 1998; Olson 1998a,b,c; Snetsinger 1998; Baker and Baker 2000a,b; Sykes et al. in press). rotbill with two other taxa that have strongly hooked bills (')'fi and hookbill), but different tongues. This grouping could easily be viewed as the result of homoplasy or just superficial re- semblances. It is reminiscent of Raikow's (1977: 113) clustering of the parrotbill with the 'O'fi on the basis of their vaguely similar bill shape and the fact that such placement was "not refuted by other characteristics." That placement is now re- futed by many other characters, and the parrot- bill, despite its large bill, clearly belongs among the thin-billed taxa. However, I do not suggest merger of Pseudonestor and Hemignathus until the relationships are better understood, even though my findings seem to show that, with Pseudonestor excluded, Hemignathus is para- phyletic. HEMIGNATHUS AND moxoPs Except for the Pseudonestor problem, the above results clearly support current AOU (1998) taxonomy that restricts Loxops to the 'akepas and groups the 'amakihis, 'akialoas, and heterobills in Hemignathus. However, the cur- rent inclusion of the 'Anianiau in the latter ge- nus is not justified. For a detailed discussion of the reasoning behind these conclusions, see Co- nant et al. (1998). DNA studies also support rec- ognition of a monotypic Magumma for the 'An- ianiau. Tarr and Fleischer's (1995) restriction- 88 TABLE 2. TABLE 1 STUDIES IN AVIAN BIOLOGY NO. 22 DATA MATRIX FOR PAUP* ANALYSIS OF HAWAIIAN HONEYCREEPERS USING CHARACTER STATES FROM Character state Taxon 1 2 3 4 5 6 7 8 9 10 11 12 13 14 lfi 16 Chaffinches 0 0 0 I 0 0 0 0 0 0 0 0 0 0 0 0 Cardueline finches 0 0 0 0 1 0 0 1 0 I 1 I 1 1 0 0 Telespiaa finches 1 I I 1 ? 0 0 1 0 l 1 1 I I 0 0 Palila I I I ? ? ? ? ? ? ? 1 1 1 ? 0 0 koa finches 1 I ? ? ? ? ? ? ? ? 1 I 1 ? 0 0 Kona Grosbeak I 1 ? ? ? ? ? 9 ? ? I 1 I ? 0 0 '6'a I 1 I 0 1 0 0 1 0 ? I I 1 ? 0 0 Lfina'i Hookbill ? ? ? ? ? ? '? ? ? ? I 1 1 1 0 0 Po'ouli 0 0 0 '? ? ? ? ? ? ? 2 2 1 ? 4 0 KfikSwahie/'alauahios 0 0 0 ? ? ? ? ? ? 0 0 0 1 2 2 1 Maui Parrotbill 1 1 I ? ? 1 ? ? ? ? 2 2 I ? I 2 Hawai'i Creeper 1 1 1 ? ? '? ? ? 1 ? 2 2 1 I 2 1 'Akikiki 1 1 I 0 I ? 0 l 0 ? 2 2 1 ? 2 1 'kepas/'Akeke'e 1 I 1 ? ? ? ? ' ? I 2 2 1 1 3 3 'Anianiau 1 I I ' ? ? ? ? ? ? 2 2 I ? 3 3 Greater 'Amakihi I 1 ? ? ? ? ? ? ? ? 2 2 l ? 3 3 'amakihis I I 1 0 1 0 I 1 1 ? 2 2 1 2 3 3 'akialoas 1 I ? 0 I 0 I I 0 I 2 2 I 1 3 3 nukupu 'us 1 1 ? 0 I ? ? ? ? ? 2 2 I ? 3 3 'Akiap61fi'au 1 I 1 0 1 1 0 1 0 ? 2 2 I ? 3 3 'Ula-'ai-hawane l 1 ? ? ? ? ? ? ? ? 2 2 I ? 3 3 'Apapane 1 I 1 0 I 0 1 I 1 1 2 2 I 1 3 3 ',kohekohe I 1 1 I 1 0 1 1 I ? 2 2 1 ? 3 3 'I'iwi 1 I l 0 1 0 0 I 1 1 2 2 I 2 3 3 Black Mamo I 1 '7 ? ? ? ? ? ,? ? ? ? 1 ? 3 3 Hawai'i Mamo 1 1 ? ? ? ? ? ? ? ? ? ? 1 ? 3 3 fragment mtDNA study of a limited number of taxa found the 'Anianiau widely separated from the 'amakihis in a clade of its own. Fleischer et al.'s (1998, this volume) mtDNA sequencing study included additional taxa and grouped the 'Anianiau with the heterobilled 'AkiapOl'au (H. munroi) and the parrotbill. James's (in Fleischer et al., this volume) osteological phylogeny, how- ever, maintains the grouping of the 'Anianiau with the 'amakihis, which may reflect the su- perficial resemblance that led to the former name "Lesser 'Amakihi" and my own (Pratt 1979) uncritical placement of this species in Hemignathus before closer scrutiny (Conant et al. 1998). James and Olson (1991: Table 14) restricted Hemignathus to 'akialoas ald the heterobills, and later (Olson and James 1995) subdivided it and placed the former in a new genus Akialoa. They grouped the 'amakihis with the 'rkepas, 'Anianiau, and Hawai'i Creeper in Loxops. Thus constituted, Loxops would be close to Amadon's (1950) characterization (Pratt 1979, Conant et al. 1998). James's (1998) newly analyzed osteolog- ical data (Fleischer et al., this volume) provide no support for such an arrangement. In fact, her phylogeny not only supports restriction of Lox- ops to 'gkepas, but can be interpreted as sup- porting a large Hemignathus as currently rec- ognized. The 'amakihis, heterobills, and akialoas are members of a single clade even on osteolog- ical grounds, but the picture is complicated by the inclusion of the "red" honeycreepers in the same clade. This result reveals one of the weak- nesses of single-character or single-complex analyses. With only one suite of characters, the computer program has no way of distinguishing homoplasy or parallelism from synapomorphy. The bill morphologies among the "red" birds (i.e., short down-curved bills, long sickle-bills, etc.) parallel those found in Hemignathus, but other characters (i.e., behavior, plumage type, sequence of plumages, and vocalizations) show that these resemblances are not synapomorphic with similar morphologies among the "green" birds (Perkins 1903, Amadon 1950). I suspect that a combination of the osteological data with my own would resolve this discrepancy and bring James's (1998) phylogeny and mine into substantial agreement. With the red birds re- moved, James's uppermost clade fairly closely approximates Hemignathus as currently delim- ited (AOU 1998). Fleischer et al.'s (1998) mtDNA sequence phylogeny supports neither an enlarged Hemig- nathus nor an enlarged Loxops. In it, the heter- HAWAI'I CREEPER--Pratt 89 TABLE 2. EXTENDED. Character state 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 0 0 0 0 0 0 I 0 I 0 0 0 0 4 ? 2 0 0 1 0 0 0 0 0 0 0 0 0 0 1 0 I 0 0 0 0 4 0 0 0 I 0 0 0 1 1 1 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 1 0 2 0 0 0 0 I 0 I I 0 0 0 0 0 0 0 0 1 0 0 1 1 1 2 0 0 0 0 0 1 1 1 0 0 0 0 0 ? ? ? ? ? 0 ') ? ? 2 0 0 0 0 I 0 1 0 0 0 0 0 0 0 0 ? 1 ? 0 9 ? 1 1 1 1 1 0 0 0 1 1 0 0 0 0 0 0 0 0 0 ? I  ? 1 1 1 2 1 0 0 0 1 ? 0 0 0 0 0 v ? ? ? ? v v ? ? I ? I 0 0 0 0 1 I 0 0 0 0 2 2 0 0 0 I 4 2 ? 0 1 1 2 0 0 0 1 1 1 0 0 0 0 0 3 1 0 2 1 3 ? I 1 1 0 2 2 0 0 0 1 1 2 0 0 0 0 4 2 2 2 0 4 1 0 0 1 1 2 0 0 0 2 I 2 0 0 0 0 3 4 2 1 0 1 4 1 I 0 I 1 2 0 0 0 2 1 2 0 0 0 0 3 4 2 I 2 1 4 ') ? ' 1 2 2 3 0 0 2 1 1 0 0 0 0 0 4 2 0 2 0 4 0 I 1 I 2 2 0 0 0 2 1 1 0 0 0 0 0 4 2 0 2 1 3 1 1 0 1 2 2 0 0 0 2 1 0 3 0 0 0 0 4 2 ? ? ? 4 ' ' ' I 2 2 0 I 0 3 I 1 I 0 0 0 0 4 2 0 2 0 3 1 0 1 1 2 2 0 1 0 3 1 1 1 0 0 0 0 4 2 ? ? ? 3 ' ? ? I 2 2 4 1 0 3 I 1 I 0 0 0 0 4 2 '? ? ? 4 ' ? ' 1 2 2 4 I 0 1 I 1 I 0 0 0 0 4 2 2 2 1 4 0 ? 0 1 2 1 0 0 0 1 1 2 0 ? 0 1 1 ' ' ? ? ? ' ? ? ' I 2 2 0 0 0 2 1 2 0 I 1 1 I I 0 0 0 I 2 2 1 1 1 2 2 0 0 0 2 1 2 0 1 0 1 1 1 0 0 2 1 2 0 I 1 1 2 2 0 2 0 3 1 2 0 1 1 1 1 1 0 0 0 I 2 I I 0 I 2 2 0 2 0 3 1 0 0 1 0 0 I I ' ? ? ? 2 ? ? ' I 2 2 0 2 0 3 1 0 0 1 0 1 1 1 ' ? ? ? 2 ? ? ' obills group with the parrotbill and 'Anianiau, the 'amakihis are sister-group to the red birds, and the Hawaii Creeper is sister-group to the 'akepas. The analysis does not include the Greater 'Amakihi or the 'akialoas. According to R. Fleischer (pers. comm.) the branching se- quence among the thin-billed honeycreepers is not well defined by the techniques used in their study, so I believe we should await further de- velopments before tinkering with a taxonomy so well supported by phenotypic characters. Although the phenotypic data support a large Hemignathus, they also support the recognition of four (or five if Pseudonestor is included) sub- genera within it: Hemignathus for the hetero- bills; Akialoa for the 'akialoas; Chlorodrepanis for the "typical" 'amakihis; and Viridonia for the Greater 'Amakihi. The latter two cannot be combined as has been done in the past (Green- way 1968) because such a construct would be paraphyletic. In fact, future studies should con- sider the possibility that the Greater 'Amakihi, like the 'Anianiau, warrants a genus of its own. THE "RED-AND-BLACK" GENERA Every study reviewed herein shows that the members of this subgroup, recognized from the time of Perkins (1903), do indeed form a well- defined clade. R. C. Fleischer (pers. comm.), on the basis of the small degree of genetic differ- ence between them, believes all of the "red" genera could justifiably be merged. On pheno- typic grounds, the genera Vestiaria and Drepan- is differ solely on a relatively minor red-to-yel- low color shift, hardly a generic-level distinction by modern standards, but my earlier suggestion (Pratt 1979) that they should be merged was not accepted by Berger (1981). Also, the 'Apapane and ',kohekohe (Palmeria dolei) are close structurally and behaviorally, although the lat- ter's unique plumage features make it look su- perficially rather different. The lumping of Him- atione and Palmeria is not as strongly supported by my phylogeny as the Vestiaria/Drepanis merger. HAWAI'I CREEPER So now we come to the one species whose taxonomic position is the subject of the widest disagreement among competing evolutionary hypotheses and hence the namesake of this pa- per. The Hawai'i Creeper is a small, drab Ha- waiian honeycreeper endemic to the island of Hawai'i (Scott et al. 1979). Its dull gray-green 90 STUDIES IN AVIAN BIOLOGY NO. 22 Majority rule 7.2_7 100 100 7..7 '100 60 100 Majority rule 10o 10o 100 100 IO0 IO0 100 100 Chaffinches Carduelme tinches Telespza finches Palila Kona Grosbeak koa finches Lana'i Hookbill Po'o-uli -- Maul Parrotbill 'akialoas -- 'akepas/'Akeke'e Hawai'i Creeper 'Akikiki 'Apepene -- 'riwi Hawai'i M arno Kakawahie/alauahios Chaffinches Cardueline finches Telespiza finches 1100L Palila Kona Grosbeak koa linches Lana"i Hookbill Maui Parrotbill 'Akiapola'au 'amakhis Greater 'Amakihi 'akepasfAkeke'e Hawai" Creeper Akikiki Anianiau Ula-'ai-hawane Apapane Akohekohe Back Mamo HawarIMamo Po'o-uli Kakawahie/alauahios Stnct b Strict Chaffinches Carduehne tinches Telespiza finches Palila Kona Grosbeak koa finches Po'o-uli Mau parrotbill Hawai'i Creeper 'Apapane Hawah M arno Chaffinches Cardueiine finches -- r- Loxioides Psittirostra Dy,morodrepania Peudonextor Hemigthus (Viridonia)  Hemigthus I I  Hemigmthus (Akialoa)  I Hemigthus (H.)lucidm I I ' Hemigmthm (H.)munroi I  0 .... ysris ....   Oreomvstia irdi - Hemignathu parvus  iri;ps  Hi--lone  Pabr DrepanB,erea Drenia p'ca Melampros()p. Poreomvza HAWAI'I CREEPER--Pratt 91 coloration and generally inconspicuous behavior may have contributed to the fact that the Ha- waiians did not distinguish it from the Hawai'i 'Amakihi (Hemignathus virens; Perkins 1903). It was first described (Wilson 1891) as Hima- tione mana, but Amadon (1950) included it in his large genus Loxops as one of the subspecies of the "Creeper," a "species" subsequently shown to be a grouping of five species in either 2 (Pratt 1979, 1992b) or 3 (Olson and James 1982b, James and Olson 1991) genera. The O'ahu, Moloka'i, and MauiFLana'i components of Amadon's "Creeper" are now placed in the enigmatic genus Paroreomyza, which now ap- pears to represent a very early divergence in the evolution of the honeycreepers (Tarr and Fleischer 1995, Fleischer et al. 1998, this study). The genus Oreomystis comprises the remaining two species, the 'Akikiki or Kaua'i Creeper, O. bairdi, and the Hawai'i Creeper, O. mana. John- son et al. (1989), Feldman (1997; phylogeny re- produced in Freed 1999) and Fleischer et al. (1998, this volume) present strong allozyme, mtDNA, and osteological evidence that O. bair- di is the sister-group of Paroreomyza, although the placement of that clade varies among the studies. For this relationship to hold, the hon- eycreepers' squared-off tongue base (Character 2) would have to have evolved twice indepen- dently, an unlikely prospect as discussed earlier. This study achieved very different results (Fig. 1 ) in which Paroreomyza and Oreomystis bairdi are as fax apart as any other two drepanidine genera. Tarr and Fleischer's (1995) restriction- site study supports this finding, but is out of step with their later mtDNA sequence analyses. On osteological grounds, Olson and James (1982) and James and Olson (1991) place the Hawai'i Creeper in their large Loxops and con- sidered it closely related to the 'amakihis (Olson and James 1995). However, James's (1998) phy- logeny (see Fleischer et al., this volume) shows it only as a sister group to most of the other thin- billed honeycreepers, a position rather close to where it appears in my study (except that the 'Akikiki is paired with it). Thus the osteological phylogeny and mine actually differ more strik- ingly on the placement of O. bairdi than on that of the Hawai'i Creeper. The osteological phy- logeny, if correct, would require either the cre- ation of a new monotypic genus for the creeper or the recognition of a huge genus Drepanis that would include everything f¾om the creeper to heterobills to mamos. If the red birds were re- moved from this assemblage as suggested above, the creeper could be in Hemignathus. In- terestingly, Feldman's (1994) independent mtDNA study showed the Hawai'i Creeper as sister group to the red honeycreepers which clade in turn formed an unresolved trichotomy with the 'amakihis and 'gkepas. Although dis- tinctive, this hypothesis is closer to those de- rived from osteology and this study than to the other mtDNA results. Fleischer et al. (1998, this volume) hypothesize on the basis of mtDNA se- quencing that the Hawai'i Creeper forms a clade with the 'gkepas which in turn is sister to an odd assemblage that includes the heterobills, parrot- bill, and 'Anianiau. So is this enigmatic little bird an odd offshoot of its own, sister to the 'Apapane (Himatione sanguinea) and 'I'iwi, a non-crossbilled 'gkepa, or an Oreomystis? The question of whether Oreomystis is related to Paroreomyza is independent of whether the Hawai'i Creeper and the 'Akikiki are congeners. So numerous are the phenotypic similarities of the Hawai'i Creeper to the 'Akikiki that manu- script reviewers of Pratt (1992b) questioned even considering them separate species, let alone members of different genera. The Hawai'i Creeper is vaguely similar in overall coloration to female and juvenile 'amakihis, female 'dcepa (Loxops coccineus), and both sexes of 'Akeke'e (L. caeruleirostris; Scott et al. 1979, Pratt et al. 1987), but differs in important details. Unlike 'amakihis and 'rakepas, adults are not sexually dichromatic. They have a broad gray mask, shaped more like the black mask of L. caerulei- rostris than the narrow black lores of 'amakihis. Unlike 'amakihis but resembling 'gkepas, nei- ther adults nor juveniles ever have wing-bars. And unlike both 'amakihis and 'gkepas, juve- niles have a distinctive plumage with pale feath- ering in the lores and over the eye. In plumage features, the Hawai'i Creeper closely resembles Oreomystis bairdi, which also lacks sexual di- chromatism as an adult, has a distinctive pale- faced juvenile plumage, and lacks wing-bars. The creeper's bill is nearly straight with a concave gonys (Pratt 1992b), pale except for a dusky tinge, variable in extent, along the cul- men. In overall shape it is somewhat interme- diate between that of an 'amakihi and that of an 'gkepa (without crossed tips) and resembles that FIGURE l. Phylogenetic trees of Hawaiian honeycreepers: a) unweighted tree, 50% majority-rule consensus; b) unweighted strict consensus tree; c) majority rule tree with Characters 1-3 weighted 2; d) strict consensus of weighted trees, with AOU (1998) scientific name equivalents and Hemignathus divided into four subgenera. See Tables 1 and 2 for data and coding. See text for analysis details. 92 STUDIES IN AVIAN BIOLOGY NO. 22 lO 5 Oreomystis bairdi 0. mana 5 -- o I I 0 2 4 Time (seconds) FIGURE 2. Juvenile begging calls of the 2 species of Oreomystis. O. bairdi recorded 6 August 1997 by David Kuhn near the Mohihi Trail above Koa'ie Stream, Alaka'i Wilderness Preserve, Kaua'i (not archived). O. maria recorded by the author 4 May 1977 at Keauhou Ranch, Ka'u District, Hawai'i (Cornell Laboratory of Orni- thology, Library of Natural Sounds No. 05274). Audiospectrograms prepared on a Macintosh computer using Canary¸ software program. of Oreomystis bairdi in nearly every detail ex- cept that it is somewhat thinner, light gray rather than pale pink, and has somewhat more dark pigment above (Pratt et al. 1987). Because their bills are nearly identical in shape, the most par- simonious hypothesis would seem to be that the two creepers share a common ancestry, but bill shape does not argue strongly against an 'akepa relationship for them both. The nuthatch-like foraging of the Hawai'i Creeper differs from that of Oreomystis bairdi only in that the chosen substrates average larger for the latter (Pratt 1992b). Of all the 'amakihi species, the Kauai 'Amakihi (H. kauaiensis) is the most frequent bark-picker, but it would never be characterized as nuthatch-like (Conant et al. 1998). Nor does the Hawai'i Creeper forage in any way resembling the feeding of either species of Loxops (Lepson and Pratt 1997, Lepson and Freed 1997). Following fledging, tightly struc- tured family groups of both Hawai'i Creeper (Scott et al. 1979, Pratt et al. 1987) and 'Akikiki (Pratt 1992b, Conant et al. 1998) forage together with frequent begging notes from the juveniles. Both may eventually join larger mixed-species flocks with 'amakihis, 'akepas, and other species (Pratt et al. 1976, Lepson and Freed 1997, pers. obs.). Similar tightly structured family foraging groups with distinctive calls have not been re- ported in 'amakihis or 'akepas (Lepson and Pratt 1997, Lepson and Freed 1997), although they both join looser flocks. Because the hypothe- sized ancestor of the drepanidines was a seed- eating cardueline finch, the nuthatch-like forag- ing of the two creepers can be viewed as a syn- apomorphy. The song of the Hawai'i Creeper is a short trill similar to that of O. bairdi, but many other drepanidine species also sing short trills, so adult songs reveal little about relationships (Scott et al. 1979, Pratt et al. 1987, Pratt 1992b, Pratt 1996). One noteworthy difference is that songs of both Oreomystis are highly stereotyped, whereas those of such potential relatives as HAWAI'I CREEPER--Pratt 93 'amakihis and 'akepas are highly variable even when uttered by the same individual (Pratt 1979,1996; Pratt et al. 1987). The begging notes of Hawai'i Creeper juveniles flocking with their parents after fledging are very similar to those of juvenile 'Akikiki (Fig. 2) in similar context, which were first recorded in 1997 and are thus not included in recently published tapes (Pratt 1996). The individual notes of 'Akikiki juve- niles are slightly shorter and cover a somewhat wider frequency range than those of the Hawai'i Creeper, but they have a similar syncopated rhythm, with notes grouped in short bursts (Fig. 2). Although a few other Hawaiian honeycreep- ers (e.g. Pseudonestor xanthophrys, Hemigna- thus munroi) have distinctive juvenile begging notes that persist long after fledging, none have the same sound or rhythmic pattern of the two creepers. No long-persisting juvenile begging notes have been reported among either 'amakih- is or 'akepas, nor among cardueline finches, and thus the juvenile calls appear to be another syn- apomorphy linking the two Oreomystis. But it is the tongues that present the most enigmatic observations. The Hawai'i Creeper's tongue is narrow and nontubular, with a notched, slightly frayed tip (Richards and Bock 1973) and resembles the tongue of O. bairdi in virtually every detail (Pratt 1992a). Such a tongue tip dif- fers strikingly from that of the hypothetical an- cestral Hawaiian honeycreeper (Raikow 1977), is found only in the Hawai'i Creeper and the 'Akikiki, and, unlike that of the parrotbill, is dif- ficult to envision as a derivative of the highly derived drepanidine tubular type. The most like- ly explanation for two taxa sharing in detail such a complex derived morphology is that they both inherited it l¾om a common ancestor. The sim- ple, notched tongue certainly appears to be a de- fining synapomorphy in Oreomystis. If Raikow (1977, 1985, 1986) is correct that the tubular drepanidine tongue defines a major clade of the Drepanidinae that includes both the "green" and "red" groups, Oreomystis cannot belong to it unless its distal tongue morphology is secondarily derived from the tubular form. Of course, such derivation is clearly possible. Both the DNA and osteology trees of Fleischer et al. (this volume) require this secondary derivation for the Hawai'i Creeper but not the 'Akikiki. My unweighted tree (Fig. la) shows the two-mem- ber Oreomystis as one branch of an unresolved trichotomy with the "red" clade on the one hand and the "green" clade on the other, but my weighted tree (Fig. l c) places it, like both of those of Fleischer et al. (this volume), in a po- sition that requires secondary derivation of the Oreomystis tongue from a tubular ancestor. This result prompted me to conduct an addi- tional analysis that focused on the "green" birds, including all species-level taxa and addi- tional characters (40-46 in Table 1) that, for rea- sons mentioned earlier, could not be used with the broader sample of taxa. I included the three Paroreomyza species and the monotypic Psitti- rostra for comparative purposes and so that the relationships of Pseudonestor would also be re- examined. All characters were unweighted in this analysis, and Character 35 (nest sanitation) was ordered rather than unordered as previously. Table 3 is the data matrix for this analysis. A heuristic search of the 46 characters produced 180 trees, from which majority-rule and strict consensus trees (Fig. 3; L = 109, CI = 0.661, RI = 0.732) were derived. This time, the two Oreomystis sorted out as the sister group to the entire clade defined by the tubular tongue (but including Pseudonestor), which I believe is a reasonable placement for it. Note that the earlier pairing of Oreomystis with 'Anianiau did not hold up in this more detailed analysis, and I re- gard it as an artifact. Problems of possible homoplasy complicate analysis of another anatomical feature that has figured prominently in the taxonomic history of the creepers. Raikow (1976) found that some Hawaiian honeycreepers, like many other pas- serines, have lost the plantaris, a minor muscle of the shank. Of the taxa he studied, only the 'amakihis and the "red" genera Himatione, Pal- meria, and Himatione lacked the plantaris. Un- fortunately, he included neither an 'akepa nor any of the "creepers" (which were all then con- sidered conspecific) other than Oreomystis bair- di. Nevertheless, Raikow (1977) separated "the Creeper" generically from the 'amakihis based on the loss of the plantaris in the latter. If the loss of the plantaris is a uniquely derived char- acter state within the honeycreeper taxon, then the logical conclusion is that the taxa that share this condition form a clade ('amakihis plus the red-and-black birds), a grouping that appears in Fleischer et al.'s (1998) mtDNA tree. Subse- quent dissections (S. L. Olson, pers. comm.) re- vealed that the Hawai'i Creeper lacks the plan- taris, a result that might also seem to support a relationship to 'amakihis. How useful is loss of the plantaris as a key to phylogeny? Clearly, it cannot be considered a synapomorphy in any broad sense, because it has occurred several times among passetines generally, and at least twice among the Carduelinae (Raikow 1976, 1977, 1978). Furthermore, avian muscles have been shown to be subject to evolutionary rever- sals (i.e., to become re-established in a lineage after loss; Raikow et al. 1979) as well as suffi- ciently variable individually to present problems for phylogenetic studies based on few specimens 94 STUDIES IN AVIAN BIOLOGY NO. 22 TABLE 3. DATA MATRIX FOR PAUP* ANALYSIS OF "HEMIGNATHINE" SPECIES OF HAWAIIAN HONEYCREEPERS USING CHARACTER STATES FROM TABLE ] Character state Taxon I 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 Psittirostra 1 1 1 1 1 0 0 I 0 ? 1 1 1 .9 0 0 1 1 1 I 0 Paroreomyza maculata 0 0 0 ? ? ? ? ? ? 0 0 0 1 2 2 1 1 1 2 0 0 Paroreomyza montana 0 0 0 ? ? ? ? ? ? 0 0 0 1 2 2 1 1 1 2 0 0 Paroreomyza fiammea 0 0 0 ? ? ? ? ? ? 0 0 0 1 2 2 1 I I 2 0 0 Pseudonestor 1 1 1 ? ? I ? ? ? '? 2 2 1 ? 1 2 1 0 2 2 0 Oreomystis maria Oreomystis bairdi Loxops cocci- Loxops caeru- leirostris Hemignathus sagittirostris 1 I ? ? ? ? ? ? ? ? 2 2 1 ? 3 3 I 2 2 0 0 Hemignathus virens Hemignathus tiara Hemignathus kauaiensis Hemignathus parvus Hemignathus ellisianus Hemignathus obscurus Hemignathus lucidus han- apepe Hemignathus lucidus luci- dus Hemignathus lucidus affin- is Hemignathus munroi I 1 1 ? ? ? ? '? I ? 2 2 1 I 2 1 1 1 2 0 0 1 1 1 0 1 ? 0 1 0 ? 2 2 1 ? 2 I 1 1 2 0 0 I 1 I ? ? ? ? ? ? 1 2 2 I 1 3 3 1 2 2 3 0 1 1 1 ? ? ? ? ? .9 ? 2 2 1 .9 3 3 I 2 2 3 0 1 I 1 0 I 0 I 1 1 ? 2 2 1 2 3 3 1 2 2 0 1 1 1 1 0 I 0 ? 1 1 ? 2 2 1 2 3 3 I 2 2 0 1 1 1 I 0 1 0 ? I 1 ? 2 2 1 2 3 3 1 2 2 0 1 l 1 1 ? ? ? ? ? ? ? 2 2 I ? 3 3 1 2 2 0 0 1 1 ? 0 1 0 1 1 0 1 2 2 I I 3 3 1 2 2 0 1 1 1 ? ? ? ? ? 1 0 1 2 2 1 1 3 3 1 2 2 0 I 1 1 ? ? ? ? ? ? ? ? 2 2 1 ? 3 3 I 2 2 4 1 I 1 ? ? ? ? ? ? ? ? 2 2 1 ? 3 3 1 2 2 4 I 1 1 ? ? ? ? ? ? ? ? 2 2 I ? 3 3 1 2 2 4 1 I 1 I 0 1 I 1 1 0 ? 2 2 1 ? 3 3 I 2 2 4 1 (Raikow et al. 1990). Further complicating mat- though the placement of that genus among the ters is the lack of information on the plantaris others remains controversial. The phenotypic ev- condition of 'akepas, the Greater 'Amakihi, and idence in this case, which includes certain and the 'Anianiau. Thus the hypothesis that the plan- probable synapomorphies of plumage sequence, taris has been lost more than once in drepanidine coloration, bill and tongue morphology, vocali- evolution is by no means far-fetched, and the zations, social behavior, and ecology are too nu- usefulness of this character in reconstructing merous and varied to be dismissed out of hand, phylogeny is severely compromised. Neverthe- as has been done in recent molecular studies, less I included it (Character 9) in my analyses none of which have even mentioned this striking as an unordered character. conflict of genetic and phenotypic data. Nor in The case for inclusion of the Hawai'i Creeper my opinion can so many similarities be credibly in Oreomystis based on "traditional" taxonomic attributed to convergence or homoplasy. data is straightforward, unequivocal, and sup- R.L. Fleischer (pets. comm.) has suggested that ported by every tree topology in this study, al- a past hybridization event could produce the re- HAWAI'I CREEPER Pratt 95 TABLE 3. EXTENDED. Character state 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 0 0 1 1 0 0 0 0 0 0 0 0 0 ? 1 ? 1 ? 1 0 1 0 0 1 0 0 1 1 1 0 0 0 0 0 3 1 0 2 1 3 ? ? ? I 1 0 3 1 3 2 0 1 1 1 0 0 0 0 0 3 1 0 2 1 3 ? 1 1 1 1 0 3 2 1 0 0 1 1 1 0 0 0 0 0 3 1 0 2 1 3 ? ? ? 0 1 0 1 0 0 0 0 0 1 1 2 0 0 0 0 4 2 2 2 0 4 1 0 0 1 1 2 3 2 1 2 0 2 1 2 0 0 0 0 3 4 2 1 0 1 4 1 0 0 1 1 0 2 1 0 0 0 2 1 2 0 0 0 0 3 4 2 1 2 1 4 ? ? ? 1 1 0 1 0 0 0 0 2 1 1 0 0 0 0 0 4 2 0 2 0 4 0 1 1 0 0 0 1 0 0 0 0 2 1 1 0 0 0 0 0 4 2 0 2 1 4 2 ? 0 1 0 0 3 1 0 3 0 2 I 0 2 0 0 0 0 4 2 ? ? ? 4 ? ? ? 1 3 3 2 2 0 1 0 3 1 1 1 0 0 0 0 4 2 0 2 0 3 0 1 1 1 3 2 3 2 2 1 0 3 1 1 1 0 0 0 0 4 2 0 2 ? 3 1 ? ? 1 3 2 3 2 3 4 0 3 1 1 1 0 0 0 0 4 2 0 2 1 3 1 0 0 1 2 3 3 2 2 1 0 2 1 1 0 0 0 0 0 4 2 0 2 I 3 1 1 0 1 1 1 4 0 0 0 0 3 1 1 1 0 0 0 0 4 2 ? ? ? 3 ? ? ? 1 2 4 3 2 1 1 0 3 1 1 1 0 0 0 0 4 2 ? ? ? 3 ? ? ? 1 2 4 2 2 2 1 0 3 1 1 1 0 0 0 0 4 2 ? ? ? 4 ? ? ? 1 4 4 3 2 1 0 0 3 1 1 1 0 0 0 0 4 2 ? ? ? 4 ? ? ? 1 3 4 3 2 1 0 0 3 1 1 1 0 0 0 0 4 2 ? ? ? 4 ? ? ? 1 4 4 3 2 1 2 0 1 1 1 1 0 0 0 0 4 2 2 2 1 4 1 ? 0 1 4 4 3 2 1 0 sults seen here, but considers convergence more has clearly played a role (Grant 1986, 1994), and likely. The name Oreomyza ( = Oreomystis) per- as Freeland and Boag (1999:584) pointed out, "it kinsi was based on a possible hybrid specimen of is extremely difficult with existing data to differ- which one parent was a Hawai'i Creeper (Amadon entiate between the effects of lineage sorting and 1950:176-177), so hybridization is neither unprec- hybridization." Recently, P. R. Grant (pers. comm. edented nor unreasonable. Furthermore, if Tarr and fide Thane Pratt) reported a pattem of hybridiza- Fleischer (1995) and Fleischer et al. (1998) are tion and subsequent backcrossing among the correct that the drepanidine radiation resulted from Geospizinae that, if it occurred among Hawaiian a recent rapid burst of speciation, then hybridiza- honeycreepers, could explain the apparent conflict tion need not indicate "next-of-kin" relationship, of phenotypic and genotypic data for the Hawaii especially because intergeneric hybrids are fairly Creeper. In such a scenario, hybrids would involve frequent in birds (Bledsoe 1988a). In the similarly primarily, or only, male Oreomystis mating with rapidly evolving Darwin's finches, hybridization female '5.kepas or 'amakihis. Given the song vari- 96 STUDIES IN AVIAN BIOLOGY NO. 22 Pffttirostra  Paroreomycamaculata 100 Paroreomyzamontana Par or eornyza flammea 88 Pseudønestør Hemignathus lucidus affinis Hemignathus lucidus hanapepe  I Hemignathus munroi I  Hemignathus lucidus lucidus '- 6 .em ignathm virens I 1[  Hemignathusfiava L4  Hemignathus kauaiensis 72[' tternignathux ellisianus - Hemignathus ob*curus Hemignathus sagittirostris Hemignathus lucidus hanapepe Hemignathus lucidus lucidus .  Hernignathus virens Hernignathus fiava Hernignathus kauaiensis -- Hemignathus ellBianus Hemgnathu sagittirostri, 100 -- t___ Loxop caeruleirostris Loxop5 caeruleirostris Hemignathus parvu,* Hemignathus parvu.* lOO [-- Oreom¾stis mana  Oreomstis bairdi  Oreomystis mana ß Oreom?tis bairdi FIGURE 3. Species-level phylogeny of "hemignathine" Hawaiian honeycreepers plus Paroreomyza and Psit- tirostra. Left, 50% majority-role consensus tree; right, strict consensus tree. Taxonomy follows AOU (1998). ation of the latter two groups and the relative uni- formity of Oreomystis songs, non-Oreornystis fe- males might be more likely to mate with a male Oreomystis than Oreomystis females to mate with a non-Oreomystis male. Offspring of such matings would then mate preferentially with Oreomystis or hybrids because males would sing the songs of their fathers and females would respond to songs of their fathers. If the birds with mixed ancestry became the ancestors of the Hawai'i Creeper, then they could retain all of the phenotypic synapo- morphies of Oreomystis but possess mtDNA, which represents solely the female line of descent, "stolen" from another species. Ongoing studies of nuclear DNA (R. Fleischer, pers. comm.) may help to solve this problem. Of course, the past hybrid- ization event might not have involved the Hawai'i Creeper at all; it could instead be the reason why the 'Akikiki tums up in the "wrong" place in some phylogenies. Indeed, the molecular and os- teological phylogenies reviewed here are more similar in their placement of the Hawai'i Creeper than the 'Akikiki. Removal of the Hawai'i Creeper from the genus Oreomystis at this stage would clearly be prema- ture, especially because we would have no un- equivocal alternative. At present, the DNA labo- ratories offer us three different hypotheses. This analysis of phenotypic characters shows very strong support for the current taxonomy, which is somewhat weakly corroborated by osteological studies and one mtDNA analysis. Furthermore, plausible hypotheses can be offered to explain the observed lack of genetic and phenotypic congru- ence. Until nuclear DNA studies are completed and possible hybridization is addressed, the pro- dent course is to avoid taxonomic changes based solely on molecular data. If future studies prove that the evolution of the Hawai'i Creeper was en- tirely independent of Oreomystis bairdi, then the large number and varied character of apparent syn- apomorphies of these two species will represent one of the most remarkable and noteworthy ex- amples of convergence ever demonstrated. That finding would be exciting, but the burden of proof clearly lies with those who would remove the Ha- wai'i Creeper t¾om Oreomystis. Why is the Ha- wai'i Creeper an Oreomystis? Because that is what the most consistent available evidence shows it to be. HAWAI'I CREEPER--Pratt 97 SUMMARY This study shows that the alpha taxonomy of the Hawaiian honeycreepers currently in use (AOU 1998) has a solid foundation in pheno- typic characters. None of the taxa, with the pos- sible exception of Hemignathus, are paraphylet- ic, and generic limits, with a few minor excep- tions, are reasonable. Hypothesized relationships at variance with current usage and based on ge- netic studies must be considered preliminary and tentative until consistent results are achieved. Taxonomic and sequence changes suggested by these results include: 1) the merger of Chlori- dops and Loxioides, or at least adjacent place- ment in the taxonomic order; 2) removal of the 'Anianiau from Hemignathus and classification as Magumma parva; 3) recognition of four sub- genera of Hemignathus (Hemignathus, Akialoa, Chlorodrepanis, and Viridonia); 4) the place- ment of Pseudonestor adjacent to Hemignathus in taxonomic sequence, or even merger of the two genera; 5) lumping of Vestiaria into Dre- panis and probably also Palmeria into Hima- tione; and 6) movement of Melamprosops and Paroreornyza to the beginning of the sequence, preceding Telespiza. ACKNOWLEDGMENTS I thank D. Kuhn for recording the 'Akikiki juveniles for this study. Students and staff of the Louisiana State University Museum of Natural Science helped in nu- merous ways: D. Lane and M. Cohn-Haft assisted in preparation of the audiospectrograms, F. Burbrink taught me how to use PAUP and MacClade, and J. V. Reinsen and F. H. Sheldon made numerous suggestions that improved the manuscript. Colleagues R. L. Fleischer, H. James, S. Olson, L. Freed, T Pratt, J. Lepson, B. Slikas, C. Mcintosh, and R. Raikow made valuable input by their challenging comments and sharing of unpublished information. J. M. Scott, S. Co- nant, and J. Rotenberry made helpful comments on earlier versions of the manuscript. Studies in Avian Biology No. 22:98-103, 2001. PHYLOGENETIC PLACEMENT OF THE PO'OULI, MELAMPROSOPS PHAEOSOMA, BASED ON MITOCHONDRIAL DNA SEQUENCE AND OSTEOLOGICAL CHARACTERS ROBERT C. FLEISCHER, CHERYL L. TARR, HELEN F. JAMES, BETH SLIKAS, AND CARL E. MCINTOSH Abstract. The Po'ouli (Melamprosops phaeosoma) is a small oscine songbird first discovered on Maui in the early 1970s and originally described as a member of the Drepanidini (Hawaiian honey- creepers). A recent study suggested that the Po'ouli may not be a drepanidine because it lacks most of a small set of drepanidine synapomorphies (e.g., specialized tongue morphology and distinctive odor). We conducted phylogenetic analyses of the Po'ouli and a number of drepanidine and potentially related songbird taxa. Our character sets included mitochondrial DNA sequences (obtained for Me- lamprosops via PCR of DNA isolated from museum specimens) and osteological characters. Analyses support the placement of the Po'ouli within the drepanidine clade, although the position of the Po'ouli within the clade is not strongly supported by either data set. Our results indicate that the Po'ouli is relatively distinct phylogenetically among drepanidines. If a goal of biodiversity conservation is to retain as much genetic diversity as possible then the Po'ouli should be considered a species of very high priority for conservation efforts. Key Words: ancient DNA; Drepanidini; Melamprosops phaeosoma; mitochondrial DNA; osteology; phylogeny; Po'ouli. In 1973 a new genus and species of Hawaiian bird was discovered by a group of student re- searchers in a small area of rainforest on the north slope of Haleakal Volcano on Maui. It was described from two collected specimens as the first new, living species of Hawaiian hon- eycreeper (Drepanidini) to be found in over 50 years (Casey and Jacobi 1974). Later, however, doubts arose concerning whether this small, brown, snail-eating bird is a drepanidine or some other type of songbird (Pratt 1992a). It was giv- en the scientific name Melamprosops phaeoso- ma, and the common Hawaiian name Po'ouli (which means "black-faced" in reference to its prominent black mask). The Po'ouli is now on the verge of extinction. Recent and intensive ef- forts to locate the species has resulted in detec- tion (and marking) of only three individuals (S. Reilly and M. Collins, pers. comm.; Reynolds et al. this volume). It is possible that this number represents the entire living population for the species. Although the Po'ouli differs in morphology, behavior and ecology from other living Hawai- ian birds (Pratt et al. 1997b), its phylogenetic uniqueness and closest relatives remain uncer- tain (Bock 1978, Pratt 1992a). According to Pratt (1992a), Melamprosops completely lacks the few synapomorphies that define the Drepan- idini, most notably the unique musty odor and specialized tongue characteristics. It also differs from all known drepanidines in plumage color and pattern, bill morphology, vocalizations, diet (i.e., specialization on snails), and other aspects of behavior (Pratt 1992a). Knowledge about the relationships and phylogenetic uniqueness of the Po'ouli will help in deciding how much effort should be expended to recover the species (Faith 1992, Krajewski 1994). Here we present cladis- tic analyses of mitochondrial DNA sequences and skeletal morphology that indicate that this troubling (and troubled) little bird is a Hawaiian honeycreeper, albeit an extremely distinctive one. METHODS SAMPLED TAXA We compared DNA and skeletal characters of Me- lamprosops to a sampling of taxa from within the Dre- panidini, Carduelini, Fringillini, Emberizinae, and oth- er outgroups. Common and scientific names of North American and Hawaiian taxa follow the AOU Check- list (1998). Common and scientific names of other taxa, and subfamily classifications, are from Monroe and Sibley (1993). Drepanidini analyzed for mtDNA sequence or os- teology (see Figs. I and 2) include Nihoa Finch, Te- lespiza ultima; Lay_san Finch, T. cantans; Palila, Lox- ioides bailleui; '0'0, Psittirostra psittacea; Lfina'i Hookbill, Dysmorodrepanis munroi; Maui Parrotbill, Pseudonestor xanthophrys; Kaua'i Creeper, Oreomys- tis bairdi; Hawai'i Creeper, O. mana; Maui 'Alauahio, Paroreomyza montana; 'Akeke'e, Loxops caeruleiros- tris; 'kepa, L. coccineus; 'Akiap61'au, Hemignathus munroi; Lesser 'Akialoa, H. obscurus; 'Anianiau, H. parvus; Kaua'i 'Amakihi, H. kauaiensis; O'ahu 'Ama- kihi, H. fiavus; Maui 'Amakihi, H. virens wilsoni; Ha- wai'i 'Amakihi, H. v. virens; 'I'iwi, Vestiaria cocci- nea; Hawai'i Mamo, Drepanis pacifica; 'Apapane, Himatione sanguinea; and ',kohekohe, Palmeria do- lei. Carduelini analyzed include the White-browed Ro- sefinch (Carpodacus thura, Genbank number 98 PLACEMENT OF PO'OULI Fleischer et al. 99 69 92 ! White-browed Rosefinch Carduelini Buff-barred Warbler FIGURE l. Phylogenetic tree constructed using a maximum parsimony criterion from mitochondrial DNA cytochrome b sequences. The phylogram is one of two maximum parsimony trees of (weighted) length 1255 and CI of 0.53. The numbers at particular nodes are the percentage of trees containing the node following a 500 repetition bootstrap. Nodes with percentages below 50% are not noted. These nodes are assumed to be unresolved and their branches collapse to a polytomy. See Methods for scientific names of taxa exhibited here. AF015765), House Finch (C. mexicanus; Fleischer et al. 1998), Common Rosefinch (C. erythrinus), Purple Finch (C. purpureus), Spot-winged Grosbeak (Mycer- obas melanozanthos), Evening Grosbeak (He6Ieri- phona ve67ertina), Desert Finch (Rhodopechys obso- leta), Golden-winged Grosbeak (Rhynchostruthus so- cotranus), European Greenfinch (Carduelis chloris), Pine Siskin (C. pinus) Red Crossbill (Loxia curviros- tra), Yellow-fronted Canary (Serinus mozambicus), Grey-headed Bullfinch (Pyrrhula erythraca), Pine Grosbeak (Pinicola enucleator), and Asian Rosy Finch (Leucosticte arctoa). The Common Chaffinch (Frin- gilla coelebs) is a fringilline outgroup. Emberizines include the Green Honeycreeper (Chlo- tophanes spiza; Fleischer et al. 1998), Scarlet-rumped Tanager (Ramphocelus passerinii; U15717), Summer Tanager (Piranga rubra; U15725), Prothonotary War- bier (Protonotaria citrea; this study), Brown-headed Cowbird (Molothrus ater; this study), Northern Car- dinal (Cardinalis cardinalis), Black-and-white Warbler (Mniotilta varia), Vesper Sparrow (Pooecetes grami- neus), White-lined Tanager (Tachyphonus rufus), Red- winged Blackbird (Agelaius phoeniceus), and Saffron Finch (Sicalisfiaveola). Outgroups are the House Spar- row (Passer domesticus) and the Buff-barred Warbler (Phylloscopus pulcher; Y10732). MITOGHONDRIAL DNA DNA was isolated from samples taken from the only two Melamprosops museum specimens that exist. The tip of one small secondary feather was removed from the B. P. Bishop Museum specimen (holotype: BBM- X147112; under the care of C. Kishinami and A. Al- lison), and a small piece of skin from the ventral open- 100 STUDIES IN AVIAN BIOLOGY NO. 22 e  Desert  :inch Golden-winged Grosbeat{ Carduelini Europ. ean Greenfinch 'ine Siskin [-- Red Crossbill [--Yellow-fronted Canary Common Rosefinch __ Grey-headed Bullfinch -- Pine Grosbeak ß -- Purple Finch i ................ -:....A,s.i .a,n,,..R.,o. sy .F..i..n .c..h. ............ ............................ L;n'fhficn ......... parrow FIGURE 2. Phylogenetic tree constructed using a maximum parsimony criterion from a matrix of osteological characters. A strict consensus of 128 optimal trees found by repeated random searches of these data (500 replicates, closest addition sequence with ten trees held at each step, initial tree improved upon with TBR branch swapping; optimal tree length 286 steps). See Methods for scientific names of species included in the tree. ing was taken from the American Museum of Natural History specimen (paratype: AMNH-810456; under the care of G. Barrowclough). Museum specimen DNA was isolated in a small laboratory dedicated to ancient DNA analyses using "ancient DNA" proce- dures (e.g., Cooper et al. 1996, Paxinos et al. 1997). Modem DNA analyses were conducted in a laboratory separated by >500 m from our ancient laboratory. Briefly, DNA was isolated by digesting skin or feather pulp overnight at 55 ø C in a DTT-SDS-EDTA buffer with proteinase K, followed by phenol and chloroform extractions and centrifugal dialysis to remove buffer and other solutes (as in Paxinos et al. 1997). We amplified and sequenced two regions of mtDNA from the museum and modern specimens (Fig. 1) using the polymerase chain reaction and specific primers: (1) 675 bp of the Cytochrome b (Cyt b) gene in two over- lapping pieces (see Fleischer et al. 1998); and (2) 224 bases of the 5' end of the mitochondrial control region (CR; Tarr 1995). Cyt b and CR sequences were also obtained for some non-drepanidine songbird species from Genbank (see Fig. 1). The Cyt b sequence was PLACEMENT OF PO'OULI Fleischer et al. 101 amplified only from the AMNH specimen, and ana- lyzed with 18 other drepanidine taxa as reported in Fleischer et al. (1998). The CR segment was amplified from the BPBM specimen only, and from an additional 6 drepanidine species. PCR controls were negative (i.e., no apparent product produced) for the study skin amplifications for both Cyt b and CR. Sequences were produced either manually as in Fleischer et al (1998) or on an ABI-373 automated DNA sequencer as in Greenberg et al. (1998), and were aligned with Se- quencher 3.0. Phylogenetic reconstructions and other analyses utilized PAUP*4.0d64 (D. Swofford, pers. comm.) and MacClade 3.01 (Maddison and Maddison 1992), and are described in the results section below. OSTEOLOGY A subset of data froin a separate study of cranial osteology and phylogeny in the drepanidines (James 1998) was used to determine if the Po'ouli is supported as part of the drepanidine clade. The original study involved 72 characters and 55 species of drepanidines, including 17 fossil species that became extinct follow- ing hutnan settlement of the archipelago less than two thousand years ago (James and Olson 1991). For the present study, the fossil taxa were excluded in order to specifically examine the phylogenetic placement of Melamprosops relative to extant or historically extinct drepanidines. Twenty-one other species of nine-pri- maried oscines were included so that other potential relationships might be revealed. Passer domesticus was included as an outgroup. The resulting matrix had 45 terminal taxa and 57 informative characters. The osteological matrix was analyzed using a par- simony criterion. All characters were mn as ordered characters except for seven multistate characters that were run as unordered because the states were not judged to be sequential. Ten characters had an essen- tially binary distribution of states except that a few taxa showed intermediate conditions. In these in- stances, the intermediate condition was scored as a third state, but the character was assigned a weight of 0.5 for the parsimony analyses, to prevent intermediate conditions froin exerting an undue influence on tree length. All other characters were unweighted. RESULTS MITOCHONDRIAL DNA Cladistic parsimony analyses of the Cyt b se- quences consistently place the Po'ouli within the Drepanidini (Fig. 1). We initially ran a heuristic search in PAUP* with replicated, random addi- tion and no character weighting, and obtained seven equally most parsimonious trees for dre- panidines and carduelines. A maximum likeli- hood (ML) estimate of the transition-to-trans- version ratio was then made using the tree with the lowest ML score (ts:tv -- 4.0:1). This ratio was used to weight transversional changes, and a heuristic search generated two maximum par- simony trees of length 1255 (unweighted for the same topology is 685 steps) and a consistency index of 0.53 (Fig. 1). Placement of Melampro- sops within the Drepanidini, however, occurs re- gardless of whether transversions are weighted 4.0:1, 10.0:1, or unweighted relative to transi- tions (although weighting and additional out- group taxa does affect the topology of drepani- dine relationships). Forcing the Po'ouli from the Drepanidini to the Carduelini (in MacClade; Maddison and Maddison 1992) increases the length of the tree (unweighted) in Figure 1 by 12 additional steps. This constrained tree is sig- nificantly longer than that of Figure I based on both parsimony (Kishino-Hawegawa test, t = 2.69, P = 0.0072; Kishino and Hasegawa 1989) and maximum likelihood (G = 52.71, P < 0.001; Felsenstein 1988) tests. Making Melam- prosops the sister to each emberizine clade also significantly increases tree length (by 20-27 ad- ditional steps; Kishino-Hasegawa test, t = 3.56, P < 0.001). Distance analyses further support a drepani- dine relationship for Melamprosops. Kimura 2- parameter and gamma-corrected distances were lower for comparisons of the Po'ouli and dre- panidines (0.086 -+ 0.002, range 0.062-0.102) than for comparisons of the Po'ouli and cardue- lines (0.147 + 0.005, range 0.142-0.152) or berizines (0.196 -+ 0.010, range 0.170-0.218). The CR sequence analyses also place the Po'ouli within the Drepanidini. First, three sin- gle-base deletions found in the Fringillini and Carduelini CR sequences do not occur in dre- panidines nor in Melamprosops CR sequence (Table 1). Second, 1000 replication bootstraps of maximum parsimony trees (with gaps and trans- versions weighted 10:1 or 5:1 over transitions; heuristic search) reveal 88% and 90% support, respectively, for monophyly of the drepanidines, including the Po'ouli. Last, forcing the Po'ouli from the Drepanidini into the Carduelini (i.e., sister to Carduelis chloris) or Emberizinae (i.e., as a sister to Melospiza georgiana) increases un- weighted tree length by 4 and 10 steps, respec- tively. The constrained trees are significantly longer (Kishino-Hasegawa test, t = 2.15, P = 0.032 when sister to Carduelis; t = 2.32, P = 0.021 when sister to Melospiza). OSTEOLOGY Parsimony analysis produced 128 equally most parsimonious trees from which we derive a strict consensus tree (Fig. 2). The Po'ouli is nested within the drepanidine clade in all of the 128 trees. Moving the Po'ouli outside the dre- panidine clade to a position as sister to either cardueline terminal taxa or cardueline resolved clades adds 9 to 23 additional steps to the total tree length. Making the Po'ouli a sister taxon to Fringilla or any emberizine outgroup adds 13 to 20.5 steps. 102 STUDIES IN AVIAN BIOLOGY NO. 22 TABLE 1. LISTED ARE 67 VARIABLE NUCLEOTIDE SITES (OF 224 TOTAL) FROM THE 5'-END OR LEFt DOMAIN OF THE MITOCHONDRIAL CONTROL REGION ASSESSED FOR ONE EMBERIZINAE (MELOSPIZA GEORGIANA; GREENBERG ET AL. 1998) AND TEN FRINGILLINAE, INCLUDING THREE FRINGILLINI (FRINGILL4; MARSHALL AND BAKER 1997), ONE CAR- DUELINI (CARDUELIS; MARSHALL AND BAKER 1997), AND SIX EXTANT MEMBERS OF THE DREPANIDINI (TARR 1995) Melospiza georgiana Fringilla teydea Fringilla montifringilla Fringilla coelebs Carduelis chloris Paroreomyza montana Loxioides balleui Telespiza cantans Hemignathus parvus Hemignathus kauaiensis Himatione sanguinea Melamprosops phaeosoma 1 2 3 4 5 6 1234567890123456789012345678901234567890123456789012345678901234567 TAGCCACGACACCTTATTATGAA-CCACTAGTGA-A-AACACTCCCGTAGGTATATTCAATAGATAG .... TGTA.-.T ..... A.C..TA..T .... A..-.-.G.TA..T...T .... -.GCTTC.TA.C.. .... T.TAG-..AC ......... -..T.CC.GA.-.-.G.TA ...... T .... -.GCTTC.TA.C.. C .... G...T .......... A..-A.T .... A..-.-.G.TA ...... T .... -.GCTTC.TAGC.A .CAAT.A .... GT ....... A.TAA.CT...GA.-G-.GA.A..T .... ACAT-GCCTGCCTAGC.. .CA...A...GATC ..... C.CTA.AC.AG.GAGG.TGG ....... ACT ....... C..C.T..C.. .CA.T.A .... G .... C..C.CCAA.T.AC.C..G.G.G ........ GT ...... C...C.T ..... .CA...A...GGT.C.C...ACCA.AC.A ..... G.G.G ..... NNNNN ...... C...C.TT.C.. .CA...A...GA ...... G.ATCAAAC.A..A..G.NNG ....... AGT .... G .... TC.T..C.. .CA.T.A...GA ........ ACCCAAC.A..A..A.A.G...C...CAC .... AG.C.TC.T..CC. .C..TTA...G ........ C.CTAAAC.ATCAC.G.NNG .... A.T..T...CG..C.TC.T..C.. .CA.T.A ........ C ..... TTC.AT.A..A..G.NNG...NN-NNNNN ...... GCTTC.T..C.. Note: Melamprosopx phaeosoma sequence is froin this study. A "." indicates identity of the nucleotide to the topmost base and an "N" indicates a base that could not be calle& A "-" indicates a gap or deletion in the sequence. Note the three insertions found in all drepanidines relative to l¾ingillines (at sites 35, 37, and 54). In addition, there are three drepanidine transversional synapomorphies (22, 26, and 29). See Fig. 1 for common names of drepanidine taxa. DISCUSSION In spite of Pratt's (1992a) assessment that the Po'ouli might not be a drepanidine, we find con- sistent evidence to the contrary. Pratt (1992a) notes that the Po'ouli should be considered a "nine-primaried oscine of uncertain affinities," and that it "does not look, smell, act, or sound like a Hawaiian honeycreepen" Our DNA evi- dence places Melamprosops within the drepani- dines, and osteological characteristics indeed make the Po'ouli "look" like a honeycreeper. How does one reconcile the apparent morpho- logical, ecological, and behavioral distinctive- ness of the Po'ouli (Pratt 1992a; Pratt et al. 1997b, this volume) with our results? Two ex- planations may account for this: (1) some of the phenotypic traits that Pratt emphasizes (i.e., those associated with foraging mode and feed- ing) may be affected by adaptive radiation and thus we might expect to see wide diversity in their character states; and (2) some of the 17 extinct drepanidine species known only from fossils may have shared these traits with the Po'ouli, thus making it different only in the con- text of living or historically extinct taxa. We do not know what factors effected the evolution of the brownish coloration and the black facial mask, nor why Melamprosops (and apparently Paroreomyza; Pratt 1992b) lack the distinctive drepanidine odor. While our results indicate that the Po'ouli is a Hawaiian honeycreeper, the relationships of the Po'ouli within the drepanidines are not well resolved by the mtDNA data (Fig. 1). Majority rule bootstrap analysis results in collapse of sup- porting branches such that Melamprosops be- comes a basal drepanidine lineage. On the strict consensus for the morphological trees, the Po'ouli joins at a node proximal to the finch-like species but distal to most other living drepani- dines. It is not depicted as the sister group of any living drepanidine species. Thus, in both mtDNA and osteological trees the Po'ouli ap- pears to represent a unique drepanidine lineage. Its lineage may have diverged from other dre- panidine lineages prior to evolution of the syn- apomorphic characters defined by Pratt (1992a). How phylogenetically distinct is the Po'ouli among living drepanidines? To answer this we estimated the contribution of each taxon to the PLACEMENT OF PO'OULI--Fleischer et al. 103 total minimum evolution score for the Cyt b tree in Figure 1. In PAUP*, we constrained the tree topology, pruned a taxon from the tree, then re- calculated the ME score. The process was re- peated for each drepanidine taxon; each ME score was subtracted from the total ME score to provide a phylogenetic "distinctiveness" score (U) for the taxon (essentially that of Faith 1992). The Po'ouli had the highest U (0.044) among the 19 drepanidines (mean and SE of U for the other 18 taxa was 0.015 +_ 0.002). To evaluate the Po'ouli's distinctiveness in the osteology- based tree we constrained the tree in Figure 2 in MacClade 3.01. A drepanidine taxon was re- moved and the length of the reduced tree was subtracted from the length of the total tree. The procedure was repeated for each of the 23 dre- panidines, and revealed that the Po'ouli was the fourth most distinctive taxon based on osteology (after Maul and Kauai creepers and the 'Akia- p61'au). Thus we consider the Po'ouli to be phylogenetically unique among the drepanidi- nes, and the taxon that individually contributes most to extant drepanidine phylogenetic diver- sity. The closest corrected genetic distance be- tween the Po'ouli and other drepanidines is 0.062. Applying a corrected internal rate cali- bration for Cyt b in honeycreepers of about 0.016 +_ 0.005/MY (from Fleischer et al. 1998) suggests that the Po'ouli split from its nearest living drepanidine relative about 3.8 +_ 0.9 MY ago (fairly early in the drepanidine radiation; Tarr and Fleischer 1995, Fleischer et al. 1998). Of course extinct fossil drepanidines (James and Olson 1991) not included here, such as Xestos- piza, may turn out to be more closely related genetically. Nonetheless, in comparison to other extant drepanidines, the Po'ouli has had a long, independent evolutionary history. This long pe- riod of independent evolution can perhaps ex- plain some of Melamprosops' unique phenotyp- ic characteristics. Such phylogenetic distinctive- ness also increases the Po'ouli's conservation value, in that the species represents a significant fraction of the genetic diversity of the drepani- dines (Faith 1992, Krajewski 1994). Along with its singular ecological, behavioral, and morpho- logical characteristics, the Po'ouli's unique evo- lutionary history convinces us that serious ef- forts should be undertaken to avoid its impend- ing extinction. ACKNOWLEDGMENTS We gratefully acknowledge C. Kishinami and A. Al- lison of the B. P. Bishop Museum, and G. Barrow- clough of the American Museum of Natural History, for allowing us to sample from the precious Po'ouli specimens under their care. The Smithsonian Institu- tion's Walcott and Wetmore Funds, the National Geo- graphic Society, and the Friends of the National Zoo provided funding for our Hawaiian honeycreeper re- search. We thank T Pratt and S. Reilly for providing information and discussion, and D. Pratt, C. van Riper and S. Conant for reading and evaluating the results in the manuscript. Status and Trends Studies in Avian Biology No. 22:106-107, 2001. STATUS AND TRENDS--INTRODUCTION J. MICHAEL SCOTT AND CHARLES VAN RIPER, III The first postcontact attempts to assess the status of Hawai'i's birds were the collection of birds by naturalists of Cook's third voyage of exploration (Medway 1981). Additional 18 th and 19 th century attempts to document the occur- rence of birds in Hawai'i were sporadic and in- completely reported, and are documented in de- tail elsewhere (Olson and James 1994a). In the last decade of the 19 th century and in the first years of the 20 th century, there was renewed in- terest in the birds of HawaiT Henry Palmer col- lected for Walter Rothschild, and S. B. Wilson obtained specimens that resulted in publication of his and Evans's monumental works on the avifauna of Hawai'i (Wilson and Evans 1890- 1899, Rothschild 1893-1900). The Nihoa Finch (Telea19iza ultima) was described in 1917 (Bryan 1917) and the Nihoa Millerbird (Acrocephalus familiaris) in 1924 (Wetmore 1924), but the Po'ouli (Melamprosops phaeosoma) would not be described until 1973 (Casey and Jacobi 1974). During this same period, Henshaw (1902a) and Perkins (1903) added much to our knowledge of the turn of the century status and distribution of birds in Hawai'i. It was not until George C. Munro's efforts to survey the avifau- na of the islands from 1935 to 1937 that anyone would attempt to systematically ascertain the 20 th century status of HawaiTs native avifauna (Munro 1944). The husband and wife team of Charles and Elizabeth Schwartz conducted an 18-mo survey of the game birds of the territory of Hawai'i (Schwartz and Schwartz 1949). The objectives of this survey were "to ascertain the game birds present on the Hawaiian Islands, their distribu- tion and abundance and factors upon which their welfare depends." Several surveys of the Lee- ward Islands followed (Bailey 1956, Amerson 1971, Amerson et al. 1974, Clapp et al. 1977, Woodward 1972). Richardson and Bowles (1964) conducted an exhaustive survey of Kaua'i, one that resulted in the last documented field observations of the 'Akialoa (Hemignathus ellisJanus). Their observation that all of the spe- cies known to have occurred on Kaua'i could still be found there resulted in the state of Ha- wai'i setting aside the Alaka'i Swamp as a re- serve. John Sincock and Gene Kridler, both of the U.S. Fish and Wildlife Service, set up a statis- tically defensible set of transects allowing an es- timate of the population size of Laysan Finch (Tele19iza cantans) and the Nihoa Millerbird in the Leeward Islands (Conant et al 1981, Conant and Morin this volume). These transects have since been monitored continuously and consti- tute the first estimate of the numbers of Hawai- ian birds that included variances. Sincock fbl- lowed his efforts in the Leeward Islands by es- tablishing a set of transects in the Alaka'i Swamp that were used to establish the popula- tion size of the endangered forest birds of Kaua'i (reported in Scott et al. 1986). Interagency ef- forts were initiated in the 1950s to monitor the numbers of waterbirds (Engilis and Pratt 1993) and to assess the number and distribution of the NSn8 (Hawaiian Goose, Branta sandvicensis; Black and Banko 1994). In 1976, nearly 100 years after its discovery (Wilson and Evans 1890-1899), the first ever attempt to estimate the population size of the Palila (Loxiodes bail- leui) was conducted (van Riper et al. 1978). That effort established that the Palila was more abun- dant than previously thought, thus documenting the value of statistically based surveys of the entire range of a species. Winston Banko pro- vided an exhaustive review of the literature on Hawaiian birds and documented all known re- cords (Banko 1979, 1980a,b,c,d; 1981a,b; 1984a,b; 1986) The second range-wide survey of the Palila occurred in 1980 (Scott et al. 1984) and used the variable circular count. This census method has been used in all subsequent attempts to estimate population size of the Palila (Jacobi et al. 1996). It was in part the success of the Palila surveys that prompted the Hawaiian Forest Bird Survey (HFBS) 1976-1981 (Scott et al. 1986), an effort to survey all the forest bird habitat in Hawai'i. The HFBS was initiated in the forests of Ka'fi on Hawai'i in 1976 and ended deep in the heart of the Alaka'i Swamp on Kaua'i in 1981. The objectives of this survey were to determine the numbers, distribution, habitat associations, and possible limiting factors of the endangered forest birds of the high islands of HawaiT The only islands not surveyed were O'ahu (Shallenberger and Vaughn 1978) and the privately owned Ni'ihau. Since completion of this HFBS, seg- ments of HFBS transects have been surveyed irregularly (Reynolds et al. this volume). The challenges of estimating the number of birds in Hawai'i were the motivation for an international symposium on estimating the number of terres- trial birds (Ralph and Scott 1981). 106 STATUS AND TRENDScott and van Riper 107 Authors in this section report on more recent efforts to assess the numbers of Hawai'i's avi- fauna. David Ainley and his coauthors use a combination of field observations and modeling to assess the status of the Hawaiian subspecies of Townsend's Shearwater (Puffinus auricularis newelli), hereafter referred to as Newell's Shear- water, whereas the late Miklos Udvardy and An- drew Engilis report on 50 years of data on the migratory Northern Pintail (Anas acuta). Mich- elle Reynolds and Thomas Snetsinger describe their eftbrts to monitor the status of the rarest birds in Hawai'i, reporting on thousands of per- son-days of field effbrt. Paul Baker describes the status and distribution of the rarest of Hawai'i's terrestrial birds, the Po'ouli (Melamprosops phaeosoma) and finds three birds remaining. The dilemma of what management actions are dictated by such a rare species has challenged the talents of scientists and managers alike. Studies in Avian Biology No. 22:108-123, 2001. THE STATUS AND POPULATION TRENDS OF THE NEWELL'S SHEARWATER ON KAUA'I: INSIGHTS FROM MODELING DAVID G. AINLEY, RICHARD PODOLSKY, LEAH DEFOREST, GREGORY SPENCER, AND NADAV NUR Abstract. We assessed the status of the endenic subspecies of Townsend's Shearwater, hereafter referred to as Newell's Shearwater (Puffinus auricularis newelli), on Kaua'i, Hawaiian Islands, where the only sizable population of this species renains. First, to index recent population trends, we ana- lyzed data gathered on the 1,000-2,000 fledglings attracted to lights and picked up annually by the "Save Our Shearwaters" (SOS) Program over a 17-year period, 1978 1994. Second, to calibrate and to provide a denographic context to these data, we quantified breeding productivity and ruortality in a nountain colony and nortality due to anthropogenic factors in the urban corridor that encircles the breeding areas during seven years: 1980 1985 (summary of previous study), 1993, and 1994. Finally, we entered rates of productivity and nortality into a Leslie ruodel to integrate these data, to evaluate the demographic importance of different sources of nortality, and to assess the utility of SOS in mitigating nortality frown anthropogenic factors. During 17 years of data collection, an average 1,432 fledglings that were attracted to lights were picked up by SOS each year; 90% were banded and released alive. Considering all of Kaua'i during the study period, more fledglings were picked up, if breeding effort and success were higher, and the full noon occurred in early October well before the mid-nonth peak of fledging. Overall, the annual totals of fledglings (1) gradually decreased on the southern shore, where the level of urbanization (and lighting) has grown to double that of the entire remainder of the island; (2) remained approxinately stable on the eastern shore (moderate urbanization); but (3) increased narkedly on the northern shore, where urbanization is low but grew dranatically during the study period. The relationships to urban- ization were corroborated by natural experiments when lighting was curtailed. Research in the breeding colony revealed (1) a high incidence of nonbreeding (46% of burrow occupants) even among expe- rienced adults, typical of nany petrel species; (2) predation (2.5% of individuals) on subadults and adults in the colonies by introduced house cats (Felts catus) and Barn Owls (Tyro alba); and (3) breeding success (0.66 chicks/pair) conparable to other shearwaters with stable populations. Research in the urban corridor revealed, conservatively, that (1) about 15% of an estimated 9,600 fledglings produced each year are picked up by SOS, (2) annual nortality of fledglings following light attraction during autmnn is about 10%, and (3) annual mortality to adults and subadults frown collisions with power lines during spring and smnner (without light attraction) is 0.6-2.1%/yr. Only 15 of the 23,000 fledglings (<0.1%) initially banded by SOS have been recovered in subsequent years, but recoveries show that first breeding occurs at about 6 yrs of age and that 1-yr-olds do not visit Kaua'i. A Leslie model, using paraneters deterruined for the Newell's Shearwater, suppleruented by those frown the very closely related Manx Shearwater (P. p. puffinus), indicated a balanced/stable population when extrinsic nortality of anthropogenic origin was excluded. Factoring in predation on adults and subadults in the colonies and mortality of fledglings and adults/subadults due to collisions with human- nade structures produced decadal declines of 30-60% in the population, with variation depending on the parameter values used. The nodel also showed that the SOS prograin is critical to reducing the rate of population decline. Predation frown introduced animals proved to be the nost important cause of decline, but collisions with structures by adults and nortality of fledglings following light attraction were also significant. Key Words: bird inpacts; cat predation; Hawai'i; Kaua'i; light attraction; Newell's Shearwater; oce- anic island; population nodel; Puffinus auricularis; transnission line; urbanization. Many populations of tropical seabirds that nest on oceanic islands with large human populations have been decimated by introductions of mammalian predators, habitat destruction, and urbanization, though the details are known only generally. Sev- eral large tropical petrels are now endangered or recently extinct, for example, the Bermuda Petrel (Pterodroma cahow), Jamaican Petrel (Pt. hasita- ta), Madeiran Petrel (Pt. mollis madeira), Fiji Pe- trel (Pt. macgillivrayi), and Magenta Petrel (Pt. magentae, of New Zealand; Croxall et al. 1984, Warham 1990, Ehrlich et al. 1992, Nettleship et al. 1994). Included in this group are those petrels nesting among the main Hawaiian Islands, the en- demic subspecies of Townsend's Shearwater (Puf- finus auricularis newelli), hereafter referred to as Newell's Shearwater, and Dark-romped Petrel (Pt. phaeopygia sandwichensis), both of which are list- ed by the U.S. Endangered Species Act (USFWS 1982a). Newell's Shearwater and Dark-romped Petrel have been extirpated from most of their for- mer nesting islands, but on Kaua'i they are still relatively abundant (Telfer et al. 1987, Harrison 1990). 108 NEWELL'S SHEARWATER DEMOGRAPHYAinley et al. 109 The Newell's Shearwater, or 'A'o, was con- sidered extinct as of 1908, but on Kaua'i in 1947 it was rediscovered and, in 1967, confirmed to be breeding (King and Gould 1967, Sincock and Swedberg 1969). A small breeding population has been confirmed recently on the island of Ha- wai'i (Reynolds et al. 1997a, Reynolds and Ritchotte 1997), and the species may also nest in very low numbers on Moloka'i and O'ahu (Harrison 1990). Rediscovery of Newell's Shearwater coincided with rapid growth in ur- ban development on Kaua'i, when hundreds of fledglings were found, having been attracted to and, typical of all petrels (Reed et al. 1985), ap- parently blinded by man-made lighting as the birds made their way from nest to ocean on their nocturnal fledgling flight (King and Gould 1967). This annual "fallout" became a major source of mortality, because fledglings die after being run over by cars or colliding with lights, utility poles and wires, and buildings (Byrd et al. 1984, Telfer et al. 1987). Shielding lights re- duced attraction by as much as 40% in experi- mental areas (Reed et al. 1985); for example, a reduction in the intensity of yard lights at the Hanalei Plantation Hotel in 1965 reduced the fallout there significantly (King and Gould 1967). New building codes established in the late 1980s request measures to shield lights (State of Hawaii 1987); however, compliance has been inconsistent (D. Ainley and R. Podol- sky, pers. obs.). Attempting to decrease the mortality associ- ated with fallout, the U.S. Fish and Wildlife Ser- vice (USFWS) and the State of Hawaii, Depart- ment of Land and Natural Resources (DLNR), organized the "Save Our Shearwaters" (SOS) Program in 1978 (Telfer et al. 1987, Rauzon 1991). Residents who found fallen shearwaters were encouraged, by advertisements in the news media, to place them in bird boxes at "Shear- water Aid Stations." The captured birds were then picked up each morning and taken for re- lease from a coastal cliff. In the 17 years through 1994, about 23,000 shearwaters have been re- trieved, banded, and released (T Teller, unpubl. data). The current relatively high abundance and easy access of the Newell's Shearwater on Kaua'i provided the opportunity to understand the species' ecology in the context of interac- tions with human activity, and to test the utility of SOS before the species' status becomes des- perate and conservation attempts costly. What is learned may help to protect this and similar sea- birds as development and tourism spread to more and more tropical islands (e.g., Croxall et al. 1984, Croxall 1991). We report here our find- ings during a study that included both fieldwork and analysis of existing unpublished data gath- ered by SOS and by government researchers since the late 1970s. The assembled information provided inputs into a demographic model of population growth under various scenarios of mortality. The model was used to evaluate the impact of three important factors indicated in the field studies: predation of adults from introduced animals, mortality of fledglings after fallout, and mortality of adults from collisions with power lines. In addition, we use the population dynam- ic model to project long-term stability of the Newell's Shearwater population on Kaua'i. METHODS FIELDWORK We conducted fieldwork in a mountain colony above Kalheo (Fig.l), where the species breeding biology was studied in the early 1980s as part of an effort to determine whether Newell's Shearwaters could be cross-fostered by the much more abundant Wedge- tailed Shearwater (P. pacificus; Byrd et al. 1984). Ac- cess to the colony was difficult but nevertheless was easier by far than to any other colony known for New- ell's Shearwater. Elevation of the Kalheo colony is about 600 m. We searched for burrows among the veg- etation on the >65 degree slopes between May and November 1993. Burrows were marked and a line of small sticks was erected across entrances to indicate burrow use when brushed aside by entering or depart- ing birds; we also noted the presence or absence of excrement and feathers. We used a miniature infrared TV camera (Furhman Diversified, Inc.) on a stiff co- axial cable "snaked" down each burrow to determine the presence of eggs or chicks. Once an egg or chick was found, we rechecked the nest's status monthly. We attempted to set up a second study colony at a site called Kaluahonu, on the southern part of the island, but found that few birds still nested there compared with the early 1980s. In 1994, due to a shortfall in funding and a request from the committee overseeing the project (see Acknowledgments), we diminished work in the Kalheo colony and allocated our efforts elsewhere. Therefore, we checked contents of burrows found the previous year on four occasions between late August and mid-November. We compared our findings on breeding productivity with the results of Teller (1986), who participated in the cross-fostering studies from 1981 to 1985. Results over the seven seasons were combined in the demographic model described below. To guard against intrusions of feral cats (Fells catus) and rats into our study colony, we placed a network of live-capture traps at the entrance to our ridge-top trails. Traps were baited every three days. In addition, we carried no food of our own into the colony for fear of attracting mammals. To assess survival from the proportion of colony occupants that may have been banded by SOS (when birds were fledglings), we captured adult shearwaters by blocking the burrow entrance just before dark and waiting nearby. Upon arrival, the birds sat by the en- trance and could be picked up easily. We checked 110 STUDIES IN AVIAN BIOLOGY NO. 22 NORTHSHORE PRINCEVILLE THE HAWAllAH these birds for bands and banded them if none was present. We quantified mortality due to collisions with power lines during summer and to fallout during au- tumn; results are reported elsewhere (Ainley et el. 1995, Podolsky et al. 1998). ANALYSIS OF SOS DATA To assess temporal and regional trends in the num- ber of birds retrieved, we analyzed data contained in annual reports of the SOS program from 1980 to 1993 (project W-18-R, Hawaii DLNR), as well as raw data computerized by SOS from 1987 to 1993. We did not use data from 1978 or 1979 in most analyses because effort by the citizenry was reduced in the first two years of the program relative to subsequent years (cit- izens learned of the program each fall through adver- tisements in newspapers and radio). A ledger on which persons could record the place where each bird was found was provided by SOS at each shearwater station. In 10-15% of cases the spe- cific pickup locality was not recorded, and in some of these (e.g., when a citizen was commuting to/from work) it was likely that birds were turned in at stations some distance from the pickup locality. Beginning in 1982, to determine geographic variation in the relative strength of fallout, SOS divided Kaua'i into ten dis- tricts (Fig. 1) and apportioned the birds of unknown locality to the various districts according to the SOS station at which these birds were turned in (Telfer et al. 1987). We combined the districts into broader regions, a procedure that further diluted the effect of any incorrect apportionment. The regions, and the dis- tricts/drop-off stations comprising them, were (see Fig. 1): (1) Northshore--Hanalei-Princeville; (2) East- shore--K¾1auea-Anahola, Kapa'a, LYhu'e, Westin La- goons (Kaua'i Sur0 Hotel; and (3) Southshore--K- loa-Po'ipfi, KalSheo, Hanapfipfi-Waimeae, MSnS-Kek- aha (including Barking Sands Naval Air Station). NS- pali-K6ke'e is included in the Northshore, but being mostly wild land, it contributed little to SOS data. In analyses where year was an important consider- ation, we did not use data from autumn 1992 or from 1993, because Kaua'i was much different in ways crit- ical to our study. Hurricane Iniki devastated human structures on the island in September 1992, just before the shearwaters had begun to fledge and SOS would have swung into action. The hurricane obliterated all bright lights; all hotels were closed due to damage and fewer than 10% of street lights or power lines were left standing. Life on Kaua'i did not return to normal until summer 1994. Hurricane Iwa, in 1982, did not pass over Kaua'i until November, after shearwater fall- out had been completed, so the fallout data were not affected and were included in analyses. To maintain robust sample sizes in the data, we made some reasonable assumptions to categorize cer- tain data rather than discarding them from analysis. First, dead adults were distinguished from dead fledg- lings in the 1987-1993 SOS data, but this was not the case in the 1980-1986 data. With no organized search effort in 1987-1990, the average number of dead adults/subadults found was 17/yr (see Results). So, to estimate the number of dead fledglings reported by SOS each autumn, 1980-1986, we assumed that the age ratio and search effort were the same as in 1987- 1990 and, therefore, subtracted 17 "adults" from the total number of dead shearwaters reported in each of those years. Second, dead adults reported during spring and sum- mer were logged by SOS beginning in 1987. It was not until 1991-1992, however, that a concerted search effort for adults/subadults was made, in effect, equal NEWELL'S SHEARWATER DEMOGRAPHY--Ainley et al. 1 11 to the effort for fledglings in autumn. In 1991, some especially interested and knowledgeable citizens (C. Berg, C. Orr, and K. Viernes) undertook this task and it was continued by us in 1993 and 1994. We assumed that patterns revealed in 1991-1994 were similar to those in the older SOS data. Next, we assumed that "adults" reported as dead in the SOS data after 15 September of each year included many individuals in- correctly aged for two reasons. First, in SOS records, peaks in number of dead birds recorded as "adults" (many of which are flattened and thus hard to assess) corresponded exactly to peaks of fledglings (Ainley et al. 1995). Second, few adults visit the colonies and no banded adults/subadults have been found after this date (see Results). Adult shearwaters desert their young a week or two before the fledgling departs (see Warham 1990); therefore, we considered all birds found after 15 September (the beginning of the fledging and fall- out period) to be fledglings. To be sure, a few adults are found after that date (T Teller, pers. comm.). Finally, it was not until 1982 that stainless steel bands were used by SOS on all fledglings. Prior to then, most were banded with monel bands. Therefore, our analyses based on return rates of banded birds do not include the data for the 1978-1981 cohorts, assum- ing that the monel bands were lost rapidly as a result of immersion in sea water (Boekelheide and Ainley 1989). To assess trends in SOS totals in the context of ur- banization, we indexed the urbanization of Kaua'i in two ways. Ultimately, we were interested in the num- ber and dispersion of shearwaters, the number and dis- persion of lights to attract them, and the number of people available to report birds or carcasses to SOS. Not having direct data on urbanization (e.g., the rate at which building permits were issued), we chose two surrogates. First, we used growth in numbers of year- round human residents (data from the U.S. Census Bu- reau, 1930-1990), and compared these among the three regions to which the SOS data had been parti- tioned (Fig. 1). From this population are the persons who participate in SOS, with participation depending only on the acts of encountering a shearwater, picking it up, and delivering it to an SOS station. In the small, close community of residents (currently 48,000 per- sons), more and more persons would know about SOS as the years passed and the proportion of interested persons would not decrease. Efforts to advertise SOS remained constant throughout the period. Next, to in- dex trends in growth of the infrastructure developed for the tourist industry (i.e., coastal hotels, condomin- iums, lighted tennis courts and driving ranges, etc.), which would not necessarily track the requirements of permanent residents, we obtained data from the state of Hawai'i on the number of passengers using the Lu'e Airport each year from 1960 to 1993. This in- frastructure (and attendant lights) would be the source of fallout. Tourists would not know about SOS. The following assumptions were used to relate trends in the SOS data to urbanization. First assump- tion: the number of fledglings retrieved by SOS in any year is proportional to breeding population size and reproductive success. Reproductive output, or at least SOS totals, appears to have exhibited no continuous trend through time, except for the occasional outlying year (see Results). Second assumption: the number of fledglings reported to SOS is strongly affected by the number and distribution of lights to attract them. This effect of lights, proposed also by Teller et al. (1987), was verified experimentally when lighting was severe- ly reduced at the Hanalei Plantation Hotel in the 1960s (King and Gould 1967), at the Kaua'i Surf/Westin La- goons Hotel after 1983, and throughout Kaua'i as a result of Hurricane Iniki during 1992-1993 (see Re- sults). Third assumption: the number of citizens pres- ent on Kaua'i also directly affects the number of birds reported. The latter two factors (i.e., number and dis- tribution of lights plus number of persons available to encounter birds) would determine the proportion of fledglings produced that were attracted to lights, went aground, and were picked up. Final assumption: be- cause the shearwater population incurs a cost through mortality from fallout (i.e., some birds die regardless of SOS), the cost, if high enough, can lead to popu- lation decline (i.e., too many fledglings die due to ef- fects of urbanization). It is possible that the proportion of fledglings attracted and picked up could become saturated (i.e., an asymptote is reached whereby ad- ditional lights and people do not lead to more birds retrieved). This would argue also, however, for fallout cost to reach a maximum early in the growth of urban development (an important consideration; see below). MODELING We developed a population-dynamic model for the Newell's Shearwater, using assumptions similar to those used by others in analogous contexts (e.g., Si- mons 1984, Beissinger 1995, Shannon and Crawford 1999), to project population trajectory with and with- out mortality due to anthropogenic factors and to quan- tify the relative impact of those threats. We used a Leslie model (Leslie 1945), which combines age-spe- cific fecundity and survival to estimate population growth rates. Due to lack of information about year- to-year variation in demographic parameters, which is the case for the vast majority of demographic studies of wild, long-lived vertebrates, we assumed average (constant) values for the parameters. Owing to lack of data regarding age-related variation in demographic parameters among shearwaters and other procellari- iforms (e.g., Bradley et al. 1989, Wooller et al. 1989), and consistent with the efforts of other researchers, we also made the simplifying assumption of age-constant survival and reproductive success for individuals that have reached adulthood. Our approach, first, was to determine the combina- tion of parameter values that produced a stable popu- lation. Against this, current population parameters could be compared to show that, in the absence of recent anthropogenic activity, Newell's Shearwaters can maintain their numbers. Second, we used conser- vative, best estimates for each parameter and compared population projections that did and did not include var- ious factors affecting population growth. The factors considered were: (1) mortality of fledglings attracted to lights and subsequently grounded during autumn (fallout), (2) mortality of adults and subadults that col- lide with utility structures during spring and summer, (3) predation of adults and subadults in the breeding 112 STUDIES IN AVIAN BIOLOGY NO. 22 F- ESTIMA]EO / MEASURED 130 .110  1 oo 20 I 90  o lO  4o  2o  o lo o 'o B u MEAN 1.7% D m 1 91-94MEAN61/YR .! I  ] 87-90 MEAN 17/Y"  o 8%90 MEAN 20 < 1980 1988 1990 1980 1985 1990 FIGURE 2. Summary of SOS data for Newell's Shearwaters on Kaua'i, Hawaiian Islands: (A) total fledglings retrieved annually, 1980-1994; (B) percentage of fledglings that died in captivity during those years; (C) the number of fledglings, and (D) number of adults, respectively, as reported dead on the road (not retrieved). The number of dead fledglings was estimated for years prior to 1987 (see Methods). colonies, and (4) reduction of mortality to fledglings as a result of the SOS program in autumn. For all analyses, we used the computer package STATA (Computer Resource Center 1993). Averages are reported with + 1 SE. RESULTS BREEDING EFFORT AND SUCCESS Telfer (1986) monitored 36-47 burrows in the Kalaheo colony during 1981-1985, and we monitored 58-65 burrows, including many in Telfer's sample, in 1993-1994. Among the bur- rows checked in 1981-1985, the proportion in which breeding adults occurred (i.e., eggs or chicks found) averaged 46.5% + 6.4% (range 30% to 62%). In 1993, the proportion was 26%, although this is a minimum as some eggs prob- ably were lost before we finished our search for burrows (which took two months). In 1994, our effort was insufficient to derive an estimate of reproductive effort (see Methods). In 1993, 58 burrows were visited by shearwaters (88%); thus, a high level of nonbreeding (no eggs laid) was apparent. Not determined in 1981-1985 was the proportion of burrows that actually were ac- tive (i.e., used regularly regardless of whether an egg was laid). Among nests in which eggs were laid, an av- erage 66.0% _+ 6.4% (range 49-75%) succeeded each year, from 1981 to 1985; in 1993 only 27% succeeded and in 1994 81% succeeded. Like Telfer in 1981-1985, we could not ascertain the cause of mortality of most chicks. Only three of the 1994 chicks were from burrows in which eggs were laid in 1993; conversely, among the sites that produced chicks in 1994, 11 were ac- tive but none of these produced eggs or chicks in 1993. In total, the Newell's Shearwater pro- duced 0.66 chicks/breeding pair/yr during the 1981-1985 period. On average, 1,432 fledglings were reported to SOS each year, ranging from 950 (1992) to 2,200 (1987; Fig. 2A). Some of the variation was explained by differences in the timing of moon phases from one year to the next. It ap- pears that when the full moon occurs in mid- October, the peak of fledging (Telfer et al. 1987, Ainley et al. 1997b), as in 1981, the total num- ber of individuals found during all of the fledg- ling period (mid-September to early November) is much lower than if the full moon occurs at the periphery of peak fledging, i.e., in early or late October (Fig. 3). Breeding effort and suc- cess probably also affect the number of fledg- lings picked up; for instance, 1987 was a year when ocean productivity in the shearwaters' feeding grounds was unusually high (see Dis- cussion) and the number of fledglings picked up was higher than expected. Nineteen eighty-seven NEWELL'S SHEARWATER DEMOGRAPHY--Ainley et al. 113 2OO0 $? ¸ lOO0 FIGURE 3. DATE OF FULL MOON IN OCTOBER The total number of fledgling Newell's Shearwaters on Kaua'i, Hawaiian islands, retrieved by SOS each autumn (Sept-Nov), 1980-1994, as a func- tion of how closely the full moon coincided with the peak of fledging (mid-October). The point for 1987 was not used to generate the regression line (r 2 = 0.572, F.  = 7.36, P = 0.009; see text). Number by black and white circles indicates yean was also a year when the full moon did not oc- cur during the middle of fiedging and, thus, the two factors (high ocean productivity, timing of full moon) combined to produce high fallout numbers. Finally, curtailment of lighting, as Hurricane Iniki accomplished in 1992-1993 (and even into 1994 somewhat), brought fewer fledglings to ground (Fig. 2A). The same pattern can be seen locally at the Kaua'i Surf/Westin Lagoons Hotel when lighting was adjusted dur- ing renovations in 1983 (Fig. 4B) as subsequent fallout was much lower. MORTALITY Breeding colony We found one fresh adult carcass and six skel- etons of adults or subadults in the colony during 1993. In 1994, we found 23 dead shearwaters. All were skeletons of adults that had been killed in the early spring during courtship two months before our first visit, and each had marks on the sternum to suggest eating by a cat. Almost all dead birds were found in the lower two-thirds of the study area indicating that the cat entered the colony from the sugarcane fields below the colony rather than using our access above the colony. Telfer (1986) found cat predation to be significant especially during the second year of his five year study. We caught one cat and eight rats during 1993, but caught neither rats nor cats in 1994. Each year, we found rat droppings deposited through- out the colony before our arrival. During our work at night in 1993, we often saw or heard introduced Barn Owls (Tyro alba). Barn Owls prey on Newell's Shearwaters (Byrd and Telfer 1980), and it was clear that they homed in on the Newell's Shearwater vocalizations that we occasionally played from a tape recorder (Ainley et al. 1995). During the day or evening only, we infrequently saw Short-eared Owls (Asio fiam- meus), but whether they prey on shearwaters is not known. Urban corridor The number of fledglings that died each year during SOS processing averaged 1.7% of the to- tal turned in (Fig. 2B). The number of fledglings logged by SOS as dead on the road, but not de- posited at SOS stations, averaged an additional 6% of the total each year (Fig. 2C). Almost all of these birds were checked for bands. During 1991-1994, when a concerted search for dead adults was conducted, 42-72 were found in spring and summer each year (mean = 61 _+ 7/yr; Fig. 2D; see also Ainley et ah 1995, Podolsky et al. 1998). Before the directed search, an average 17 _+ 2 dead adults were re- ported per year by SOS (1987-1990). In 1993- 1994, among 30 adults that could be sexed (not overly smashed), the male:female ratio was 8:9, and 7 (23%) were breeders. The average mass of dead adults was 381 _+ 8 g (N = 35), a value important to our estimate of adult survival (see below). RATES OF BAND RECOVERIES Recoveries and band-return rates of fledgling and adult shearwaters were unexpectedly low. None of 15 fledglings banded in 1993-1994 and none of 52 banded in the study colony during 1980-1985 were picked up subsequently by SOS. In 1993, we captured nine adults in the col- ony, but none had been banded previously. Only 1 of 30 adults/subadults found dead in 1993- 1994 was banded. That one individual had been banded as a fledgling by SOS during fallout on the Southshore. Thus, we found 1 (2.6%) banded birds among 39 adults/subadults examined in the colony in 1993-1994. Similarly low band re- turns are evident in a sample of 14 adults banded in 1983 (T Telfer, unpubl. data). These birds were attracted one night to a camp light in the K6ke'e forest (Fig. 1). One of these birds (7.1%) was subsequently recovered upon hitting a pow- er line. An equally low recovery rate is evident among adults found dead along power lines and roadways. Thus far, only 15 of the 23,000 fledg- lings banded and released by SOS have been recovered as adults or subadults during subse- quent years (Table 1). Excluding data from the 114 STUDIES IN AVIAN BIOLOGY NO. 22 EASTSHORE WITHOUT WESTIN 769 - 54 400 - A 300 - 200- NORTHSHORE 217 ß 31 C -200 SOUTHSHORE 357 & 34 400 E 300 200 0 -200 YEAR WESTIN ONLY 240 ñ 53 400 - B 300 - -200 - * EASTSHORE I009 & 60 400- D 300 ' 200 ' -200 ' YEAR FIGURE 4. Deviations from the mean number of Newell's Shearwater fledglings on Kaua'i, Hawaiian Islands, retrieved by SOS each year, 1980-1994, on the: (A) Eastshore with data from the Kaua'i SurfF½/estin Hotel (district) removed [Y = -2201 + 26.2X; r 2 0.365, se 10.9, P = 0.038]; (B) Kaua'i Surf F½/estin Lagoons Hotel only [Y = 2723-31.6X; r 2 - 0.381, se 12.7, P = 0.032]; and totals for (C) Northshore [Y = -2062 + 24.1X; r 2 - 0.543, se 6.9, P = 0.006], (D) Eastshore [P 0.7], and (E) Southshore [P = 0.8]. first few years, when weak monel bands were used, the recovery rate was only 0.1% (12 of 12,443 birds banded in 1982-1990 and recov- ered in 1989-1994). Looked at in another way, among 351 adults reported to SOS, from 1987 to 1994 (when search effort was quantified), 12 (3.4%) had been banded. Three-fourths of the recoveries occurred during the past four years, when search effort was much greater than it had been (Ainley et al. 1995). No birds <2 yrs of age have been recovered. POPULATION TRENDS For all of Kaua'i, numbers of fledglings picked up each year were about the same during the period 1980-1990 (Fig. 2A). Thereafter, even after 1987 numbers declined each year (in- cluding years beyond those of this study, through 1997; SOS unpubl. data, T Telfer, pers. comm.). Results separated by region of retrieval showed a steeply growing number of fledglings for the Northshore (Fig. 4C, D, and E). No slop- ing trend was evident for the Eastshore (Fig. 4D), unless data for the Kaua'i Surf/Vqestin La- goons Hotel were analyzed separately (Fig. 4B). Then, positive growth in the number of fledg- lings was evident (Fig. 4A). Similarly, no slop- ing trend was evident for the Southshore overall, although a decline not evident in the other regions is apparent after 1988 (Fig. 4E). The growing number of SOS-processed birds on the East- and especially the Northshore oc- curred in concert with the doubling and quadru- pling, respectively, of urban development (size NEWELL'S SHEARWATER DEMOGRAPHYAinley et al. 115 TABLE 1. THE TiME OF YEAR THAT BANDED NEWELL'S SHEARWATERS OF KNOWN AGE WERE RECOVERED ON KAUA'I FROM 1980 TO 1994 a Age (Yr) May June July August <18 18 25 1 8 15 22 29 6 13 20 27 3 10 17 24 31 2-3 I I I I 2 I 1 4-5 1 2 1 1 ->6 1 1 a Dates represent the first day of one-week periods. of the human population) since 1970 (Fig. 5A). On the Southshore, where the human population has always exceeded that elsewhere on Kaua'i, it also increased during 1970-1990, but in this case it was returning to a level reached previ- ously in the 1940s. The infrastructure to support tourists (lights included), indexed by the number of persons passing through the Lihu'e Airport, increased more than 12-fold in recent decades 3O 2O 15 l0 0 ß NORTHSHORE ¸ EASTSIIORE ß SOUTHSHORE 1930 1960 1990 30 2.0 . ß 197¸ 90 Indices to urbanization of Kaua'i, Ha- 1960 FIGURE 5. waiian Islands: (A) Number of permanent residents on North-, East-, and Southshores, 1930 1990 (cf. Fig. 1); (B) Number of passengers at Lu'e Airport (mostly tourists who need to reside at hotels, condominiums, etc.), 1960-1993 (data from State Airports Commis- sion). Another commercial but private airport opened in Princeville ca. 1980 (but no data on passengers are available). (Fig. 5B). We hypothesize that this growth, too, with its accompanying lights, probably affected the ability of urban areas to attract fledglings. Many coastal hotels, restaurants, sporting facil- ities, etc., have been built to accommodate these tourists and the resident population to service them. To summarize, then, on portions of Kaua'i where urbanization has been increasing recently, more and more fledglings have been recovered by SOS; where urbanization has been even dens- er and more widely spread for a long time, no trend in SOS retrievals has occurred. We hy- pothesize that either the proportion of fledglings attracted to lights and the retrieval capabilities of SOS have become saturated in those areas, or an increasingly greater proportion of fledglings are being attracted and the shearwater popula- tion has suffered greater mortality due to fallout and, in effect, has declined (see Discussion). In other words, the decline is masked because an increasing proportion of fledglings are being at- tracted to lights. POPULATION MODELING To put our results into perspective, we devel- oped a Leslie model (Leslie 1945), that incor- porates the various parameters of productivity and mortality. Before doing this, and in order to estimate mortality rates, we had to estimate the total number of fledglings produced on KauaT The average 9,636 fledglings/yr was derived by multiplying three values: (1) 84,000, the esti- mated total population of Hawaiian Newell's Shearwaters (excluding fledglings; Spear et al. 1995); (2) 0.637, the proportion of total popu- lation of breeding age, i.e., 6 yrs or older, de- rived from the stable age distribution (see be- low); and (3) 0.547, the proportion of adults that bred in any given year (see below). The result was 14,600 breeding pairs, which produced 0.66 fledglings per pain Our estimate of fledgling numbers does not correct for the few that would occur on Hawai'i (where radar studies indicate far fewer Newell's Shearwaters than on Kaua'i; Ainley et al. 1997b, Reynolds et al. 1997a). PARAMETERS USED IN THE POPULATION MODEL Five demographic parameters were required in the Leslie model: survival of adults (i.e., those 116 STUDIES IN AVIAN BIOLOGY NO. 22 32 64 q28 256 5:2 qooo 4000 8ooo LOG BODY MASS FIGURE 6. Relationship between body mass and an- nual adult survival among procellariiforms; data ex- tracted from Galllard et al. (1989) and Dunning (1992). A significant relationship (P - 0.045) exists between log (adult survival) and log (body mass). birds physiologically mature); survival of juve- niles and subadults (i.e., birds between fledging and 12 months of age and those after the first year of life but before adulthood, respectively); age of first breeding; reproductive success; and breeding probability (i.e., the probability that an adult will breed in a specific year). Before dis- cussing model results, we present values for these parameters here, incorporating empirical results and those from the literature. Annual survivorship Annual survivorship, as in most seabirds, has not been studied in Newell's Shearwater. We es- timated annual adult survival to be 0.905, a val- ue reported for a population of the very closely related Manx Shearwater (P. p. puffinus; taxon- omy summarized in Ainley et al. 1997a), whose numbers have been stable and which has been exhaustively studied since the 1950s (Brooke 1990). This value is consistent with those re- ported for procellariiforms of similar mass (Croxall and Gaston 1988) and with an allome- tric relationship to body mass (381 g; see above) among procellariiforms (Fig. 6). From this re- gression, the predicted value for adult survival of a Newell's Shearwater was 0.904 -+ 0.017, with an approximate 95% prediction interval of 0.870-0.934. Juvenile and subadult survival Juvenile and subadult survival also have not been studied in Newell's Shearwater and are poorly known in procellariiforms and most wild birds. The well-studied Manx Shearwater, again, can provide some insight. After adjusting for dispersal, Brooke (1990) estimated that 33.3% of Manx Shearwater fledglings survived from fledging to breeding age (age 6 yrs or older). We incorporated this value into the simulations for Newell's Shearwaters, after considering the fol- lowing patterns in the few other seabird species for which empirical data are available. Annual survival of juvenile and subadult alcids (e.g., Common and Thick-billed murres [Uria aalge and U. Lornvia], the size of which is similar to Newell's Shearwater) at ages 1, 2, and 3 yrs, respectively, is 60%, 82-83%, and 95-96% of the adult value; from the fourth year on, suba- dults have attained 100% of the adult value (Nur 1993, De Santo and Nelson 1995; S. Beissinger and N. Nur, unpubl. data). A similar pattern has been observed among male Western Gulls (Lar- us occidentalis; Spear et al. 1987), South Polar Skuas (Catharacta maccormicki; Ainley et al. 1990), also similarly sized to Newell's Shear- water, as well as among the heavier-bodied Ad- lie Penguin (Pygoscelis adeliae; Ainley and DeMaster 1980) and African Penguin (Sphenis- cus dermersus; Shannon and Crawford 1999). This pattern of age-specific survival was main- tained by us for the Newell's Shearwater while scaling survival upward to achieve a total sur- vival of 0.333 between fledging and age 6 yrs. The result was annual survival estimates of 0.654, 0.78, 0.89, and 0.905 in the first four years of life, and 0.905 for each year of life thereafter (within 1 SE of 0.904, the value ob- tained from the allometric regression, above). These survival values, consistent with those for other seabird species, if anything, may be a bit high rather than too low. For example, sur- vival from fledging to age 6 yrs in a growing population of Cory's Shearwater (Calonectris cliomedea; Mougin et al. 1987) was estimated to be in the interval 0.230-0.334; Simons (1984) assumed survival from fledging to breeding age of 0.268 for a stable population of Dark-rumped Petrels; and for five alcid species, survival to average breeding age ranged 0.244-0.345 (Hud- son 1985). As pointed out below, given an adult survival of 0.905, survival from fledging to breeding age would need to be at least 0.333 to produce a stable population; therefore, we re- tained this estimate. Age of first breeding On the basis of an average age of first breed- ing in the Manx Shearwater of six to seven years (Brooke 1990) and data presented in Table 1, we assumed that no Newell's Shearwater breeds be- fore age 6 yrs, and from age 6 yrs onward, all individuals breed with probability, p (see be- low). Among 15 banded, known-age Newell's Shearwaters recovered by SOS during the past several years (Table 1), essentially all <5 yr of age were found t¾om the period of late-egg lay- ing onward, suggesting that they did not breed. NEWELL'S SHEARWATER DEMOGRAPHY--Ainley et al. 117 Of the two birds 6-7 yrs old, one was found in the prelaying period, consistent with a bird ar- riving early enough to breed. We further esti- mated that 1-p fraction of 6-yr-old Newell's Shearwaters would not breed in a given year. The result of this assumption is that the actual mean age of first breeding is between 6 and 7 yrs in all of our simulations, which is consistent with values not just for Manx Shearwater but also for the other shearwater species for which empirical data are available: 7 yrs in the Short- tailed Shearwater (Bradley et al. 1989) and 9 yrs in Cory's Shearwater (Mougin et al. 1987). Longevity We assumed a maximum age of 36 yrs for Newell's Shearwater. This corresponds to the maximal age observed among other shearwaters (e.g., Bradley et al. 1989). Productivity We used a breeding success value of 0.66 fledglings/breeding pair, a value determined in our study, to simulate the current trajectory of the Kaua'i population. In the Manx Shearwater, reproductive success was 0.70 (Brooke 1990), a value consistent with that reported for Short- tailed Shearwater (Wooller et al. 1989). We used 0.70 to simulate a balanced Newell's Shearwater population. The low numbers of fledglings picked up by SOS during 1992 and 1993 may be due to sev- eral factors: (1) strong E1 Nifio conditions that negatively affected food availability and, thus, shearwater breeding success (see below); (2) the possibility that Hurricane Iniki killed many birds, forcing a need for much new pairing and construction of burrows, two factors that result in lower breeding success in other seabirds (but this is unlikely; see above); (3) an absence of bright lights (which attract fledglings) on Kaua'i for many months after the storm (see above); and (4), at least for 1992, the effect of the full moon during fledging (Fig. 3). None of these explanations are likely to explain the pattern en- tirely, however, because the numbers of fledg- lings found by SOS continues to decline even through 1998 (SOS, unpubl. data; T Telfer, pers. comm.). We hypothesize that recently we have begun to see the effects of the costs of fallout and adult mortality on the stability of the shear- water population (see below). The low number of birds turned in during fall 1978 was certainly a result of the start-up nature of the SOS program. The large number found in 1987 was unusual, but is consistent with that year being at the start of one of the strongest La Nifias of recent decades. At that time, unusually productive waters existed in the eastern tropical Pacific, where Newell's Shearwaters feed (cf. Ribic et al. 1992, Spear et al. 1995). Thus, in 1987, breeding success may have been unusu- ally high, and, in that year, the timing of the full moon would not have decreased the numbers of fledglings found. It is not known whether the fewer fledglings found in some years, as a function of moon phase, is due to their greater ability to see struc- tures in the moonlight (hence, fewer crashes), or whether fledglings are attracted away from civ- ilization by the very bright moon (as suggested by Reed et al. 1985). The moon is clearly the brightest light source around and is low on the horizon just after sunset; our surveys indicated that most fallout occurs during the three hours after sunset (Ainley et al. 1995). Also hypothe- sized as a possibility by Reed et al. (1985), but determined to be false by us, is that moon phase affects the fledging rate, i.e., young may not fledge during the bright full moon (Ainley et al. 1995). Breeding probability The breeding probability parameter refers to the proportion of adults occupying a burrow in which no egg is laid. In Newell's Shearwater, 46% of occupied burrows produced an egg dur- ing the 1981-1985 period. Some of these bur- rows were surely occupied by prebreeding in- dividuals. Assuming that (1) all 4- and 5-yr-old Newell's Shearwaters occupied burrows but did not breed (see above), (2) all individuals >5 yr of age occupied burrows and bred with proba- bility p (see above), and (3) 4- and 5-yr-olds composed 15.9% of all individuals 4 yr or older (as determined from simulations described be- low), we can solve for the fraction of breeding- aged individuals that bred. Dividing 46% by 84.1% (proportion of burrow-holding population that has the potential to breed) yields an annual breeding probability of 0.547. In the Manx Shearwater, 20% of adults that had bred previ- ously do not breed in a given year (Brooke 1990); in Short-tailed Shearwater, 12% of adults do not attend the colony and 19% maintain bur- rows but do not lay an egg (i.e., breeding prob- ability is 0.69; Wooller et al. 1989). There are two sources of uncertainty concern- ing our estimate of breeding probability. First, sampling error is associated with the estimate of 46% of pairs breeding among those occupying burrows (SE = 0.035; 95% CI = 0.39-0.53). Second, uncertain is our assumption that 15.9% of burrows are occupied by prebreeding individ- uals (i.e., that all 4- and 5-yr-olds occupy bur- rows but do not breed and that 2- and 3-yr-olds do not occupy burrows). If 3-, 4- and 5-yr-olds occupy burrows, this implies that 22% of bur- 118 STUDIES IN AVIAN BIOLOGY TABLE 2. PARAMETER ESTIMATES USED IN LESLIE MODELS UNDER DIFFERENT SCENARIOS NO. 22 Best estimate: Newell's Shearwater Balanced: W/O Power line W/Power line W/Power line Manx mortality, predation, or mortality, predation, mortality, and fallout, Shearwater fallout and fallout but W/O predation Survival: adult 0.909 0.905 0.896 0.904 Survival: fledgling to adulthood 0.333 0.333 0.239 0.327 Age first breeding 6-7 6-7 6-7 6-7 Breeding success 0.70 0.66 0.634 a 0.634 a Breeding probability 0.80 0.547 0.547 0.547 X 1.000 0.968 0.939 0.963 0.66 chicks fledge per breeding pair but 4% of fledged chicks die in fallout (not rescued by SOS); therefore, 0.66 x 0.96 - 0.634. rows are occupied by prebreeders; alternatively, if only 5-yr-olds occupy burrows, this implies that 11% of burrows are occupied by prebreed- ers. In turn, this implies that breeding probabil- ity may vary between 0.60 (if only 5-yr-olds hold burrows) and 0.50 (if 3-, 4- and 5-yr-olds all hold burrows). We used 80% breeding prob- ability for simulating a balanced population (Manx Shearwater), but 54.7% for simulating the contemporary Newell's Shearwater popula- tion. The factors that affect breeding probability in Newell's Shearwater are not known for certain. Why reproductive effort was so low especially in 1993 is difficult to ascertain. As in 1983, the year that Telfer (1986) found the fewest burrows with eggs, 1993 was a year of major E1 Nifio. Characteristic of such years, seabirds forgo re- production because of a lack of food reserves (Schreiber and Schreiber 1984, Ainley and Boe- kelheide 1990). The 1993 breeding season also closely followed the devastation of Hurricane Iniki (September 1992). We saw some evidence of terrain slumping and a few uprooted trees at Kalaheo. Thus, the high level of nonbreeding could have been related to storm damage, but we saw little evidence of major burrow exca- vation, and many burrows used in 1993-1994 was not unusual for this population. It is not clear why breeding probability is low in the Newell's Shearwater (54%), but it may result from a high level of mate loss (itself a result of excessive mortality, see below) because, among seabirds, breeders who have lost their mates usu- ally cannot obtain a new one quickly (e.g., Ain- ley and DeMaster 1980, Boekelheide and Ainley 1989). It could be, too, that our estimates are biased. SIMULATION OF A BALANCED POPULATION Incorporating values from the Manx popula- tions (Table 2), our model produced a population that was nearly balanced but still declined slow- ly at 0.65% per year (X, the finite population growth rate = 0.994). Thus, after 10 years, the population will have declined by 6.3%. Increas- ing adult survival from 0.905 to 0.909, however, produced a stable population: X = 1.000. An adult survival rate of 0.909, within 1 SE of Broo- ke's (1990) estimate of 0.905, is statistically rea- sonable. Substituting a breeding probability of 0.547 and a reproductive success of 0.66 in the model, i.e., Newell's Shearwater values, produced a population that declined at 3.2%/yr (X = 0.968). This is our best estimate of the current popula- tion trajectory of the Newell's Shearwater in the absence of additional mortality due to fallout, collisions with power lines, or from introduced predators (see below). In other words it is an idealistic scenario. The main factor affecting the declining growth rate was the fact that breeding probability was 0.547, rather than 0.8. Substi- tuting 0.547 into the model was by itself suffi- cient to reduce population growth rate from 1.000 to 0.978. A breeding success of 0.66 (ver- sus 0.70) and adult survival of 0.905 (versus 0.909) were of minor influence in lowering pop- ulation growth rate, accounting for an additional drop from 0.978 to 0.968. POPULATION STABILITY WITH MORTALITY OF FLEDGLINGS DURING FALLOUT We next added mortality to fledglings during fallout to the simulation, i.e., attraction to lights and subsequent death owing to a complex of fac- tors (see Introduction). This mortality occurs in spite of the efforts of SOS. On the basis of the SOS data gathered by an opportunistic effort, the percent of fledglings that died annually, among those encountered by SOS, was 7.7% (see above; Fig. 2). However, on our night surveys in which search effort was quantified, 43% of fledglings were found dead (Ainley et al. 1995, Podolsky et al. 1998). The discrepancy with SOS must be due partly to dif- ferent areas being surveyed, our sampling of ar- eas that were less frequented by citizens (e.g., NEWELL'S SHEARWATER DEMOGRAPHY--Ainley et al. 119 TABLE 3. FLEDGLING MORTALITY AS A FUNCTION OF MORBIDITY AND DISCOVERY RATES a OF NEWELL'S SHEARWATERS ON KAUA'I, HAWAIIAN ISLANDS Discovery rate Morbidi- ty rate 100% 80% 67% 50% 7.7% 0.011 0.014 0.017 0.023 15% 0.022 0.028 0.033 0.044 25% 0.037 0.046 0.056 0.074 43% 0.064 0.080 0.096 0.127 a Morbidity - percentage dead among downed fledglings; discovery - percentage of downed fledglings found by SOS. sugarcane fields, secondary roads), and the re- luctance of the public to salvage dead birds for SOS. The true mortality could be approximated better if we knew the number of birds that citi- zens rescued from our circuits each night before we passed through. Our regular checks of SOS shearwater drop-off stations in the vicinity of our circuits, however, revealed a few (1-5) but not disproportionately large numbers of addi- tional live birds. Clearly, a greater proportion of each year's fledgling cohort dies than is revealed by SOS data. This hypothesis is supported by the fact that SOS reported none of 44 dead birds tagged and left in place by us during autumn 1993 and 1994 (Ainley et al. 1995, Podolsky et al. 1998). Thus, the true morbidity, i.e., proba- bility that a downed fledgling dies, is likely be- tween 7.7% and 43% of all fledglings. In our simulations, we considered these extremes as well as intermediate values of 15% and 25%. Finally, we estimated the proportion of all downed fledglings encountered by the public and (if alive) brought to SOS stations, i.e., dis- covered. An extreme assumption would be that citizens discovered (and recorded) all downed fledglings. This scenario is unlikely, because some fledglings fall in inaccessible areas, such as sugarcane fields (which occupy a huge pro- ponion of Kaua'i's coastal plain and are crossed by many kilometers of power lines), as well as other factors that could prevent discovery (e.g., birds moved by predators, birds hiding in the bushes). On the other hand, without recording them, some citizens find birds and release them into the ocean at the beach (probably jeopard- izing the shearwaters, which are not anatomi- cally prepared to deal with surf). The proportion of individuals that escape on their own are not our concern here. Therefore, we have considered four scenarios: 100%, 80%, 66.7%, and 50% of all downed fledglings are discovered by SOS. Combining four levels of morbidity and four levels of discovery yields 16 combinations of total fledgling mortality (Table 3). We recognize that the two dimensions, morbidity and discov- TABLE 4. POPULATION GROWTH RATES ()k) IN RELA- TION TO FLEDGLING MORTALITY, ADULT, AND SUBADULT MORTALITY AND PREDATION OF NEWELL'S SHEARWATER ON KAVA'l, HAWAIIAN ISLANDS Adulffsubadult power line-caused mortality Fledgling mortality None Low Medium High Without predation from introduced animals: 0.02 0.966 0.965 0.963 0.04 0.965 0.963 0.962 0.06 0.964 0.962 0.960 0.08 0.963 0.961 0.959 0.10 0.961 0.960 0.958 With predation from introduced animals: 0.02 0.941 0.939 0.04 0.939 0.938 0.06 0.938 0.937 0.08 0.937 0.936 0.10 0.936 0.934 0.962 0.960 0.959 0.958 0.957 ery, are likely related: the more fledglings that come down in areas not covered efficiently by citizens, the higher the level of morbidity, since many fledglings will not be able to recover (e.g., it would take days, if ever, for a shearwater to extricate itself from the tall, dense foliage of a sugarcane field). However, our intention is mere- ly to indicate the range of fledgling mortality likely to be sustained by this population. Total fledgling mortality due to fallout for the 16 dif- ferent combinations of morbidity and discovery ranged 1.1% to 12.7%. Thus, in the most opti- mistic scenario, 1,432 out of 9,636 fledglings are downed and 7.7% of the 1,432 die (110/9,636 = 0.011). In the most pessimistic scenario, 2,864 fledglings are downed and 43.1% of these die (1,232/9,636 = 0.128). We simulated the effects of fledgling mortality due to fallout, allowing the fraction of fledglings dying to vary from as low as 0.02 to as high as 0.10, where all other parameter values corre- sponded to our best estimate model (Table 2). High fledgling mortality (0.10) lowered 2t by 0.5%, compared to low fledgling mortality (0.02; Tables 4, 5; Fig. 7A). POPULATION STABILITY WITH SUBADULT AND ADULT MORTALITY DUE TO POWER LINE COLLISIONS As indicated above, about 61 subadults and adults have been found dead as a result of power line collisions each year. This by no means in- cludes all such individuals, as an adequate search of inland power lines, of which there are about 40 km across sugarcane fields, was be- yond our resources (Ainley et al. 1995, Podolsky et al. 1998). We assumed true island-wide mor- tality to be either 122 birds (i.e., twice the mea- sured level: "low power line mortality"), 244 120 STUDIES IN AVIAN BIOLOGY NO. 22 TABLE 5. COMPARISON OF POPULATION GROWTH RATES () AT DIFFERENT LEVELS OF SPONTANEOUS ES- CAPEMENT BY DOWNED NEWELL'S SHEARWATER FLEDG- LINGS ON KAUA'I, HAWAIIAN ISLANDS, WITH AND WITH- OUT PREDATION FROM INTRODUCED ANIMALS AND WITHOUT THE SOS PROGRAM 25% of dovned fledglings 0% of dovned fledglings es escape cape Population Differcnce  Population Difference a Fledgling mortality Grovth rate Grovth rate 1000 - 800 - 600 ' Low-level po;ver line mortality for adults/subadults,  oo vithout predation:  0.02 0.958 0.0071 0.955 0.0096  0.06 0.955 0.0073 0.952 0.0099 - 0.10 0.952 0.0075 0.949 0.0102 ' Lov-level pover line mortality for adults/subadults,  000- with predation: - 0.02 0.934 0.0062 0.932 0.0085 0.06 0.932 0.0063 0.930 0.0087 0.10 0.929 0.0066 0.926 0.0091 Boo - a Absolute difference in h, comparing population growth rate for a pop- ulation with (see Table 4) and without SOS program (in which 25% or 0%. respectively, of thc 1,432 fledglings turned in each year would escape on their own). birds (i.e., 4 X 61, "medium power line mor- tality"), or 350 dead birds ("high power line mortality"; see Ainley et al. 1995, Podolsky et al. 1998, for derivation). In addition, such mor- tality is apparently age specific: subadults appear more vulnerable than breeding adults. First, as noted above, 20% of the birds found and nec- ropsied by us were active breeders, yet an esti- mated 35% of such birds exist in the population (on the basis of the model, 0.637  0.547). Sec- ond, an additional sample of 15 known-aged (banded) subadult and adult individuals killed by power lines (Table 1) indicated that only two (13%) were 6 yrs of age or older (i.e., of breed- ing age); the remainder were 2-5 yrs of age (subadults). Thus, the two samples yielded sim- ilar adult:subadult ratios. On the basis of these data, we assumed either 24, 48, or 70 dead breeders per year (122, 244, or 350 X 0.2). Dividing 24, 48, and 70 by the total number of adults at the colony yielded mor- tality rates of 0.046%, 0.092%, and 0.131%, re- spectively, for the three levels of power line mortality. For subadults ages 2-5 yrs (Table 1), depending on the level of power line mortality, we derived mortality rates of 0.60%, 1.20%, and 1.72%, respectively. We simulated population growth rate incor- porating these levels of subadult and adult mor- tality, together with a range of fledgling (fallout) mortality values (Table 4). The effect of high subadult and adult power line mortality com- pared to no such mortality was to lower popu- lation growth rate by 0.5% for a given level of 600 400 COLLISIONS A FLEDGLING MORTALITY: 0 4 lb 1 2b . 8 ADULT MORTALITY: NONE ß HIGH A 2.1% YEAR FIGURE 7. Results of simulations shoving effects on population growth in the Nevell's Shearvater caused by: (A) fledgling mortality of 0, 4, and 10% due to fallout; and (B) no versus high adulffsubadult mortality due to collisions vith pover lines. The hor- izontal line indicates 50% population level. fledgling mortality. Low and intermediate sub- adult and adult power line mortality generated intermediate levels compared to high and no power line mortality. It appears that the magni- tude of the effect of power line mortality on population growth rate is roughly comparable to the estimated effect of fledgling mortality (i.e., depressing the population growth rate by as much as 0.5%). We considered the high level of power line mortality to be the best estimate of such (Ainley et al. 1995, Podolsky et al. 1998). Nevertheless, for analyzing effects of predation and efficacy of the SOS program, to err on the side of caution given the uncertainties involved, we used our most conservative estimates of power line mortality. POPULATION STABILITY WITH PREDATION OF BURROW OCCUPANTS Mortality due to predation from introduced animals should be considered additional to mor- tality already discussed, since most studies of shearwaters have been conducted at sites where NEWELL'S SHEARWATER DEMOGRAPHY---Ainley et al. 121 predation of subadults and adults is low. Some Manx Shearwaters are taken by Great Skuas (Catharacta skua) and large gulls (Larus spp.), but numbers of these avian predators are ex- tremely low because of control programs (Fur- ness 1987). Predation by humans had a marked effect on the Cory's Shearwater population (Mougin et al. 1987). We found 30 dead subadults and adults among the estimated 600 individuals in the Kalheo colony (about 150 burrows X 2 yrs X 2 individ- uals/burrow/yr; Ainley et al. 1995). This yielded a crude estimate of 5% mortality among burrow holders. We modeled this as an extra 2.5% mo- rality averaged over all adults and subadults (ex- cluding 1-yr-olds). We chose to use 2.5% mor- tality (rather than 5%) to reflect the fact that some individuals killed might have been tran- sients and not burrow holders, and some non- breeders of breeding age might not have been present at the burrows at all, as in the Short- tailed Shearwater (Wooller et al. 1989; see above). As shown below, even 2.5% mortality of subadults and adults has a dramatic effect on population growth rate. We have not considered mortality of adults or chicks due to predation by rats, for we have no data on rat predation, a most difficult factor to quantify (Thompson 1987, Sero 1995, Sero and Conant 1996). We also considered that predation, especially from owls, is age specific. Active breeders are inconspicuous, so we assumed that they incurred a very low predation rate, whereas 4- and 5-yr- olds, who attempt to gain both a burrow and a mate, are most conspicuous of all and suffered the highest predation rate. We assumed that in- active breeders (individuals that bred in a pre- vious year, but not the current year) and 2-and 3-yr-olds incur intermediate levels of predation. Taking into account our subjective assessment of predation risk, 2- and 3-yr-olds were assigned a mortality rate due to predation of 5%; 4- and 5- yr-olds a rate of 10%; and breeding age individ- uals (whether active or inactive) a predation rate of 1%. Averaged over all individuals 2 yrs of age or older, mortality due to predation was 2.5%. The effect of predation on population growth was dramatic (Tables 4, 5). Simulations indicat- ed a decline of 0.023-0.024 in the finite popu- lation growth rate, depending on the level of mortality assumed for fledglings, subadults, and adults. Thus, the two most important factors in determining population growth (and in this case, decline) were the low breeding probability com- pared with that of stable shearwater populations (0.547 versus 0.80) and the apparently high mor- tality rate due to introduced predators. The two may well be related; loss of mates (due to pre- dation, hurricanes, or power line collision, see above) may lead to a reduced breeding proba- bility for the current or subsequent breeding sea- son. POPULATION STABILITY WITH SOS REDUCTION OF FLEDGLING MORTALITY There is little information regarding the num- ber of fledglings that come to ground but then, in the absence of SOS, spontaneously escape to the sea. Here, to assess the impact of SOS, we consider two possibilities: 0% and 25% of downed fledglings escape on their own. Telfer et al. (1987) proposed that few fledglings that fallout would be capable of survival on their own. In fact, we observed two fledglings who took off after being grounded (it was windy and they were in a large, unobstructed expanse-- empty parking lots; Ainley et al. 1995). In the simulations (Table 5), 2.0-10.0% of all fledglings were assumed to have died as a result of hitting power lines, etc., just as was imple- mented in the simulations shown in Table 4, and then an additional 1,432 (due to not being res- cued by SOS participants, assuming 0% spon- taneous escape) or 1,074 (assuming 25% spon- taneous escape) fledglings die. The decline in population growth rate in the absence of SOS was 0.62-0.75% if 25% of downed fledglings escaped on their own, and ranged 0.85-1.02% in the absence of spontaneous escape. Therefore, the SOS program has had a significant effect on population growth of the Kaua'i population: fledgling mortality, in the presence of SOS, low- ered the population growth rate by 0.12-0.62% but, in the absence of SOS, lowered it by an additional 0.62-1.02%. MODELED POPULATION TRAJECTORIES Finally, we modeled the population trajectory for the Kaua'i population of Newell's Shear- waters under four scenarios, assuming an arbi- trary starting population size of 1,000 individu- als (of all ages >1 yr; Fig. 8A). All scenarios assumed a "low" level of subadult and adult mortality due to power lines and 4% mortality of fledglings due to fallout (see above). Scenar- ios 1 and 3 assumed the continued operation of the SOS program, but scenarios 2 and 4 assumed no such program, and further assumed that 25% of downed fledglings spontaneously escape (see above). Scenarios 1 and 2 included no provision for mortality due to introduced predators; sce- narios 3 and 4 included such mortality. We also considered results for these scenarios with values of breeding probability and repro- ductive success from the Manx Shearwater (Ta- ble 2, Fig. 8B; Brooke 1990). After all, Newell's Shearwater eggs raised in the absence of pred- 122 STUDIES IN AVIAN BIOLOGY NO. 22 lOOO 800 600 400 2OO 000 800 600 400 %.. A -- No Introduced Predators z No Introduced Predators, No SOS ß Introduced Predators D Introduced Predators. No SOS NEWELL'S SHEARWATER  B MANX SHEARWATER YEAR FIGURE 8. Results of simulations showing effects on population growth of shearwaters in the face of low-level mortality to fledglings due to fallout (SOS in operation), low levels of mortality to adults/suba- dults due to collisions, and low levels of predation on adults/subadults in the breeding colonies, assuming: (A) demographic parameters estimated currently for the Newell's Shearwater, and (B) demographic param- eters estimated for the Manx Shearwater. ators (by Wedge-tailed Shearwaters) attained a success equal to that of the Manx Shearwater (Byrd et al. 1984). Moreover, Manx values, when combined with other parameter values, produced a population declining slightly, where- as Newell's values produced a population de- clining steeply (see above). If one concludes that the Newell's population is declining slightly rather than steeply, one should adopt Manx val- ues. Whatever values one uses, however, quali- tatively similar results are produced: the cessa- tion of the SOS program would accelerate the decline of the Newell's Shearwater population by two fold (in the absence of predation). DISCUSSION Comparing the spatial and temporal patterns in the SOS data with those evident in urbaniza- tion, as well as modeling results, we interpret the trends seen in fallout as follows. The shear- water population on the Southshore is decreas- ing. The increased urbanization there, which is compensated somewhat by the slightly increased use of shielded lights (since 1987), should lead to more shearwaters being found, all else being equal. The opposite pattern observed, however (no increase in fallout), is consistent with an added cost (mortality that is not uncompensated) and a declining population. The severe reduction in the size of the Southshore colony at Kalu- ahonu (few occupied burrows present in 1992- 1993 compared to the early 1980s; Ainley et al. 1995) is consistent with this trend. In fact, be- cause our inputs to the demographic model were gathered on the Southshore (Kalaheo colony, routes to quantify mortality), our model results duplicate well what we propose is happening to the Southshore Newell's shearwater population on the basis of SOS results. In contrast to the Southshore, shearwater col- onies on the Eastshore and Northshore are fac- ing increased urbanization (well beyond histor- ical levels; Fig. 5A) and, as predicted with more lights, more birds are being reported to SOS (Fig. 4A, C). The very recent growth in urban- ization is so dramatic that the increased use of shielded lights (although still minimal) must be having little compensatory effect on shearwater fallout. Due to mortality and ensuing population decline, the fallout pattern for the North- and Eastshore eventually should duplicate the trend seen on the Southshore: level or decreasing fall- out. Indeed, following our study, the number of fledglings found in 1995, 1996, 1997, and 1998 (T. Telfer, pers. comm.) continued the "unex- plained" gradual lowering of SOS totals that be- gan in 1992 (or even 1987). In the absence of fallout, power line-caused mortality, and introduced predators, the model showed that the Kaua'i population of Newell's Shearwaters should be able to maintain its num- bers, i.e., no other important factors affect pop- ulation instability. The SOS program goes far to reduce one of these mortality factors, death of fledglings due to fallout. Even with SOS, how- ever, there is significant mortality of fledglings; >2% and as much as 10% or more of fledged shearwaters likely die as a result of fallout. Mor- tality of subadults and adults due to power line collisions also depresses population growth, but depending on the actual rates obtained, it may or may not be as important. Firm quantification of the significance of power line-caused mortal- ity among subadults and adults awaits further study. In the absence of the SOS program, how- ever, fallout-caused mortality of fledglings NEWELL'S SHEARWATER DEMOGRAPHY--Ainley et al. 123 would likely be more important than power line- caused mortality of subadults and adults. Evidence points clearly to two factors that im- portantly affect population growth of the New- ell's Shearwater: low breeding probability and high rates of predation on adults and subadults. The cause of the low breeding probability are not readily apparent, but rates would be exac- erbated by mortality of breeders and prebreeders due to predation, disturbance by predators, and collisions with power lines. Otherwise, even if the Newell's Shearwater breeding population is not currently declining (i.e., the model is wrong and the SOS results are not a valid index of pop- ulation size), our results indicate the vulnerabil- ity of the Newell's Shearwater population. A re- duction in the production and survival of fledg- lings will only be felt many years later, at the time when such fledglings would have begun breeding. Remember, the longevity of this spe- cies is about 30 yrs, and not even one generation has passed since urbanization began to expand rapidly. We ask, Are the low SOS totals con- tinuing past 1987 and the unusually low banded- bird recovery rates finally indicating decreased survival? Seen in this context, mortality of adults and subadults due to collisions is still of great concern for recovery of Newell's Shear- water (see USFWS 1982a). Alternative hypotheses exist, of course, to ex- plain some of the trends revealed by our re- search and simulations. The regional difference in trends could be a result of an increasing pop- ulation of shearwaters on the Northshore due ei- ther to immigration from colonies on the South- shore (in turn to help explain the decrease there) or much better breeding success on the North- shore than on the Southshore. A shift from the Southshore to the Northshore is problematic giv- en the high degree of philopatry characteristic of procellariiforms (Warham 1990). The very low recovery rate of shearwaters initially band- ed as fledglings by SOS could be a result of a lower-than-natural survival rate of these birds (deemed to have been "rescued" by SOS only because they were able to fly away). Another possibility is that the large majority of fledglings picked up by SOS were produced on the North- shore--and eventually recruited to Northshore colonies as adults--but having reached the sea were attracted back to land by coastal lights on the more brightly lighted South- and Eastshores. Until 1995, the Northshore had lacked the power lines that effectively "sample" adults and su- badults in the population, although following completion of our study high, deep arrays of lines have been installed. Thus, sampling effi- ciency may have increased and we can see whether or not the number of banded birds found also increases. Additional research in the colonies on the Northshore also could easily de- termine whether many banded shearwaters nest there. In conclusion, then, the population of New- ell's Shearwaters on Kaua'i appears to be de- clining. On the basis of demographic modeling, the prospects appear to be poor for the continued existence of a robust population of this species on this island. A reversal of the indicated pop- ulation trends will be possible only with more strict controls of lighting (such as on the Big Island, where the astronomical observatories re- quire minimal upward light radiation), fencing and predator control in several important shear- water breeding areas, and, possibly, the burying of power lines in a few especially critical areas (Ainley et al. 1997a, Podolsky et al. 1998). ACKNOWLEDGMENTS T. C. Telfer (Hawaii DLNR) introduced us to shear- waters and petrels on Kaua'i, contributed much in- sight, lent equipment to us, and provided access to past data from the SOS program. He chose not to accept authorship of this paper. Assistance, too, was provided by R. Voss and K. Viernes (USFWS), Kilauea Point National Wildlife Refuge. C. Berg was very helpful and supplied much information early on that improved our knowledge of Kaua'i. We also thank Kaua'i Elec- tric Company and the Electric Power Research Insti- tute (EPRI) for funding, and the coordination provided by J. Huckabee and M. Fraser (EPRI). The counsel received from the EPRI Scientific Advisory Panel was also of great value: D. Boersma (University of Wash- ington), G. Breece (Georgia Power and Southern Com- pany Services), S. Conant (University of Hawai'i), E. Colson (Pacific Gas and Electric, and Colson and As- sociates), E. Flint (Pacific Islands Office, USFWS), L. Ginzburg (Applied BioMathematics), and S. Kress (National Audubon Society). Finally, B. A. Cooper, R. H. Day, M. Fraser, T C. Telfer, L. B. Spear, S. Conant, and several anonymous reviewers provided many good comments on the manuscript. Studies in Avian Biology No. 22:124-132, 2001. MIGRATION OF NORTHERN PINTAIL ACROSS THE PACIFIC WITH REFERENCE TO THE HAWAIIAN ISLANDS MIKLOS D. E UDVARDY AND ANDREW ENGILIS, JR. Abstract. Northern Pintails (Arias acuta) regularly occur as winter visitors on most Pacific islands with suitable habitat. Their breeding distribution includes both sides of the Pacific Rim. While large populations breed in Siberia and winter in California, numerous North American breeders also winter in areas near the Sea of Japan, Hawaiian Islands, and other Pacific island groups. Though pintail flights across the Pacific have not been well documented, scrutiny of banding returns shows that an exclusive Califomia-Hawai'i flyway does not exist, as was earlier proposed. Data support a more complex movement of birds from numerous breeding locations in the Holarctic. We summarize movements of Holarctic nesting pintails to wintering grounds in the Hawaiian Islands that include birds originating from northeastern Siberia, Alaska, and the interior prairie provinces and states of North America. We also summarize pintail movements to other Pacific archipelagoes. Finally, to close the circle around the North Pacific, we summarize movements of birds between Canadian and Alaskan breeding grounds to wintering sites in Japan. We also discuss other panmictic, Holarctic migrants and their colonization attempts in Hawai'i. Key Words: Anas acuta; banding return; Holarctic; migration; Northern Pintail; Oceania; panmixis. The primary interest of a faunist is in establish- ing the list of species that regularly occur in the area under scrutiny. Data of a species' regular occurrence increase knowledge of their total dis- tribution, which is the aim of the zoogeographer. Regularly occurring species are recognized as influential members of local ecosystems; thus, they play a prominent role in ecogeographical studies. Often less attention is paid to scarce, rare, or irregularly occurring species, for chance seems to determine their detection, and their role in community ecology appears negligible. Regarding these "lesser" elements of local fauna, interest increases when a chance visitor comes from afar. Lately, the study of rarities be- came important on two accounts. First, it is re- alized that bird species are to an extent dynamic; the "stray" individuals caught outside of their regular distributional range are all potential col- onists. The trends in their occurrence outside the "normal" range and throughout a longer time period may reveal the nature and extent of the pioneering tendency of the species. Second, it is also realized that species composition of faunas fluctuates; thus rare visitors may reveal trends in faunal changes. Holarctic waterfowl are among the most suc- cessful colonizers owing to their exceptional powers of flight between breeding and non- breeding areas. Their ability to move long dis- tances and tendency for dispersal have resulted in establishment of waterfowl on many remote land masses where food and freshwater re- sources are available (Weller 1980). As with all remote oceanic islands, the Ha- waiian Archipelago received its endemic avifau- na through over-water dispersal and subsequent local speciation. The Hawaiian avifauna consists of year-round residents (the landbirds) and sea- sonal but regular visitors (seabirds that come to breed, and Anseriformes and Charadriiformes that winter in Hawai'i). Thirty-three species of migratory waterfowl have been recorded in the Hawaiian Islands (Pyle 1997). Ten species are annual visitors with Northern Pintail (Anas acu- ta), Northern Shoveler (Anas clypeata), Lesser Scaup (Aythya affinis), American (Anas ameri- cana) and Eurasian (A. penelope) wigeons, and Green-winged Teal (Anas crecca) accounting for 95% of those birds wintering in the islands (En- gilis 1988). Our focus in this paper, the pintail, is a reg- ularly occurring winter visitor in Hawai'i and is a scarce or irregular visitor to other Pacific is- land groups. Reliable but general historical ac- counts claim that pintail came in large numbers to Hawai'i (Munro 1944). Earlier evidence is suggested by the fact that the Hawaiians recog- nized two species by name: pintail (Koloa Mapu) and shoveler (Koloa Moha), indicating that they were an obvious component to the Ha- waiian avifauna before Captain Cook discovered the islands in the 1770s. Surveys have docu- mented migratory ducks exceeding 10,000 birds in the mid-1950s (Medeiros 1958). We exam- ined the data from biannual waterbird surveys conducted on most lowland wetlands since the 1940s (Table 1). We omit data collected from 1960 through 1977 (Ni'ihau, Hawai'i, and Mo- 1oka'i not regularly surveyed during those peri- ods). These data, summarized in Engilis (1988), confirm that the population size of wintering pintails in Hawai'i have declined tenfold. This decline has led to added interest by conserva- tionists to address habitat needs in the Hawaiian Islands benefiting migratory waterfowl and 124 PINTAlL MIGRATION IN THE PACIFIC--Udvardy and Engilis 125 TABLE 1. CENSUS OF PINTAILS IN HAWAI'I Year Total Pintails 1950 1,593 195l 1,875 1952 7,094 1953 8,226 1954 1,950 1955 2,653 1956 3,045 1957 1,619 1958 1,126 1959 1,249 1978 897 1979 490 1980 923 1981 377 1982 150 1983 60 1984 235 1985 150 1986 501 1987 203 Notes: Data from 1950 to 1959 l¾om Meideros (1950-1959). Counts were taken on Maul, Hawai'i, O'ahu, and Kaua'i. Data from 1978 to 1987 from Engilis (1988). During 1960-1977 not all islands were sur- veyed and records are sketchy. The period of 1978 1987 represents the best modern data set as all eight main islands including Ni'ihau were surveyed. shorebirds. Understanding pintail movements to Hawai'i will assist in these efforts. Medeiros (1958) documented the movement of pintails between the Hawaiian Islands and North America, speculating a California-Hawai'i flyway. Although this connection is correct, the true migration patterns are more complex. We analyzed banding data from the U.S. Migratory Bird Management Office (MBMO), Yamashina Institute for Ornithology, Japan, and the Russia Bird Ringing Center. Included in our data col- lection was a summary of available literature, examination of specimens from the American Museum of Natural History (AMNH), National Museum of Natural History (USNM), and Bern- ice P. Bishop Museum (BPBM), examination of bird observation records from the Hawaii Rare Bird Database (HRBD), and fieldwork conduct- ed by us (Udvardy 1958-1960 and Engilis 1984-1997). These sources enabled us to gather considerable amounts of data indicating that pin- tails from at least half of the species circumpolar distribution are potential winter visitors to Ha- wai'i and that their movements across the Pacific are complex. In the following discussion we try to document these assumptions. NORTHERN PINTAlL MIGRATION TO THE HAWAIIAN ISLANDS Of the 2,811 pintails banded in Hawai'i, 107 have been recovered on the North American mainland and 16 have been retrapped on the is- lands. Additionally, a pintail banded on Maui in October 1952 was reported taken a month later from Pukapuka (Danger) Atoll in the Tuamotu Archipelago. Significantly, the Tuamotus are al- most due south from the Hawaiian chain, as are the Line Islands, where two pintails were recov- ered two to three months after same-year autum- nal banding in North America (MBMO data). Medeiros' analysis of these returns lead him to the conclusion that the islands' wintering pintail population is not blown off course but are delib- erately flying from central California to, and re- turn there from, their wintering areas in Hawai'i. Of the above mentioned 107 Hawai'i-banded pintails, 45 were recovered in the San Francisco Estuary, California (Fig. 1). These returns also confirmed that pintails return to the islands one or several years after the initial banding there. Thus, pintails repeatedly and deliberately visit Hawai'i to spend the winter, with some flying further southward after having used the islands in transit (Medeiros 1958). Medeiros speculated that the autumnal flight probably used the north- erly trade winds that originate outside central California, while for the return flight in the spring the ducks probably are helped by the westerlies. According to the MBMO banding/recovery data, 165 pintails have been banded in Hawai'i and recovered (including 16 in Hawai'i) be- tween 1953 and 1960 (Fig. 1). The data reveal that the high number of California returns in the total of Hawai'i-banded ducks matches the dis- tribution pattern, at banding, of 14 pintails band- ed from 1951 to 1954 in North America and later recovered in the Pacific (Fig. 2). In addi- tion, the proportion of California's share in the total of 165 records is 77.6% against all other localities; if we compare California only with the coastal entities of Alaska, British Columbia, Washington, and Oregon, the proportions are 128 against 24, or 84.2%. In order to assess the relation of mainland populations of pintail to the population visiting the Hawaiian Islands, according to the banding and recovery results, we have compared the fig- ures of banding effort, recoveries, and hunting pressure on the Pacific coastal areas of North America for the years of Medeiros's project (Ta- bles 2, 3). We excluded Alaska from these tables because there were no data available for hunting pressure or banding efforts in Alaska for the 1950s. Comparing the data in Tables 3 and 4, we concluded that during the 1950s, California pin- tails were indeed providers of over 90% of the birds annually harvested by hunters in the tem- perate Pacific Coast of North America and also 126 STUDIES IN AVIAN BIOLOGY NO. 22 FIGURE 1. Northern Pintails banded in Hawai'i and recovered anywhere. of pintails annually banded there. The recoveries in California are predominately from the fall when hunting pressure is at its highest. Also, pintails arrive in California earlier than most species of migratory waterfowl, boosting the California figures (Miller 1985). These facts, overlooked by Medeiros, contributed to the pre- dominance of California in the Hawaiian band- ing and recovery data. However, California re- mains a critical area for pintail, supporting over 50% of those wintering in the United States (Heitmeyer et. al. 1989); thus it probably serves as a principle staging area for Hawaiian-bound pintails. This still needs to be confirmed through modern marking and tracking studies. We note that banding recoveries support the notion that pintails could equally originate from other Pa- cific Coast localities such as Mexico, Oregon, or Washington (Figs. 1, 2). A second pattern of movement can be seen from birds banded in Hawai'i and recovered in the Arctic. Five birds banded in Hawai'i in the 1950s were recovered in the Arctic: one in the Aleutian Islands; another in the Yukon-Kuskok- win Delta, an important breeding ground in western Alaska; and two on Alaska's South Coast (Fig. 1). One bird was recovered in the Anadyr Region of eastern Russia (lat. 62 ø 5' N, long. 179 ø 1' E). The later bird was a hatching- year male banded on Maui, Hawai'i, 22 Febru- ary 1954. It was shot on the breeding grounds 29 May 1960. These multiple recoveries strad- dling the Bering Sea provide another migration link from the Holarctic to the Hawaiian Islands. We speculate that Arctic nesting pintail probably make the transoceanic flight direct from south- ern Alaska/Siberia to the Hawaiian Islands, in- tercepting the leeward islands (e.g., Midway and Laysan), resting, and then moving to the main islands. Just as plausible, however, is a move- ment of Alaskan birds south along the Pacific Coast of North America, into California, and then across the Pacific. This movement may be indirectly supported by banding evidence of Alaskan birds as nearly 80% of those recovered have been taken in California (Austin and Miller 1995). The early arrival of pintails to Califor- nia-males arrive in numbers by late August (Miller 1985)-could allow time for birds to re- fuel and make the flight to the Hawaiian Islands. Again, this high return rate of Alaskan-banded PINTAIL MIGRATION IN THE PACIFIC--Udvardy and Engilis 127 FIGURE 2. Northern Pintails banded anywhere and recovered in the Pacific Ocean (triangles recovery location, dots = banding location). pintails can be biased by the high number of birds shot in California. POPULATION DEMOGRAPHY OF PINTAIL WINTERING IN THE HAWAIIAN ISLANDS From Medeiros's banding data we note that Hawai'i had a sex ratio skewed towards females (Table 4). This is atypical for what has been re- ported for pintails (and other ducks) of North America, where males tend to outnumber fe- males in most studies on the wintering and breeding grounds (Bellrose et al. 1961, Miller 1985, Rienecker 1987, Austin and Miller 1995, Migoya and Baldassarre 1995). The higher num- ber of males recorded in waterfowl populations has been speculated to be the result of a high mortality rate (increased predation due to habitat fragmentation) of adult females during the TABLE 2. HUNTING PRESSURE a ON PINTAIL 1950--1956 AT PACIFIC COASTAL AREAS Year British Columbia Washington Oregon California 1950 69,600 109,500 (est.) -- 1,945,300 1951 94,830 114,900 -- 2,966,000 1952 72,620 111,250 -- 4,659,000 1953 94,940 97,800 -- 4,599,500 1954 93,940 112,600 -- 3,461,600 1955 70,490 128,200 -- 3,312,700 1956 71,940 117,700 -- 3,526,000 Totals 568,360 791,950 913,620 (est.) 24,470,100 Yearly Mean 81,194 113,136 130,517 3,495,729 a Figures represent reported birds taken by hunters during the legal hunting season of each year. Source: state and provicial hunting records obtained in writing by M.D.E Udvardy. 128 STUDIES IN AVIAN BIOLOGY NO. 22 TABLE 3. BANDING OF PINTAIL 1950--1956 AT PA~ TABLE 4. SEX RATIOS OF BIRDS BANDED IN THE HA- CIFIC COASTAL AREAS WAIIAN ISLANDS (MEDEIROS 1950-1959) British Sex ratio Year Columbia Washington Oregon Calirnia Year N (Males to Females) 1950 28 110 234 9,334 1951 417 0.63 1951 26 774 544 19,360 1952 856 0.84 1952 31 656 102 17,570 1953 644 0.65 1953 0 433 574 16,737 1954 446 0.94 1954 5 143 1,000 16,514 1955 478 0.50 1955 5 625 2,931 21,475 1956 0 988 1,651 15,759 Total 95 3,729 7,036 116,749 Source: U.S. Migratory Bird Management office records. breeding season (Johnson and Sargeant 1977). The disproportionate numbers of females seen in Hawai'i may therefore be the result of female pintail's tendency to exhibit philopatty to their winter quarters (Rienecker 1987, Anderson et al. 1992), coupled with the effort required to reach the Hawaiian Islands. In addition, pintails un- dergo a sex-segregated migration as males move to molting grounds earlier than females, in some cases arriving months earlier (Fuller 1953, Oring 1964, Salomonsen 1968, Bellrose 1976). Both sexes appear prone to wander, particularly young birds, as is revealed in the specimen record. Of the 42 pintail specimens examined from Pacific islands, 25 were hatching-year birds and 17 were adults. Medeiros's trapping and banding data also revealed a decline in pintail age ratio throughout his study (Table 5). This decline was also reflected in the Pacific flyway pintail pop- ulation and was the result of a severe drought in the prairie provinces of Canada depressing con- tinental waterfowl populations (Ducks Unlimit- ed 1990). The timing of pintail migration to Hawai'i has apparently changed in the past five decades. The decline of pintails in North America has been well documented, and we have seen a similar decline in Hawai'i (Engilis 1988, Ducks Unlim- ited 1990, Austin and Miller 1995). Not only has there been a decline in numbers, but the period of arrival has decreased as well. In the 1950s, Medeiros documented birds arriving, in num- bers, as early as mid-September. His banding re- cords revealed that the early arrival was marked by small flocks of males, followed by females and hatching-year birds that arrived in October. Pintail numbers peaked in November. This sex- segregated migration pattern has been docu- mented for other waterlbwl in North America, particularly in California where male pintail comprised over 90% of the total birds arriving in August but only 53% of the total wintering population once females arrived (Miller 1985). By the mid-1980s to present, pintail arrival patterns changed in HawaiT A more abbrevi- ated migration occurs with the main bulk of pin- tail arriving in the islands, marked by hatching- year birds (based on the timing of their body molt; A. Engilis, unpubl. data) by late October, peaking in November, and stabilizing at a few hundred birds through the winter. We speculate that the early arrival of male pintails to Hawai'i was lost during the years of continental decline (mortality?) from 1975 to 1985 leading to the observed, abbreviated migration and decline in Hawai'i. In the late 1990s, a few early flocks have again been observed in late September; most are comprised of male birds (A. Engilis and A. J. McCafferty, pets. obs.). During the same period, pintail numbers have increased on the continent (USFWS 1996b). MOVEMENT OF NORTHERN PINTAlL ACROSS THE PACIFIC To complete the assessment of pintails mi- grating across the Pacific, we assembled data for pintail banded in North American and recovered in Eurasia. One movement of birds between the continents has been documented, with part of the population breeding in eastern Siberia and win- tering in the western United States (Dement'ev and Gladkov 1952, Henny 1973). Again the banding recoveries (N = 423) yield a more com- plex pattern of movement across the North Pa- cific than first thought. To make sense of these data, we combined the patterns of movement into three groups. TABLE 5. RATIOS OF WINTERING NORTHERN PINTAlL ADULTS TO JUVENILES IN THE PACIFIC FLywAY AND THE HAWAIIAN ISLANDS BASED ON BANDING RECORDS (MED- EIROS 1950--1959) Year Hawai'i Pacilic Flyway a 1951 1.19 3.50 1952 2.07 3.70 1953 0.51 0.50 1954 0.72 0.50 a Extrapoled from Bellrose et al. 1961. PINTAlL MIGRATION IN THE PACIFIC--Udvardy and Engilis 129 FIGURE 3. Northern Pintail banding recoveries in Asia below 50 ø N; birds banded in North America (triangles = recovery location, dots = banding location). GROUP 1 Three birds were recovered in Europe and one in western Siberia. The first bird, a drake, was banded in northern California and shot eight years later in western Siberia. Another drake, also from California, and was recovered two years later from the Arctic coast of Russia's Kara Sea. A third, an immature drake from the Canadian maritime province of Nova Scotia, was found in Chechia two and a half years later. These records provide an example of the mech- anism whereby these circumpolar, wetland spe- cies mix their genotype so that no specialization could occur, supporting the notion that the Hol- arctic pintail population remains panmictic and opportunistic, thus adapted to varying climate conditions (Udvardy 1969:180-181). The last of these cases defies all speculations; an adult fe- male from northern California that was found six years later in the Ukraine (Rienecker 1987, 1988). The remaining 420 pintails mentioned above were divide into two groups: those recovered in Asia below 50 ø N (group 2) and those above it (group 3). GROUP 2 Below the 50 ø N parallel, 21 North American- banded pintails have been recovered in Japan, 1 in Korea, and 2 in Sakhalin Island, Russia. Of these 24 birds, all were winter visitors: 5 were banded on the Aleutian Islands; a scattering came from the tundra or northern parklands of Canada; 11 were in a cluster from the southern Canadian and northern U.S. prairies; and another scattering originates in California and other western states (Fig. 3). These data corroborate Henny's data (1973) and, in addition, show that there is an unknown, but sizable number of North American pintails that regularly winter in the region of the Sea of Japan, the area which is also a wintering ground for some portion of the East Asian breeding population (Dement'ev and Gladkov 1952, Ornithological Society Japan 1974, Meyer de Schauensee 1984). Further, four female pintails, all banded within 5 days of one another on the Aleutian Islands, were recovered in Japan: three of them 40, 52, and 64 days after their banding date, respectively. The fourth was recovered, also in Japan, a year later. These tour females, banded in "immature" plumage, ex- 130 STUDIES IN AVIAN BIOLOGY NO. 22 2O ø 2½ FIGURE 4. Localities where Northern Pintails have been recorded in the Pacific Ocean (banding, sight and specimen records). Sources: Reichenow (1899, 1901), Schnee (1901), Baker (1946), Gallagher (1958-1959), Yocum (1964), Fosberg (1966), Amerson (1969), Ely and Clapp (1973), Palmer (1976), Pratt et al. (1987), Engilis (1988), Stinson et al. (1997), specimens from American Museum of Natural History, National Museum of Natural History, Bernice P. Bishop Museum, and sight observation records Hawaii Rare Bird Database). emplify the regularity of visiting and returning to winter grounds (cf. philopatry of Rohwer and Anderson 1988), reminding us of similar data from Medeiros's banding returns in the Hawai- ian Islands. It is tempting to suggest a Canada- Japan flyway on a great circle route from the North American prairies through the Aleutian Chain, Kamchatka, and the Kuri! Islands. GROUP 3 The remaining 396 banded birds recovered in Asia were there predominantly as spring-sum- mer arrivals because 338 of them were found from April to July, 55 in the fall months, and only 3 in the winter. Beside Henny (1973), a number of publications deal with drought dis- placement of pintails to Alaska and beyond (Derksen and Eldridge 1980; Hestbeck 1995, 1996). Thus, there is a sizable movement be- tween breeding grounds in eastern Siberia and wintering areas in North America. It is not known whether these birds fly over the ocean or in a great circle route or follow a coastal route along the Pacific Rim. To close the circle around the pintails of the Hawaiian Islands, we looked at the rest of Oce- ania (Fig. 4). Our scrutiny of the pertinent lit- erature, banding records, and museum speci- mens shows that every island group of central Oceania has received pintail visitors, often in numbers. We mention here two special cases as extremes. During the period when the Marshall Islands were German colonies, Anton Reichen- ow reported in 1899 about an autumnal duck migration viewed at Jaluit Atoll by reliable pub- lic officers and documented by specimens sent to the Berlin Museum as pintails, Green-winged Teals, and Canvasbacks (Aythya valisineria). "In ununterbrochener Folge ungeheure keilfor- mige Schwiirme" (large numbers in uninterrupt- ed sequence of enormous v-shaped flocks) moved over the atolls of Bikar, Utirik, Ailuk, Jemo, Likiep, and Wotje from north to south in the fall, and back again in May (Reichenow PINTAIL MIGRATION IN THE PACIFIC--Udvardy and Engilis 131 1899, 1901). Another observed migration to- ward north and north-east in the vicinity of Kwajalein Atoll was documented in May 1900 (Schnee 1901). The pintail is also an uncommon, but regular winter visitor to the Mariana Islands, occurring regularly on the main islands of Guam (numer- ous sites), Saipan (Lake Susupe), and Tinian (Hagoi Marsh; Stinson et al. 1997). Kuroda (1961) linked the pintail that reach Micronesia to the "Nearctic Hawaiian Flyway" (cf. Baker 1953), referring to the now unrecognized North American race (A. acuta tzitzihoa). However, with the prevailing storms across Japan moving east and southeast, it is conceivable that the North American connection to the Marianas are actually birds originating from the "Canada-Ja- pan" corridor. SUMMARY OF COLONIZATION EVENTS BY HOLARCTIC MIGRANTS IN HAWAI'I The winter range of Northern Pintail is per- haps the most widespread distribution area of all species of waterfowl (Palmer 1976, Austin and Miller 1995). Pintail are prone to disperse and wander as is evident by the banding, observa- tion, and specimen information synthesized here. The species has been recorded on all con- tinents except Antarctica and shares ancestry with the endemic island form in the Southern Hemisphere, Eaton's Pintail (Arias eatoni). Hol- arctic species prone to wandering have given rise to the majority of known endemic water- birds and most landbirds of North Pacific islands (Fleischer and Mcintosh this volume). The ma- jority of the species that have colonized are those whose resources naturally fluctuate, both on a regional and seasonal pattern. Many of these are representative of highly volatile spe- cies such as fringillid finches, frugivorous thrushes, waterbirds (rallids, shorebirds, and ducks), and raptors, the latter whose populations erupt relative to fluctuating small mammal num- bers. Colonization events have been rarely doc- umented on island groups, and although pintail have yet to be recorded nesting in Hawai'i, other Holarctic migrants have. We summarize three cases where colonization has led to, or is sus- pected to have lead to, a Hawaiian breeding pop- ulation of a Holarctic migrant. FULVOUS WHISTLING-DUCK The Fulvous Whistling Duck (Dendrocygna bicolor) apparently reached Hawai'i under its own power in 1982 when a flock of six birds suddenly appeared on O'ahu (Leishman 1986). They began nesting on O'ahu's North Shore, ex- panding to nearly 30 birds in under five years. Dispersal records of individual birds were doc- umented on Moloka'i, Maui, and Kaua'i during the late 1980s. After the decline of wetlands and aquaculture on O'ahu's North Shore in 1992, the population of Fulvous Whistling Ducks crashed dramatically, so that by 1998 only one bird re- mained on the James Campbell National Wild- life Refuge (A. Engilis, Jr., pets. obs.). It is of interest to note that the whistling duck has high populations on the Pacific Coast of North Amer- ica only in western Mexico. Thus it is conceiv- able that these birds originated from there, as could migratory pintail as stated earlier. A Mex- ico-Hawai'i tie is also suggested by other va- grants that have occurred in Hawai'i: e.g., Little Blue Heron (Egretta caerulea), Laughing Gull (Larus atricilla), and Great-tailed Grackle (Quiscalus mexicanus; Pyle 1997). PIED-BILLED GREBE The Pied-billed Grebe (Podilymbus podiceps) has bred in Hawai'i since the mid-1980s. A sin- gle bird arrived to overwinter in 1984 on 'Ai- makapfi Pond, located on the Kona Coast of Ha- wait It left in the spring of 1985. Two birds returned the following fall, remained, and gave rise to a population on the pond that remained stable at a dozen birds throughout the late 1990s (R. David, unpubl. data). These two birds re- mained and have given rise to a population on the pond that remains stable at about a dozen birds. Dispersal records on Kaua'i, Maui, and on other wetlands of the island of Hawai'i are be- coming more frequent in recent years, probably representing young birds that may have origi- nated from 'Aimakapfi Pond (HRBD, unpubl. data). GREAT BLUE HERON AND WHITE-FACED IBIS Although they have not yet been recorded breeding, two Holarctic ciconids are now regular residents in small numbers, the Great Blue Her- on (Ardea herodias) and White-faced Ibis (Ple- gadis chihi). Great Blue Herons continue to wander through the chain and at times form small groups, often roosting among nesting col- onies of Cattle Egrets (Bubulcus ibis) and Black- crowned Night Herons (Nycticorax nycticorax). What drives these birds to disperse to the Pa- cific islands remains unclear, as do the mecha- nisms of how they navigate to the islands year after year (wintering shorebirds and waterfowl). Mayr (1953) discussed the migration of birds across the Pacific speculating that historically the islands of the Pacific were more massive, thus providing better opportunity for coloniza- tion. He also suggested two patterns affecting Holarctic birds, that of route abbreviation and route prolongation. The Northern Pintail might more readily fall into the latter group of mi- 132 STUDIES IN AVIAN BIOLOGY NO. 22 grants, a species whose migration patterns have been elongated as a result of global climate changes and expanding breeding range north- ward, away from traditional wintering grounds. Baker (1953) further alluded to the fact that oce- anic islands provide excellent wintering grounds due to the absence of mammal and reptilian predators. Both authors discussed their findings with an emphasis on northern nesting shore- birds, including those species where the majority of the known population winters in the Pacific: Bristle-thighed Curlew (Numenius tahitiensis), Pacific Golden Plover (Pluvialisfulva), Wander- ing (Heteroscelus incanus) and Gray-tailed (H. brevipes) Tattlers, and Bar-tailed Godwit (Li- mosa lapponica; Baker 1951, Mayr 1953). Finally, migration out over the Pacific Ocean has rarely been observed, owing to paucity of observers and opportunities. A great "corridor" of migrating Alaska pintails was observed in the fall in southern British Columbia, and another one moving from Alaska southwest across the Pacific has been postulated (Bellrose 1976, Campbell 1990) and backed by observations (Martin and Myres 1969). We uncovered one specimen of American Wigeon (AMNH 131716) collected by C. H. Townsend in 1891 from the USS Albatross, "500 miles NW of O'ahu", documenting yet another species' movement across the Pacific. Perhaps new tech- nologies for tracking large birds (satellite telem- etry and Doppler radar) may help shed light on these movements. The evidence bears out a complex setting for migratory waterfowl in the Pacific, fortunately observers of the past had the foresight to band pintails to help us elucidate these movements herein. What has become clear is that pintail remain a regularly occurring com- ponent of the Pacific island avifauna, represent- ing a link to the mechanics of island coloniza- tion from the Holarctic faunal region. ACKNOWLEDGMENTS In assembling the data for this paper we received enthusiastic help and data from many fellow wildlifZ researchers such as C. Kessler, B. H. Powell, J. M. Sheppard, K. Smith, and G. J. Wiles, all of the U.S. Geological Survey. The following provided banding data for our studies: B. Trost of the U.S. Fish and Wildlife Service for information on pintail banded in North America; M. Yoshii of the Yamashina Institute for Ornithology for information on pintail in Japan; in the 1960s, A. Vinckurov of the Russia Ringing Center, Moscow, for data on pintails in Russia. The original hunter and banding data frmn the 1950s has been cour- teously provided by the departments of wildlife of Ha- wai'i, Alaska, British Columbia, California, Oregon, Washington, and the U.S. Fish and Wildlife Service. B. O'Hara and A. Morton of Ducks Unlimited created all figures and M. Doyle helped with many of the tech- nical details while working on the data. We would also like to thank M. LeCroy (AMNH), P. Angle (USNM), and C. Kishinami and R. L. Pyle (BPBM) for access to specimens and assistance while working with col- lections. K. Evans and J. M. Scott provided helpful comments on the manuscript. The junior author thanks M. Udvardy and family for their support of the manu- script after M.D. E's untimely passing. Studies in Avian Biology No. 22:133-143, 2001. THE HAWAI'I RARE BIRD SEARCH 1994-1996 MICHELLE H. REYNOLDS AND THOMAS J. SNETSINGER Abstract. We compiled the recent history of sightings and searched for 13 rare and missing Hawaiian forest birds to update status and distribution information. We made 23 expeditions between August 1994 and April 1996 on the islands of Hawai'i, Maui, Moloka'i, and Kaua'i totaling 1,685 search hours, 146 field days, and 553 person days. During our surveys we found four critically endangered birds: the Po'ouli (Melamprosops phaeosoma, five to six individu',fis), Maui Nukupu'u (Hemignathus lucidus affinis, one individual), 'I'iwi (Vestiaria coccinea) on Moloka'i (one individual), and the Puaiohi (Myadestes palmerk 55-70 individuals). Detection rates for each species were 0.013, 0.002, 0.012, and 0.318 detections/hr, respectively. Although not visually confirmed during our surveys, auditory detections, unconfirmed sightings, and other reports suggest the possible existence of '0'fi (Psittirostra psittacea) on Hawai'i, Kaua'i Nukupu'u (Hemignathus lucidus hanapepe), and Maui ',kepa (Loxops coccineus ochraceus) in perilously low numbers. Six undetected forest bird popula- tions, Kfima'o (Myadestes myadestinus), Kaua'i ')'6 (Moho braccatus), Bishop's '6'0 (Moho bishopi), '0'fi on Kaua'i, Greater 'Akialoa (Hemignathus ellisianus), and Kfikfiwahie (Paroreomyza fiammea) have high probabilities of being extinct. Oloma'o (Myadestes lanaiensis) from Moloka'i are probably extirpated from the areas searched on that island but may persist on the unsurveyed Oloku'i Plateau. Key Words: bird survey; critically endangered; extinct; Hawai'i; 'I'iwi; Nukupu'u; Po'ouli; Puaiohi. Descending from a small number of original col- onizers, Hawai'i's native plants and animals are an evolutionary panoply. Species underwent ex- plosive adaptive radiation and specialization in the world's most isolated island chain (Carlquist 1974, Scott et al. 1986, Freed et al. 1987a, Ho- warth et al. 1988, James and Olson 1991, Olson and James 1991, Wagner and Funk 1995, Pratt and Pratt this volume). Striking examples of spe- ciation occurred among the lobelioids, fruit flies, land snails, and Hawaiian honeycreepers (Frin- gillidae: Drepanidinae), with more than 50 known species having evolved from one cardue- line finch colonizer (Johnson et al. 1989, Tarr and Fleischer 1995). The isolation that allowed such unique adap- tations also predisposed the ecosystem to vul- nerability due to human caused and stochastic natural disturbances. Multiple pressures have re- sulted in catastrophic species extinctions; habitat destruction and nonnative species introductions, including ungulates, mammalian predators, pathogens, and disease vectors, have had the most extensive and detrimental effects on Ha- wai'i's island ecosystem (Atkinson 1977, Ralph and van Riper 1985, Scott et al. 1986, Loope et al. 1988, Atkinson et al. 1995). Recent fossil ev- idence indicates at least 50% of the original avi- fauna went extinct after the arrival of the Poly- nesians about 400 AD, and today 75% of the historically known native birds are either extinct or endangered (James and Olson 1991, Olson and James 1991, Ehrlich et al. 1992). Coincident with increased human develop- ment and the spread of the Culex mosquito since the 1900s, HawaiTs remaining native avifauna has experienced a steady decline with low-ele- vation and specialized species suffering partic- ularly heavy losses (Baldwin 1953, Warner 1968, Scott and Kepler 1985, van Riper et al. 1986, Pratt 1994). Many species that were abun- dant or common into the early 1900s had low population densities during the extensive U.S. Fish and Wildlife Service (USFWS) Hawaiian Forest Bird Surveys (HFBS) of the 1970s and 1980s (e.g., '(5'fi [Psittirostra psittacea], Maui 'kepa [Loxops coccineus ochraceus], '(5'0 [Moho spp.], Hawaiian Crow [Corvus hawaiien- sis] or 'Alalfi, Moloka'i's Oloma'o [Myadestes lanaiensis rutha], and Kfima'o [Myadestes my- adestinus]; Bryan and Seale 1901, Henshaw 1902a, Perkins 1903, Bryan 1908; Banko 1980a, 1980b, 1981a, 1984a, 1986; Scott et al. 1986). Today the existence of more than half Hawai'i's critically endangered (Mace and Lande 1991) birds is seriously in question (Pratt 1994, USFWS 1996a). The Convention on International Trade in En- dangered Species and the World Conservation Union (WCU 1982) have set 50 years of no sightings as the arbitrary limit to declare species extinction. This may be a useful definition in some cases, but it is hardly appropriate when periodic intensive search effort or surveys by qualified personnel make it possible to evaluate the likelihood of extinction objectively. While most of Hawai'i's endangered endemics are rare, often cryptic species that inhabit remote, rainy, and treacherous terrain where search effort is ir- regular (further complicated by difficulties in gaining access to rare bird habitat on both public and private lands), the periodic survey and in- tensive search methodology initiated in Hawai'i in the 1960s (Richardson and Bowles 1964, Sin- 133