Citation
Speciation and Extinction of Flightless Rails (Aves: Gallirallus) in Oceania

Material Information

Title:
Speciation and Extinction of Flightless Rails (Aves: Gallirallus) in Oceania
Creator:
KIRCHMAN, JEREMY J.
Copyright Date:
2008

Subjects

Subjects / Keywords:
Archipelagos ( jstor )
Birds ( jstor )
Bones ( jstor )
DNA ( jstor )
Femur ( jstor )
Fossils ( jstor )
Humans ( jstor )
Museums ( jstor )
Species ( jstor )
Toes ( jstor )
Florida Museum of Natural History ( local )

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
Copyright Jeremy J. Kirchman. Permission granted to the University of Florida to digitize, archive and distribute this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
Embargo Date:
2/28/2007
Resource Identifier:
649814480 ( OCLC )

Downloads

This item is only available as the following downloads:


Full Text

PAGE 1

SPECIATION AND EXTI NCTION OF FLIGHTLESS RAILS (AVES: Gallirallus ) IN OCEANIA By JEREMY J. KIRCHMAN A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2006

PAGE 2

Copyright 2006 by Jeremy J. Kirchman

PAGE 3

Dedicated to the many ornithol ogists and archaeologists who collected the specimens that are the basis of this research.

PAGE 4

iv ACKNOWLEDGMENTS I wish to thank my supervisory committee members (Pamela Soltis, Gustav Paulay, Brian McNab, Rebecca Kimball; and chair David Steadman). Their advice and encouragement were invaluable at every stag e. I am indebted to P. Soltis and D. Steadman for logistical and financial support of the ancient DNA lab; to R. Kimball for use of her DNA sequencing lab and for critical discussions regarding lab work; to Andy Kratter, Chris Filardi, Sto rrs Olson, Helen James, and Robert Fleischer for additional advice and logistical support. I thank Rebecca Kimball, James Frankli n, Andrew Cox, Ben Burkley, and Padi Tester for their kind assist ance with DNA data collection; and Rebecca Kimball, Ed Braun, and David Reed for guidance with DNA data analyses. I thank Donna Kalfatak and Ernest Bani of the Republic of Vanuatu Environment Unit for permits to conduct research in Vanuatu; and Andy Kratter, Mary Hart, David Steadman, Roy Palo, Saul Loi, Saki Naisak, Numa Fred, a nd Ralph Regenvanu for field assistance in Vanuatu. Tissue samples were kindly provided by Sharon Birks at the Burke Museum of Natural History, University of Washington; Andy Kratter and David Steadman at the Florida Museum of Natural Hist ory, University of Florida; James Dean and Storrs Olson at the National Museum of Natural History, Smithsonian Institution; David Willard, Shannon Hackett, and John Bates at the Field Mu seum of Natural History; Joel Cracraft, Chris Filardi, and Paul Sweet at the Amer ican Museum of Natural History; Carla Kishinami at the Bernice P. Bishop Museum; and Jeremiah Trimbal and Scott Edwards at

PAGE 5

v the Museum of Comparative Zoology, Harv ard University (MCZ). For access to specimens I thank Tom Webber, Andy Kratter, and David Steadman (UF); Janet Hinshaw University of Michigan Museum of Zool ogy (UMMZ); James Dean and Storrs Olson (USNM); David Willard, Shannon Hackett, and John Bates (FMNH); Paul Sweet and Joel Cracraft (AMNH); Sievert Rohwer and Chris Wood (UWBM); Kristof Zyskowski Yale Peabody Museum (YPM); and Jeremiah Trimbal and Scott Edwards (MCZ). Financial support for this research came fr om the University of Florida Graduate Student Council Mentorship Opportunity Fund, the Frank Chapman Fund of the American Museum of Natural History, a Grinter Fellowship and a McLaughlin Fellowship from the University of Florida College of Liberal Arts and Sciences, a Graduate Fellowship from the Smithsonian Institution, a Grant-in-Aid research award from Sigma Xi, an Alexander Wetmore Award from the American Or nithologistsÂ’ Union, and from Teaching Assistantships from the Un iversity of Florida Department of Zoology. Finally, I am most grateful to my wife Laurie for her kindness and support and for logistical assistance of every kind.

PAGE 6

vi TABLE OF CONTENTS page ACKNOWLEDGMENTS.................................................................................................iv LIST OF TABLES...........................................................................................................viii LIST OF FIGURES.............................................................................................................x ABSTRACT.....................................................................................................................xi ii CHAPTER 1 EVOLUTION OF FLIGHTLESS RAILS ON PACIFIC ISLANDS...........................1 2 HISTORICAL BIOGEOGRAPHY OF FLIGHTLESS RAILS ( GALLIRALLUS ) IN OCEANIA: A SURVEY OF THE ZOOARCHAEOLOGICAL AND PALEONTOLOGICAL RECORD..............................................................................8 Introduction................................................................................................................... 8 Materials and Methods...............................................................................................11 Results........................................................................................................................ .14 Comparative Osteology and Systematics............................................................14 Order Gruiformes, family Rallidae..............................................................14 Genus Gallirallus Lafresnaye, 1841............................................................15 Geographical survey of Gallirallus fossils in Oceania........................................16 Discussion...................................................................................................................41 3 PARALLEL EVOLUTION OF FLI GHTLESSNESS IN RAILS FROM OCEANIA: PHYLOGENY AND SYSTEMATICS OF GALLIRALLUS...............87 Introduction.................................................................................................................87 Methods......................................................................................................................91 Results........................................................................................................................ .95 Discussion...................................................................................................................99 4 PHYLOGENETIC AND DEMOGRAP HIC TESTS OF RAPID PARALLEL SPECIATION OF FLIGHTLESS RAIL S (AVES: RALLIDAE) FROM A VOLANT ANCESTOR, GALLIRALLUS PHILIPPENSIS ......................................114 Introduction...............................................................................................................114

PAGE 7

vii Methods....................................................................................................................118 Results.......................................................................................................................1 23 Discussion.................................................................................................................127 LIST OF REFERENCES.................................................................................................141 BIOGRAPHICAL SKETCH...........................................................................................152

PAGE 8

viii LIST OF TABLES Table page 2-1 Prehistoric sites in O ceania containing remains of Gallirallus . .............................47 2-2 Skeletal measurements (in mm) of † Gallirallus ernstmayri and extant, flightless congeners from Near Oceania. ...............................................................................48 2-3 Skeletal measurements (in mm) of extinct (†) Gallirallus species from the Mariana Islands. ......................................................................................................49 2-4 Correlation coefficients from a principal components analysis of † Gallirallus pendiculentus and eight Gallirallus species (see Figure 2-8)..................................53 2-5 Skeletal measurements (in mm) of † Gallirallu vekamatolu, † G. storrsolsoni , and seven congners. ......................................................................................................54 2-6 Correlation coefficients from a principal components analysis of † Gallirallus vekamatolu and seven congeners (see Figure 2-11).................................................57 2-7 Principal components analysis correlation coefficients of † Gallirallus storrsolsoni and seven congeneric sp ecies (see Figure 2-17)..................................57 2-8 Skeletal measurements (in mm) in † Gallirallus roletti and select congeners..........58 2-9 Skeletal measurements (in mm) of † Gallirallus gracilitibia and selected congeners..................................................................................................................60 2-10 Skeletal measurements (in mm) of † Gallirallus epulare and selected congeners...61 3-1 Specimen information and DNA sequence lengths for all control region (CR), cytochrome b (Cyt b ), and 12S rDNA (12S) sequences........................................105 3-2 Fossil bone specimens of prehistorically extinct Gallirallus species sampled for ancient DNA. .........................................................................................................107 3-3 Primers used in PCR and sequencin g of samples in this study. ...........................108 3-4 Mean percent sequence diverge between Gallirallus species for cyt b (above diagonal) and combined cyt b , CR, and 12S (below diagonal)..............................109

PAGE 9

ix 3-5 Summary of all known species c onsidered members of the genus Gallirallus. .....110 4-1 Voucher specimen information for all individuals from which control region sequences were obtained. .....................................................................................132

PAGE 10

x LIST OF FIGURES Figure page 1-1 Map of Oceania, showing all major arch ipelagoes and the line separating Near Oceania and Remote Oceania....................................................................................7 2-1 Map of Western Oceania, indicating islands with archaeological sites containing Gallirallus fossils.....................................................................................................62 2-2 Humeri of Gallirallus species known from Near Oceania in anterior (upper) and posterior (lower) aspects..........................................................................................63 2-3 Distal tibiotarsi of of Gallirallus species known from Near Oceania in anterior (upper) and posterior (l ower) aspects. ....................................................................64 2-4 The coracoids (upper row) and carpometacarpi (lower row) of Gallirallus rails from the Mariana Islands. .......................................................................................65 2-5 The sternum of Gallirallus in lateral (left) and ve ntral (right) aspects. .................66 2-6 Wing elements of Gallirallus. ..................................................................................67 2-7 Size variation in femora (upper) and tarsometatarsi (lower) of † Gallirallus pendiculentus, Tinian, G. owstoni, Guam, and G. philippensis, Tutuila..................68 2-8 Plot of † Gallirallus pendiculentus and eight other species of Gallirallus on the first two principal components of postcranial skeletal measurements. .................69 2-9 Tonga, showing islands with archae ological sites contai ning fossil bones of Gallirallus ................................................................................................................69 2-10 Gallirallus tarsometatarsi in acrotarsial ( upper) and plantar (lower) aspects..........70 2-11 The rostrum (A, B), humerus (C, D), ul na (E, F), carpometacarpus (G, H), femur (I, J), and tibiotarsus (K, L) of † Gallirallus vekamatolu, ‘Eua................................71 2-12 Plot of † G. vekamatolu and seven other species of Gallirallus on the first two principal components of 14 post-cranial sk eletal characters (see Table 2-6). ........72 2-13 Society Islands with inset of Huahin e, showing the location of the Fa’ahia archaeological site....................................................................................................73

PAGE 11

xi 2-14 Lateral (left) and dorsal (r ight) views of the skulls of Gallirallus. ........................74 2-15 Ventral (upper) and lateral (l ower) views of the sterna of Gallirallus . ..................75 2-16 Humeri (A-C), ulnae (D-F), carpomet acarpi (G-I), tibiotarsi (J-L), and tarsometatarsi (M-O) of Gallirallus. ........................................................................76 2-17 Plot of mean scores for eight species of Gallirallus on the first two principal components of 16 post-cranial skeletal measurements(see Table 2-7). .................77 2-18 The Marquesas Islands, showing the lo cation of the four archaeological sites discussed in this chapter...........................................................................................78 2-19 A. † Gallirallus roletti holotype with the same elements from B. G. philippensis (UF 39855, Tutuila) and C. G. owstoni (UF 39921, Guam). .................................79 2-20 Rostra of † Gallirallus roletti (BPBM 166456, Tahuata) in lateral (left) and dorsal (right) aspects. .............................................................................................80 2-21 Femora of A. † Gallirallus roletti (holotype BPBM 166447, Tahuata) in dorsal (top) and ventral (bottom) aspects............................................................................81 2-22 Tibiotarsi of A. † Gallirallus roletti (holotype BPBM 166446, Tahuata) in dorsal (top) and ventral (bottom) aspects. .........................................................................82 2-23 Tarsometatarsi of A. † Gallirallus roletti (holotype BPBM 166448, Tahuata,) in acrotarsial (top) and planta r (bottom) aspects. .......................................................83 2-24 Scapulae of A. † Gallirallus undescribed sp., B. G. philippensis , and C. G. owstoni , in dorsal aspect. ........................................................................................84 2-25 Humeri of A. † Gallirallus storrsolsoni , B. † G. epulare , C. † G. gracilitibia , D. G. philippensis , and E. G. owstoni ...........................................................................85 2-26 Ulnae of A. † Gallirallus storrsolsoni , B. † G. epulare , C. G. philippensis , and D. G. owstoni in dorsal aspect.......................................................................................86 3-1 Maximum likelihood phylogeny of comb ined 12S, cytochrome b, and control region data..............................................................................................................112 3-2 Majority-rule consensus tree of 10,00 0 maximum parsimony trees (length 399 steps).......................................................................................................................11 3 4-1 Oceania, showing the distribution of Gallirallus philippensis (dotted outline). Arrows point to islands from which G. philippensis was sampled. .....................136 4-2 The Gallirallus mitochondrial control region, showing the location of PCR primers used to amplify a 325 bp portion of Domain I..........................................137

PAGE 12

xii 4-3 Maximally parsimonious network (u nrooted) of 38 sampled (ovals) and 21 unsampled (circles) control region haplotypes.......................................................138 4-4 Frequency distributions of the numbe r of pairwise nucleotide substitutions between A) all 71 G. philippensis , and B) 61 sampled outside of the Philippines.139 4-5 Maximum likelihood phylogeny of 38 Gallirallus philippensis haplotypes and individuals from ten other Gallirallus species.......................................................140

PAGE 13

xiii Abstract of Dissertation Pres ented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy SPECIATION AND EXTI NCTION OF FLIGHTLESS RAILS (AVES: Gallirallus ) IN OCEANIA By Jeremy J. Kirchman August 2006 Chair: David W. Steadman Major Department: Zoology The evolution of flightless birds on predat or-free islands has lo ng been regarded as a compelling example of adaptation by natura l selection. Flightlessness has evolved independently in at least 15 predominantly volant avian families, and bones from prehistoric sites on Pacific is lands reveal the recent extinc tion of hundreds or perhaps thousands of flightless rail (F amily Rallidae) species after colonization by humans and their commensals. Comparative studies implicate selection for reduced energy expenditure as the force drivi ng the evolution of insular fli ghtlessness, and suggest that flightless rails endemic to single islands probably evolve d rapidly from widespread, volant colonists. My study presents or iginal data from paleontology, molecular systematics, and population genetics to infer the geographical and temporal patterns of speciation and extinction in the genus Gallirallus , a diverse lineage of volant and flightless species. I examined 1,789 pr eviously undescribed fossil bones from archaeological sites on 19 islands in Oceania , referring 1,487 of these specimens to nine

PAGE 14

xiv newly described flightless Gallirallus species. I also id entified and discussed osteological characters that define the family Rallidae, Gallirallus , and all new species. To address the question of how flightless species are rela ted to each other and to widespread volant species, I re constructed the phylogeny of Gallirallus species from sequences of the mitochondrial control region, cytochrome b , and 12S genes obtained from modern and ancient sources. Flightless species appear to have evolved rapidly and repeatedly throughout the Pleistocene in a series of independent colonizations of archipeligoes by G. philippensis and other, older ancestors. Taxonomic implications are discussed. To address the mechanistic question regarding how (and how fast) populations of volant colonists become genetically structured , I examined control region variation among 71 individuals of the ancestral volant species Gallirallus philippensis collected from throughout its range. Popula tions from the Philippines appear to be genetically distinct and may be ancestr al to populations that expanded throughout Oceania only in the last 30,000 to 75,000 years, perhaps facilitated by human alteration of island environments.

PAGE 15

1 CHAPTER 1 EVOLUTION OF FLIGHTLESS RAILS ON PACIFIC ISLANDS The evolution of flightless species in pr imarily volant groups of animals such as birds, bats (hypothetically), and insects has at tracted the attention of evolutionists since Darwin (1859), who devoted much thought to the subject, considering it to be a particularly compelling example of adaptati on by natural selection. He argued that secondarily flightless species evolve in response to the ec onomizing power of selection: “If under changed conditions of life a structur e before useful becomes less useful, any dimunition, however slight, in its development, will be seized on by natural selection, for it will profit the individual not to have its nutriment wast ed in building up an useless structure.” (Darwin 1859, p. 186). Thus, the econo mics of nature may lead to the loss of energetically demanding flight capability in volant organisms, but only if loss or reduction of flight does not grea tly increase the cost of obtai ning food or greatly increase mortality via predation. The complete relian ce of bats on flight for foraging (and their inept terrestrial locomotion) may explain w hy there are no known flightless bats. By contrast, many species of insects and bi rds have become flightless, and many comparative and experimental studies of th ese cases have supported Darwin’s hypothesis of “nutriment” economy. Experiments in insects show that wingless morphs are substantially more fecund than volant morphs, devoting energy “saved” on flight to reproduction (Roff 1994). The importance of energy conservation in the evolution of flightless birds was strongly supported by McNab (1994a, 2002) and McNab and Ellis

PAGE 16

2 (2006) who showed decreased basal metabolic rates and decreased pectoral muscle mass in flightless versus volant birds. Whereas the ultimate cause of flightlessness is the same in birds and insects, the two groups differ with respect to proximat e causes related to habitat and geographic distribution. Roff (1990) showed that flightlessness in insect s is correlated with habitat stability, mating system, and mode of metamo rphosis, but not with insularity. The proximate causes are less complicated in birds, and can be explained by the high incidence of flightlessness in aquatic species and especially in insular species (Roff 1990). These circumstances highlight the importance of predator avoidance in maintaining flight structures in birds. Th e flightless condition has evolved independently in at least 15 predominantly volant avian fa milies (not including penguins and ratites). A few aquatic species have become flightless on continents (three gr ebes, one duck), but nearly all flightless birds evol ved on predator-free islands. Interestingly, flightless birds have evolved in sympatry with marsupial pr edators (on Tasmania), but never on islands with placental carnivores. Flight ability in volant island colonists may be reduced by selection for decreased dispersal as well as for energy conservation (Olson 1973a). Although reduced energy expenditure is known to be an important su rvivorship strategy in many island vertebrates (McNab 1994b) , the role of reduced dispersal per se is more speculative. Birds are particularly sensitive to selection for energy conservation because of their large anatomical and metabolic i nvestment in the energy-demanding muscles of the flight apparatus, which comprise as much as 25% of total body mass (Gill 1995). Insular flightlessness is especially prevalent in members of the avian family Rallidae, in which 31 of the approximately 150 historically known species are or were

PAGE 17

3 flightless species endemic to si ngle islands, or to sets of is lands that were connected in the Pleistocene during periods of lowered sea-le vels (Taylor 1998). In addition, scores of prehistorically extinct fligh tless species are known from fossil bones from archaeological and paleontological sites on Earth’s o ceanic islands (Steadman 1995, 2006, Olson and James 1991, Olson 1973a, Worthy and Holdaway 2002). This record reveals extensive Pleistocene and Holocene extinction of fligh tless rails after coloni zation by humans and their commensals (primarily dogs, pigs, and rats). Losses were greatest in Oceania (Figure 1-1), where flightless rail species ma y have numbered in the “hundreds” (Livezey 2003) or perhaps as many as 1600 (Steadman 2006). No island in Oceania except for New Caledonia and New Zealand ever was connected to continental land, so each flightless species of ra il probably had an independent tran sition to flightlessness. Thus, the flightless rails of Oceania are the larg est and yet least studied example of adaptive radiation in island birds, and may also be the most extensive example of convergent evolution among vertebrates (McN ab and Ellis 2006). Their mass extinction in the last few thousand years makes Darwin’s statemen t above regarding “changed conditions of life” doubly true, as a once-adaptive survival strategy became a maladaptive liability when the people showed up. Why did so many species of rails, in part icular, evolve the flightless condition? Rails appear to possess a suite of mo rphological, ecological, behavioral and developmental characteristics that predispose them to flightlessness. Even volant rails are characterized by relatively small pector al muscle masses, which average only 12.9% of total body mass (Livezey 2003) . All rails nest and forage on the ground and rarely fly even to escape potential predators, fro m which they typically run for cover.

PAGE 18

4 Paradoxically, these weak, reluctant flyers are great dispersers, as evidenced by the frequency with which they turn up as vagran ts outside of their regular ranges (Remsen and Parker 1990), and by their occurrence on even the most isolated of oceanic islands. Many temperate species are long-distance mi grants. In developing rails, the sternum does not ossify until well after hatching, and a ll rails are effectively flightless as chicks and as nearly adult-sized birds. Insular flightlessness is thought to result from heterochronic (more specifically neotenic) dela y in the development of pectoral flight apparatus (Olson 1973a, Feduccia 2000), and so may be under the control of one or a very few regulatory genes. Biogeographi c evidence also indicates that insular flightlessness may evolve rapidly in rail s. Flightless species closely resemble widespread, polytypic, volant species, and flightless species are thought to derive from populations of still extant volant congeners. These facts led Olson (1973a) to propose a model for the evolution of Oceanic rail divers ity in which multiple closely related species may descend from one or a few (perhaps still extant) ancestral species as natural selection rapidly pushes one island-colonizing populati on after another down parallel paths to flightlessness. Phylogenetic tests of OlsonÂ’s rapid parallel speciation m odel of rail evolution have been difficult because there is extensiv e homoplasy due to convergent evolution of flightlessness (in the case of osteological data) and most species are extinct (in the case of genetic data). Limited support has come from molecular phylogenetic studies of mitochondrial DNA in species of crakes ( Porzana, Amaurornis, Poliolimnas, Slikas et al. 2002), and rails from New Zealand (Trewick 199 7). Both studies found that single-island endemic flightless species were derived from within paraphyletic widespread, volant,

PAGE 19

5 polytypic species. Slikas et al. (2002) also showed that flightless species occupy short branches on the phylogeny, indicating recent di vergence from volant ancestors. Trewick (1997) showed that flightless swamphen s from New Zealand’s North Island ( Porphyrio mantelli ) and South Island ( P. hochstetteri ) evolved the f lightless condition independently from different co lonizations by the widespread P. porphyrio . Phylogenetic studies that include flightless and volant species of rails support the hypothesis that flightlessness can evolve ra pidly and repeatedly, but much remains unanswered. Most of the fossils on which estim ates of rail alpha-dive rsity are based have never been described, precluding even an understanding of the taxonomic and geographic scope of the flightless rail radiation. No genetic study ha s sampled exhaustively within putative ancestral species to examine their genetic architecture or the temporal and geographic context of speciati on events. Further, previous phylogenetic studies have focused on the isolated archipelagos of Ha waii (Slikas et al. 2002) and New Zealand (Trewick 1997). The goal of my study was to combine orig inal data from paleontology, molecular systematics, and population genetics to infer the patterns of specia tion and extinction in one rail lineage that has radiated exte nsively in Oceania, and thereby obtain a comprehensive understanding of the tem po and mode of evolution of insular flightlessness. My study focused on the “typical long-billed rails” of the genus Gallirallus sensu lato . This lineage provides an excel lent opportunity for such an integrative study because it includes multiple fl ightless species endemic to single islands in Oceania, two widespread volant species, either of which may have given rise to flightless species (O lson 1973b), and undescribed fossil bones from archaeological sites

PAGE 20

6 throughout Oceania. I addressed th ree major questions: 1) How many Gallirallus species were there and where did they live? I a ddressed this question by surveying the many hundreds of undescribed fossils attributed to Gallirallus from prehistoric sites on 19 islands in 8 major archipelagos in Oceania. 2) How are flightless sp ecies related to each other and to the widespread vol ant species, and when did they diverge? I addressed this question by reconstructing the phylogeny of Gallirallus species from modern and ancient DNA sequence data and estimating divergence ti mes from a locally calibrated molecular clock. 3) How are populations of volant coloni sts genetically struct ured, and how fast do they become isolated? I addr essed this question by survey ing the population genetics of the ancestral volant species Gallirallus philippensis. Questions 1 and 2 concern patterns of extinction and speciation in Gallirallus, whereas question 3 con cerns the process of speciation. My analyses provide a more comprehensive understanding of the temporal and geographic patterns of e volution in an adaptive radi ation for which the natural selection regime is well characterized.

PAGE 21

7 Figure 1-1. Map of Oceania, showing all major archipelagoes a nd the line separating Near Oceania and Remote Oceania.

PAGE 22

8 CHAPTER 2 HISTORICAL BIOGEOGRAPHY OF FLIGHTLESS RAILS ( Gallirallus ) IN OCEANIA: A SURVEY OF THE ZOOARCHAEOLOGICAL AND PALEONTOLOGICAL RECORD Introduction It is well known that island species are especially vulnerable to extinction and that the current anthropogenic mass extinction event has disproportionately affected island ecosystems. Indeed, nearly all of the 75 sp ecies of birds known to have become extinct since 1600 lived on islands (Fuller 1987). Less well-known are the hundreds or perhaps thousands of bird extinctions that occurred prehistorically after human colonization of EarthÂ’s islands. The fossil record of bones from archaeologi cal (cultural) and paleontological (non-cultural) sites documents extensive prehistoric bird extinctions on oceanic islands throughout the world. These lo sses were greatest in the Pacific, where the fossil record is rich, and where exploi tation of native birds by humans and their commensals had catastrophic consequences fo r indigenous bird communities (Steadman 1995, 2006, James and Olson 1991, Olson and James 1991, Worthy and Holdaway 2002). The colonization of Oceania (Figure 1-1) was one of the last chapters in the history of human global expansion. The chronology of colonization, as reconstructed mainly from archaeological evidence, is thought to have consis ted of two major waves of population expansion into regions now called Near Oceania and Remote Oceania (Kirch 2000). The first expansion began at leas t 40,000 radiocarbon years before present (yr BP) with the first purposive human voyages out of Southeast Asia across the deep-water gap separating the Sunda shelf from Sahul , the landmass that included Australia,

PAGE 23

9 Tasmania, and New Guinea. The voyaging te chnology that enabled co lonization of Sahul also enabled the rapid subse quent colonization of the arch ipelagoes of Near Oceania, namely the Bismarck Archipelago and the So lomon Islands. Early occupation sites in Near Oceania, especially on New Ireland and B uka, indicate that this entire region was colonized by ca. 35,000 yr BP (Allen et al. 1988, Wickler and Spriggs 1988). The second expansion began ca. 3500 yr BP when the pot tery-making, Austronesian-speaking people known as the Lapita cultural complex (named for the Lapita Site in New Caledonia where their pottery was first discovered) undert ook long-distance voya ges within and beyond Near Oceania to the archipelagoes of Remo te Oceania. This wave of colonization, probably enabled by improved sailing technolo gy and a portable agricultural tool kit, reached New Caledonia and Tonga within a few centuries, and came to include nearly every inhabitable island in Oceania by ca. 1000 yr BP. Although the severity of ecological distur bance by human colonists differed from island to island, human exploitation of native avifaunas is evident throughout the Oceanic zooarchaeological record (Steadman 2006). At many sites, burned and broken bones from kitchen middens, especially of la rge-bodied species, suggest direct human exploitation of native birds for food. Conversion of forested habitats to agricultural use, and introduction of rats, pigs, and dogs also surely were devastating to native bird populations. Avian species-level extinction was least severe on the large, rugged islands of Near Oceania where the vulnerability of bird populations may have been buffered by the presence of indigenous rodents and suppression of human populations by malaria (Steadman and Kirch 1998). Conversely, on most islands in Remote Oceania, 50 to 100% of the species of landbirds that existed at human contact do not survive there today

PAGE 24

10 (Steadman 1995). All ground-nesting birds were affected, as were many species hunted for food including pigeons, parrots, and mega podes, but especially prevalent among the extinct species are rails (Order Gruiformes , Family Rallidae), most of which were flightless species endemic to single islands, or to multiple islands connected in the Pleistocene during periods of lowered sea-levels. Flightless rail extinction – The fossil record of rails in Oceania reveals the largest (in terms of species) and yet least-understood example of adaptive radiation in island birds. Nearly every island in the tropical Pacific that has a decent prehistoric record of birds has yielded 1 to 4 endemic species of flightless rail. Crude estimates (based on extrapolation from the known fossil record) place the number of flightless rails that once existed across the Pacific in the “hundr eds” (Pimm et al. 1994, Livezey 2003), 500 to 1600 (Steadman 2006) to nearly 2000 species (Steadman 1995). Despite the high estimates of prehistoric flightless rail di versity, fewer than 20 species have been described (Worthy 2004). Nearly all of the na med extinct (†) species of flightless rails have been placed in widespread, extant genera, including 7 to 10 species of Porzana from the Hawaiian Islands alone (Olson and Jame s 1991); one to several species each of Gallirallus , Gallinula , Porphyrio, and Fulica from New Zealand (summarized in Worthy and Holdaway 2002); and many species of Porzana , Gallirallus , and Porphyrio from the rest of Oceania (Steadman 1987, 1988, 2006, Steadman et al. 2000). Exceptions include two long-billed, flightless sp ecies in monotypic genera: † Vitirallus watlingi , a “ Gallirallus -like” rail described from Late Plei stocene and Holocene deposits on Fiji (Worthy 2004); and † Capellirallus karamu from Holocene cave deposits on North Island,

PAGE 25

11 New Zealand (Falla 1954). Undoubtedly, most of the flightless rail spec ies that existed in Oceania at the time of human contact remain undescribed. Whereas most species disappeared without leaving a trace, fossil remains of perhaps dozens of still undesc ribed rail species have been recovered from early human occupation sites on most major archipelagoes in Oceania. These fossils, identified to genus by D. W. Steadman and housed in the Division of Ornithology, Florida Museum of Natural History, document the ex istence of many more species of possibly flightless rails in Oceania prior to the era of European exploration and col onization. In this chapter, I interpret the species-level diversity of specimens assigned to the genus Gallirallus , the “typical rails,” and establish the alpha taxonom y of this radiation. This descriptive research will begin to answer the historic al biogeographic question, How many species of Gallirallus existed prior to human contact, and where did they live? In this dissertation I do not propose any new species names. Newl y named taxa have been described in a series of papers published in peer-reviewed journals in 2005 and 2006 so that new names based on this research will have proper taxonomic priority. Materials and Methods Undescribed fossil rail specimens from archaeological sites on 19 islands in 7 major archipelagoes (Table 2-1) referred to as “ Gallirallus new sp.” in faunal summaries by D. W. Steadman and colleagues (Steadman 1993, 1995, 1999, 2006, Steadman and Pahlavan 1992, Steadman and Rolett 1996, Steadman et al. 1999) were examined in the Division of Ornithology, Florida Museum of Natural History, University of Florida, Gainesville, FL. All uncataloged fossils we re assigned UF catalog numbers. Specimens from four islands in the Marquesas belong to the Bernice P. Bishop Museum, Honolulu, and were examined while on loan to D. W. Steadman.

PAGE 26

12 Measurements were taken with electronic digital calipers, rounded to the nearest 0.1 mm. To assess the degree of flightlessness in fossil species for which elements from both pectoral and pelvic regi ons are known, I performed a series of principal component analyses (PCA) using the software packag e SPSS (v. 13.0). Unweighted character means (natural [base e ] log-transformed) for each species were used for PCA. The characters in each analysis were chosen on the basis of availability in fossil specimens. I used PCA to separate flightless and volant taxa graphically rather than employing a statistical classification technique such as descriminant function analysis because the data were inappropriate for this method in three importa nt ways: 1) Groups (i.e., “volant” versus “flightless”) contained specimens from multip le species and thus lacked phylogenetic independence; 2) Box’s M test s (results not shown) indica ted that groups (volant vs. flightless) had unequal covarian ce matrices; 3) sample sizes were small and were unequal among taxa. Osteological terminology follows Baumel and Witmer (1993). Radiocarbon dates are expressed as radiocarbon years befo re present (yr BP), or as calenderically calibrated years before present (cal BP). Skeletons used for comparison with fossils were from the American Museum of Natural History (AMNH), Florida Museum of Natural History, University of Florida (UF), National Museum of Natural Histor y, Smithsonian Institution (USNM), National Museum of New Zealand (NMNZ), Départment de Archéologie, Cent re Polynésien des Sciences Humaines, Tahiti (DAPT), Univer sity of Michigan Museum of Zoology (UMMZ), Thomas Burke Memorial Museum, University of Washington (UWBM), and Yale Peabody Museum (YPM). I examined these modern specimens (following Taylor 1998 for subspecies-level taxonomy of polytypic Gallirallus and Porphyrio species):

PAGE 27

13 Porzana tabuensis UWBM 42501, 42528; Porzana [ Poliolimnas ] cinerea DMNH 72836, 72906; Rallus longirostris UF 40956, 24200; Rallus striatus USNM 85892, 343214, 559919, YPM 107205; Gallirallus insignis AMNH 27136; G. t. torquatus UMMZ 228275, 228279, 228280, AMNH 17715-17717, USNM 290445; G. owstoni UF 3991839921, 39256, 42967-42969, 44363, 44364, 44367-44377, UMMZ 215472, USNM 561968, 611816, 612616, 613738-614744, 614233-614235, 614771, 614772; G. australis greyi UF 24326, 24327, G. a. australis YPM 102249, 110760, 110789, 110790, 110844; G. philippensis goodsoni UF 39854, 39855, G. p. sethsmithi UF 42878, 42902, 42933 42935, 43180, 43180, 43182, 43222, 43223, 43176-43178, G. p. philippensis USNM 560651, UMMZ 233050, G. p. yorki USNM 560791, G. p. mellori USNM 620196, G. p. pelewensis YPM 103082, 103087, G. p. ecaudatus UWBM 42863, 42865, 42866; G. rovianae AMNH cef878; G. [“ Nesoclopeus” ] woodfordi immaculatus UF 39399, 39406, 39409, 39547, 39556, 39574; Amaurornis olivaceus UF 40216; A. phoenicurus UF 24387; Porphyrio porphyrio samoensis UF 39332, 39388, 39407, USNM 561547, 561549, 541551, P.p. poliocephalus USNM 34212, P.p. pulverulentus USNM 226035, 292296, 292297; P. martinicus UF 39927, 42418, 42419, Gallinula chloropus UF 39927. In addition to the previously undescribed specimens that are th e subject of this chapter, I also examined fossil specimens of these extinct species: † Gallirallus ripleyi rostrum UF 55596, coracoids UF 54711, USNM 402896 (paratype), humeri UF 51402, 55752, ulnae UF 54901, 55215, carpometacarpi UF54700, 54988, femur UF 51320, tibiotarsi UF 55283, 59574, USNM 402895 (holotype), ta rsometatarsi UF 54761, 55223, USNM 402895 (holotype); † Porphyrio paepae femora BPBM 165649, 166434, tibiotarsus BPBM 165651.

PAGE 28

14 Results Of 1789 fossil specimens of Gallirallus examined, 1487 were assigned to nine newly described species, 306 to extant species (288 to G. philippensis, 7 to G. woodfordi , 11 to G. rovianae ), and 4 to cf. Gallirallus sp. In this section I describe the osteological characters used to refer fossil specimens to the Family Rallidae and to the Genus Gallirallus . This is followed by the alpha taxonomy of fossil Gallirallus from each archipelago. Comparative Osteology and Systematics Order Gruiformes, family Rallidae The phylogenetic relationships among genera in the Rallidae are not well resolved (Ripley 1977, Olson 1973b, Livezey 1998) and ther e is little consensus regarding the composition of rail genera (T aylor 1998). As a starti ng point for generic level classification, I have regarded all “typical long-billed rails” from O ceania as species of Gallirallus sensu lato , distinguishing them from the ot her similarly sized rallids called swamphens ( Porphyrio ), moorhens ( Gallinula ), and coots ( Fulica ), and from the much smaller rallids called crakes ( Porzana, Amaurornis and Poliolimnas ) that are also found in Oceania. Published descriptions of the long-billed species Vitirallus watlingi (Worthy 2004) and Capelirallus karamu (Worthy and Holdaway 2002) neither which I have examined, convince me that these species ar e well placed in monotypic genera apart from Gallirallus . My treatment of Gallirallus departs slightly from th e classifications of Olson (1973b) and Taylor (1998), both of whom retained the genus Nesoclopeus for woodfordi of the Solomon Islands and † poecilopterus of Fiji, although stressi ng the close affinity of these species with Gallirallus . Taylor (1998) departed fr om Olson (1973b) by placing pectoralis of New Guinea and Australia in Lewinia . Livezey (1998, 2003) also

PAGE 29

15 acknowledged the difficulty of establishing ge neric-level relationshi ps in this group on the basis of osteology but advocated dividi ng the typical rails of Oceania among the genera Gallirallus , Nesoclopeus, Habropteryx and Tricholimnas, and the monotypic Cabalus . As defined herein, Gallirallus s.l . comprises 19 previously named species including eight extant flightless species ( G. australis of New Zealand, G. sylvestris of Lord Howe Island, G. owstoni of Guam, G. okinawae of Okinawa, G. insignis of New Britain, G. woodfordi of the Solomon Islands [minim ally Bougainville, Isabel, and Guadalcanal], G. rovianae of New Georgia, Solomon Islands, and G. calayanensis of Calayan, Philippines), five historically extinct flightless species († G. wakensis of Wake Island, † G. lafresnayanus of New Caledonia, †G. modestus of the Chatham Islands, New Zealand, †G. poecilopterus of Viti Levu, Fiji, and † G. pacificus of Tahiti, Society Islands), two previously named pr ehistoric (i.e., fossil) species († G. huiatua of Niue, and † G. ripleyi of Mangaia, Cook Islands) and four extant, volant species G. philippensis , G. torquatus , G. striatus , and G. pectoralis. The flightless species G. calayanensis was discovered living on the island of Calayan, Ph ilippines, in 2004 (Allen et al. 2004). Genus Gallirallus Lafresnaye, 1841 I refer fossils from Oceania to Gallirallus rather than to other genera of Oceanic rails on the basis of the following characters. Skull: frontals narrow, concave. Rostrum: long, narrow, and shallow with elongate nare s. Mandible: cotyla lateralis narrow and concave, fossa for condylus medialis quadratum (the main articulation surface in the os articulare) shallow and wide. Scapula: facies articularis clavicularis relatively small and oriented at a more obtuse angl e from corpus scapulae. Coracoid: acrocoracoid extends medially over the sulcus muscul o supracoracoidei such that the foramen triosseum is less open cranially. Humerus: crista pectoralis thicker proximally in pr oximal aspect; corpus

PAGE 30

16 humeri in dorsal aspect uniformly thick, rather than thicker proximally; tuberculum supracondylare ventrale low, making the condyl us ventralis more contiguous with the epicondylus ventralis; fossa pneumotricipitali s deep and wide with prominent crus ventrale fossae. Ulna: thin in cranial aspect with rectangu lar (rather than rounded) margo cranialis. Pelvis: ala preacet abularis ilii broadly continuous with crista dorsalis of synsacrum. Femur: distal end of corpus femoris becomes gradually wider; condylus medialis subcircular in medial aspect; impr esso ansae musculo iliofibularis abuts sulcus fibularis; rotular groove broad. Tibiotarsus: cranio-lateral and cranio-medial margins of corpus tibiotarsi rounded rath er than sharp; impresso lig amentum collateralis medialis deep and wide; facies articularis femoris la rge; depressio epicondylaris lateralis deep; condylus medialis subcircular in medial aspect . Tarsometatarsus: corpus tarsometatarsi much wider than deep; medial sulcus hypot arsi not enclosed; fossa parahypotarsalis medialis shallow in proximal aspect; fossa me tatarsi I short and shallow; crista plantaris mediana slopes gradually (not steeply) to hypotarsi s; distal end of troc hlea metatarsi tertii sloped toward medial trochlea; cotyla medialis is rectangular in proximal aspect with flat (not rounded) dorsal margin. Pe dal phalanges are referred to Gallirallus on the basis of size and proportions (gener ally stouter than in Porzana , Poliolimnas , or Porphyrio ). Geographical Survey of Gallirallus Fossils in Oceania Bismarck Islands, New Ireland Bird bones recovered from early human occupation sites in the Bismarck Islands and Solomon Islands are generally more fragmentary and less abundant th an those from the much yo unger archaeological sites in Remote Oceania. From four archaeologica l sites in caves and rockshelters on New Ireland (Figure 2-1), C. Gosden and J. Alle n excavated 66 bones of rails identified by Steadman et al. (1999) as bel onging to six extant species a nd two undescribed species.

PAGE 31

17 Twenty-one specimens referred to Gallirallus new sp. (Steadman et al. 1999) from the Panakiwuk and Balof rockshelter sites a nd from the Matenkupkum and Matenbek cave sites are assigned to the new species, † Gallirallus unnamed A Kirchman and Steadman (2006c). Holotype Distal humerus UF 62983 (Figur e 2-2), from Balof site, New Ireland. Paratypes –Balof archaeological site : Coracoids UF 63716, 63717, scapulae UF 63714, 63715, pelves UF 63712, 63713, distal tibiotarsi UF 62978, 62979, 62981, 62982, 62984, 62987, proximal tibiotarsus UF 62986, and distal tarsometatarsus UF 62980. Matembek Site: Femoral shaft UF 62976, part ial tibiotarsi UF62975, 62977 (Figure 2-3). Panakiwuk Cave: Proximal humerus UF 62973. Matenkupkum Cave: Tibiotarsal shaft UF 62974. Etymology – Named for Ernst Mayr, to honor his memory and to recognize his unrivaled contributions to ornith ology in the Western Pacific. Diagnosis – A large rail (Table 2-2) that differs in many osteological characters from the similarly sized a nd perhaps closely related Gallirallus insignis from nearby New Britain. The distal humerus of G. unnamed A has a narrow, ovoid fossa musculo brachialis similar to that in G. insignis , but is deeper and more clearly emarginated, as in fully flightless species. The holotype UF 62983 (Figure 2-2) di ffers from humeri of all other species of Gallirallus in having a condylus ventralis separated from the condylus dorsalis by a deep furrow, making the condylus ventralis more prominent in cranial and distal aspects; corpus humeri extremita s distale rounder in cr oss-section with a width:depth ratio higher than in all congeners (3.1:3.8 mm in † G. unnamed A versus e.g., 2.5:3.6 in G. insignis , 2.6:3.5 in G. philippensis , and 2.7:4.2 in G. woodfordi ). The

PAGE 32

18 proximal humerus UF 62973 is similar to that of G. woodfordi in overall size and in the shape of the crista deltopectoralis, which is deeply concave in caudal aspect in both species. It differs from that of G. woodfordi in having a depression at the proximal margin of the crista deltopect oralis visible in proximal aspect. The tibiotarsus of † G. unnamed A (Figure 2-3) is of similar size as G. insignis but differs from all congeners in the following characters: foramen interosseum proximale very short; impressio ligamentum collateralis medialis deeper and w ith a prominent margin on facies caudalis; furrow between corpus tibiotarsi and cris ta fibularis deeper and longer; pons supratendineus expanded craniall y, and more visible in medial aspect. Tarsometatarsus: corpus tarsometatarsi distale wider but not deeper than in G. insignis ; groove proximal to foramen vasculare distale short; fossa metata rsi I relatively shorter than in all other species. Remarks – The prominent fossa musculo brac hialis and the small, rotated crista deltopectoralis of the humerus of † G. unnamed A suggest reduced flight capability relative to fully volant conge ners. Unfortunately, the st ernum is unknown, and both the forelimb and hindlimb specimens are too frag mentary to permit morphometric analyses of this species’ flight ability. Solomon Islands, Buka Fragments of 18 rail limb bones from the late Pleistocene Kilu Cave site are difficult to distinguish from specim ens of the two large flightless rail species that inhabit the Solomon Islands today, G. rovianae and the larger G. woodfordi . Only the distal tarsometatarsi UF 62991 and 62992 can be assigned with confidence, and I refe r these specimens to G. rovianae . I refer the femur shafts UF 63000, 63002, tibiotarsi shafts UF 63001, 63003-63005, a nd the tarsometatarsus shaft UF

PAGE 33

19 62999 to G. cf. woodfordi on the basis of their larger size. The ulnar shaft UF 62998, femora UF 62989, 62990, 62994-62997, tibiotarsus UF 62993, and tarsometatarsus UF 62988 are refered to G. cf. rovianae on the basis of their smaller size. Vanuatu, Efate Unlike early occupation sites throughout Remote Oceania, sites dating to the Lapita period (ca. 3000-2800 cal BP) excavated by M. Spriggs and S. Bedford on multiple islands in Vanuatu have not yielded bone deposits that document extensive prehistoric bird extinctions. Bird bones from Arapus, an open, coastal village site on Efate (Figure 2-1), incl ude remains of the introduced Gallus gallus , as well as other large-bodied, native species typically hunted for food (Spriggs and Bedford 2001). Twelve rail bones from Arapus are not qua litatively different from specimens of G. philippensis , a volant species that is extant in Vanuatu today. Northern Mariana Islands Sites on four islands in the Marianas (Figure 2-1) document a history of human exploitation of birds beginning ca. 2500 yr BP, which probably is about 1000 years after the first arrival of humans in the Mariana Islands (Kirch 2000). Although these bone assemblages fail to record the period of initial human contact with the native avifauna, they contain the remains of 12 extirpat ed bird species, as well as undescribed extinct species of rails , ducks, parrots, and passerines (Steadman 1999). Rail remains from the Pisonia Rock shelter on Aguiguan, and the Railhunter Rockshelter on Tinian incl ude over 1100 specimens of Porzana [ Poliolimnas ] cinereus , Porzana sp., and Gallirallus sp . (Steadman 1999). In addi tion, hundreds of specimens of Porzana and Gallirallus were recently excavated by John Craib at the Unai Chulu site (Craib 1993) on Tinian and the Mochong site on Rota.

PAGE 34

20 Aguiguan I examined 219 specimens of Gallirallus from the Pisonia Rockshelter, representing most cranial and post-cranial skeletal elements . Nearly all of these bones are burned; they are dominated by pedal pha langes and average more fragmentary than the bones of Gallirallus from nearby Tinian, to be disc ussed below. The rail from Aguiguan represents a new species, † Gallirallus unnamed B Kirchman and Steadman 2006c . Holotype Complete coracoid UF 62934 (F igure 2-4) from the Pisonia Rockshelter archaeological site, Aguiguan, Commonwealth of the Northern Mariana Islands. Paratypes Rostrum UF 62918, mandibles UF 62912-62914, 62926, 62946, 62947, quadrates UF 62888, 62889, 62941, vertebrae UF 62880, 62890, 62919 (lot of 10), 62933, 62943, 62955, sterna 62892/62893 (Figure 2-5, two pieces glued together), 62922, coracoids UF 62881, 62884, 62910, 62911, 62931, 62939, 62940, scapulae 62878, 62915, 62916, humeri UF 62887, 62902, 62903, ulna UF 62925, radius UF 62953, carpometacarpi UF 62882, 62883, 62917 (Figure 2-4), 62923, 62924, phalanx 1 of digitus majoris UF 62938, synsacrum UF 62948, femur UF 62875, tibiotarsi UF 62876, 62894-62901, 62927, 62937, 62944, fibulae UF 62932, 62942, 62954, tarsometatarsi UF 62877, 62879, 62904-62909, 62928, 62929, 62949-62952, pedal phalanges UF 62870-62873, 62874 (lot of nine), 62855, 62886 (lot of six), 62891 (lot of 48), 62920, 62921 (lot of 13), 62930 (lot of nine ), 62935, 62936 (lot of 22), 62945 (lot of 24), 62956. Diagnosis A small to medium-sized rail (smaller than Gallirallus philippensis ; Table 2-3) with the following unique combin ation of characters: Mandible: fenestra mandibulae caudalis large. Sternum: angle of cris ta lateralis relative to dorsal margin, in

PAGE 35

21 lateral aspect, ca. 50°; height of carina sterni higher than in G. owstoni but lower than in G. philippensis . Coracoid: facies articularis sterna lis narrow, shallow; impressio musculo sternocoracoracoidei less concave; corpus cora coidei, in dorsal aspect, wide relative to processus procoracoideus; facies articularis clavicularis with little protrundence from corpus coracoidei. Scapula: in proximal asp ect, area between facies articularis humeralis and facies articularis clavicularis shallow. Humerus: caput humeri shallow in proximal aspect; incisura capitis shallow. Carpomet acarpus: short and slender relative to leg elements. Etymology Named after the type locality, Pi sona Rockshelter, which in turn was named after an immense Pisonia grandis tree that grew nearby. Rota Rail bones from the Mochong Site a nd the Route 100 Site represent a new species, † Gallirallus unnamed D Kirchman and Steadman 2006c. Holotype Complete coracoid UF 6 3302 (Figure 2-4) from the Route 100 archaeological site, Rota, Commonwealth of the Northern Mariana Islands. Paratypes Route 100 site: Rostrum UF 63296, skull UF 63293, vertebrae UF 63301, 63305, coracoid UF 63292, scapula UF 63300, ulna UF 63304, carpometacarpus UF 63299 (Figure 2-4), tibiotarsus UF 63319, tarsometatarsus UF 63303. Mochong Site: Distal femur UF 62964, proximal tibiotarsus UF 62965, complete tarsometatarsus UF 62962, distal tarsometatarsus UF 62963. Diagnosis A medium-sized ra il, slightly smaller than Gallirallus philippensis (Table 2-3), with the following unique comb ination of characters. Coracoid: facies articularis sternalis narrow, shallow; fora men nutrium supracoracoidei large; lateral margin of facies articularis short and rounded in humeral aspect. Scapula: in proximal

PAGE 36

22 aspect, area between facies articularis hume ralis and facies arti cularis clavicularis shallow. Carpometacarpus: length from proce ssus pisiformis to proximal end of spatium intermetacarpalis short. Etymology From the Latin tento ( tempto ), which means “try, prove, put to the test” (Brown 1956:819). The name unnamed D thus refers to how trying it has been for rails of the genus Gallirallus to survive on Rota. Gallirallus unnamedB became extinct in prehistoric times, whereas biologists have attempted to establis h a wild population of G. owstoni on Rota since 1990, with little suc cess in spite of valiant effort. Saipan Six partial tibio tarsi UF 62967-62972 and one pedal phalanx UF 62966 from the Chalan Piao site belong to Gallirallus philippensis . In spite of minor differences, nine fragmentary bones from th e Unai Bapot site are also referred to G. philippensis , namely a narial bar UF 63744, humerus UF 63740, ulna UF 63736, three tibiotarsi UF 63737-63739, and three tarsomet atarsi UF 63741-63743. The leg bones of the Saipan birds are rather small, in the size range of modern female specimens of G. philippensis . While it could be that a larger seri es of bones from Saipan might reveal characters to distinguis h the population there from G. philippensis , the bones in hand do not have adequate characte rs to provide a diagnosis. Tinian Two sites on Tinian contain abundant remains of a medium-sized rail that represents a new species, † Gallirallus unnamed C Kirchman and Steadman 2006c. Holotype Complete humerus UF 63419 (F igure 2-4), Unai Chulu archaeological site, Tinian, Commonwealth of the Northern Mariana Islands. Paratypes Bones from two archaeo logical sites on Tinian belong to G. unnamed C . Unai Chulu Site: Rostra UF 62140, 62300, 62301, 63385, 63462, quadrates UF

PAGE 37

23 63393, 63468, 63516, 63552, mandibles UF 62378, 62664, 62665, 63366, 63447, 63495, 63559, vertebra UF 63440, sterna UF 62295, 62320, 62334, 62354, 62377, 62387, 62669, 63432, scapulae UF 62091-62094, 62164, 62319, 62353, 62391, 62634, 62683, 63388, 63433, 63459, 63467, 63473, 63494, 63515, 63546, 63547, coracoids UF 61942, 61951, 61974, 62009, 62022, 62046-62048, 62078, 62124, 62160, 62167, 62198, 62227, 62307, 62313, 62359-62361, 62380, 62393, 62425, 62431, 62635, 62675, 62687, 63361, 63368, 63386, 63387, 63402, 63423, 63430, 63503, 63504, 63600, 63601, 63617, 63618, 63696, humeri UF 61905, 61908, 61915, 61952, 62003, 62008, 62015, 62024, 62036, 62037, 62081, 62090, 62117, 62126-62128, 62133, 62146, 62179-62182, 62262, 62263, 62335, 62365, 62374, 62381, 62383, 62413, 62439, 62441,62636-62638, 62639, 62759, 62765, 63367, 63370, 63383, 63392, 63398, 63412, 63426, 63498-63503, 63522, 63523, 63529, 63534, 63543-63545, 63553, 63557, 63570, 63580, 63585,63591, 63598, 53599, 63610, 63614-63616, 63644, 63645, 63673, 63674, 63694, 63695, ulnae UF 61907, 61916, 61917, 61945, 62209, 62312, 62414, 62264, 62265, 63375, 63404, 63429 (Figure 2-4), 63521, 63530, 63548, 63556, radii UF 62088, 62089, 62171, 62616, 62666, 63456,63549, 63612, 63646, 63708, carpometacarpi UF 61939, 61941, 62049, 62050, 62095, 62134, 62135, 62150, 62199, 62366, 62394, 62415, 62416, 62424, 62432, 62274, 62275, 62668, 62711, 63372 (Figure 2-4), 63384, 63442, 63480, 63481, 63588, 63594, 63609, 63641-63643,63676-63678, phalanx 1 of digitus majoris UF 62384, pelvis UF 63452, femora UF 61894, 61904, 61922, 61934, 61972, 61975, 61976, 62016, 62020, 62031, 62043, 62080, 62168, 62178, 62228, 62294, 62327, 62331, 62332, 62350, 62358, 62376, 62389, 62619, 62682, 62715, 62746, 62761, 62769, 62775, 63357, 63362, 63369, 63376, 63377, 63380, 63400, 63431, 63443, 63448, 63451, 63463, 63466, 63469, 63472,

PAGE 38

24 63475, 63490-63493, 63511, 63517, 63518, 63525, 63528, 63531, 63541,63560, 63592, 63613, 63647-63651, 63700, tibiotarsi UF 61896-61899, 61902, 61903, 61909, 61910, 61913, 61914, 61920, 61938, 61947, 61963/61964 (two pieces glued together), 61977, 61988, 61992, 61997, 61998, 62002, 62005, 62006, 62010, 62014, 62018, 62019, 62023, 62025, 62027-62030, 62032-62034, 62038-62042, 62051-62054, 62056, 62057, 6205962061, 62082-62087, 62112-62016, 62130-62132, 62141-62145, 62151, 62153, 62156, 62157, 62161-62163, 62169, 62170, 62184-62186, 62188, 62196, 62197, 62201-62208, 62229-62232, 62255-62261, 62291, 62311, 62315-62317, 62328-62330, 62349, 62356, 62357, 62367, 62369, 62379, 62382, 62392, 62400, 62401, 62410-62412, 62428, 62429, 62437, 62438, 62502, 62600, 62633, 62655-62663, 62701, 62703, 62704, 62712, 62716, 62717, 62720, 62737, 62763, 62764, 62771, 62772, 62777, 62779, 63358, 63359, 63363, 63371, 63374, 63378, 63379, 63382, 63389, 63399, 63405-63410, 63413-63416, 63420, 63427, 63434, 63435, 63438, 63439, 63444, 63446, 63449, 63464, 63465, 63471, 6348263485, 63488, 63489, 63504-63510, 63512, 63519, 63524, 63538-63540, 63550, 63551, 63561-63569, 63577-63579, 63586, 63587, 64595-63597, 63608, 63619-63626, 6365263658, 63679-63682, 63691, 63692, 63695, tarsometatarsi UF 61895, 61901, 61921, 61944, 61999, 62001, 62007, 62017, 62026, 62035, 62044, 62045, 62058, 62077, 62079, 62120-62123, 62125, 62147-62149, 62152, 62154, 62158, 62159, 62165, 62166, 6217262177, 62187, 62195, 62210-62217, 62219, 62233-62237, 62266-62272, 62308-62310, 62314, 62336-62340, 62346-62348, 62355, 62362-62364, 62386, 62370, 62375, 62386, 62390, 62399, 62423, 62430, 62436, 62617, 62648-62654, 62686, 62729, 62753, 6275662758, 62766, 62773, 63360, 63364, 63365, 63381, 63391, 63411, 63421, 63425, 63428, 63436, 63453, 63460, 63461, 63470 (Figure 2-7), 63477-63479, 63496, 63497, 63513,

PAGE 39

25 63514, 63532, 63533, 63571-63573, 63576, 63589, 63590, 63593, 63627 (Figure 2-7), 63628-63633, 63659-63663, 63683, pedal phalanges UF 61923, 61924, 61928, 61933, 61937, 61956-61958, 61970, 61971, 61978-61981, 61986, 61987, 61991, 62013, 62055, 62062-62076, 62100-62110, 62129, 62136-62139, 62155, 62188-62193, 62220-62226, 62238-62254, 62276-62290, 62302-62305, 62318, 62341-62345, 62351, 62352, 6237162373, 62395-62398, 62402-62404, 62417-62422, 62426, 62427, 62433-62435, 62608, 62611-62613, 62632, 62640-62647, 62691, 62692, 62695, 62697, 62739, 62747, 62450, 62455, 62474, 62762, 63373, 62778, 63394-63397, 63401, 63403, 63417, 63437, 63441, 63520, 63535-63537, 63575, 63581, 63602-63605, 63611, 63634-63640, 63664-63673, 63684-63689, 63690, 63476. From Railhunter Rockshelter I refer the fo llowing additional paratypes: Rostra UF 60023, 60027, 60042, 60044, 60071, 60147, 60143, 60148, 60199, mandibles UF 60027, 60108, 60129, 60147, 60148, 60198, 60210, vertebra UF 60209, sterna UF 60072, 60109, 60142 (Figure 2-5), 60149, 60166, coracoids UF 60042, 60045, 60058, 60078, 60130, 60143, 60178, 60110 (Figure 2-4), 60111, 60217, 60219, 60225, 60226, scapulae UF 60029, 60030, 60059, 60073-60077, 60112, 60146, 60150, 60179, 60180, 60188-60190, 60224, humeri UF 60046, 60079-60081, 60131, 60151, 60167, 60181, 60191, 60212, ulnae UF 60083, 60084, 60185, radii UF 60113, 60132, 60133, 60152, 60192, carpometacarpi UF 60085, 60060, pelvis UF 60086, femora UF 60049, 60061, 6013460137, 60153, 60157, 60168, 60177, 60182, 60193 (Figure 2-4), 60194, 60158, 60114, 60115/60116 (Figure 2-5, two pieces glued to gether), tibiotarsi UF 60031, 60032, 60047, 60048, 60050, 60051, 60062-60066, 60087-60096, 60117-60119, 60126-60128, 60138, 60175, 60186, 60200-60203, 60207, 60213, 60220, 60227, fibulae UF 60154, 60195,

PAGE 40

26 tarsometatarsi UF 60033-60035, 60052, 60067, 60097-60102, 60139, 60144, 60145, 60155, 60156, 60159-60161, 60169-60172, 60176, 60187, 60196, 60214, 60215, pedal phalanges UF 60036-60041, 60082, 60103-60107, 60120-60125, 60140, 60141, 6016260165, 60173, 60174, 60183, 60197, 60204-60206, 60208, 60216, 60218, 60220, 60221. Diagnosis A medium-sized ra il, slightly smaller than Gallirallus philippensis (Table 2-3), and with disproportionately shor ter wing elements. Elements of the pectoral girdle and wing have many unique characters that distinguish † G. unnamed C from all other known species of Gallirallus , as follows. Mandible: large relative to its body size. Sternum: spina externa of rostrum sterni absen t; margo cranialis of carina sterni shifted dorsally and caudally relative to volant conge ners; margo cranialis of labrum dorsale deeply notched (<90°) to form a v-shaped me dial separation between the sulci articularis coracoidei. Coracoid: corpus coracoidei stout, relatively shor t (Table 2-3); humeroventral half of corpus coracoidei wide, w ith much protrudence of facies articularis clavicularis; lateral margin of facies arti cularis humeralis shor t, rounded in humeral aspect. Humerus: short relative to leg elemen ts; distal end wide re lative to length and width of corpus humeri (Table 23); as in the flightless species G. owstoni , G. australis , G. ripleyi, the corpus humeralis with a straight margo cranialis (rather than sigmoid) in dorsal aspect; crista bicipitalis reduced distally ; distal margin of crista pectoralis slopes more gradually to corpus humeri; fo ssa musculo brachialis deep; processus supracondylaris dorsalis large. Ulna: similar in size and proportion to in G. striatus and small female G. philippensis ; depressio musculo brachiali s relatively deep and more distinctly emarginated than in G. philippensis , G. torquatus , G. striatus , G. owstoni , or G. australis , resembling the condition seen in † G. ripleyi and † G. storrsolsoni , the latter

PAGE 41

27 having the deepest, most prominent depressio musculo brachialis; tuberculum carpale juts abruptly out from corpus ulnare (rather than gradually sloping to corpus ulnare) and is triangular rather than circular in ventral aspect. Carpometacarpus: shorter but not more slender than in female G. philippensis , but less shortened than in G. owstoni . Femur and Tibiotarsus: not smaller (Table 2-3) or qualitatively different than in G. philippensis . Tarsometarsus: slightly more robust (s horter on average but not wider) than in G. philippensis; trochlea metatarsi III relati vely wide in dorsal aspect. Etymology Derived from the Latin word pendiculus , meaning a noose or snare (Brown 1956: 812). The name unnamed C is an adjective (“of the snare”) modifying the masculine Latin noun Gallirallus. It refers to the manner by which the remains of this flightless species were probably accumulated, that is by hunters us ing leg-hold snares. Remarks Correlation coefficients from a principal components analysis of postcranial skeletal measurements (Table 2-4) from nine Gallirallus species (Figure 2-8) indicate that the first component describes the overall body size of each species, whereas the second component describes the degree of flightlessness, with high positive values corresponding to reduced length of wing elements and high negative values corresponding to robust (wide) leg elemen ts. The first and second components summarize 89.42% and 8.08% of morphological va riance, respectivel y. Each of the subsequent components accounts for less than 1% of the variance. Species mean values plotted along these axes (Fi gure 2-8) separate known f lightless species from known volant ones. The position of † G. unnamed C lies on or just above the threshold for flightlessness, suggestin g that it was likely to have been flightless.

PAGE 42

28 Characters of the sternum associated with flightlessess indicate that † G. unnamed C had reduced flight capability relative to the three similarly sized, volant species ( G. philippensis, G. striatus , and G. torquatus ), but a stronger degree of volancy than in the fully flightless species G. owstoni , † G. wakensis , G. woodfordi , and especially G. australis . The apex of the carina sterni is not preserved in any specimen of G. unnamed C , precluding measurement of carina depth. Nevertheless, the condi tion of the sternum corroborates the hypothesis that † G. unnamed C was near the thresh old of flightlessness on the basis of PCA of other postc ranial elements (Figure 2-8). Tonga Archaeological sites on seven Tonga n islands (Figure 2-9) excavated by D. V. Burley, D. W. Steadman, and asso ciates (Burley 1999, St eadman et al. 2002a, 2002b), contain abundant vertebrate fossils th at reveal the extinction (species-level losses) or extirpation (population-level loss es) of four species of lizards (Pregill 1993, Pregill and Steadman 2004), four species of bats (Koopman and Steadman 1995), eight species of seabirds, and 23 species of la ndbirds (Steadman 1993, 2006, Steadman et al. 2002a, b). Prehistoric bones of Gallirallus are known from seven archaeological sites on five islands in the Ha’apai Group of Tonga. E ach of these islands ha s a single site that dates to the Lapita period of first hum an contact, approximately 2900-2700 cal BP (Burley 1994, 1999, Steadman et al. 2002a). Li fuka also has two younger sites dating to 2800-2500 cal BP and 500-200 cal BP (Steadman et al. 2002b). Except for two specimens from Ha’afeva, all Gallirallus bones from sites in Ha’apai are qualitatively indist inguishable from, and within the range of size observed in, my sample of specimens of G. philippensis, to which I refer all specimens from Foa, Lifuka, ‘Uiha, and Ha’ano, and 14 of the 16 bones from Ha’afeva.

PAGE 43

29 A proximal tibiotarsus (UF 56463) and a di stal tibiotarsus (UF 56755) from the Mele Havea site on Ha’afeva are larg er than the largest specimens of G. philippensis . UF 56463 differs further from the tibiotarsus of G. philippensis in having the junction of the crista fibularis and corpus tibiotarsi more concave in dorsal aspect, and in having a greater distance between crista fi bularis and facies articularis lateralis. I speculate that these specimens are the only known remains of an undescribed species of Gallirallus , and these specimens are referred to † Gallirallus undescribed sp. Prope r description of this species requires additional fossil material. Tongatapu Each of the 32 bones of Gallirallus from the Lapita period (29002700 cal BP) Ha’ateiho archaeol ogical site are referred to G. philippensis . ‘Eua The ‘Anatú site on ‘Eua is the only prehistoric site in Tonga that pre-dates the arrival of people (Steadman 1993, 1995). La yer I contains cultural artifacts and bones of human-introduced pig, dog, rat, and chicke n. Radiocarbon dates from Layer I range from 570 ± 70 to 2710 ± 70 cal BP. Pre-cultural Layers II and III are separated by a thin layer of calcite flowston e with uranium series dates of 60,000 ± 3000 to 78,800 ± 2700 years old. Bones of Gallirallus are found in all three Layers. The 14 bones from Layer I are all identifiable as G. philippensis , which is absent in the underlying Layers II and III. Layers II and III instead contained 78 bones of † Gallirallus vekamatolu Kirchman and Steadman (2005). Holotype Complete tarsometatarsus UF 51991 (Fig 2-10) from the ‘Anatu site on ‘Eua. The holotype and a paratype (ulna UF 51734) were referred to as “ Gallirallus new sp. (‘Eua, Tonga)” by Steadman et al. (2000: Figs 11, 12).

PAGE 44

30 Paratypes Rostra UF 52285, 52292 (Fi gure 2-11), quadrates UF 52341, 52711, mandibles UF 51836, 52525, vertebrae UF 50513, 51799, 51876, 52031, 52032, 52033, 52188, 52419, coracoids UF 52024, 52204, 52409, 52966, scapulae UF 51726, 52036, 52179, 52318, 52956, 52985, humeri UF 52333, 52707 (Figure 2-11), ulnae UF 51734 (Figure 2-11), 51736, 52552, 52868, radius UF 52332, carpometacarpi UF 51743 (Figure 2-11), UF52011, wing phalanges UF 52234, 52789, femora UF 52020 (Figure 2-11), 52052, 52058, 52079, 52095, 52105, 52354, 52518, tibiotarsi UF 51729 (Figure 2-11), 52045, 52077, 52127, 52202, 52211, 52320, 52592, 52951, tarsometatarsi UF 50484, 51732, 52002, 52047, 52106, 52137, 52501, 52577, 52876, pedal phalanges UF 50483, 50623, 51828, 51832, 51979, 52018, 52447, 52460, 52486, 52499, 52516, 52524, 52581, 52683, 52735, 52788, 52955. Diagnosis – A medium-sized (T able 2-5), stocky species of Gallirallus , distinguished from congener s by the following characters. Dentary: flat symphysis mandibulae; rapidly tapering tip of rostrum mandibulae. Humerus: fossa muscularis brachialis deep as in other f lightless species; crista deltopect oralis very thick and rotated cranially. Ulna: cotyla ventralis ovoid, extending onto olecranon. Carpometacarpus: short symphisis metacarpalis distalis; os met acarpale alulare short and wide in proximal aspect, as in † G. ripleyi . Femur: prominent fossa lateral to crista supracondylus medialis; prominent fossa on medio-distal corpus femori s. Tibiotarsus: in cisura tibialis deep, opening caudally; fossa laterodistal to crista cnemialis lateralis shallow, poorly defined. Tarsometatarsus: corpus ta rsometatarsi robust as in † G. ripleyi , but not so robust as G. australis ; facies dorsalis of corpus ta rsometatarsi concave, as in G. torquatus .

PAGE 45

31 Etymology – From the Tongan words veka , which refers to the extant rail G. philippensis , and matolu , meaning stout (Churchward 1959). The name vekamatolu is noun of neuter gender. Remarks Elements of the skull and legs of † Gallirallus vekamatolu are larger and more robust than in G. philippensis (Figure 2-11). Conversely, the wing elements, though differing in detail, are similar in size and only very slightly stouter than in G. philippensis . The degree of reduction of wing elements in † G. vekamatolu is similar to that of flightless G. owstoni , but less than in G. australis. Results of principal components analysis of 14 post-cranial skeletal characters (Table 2-6) also support the hypothesis that † G. vekamatolu was flightless. Correlation co efficients indicate that PC 1 describes variation in overall si ze and the degree of reduction of the ulna, and that PC 2 is a mixture of wing-element reduction and tarsomet atarsal robustness. A plot of PC 1 vs PC 2, summarizing 95.65% of mor phometric variance, clusters † G. vekamatolu with flightless rather than volant mode rn congeners (Figure 2-12). Bird bones from ‘Anatú and other smaller sites on ‘Eua represent 23 species of land birds now gone from the island. † Gallirallus vekamatolu is one of 11 extinct species, with 12 others no longer on ‘Eua but still existing on one or more islands elsewhere (Steadman 1993, 2006). The flightless † G. vekamatolu is the only species of rail recorded on ‘Eua in pr e-cultural strata. After human arri val, bones of thr ee volant species of rails, Porzana tabuensis , G. philippensis, and Porphyrio porphyrio, are recorded commonly, although only the last tw o species survive in Tonga. Society Islands, Huahine Prehistoric birds from the So ciety islands (Figure 2-13) are known from one site in on Huahine, an eroded volcanic island ca. 3 million years old

PAGE 46

32 (Dickinson 1998). The Fa’ahia site (Figure 2-13) was excav ated by Y. H. Sinoto and colleagues from 1973 to 1984 in cooperation with the Départment de Archéologie, Centre Polynésien des Sciences Humaines, Tahiti (DAP T). The cultural deposits at Fa’ahia were found submerged below the modern water ta ble, and contained exceptionally well preserved organic materials including wooden adze handles and parts of canoes, a wide variety of artifacts made of bone, shell, or stone, and non-human bones representing food items of all sizes (Sinoto 1975, 1979). By East Polynesian standards, Fa`ahia is an early occupation site with radiocarbon dates rangi ng from ca. 1250 to 750 yr BP (Sinoto 1983). Despite uncertainty in the chronology of first human a rrival (Sinoto 1970, Kirch 1984, 1986, 2000, Spriggs and Anderson 1993, Conte and Anderson 2003), bird bones from Fa`ahia mainly represent extinct species, cons istent with the pattern of heavy exploitation of native birds early in the archaeologi cally-preserved cultural sequence in East Polynesia, which typically begins at ca. 1000 yr BP (Dye and Steadman 1990, Kirch et al. 1995, Steadman and Rolett 1996). Of over 300 bi rd bones obtained at Fa`ahia including 53 rail bones identified by Steadman a nd Pahlavan (1992), three were from Porzana tabuensis , a small crake that is widespread in Polynesia but that no longer occurs on Huahine. The other 50 bones were assigne d to a larger, presumably undescribed, flightless rail, referred to as Gallirallus new sp. Of these, 47 specimens belong to the new species † Gallirallus storrsolsoni Kirchman and Steadman 2006a. Holotype Complete cranium and rost rum BPBM 166036 (Figure 2-14) excavated from the Fa’ahia site, Huahine by Y. H. Sinoto. Paratypes Crania BPBM 166021, 166026, 168015, DAPT 139, rostrum DAPT 21, vertebrae BPBM 166024, 166025, 166035, 168001, 168002, 168078, DAPT 13, 14, 25,

PAGE 47

33 60, 61, 122, 143, 144, 145, 163, rib BPBM 166018, sterna BPBM 166017, 166027 (Figure 2-15), humerus BPBM 166022 (Figur e 2-16), ulnae BPBM 166033 (Figure 2-16), 168121, 168150, radius BPBM 168056, carpometaca rpus BPBM 168165 (Figure 2-16), synsacrum BPBM 166020, femora BPBM 168131, DAPT 27/105, tibiotarsi BPBM 166023 (Figure 2-16), 166032, 168028, 168046, 168123, 168149, 168170, DAPT 47, 55, 119, tarsometatarsi BPBM 166034, 168124, DAPT 7 (Figure 2-16). Diagnosis A medium-sized species of Gallirallus (Table 2-5) distinguished from congeners in Oceania as follows. Skull: fo ssa temporalis deeply excavated, clearly emarginated by crista temporalis and exte nds more caudally; cranium with prominent crista nuchalis transversae and a low, broad calveria (as in G. woodfordi ); in dorsal aspect, the posterior margins of the orbits abruptly angle away from the midline; the lamina parasphenoidalis is well emarginated cau dally. Sternum: spina externa of rostrum sterni absent. Humerus: incisura capitus narrows proximally in caudal aspect; crista deltopectoralis rectangular, para llel to corpus humeri; corpus humeri thin, round in crosssection. Ulna: corpus ulnaris straight and dorso -ventrally flattened, more so than even in flightless G. owstoni , † G. vekamatolu , or G. australis ; impressio brachialis deep and clearly emarginated. Synsacrum: broad in ve ntral aspect, gradually narrowing caudally. Femur: corpus femoris robust, approachi ng but not surpassing the stoutness of † G. vekamatolu . Tibiotarsus (Figure 2-15): proportionally short (as in G. australis ); incisura intercondylaris wide, resulting from an obtus e angle between the condylus medialis and condylus lateralis; juncture of condylus medialis with facies caudalis of corpus tibiotarsis abrupt rather than gradually sloping. Tarsometatarsus (Figure 2-15): corpus tarsometatarsi dorso-ventrally flattened with a width-to-depth ratio (1.46) greater than in

PAGE 48

34 other Gallirallus species (1.21-1.36); viewed medially, the proximal one-third of corpus tarsometatarsi slopes toward the hypotarsus rather than be ing perpendicular to facies dorsalis. Etymology Named after Storrs L. Olson in recognition of his unparalleled contributions to the evolution, systematics, a nd paleontology of flightless rails on islands. Remarks The material from Fa’ahia repres ents 6 individuals, minimally. Because of the excellent preservational environment at the Fa’ahia site, even the most delicate elements of the skeleton of † G. storrsolsoni are known, including one partial sternum (UF 166027, Figure 2-15) that still retains the ante rior margin of the carina. The greatly reduced carina sterni, small wing elements, a nd stout leg elements support a diagnosis of flightlessness, as do the results of PC A of 16 skeletal measurements from G. storrsolsoni and seven other species of Gallirallus that are known to be eith er volant or flightless. Correlation coefficients of the first four principal components (PC), which together account for 99.4% of morphological variance, indicate that PC 1 describes variation in overall size and the degree of reduction of the carina sterni, ulna, and carpometacarpus, and that PC 2 is a description of keel a nd wing reduction, and leg-bone robustness (Table 2-7). A plot of PC 1 vs. PC 2, summarizing 96.5% of mor phometric variance, clusters G. storrsolsoni with flightless rather than volant congeners (Figure 2-17). Three femora from the Fa`ahia archaeological site (BPBM 166031, DAPT 39, 53) were referred to Porphyrio Brisson 1760 rather than the othe r genera of large Pacific rails ( Gallirallus , Gallinula , Fulica ) by Kirchman and Steadman (2006a) because of these characters: in proximal aspect, more obtus e angle formed at the junction of the impressiones obturatoriae and trochanter fe moris; impressiones obturatoriae more

PAGE 49

35 prominent, leading to a more concave proximoposterior area of corpus femoris; similar size and position of the impressiones iliotrochanteria and linea intermuscularis caudalis; corpus femoris overall more sl ender; distal end of corpus femoris not expanded laterally until the epicondylus lateralis is reached; rotula r groove more narrow; in posterior aspect, medial margin of the condylus medialis orient ed roughly parallel to the shaft rather than diagonal. These specimens were referred to the new species † Porphyrio mcnabi , Kirchman and Steadman (2006a). Marquesas Islands, Tahuata Fossil Gallirallus are known from sites on four islands in the Marquesas Islands (Figure 2-18). The sites represent early human occupations developed in coastal calcareous sands on Tahuata, Hiva Oa, Ua Huka, and Nuku Hiva. A total of 53 bones from these four sites were referred to “ Gallirallus new spp.” or “ Gallirallus undescribed spp.” by Steadman and Rolett (1996) and Steadman (1989a, 2006). The Hanamiai site on Tahua ta was excavated by B. V. Rolett in 19841985 (Rolett 1998). The sediment excavated at Hanamiai was screen-washed through 1/8 inch mesh, producing 716 identifiable bi rd bones, among which are 70 landbird bones from 10 species (Steadman and Rolett 1996). Radiocarbon dates from the lowest stratigraphic levels of Hanamiai place the be ginning of human occupation of Tahuata at ca. 1000 yr BP (Rolett 1998). The 22 Gallirallus bones from Hanamiai are referrred to the new species † Gallirallus roletti Kirchman and Steadman (2006b). Holotype Associated complete ri ght femur BPBM 166447, right tibiotarsus BPBM 166446, right tarsometatarsus BP BM 166448, and pedal phalages BPBM 166449, 166450 (Figures 2-19, 2-21). Collected by B. V. Rolett and colleagues from the Hanamiai archaeological site, Tahuata, Marquesas Islands.

PAGE 50

36 Paratypes Rostrum BPBM 166456 (Figur e 2-20), articulars BPBM 166458, 166542, dentaries BPBM 166436, 166439, 166457, coracoid BPBM 166455, femora BPBM 166435, 166445, tibiotarsi BPBM 166437, 166438, 166444, 166452, tarsometatarsi BPBM 166440, 166441, 166451, 166454. All are incomplete and from the same locality as the holotype. Diagnosis A medium-size d, flightless species of Gallirallus (Table 2-8) distinguished from congeners in Oceania as follows. Rostrum: robust and deep with height to width ratio (at anteri or margin of nares) of 1.24 (< 1.03 in all other species of Gallirallus ); in ventral aspect, trough in os prem axilare deep and wide, crista tomialis thin and sharp. Mandible: in caudal aspect , fossa caudalis with straight lateral and medio-ventral sides and a deep, narrow groove on top (ventral) side; pars symphysialis long, with rami sloped steeply to form a V-shaped (rather than U-shaped) trough in cranial aspect. Femur: medio-distal margin of neck deeply excavated below facies articularis acetabularis, forming a sulcus in anterior aspect; in ventral aspect, trochanter femoris forms a prominent right angle with faci es articularis antitrochanterica; the most proximal impressiones obturatori ae deep and long, forming a gr oove parallel to the crista trochanteris; in medial aspect , corpus femorus stout and st raight, especially on leading edge (facies cranialis). Ti biotarsus: fossa retropatellari s narrow but deep; impresso ligamentum collateralis medialis shallow; fossa flexoria shallow; crista fibularis short but broad, projecting farther from corpus tibiota rsis at distal end than in all except G. owstoni ; tuberculum retinaculi musculo fibularis prominent (as in † G. storrsolsoni and G. torquatus ) but not forming a tube (as in G. woodfordi ); condylus medialis and condylus lateralis large relative to widt h and depth of corpus tibiotars us. Tarsometatarsus (Figures

PAGE 51

37 2-19, 2-23): proportionately stout, although not so much as in † G. vekamatolu or † G. ripleyi ; sulcus extensorius deeply concave; corp us tarsometatarsi shallow relative to its breadth; facies medialis thin in medial as pect, especially proxim ally (approaching the condition in Amaurornis olivaceus or Porphyrio spp.); sulcus proximal to the foramen vaculare distale short and shallow; distal trochleae large and widely splayed. Etymology Named in honor of Dr. Barry V. Rolett, whose outstanding research in the Marquesas Islands has been of great impor tance to both biologists and archaeologists. In particular, his careful excav ations at the Hanamiai site yielded the most extensive and useful series of Gallirallus bones from the Marquesas Islands. Remarks The material from Hanamiai re presents four individuals, minimally. Lacking the sternum, scapula, and elements of the forelimb, the only evidence that † G. roletti was flightless comes from the relative ly small size of the coracoid (BPBM 166455), and the larger, more open shape and more medial postition of the cotyla scapularis on the dorsal surface of the co racoid, which more closely resembles the condition in flightless rath er than volant species of Gallirallus . This specimen is pitted over much of its surface, indicating that it ma y represent a juvenile bird. Comparing this coracoid among those of known juveniles and adults in G. philippensis (volant) and G. owstoni (flightless) suggests that BPBM 166455 is from a bird three to four months old and that the coracoid’s small si ze (relative to leg elements) is due to flight lessness rather than to the age of the bird. Steadman a nd Rolett (1996) referred 24 specimens to “ Gallirallus new sp. (Tahuata Rail),” which was named G. roletti based on 22 specimens. The discrepancy is accounted by an ungual phalanx of † Porphyrio paepae (BPBM 166442) being mistakenly liste d among the specimens of Gallirallus new sp. in Steadman

PAGE 52

38 and Rolett (1996) and by another pedal pha lanx (BPBM 166453) whose identity cannot now be determined. Hiva Oa The Hanatekua Rockshelter on Hi va Oa was excavated in 1967-1968 by Y. H. Sinoto and P. Bellwood (Bellwood 1972) , yielding 146 landbird bones from seven species (Steadman 2006). Radiocarbon date s place the beginning of human occupation on Hiva Oa at ca. 1000 yr BP (Rolett 1998). Two rail bones from the Hanatekua Rockshelter on Hiva Oa, a scapula, BPBM 165655 (Figure 2-24) and a femur, BPBM 168539 (Figure 2-21) were referred to † Gallirallus sp. by Kirchman and Steadman (2006b). These two bones re present a large, probably flight less species of Gallirallus . The specimens were an inadequate basis for describing a new species, but the scapula differs from that of congeneric species in having a relatively thick area between the facies articularis humeralis and facies articularis clavicularis, in proximal aspect, and in having a distinctive flange on ventral surface of corpus scapulae ne ar extremitas caudalis. The femur lacks both the proximal a nd distal ends, but is remarkable in having a relatively wide divergence of the two linea intermusculari s caudalis. The femoral shaft from Hiva Oa (BPBM 168539) is too large to belong to † P. paepae , and is larger than that of † G. roletti , suggesting that the Gallirallus species from Hiva Oa was quite large, and that Gallirallus from the southern islands were larger than those of the two northern islands. Ua Huka The Hane Dune site on Ua Huka , excavated by Y. H. Sinoto in the 1960s (Sinoto 1966, 1970, 1979), is the richest Marquesan site in terms of faunal remains, having yielded more than 12,000 identifiable bird bones, including 2187 landbird bones from 17 species (Steadman 2006) . Radiocarbon dates from the earliest strata of the Hane Dune site suggest that the initial human occupation of Ua Huka was ca.

PAGE 53

39 1350 yr BP (Rolett 1998). Hane Dune is the earliest well-dated site in the Marquesas. The 21 rail bones from Hane Dune site on Ua Huka represent a new species † Gallirallus gracilitibia Kirchman and Steadman (2006b). Holotype Right tibiotarsus lacki ng proximal end, BPBM 166013, 176387 (Figure 2-22; two pieces with different catalogue numbers glued together). Collected from the Hane Dune site, Ua Huka, by Y. H. Sinoto. Paratypes Coracoid BPBM 166015, humeri BPBM 163130 (Figure 2-25), 166014, femur BPBM 176974 (Figure 2-21), tibiotarsi BPBM 163240, 166008, 166012, 166016, 169145, 169146, 170909, 171294, 175384, 175145, 175322, 175395, 176725, 176746, 176971, 176972. All are incomplete and from the same locality as the holotype. Diagnosis A small to medium-sized (Table 2-9), flightless species of Gallirallus that differs from congeneric species as follows. Coracoid: sulcus musculo supracoracoidei relatively shallower and wide r than in volant species. Humerus: crista bicipitalis small; corpus humeri thin and curv ed; distal junction of crista pectoralis and corpus humeri abrupt, rather th an gradually sloping; sulcus ligamentum transversus deep. Femur: corpus femoris gracile. Tibiotarsus: corpus tibiotarsus slender relative to its length; depressio epicon dylaris medialis deep. Etymology From the Latin words gracilis (slender, slim, thin) and tibia (the shinbone, tibia); see Brown (1956:469, 791). The name gracilitibia is a feminine noun in apposition to Gallirallus . It refers to the distinctively slender tibiotarsus in this species, especially compared with that of † G. roletti . Remarks A minimum of eight individuals are represented. The length-to-width ratios of tibiotarsi show † G. gracilitibia to have the thinnest shaf t relative to length of any

PAGE 54

40 species of Gallirallus , whether flightless or volant (Table 2-9). Features of the coracoid and humerus of † G. gracilitibia described above, as well as their size relative to hindlimb elements indicate that this species may have been flightless. Nuku Hiva The Ha’atuatua Dune site was ex cavated initially by R. Suggs in 1956 and 1958 (Suggs 1961) and in 1992-1994 by B. V. Rolett and E. Conte (Rolett 1998). The combined excavations at Ha’atuatua Dune yielded 27 bones of landbirds belonging to nine species (Steadman 2006). Radio carbon dates place the beginning of human occupation of Nuku Hiva at ca. 1000 yr BP (Rolett 1998). The eight Gallirallus bones from Ha’atuatua site on Nuku Hi va belong to the new species † Gallirallus epulare Kirchman and Steadman (2006b). Holotype Nearly complete left ulna , BPBM 181658 (Figure 2-26). From the Ha`atuatua archaeological site, Nuku Hiva, Marquesas Islands. Collected by B. V. Rolett, E. Conte, and colleagues in 1994-1995. Paratypes Humerus BPBM 181657 (Figur e 2-25), femur BPBM 181659 (Figure 2-21), tibiotarsi BPBM 167191, 181660, 181661 (F igure 2-22), 181662, tarsometatarsus BPBM 167119 (Figure 2-23). All are from the same locality as the holotype. Diagnosis A small species of Gallirallus (Table 2-10) distinguished from congeneric species as follows. Humerus and ulna very small and sl ender relative to leg elements, especially compared to those of † G. gracilitibia from nearby Ua Huka. Ulna: straighter than in volant species. Tarsom etatarsus: facies dorsalis of corpus tarsometatarsi highly convex proxi mal to trochlea metatarsi III; foramen vasculare distale relatively large and round.

PAGE 55

41 Etymology From the Latin epularis (pertaining to a banquet, belonging to a banquet; Brown 1956:292). The name epulare is an adjective that modifies the masculine Gallirallus . It refers to the archaeological context in which the specimens of † G. epulare were found, namely that of a k itchen midden dominated by the bony and shelly remains of foods eaten by prehistoric Polynesians. Remarks The humerus, femur, and tibiotarsi lack both the proximal and distal ends, and therefore also lack diagnostic features. The most striking feature of † Gallirallus epulare is its tiny wing elements relative to its leg elements; whereas the leg elements closely resemble those from a small female G. owstoni , the humerus and ulna are much smaller than those of G. owstoni , and indeed more clos ely resemble those of † G. ripleyi and † G. wakensis . Discussion As is true for rails in general (Olson 1973a, Livezey 2003) there are many parallel aspects to the evoluti on of flightlessness in Gallirallus throughout Oceania. Features associated with flightlessness include an overall reduction and st raightening of wing elements, and modifications of the humerus (deep fossa musculo brachialis, cranially rotated crista deltopectoralis, smaller crista bicipitalis) and sternum (reduced and caudally displaced carina sterni). The delicate sternum is unknown in most fossil species, although excellently preserved bone deposits at the Fa’ahia site (Huahine, Society Islands), Pisonia Rockshelte r (Rota, Mariana Islands), and Railhunter Rockshelter (Tinian, Mariana Islands) contain partial sterna of the fully flightless † G. storrsolsoni and marginally flightless † G. tempatatus and † G. unnamed C , respectively. Other newly described species (e.g., † G. vekamatolu, † G. erstmayri, † G. epulare, † G. gracilitibia ) were inferred to be flightless based on th e condition of the humerus and coracoid, and on

PAGE 56

42 the relative proportions of wing and leg elements. As Livezey (1998, 2003) has shown, the degree of reduction of pectoral elements va ries greatly even among flightless species. Morphometric analyses of Gallirallus skeletons indicates that some species, such as G. australis, have greatly reduced carina sterni a nd wing elements, whereas other extant species known to be flightless, such as G. owstoni , seem to be near the threshold of flight (Figures 2-7, 2-11, 2-17). The graded re duction in energy expenditure observed among species of rails with varyi ng powers of flight (McNab a nd Ellis 2006) appears to be mirrored by the continuum of wing re duction among flightless species of Gallirallus . Whereas the evolution of flightlessne ss in many island species of birds is associated with gigantism (e.g., megapodes, ow ls, pigeons, parrots, and certain waterfowl and rails of the genus Porphyrio ), Gallirallus shows no consistent trend in overall body size among flightless species. Assuming that the ancestral, colonizi ng species were of similar size to extant volant G. philippensis or G. torquatus , some flightless species have become larger (e.g., G. australis, † G. unnamed A, G. woodfordi ), some have become smaller († G. wakensis , † G. ripleyi ), and some have changed very little in overall size († G. vekamatolu, G. owstoni, † G. storrsolsoni, † G. epulare, † G. gracilitibia, † G. roletti, † G. unnamed C ). Biogeography Of the 26 living and extinct species of Gallirallus sensu lato that have now been named, all but four are flight less species endemic to single islands or on multiple islands that were connected during the late Pleistocene period of lowered sea levels (Mayr and Diamond 2001, Steadman 2006, Taylor 1998). The two extant, volant species, G. torquatus and G. philippensis , are sympatric in Philippines, Sulawesi, and New Guinea, although the latter species is al so distributed south to Australia and New

PAGE 57

43 Zealand, and east in Oceania to Samoa. Flightless species of Gallirallus have evolved on nearly all major archipelagos in Oceania from the Ryukyu Islands of southern Japan, south to New Zealand’s Chatham Islands, and ea st to the Marquesas. The species from the Marquesas Islands represent the northwestern limit of Gallirallus . The very rich prehistoric record of birds (15,000+ bones fr om 12 archaeological sites) from Henderson Island (Pitcairn Group) has yielded no evidence of Gallirallus (Wragg 1995, Steadman and Olson 1985). Between the Cook Islands (where † G. ripleyi lived) and Henderson Island lie the Austral (Tubuai) Islands, wher e the single prehistori c landbird bone known represents an undescribe d, extinct species of Ptilinopus (Columbidae; Steadman 2006). Likewise, the well-studied and ve ry numerous fossils from dune s, lava tubes and lakes in the Hawaiian Islands, which have yi elded 7-10 flightless species of Porzana rails, lack evidence of Gallirallus (Olson and James 1991). No species of Gallirallus exist anywhere today in Eastern Polynesia, where all known species are flightless and extinct, and where only † G. pacificus of Tahiti is known to have survived to the period of European contact. Many archipeligoes had multliple flightless species of Gallirallus endemic to single islands . Newly described species from the Bismarck Islands († G. unnamed A of New Ireland), Mariana Islands († G. unnamed C of Tinian) and Society Islands († G. storrsolsoni of Huahine) double the diversity of flightless Gallirallus known from those archipelagoes. In the Bismarcks, G. insignis is still found on New Britain, which was never connected to New Ireland during periods of lowered sea-level in the Pleistocene. In the Marianas, G. owstoni was common on Guam but began to decline rapidly beginning in 1968 following the introduction to that island of the brown tree snake

PAGE 58

44 ( Boiga irregularis ). It now survives in captive breeding programs and has been reintroduced to nearby Rota, which is snake-free. In the Society Islands, † G. pacificus was discovered on Tahiti by naturalis ts from Captain James Cook’s 2nd voyage (1777), but no specimens exist, and the species is known only from a painting by Georg Forster. The plumage, soft-part colors, and bill shape make it clear that pacificus is correctly placed in Gallirallus . Measurements made by Storrs L. Olson in 1998 (pers. comm.) from Forster’s original, full-scale painting in the British Museum of Natural History make it clear that † G. pacificus was a much smaller rail than † G. storrsolsoni and was probably flightless. Huahine and Tahiti are seperated by a deep water trench and were never connected during periods of lowered sea-level. The possible former existence of † G. pacificus on Mehetia (Taylor 1998), a small island 110 km east-southeast of Tahiti, is unsubstantiated and doubtful. The rail bone s from archaeological sites on three Marquesan islands (Tahuata, Nuku Hiva, and Ua Huka) represent three new flightless species of Gallirallus , one endemic to each island. A probable fourth species, from Hiva Oa, requires more material before a description is feasible. Chronology of extinction – In Tonga, † G. vekamatolu is unknown from cultural contexts, but is replaced, even in the earl iest (2900-2700 yr BP) cultural strata, by G. philippensis , an extant, widespread, volant, non-for est rail. Thus it is unknown whether † G. vekamatolu survived, however precariously, un til the arrival of humans and was quickly exterminated, or if it went extinct “naturally” befo re ‘Eua was colonized. The fact that † G. vekamatolu would be unknown if not for the existence of much older (ca. 60,000-70,000 yr BP) pre-cultural strata at ‘Ana tu suggests that there may have been

PAGE 59

45 other prehistorically extinct species of Gallirallus in places such as Vanuatu and Samoa that lack pre-cultural fossil records. In most archipelagoes howe ver, flightless species of Gallirallus appear in the earliest cultural strata and become extinct s oon thereafter. In thes e cases extinction can unambiguously be attributed to direct e xploitation by humans and their introduced commensals. The large tibiotarsi (UF 56463, 56755) from Ha’afeva in Tonga represent an extinct, presumably flightless species just barely sampled by the archaeological excavations at the Mele Havea site. This rail must have disappeared very soon after human arrival, leaving behind only two bones from Levels 10 and 11 of Mele Havea, which bear Lapita-style potte ry and dates to 2900-2700 cal BP . Like the Tongoleleka site on Lifuka (Steadman et al. 2002a) the highly stratified Mele Havea site records the extirpation of many landbird species coincident with human arrival on the island (Kirchman and Steadman 2005). The extinction of flightless rails in the Marquesas also probably took place in prehistoric times. On Tahuata, all but one of the 22 specimens of † Gallirallus roletti were excavated from Levels G, GH, or H (Phases I and II) at Hanamiai, which represent the early occupation of the site at ca. 1000-700 yr BP (Rolett 1998). A single bone of † G. roletti is from Phase III (Level F), which dates to 700-550 yr BP. This may be the approximate time of extinction for † G. roletti , bones of which were not recovered in the younger Levels A-D (Phases IV, V). Similarly on Ua Huka, all of the bones of † G. gracilitibia were from strata dated to >800 yr BP (Steadman 1991). Specimens of G. philippensis are known from cultural contexts in archaeological sites throughout its current range including New Caledoni a (Balouet and Olson 1989), New Zealand (Worthy and Holdaway 2002), Va nuatu, and Tonga, and outside its current

PAGE 60

46 range in Marianas Islands (Kirchman a nd Steadman 2006c). It does not occur in precultural strata in New Caledonia and Tonga indicating that it may be a recent arrival in Oceania where the extermination of flightle ss competitors and the opening of forest habitats by humans may have enabled its colonization.

PAGE 61

47Table 2-1. Prehistoric sites in Oceania containing remains of Gallirallus . Age range refers to earliest and latest estimates based dates obtained from bones or charcoal. * = site s lacking radiocarbon dates but determined to post-date human arrival. See references cited for individual radiocar bon dates from specific sites and details regarding excavation methods and cultural context. NISP = number of identified Gallirallus specimens examined. Island, Archipelago Site Age Range (yr BP) References NISP New Ireland, Bismarck Islands Balof 1 and 2 14,000 Allen et al. (1988), St eadman et al (1999) 15 Matenbek 20,000-6,000 Allen et al. (1988), Steadman et al (1999) 3 Pankiwuk Cave 15,000-8,000 Allen et al. (1988), Steadman et al (1999) 1 Matenkupkum 14,000-10,000 Allen et al. (1988), Steadman et al (1999) 1 Buka, Solomon Islands Kilu Cave 28,000-7,000 Wickler and Spriggs 1988 18 Aguiguan, Mariana Islands Pisonia Rockshelter 1780 540 Steadman (1999) 219 Rota, Mariana Islands Mochong 2500 250 John Craib (pers. comm.) 3 Route 100 <2500* David Welch (pers. comm.) 11 Tinian, Mariana Islands Unai Chulu 3500 2800 Craib (1993) 855 Railhunter Rockshelter 2460 1880 Steadman (1999) 192 Saipan, Mariana Islands Chalan Piao 3550 3000 Butler (1994), Amesbury et al. (1996) 7 Unai Bapot 3550 3000 Mike Carlson (pers. comm.) 9 Efate, Vanuatu Arapus 3000 2800 Spriggs and Bedford (2001) 12 Foa, Tonga Faleloa 2900-2700 Burley (1999) 35 Ha’afeva, Tonga Mele Havea 2900-2700 Burley (1999) 16 Ha’ano, Tonga Pukotala 2900-2700 Burley (1999) 36 Lifuka, Tonga Tongaleleka 2900-2700 Steadman et al. (2002a) 7 Holopeka 2800-2500 Burley (1999) 5 Toumu’a Well 500-200 Burley (1999) 68 ‘Uiha, Tonga Vaipuna 2900-2700 Burley (1999) 21 Tongatapu, Tonga Ha’ateiho 2900-2700 Steadman et al (2002a) 32 “Eua, Tonga ‘Anatu Layer I 2900-500 Steadman (1993) 14 ‘Anatu Layers II and III 78,000 -60,000 Steadman (1993) 78 That Cave <2900* Steadman (unpublished) 18 Tupou Cave <2900* Steadman (unpublished) 6 ‘Anatuli <2900* Steadman (unpublished) 1 Midden Cave <2900* Steadman (unpublished) 2 Unnamed Cave <2900* Steadman (unpublished) 1 Huahine, Society Islands Fa’ahia 1250-750 Sinoto (1983) 50 Tahuata, Marquesas Islands Hanamaia 1000 Rolett (1998) 22 Hiva Oa, Marquesas Islands Hanatekua Rockshelter 1000 Bellwood (1972) 2 Nuku Hiva, Marquesas Islands Ha’atuatua 1000 Suggs (1961), Rolett (1998) 8 Ua Huka, Marquesas Islands Hane Dune 1350 Sinoto (1970) 21

PAGE 62

48Table 2-2. Mean, range, and sample size of skeletal measurements (mm) of †Gallirallus unnamed A and extant, flightless congeners from Near Oceania. M = male, F = female, U = unknown sex. † G. unnamed A U New Ireland, Bismarcks G. insignis F New Britain, Bismarcks G. rovianae M New Georgia, Solomons G. woodfordi M Isabel, Solomons G. woodfordi F Isabel, Solomons Humerus Distal width 8.8 1 7.2 1 8.2 1 9.1 8.9-9.2 3 8.8 8.6-8.9 3 Femur Shaft width 4.5 1 --4.9 1 5.2 5.2-5.3 3 4.8 4.6-5.1 3 Tibiotarsus Distal width 8.4 1 8.8 1 8.4 1 10.0 9.9-10.1 3 9.5 9.4-9.6 3 Shaft width 3.8 1 4.5 1 4.7 1 5.4 5.2-5.5 3 5.2 5.0-5.3 3

PAGE 63

49Table 2-3. Mean, range, and sample size of skeletal measurements (mm) of extinct (†) Gallirallus species from the Mariana Islands and select living and extinct congeners. M = male, F = female, U = unknown sex, --= can not be evaluated. G. philippensis is volant, all others are flight less. In all Tables, specimens of all available subspecies of G. australis and G. philippensis are combined, given that subspecifi c differences in size are much smalle r than size differences between males and females † G. unnamed C U Tinian, Marianas † G. unnamed B U Aguiguan, Marianas † G. unnamed D U Rota, Marianas G. philippensis M Oceania* G. philippensis F Oceania* G. owstoni M Guam G. owstoni F Guam Coracoid Total length 21.7 20.2-23.7 7 24.4 1 22.3 1 26.8 26.2-27.5 6 23.2 21.8-24.5 6 23.6 22.9-24.7 7 22.8 21.6-24.3 11 Sternal facet width 5.6 5.3-5.9 5 6.0 1 5.2 1 6.6 6.1-7.3 6 5.5 5.1-5.9 6 6.2 6.1-6.3 7 5.9 5.2-6.2 11 Minimum shaft width 2.4 2.2-2.7 9 2.7 1 2.5 1 3.0 2.7-3.2 6 2.8 2.5-3.0 6 2.6 2.5-2.8 7 2.6 2.3-3.0 11 Humerus Total length 41.9 41.2-42.7 2 ----51.3 49.2-53.0 6 45.6 41.5-49.3 6 47.5 45.3-51.1 7 44.7 42.9-46.8 11 Distal width 6.2 5.7-7.1 17 ----7.1 6.9-7.5 6 6.1 5.9-6.3 6 6.5 6.3-6.7 7 6.1 5.9-6.3 11 Minimum shaft width 2.8 2.6-3.1 13 ----3.3 3.2-3.4 5 2.9 2.7-3.3 6 2.8 2.6-3.1 7 2.7 2.6-3.0 11 Ulna Total length 36.0 34.6-38.0 3 ----43.9 41.4-44.8 6 39.2 35.2-43.6 6 39.1 36.8-41.5 7 36.6 34.8-38.7 11 Minimum shaft width 2.4 2.2-2.5 9 ----3.0 2.8-3.1 6 2.7 2.6-2.8 6 3.0 2.8-3.2 7 2.7 2.6-3.0 11

PAGE 64

50 Table 2-3 Continued † G. unnamed C U Tinian, Marianas † G. unnamed B U Aguiguan, Marianas † G. unnamed D U Rota, Marianas G. philippensis M Oceania* G. philippensis F Oceania* G. owstoni M Guam G. owstoni F Guam Carpometacarpus Total length 23.3 1 ----29.0 27.0-30.3 4 26.3 24.4-29.4 6 25.0 23.0-26.7 7 23.6 22.1-25.1 11 Femur Total length 50.8 47.5-53.2 5 ----53.8 51.3-54.9 6 47.3 45.0-49.7 6 56.0 54.0-59.2 7 52.4 49.2-54.6 11 Distal width 8.1 7.5-9.1 9 ----8.7 8.4-9.1 6 7.4 7.0-7.9 6 9.5 8.9-9.9 7 8.7 8.4-9.0 11 Minimum shaft width 3.4 3.1-3.5 7 ----4.0 3.7-4.2 6 3.3 3.1-3.5 6 3.8 3.3-4.0 7 3.6 3.4-3.9 11 Tarsometatarsus Total length 41.6 38.4-46.5 7 --47.4 1 48.0 45.3-49.0 6 42.5 40.8-45.2 6 51.9 48.7-54.9 7 48.5 46.2-50.6 11 Proximal width 6.2 5.6-7.3 17 5.7 5.6-5.7 2 6.6 6.4-6.8 2 7.0 6.7-7.2 6 6.1 5.8-6.6 6 7.6 7.0-8.2 7 7.1 6.7-7.3 11 Distal width 6.5 6.1-6.8 14 6.1 1 7.4 1 7.1 6.7-7.5 6 6.3 6.0-6.9 6 8.0 7.4-8.5 7 7.4 7.2-7.6 11 Minimum shaft width 3.1 2.7-3.7 17 3.1 3.0-3.1 2 3.4 3.3-3.5 2 3.5 3.1-3.6 6 3.0 2.8-3.3 6 3.8 3.5-4.0 7 3.4 3.2-3.8 11

PAGE 65

51Table 2-3 Continued G. woodfordi M Isabel, Solomons G. woodfordi F Isabel, Solomons † G. wakensis M Wake Is. † G. wakensis F Wake Is. † G. ripleyi U Mangaia, Cook Islands Coracoid Total length 33.0 31.8-33.7 3 30.9 30.5-31.4 3 17.3 1 16.2 1 15.9 1 Sternal facet width 9.7 9.4-10.2 3 9.2 9.0-9.3 3 5.3 1 5.0 1 5.4 1 Minimum shaft width 3.9 3.6-4.4 3 3.5 3.1-3.8 3 1.9 1 1.9 1 2.1 1 Humerus Total length 59.4 57.5-61.0 3 57.2 55.8-58.9 3 33.4 1 30.7 1 31.5 1 Distal width 9.1 8.9-9.2 3 8.8 8.6-8.9 3 5.1 1 4.7 1 4.8 1 Minimum shaft width 3.8 3.7-3.8 3 3.6 3.5-3.7 3 2.0 1 1.8 1 2.1 1 Ulna Total lenght 51.6 50.7-52.8 3 49.5 47.8-50.6 3 27.6 1 25.0 1 22.9 22.6-23.1 2 Minimum shaft width 3.6 3.5-3.8 3 3.3 3.2-3.5 3 1.9 1 1.9 1 1.8 1.7-1.8 2 Carpometacarpus Total length 32.7 32.6-32.8 3 32.1 30.0-33.5 3 18.5 1 16.4 1 14.6 14.5-14.7 2 Femur Total length 72.1 72.0-72.3 3 69.6 69.0-70.3 3 37.9 1 36.7 1 39.8 1

PAGE 66

52Table 2-3 Continued G. woodfordi M Isabel, Solomons G. woodfordi F Isabel, Solomons † G. wakensis M Wake Is. † G. wakensis F Wake Is. † G. ripleyi U Mangaia, Cook Islands Distal width 13.3 13.2-13.4 3 12.7 12.5-12.9 3 6.9 1 6.1 1 6.7 1 Minimum shaft width 5.2 5.2-5.3 3 4.8 4.6-5.1 3 2.8 1 2.6 1 3.0 1 Tarsometatarsus Total length 70.9 69.8-72.3 3 68.7 66.7-72.0 3 35.6 1 33.3 1 33.1 1 Proximal width 10.5 10.2-10.7 3 10.0 9.9-10.1 3 6.0 1 5.5 1 5.7 1 Distal width 10.8 10.7-11.0 3 10.5 10.5-10.6 3 6.2 1 5.9 1 6.4 6.3-6.5 3 Minimum shaft width 4.9 4.8-5.0 3 4.6 4.4-4.7 3 2.9 1 2.5 1 3.0 2.8-3.1 2

PAGE 67

53Table 2-4. Correlation coefficients from a principal co mponents analysis of skel etal measurements from † Gallirallus unnamed C and three volant and five flightless Gallirallus species (Figure 2-8). Skeletal measurements PC1 PC2 PC3 PC4 Coracoid, length .970-.232.028.014 Coracoid, sternal facet width .926.222.260.129 Coracoid, shaft width .977.095.042-.109 Humerus, length .954-.289.028-.036 Humerus, distal width .986-.109.040.042 Humerus, shaft width .977-.106-.001-.166 Ulna, length .857-.499.069-.063 Ulna, proximal width .984-.099-.094.082 Ulna, shaft width .976-.131-.073-.041 Carpometacarpus, length .869-.484-.019.086 Femur, length .987.076-.005-.086 Femur, distal width .958.282-.030-.006 Femur, shaft width .965.014-.235.107 Tarsometatarsus, length .983-.025.059.073 Tarsometatarsus, proximal width .926.368-.024.063 Tarsometatarsus, dist al width .887.459-.006.004 Tarsometatarsus, shaft width .876.462-.025-.091 Percent total variance explained 89.42 8.08 0.90 0.68

PAGE 68

54Table 2-5. Skeletal measurements (mm) of † Gallirallus vekamatolu, † G. storrsolsoni , and seven congeners, with mean, range, and sample size. F, female; M, male; U, sex unknown. . † G. vekamatolu U † G. storrsolsoni U G. owstoni M G. owstoni F G. australis M G. australis F † G. ripleyi U Cranium length --33.7 32.4-34.8 4 33.3 32.8-34.2 4 31.6 30.9-32.3 10 42.9 42.2-43.5 2 40.0 39.0-41.5 4 --Rostrum length --38.8 36.5-41.0 2 40.7 36.9-43.3 6 37.1 35.5-39.3 10 52.0 50.6-53.3 2 46.0 40.0-50.4 4 --Sternal carina depth --9.1 1 10.7 9.7-11.6 7 10.4 9.9-11.4 10 12.0 11.5-12.5 2 10.6 9.9-12.0 4 --Sternum width at coracoids --14.6 14.5-14.8 2 11.7 10.7-12.2 7 11.3 9.6-11.9 9 20.9 20.3-21.5 2 19.2 18.6-19.9 4 --Humerus shaft width --2.2 1 2.8 2.6-3.1 7 2.7 2.6-3.0 11 4.3 3.9-4.6 3 3.8 3.5-4.2 4 2.1 1 Ulna length 41.3 1 37.5 1 39.1 36.8-41.5 7 36.6 34.8-38.7 11 43.7 43.0-44.2 3 38.2 36.0-41.4 4 22.9 22.6-23.1 2 Ulna proximal width 5.4 1 4.8 1 4.6 4.3-4.8 7 4.3 4.1-4.5 11 7.0 6.5-7.3 3 6.0 5.5-6.8 4 2.8 2.7-2.9 2 Femur length --49.4 1 56.0 54.0-59.2 7 52.4 49.2-54.6 11 80.6 78.7-82.3 3 69.9 66.1-73.4 4 39.8 1 Femur distal width 10.9 10.7-11.0 2 9.2 1 9.5 8.9-9.9 7 8.7 8.4-9.0 11 16.5 15.9-17.6 3 14.3 13.2-15.0 4 6.7 1 Tibiotarsus length -68.6 1 79.9 76.1-84.6 7 75.5 72.9-79.8 11 117.4 116.6-118.7 3 100.0 93.4-105.0 4 --

PAGE 69

55Table 2-5 Continued † G. vekamatolu U † G. storrsolsoni U G. owstoni M G. owstoni F G. australis M G. australis F † G. ripleyi U Tibiotarsus distal width 8.0 7.9-8.0 2 6.9 1 7.4 6.8-7.9 7 6.8 6.5-7.1 11 12.5 12.1-13.1 3 10.8 10.4-11.2 4 6.0 5.6-6.3 3 Tarsometatars us total length 45.1 1 --51.9 48.7-54.9 7 48.5 46.2-50.6 11 67.8 66.5-68.5 3 59.2 56.3-62.2 4 33.1 1 Tarsometatars us proximal width 8.3 7.8-8.7 2 7.4 1 7.6 7.0-8.2 7 7.9 6.7-7.3 11 12.6 12.2-13.4 3 11.0 10.7-11.7 4 5.7 1 Tarsometatars us shaft width 4.3 4.0-4.6 2 4.1 1 3.8 3.5-4.0 7 3.4 3.2-3.8 11 6.1 5.6-6.4 3 5.4 5.2-5.6 4 3.0 2.8-3.1 2 Tarsometatars us shaft depth 3.2 3.0-3.3 2 2.8 1 2.8 2.5-3.0 7 2.5 2.4-2.8 11 4.8 4.5-5.0 3 4.1 3.8-4.6 4 2.2 2.1-2.3 2 G. philippensis M G. philippensis F G. striatus M G. torquatus M G. torquatus F G. woodfordi M G. woodfordi F Cranium length 32.1 31.0-32.9 4 29.9 28.4-31.0 6 28.7 28.2-29.0 4 33.9 32.4-35.2 5 31.0 1 39.8 39.0-40.5 3 39.0 38.2-39.5 3 Rostrum length 36.0 33.5-38.6 4 29.5 27.0-31.6 6 35.7 33.0-38.1 4 42.2 39.7-45.8 4 39.4 1 49.1 47.8-49.9 3 44.8 44.6-45.0 3 Sternal carina depth 12.7 12.0-13.5 6 12.3 11.6-13.5 6 12.2 11.6-13.0 3 13.1 12.2-14.1 5 10.9 1 13.2 12.6-13.8 3 12.6 12.2-13.4 3 Sternum width at coracoids 11.1 9.9-12.0 6 10.0 9.4-10.8 6 9.0 8.4-10.0 4 11.9 11.1-12.4 5 11.7 1 18.8 18.6-19.0 3 17.7 16.5-18.8 3

PAGE 70

56Table 2-5 Continued G. philippensis M G. philippensis F G. striatus M G. torquatus M G. torquatus F G. woodfordi M G. woodfordi F Humerus shaft width 3.3 3.2-3.4 5 2.9 2.7-3.3 6 2.6 2.5-2.7 4 3.2 3.0-3.4 5 2.9 1 3.8 3.7-3.8 3 3.6 3.5-3.7 3 Ulna length 43.9 41.4-44.8 6 39.2 35.2-43.6 6 36.2 33.7-37.9 4 44.0 43.0-45.3 5 40.0 1 51.6 50.7-52.8 3 49.5 47.5-50.6 3 Ulna proximal width 5.0 4.7-5.3 6 4.3 4.1-4.4 6 4.0 3.8-4.1 4 5.1 4.8-5.4 5 5.0 1 6.4 6.3-6.5 3 5.9 5.9-6.0 3 Femur length 53.8 51.3-54.9 6 47.3 45.0-49.7 6 45.1 42.2-46.9 4 55.5 53.3-57.5 5 50.6 1 72.1 72.0-72.3 3 69.6 69.0-70.3 3 Femur distal width 8.7 8.4-9.1 6 7.4 7.0-7.9 6 45.1 42.2-46.9 4 9.3 8.6-9.9 5 8.5 1 13.3 13.2-13.4 3 12.7 12.5-12.9 3 Tibiotarsus length 76.9 71.7-80.3 6 66.7 63.0-70.6 6 62.1 58.2-64.3 4 84.2 80.9-87.5 5 74.6 1 105.0 100.6-108.8 3 102.8 99.0-105.8 3 Tibiotarsus distal width 6.9 6.5-7.2 6 6.1 5.6-6.7 6 5.3 5.1-5.6 4 7.2 6.7-7.6 5 6.6 1 10.0 9.9-10.1 3 9.5 9.4-9.6 3 Tarsometarsus total length 48.0 45.3-49.0 6 42.5 40.8-45.2 6 39.0 36.7-39.9 4 54.0 52.0-56.6 5 46.4 1 70.9 69.8-72.3 3 68.7 66.7-72.0 3 Tarsometatarsu s proximal width 7.0 6.7-7.2 6 6.1 5.8-6.6 6 5.4 5.3-5.5 4 7.4 7.0-7.9 5 6.6 1 10.5 10.2-10.7 3 10.0 9.9-10.1 3 Tarsometatarsu s shaft width 3.5 3.1-3.6 6 3.0 2.8-3.3 6 2.7 2.5-2.9 4 3.6 3.3-3.8 5 3.3 1 4.9 4.8-5.0 3 4.6 4.4-4.7 3 Tarsometatarsu s shaft depth 2.8 2.6-3.0 6 2.5 2.3-2.8 6 2.1 2.0-2.2 4 2.7 2.2-2.9 5 2.6 1 4.0 3.9-4.2 3 3.8 3.7-4.0 3

PAGE 71

57 Table 2-6. Correlation coeffici ents from a principal components analysis of 14 skeletal measurements from † Gallirallus vekamatolu and seven congeners (Figure 211). Skeletal measurements PC 1PC 2PC 3PC 4 Coracoid, sternal facet width .934.275-.089-.160 Humerus, distal width .843.430-.294-.035 Ulna, length .653.715.203.135 Ulna, proximal depth .974.146-.141.053 Ulna, shaft width .983-.062-.105-.107 Femur, distal width .988-.117-.041.049 Tibiotarsus, distal width .992-.103-.030.026 Tibiotarsus, medial condyle depth .987-.136.044.019 Tibiotarsus, lateral c ondyle depth .974-.175.120-.034 Tarsometatarsus, total length .885.070.445-.101 Tarsometatarsus, proximal width .994-.095-.027.023 Tarsometatarsus, dist al width .980-.191.015-.025 Tarsometatarsus, minimum shaft width .956-.239-.024.154 Tarsometatarsus, minimum shaft depth .957-.206-.010.029 Percent total variance explained 88.367.29 2.75 0.71 Table 2-7. Correlation coeffici ents from a principal components analysis of 16 skeletal measurements from † Gallirallus storrsolsoni and seven congeners (Figure 217). Skeletal measurements PC1 PC2 PC3 PC4 Sternum, width at coracoids .265-.093-.045.019 Sternum, keel depth .097.177.037.027 Humerus, shaft width .219.089.077-.014 Ulna, total length .146.110-.035.003 Ulna, proximal width .217.024-.013.023 Ulna, shaft width .203.049.025.030 Carpometacarpus, intermetacarpal space length .163.165-.098-.022 Femur, total length .224.021.021-.002 Femur, distal width .293-.047.004-.002 Femur, shaft width .255-.024-.002.015 Tibiotarsus, total length .222.014.023-.031 Tibiotarsus, distal width .263-.048.018-.011 Tibiotarsus, shaft width .263-.012.002-.030 Tarsometatarsus, proximal width .258-.061.004-.013 Tarsometatarsus, shaft width .253-.067-.029.018 Tarsometatarsus, shaft depth .262-.037.004.004 Percent total variance explained 85.81 10.65 2.33 0.61

PAGE 72

58Table 2-8. Skeletal measurements (mm) of † Gallirallus roletti and select congeners, with mean, range , and sample size. F, female; M, male; U, sex unknown; ---, can not be evaluated. † G. roletti U † G. storrsolsoni U G. owstoni M G. owstoni F G. australis M G. australis F † G. ripleyi U Rostrum: height at anterior margin of nares 5.2 1 3.7 3.5-3.9 2 3.4 3.2-3.8 5 3..3 3.1-3.5 4 5.3 5.1-5.4 3 4.7 4.4-5.0 4 2.9 1 Rostrum: width at anterior margin of nares 4.2 1 4.1 4.0-4.2 2 4.4 4.0-4.7 5 4.2 3.8-4.5 4 5.5 5.1-5.7 3 5.0 4.5-5.5 4 2.8 1 Femur: length 54.5 1 49.4 1 56.0 54.0-59.2 7 52.4 49.2-54.6 11 80.6 78.7-82.3 3 69.9 66.1-73.4 4 39.8 1 Femur: distal width 9.8 1 9.2 1 9.5 8.9-9.9 7 8.7 8.4-9.0 11 16.5 15.9-17.6 3 14.3 13.2-15.0 4 6.7 1 Femur: min shaft width 4.0 4.0 2 3.9 1 3.8 3.3-4.0 7 3.6 3.4-3.9 11 6.7 6.5-6.9 3 5.6 5.1-5.9 4 3.0 1 Tibiotarsus: length 79.5 1 68.6 1 79.9 76.1-84.6 7 75.5 72.9-79.8 11 117.4 116.6-118.7 3 100.0 93.4-105.0 4 --Tibiotarsus: distal width 7.8 7.7-7.9 2 6.9 1 7.4 6.8-7.9 7 6.8 6.5-7.1 11 12.5 12.1-13.1 3 10.8 10.4-11.2 4 6.0 5.6-6.3 3 Tarsometatarsus: proximal width 7.9 1 7.4 1 7.6 7.0-8.2 7 7.0 6.7-7.3 11 12.6 12.2-13.4 3 11.0 10.7-11.7 4 5.7 1 Tarsometatarsus: distal width 8.7 1 --8.0 7.2-8.5 7 7.4 7.2-7.8 11 14.0 13.6-14.3 3 12.1 11.2-13.0 4 6.4 6.3-6.5 3 Tarsometatarsus: shaft width 3.5-3.9 3.7 2 4.1 1 3.8 3.5-4.0 7 3.4 3.2-3.8 11 6.1 5.6-6.4 3 5.4 5.2-5.6 4 3.0 2.8-3.1 2

PAGE 73

59Table 2-8 Continued † G. roletti U † G. storrsolsoni U G. owstoni M G. owstoni F G. australis M G. australis F † G. ripleyi U Tarsometatarsus: shaft depth 2.7 2.5-2.8 2 2.8 1 2.8 2.5-3.0 7 2.5 2.4-2.8 11 4.8 4.5-5.0 3 4.1 3.8-4.6 4 2.2 2.1-2.3 2 G. philippensis M G. philippensis F G. striatus M G. torquatus M G. torquatus F G. woodfordi M G. woodfordi F Rostrum: height at anterior margin of nares 3.1 2.8-3.4 5 2.8 2.6-2.9 5 2.8 1 3.2 3.1-3.3 2 --4.3 4.2-4.6 3 4.2 3.9-4.4 3 Rostrum: width at anterior margin of nares 3.7 3.4-4.2 5 3.2 3.1-3.4 5 3.1 1 3.4 3.4-3.4 2 --4.7 4.6-4.9 3 4.6 4.5-4.8 3 Femur: length 53.8 51.3-54.9 6 47.3 45.0-49.7 6 45.1 42.2-46.9 4 55.5 53.3-57.5 5 50.6 1 72.1 72.0-72.3 3 69.6 69.0-70.3 3 Femur: distal width 8.7 8.4-9.1 6 7.4 7.0-7.9 6 45.1 42.2-46.9 4 9.3 8.6-9.9 5 8.5 1 13.3 13.2-13.4 3 12.7 12.5-12.9 3 Tibiotarsus: length 76.9 71.7-80.3 6 66.7 63.0-70.6 6 62.1 58.2-64.3 4 84.2 80.9-87.5 5 74.6 1 105.0 100.6-108.8 3 102.8 99.0-105.8 3 Tibiotarsus: distal width 6.9 6.5-7.2 6 6.1 5.6-6.7 6 5.3 5.1-5.6 4 7.2 6.7-7.6 5 6.6 1 10.0 9.9-10.1 3 9.5 9.4-9.6 3 Tarsometatarsus: proximal width 7.0 6.7-7.2 6 6.1 5.8-6.6 6 5.4 5.3-5.5 4 7.4 7.0-7.9 5 6.6 1 10.5 10.2-10.7 3 10.0 9.9-10.1 3 Tarsometatarsus shaft width 3.5 3.1-3.6 6 3.0 2.8-3.3 6 2.7 2.5-2.9 4 3.6 3.3-3.8 5 3.3 1 4.9 4.8-5.0 3 4.6 4.4-4.7 3 Tarsometatarsus: shaft depth 2.8 2.6-3.0 6 2.5 2.3-2.8 6 2.1 2.0-2.2 4 2.7 2.2-2.9 5 2.6 1 4.0 3.9-4.2 3 3.8 3.7-4.0 3

PAGE 74

60Table 2-9. Skeletal measurements (mm) of † Gallirallus gracilitibia and select congeners with me an, range, and sample size. F, female; M, male; U, sex unknown; ---, can not be evaluated. † G. gracilitibia U † G. storrsolsoni U G. philippensis M G. philippensis F G. owstoni M G. owstoni F G. torquatus M G. torquatus F Humerus: shaft width 2.2 2.1-2.3 2 2.2 1 3.3 3.2-3.4 5 2.9 2.7-3.3 6 2.8 2.6-3.1 7 2.7 2.6-3.0 11 3.2 3.0-3.4 5 2.9 1 Tibiotarsus: shaft width 3.0 2.9-3.3 8 3.4 3.2-3.7 3 3.6 3.3-4.0 6 3.1 2.9-3.3 6 4.1 3.8-4.5 7 3.8 3.4-4.1 11 3.8 3.4-4.0 5 3.3 1 Tibiotarsus: length to fibular crest* 54.5 1 50.2 50.2-50.2 2 52.8 51.9-55.3 5 47.1 43.3-51.6 6 57.8 55.2-61.2 7 54.4 52.9-55.9 11 59.3-61.5 2 59.0 1 Tibiotarsus: distal width 6.4 6.3-6.5 4 6.9 1 6.9 6.5-7.2 6 6.1 5.6-6.7 6 7.4 6.8-7.9 7 6.8 6.5-7.1 11 7.2 6.7-7.6 5 6.6 1 *distal end of bone to dist al edge of fibular crest

PAGE 75

61Table 2-10. Skeletal measurements (mm) of † Gallirallus epulare and select congeners with mean, range, and sample size. F, female; M, male; U, sex unknown; ---, can not be evaluated. † G. epulare U † G. storrsolsoni U G. philippensis M G. philippensis F G. owstoni M G. owstoni F G. torquatus M G. torquatus F Ulna: length 25.5* 1 37.5 1 43.9 41.4-44.8 6 39.2 35.2-43.6 6 39.1 36.8-41.5 7 36.6 34.8-38.7 11 44.0 43.0-45.3 5 40.0 1 Ulna: shaft width 2.0 1 2.7 1 3.0 2.8-3.1 6 2.7 2.6-2.8 6 3.0 2.8-3.2 7 2.7 2.6-3.0 11 3.0 2.8-3.1 5 2.8 1 Tibiotarsus: shaft width 3.4 1 3.4 3.2-3.7 3 3.6 3.3-4.0 6 3.1 2.9-3.3 6 4.1 3.8-4.5 7 3.8 3.4-4.1 11 3.8 3.4-4.0 5 3.3 1 Tarsometatarsus: shaft width 3.5 1 4.1 1 3.5 3.1-3.6 6 3.0 2.8-3.3 6 3.8 3.5-4.0 7 3.4 3.2-3.8 11 3.6 3.3-3.8 5 3.3 1 Tarsometatarsus: shaft depth 2.4 1 2.8 1 2.8 2.6-3.0 6 2.5 2.3-2.8 6 2.8 2.5-3.0 7 2.5 2.4-2.8 11 2.7 2.2-2.9 5 2.6 1 Tarsometatarsus: distal width 6.5 1 --7.1 6.7-7.5 6 6.3 6.0-6.9 6 8.0 7.2-8.5 7 7.4 7.2-7.6 11 7.8 7.3-8.4 5 7.0 1 *estimated

PAGE 76

62 Figure 2-1. Map of Western Oceania, indica ting islands with archaeological sites containing Gallirallus fossils.

PAGE 77

63 Figure 2-2. Humeri of Gallirallus species known from Near Oceania in anterior (upper) and posterior (lower) aspects. A. † G. unnamed A UF 62983, new species from New Ireland. B. G. insignis AMNH 27136, from New Britain. C. G. philippensis UF 39855, widespread in Oceania, specimen from Tutuila, American Samoa. D. G. rovianae AMNH CEF878, from New Georgia. E. G. woodfordi UF 39409, Isabel. Scale = 40 mm.

PAGE 78

64 Figure 2-3. Distal tibiotarsi of of Gallirallus species known from Near Oceania in anterior (upper) and posterior (lower) aspects. A. † G. unnamed A UF 62977, new species from New Ireland. B. G. insignis AMNH 27136, New Britain. C. G. philippensis UF 39855, Tutuila, American Samoa. D. G. rovianae AMNH CEF878, New Georgia, Solomon Islands. E. G. woodfordi UF 39409, Isabel, Solomon Islands. Scale = 20 mm.

PAGE 79

65 Figure 2-4. The coracoids (upper row) and carpometacarpi (lower row) of Gallirallus rails from the Mariana Islands. A. † G. unnamed B new species, UF 62934, UF 62917, Aguiguan. B. † G. unnamed D new species, UF 63302, UF 63299, Rota. C. † G. unnamed C new species, UF 60110, UF 63372, Tinian. D. G. owstoni , UF 39921, Guam. E. G . philippensis , UF 39855, Tutuila, American Samoa. Scale = 30 mm.

PAGE 80

66 Figure 2-5. The sternum of Gallirallus in lateral (left) and vent ral (right) aspects. A. † G. unnamed B new species, UF 62892/62893, Aguiguan. B. † G. unnamed C new species, UF 60142, Tinian. C. G. owstoni , UF 39920, Guam. D. G. philippensis , UF 43222, Efate, Vanuatu. Scale = 40 mm.

PAGE 81

67 Figure 2-6. Wing elements of Gallirallus , including caudal (uppe r row) and cranial (lower row) aspects of the humerus (A-C ), and ventral (upper row) and dorsal (lower row) aspects of th e ulna (D-F). Shown are † G. unnamed C new species, Tinian (A,UF 63419; D, UF 63429), G. owstoni , Guam (B, E, UF 39920), and G. philippensis , Tutuila, American Samoa (C, F, UF 39855). Scale = 40 mm.

PAGE 82

68 Figure 2-7. Size variation in femora ( upper) and tarsometatarsi (lower) of † Gallirallus unnamed C, Tinian (A, B), G. owstoni, Guam (C, D), and G. philippensis, Tutuila, American Samoa (E, F). A, femur UF 60193 and tarsometatarsus UF 63627, and B, femur UF 60115/60116 and tarsometarsus UF 63470 all of unknown sex. C, female UF 39220 and D, male UF 39221. E, female UF 43222 and F, male UF 39855. Scale = 40 mm.

PAGE 83

69 Figure 2-8. Plot of † Gallirallus Unnamed C and eight other species of Gallirallus on the first two principal components of postcranial skeletal measurements (see Table 2-4). Filled symbols are flightless species, hollow symbols are volant species. The line represents an hypothesi zed threshold for flightlessness. Figure 2-9. Tonga, showing islands with ar chaeological sites cont aining fossil bones of Gallirallus .

PAGE 84

70 Figure 2-10. Gallirallus tarsometatarsi in acrotarsial (uppe r) and plantar (lower) aspects. A, † G. vekamatolu holotype UF 51991, ‘Eua, Tonga. B. † G. ripleyi holotype USNM 402895, Mangaia, Cook Islands. C. G. philippensis UF 43221, Efate. D. G. owstoni UF 39920, Guam. E. G. woodfordi UF 39574, Isabel. F. G. australis UF 33524, New Zealand. Scale = 30 mm.

PAGE 85

71 . Figure 2-11. The rostrum (A, B), humerus (C , D), ulna (E, F), carpometacarpus (G, H), femur (I, J), and tibio tarsus (K, L) of † Gallirallus vekamatolu, ‘Eua (paratypes: A, UF 52292; C, UF 52707; E, UF 51734; G, UF 51743; I, UF 52020; K, UF 51729) and G. philippensis (B, D, F, H, J, L all from UF 43221, Efate). Scale = 30 mm.

PAGE 86

72 Figure 2-12. Plot of † G. vekamatolu and seven other species of Gallirallus on the first two principal components of 14 post-cran ial skeletal characters (see Table 26). Filled symbols are flightless species, hollow symbols are volant species. The line represents an hypothesized threshold for flightlessness.

PAGE 87

73 Figure 2-13. Society Islands w ith inset of Huahine, showi ng the location of the FaÂ’ahia archaeological site.

PAGE 88

74 Figure 2-14. Lateral (left) and dorsa l (right) views of the skulls of Gallirallus. A. G. philippensis , UWBM 42866, ‘Eua, Tonga. B. † G. storrsolsoni , holotype, BPBM 166036, Huahine, Society Islands. C G. owstoni , UF 39918, Guam. D. G. torquatus , UMMZ 228275 Luzon, Philippines. E. G. woodfordi , UF 39399, Isabel, Solomon Islands. Scale = 50 mm.

PAGE 89

75 Figure 2-15. Ventral (upper) and late ral (lower) views of the sterna of Gallirallus . A. G. philippensis , UWBM 42866, ‘Eua. B. † G. storrsolsoni , BPBM 166027, Huahine. C. G. owstoni , UF 39918, Guam. Scale = 50 mm.

PAGE 90

76 Figure 2-16. Humeri (A-C), ulnae (D-F), carpometacarpi (G-I), tibiotarsi (J-L), and tarsometatarsi (M-O) of Gallirallus philippensis (A., D., G., J., M., all from UF 39885, Tutuila), † G. storrsolsoni (B. BPBM 166022, E. BPBM 166033, H. BPBM 168165, K. BPBM 166023, N. DAPT 7), and G. owstoni (C., F., I., L., O., all from UF 42968, Guam). Scale = 50 mm.

PAGE 91

77 Figure 2-17. Plot of mean scores for eight species of Gallirallus on the first two principal components of 16 post-cranial skeletal m easurements(see Table 2-7). Hollow symbols are volant species; filled sym bols are flightless species. The line indicates an hypothesized thre shold for flightlessness.

PAGE 92

78 Figure 2-18. Marquesas Islands, showing the location of the four archaeological sites discussed in this chapter.

PAGE 93

79 Figure 2-19. Holotype of A. † Gallirallus roletti, consisting of associated femur (BPBM 166447), tibiotarsus (BPBM 166446), ta rsometatarsus (BPBM 166448), and two pedal phalanges (BPBM 166449, 166450) , top to bottom, respectively, Tahuata, Marquesas Islands. Shown w ith the same elements from B. G. philippensis (UF 39855, Tutuila) and C. G. owstoni (UF 39921, Guam). Scale = 50 mm.

PAGE 94

80 Figure 2-20. Rostra of A. † Gallirallus roletti (BPBM 166456, Tahuata), B. † G. storrsolsoni (holotype BPBM 166036, Huahine), C. G. philippensis (UF 39855, Tutuila,), and D. G. owstoni (UF 39921, Guam) in lateral (left) and dorsal (right) aspects. Scale = 20 mm.

PAGE 95

81 Figure 2-21. Femora of A. † Gallirallus roletti (holotype BPBM 166447, Tahuata), B. † Gallirallus undescribed sp . (BPBM 168539, Hiva Oa), C. † G. gracilitibia (BPBM 176974, Ua Huka), D. † G . epulare (BPBM 181659 Nuku Hiva), and E. G. philippensis (UF 39855, Tutuila, Samoa) in dorsal (top) and ventral (bottom) aspects. Scale = 50 mm.

PAGE 96

82 Figure 2-22. Tibi otarsi of A. † Gallirallus roletti (holotype BPBM 166446, Tahuata), B. † G. storrsolsoni (BPBM 166023, Huahine), C, † G. epulare (BPBM 181661, Nuku Hiva), D, † G. gracilitibia (holotype BPBM 166013/176387, Ua Huka), E, G. philippensis (UF 39855, Tutuila), F, and G. owstoni (UF 39921, Guam,) in dorsal (top) and ventral (bottom) aspects. Scale = 50 mm.

PAGE 97

83 Figure 2-23. Tarsometatarsi of A. † Gallirallus roletti (holotype BPBM 166448, Tahuata,), B. † G. storrsolsoni (DAPT 7, Huahine), C. † G. epulare (BPBM 167119, Nuku Hiva), D. G. philippensis (UF 39855, Tutuila), and E. G. owstoni (UF 39921, Guam) in acrotarsial (top) and plantar (bottom) aspects. Scale = 50 mm

PAGE 98

84 Figure 2-24. Scapulae of A. † Gallirallus undescribed sp. (BPBM 165655, Hiva Oa, Marquesas Islands), B. G. philippensis (UF 39855, Tutuila), and C. G. owstoni (UF 39921, Guam) in dorsal aspect. Scale = 50 mm.

PAGE 99

85 Figure 2-25. Humeri of A. † Gallirallus storrsolsoni (BPBM 166022, Huahine), B. † G. epulare (BPBM 181657, Nuku Hiva), C. † G. gracilitibia (BPBM 163130U, Huka), D. G. philippensis (UF 39855, Tutuila), and E. G. owstoni (UF 39921, Guam) in caudal (top) and cranial (bottom) aspects. Scale = 50 mm.

PAGE 100

86 Figure 2-26. Ulnae of A. † Gallirallus storrsolsoni (BPBM 166033, Huahine), B. † G. epulare (holotype BPBM 181658, Nuku Hiva), C. G. philippensis (UF 39855, Tutuila), and D. G. owstoni (UF 39921, Guam) in dorsal aspect. Scale = 50 mm.

PAGE 101

87 CHAPTER 3 PARALLEL EVOLUTION OF FLI GHTLESSNESS IN RAILS FROM OCEANIA: PHYLOGENY AND SYSTEMATICS OF Gallirallus Introduction The evolution of flightless species within primarily volant groups of animals (i.e., secondarily flightless sp ecies) has attracted th e attention of evolu tionists since Darwin (1859), who regarded the phenomenon as a co mpelling example of adaptation by natural selection. Many insects and birds have b ecome flightless, and many comparative and experimental studies of these cases have implicated selection for reduced energy expenditure as the force drivi ng the evolution of flightlessn ess. Strong support for the importance of energy conservation in the evol ution of flightless bi rds comes from McNab (1994a, 2002) and McNab and Ellis (2006) who demonstrated decreased basal metabolic rates and pectoral muscle mass in flightless versus volant birds. The flightless condition has evolved independently in at least 15 predominantly volant avian families (not including penguins and ratites). A few a quatic species have become flightless on continents (three grebes, one duck), but nearly all flightle ss birds evolved on predatorfree islands, highlighting the importance of pr edator avoidance in the maintenance of flight structures in birds. Birds are pa rticularly sensitive to selection for energy conservation because of their large anatomical and metabolic investment in the energydemanding muscles of the flight apparatus, which may comprise up to 25% of total body mass.

PAGE 102

88 Insular flightlessness is especially prevalent in members of the avian family Rallidae, in which 31 of the approximately 150 historically known species are or were flightless species endemic to single islands, or to islands connected in the Pleistocene during periods of lowered sea-levels (Taylor 199 8). In addition, scores of prehistorically extinct flightless species are known from fossil bones from archaeological and paleontological sites on Eart h’s oceanic islands (Stead man 1995, 2006, Olson and James 1991, Olson 1973a, Worthy and Holdaway 2002). This record re veals extensive Pleistocene and Holocene extinction of fli ghtless rails following colonization by humans and their commensals (primarily dogs, pigs, and rats). Losses were greatest in Oceania (Figure 1-1), where flightless rail species ma y have numbered in the “hundreds” (Livezey 2003) or perhaps as many as 1600 (Steadman 2006). No islands in Oceania except for New Caledonia and New Zealand were ever c onnected to continen tal land masses, so each flightless species likely represents an inde pendent transition to flightlessness. Thus, the flightless rails of Oceania are the larg est and yet least studied example of adaptive radiation in island birds, and may also be the most species-rich example of convergent evolution among vertebrates (M cNab and Ellis 2006). The high incidence of insular flightlessne ss in rails may be due to morphological, ecological, behavioral, and developmental char acteristics that predis pose them to evolve the flightless condition. Volant rails are char acterized by relatively sm all pectoral muscle masses, which average only 12.9% of total body mass (Livezey 2003). All rails nest and forage on the ground and rarely fly even to es cape potential predat ors, from which they typically run for cover. In developing rail s, the sternum does not ossify until well after hatching and all rails are effectively flightless as chicks and as nearly adult-sized birds,

PAGE 103

89 suggesting that neotenic delay in the developm ent of the flight apparatus, perhaps under the control of one or a few regulatory genes, may result in fully f lightless adults (Olson 1973a, Feduccia 2000). Paradoxically, these weak, reluctant flyers are great dispersers as evidenced by the frequency with which they tu rn up as vagrants outside of their regular ranges, and by their occurrence on even th e most isolated of oceanic islands. Biogeographic evidence also indicates that in sular flightlessness ma y evolve rapidly in rails; flightless species closely resemble wide spread, polytypic, volant species. Flightless species are thought to derive fr om populations of still extant volant congeners. These facts led Olson (1973a) to propose a model for the evolution of Ocean ic rail diversity in which multiple closely related species may descend from one or a few, perhaps still extant ancestral species as natural sel ection rapidly pushes one island-colonizing population after another down parallel paths to flightlessness. Phylogenetic tests of OlsonÂ’s rapid parallel speciation m odel of rail evolution based on morphological data have been hamp ered because of extensive homoplasy characterizes convergent evolution of f lightlessness (Olson 1973b, Livezey 1998, 2003). Osteological differences are slight among species within the few genera that have produced nearly all the flightless species in the Pacific ( Porphyrio, Porzana, Gallirallus , and, to a lesser extent, Gallinula ), and species-level relations hips in these groups remain poorly resolved (Livezey 2003). Recent phyl ogenetic analyses of mitochondrial DNA that have included ancient DNA from extinct flightless species have clarified rates and patterns of evolution in the crakes ( Porzana, Poliolimnas, and Amaurornis; Slikas et al. 2002), and rails from New Zealand (Trewick 1997). Both studies support the hypothesis that single-island endemic flightless rail s are derived from within paraphyletic

PAGE 104

90 widespread, volant, polytypic sp ecies. Slikas et al. (2002) also showed that the extinct ( † ), flightless species † Porzana palmeri (from Laysan) and the flightless P. atra (Henderson Island) occupy shor t branches on the phylogeny, i ndicating recent divergence from their volant ancestors P. pusilla and P. tabuensis , respectively. Trewick (1997) showed that flightless swamphens from New Zealand’s North Island ( Porphyrio mantelli ) and South Island ( † P. hochstetteri ) evolved the flightless co ndition independently from different colonizations of New Zeala nd by the widespread, volant species P. porphyrio . In this chapter I infer the relationships among flightless and volant species in one rail lineage that has radiated extensively in Oceania using phylogenetic analyses of modern and ancient DNA sequence data, and es timate divergence ti mes within the group using a locally calibrated mol ecular clock. The research focuses on the “typical longbilled rails” of the genus Gallirallus sensu lato . There has been little consensus regarding the composition of the genus Gallirallus (Ripley 1977, Olson 1973b, Livezey 1998, Taylor 1998), and as a starti ng point for phylogenetic anal yses I have regarded all “typical long-billed rails” from Oceania as species of Gallirallus sensu lato , distinguishing them from the similarly sized swamphens ( Porphyrio ), moorhens ( Gallinula ), and coots ( Fulica ), and the much smaller crakes ( Porzana, Amaurornis, and Poliolimnas ) that are also found in Oceania. This lineage provides an excellent opportunity for such a study because it include s seven living and ca. 17 extinct, putatively flightless species endemic to si ngle islands in Oceania, and four extant, volant species, any of which may have given rise to flightless species. This treatment of Gallirallus s.l. departs slightly from the classification of Olson (1973b) , who retained the genus Nesoclopeus for woodfordi of the Solomon Islands and † poecilopterus of Fiji, and

PAGE 105

91 Tricholimnas for † lefresnayanus from New Caledonia. Tayl or (1998) also retained Nesoclopeus , but departed from Olson (1973b) by not recognizing Tricholimnas as distinct from Gallirallus , and by placing pectoralis of New Guinea in Lewinia . Livezey (1998, 2003) placed the volant species pectoralis and striatus in Gallirallus , and advocated recognition of the genera Nesoclopeus, Tricholimnas (including sylvestris and † lafresnayanus ) , Habropteryx (for torquatus , insignis , and okinawae ), and the monotypic Cabalus for † modestus. The phylogenetic framework I reconstruct for the Gallirallus radiation will address the following questions : 1) How are living and extinct flightless species related to each other and to extant volant species? 2) Do populations of extant volant species evolve rapidly into flightless species? and 3) How should the species of “typical” rails from Oceania be classified into genera? Methods Taxon sampling – I examined mitochondrial DNA sequences of 35 individuals representing 16 species of Gallirallus s.l. including three historically extinct species and three recently named extinct fossil species (T able 3-1). I generated sequences of the cytochrome b, 12S, and control region genes from 29 rails, and included previously published (Trewick 1997) cytochrome b and 12S sequences from 6 rails. For 10 species, more than one individual was sampled to veri fy sequence identity and to examine species monophyly. I sampled three individuals from two subspecies of the volant G. torquatus , and nine individuals from four subsp ecies of the widespread, volant species G. philippensis . DNA was extracted from frozen tissue samp les of 12 individuals provided by the Florida Museum of Natural Hist ory, University of Florida (U F), Field Museum of Natural History (FMNH), the American Museum of Natural History, New York (AMNH), and

PAGE 106

92 the Thomas Burke Museum, University of Washington (UWBM). Museum skins of 14 individuals were sampled from the collec tions of AMNH, FMNH, National Museum of Natural History, Smithsonian Institution (USNM), and the Museum of Comparative Zoology, Harvard University (MCZ). At each museum I removed small (ca. 1 mm x 3 mm) slivers of epidermal and connective tissu e from the bottom of the feet (“toe pads”) to sterile plastic tubes using disposable sterile scalpel blades and forceps sanitized with bleach. Samples of prehistorically extinct sp ecies were taken from fossil bones recovered from archaeological sites and held in the collections of the Bernice P. Bishop Museum (BPBM) and UF (Table 3-2). Chips (ca. 3 x 6 mm) were removed from shafts of limb bone specimens using a disposable cutting wh eel attachment for a Dremel® grinder. DNA extraction and sequencing – DNA extractions from toe pads and fossil bones were carried out in the dedicated An cient DNA Laboratory in the Florida Museum of Natural History, where no previous work on birds has been done. Extractions from frozen tissues and all DNA amplification and sequencing were carried out in a separate DNA-sequencing laboratory in the Department of Zoology, University of Florida. Genomic DNA was extracted from frozen tissu es using the phenol-chloroform-ethanol method (Sambrook et al. 1989). Extractions fr om toe pads followed a modified version of the phenol-chloroform-dialysis method of Flei scher et al. (2000), as follows. Toe pads were incubated for 12-24 hr with agitation at 55 û C in 750µL of an extraction buffer consisting of 1% SDS, 7.5mg DTT, 0.1mg proteinase k, 0.02M EDTA, 0.01M Tris, and 0.01M NaCl. Toe pads were removed from tube s using forceps sanitized with bleach and minced using a disposable sterile scalpel blad e and returned to thei r tubes. Additional 0.1 mg of proteinase k was added and tubes were resealed and incuba ted for 24-48 hr with

PAGE 107

93 agitation at 55 û C. DNA was isolated by two rounds of mixing and centrifugation with equal volumes of phenol and one round with equal volume of chloroform. Final DNA recovery and concentration in a final volum e of 150µL of water was accomplished using MicroCon (Centricon®) spin columns ra ther than ethanol precipitation. Extraction of DNA from bones followed a m odified version of the silica-based method described by Boom et al. (1990) a nd Höss and Pääbo (1993). Bone chips were ground to powder under liquid nitrogen with mo rtar and pestle and were incubated for 24-48 hr with agitation at 55 û C in 750µL of an extracti on buffer consisting of 7.5M guanidinium thiocyanate (GuSCN), 0.1M Tr is-HCl (pH 6.4), 0.02M EDTA (pH 8.0), and 1.3% Triton X-100. Following centrifugation, 50 0µL of the supernatant was removed to a second tube containing an additional 500µL of the GuSCN extraction buffer and 40µL of silica suspension (SiO2 in water). DNA was bound to the silica during a 10-minute incubation at 27 û C. The silica was then pelleted by centrifugation and the pellet washed twice with a buffer containing 7.5M GuSC N, 0.1M Tris-HCl pH 6.4, and 0.02M EDTA pH 8.0, and twice with 70% EtOH supplimente d with 10mM NaCl. DNA was recovered from the washed and dried pellet by elution in two 75µL volumes of TE at 60 û C. Stringent protocols to avoid and detect contamination were followed in the Ancient DNA Lab, including negative extracti on controls (containing no to e pad or bone tissue), glove changes between handling each sample, ultravio let irradiation of all plastics, exclusive use of aerosol-barrier pipette tips, and daily sanitation of all surfaces with 10% bleach solution. I amplified portions of cytochrome b , 12S, and Domain I of the control region from the extant species Gallirallus owstoni , G. woodfordi , and G. philippensis in polymerase

PAGE 108

94 chain reactions (PCRs) using primers from Gl enn et al. (1999), So renson et al. (1999), and Slikas et al. (2002). From alignments of these sequences I designed internal primers for use in amplification and sequencing of small fragments (135 bp – 200 bp) from ancient DNA extracts (Table 3-3). Five µL of DNA extract were added to a 50µL PCR which contained 0.4mM dNTPs, 0.5 mg/ml BSA, 0.5mM of each Gallirallus -specific primer, 1 unit Eppendorf Hotmaster Taq® DNA polymerase, and 5µL Eppendorf PCR buffer containing MgCl2. Negative PCR controls were carried out to highlight contamination by exogenous DNAs. All PCR pr oducts were cut from 1% agarose gels following electrophoresis and purified using a Eppendorf Perfectprep® Gel Cleanup kit. Nucleotide sequences of both the light and h eavy strands were reso lved on an Applied Biosystems Inc. ® 3100 automated DNA sequencer. Data analyses Sequences (all >75% double stra nded) were aligned and edited in Sequencher version 3.1.1 (Gene Codes Corpora tion). I performed phylogenetic analyses on an alignment of concatenated cyt b , 12S, and CR sequences using the maximum parsimony (MP) and maximum likelihood (ML) cr iteria for tree selection carried out in PAUP* version 4.0b10 (Swofford 2003). Heuris tic MP and ML searches and parsimony bootstrap analyses of node s upport (200 MP replicates, 50 ML replicates) used the tree bisection and reconnection method of br anch swapping, with 10 random sequence additions per replicate. I conducted a Bayesi an analysis of the three-gene alignment using the program MrBayes version 3.1.1 (Hue lsenbeck and Ronquist 2001), in which I ran two parallel Markov chains for 1,000,000 generations initiated with a random starting tree, and sampled the chains every 500 gene rations yielding 2000 point estimates of model parameters and tree topology. Log-lik elihood values for all sampled trees were

PAGE 109

95 plotted against generation time to determine when the chains reached an equilibrium, and all trees sampled prior to stab ilization (n=100) were discarde d as the “burn-in.” These subsamples minus the burn-in generations were used to construct a majority-rule consensus tree with posterior probabilities on nodes. The m odel of DNA evolution used in ML and Bayesian analyses was select ed by hierarchical likelihood ratio tests implemented in MODELTEST 3.6.6 (Posada and Crandall 1998). Divergence times among species were estimated by first conductin g a likelihood ratio test of clock-like evolution, and then calibrati ng the rate of evolution for the concatenated three-gene alignment by fixing the maximum age of the node connecting † G. wakensis to its sister taxon at 124,000 years ago. This date correspon ds to the time when eustatic sea level was elevated +6 m, thereby inundating lowlying Wake Island (max. elev. 6 m), as inferred from a combined analysis of marine oxygen isotope data and the ages of terraces in New Guinea’s Huon peninsul a (Chappell and Shackleton 1986). Results Partial sequences of all th ree genes were obtained for all modern and ancient samples (Table 3-1) with three exceptions: 1) No cyt b amplifications succeeded from any of three fossil bone samples of † G. unnamed C , 2) Only CR was amplified from † G. unnamed D UF 62963, and 3) cyt b and 12S PCR products from the 120 year old toe pad sample of G. australis proved to be contaminants from bird carcasses that were being prepared as specimens at FMNH on the day th at toe pad was sampled. Control-region sequences from this sample are apparently ge nuine owing to the incr eased specificity of Gallirallus -specific CR primers. In all analyses, CR sequence from FMNH 67208 was combined with published cyt b and 12S sequences from NMNZ L30653. I had no access to specimens of G. sylvestris and † G. dieffenbachii and analyses of these taxa relied on

PAGE 110

96 published sequences of 12s and cyt b , and 12S only, respectively. The complete concatenated alignment consisted of 1160 nucleotides (CR = 325, cyt b = 307, 12S = 528), with individual sequences ranging fr om 266 bp to 1157 bp. Sequences of 12S varied in length among species owing to two, single-nucleotide insertions in pectoralis, two in striatus (one shared with pectoralis ), and two, two-to-eight -nucleotide stretches that could not be aligned unambiguously and were excluded from analyses. As determined by pairwise HKY-corrected genetic distances, 12S is the slowestevolving gene, cyt b is slightly faster, and Domain I of the CR is much faster. This pattern is consistent with relative rates obser ved in many other groups of birds (Kirchman et al. 2001, Baker and Marshall 1997), although Domain I appa rently evolves at roughly the same rate as cyt b in pheasants and partridges (Kimball et al. 2001). Genetic distances calculated for all genes combined averaged very slightly higher than those calculated from cyt b alone (Table 3-4). On the basis of these genetic distances, all phylogenetic trees we re rooted using striatus (designated as pa raphyletic outgroup in PAUP*). Hierarchical likelihood ratio tests determin ed that the best-fit model of nucleotide substitution was HKY + + I, which includes separate transition and transversion probabilities, a gamma-distribution shape parameter ( = 0.6793) for among-site rate variation, a parameter for the proportion of invariant sites (I = 0.5175), and parameters for Ti:Tv ratio (6.3304) and nucleot ide frequencies (A=.3152, C=.2910, G=.1603, T=.2335). Heuristic maximum likelihood tree searches using this model found the single most likely tree (-ln likelihood = 3522.1202) sh own in Figure 3-1. This same topology was also recovered in the majority-rule cons ensus tree from the Bayesian analysis, and

PAGE 111

97 posterior probabilities are shown in Figur e 3-1. Maximum like lihood bootstrapping and Bayesian posterior probabili ties indicate moderate to high support for the monophyly of all species except Gallirallus philippensis , which is paraphyletic with respect to three flightless congeners (node 3 Figure 3-1). A Kashino-Hasegawa test with two-tailed RELL distribution (1000 reps) implemented in PAUP* strongly rejected a constrained tree in which G. philippensis was fixed as monophyletic (-ln likelihood = 4763.7903, p=0.000). The majority-rule consensus of 10,000 most-parsimonious trees, each of length 399 steps, is congruent with the ML a nd Bayesian trees, but fails to recover two nodes that received low ML bootstrap and pos terior probability s upport: a node uniting G. owstoni and † G. dieffenbachii , and the node that places G. woodfordi basal to † G. wakensis plus its sister clade (Figure 3-2). All analyses support a monophyletic Gallirallus, excluding the volant species pectoralis and striatus , at node 1 in Figures 3-1 and 32. Within this clade, branch lengths are relatively short compared to the large genetic distance to pectoralis and especially striatus . Gallirallus australis, which is the type species for the genus Gallirallus , is basal within this clade, and I c onsider all species descended from the ancestor represented by node 1 to comprise the genus Gallirallus (summarized in Table 3-5). Basal nodes within this clade, descri bing the sequence of sp eciation events since G. australis diverged from its sister taxon, are not well resolved by bootstraps and posterior probabilities. However, both model-based techniques and maximum parsimony indicate that the volant species G. torquatus and two flightless species, † G. modestus from the Chatham Islands and G. insignis of New Britain, are genetically distinct and are basal to a more recently evolved clade that includes most flightless species and the volant G.

PAGE 112

98 philippensis . There is broad support for a clade (at node 2 in Figures 3-1, 3-2) of genetically very similar (Tab le 3-4) species uniting † G. wakensis and its sister clade, which includes † G. ripley, G. sylvestris, G. philippensis , and it’s derivatives † G. unnamed C , G. rovianae , and † G. unnamed D . Within the paraphyletic G. philippensis , individuals of that species sampled from islands in the Philippines form a clade that is sister to one including G. philippensis from Australia, New Zealand, and Vanuatu, two fossil flightless species from the Mariana Islands and G. rovianae , which is extant on New Georgia, Solomon Islands. A likelihood ratio test of cl ock-like evolution failed to re ject the null hypothesis of no difference between the most likely tree (Figur e 3-1) and a clock-enfo rced tree (ln L= 3540.1890, -2 L= 36.136, df=32, p>0.05) and so the ML tree can be used to estimate divergence times within Gallirallus and to estimate the timing of the origin of G . philippensis (possibly in the Philippines). At least 1.82% HKY-corrected pairwise sequence divergence separates † G. wakensis and members of its sist er clade (Table 3-4). If this mitochondrial divergence occurred since the two clades split, and if † G. wakensis evolved on Wake Island soon after it was inundated by elevated sea-levels 124 ka (thousand years ago), the rate of evolution of the combined three-gene data set (the combined clock) is 14.7% per million years. The rate for just the cytochrome b gene (cyt b clock) is calculated as 10.6% per milli on years. The combined clock and cyt b clock estimates for the origin of Gallirallus (node 1) are 381 ka and 399 ka, respectively. This implies a very rapid ra te of speciation in Gallirallus , especially considering that many extinct species of Gallirallus known only from fossils must also have diverged within this time frame. The combined clock and cyt b clock estimates for the origin of G.

PAGE 113

99 philippensis and its flightless deriva tives (node 3) are 22.4 ka and 30.8 ka, respectively. The very recent origin of this group is c onsistent with strati graphic archaeological evidence suggesting that G. philippensis colonized New Caledonia and Tonga only in the last few thousand years following the expans ion of humans into Oceania (Balouet and Olson 1989, Kirchman and Steadman 2005). It also means that populations of G. philippensis have diverged to become fully flight less species in a span of time that is essentially too short to be recorded by mitochondrial DNA evolution. An alternative explanation for the paraphyly of G. philippensis , namely that recent or ongoing hybridization with fli ghtless congeners has resulted in transgression of mitochondrial DNA into those species, can not be ruled out without further sequencing of additional specimens and of nuclear genes. Whereas th is may be impossible with respect to fossil species, efforts are underway to investigat e genetic signatures of hybridization with G. rovianae . Discussion The success rate of ancient DNA extractions from toe pad samples, the oldest of which was collected in 1886, was 100%. Extr actions from fossil bone specimens were less successful and depended upon the preservational environment of individual archaeological sites. DNA amplifications were successful from multiple bones of † G. ripleyi (3 of 3 bones sampled) from Tangatata u Rockshelter on Mangaia, Cook Islands, and † G. unnamed C (2 of 3 sampled) from Railhunt er Rockshelter on Tinian, Mariana Islands, and from a single bone (of 2 sampled) of † G. unnamed D from Mochong Site on Rota, Mariana Islands.

PAGE 114

100 Taxonomic implications The sequence of basal nodes within Gallirallus indicate that the volant species G. torquatus from Wallacea, Philippines, and New Guinea, although monophyletic today and not genetically similar to any species (Table 3-4) may have been the source of an early radiation that gave rise to the flightless species † G. modestus and G. insignis . This is consistent with Olson’s (1973b) hypothesis that G. insignis is “derived from G. torquatus stock.” Olson (1973b) was apparently partly correct in considering the flightless species G . australis, G. owstoni, † G. wakensis, † G. pacificus, and G. sylvestris to be “derivatives of G. philippensis stock.” Despite the close resemblance of G . australis and G. owstoni to G. philippensis , the former two species are not closely related to the latter. M itochondrial data suppor t his hypothesis that pectoralis is part of a more primitive “pro-R allus” group, but not his claim that striatus is an “advanced form of Gallirallus ” convergent with the more specialized Rallus rails. On the basis of the tree topologies a nd genetic distances presented in this chapter I tentatively recommend that pectoralis be placed in Lewinia following Taylor (1998), and that striatus be placed in Rallus , contingent upon further phylogenetic analyses that include at a minimum Rallus aquaticus (the type species for Rallus ) and the other two species assigned to Lewinia by Taylor (1998), L. mirificus and L. muelleri (the type species for Lewinia is pectoralis ). My results support Trewick’s (1997) hypothesis on the basis of partial cyt b and12S sequence data from a limited set of taxa that the Chatham Islands species † G. dieffenbachii may share a more recent common ancestor with owstoni than with a clade consisting of G . philippensis and G . sylvestris , and that G. australis is only distantly related to these other typical rails, but do not corroborate his finding that the Chatham rails are sister speci es. There is also no evid ence to support Ripley’s (1977)

PAGE 115

101 inclusion of dieffenbachii as a subspecies of G. philippensis , nor is there support for the phylogenetic distinctiveness of genera recognized by Livezey (1998) including “ Tricholimnas” (including sylvestris and the unsampled † lafresnayanus ), “ Habropteryx ” (including insignis, torquatus , and the unsampled okinawae ), “ Cabalus ” ( † modestus ) or “ Nesoclopeus ” ( woodfordi and the unsampled † peocilopterus ). I propose that the above four genera be abandoned, and that the speci es placed therein by other authorities be recognized as members of Gallirallus (Table 3-5). Rates of speciation Estimated divergence times within Gallirallus and within the G. philippensis species complex support the hypothesi s that flightless rails endemic to single islands evolve rapidly fr om populations of widespread, volant congeners. Rates of gene evolution calibrated with re ference to the maximum age of † G. wakensis are approximately five-fold higher than the commonly invoked cytochrome b rate of 2% per million years calibrated with reference to an Early Pliocene (4-5 million years ago) divergence within geese (Shields and Wilson 1987). Similar or slightly slower rates have been calibrated for cyt b for passerine divergences base d on the ages of the Hawaiian islands (1.6% per million years, Fleischer et al. 1998), and for divergences among crane species based on fossils up to 16 million year s old (0.7-1.7% per million years; Krajewski and King 1996). Slikas et al. ( 2002) calibrated a faster rate of evolution for the combined mitochondrial protein-coding genes cyt b , ATP8, and COII in Paci fic crakes using the same calibration point as Fleischer et al. ( 1998), i.e., the age of the island of Hawaii (430 ka) as the maximum age of the endemic flightless rail † Porzana sandwichensis. The rate they obtained (ca. 13% /my) was similar to the ones calibrated herein for cyt b (10.6% /my) and for cyt b plus CR and 12S (14.7% /my). A lthough different rates were obtained

PAGE 116

102 in different bird groups on the basis of ages of the Hawaiian Islands, there is an apparent correlation between the age of the calibration point and the es timated rate of evolution such that younger calibrations, su ch as that used here for Gallirallus , yield faster rates. Consistent with this trend, s till faster rates of mitochondrial evolut ion (2-7 times the 2%/my rate) have been estimated non-phylogenetically from observed divergence between modern and ancient (presumed ances tral) populations of penguins separated by only 6000 years (Lambert et al. 2002). Perhaps younger calib ration points more accurately reflect the true substitution rate of mitochondrial DNA, and much older divergences yield slow rates because millions of years of selec tive sweeps, population bottlenecks, and transgressive hybridizations periodically reduce genetic variation. Alternative explanations for the faster rates (relative to phylogenetically calibrated rates) obtained herein and by Slikas et al. (2002) for two groups of Pacific rails include 1) fast generation times in tropical rails, which can re produce in as little as ca. 80 days (Jeff Sailer, pers. comm.), rather than the year that is typical of most passerines, or 1-2 years for cranes, or 2) the possibility that mtDNA divergences vastly pr edate species splits estimated by island ages in rails but not in passerines. Despite this wide range in calibrated rates even within birds, divergences among Gallirallus species certainly occurred throughout the Pleistocen e, even if the slowest rates are applied. I speculate that fluctuations in sea-levels may have facilitated sequentia l waves of colonization in Oceania by volant rails from Wa llacea and Southeast Asia. Geography of speciation Three sets of flightless Gallirallus species endemic to the same archipelago were included in phyl ogenetic analyses. None of the sets (†G. modestus and †G. dieffenbachii of the Chatham Islands, G. rovianae and G. woodfordi of

PAGE 117

103 the Solomon Islands, and G. owstoni and †G. unnamed C or †G. unnamed D of the Mariana Islands) are monophyletic, indicating that there has been little speciation within archipelagoes from single colonizations. Rather , it appears that there have been a series of multiple invasions of archipelagoes in Oceania. I speculate that in all three archipelagoes an early wave of co lonization, perhaps by the ancestor of G. torquatus , gave rise to †G. modestus , G. woodfordi , and G. owstoni , and that more recent colonizations gave rise to †G. dieffenbachii , G. rovianae , and †G. unnamed C and †G. unnamed D , the latter three being the re sult of recent colonization by G. philippensis . The temporal and geographic context of speciation in Gallirallus inferred from phylogenetic analyses of mitochondrial DNA suppor ts a cyclical view of speciation first articulated by Mayr (1942) and later by ma ny others (summarized in Grant and Grant 1998) in which double (or rarely triple) invasions of islands by the sa me ancestral species provides the basis for within-archipelago divers ification. To the extent that colonists ( G. torquatus, G. philippensis ) are broadly adapted generalist s that give rise to locally adapted specialists (flightless species), the pattern in Gallirallus is consistent with the socalled taxon cycle model of evolu tion (Ricklefs and Cox 1972). The Gallirallus radiation departs from the taxon cycle model in two im portant ways, however. First, flightless Gallirallus species from the same archipela go are not found in sympatry on single islands, precluding a direct role for ecological (competitive) reinforcement of isolating mechanisms. Second, the taxon cycle is usua lly formulated in terms of continental species colonizing satellite islands, whereas volant Gallirallus species appear to have colonized very distant archipelagos, and to have done so from the Philippines and Wallacea rather than from Australia or Asia. The very small genetic distances between

PAGE 118

104 G. philippensis and its flightless derivatives suggest a possible role of hybridization in Gallirallus . Mayr (1942) considered hybridization between p opulations in secondary contact to be a homogenizing force with li ttle creative potential for species-level divergence. Lack (1947) hypothesized the hybr idizations between species of Galapagos finches may have resulted in intermediately -sized species on nearby islands, and more recent genetic studies confirm that hybridi zation does indeed occur among species of Geospiza, and that under certain ci rcumstances hybrids may be favored by selection (Grant and Grant 1994). It may be that Gallirallus rovianae of New Georgia is a result of recent hybridization between populations des cended from multiple colonizations of G. philippensis , or that G. philippensis mitochondrial DNA was recently transgressed into G. rovianae that was descended from much earlier co lonization by some other species. Very little is known about the natu ral history of or phenotypi c and genetic variation in G. rovianae , as only two specimens exist in the wo rldÂ’s museum collections. Elucidating the role of hybridization in the formation of flightless Gallirallus species must await further study of G. rovianae genetics and ecology.

PAGE 119

105Table 3-1. Specimen information and DNA sequence lengths for all control region (CR), cytochrome b (Cyt b), and 12S rDNA (12S) sequences used to reconstruct the ph ylogeny of Gallirallus. AMNH = American Museum of Natural History, FMNH = Field Museum of Natural History, MC Z = Museum of Comparative Zoology, NMNH = National Museum of Natural History, Smithsonian Institution, NMNZ = National Museum of New Zealand, UF = Florida Museum of Natural History, University of Florida, UWBM = Thomas Burke Museum, Univer sity of Washington. Asterisk s (*) identify individuals for which sequences were obtained from Trewick 1997 (GenBank accession numbers for 12S and Cyt b, respectively, listed under “Source”). Species Specimen number Locality Source CR 12S Cyt b G. philippensis NMNH 582808 Batan, Philippines Toe pad 325 492 163 G. philippensis MCZ 194032 Lagengilang, Philippines Toe pad 325 483 307 G. philippensis FMNH 425075 Mindanao, Philippines Toe pad 325 484 307 G. philippensis UWBM 62806 NSW, Australia Frozen tissue 325 492 307 G. philippensis UWBM 62974 Australia Frozen tissue 325 523 307 G. philippensis UF 43154 Santo, Vanuatu Frozen tissue 325 475 307 G. philippensis UF 43180 Uripiv, Vanuatu Frozen tissue 325 525 307 G. philippensis UF 43223 Efate, Va nuatu Frozen tissue 325 525 307 G. philippensis* NMNZ NM23996 New Zealand U77150, U77174 --379 246 G. torquatus MCZ 269535 Malukken Greater Sunda Is Toe pad 325 482 307 G. torquatus FMNH 358233 Sibuyan, Philippines. Frozen tissue 325 525 305 G. torquatus FMNH 358235 Sibuyan, Philippines. Frozen tissue 325 522 303 G. australis FMNH 67208 North Is., New Zealand Toe pad 123 ----G. australis* NMNZ L30653 Kapiti Is., New Zealand U77148, U77177 --382 246 † G. unnamed C UF 60153 Tinian, Mariana Islands Fossil bone 266 143 --† G. unnamed D UF 62963 Rota, Mariana Islands Fossil bone 266 ----† G. wakensis MCZ 157073 Wake Island Toe pad 325 482 307 † G. wakensis AMNH 265485 Wake Island Toe pad 325 487 307 G. sylvestris* none Lord Howe Island U77152, U77176 --382 246 † G. dieffenbachii* NMNZ S-30120 Lord Howe Island U77159 --382 --G. insignis AMNH 777724 New Britain, Bismark Is. Toe pad 325 342 307 G. insignis AMNH 333800 New Britain, Bismark Is. Toe pad 325 480 307 † G. modestus MCZ 39926 Chatham Is. Toe pad 325 487 307 † G. modestus* NMNZ S-27480 Chatham Is. U77158 --379 --G. rovianae AMNH CEF 878 New Georgia, Solomon Is. Frozen tissue 325 525 307

PAGE 120

106Table 3-1 Continued Species Specimen number Locality Source CR 12S Cyt b G. owstoni UF 42967 Guam, Mariana Is . (captive) Frozen tissue 325 525 307 G. owstoni UF 42969 Guam, Mariana Is . (captive) Frozen tissue 325 506 307 G. owstoni* none Guam, Mariana Is. (captive) U77151, U77175 --379 246 G. woodfordi UWBM 69782 Isabel, Solomon, Is. Frozen tissue 325 505 307 G. woodfordi UWBM 69783 Isabel, Solomon Is. Frozen tissue 325 496 307 † G. ripleyi UF 54277 Mangaia, Southern Cook Is. Fossil bone 228 101 124 Lewinia pectoralis AMNH 802028 New Guinea Toe pad 123 505 307 Lewinia pectoralis AMNH 765544 New Guinea Toe pad 123 485 307 Rallus striatus AMNH 545058 Kuala Lumpur Toe pad 143 504 307 Rallus striatus AMNH 648249 Sarawak Toe pad 143 487 307

PAGE 121

107 Table 3-2. Fossil bone specimens of prehistorically extinct Gallirallus species sampled for ancient DNA. * = Approximate ag es of bones expressed in radiocarbon years before present Species, specimens DNA amplification ? Site Island, archipelago Age of bones* † G. unnamed C UF 60088 No Railhunter Rockshelter Tinian, Marianas 2460-1880 UF 60126 Yes Railhunter Rockshelter Tinian, Marianas 2460-1880 UF 60153 Yes Railhunter Rockshelter Tinian, Marianas 2460-1880 G. cf. philippensis UF 62972 No Chalan Piao Saipan, Marianas 3500-3000 † G. unnamed B UF 62937 No Pisonia RockshelterAguiguan, Marianas 1780-540 UF 62950 No Pisonia RockshelterAguiguan, Marianas 1780-540 † G. unnamed D UF 62963 Yes Mochong SiteRota, Marianas <2500 UF 63319 No Mochong SiteRota, Marianas <2500 † G. vekamatolu UF 52486 No ‘Anatu Site ‘Eua, Tonga 78,000-60,000 † G. ripleyi UF 51413 Yes Tangatatau Rockshelter Mangaia, Cook Islands 1000-700 UF 54277 Yes Tangatatau Rockshelter Mangaia, Cook Islands 1000-700 UF 54287 Yes Tangatatau Rockshelter Mangaia, Cook Islands 1000-700 † G. storrsolsoni DAPT 47 No Fa’ahia Site Huahine, Society Islands 1250-750 DAPT 07 No Fa’ahia Site Huahine, Society Islands 1250-750 † G. roletti BPBM 166440 No Hanamiai SiteTahuata, Marquesas 1000 BPBM 166441 No Hanamiai SiteTahuata, Marquesas 1000 † G.gracilitibia BPBM 176746 No Hane Dune SiteUa Huka, Marquesas 1350 BPBM 176971 No Hane Dune SiteUa Huka, Marquesas 1350

PAGE 122

108 Table 3-3. Primers used in PCR and sequencing of samples in this study. The letters “L” and “H” refer to light and heavy st rands of the mitochondrial genome, respectively. . Gene Primer Sequence Reference Control Region L CR 1 5’-AATCACCGCGGCATGTAATC-3’ This study L CR 2 5’-GTAATCGCTCCACCTACTG-3’ This study L CR 3 5’-GACTAAGGGTACAAATCCC-3’ This study H CR 1 5’-CCTGTACCATGTATGTTTTG-3’ This study H CR 2 5’-GATCTCTCGTGAGGTGAACG-3’ This study H CR 3 5’-GTGGTTAGTTCCA ATAACCGAG-3’ This study Cytochrome b L 14957 5’-CCATCCAACATCTCAGCATGATGAAA-3’ Slikas et al. (2002) L 15134 5’-CAATACGGCTGACTACTCCG-3’ This study H 15115 5’-CGGAGTAGTCAGCCGTATTG-3’ This study H 15295 5’-TCAGAATGATATTTGTCCTCA-3’ Sorenson et al. (1999) 12S rDNA L 1753 5’-AAACTGGGATTAGATACCCCA CTA-3’ Sorenson et al. (1999) L 1937 5’-CAGCCTACATACCGCCGTCC-3’ This study L 2116 5’-GGATTTAGCAGTAAAGGGGGG-3’ This study H 1917 5’-GGACGGCGGTATGTAGGCTG-3’ This study H 2096 5’-CCCCCCTTTACTGC TAAATCC-3’ This study H 2294 5’-CTTTCAGGTGTAAGCTGARTGCT T-3’ Sorenson et al. (1999)

PAGE 123

109Table 3-4. Mean percent sequence divergence (HKY-corrected) between Gallirallus species for the cytochrome b gene (above diagonal) and for the combined cytochrome b , control region, and 12S genes (below diagonal). phil. rovi. sylv. wake. ripl. owst . wood. torq. insi. mode. aust. G. philippensis 0.43 1.34 1.35 1.15 2.67 2.39 5.78 4.63 6.29 6.70 G. rovianae 0.64 1.34 1.32 0.79 2.67 2.34 5.70 4.78 6.20 6.78 G. sylvestris ----1.78 0.00 2.91 2.90 4.64 3.93 6.23 6.48 † G. wakensis 2.05 1.82 --1.64 2.53 2.34 5.68 4.78 6.20 7.24 † G. ripleyi -------2.07 3.40 6.84 6.03 6.05 8.68 G. owstoni 2.62 2.50 --1.92 --2.14 5.42 4.37 6.81 7.63 G. woodfordi 2.90 2.72 --2.64 --2.17 5.03 4.44 5.85 5.37 G. torquatus 5.80 5.57 --5.45 --4.72 4.99 4.99 5.30 4.23 G. insignis 4.29 4.28 --3.86 --2.97 2.17 5.34 4.77 5.93 † G. modestus 5.83 5.74 --5.35 --4.84 5.23 5.47 5.31 6.89 G. australis 6.33 6.12 --6.57 --6.04 6.09 5.60 5.49 5.92

PAGE 124

110Table 3-5. Alphabetical summary of all known species considered members of the genus Gallirallus, including status (E= extant, C= extinct in wild but breeds in captivity, †H = extinct, recorded historically, †P= ex tinct, recorded prehistorically as fossil bones), flight ability (V= volant, F= flightless), and geographic range. Species Flight Status Geographic Range G. australis F E New Zealand G. dieffenbachii F †H,P Chatham Is., Chatham Islands G. epulare F †P Nuka Hiva, Marquesas G. gracilitibia F †P Ua Huka, Marquesas G. huiatua F †P Niue G. insignis F E New Britain, Bismarck Islands G. lafresnayanus F †H New Caledonia G. modestus F †H,P Mangare Is., possibly Pitt Is., Chatham Islands G. okinawae F E Okinawa, Ryukyu Islands G. owstoni F C Guam, Mariana Islands G. pacificus F? †H Tahiti, Society Islands G. philippensis V E Cocos Is., Philippines, Sundas E. to Samoa, Australia, New Zealand G. poecilopterus F †H Taveuni, Viti Levu, and Ovalau, Fiji G. ripleyi F †P Mangaia, Cook Islands G. roletti F †P Tahuata, Marquesas G. rovianae F E New Georgia Group, Solomon Islands G. storrsolsoni F †P Huahini, Society Islands G. sylvestris F E Lord Howe Island G. torquatus V E Philippines, Sulawesi, NW New Guinea G. vekamatolu F †P ‘Eua, Tonga G. wakensis F †H Wake Island G. woodfordi F E Bougainville, Santa Isabel, and Guadalcanal, Solomon Islands G. unnamed A F †P New Ireland, Bismarcks

PAGE 125

111Table 3.5 continued. G. unnamed B F †P Aguiguan, Mariana Islands G. unnamed C F †P Tinian, Mariana Islands G. unnamed D F †P Rota, Mariana Islands G. undescribed sp. F †P Ha’afeva, Tonga G. undescribed sp . F †P Hiva Oa, Marquesas

PAGE 126

112 Figure 3-1. Maximum likelihood phylogeny of co mbined 12S, cytochrome b, and control region data. Numbers at nodes indica te ML bootstrap support/Bayesian posterior probability. * = support for monophyly of G. owstoni , not for G. owstoni plus G. dieffenbachii .

PAGE 127

113 Figure 3-2. Majority-rule consensus tree of 10,000 maximum pars imony trees (length 399 steps). Numbers at nodes indica te ML bootstrap support/Bayesian posterior probability.

PAGE 128

114 CHAPTER 4 PHYLOGENETIC AND DEMOGRAPHI C TESTS OF RAPID PARALLEL SPECIATION OF FLIGHTLESS RAILS ( AVES: RALLIDAE) FROM A VOLANT ANCESTOR, Gallirallus philippensis Introduction Species complexes consisting of geographi cally variable popul ations are often studied by biologists interested in speciation because subspecific lineages may represent incipient species. Examining the geographic context of evolutionary divergence within species is an important means of elucidati ng the mechanisms as well as the temporal context of speciation (Avise 2000). Severa l recent phylogeographic studies have shown that widespread polytypic species are ofte n paraphyletic, with peripheral populations having given rise to extant species (e .g., Omland et al. 2000, Paxinos et al. 2002, Salzburger et al. 2002). This mode of speci ation may be common in groups that include continental and island populations (Coyne and Orr 2004). In cases where island colonization is followed by strong selection for decreased dispersal ability in the population, gene flow may become severely limited and new biological species ( sensu Mayr 1963) may form, a process limited by the extent to which new colonists continue to arrive. If many such species are spawned from time to time from a single “ancestral” species, an adaptive radiation based on parall el speciation may result. This form of radiation is fundamentally different from the familiar adaptive radiations on remote archipelagos in which a single colonizing species undergoes morphological divergence, filling many different niches (Schluter 2000). Nevertheless, this mode of evolution may predominate in certain groups that are para doxically both good coloni sts and subject to

PAGE 129

115 strong selection for decreased dispersal, gr oups referred to by Mayr and Diamond (2001) as the “great speciators.” A case in point is the evolution of insular flightless species of rails (Aves, Family Rallidae) from volant colonists. Thirty-one of the approximately 150 historically known species of rails are or were flightless species endemic to single islands, or to islands connected in the Pleistocene during periods of lowered sea-levels (Taylor 1998). In addition, scores of prehistorically extinct flightless species are known from fossil bones from archaeological and paleont ological sites on Ea rth’s oceanic islands (Steadman 1995, 2006, Olson and James 1991, Olson 1973a, Worthy and Holdaway 2002). This record reveals extensive Pleistocene and Holocene extinction of flightless rails following colonization by humans and their commensals (p rimarily dogs, pigs, and rats). Losses were greatest in Oceania (Figure 1-1), where flightless rail species may have numbered in the “hundreds” (Livezey 2003) or perhaps as many as 1600 (Steadman 2006). As most Pacific islands were never connected to c ontinental land, each flightless species may be an independent transition to flightlessness. Thus, the flightless species of Rallidae in Oceania are the largest and yet least studied example of adaptiv e radiation in island birds, and may also be the most extensive exampl e of convergent evolu tion among vertebrates (McNab and Ellis 2006). Flight ability in volant island colonists may be reduced by selection for decreased dispersal as well as for energy conservation (Olson 1973a), the latter hypothesis being strongl y supported by studies showin g decreased basal metabolic rates and decreased pectoral muscle mass in flightless versus volant birds (McNab 1994a, 2002, McNab and Ellis 2006). Olson (1973a) pr oposed a model for the evolution of Oceanic rail diversity in which multiple closel y related species may descend from one or

PAGE 130

116 a few, perhaps still extant ancestral species as natural selection fo r energy conservation rapidly pushes one island-colonizing populati on after another down parallel paths to flightlessness. Support for Olson’s rapid parall el speciation model of rail evolution has come from phylogenetic studies of mitochondrial DNA in the genus Porzana (Slikas et al. 2002), and in rails from New Zealand and neighbor ing islands (Trewick 1997). Both studies found that single-island endemic flightless species were derive d from within paraphyletic widespread, volant, polytypic sp ecies. Slikas et al. (2002) also showed that flightless species occupy short branches on the phyl ogeny, indicating a recent divergence from volant ancestors. No study has sampled exhaus tively within putative ancestral species to examine their genetic architect ure or the temporal and geogr aphic context of speciation events. Surveys of intra-specific mitoc hondrial DNA variation in birds from island archipelagos have clarified historical relationships among populations, highlighted patterns of gene flow and colonization, a nd showed clearly that morphological and molecular divergence are decoupled in many sp ecies (e.g., Sato et al. 1999, Kirchman et al. 2000, Pearce et al. 2002, Warren et al. 2003, Kvist et al. 2005). In this chapter I analyze mitoc hondrial control regi on variation in Gallirallus philippensis , a complex of 22 subspecies (Taylor 1998) of volant rails distributed widely in Australasia and Oceania that is thought to be the ancestor of mu ltiple recently evolved flightless species endemic to single islands in Oceania (Olson 1973b, Ripley 1977, Taylor 1998, Chapters 2 and 3 herein). I employ a combination of phylogenetic and historical demographic analyses that have proven effective in reconstructing patterns of diversification in continen tal species (Merilä et al., 1997, Cheviron et al. 2005).

PAGE 131

117 The Buff-banded Rail, Gallirallus philippensis (Order Gruiformes, Family Rallidae), is a large (180-280 g) ground-dwel ling omnivore with weak powers of flight that is locally common in disturbed habitats (rice fields, coconut plantations, along roads) below 500 m elevation throughout its extensive range (Figure 4-1). Geographic variation in size and in plumage has been well studi ed but is highly comp lex, and subspecific taxonomy has not been stable. Peters ( 1934) recognized only 14 subspecies, lumping many named populations from Australia and New Guinea. In a series of papers analyzing the extensive material from th e Whitney South Seas Expeditions and the Rothschild Collection, both housed at the Am erican Museum of Natural History, Mayr (1933, 1938, 1949, Mayr and Guilliard 1951) descri bed eight new subspecies. Ripley (1977) recognized 26 subspecies, including a ll of Mayr’s subspecies and previous subdivisions of some of Peters’ taxa, but i ndicated that racial affinities were poorly understood, especially among western populatio ns. Discussing the subspecies from Wallacea, White and Bruce (1986) pointed ou t that rail species are characterized by marked individual and local variation in size, and that size differences in G. philippensis may be of little significance. Schodde & de Naurois (1982) called phenotypic variation in G. philippensis “highly confusing” and questioned the validity of many of the 20+ named subspecies, and yet described a new subspecies, G. p. tounelieri, which Marchant and Higgens (1993) consider ed “doubtful.” Taylor (1998), who recognized 22 subspecies, acknowledged that plumage variat ion previously thought to be geographical may in fact be due to age, specime n wear, and individual variation. Under the view that phenotypic variat ion is geographical and indicative of reproductive isolation, Mayr and Diamond (2001) included G. philippensis among their

PAGE 132

118 list of “great speciators,” t hose highly polymorphic specie s characterized by high local abundance and intermediate vagility that give rise to subspecies and eventually full biological species following colonization fr om mainland sources. They hypothesized a series of dispersal waves to islands in Nort herm Melanesia to account for evolution of G. philippensis subspecies and flightless Gallirallus species occurring there. This interpretation is consistent with Olson’s mode l for the evolution of flightlessness in rails, but is at odds with zooarchaeological evid ence from stratified prehistoric deposits in Tonga (Steadman 1993, Kirchman and Steadma n 2005) and New Caledonia (Balouet and Olson 1989), which indicate that G. philippensis colonized islands in Oceania only after the arrival of humans 2500-3000 years ago. My analyses of molecu lar genetic data will address the following questions: 1) Is genetic variation in G. philippensis geographically structured such that island or archipelago populations are independe nt lineages? 2) Did G. philippensis colonize Oceania from Australasia following the expansion of humans into Oceania? 3) How are the extant populations of G. philippensis and flightless Gallirallus species related historically? Methods Taxon sampling: I sought tissue samples of G. philippensis from as many island populations as possible, a nd from one or two individuals of each flightless Gallirallus species, including all possible extant and extinct species. The sampling strategy for G. philippensis was to obtain samples from frozen tissue collections and traditional museum specimens (skins) collected from all available archipelagos within its range (Figure 4-1), and to focus new specimen collection efforts on multiple islands in a single archipelago. This combination of broad geographic c overage and dense sampling within one archipelago was intended to elucidate the entire range of genetic diversity and the

PAGE 133

119 geographic scale of genetic st ructure. Sampling of new G. philippensis specimens was carried out on three islands in the Republic of Vanuatu during two expeditions to that archipelago in 2002 and 2003 by th e Florida Museum of Natura l History, University of Florida (UF). A total of 21 birds were collect ed with the help of local hunters on Efate (5), Espiritu Santo (11) and Uripiv (5) unde r permits issued by the Republic of Vanuatu Environment Unit. Tissue samples (muscle, heart, liver) were stored in DMSO-NaCl buffer at ambient temperature in th e field, and were stored at -80 û C upon return to UF. Additional frozen tissue samples of some congeneric flightless species and of G. philippensis from Australia were provided by UF, the American Museum of Natural History, New York (AMNH), the Field Muse um of Natural History (FMNH), and the Thomas Burke Museum, University of Wa shington (UWBM). Museum skins were sampled from the collections of FMNH, AM NH, National Museum of Natural History, Smithsonian Institution (USNM), and the Museum of Comparative Zoology, Harvard University (MCZ). At each museum I removed small (ca. 0.05g) slivers of epidermal and connective tissue from the bottom of the feet (“toe pads”) to sterile plastic tubes using disposable sterile scalpel blades and forcep s sanitized with bleach. Samples of some extinct flightless species we re taken from fossil bones recovered from archaeological sites and held in the collections of UF and the Bernice P. Bishop Museum (BPBM). Chips (ca. 3 x 6 mm) were removed from limb bone specimens using a disposable cutting wheel attachment for a Dremel® grinder. Table 4-1 summarizes the locality and voucher specimen information of all samples used in this chapter. DNA extraction and sequencing: DNA extractions from toe pads and fossil bones were carried out in the dedicated ancien t DNA laboratory at the Florida Museum of

PAGE 134

120 Natural History, where no previous work on birds has been done. Extractions from frozen tissues and all DNA amplification (P CR) and sequencing were carried out in a separate laboratory in the Department of Zoology. Genomic DNA was extracted from frozen tissues using the phenol-chlorof orm-ethanol method (Sambrook et al. 1989). Extractions from toe pads followed a modi fied phenol-chloroform protocol: Toe pads were incubated for 12-24 hr with agitation at 55 û C in 750µL of an extraction buffer consisting of 1% SDS, 7.5mg DTT, 0.1mg proteinase k, 0.02M EDTA, 0.01M Tris, and 0.01M NaCl. Toe pads were removed from tube s using forceps sanitized with bleach and minced using a disposable sterile scalpel blad e and returned to thei r tubes. Additional 0.1 mg of proteinase k was added and tubes were resealed and incuba ted for 24-48 hr with agitation at 55 û C. DNA was isolated by two rounds of mixing and centrifugation with equal volumes of phenol and one round with equal volume of chloroform. Final DNA recovery and concentration in a final volume of 150µL of ultrapure water was accomplished using MicroCon (Centricon®) spin columns rather than ethanol precipitation. Extraction from bones followed a modified version of the silica-based method described by Boom et al. (1990) and Höss a nd Pääbo (1993). Bone chips were ground to powder under liquid nitrogen with mortar and pestle and were incubated for 24-48 hr with agitation at 55 û C in 750µL of an extraction buffe r consisting of 7.5M guanidinium thiocyanate (GuSCN), 0.1M Tris-HCl (pH 6.4), 0.02M EDTA (pH 8.0), and 1.3% Triton X-100. Following centrifugation, 500µL of th e supernatant was removed to a second tube containing an additional 500µL of the GuSCN extraction buffer and 40µL of silica suspension (SiO2 in water). DNA was bound to the silica during a 10 minute incubation

PAGE 135

121 at 27 û C. The silica was then pelleted by centrif ugation and the pellet washed twice with a buffer containing 7.5M GuSCN, 0.1M Tris -HCl pH 6.4, and 0.02M EDTA pH 8.0, and twice with 70% EtOH supplemented with 10m M NaCl. DNA was recovered from the washed and dried pellet by elution in two 75µL volumes of TE at 60 û C. Contamination controls carried out in th e ancient DNA lab include nega tive extraction controls (containing no toe pad or bone tissue), glove changes between handling each sample, ultraviolet irradiation of all plastics, exclusiv e use of aerisol-barrier pipette tips, and daily sanitation of all surfaces and equipm ent with 10% bleach solution. Control region amplification: No previously published work has characterized the mitochondrial control region of any species of rail (Rallidae) and so initial polymerase chain reaction (PCR) amplifications targeted the entire region and adjacent genes using methods and primers of Glenn et al. (1999) and Sorenson et al. ( 1999). I obtained long fragments (ca. 3000 bp) that stretched from th e 3’ end of ND6 to the middle of 12S from the species Gallirallus owstoni , G. woodfordi , and G. philippensis. From an alignment of these sequences I designed three primer pair s for use in amplifica tion and sequencing of small fragments (135 bp – 200 bp) at the 3’ end of Domain I ( sensu Baker and Marshal 1997) downstream of a region of tandemly rep eated nucleotide motifs (Figure 4-2). Five µL of DNA extract were added to a 50µL PCR which contained 0.4mM dNTPs, 0.5mg/ml BSA, 0.5mM of each Gallirallus -specific primer, 1 unit Eppendorf Hotmaster Taq® DNA polymerase, and 5µL E ppendorf PCR buffer containing MgCl2. Negative PCR controls were carried out to highlight contamination by exogenous DNAs. Portions of Domain I of the mitochondrial control regi on were amplified in 50ul polymerase chain reactions (PCRs) using various combinations of the following primers: CR L1 (5’ AAT

PAGE 136

122 CAC CGC GGC ATG TAA TC 3’), CR L2 (5’ GTA ATC GCT CCA CCT ACT G 3’), CR L3 (5’ GAC TAA GGG TAC AAA TCC C 3’), CR H1(5’ CCT GTA CCA TGT ATG TTT TG 3’), CR H2 (5’ GAT CTC TCG TGA GGT GAA CG 3’), CR H3 (5’ GTG GTT AGT TCC AAT AAC CGA G 3’). All PC R products were cut from 1% agarose gels following electrophoresis and purified using a Eppendorf Perfectprep® Gel Cleanup kit. Nucleotide sequences of both the lig ht and heavy strands were resolved on an Applied Biosystems Inc. ® 3100 automated DNA sequencer. Data analyses: Sequences (all >90% double strande d) were aligned in Sequencher version 3.1.1 (Gene Codes Corporation). The program DNAsp version 4.00 (Rozas et al. 2003) was used to calculat e nucleotide diversity ( ) and haplotypic diversity (Hd) from equations in Nei (1987), and to calculate pair wise (by subspecies, by archipelago, and by island within Vanuatu) and total Fst from equations in Hudson et al. (1992). To further investigate geographic genetic structure, an unrooted parsimony network of haplotypes was constructed using TCS version 1.18 (Cleme nt et al. 2000). To test the hypothesis that G. philippensis has expanded into Oceania following the spread of humans, histograms of pairwise nucleot ide differences (mismatch dist ributions) were constructed and tested for fit to Rogers and Harpending’ s (1992) infinite alle les model of population growth, which predicts a smooth, unimodal Poisson curve when populations have undergone a recent, rapid demographic expansio n as one would expect of a species that had recently colonized many archipelagoes from a single source area. Using DNAsp, Harpending’s (1994) raggedness index, r, wh ich measures the smoothness of observed mismatch curves, was calculated and tested for significant departure from unimodality using 1000 coalescent simulations given em pirical sample sizes and estimates of , a

PAGE 137

123 measure of effective population size equal to Neµ for mitochondrial DNA (Nei 1987). Demographic expansion was also inferred by cal culating Fu and Li’s (1993) F* and D* and Fu’s (1997) Fs, which are negative when there is an excess of rare (=new) alleles (an indication of population expansion under the assumption of selective neutrality), and tested their significance by 1000 coal escent simulations as above. Phylogenetic analyses of control region alignments that included all sampled haplotypes of G. philippensis and one or more individuals of other Gallirallus species used the maximum likelihood (ML) criterion for tree selection carried out in PAUP* version 4.0b10 (Swofford 2003), and the method of Bayesian inference carried out in the program Mr. Bayes version 3.1.1 (Huelsenbe ck and Ronquist 2001). Models of DNA evolution used in ML and Bayesian analyses were selected by hi erarchical likelihood ratio tests implemented in MODELTEST 3.6.6 (Posada and Crandall 1998). Heuristic ML searches and parsimony bootstrap analyses of node support (100 re plicates) used the tree bisection and reconnection method of branch swapping, with 1 random sequence addition per replicate. In the Bayesian an alysis, I ran two para llel Markov chains for 1,000,000 generations initiated with a random star ting tree, and sampled the chains every 500 generations. Log-likelihood values for all sampled trees were plotted against generation time to determine when the chains reached an equilibrium, and discarded all trees sampled prior to stabili zation as the “burn-in.” Results Control region variation: I obtained double-stranded sequences of 325 bp of Domain I of the mitochondrial cont rol region from 71 individuals of Gallirallus philippensis and from one to four specimens each from seven other Gallirallus species. Shorter sequences (123-266 bp) were obtained from fossil bones of G. ripleyi and G.

PAGE 138

124 unnamed C , and from toe pads of G. australis and the outgroup species Lewinia pectoralis (Table 4-1). Variation among individuals of G. philippensis was high with 40 of 325 sites polymorphic. These vari able sites defined 38 haplotypes in G. philippensis and resulted in an average pairwise numb er of nucleotide differences among all haplotypes (k) of 5.419, nucleotide diversity ( ) of 0.01667, and haplotype diversity (Hd) of 0.951. The maximum parsimony network of hapl otypes (Figure 4-3) reveals little geographic structure to genetic variation, with birds from different archipelagoes sharing the two most frequent haplotypes and w ith no clustering of haplotypes from any archipelago, with the excep tion of the eight haplotypes found in 10 birds from the Philippines. Of the 21 birds sampled from three islands in Vanuatu, 11 carried the same widespread haplotype (No. 2), six birds carried singleton ha plotypes that differed from that most common one by a single nucleotide substitution, and three birds shared a more divergent haplotype. The only large genetic break in the network is a string of six unsampled haplotypes between a single bird from Halmahera and all other birds. Otherwise, few extinct or unsampled ha plotypes are represen ted in the network, indicating that all G. philippensis sampled are genetically similar. I calculated pairwise and total values of Fst grouping haplotypes by island (for Vanuatu birds only), by subspecies (followi ng Taylor 1998 as in Table 4-1, including only those with at least three individuals), and by archipelago (including Philippines, Palau, Wallacea [Sulawesi, Halmahera, Flores], Bismarck Islands [broadly defined to include Pityilu, Ninigo, Skoki, New Britain, Wanton, Boang], Australia, New Zealand, Vanuatu, and Samoa). Values were near zero between islands in Vanuatu (pairwise = 0.087 – 0.046, total = 0.026) but were higher at larger geographic scales. When

PAGE 139

125 sequences were grouped by subspecies, pairwise Fst values ranged from 0.104 – 0.634, and total Fst = 0.329. Values were higher when se quences were grouped by archipelago (pairwise = 0.014 – 0.795, total = 0.4 24) due largely to the sepa ration of birds from the Philippines (which are the most genetically distinct) from those in the Greater Sundas, considered to be the single subspecies G. p. philippensis . Average pairwise Fst between the Philippines and othe r archipelagoes was 0.612. Historical Demography: Values of Fs, D*, and F* were estimated and mismatch distributions analyzed both for the complete sample of 71 birds and for a subset that excluded the 10 individuals from the Philippines. Estimates were significantly negative for both the complete sample (Fs = -24.582, P[Fs < -24.582]=0.000; D* = -2.694 , p<0.05; F* = -2.574, p<0.05) and the non-Philippine sample (Fs = -21.324, P[Fs < 21.324]=0.000; D* = -4.054 , p<0.02; F* = -3.87 1, p<0.02) indicating excesses of rare (presumably new) haplotypes. Under the a ssumption that rare alleles are not being maintained by balancing selection, significantly negative values of Fs, D*, and F* are evidence of recent rapid demographic expansion into the sampled area. Plotted mismatch distributions (Figure 4-4) highlight the genetic break between the Philippine population and the rest of G. philippensis, and support a hypothesis of recent expansion only for the subsample of sequences sampled outside th e Philippines. The distribution for the complete sample is double-humped indicating that it contains a subset of pairwise comparisons that have higher genetic distan ces. The second hump goes away when the Philippines birds are excluded from the analys is and the resulting distribution (Figure 44B) closely matches the expected unimodal, Poisson-shaped distribution of coalescent times in a recently expanded population under Rogers and Harpending’s (1992) infinite

PAGE 140

126 alleles model. Neither curve had a significan tly low value of r, Harpending’s measure of raggedness as determined by coalescent simulations (complete sample r = 0.0123, P[r<0.0123]=0.120; non-Philippine sample r = 0.0217, P[r<0.0217]=0.316). Phylogenetics: Hierarchical likelihood ratio tests determined that the best-fit model of nucleotide substitution was HKY + , which includes separate transition and transversion probabilities, a gamma-distribution ( = 0.1935) to accomidate among-site rate variation, and parameters for Ti:T v ratio (5.5791) and variable nucleotide frequencies (A=.3397, C=.2549, G=.1384, T=.2670). Heuristic maximum likelihood tree searches using this model found the singl e most likely tree (-ln likelihood = 1260.3399) shown in Figure 4-5. Parsimony bootstra pping analyses and Bayesian posterior probabilities indicated low suppor t (below 50%) for basal nodes that resolve interspecific relationships in Gallirallus . Support was high for the monophyly of Gallirallus species, and posterior probabilities but not boot straps were high for nodes indicating G. philippensis is paraphyletic, having given ri se to the flightless species G. rovianae (endemic to New Georgia, Solomon Islands), and † G. unnamed D and † G. unnamed C (both known only from abundant fossil bones from Rota, and Tinian, respectively, in the Northern Mariana Islands). Consistent with the parsimony network of haplotypes (Figure 4-3), there are no clades in the ML tree that correspond to geographical regions, but haplotypes sampled from the Philippines form a paraphyletic cluster at the base of G. philippensis . A likelihood ratio test of cl ock-like evolution failed to re ject the null hypothesis of no difference between the most likely tree (Figur e 4-5) and a clock-enfo rced tree (ln L= 1296.4774, -2 L= 72.2747, df=55, p>0.05) and so the ML tree can be used to estimate

PAGE 141

127 divergence times within Gallirallus and estimate the timing of the expansion of G . philippensis out of the Philippines. A matrix of HKY-corrected pairwise genetic distances was constructed using PAUP*, and divergence times were estimated by evoking published molecular clock calibrations for the hypervariable Domain I of the avian control region rather than calibrating a local clock based on the age of Wake Island (as in Chapter 3) because of weak s upport in the ML phylogeny for nodes connecting G. wakensis to other species. Using QuinnÂ’s (1992) rate of 20.8% per million years calibrated with fossil anatid s (ducks and geese), the maximum HKY-corrected genetic distance among individuals of G. philippensis (0.0484) indicates that this species originated ca. 230 ka (thousand years ago). The maximum divergence among individuals sampled outside of the Philippines (0.0157) suggests a date for the hypothesized expansion of G. philippensis out of the Philippines of ca. 75 ka. These dates correspond to nodes receiving 98% posterior probability support in the Bayesian analysis, and the latter date of ca. 75 ka represents the maxi mum age for the three flightless species that are derived from within G. philippensis . Branch lengths are short within Gallirallus relative to the outgroup species, indicating a high rate of spec iation within the genus. Discussion Control region variation in Gallirallus philippensis is high both at the level of nucleotide diversity ( = 0.01667, 40 of 325 sites were vari able including 20 singleton sites and 20 parsimony-informative sites) and at the level of haplotype diversity (Hd = 0.951, 71 birds carry 38 haplotypes). The obser ved level of genetic diversity makes it possible to examine the geographic scale of genetic structure as measured by Fst calculations. At the smallest scale of vari ation among islands within the archipelago of Vanuatu, Fst values are near zero indicating no ba rriers to gene flow between islands

PAGE 142

128 separated by <100 km. The higher total Fst value for individuals grouped by archipelago (0.424) suggests that gene flow becomes restri cted, but still not eliminated, by the larger (100 -1000 km) gaps separating archipelagoe s. This validates Mayr and Diamond’s (2001) hypothesis that “great speciators” such as G. philippensis are of intermediate vagility, willing to cross little wa ter gaps but not big ones. In ter-archipelago gene flow is probably more limited in small passerine specie s, such as monarchs and white-eyes, that show extensive species-level differentiation among Pacific ar chipelagos (Cibois et al. 2004, Filardi and Smith 2005, Slikas et al. 2000 ). The only large genetic break among populations of G. philippensis occurs between the Philippines and all other archipelagoes. High pairwise Fst values and the clustering of Philippi nes haplotypes at the basal nodes of a paraphyletic G. philippensis (Figure 4-5) support the hypothesis that the population sampled from four islands in the Philippine s is the only genetica lly distinct population, and may be the ancestral populat ion of this species. Historical demographic analyses also support a Philippine origin of G. philippensis in Oceania and suggest that there has been a recent expansion out of that archipelago (Figure 4-4). Fossil specimens of G. philippensis are known from cultural contexts in archaeological sites throughout its current range (Olson and Balouet 1989, Worthy and Holdaway 2002), and outside its current range in Marian as Islands (Kirchman and Steadman 2006c) but do not occu r in precultural strata, sugge sting that it may be a very recent arrival in Oceania where the exterm ination of flightless competitors and the opening of forest habitats by humans may ha ve enabled its colonization. Radiocarbon chronologies from early occ upation sites in the Bismarck Archipelago and Solomon Islands indicate that the region was co lonized by 35,000 36,000 yr BP (Allen et al.

PAGE 143

129 1988, Wickler and Spriggs 1988). A second human expansion began only ca. 3500 yr BP when the pottery-using, Austronesian-speak ing people known as the Lapita cultural complex undertook long-distance voyages to all archipelagoes in Oceania reaching New Caledonia and Tonga with a few centuries, a nd every inhabitable island by ca. 1000 yr BP (Kirch 2000). The estimated timing of the G. philippensis expansion into Oceania based on Quinn’s (1992) molecular clock calibra tion, ca. 75 ka, predates the expansion of humans out of Australasia and into the ar chipelagoes of Oceania by ca. 40,000 years. Two reasonable, alternative calibra tions would place the expansion of G. philippensis closer to that of humans, and support th e view that humans may have enabled its colonization by exterminating flightless compe titors and converting forested habitats to agriculture. Lambert et al. ( 2001) calibrated a rate of e volution of Domain I of the control region based on measured changes between ancestral (from ancient DNA) and modern colonies of the penguin species Pygoscelis adeliae that was 2-7 times higher than the phylogenetically calibrated rate of 20.8% per million years. Applying Lambert et al.’s (2001) slowest rate (doub ling the rate calculated above ) would make the expansion of G. philippensis roughly coincident in time w ith the human expansion. A second alternative is to calibrate a local Gallirallus clock using the maximum age of the flightless species † G. wakensis, endemic to low-lying Wake Island, which is thought to have been submerged by the ca. +6 m eustat ic sea-level rise that occurred 124,000 years ago. The minimum HKY-corrected genetic distance between † G. wakensis and any of its potential sister species is 3.8%, yielding a ra te of 30.6% per million years and a time for the origin of G. philippensis of ca. 50,900 years ago. The faster rate of CR Domain I evolution obtained using the Wake Island cal ibration mirrors the faster rates obtained

PAGE 144

130 with the same calibration for the combined and cyt b clocks used in Chapter 3. Not surprisingly, the date of origin of G. philippensis obtained with this rate (50.9 ka) is closer to the estimated age of that para phyletic species (22.4 30.8 ka) based on other genes but the same calibration point. Although these molecular clock estimates of divergence times in Gallirallus are crude, it is clear from the short branch lengt hs separating species in Figure 4-5 that the time between speciation events is short in this group. It is also clear that G. philippensis must have given rise to the flightless species G. rovianae, † G. unnamed D, and † G. unnamed C very recently. Under this scenario, one following the Biological Species Concept may recognize a paraphyletic G. philippensis, but one following a Phylogenetic Species Concept might consider rovianae and unnamed C to be flightless populations of G. philippensis . A different scenario involving hy bridization may also explain the observed phylogeny by hypothesizing that m itochondrial DNA has recently transgressed into G. rovianae, † G. unnamed D, and † G. unnamed C following “secondary” contact with G. philippensis . This alternative can be inve stigated further by sampling more individuals of G. rovianae (currently only two specimens exist; genetic sampling of the type specimen is underway to verify the results presented herein) and by sequencing nuclear genes. A similar investigation of the extinct species k nown only from fossil bones from the Marianas, may be impo ssible. If hybrid ization between G. philippensis and G. rovianae is supported by nuclear DNA data, the breach of repr oductive isolation would justify a phylogenetic criter ion for species recognition. Ever since the time of Darwin and Wallace, patterns of geographic variation in birds from Pacific islands have played an important role in the formulation of

PAGE 145

131 evolutionary and biogeographic theory. Variation among bird populations in the Southwest Pacific became an important focu s of speciation resear ch after Mayr (1942, 1963) interpreted plumage differences as evid ence of reproductive is olation, but, as Mayr and Diamond (2001) recently lamented, almost no molecular genetic data have been collected to test for genetic isloation of avian populations, de spite great interest in the avifauna by ecologists, evolu tionists, biogeographers, and co nservationists. Pioneering molecular phylogenetic studies of Pacific passerines by Filardi and Smith (2005) and Filardi and Moyle (2005) have revised hypothesized relationships and colonization patterns that previously were based only on morphology, but th e present study is the first intraspecific genetic survey of non-passerine birds in Southwest P acific archipelagoes, and also is the first to examine the history of colonization and divergence in a lineage that has given rise to flightless species with in and beyond its present range. Like many previous studies of geographic variation in bi rds, the present study ch allenges traditional views of speciation based on pa tterns of plumage variation, finding little support for the validity of subspecies or for genetic di vergence of island po pulations following colonization from continental sources. Rather , my data are consiste nt with the finding of Filardi and Moyle (2005) that Australia is a target of colonization from island Oceania, although it remains a likely source of island colonists in other groups of birds (e.g., Psittacidae, Meliphagidae). Further progress testing 20th century models of speciation will come as more population genetics data shed additional light on the role of hybridization, and as better estim ates of divergence times clar ify the temporal context of population subdivision.

PAGE 146

132Table 4-1. Voucher specimen information fo r all individuals from whic h control region sequences were obtained. Taxonomy for subspecies of Gallirallus philippensis follows Taylor 1998. AMNH = American Museum of Natural History, FMNH = Field Museum of Natural History, MC Z = Museum of Comparative Zoology, NMNH = National Museum of Natural History, Smithsonian Institution, UF = Florida Museum of Na trural History, University of Florida, UWBM = Thomas Burke Museum, University of Washington. Haplotype num bers refer to those depi cted in Figures 4-2, 4-5. Taxon Specimen number Locality Date Source Sequence Length Haplotype G. p. philippensis MCZ 332899 Flores, Greater Sunda Is. 04 Jul. 1960 Toe pad 325 1 G. p. philippensis MCZ 270102 Sulawesi, Greater Sunda Is. 23 Sep. 1939 Toe pad 325 11 G. p. philippensis NMNH 248149 Sulawesi, Greater Sunda Is. 13 Dec. 1914 Toe pad 325 1 G. p. philippensis NMNH 250754 Sulawesi, Greater Sunda Is. 22 Mar. 1917 Toe pad 325 4 G. p. philippensis NMNH 582808 Batan, Philippines 01 Jun. 1985 Toe pad 325 30 G. p. philippensis MCZ 194032 Lagengilang, Philippines 12 Jan. 1937 Toe pad 325 32 G. p. philippensis FMNH 422631 Luzon, Philippines 06 Apr. 1960 Toe pad 325 31 G. p. philippensis FMNH 419840 Luzon, Philippines 19 Apr. 1959 Toe pad 325 33 G. p. philippensis NMNH 578152 Mindanao, Philippines 19 Dec. 1962 Toe pad 325 37 G. p. philippensis MCZ 64506 Mindanao, Philippines 31 Jul. 1931 Toe pad 325 36 G. p. philippensis FMNH 425072 Mindanao, Philippines 18 Oct. 1968 Toe pad 325 36 G. p. philippensis FMNH 425073 Mindanao, Philippines 20 Mar. 1968 Toe pad 325 35 G. p. philippensis FMNH 425074 Mindanao, Philippines 23 Oct. 1968 Toe pad 325 36 G. p. philippensis FMNH 425075 Mindanao, Philippines 24 Oct. 1968 Toe pad 325 37 G. p. philippensis ? UF 44633 Captive: Miami Metro Zoo 03 May 2005 Frozen tissue 325 24 G. p. mellori FMNH 407593 NSW, Australia May 1928 Toe pad 325 26 G. p. mellori FMNH 417306 Queensland, Australia 12 Apr. 1947 Toe pad 325 15 G. p. mellori UWBM 78189 Heron Island, Australia 15 Aug. 2002 Frozen tissue 325 22 G. p. mellori UWBM 62806 NSW, Australia 25 Feb. 1995 Frozen tissue 325 14 G. p. mellori UWBM 78193 Heron Island, Australia 20Jul. 2000 Frozen tissue 325 22 G. p. mellori UWBM 62963 NSW, Australia 01 Jul. 1993 Frozen tissue 325 24 G. p. mellori UWBM 62974 Australia none Frozen tissue 325 5 G. p. mellori UWBM 62975 NSW, Australia 06 Jun. 1995 Frozen tissue 325 22 G. p. mellori NMNH 405631 NT, Australia 16 Oct. 1948 Toe pad 325 21 G. p. mellori NMNH405632 NT, Australia 16 Oct. 1948 Toe pad 325 25 G. p. mellori NMNH 278019 NSW, Australia 18 Sep. 1919 Toe pad 325 26 G. p. admiralitatus NMNH 377620 Pityilu, Admiralties 15 Apr. 1945 Toe pad 325 29

PAGE 147

133Table 4-1 Continued Taxon Specimen number Locality Date Source Sequence Length Haplotype G. p. admiralitatus AMNH 336225 Ninigo, Admiralties 05 Jun. 1934 Toe pad 325 29 G. p. praedo AMNH 335161 Skoki, Sabben Is. 20 Dec. 1933 Toe pad 325 12 G. p. meyeri AMNH 333057 New Britain, Bismarck Is. 09 Sep. 1932 Toe pad 325 1 G. p. meyeri FMNH 407350 New Britain, Bismarck Is. May 1928 Toe pad 325 1 G. p. lesouefi FMNH 407170 Wanton Is., Bismarck Is. 20 Nov. 1928 Toe pad 325 13 G. p. lesouefi AMNH 335590 Boan g, Bismarck Is. 06 Feb. 1935 Toe pad 325 1 G. p. reductus AMNH 704717 New Guinea 05 May 1952 Toe pad 325 10 G. p. reductus NMNH 518916 New Guinea 18 Aug. 1960 Toe pad 325 3 G. p. christophori AMNH 227846 San Christobal, Solomon Is. 16 Dec. 1929 Toe pad 325 16 G. p. swindellsi AMNH 337536 Loyaltiy Is. 18 Dec. 1937 Toe pad 325 2 G. p. swindellsi NMNH 77060 New Caledonia 24 Dec. 1877 Toe pad 325 1 G. p. pelewnsis NMNH 384774 Garakayo, Palau Is. 19 Sep. 1945 Toe pad 325 28 G. p. pelewnsis MCZ 264024 Koror, Palau Is. 28 Sep. 1930 Toe pad 325 34 G. p. pelewnsis NMNH 385486 Peliliu, Palau Is. 28 Aug. 1945 Toe pad 325 28 G. p. yorki NMNH 572364 Halm ahera, Moluccas 01 Jan. 1985 Toe pad 325 23 G. p. assimilis MCZ 39925 North I., New Zealand Jul. 1898 Toe pad 325 15 G. p. assimilis NMNH 109118 North I., New Zealand Sep. 1882 Toe pad 325 24 G. p. assimilis NMNH 109119 North I., New Zealand May 1882 Toe pad 325 1 G. p. sethsmithi NMNH 277303 Taviuni, Fiji 30 Nov. 1923 Toe pad 325 2 G. p. sethsmithi UF 43175 Santo, Vanuatu 17 Jun. 2003 Frozen tissue 325 17 G. p. sethsmithi UF 43176 Santo, Vanuatu 18 Jun. 2003 Frozen tissue 325 2 G. p. sethsmithi UF 43153 Santo, Vanuatu 18 Jun. 2003 Frozen tissue 325 27 G. p. sethsmithi UF 43154 Santo, Vanuatu 20 Jun. 2003 Frozen tissue 325 2 G. p. sethsmithi UF 43150 Santo, Vanuatu 17 Jun. 2003 Frozen tissue 325 18 G. p. sethsmithi UF 42933 Santo, Vanuatu 07 Nov. 2002 Frozen tissue 325 27 G. p. sethsmithi UF 42934 Santo, Vanuatu 10 Nov. 2002 Frozen tissue 325 2 G. p. sethsmithi UF 42935 Santo, Vanuatu 10 Nov. 2002 Frozen tissue 325 2 G. p. sethsmithi UF 42878 Santo, Vanuatu 10 Nov. 2002 Frozen tissue 325 7 G. p. sethsmithi UF 42902 Santo, Vanuatu 10 Nov. 2002 Frozen tissue 325 2 G. p. sethsmithi UF 43177 Santo, Vanuatu 18 Jun. 2003 Frozen tissue 325 8 G. p. sethsmithi UF 43180 Uripiv, Vanuatu 20 Jun. 2003 Frozen tissue 325 27

PAGE 148

134Table 4-1 Continued Taxon Specimen number Locality Date Source Sequence Length Haplotype G. p. sethsmithi UF 43178 Uripiv, Vanuatu 20 Jun. 2003 Frozen tissue 325 19 G. p. sethsmithi UF 43179 Uripiv, Vanuatu 20 Jun. 2003 Frozen tissue 325 9 G. p. sethsmithi UF 43181 Uripiv, Vanuatu 21 Jun. 2003 Frozen tissue 325 2 G. p. sethsmithi UF 43182 Uripiv, Vanuatu 21 Jun. 2003 Frozen tissue 325 2 G. p. sethsmithi UF 43221 Efate, Vanuatu 03 Jul. 2003 Frozen tissue 325 2 G. p. sethsmithi UF 43222 Efate, Vanuatu 03 Jul. 2003 Frozen tissue 325 2 G. p. sethsmithi UF 43223 Efate, Vanuatu 04 Jul. 2003 Frozen tissue 325 2 G. p. sethsmithi UF 43224 Efate, Vanuatu 08 Jul. 2003 Frozen tissue 325 6 G. p. sethsmithi UF 43225 Efate, Vanuatu 08 Jul. 2003 Frozen tissue 325 2 G. p. goodsoni MCZ 96960 Manua, Samoa 21 Dec. 1923 Toe pad 325 20 G. p. goodsoni NMNH 576839 Tutuila, Samoa 10 Jul. 1976 Toe pad 325 20 G. p. goodsoni NMNH 493500 Tutuila, Samoa 8 Mar. 196? Toe pad 325 20 G. p. goodsoni UF 39854 Ofu, Samoa 06 Apr. 1995 Toe pad 325 20 G. torquatus MCZ 269535 Malukken, Greater Sunda Is 27 Sep. 1938 Toe pad 325 G. torquatus MCZ 270553 Sulawesi, Lesser Sunda Is. 17 Aug. 1938 Toe pad 325 G. australis FMNH 67208 North Is., New Zealand 1886 Toe pad 123 G. unnamed C UF 60153 Tinian, Mariana Is. ca. 2500 yr BP Fossil bone 266 † G. wakensis MCZ 157073 Wake I. 1923 Toe pad 325 † G. wakensis AMNH 265485 Wake I. 1892 Toe pad 325 G. insignis AMNH 777724 New Britain, Bismark Is. 30 Dec. 1958 Toe pad 325 G. insignis AMNH 333800 New Britain, Bismark Is. 10 Mar. 1933 Toe pad 325 † G. modestus MCZ 39926 Chatham Is. Mar. 1891 Toe pad 325 G. rovianae CEF 878 New Georgia, Solomon Is. 08 May 2004 Frozen tissue 325 G. owstoni UF 42967 Guam, Mariana Is. (captive) 2002-2004 Frozen tissue 325 G. owstoni UF 42968 Guam, Mariana Is. (captive) 2002-2004 Frozen tissue 325 G. owstoni UF 42969 Guam, Mariana Is. (captive) 2002-2004 Frozen tissue 325 G. woodfordi UWBM 69746 Isabel, Solomon Is. 13 Jul.1997 Frozen tissue 325 G. woodfordi UWBM 69782 Isabel, Solomon, Is. 24 Jul. 1997 Frozen tissue 325 G. woodfordi UWBM 58788 Isabel, Solomon Is. 13 Jul. 1997 Frozen tissue 325 G. woodfordi UWBM 69783 Isabel, Solomon Is. 24 Jul. 1997 Frozen tissue 325 † G. unnamed D UF 62963 Rota, Mariana Is. ca. 2500-250 yr BP Fossil bone 266 -

PAGE 149

135Table 4-1 Continued Taxon Specimen number Locality Date Source Sequence Length Haplotype † G. ripleyi UF 54277 Mangaia, Southern Cook Is. ca. 1000 yr BP Fossil bone 228 Lewinia pectoralis AMNH 802028 New Guinea 18 Jun. 1965 Toe pad 123 Lewinia pectoralis AMNH 765544 New Guinea 26 Mar. 1954 Toe pad 123

PAGE 150

136 Figure 4-1. Oceania, showing the distribution of Gallirallus philippensis (dotted outline). Arrows poi nt to islands from which G. philippensis was sampled. Underlined names indicate ar chipelagos from which endemic flightless Gallirallus species were sampled.

PAGE 151

Figure 4-2. The Gallirallus mitochondrial control region, showing the location of PCR primers used to amplify a 325 bp portion of Domain I.

PAGE 152

138 Figure 4-3. Maximally parsimonious network (u nrooted) of 38 sampled (ovals) and 21 unsam pled (circles) control region haplotype s. Each line connecting haplotypes represents one nucleotide subs titution. Names inside ovals i ndicate the place(s) from which each hapotype was sampled. Numbers in parentheses are arbitrarily assigne d haplotype numbers (see also Table 41, Figure 4-5).

PAGE 153

139 Figure 4-4. Frequency distributi ons of the number of pairwi se nucleotide substitutions between A) all 71individuals of Gallirallus philippensis , and B) 61 individuals sampled outside of the Philippines.

PAGE 154

140 Figure 4-5. Maximum likelihood phylogeny of 38 Gallirallus philippensis haplotypes (numbered as in Figure 4-2 and Table 4-1) and individuals from ten other Gallirallus species, of which only G. torquatus is volant. The tree is rooted to the outgroup species Lewinia pectoralis. Numbers above branches refer to Bayesian posterior prob abilities (above) and pars imony bootstrap support for nodes receiving >50% support.

PAGE 155

141 LIST OF REFERENCES Allen, D., C. Olineros, C. Española, G. Br oad, and J. C. T. Gonzalez. 2004. A new species of Gallirallus from Calayan island, Philippines. Forktail 20: 1-7. Allen, J., C. Gosden, R. Jones, and J. P. White. 1988. Pleistocene dates for the human occupation of New Ireland, Northe rn Melanesia. Nature 331:707-709. Amesbery, J. R., D. R. Moore, and R. L. Hunter-Anderson. 1996. Cultural adaptations and Late Holocene sea level change in th e Marianas: recent excavations at Chalan Piao, Saipan, Micronesia. Bulletin of th e Indo-Pacific Prehistory Association 15:53-69. Avise, J., 2000. Phylogeography. Harvar d University Press, Cambridge, MA. Baker, A. J., and H. D. Marshall. 1997. M itochondrial control regi on sequences as tools for understanding evolution. Pages 51-82 in Avian molecular evolution and systematics (D. P. Mindell, Ed.). Academic Press, San Diego, CA. Balouet, J. C., and S. L. Olson. 1989. Fossil birds from late Quaternary deposits in New Caledonia. Smithson. Contri butions to Zoology, No. 469. Baumel, J. L., and L. M. Witmer. 1993. Osteologia. Pages 45-132 in J. J. Baumel, A. S. King, J. E. Breazile, H. E. Evans, and J. C. Vanden Berge, eds. Handbook of Avian Anatomy: Nomina Anatomica Avium, 2nd Ed. Nuttall Ornith. Club, Cambridge, MA. Bellwood, P. 1972. Settlement pattern survey, Hanatekua Valley, Hiva Oa, Marquesas Islands. Pacific Anthropology Records, No. 17, Bernice P. Bishop Museum, Honolulu, HI. Boom, R., C. J. A. Sol, M. M. M. Salimans, C. L. Jansen, P. M. E. Wertheim-van Dillen, and J. van der Noordaa. 1990. Rapid and si mple method for purification of nucleic acids. Journal of Clinic al Microbiology 28:495-503. Brown, R. W. 1956. Composition of Scientific Words. Smithsonian Institution Press, Washington D.C. Burley, D. V. 1994. Settlement pattern and To ngan prehistory reconsiderations from Ha’apai. Journal of the Polynesian Society 103:379-411.

PAGE 156

142 Burley, D. V. 1999. Lapita settlement to th e East: New data and changing perspectives from Ha’apai (Tonga) prehistory. Pages 189-200 in The Pacific from 5000 to 2000 BP (J. C. Galipaud and I. Lilley, Eds. ,) Institut de Recherche Pour Le Développment, Paris. Butler, B. M. 1994. Early prehistoric settleme nt in the Mariana Islands: New evidence from Saipan. Man and Culture in Oceania, 10,15-38. Chappell, J., and N. J. Shackleton. 1986. Oxygen isotopes and sea level. Nature 324:137140. Cheviron, Z. A., S. J. Hackett, and A. P. Capparella. 2005. Complex evolutionary history of a Neotropical lowl and forest bird ( Lepidothrix coronata ) and its implications for historical hypotheses of the origin of Neotropical avian diversity. Molecular Phylogenetics and Evolution 36:338-357. Churchward, C. M. 1959. Tongan dictionar y. Government of Tonga, Nuku’alofa. Cibois, A., J.-C. Thibault, and E. Pasque t. 2004. Biogeography of Eastern Polynesian monarchs ( Pomarea ): an endemic genus close to extinction. Condor 106:837-851. Clement, M., D. Posada, and K. Crandall. 2000. TCS: a computer program to estimate gene geneologies. Molecular Ecology 9:1657-1660 Conte, E., and A. Anderson. 2003. Radiocarbon ages for two sites on Ua Huka, Marquesas. Asian Pe rspectives 42:155-160. Coyne, J. A., and H. A. Orr. 2004. Speciati on. Sinauer Associates, Inc., Sunderland, MA. Craib, J. L. 1993. Early occ upation at Unai Chulu, Tini an, Commonwealth of the Northern Mariana Islands. Bulletin of the Indo-Pacific Preh istory Association, 13:116-134. Darwin, C. 1859. The origin of species by mean s of natural selecti on or the preservation of favoured races in the str uggle for life. John Murray, London. Diamond, J. M. 1991. A new species of rail from the Solomon Islands and convergent evolution of insular flig htlessness. Auk 108:461-470. Dickinson, W. R. 1998. Geomorphology and geodynamics of the Cook-Austral-IslandSeamount Chain in the South Pacific Ocean: implications for hotspots and plumes. International Geologi cal Review 40:1039-1075. Dye, T. and D. W. Steadman. 1990. Polynesi an ancestors and their animal world. American Scientist 78:209-217. Falla, R. A. 1954. A new rail from cave deposits in the North Island of New Zealand. Records of the Auckland In stitute and Museum 4:241-244.

PAGE 157

143 Feduccia, A. 2000. The origin and evolution of birds, 2nd Ed. Yale University Press, New Haven, CT. Filardi, C. E., and R. G. Moyle. 2005. Si ngle origin of a pan-Pacific bird group and upstream colonization of Au stralia. Nature 438:216-219. Filardi, C. E., and C. E. Smith. 2005. Mo lecular phylogenetics of monarch flycatchers (genus Monarcha ) with emphasis on Solomon Island endemics. Molecular Phylogenetics Evolution 37:776-788. Fleischer, R. C., C. E. McIntosh, and C. L. Tarr. 1998. Evolution on a volcanic conveyor belt: using phylogeographic re constructions and K-Ar-based ages of the Hawaiian Islands to estimate molecular evolutionary rates. Molecular Ecology and Evolution 7:533-545. Fleischer, R. C., S. L. Olson, H. F. James, and A. C. Cooper. 2000. The identity of the extinct Hawaiian eagle ( Haliaeetus ) as determined by mitochondrial DNA sequence. Auk 117:1051-1056. Fu, Y.-X. 1997. Statistical tests of neutra lity of mutations against population growth, hitchhiking and background selection. Genetics 147:915-925. Fu, Y.-X., and W.-H. Li. 1993. Statistical te sts of neutrality of mutations. Genetics 133:693-709. Fuller, E. 1987. Extinct birds. Viking Press, London. Gill, F. B. 1995. Ornithology, 2nd Edition. W. H. Freeman, New York. Glenn, T. C., W. Stephan, and M. J. Bra un. 1999. Effects of a population bottleneck on whooping crane mitochondrial DNA vari ation. Conservation Biology 13:10971107. Grant, P. R., and B. R. Grant. 1994. Phenot ypic and genetic effect s of hybridization in Darwin’s finches. Evolution 48:297-316. Grant, P. R., and B. R. Grant. 1998. Speci ation and hybridization of birds on islands. Pages 142-162 in Evolution on Islands (P. R. Grant, Ed.). Oxford University Press, Oxford. Harpending, H., 1994. Signature of ancien t population growth in a low-resolution mitochondrial DNA mismatch dist ribution. Human Biology 66:591-600. Hoss, M., and S. Pääbo. 1993. DNA extraction from Pleistocene bones be a silica-based purification method. Nucleic Acids Research 21:3913-3914. Hudson, R. R., M. Slatkin, and W. P. Maddis on. 1992. Estimation of levels of gene flow from DNA sequence data. Genetics 132:583-589.

PAGE 158

144 Huelsenbeck, J. P., and F. Ronquist. 2001. MRBAYES: Bayesian inference of phylogenetic trees. Bioinformatics 17:745-755. James, H. F., and S. L. Olson. 1991. Descripti on of thirty-two new sp ecies of birds from the Hawaiian Islands: Part II. Passerifo rmes. Ornithological Monographs, No. 46. American OrnithologistsÂ’ Union, Washington, D.C. Kimball, R. T., E. L. Braun, J. D. Ligon, V. Lucchini, and E. Randi. 2001. A molecular phylogeny of the peacock-pheasants (Galliformes: Polyplectron spp.) indicates loss and reduction of ornamental tr aits and display behaviours. Biological Journal of the Linnean Society 73:187-198. Kirch, P. V. 1984. The Evolution of the Pol ynesian Chiefdoms. Cambridge University Press, Cambridge, UK. Kirch, P. V. 1986. Rethinking East Polynesian prehistory. Journa l of the Polynesian Society 95:9-40. Kirch, P. V. 2000. On the Road of the Winds : An Archaeological History of the Pacific Islands Before European Contact. Univ. California Press, Berkeley. Kirch, P. V. and J. Ellison. 1994. Palaeoen vironmental evidence for human colonization of remote Oceanic islands. Antiquity 68:310-321. Kirch, P. V., D. W. Steadman, V. L. Butler, J. Hather and M. Weisler. 1995. Prehistory and human ecology in Eastern Polynesia: excavations at Tangatatau Rockshelter, Mangaia, Cook Islands. Archaeology in Oceania 30:47-65. Kirchman, J. J., and D. W. Steadman. 2005. Rails (Aves: Rallidae: Gallirallus ) from prehistoric sites in the Kingdom of Tonga, including description of a new species. Proceedings of the Biological Society of Washington 118:465-477. Kirchman, J. J., and D. W. Steadman. 2006a. New species of rails (Aves: Rallidae) from an archaeological site on Huahine, So ciety Islands. Pacific Science 60:281-297. Kirchman, J. J., and D. W. Steadman. 2006b. New extinct rails (Aves; Rallidae) from archaeological sites in the Marquesas Islands, French Polynesia. Pacific Science 60:xxx-xxx. Kirchman, J. J., and D. W. Steadman. 2006c. Rails (Rallidae: Gallirallus ) from prehistoric archaeological and paleontol ogical sites in Western Oceania. Zootaxa xx:xx-xx. Kirchman, J. J., S. J. Hackett, S. M. Goodman, and J. M. Bates. 2001. Phylogeny and systematics of ground rolle rs (Brachypteraciidae) of Madagascar. Auk 118:849863.

PAGE 159

145 Kirchman, J. J., L. A. Whittingham, and F. H. Sheldon. 2000. Relationships among Cave Swallow populations ( Petrochelidon fulva ) determined by comparisons of microsatellite and cytochrome b data. Molecular Phylogenetics and Evolution 14:107-121. Koopman, K. F., and D. W. Steadman. 1995. Extinction and biogeography of bats on 'Eua, Kingdom of Tonga. Ameri can Museum Novitates 3125:1-13. Krajewski, C., and D. G. King. 1996. Mol ecular divergence and phylogeny: Rates and patterns of cytochrome b evolution in cranes. Mol ecular Biology and Evolution 13:21-30. Kratter, A. W., Kirchman, J. J., Steadman, D. W., In press. Upland bird communities on Santo, Vanuatu. Wilson Bulletin. Kvist, L., J. Broggi, J. C. Illera, and K. Koivula. 2005. Colonisati on and diversification of the blue tits ( Parus caeruleus teneriffae -group) in the Canary Islands. Molecular Phylogenetics and Evolution 34:501-511. Lack, D. 1947. DarwinÂ’s finches. Cambridge University Press, Cambridge, UK. Lambert, D. M., P. A. Ritchie, C. D. Mill ar, B. Holland, A. J. Drummond, and C. Baroni. 2002. Rates of evolution in ancient DNA from adelie penguins. Science 295:22702273. Livezey, B. C. 1998. A phylogenetic analysis of the Gruiformes (Aves) based on morphological characters, with an emphasi s on the rails (Rallidae). Philosophical Transactions of the Royal Soci ety of London, Series B, 353:2077-2151. Livezey, B. C. 2003. Evolution of flightle ssness in rails (Gruiformes: Rallidae): Phylogenetic, ecomorphological, and ontoge netic perspectives. Ornithological Monographs, No. 53. American Ornit hologistsÂ’ Union, Washington, D.C. Marchant, S., and P. J. Higgins. 1993. Handbook of Australian, New Zealand, and Antarctic birds, Volume 2. Oxford University Press, Melbourne. Mayr. E. 1933. Birds collected during the Whitney South Seas Expedition, XXIII. American Museum Novitates 609:3-4. Mayr. E. 1938. Birds collected during th e Whitney South Seas Expedition, XL. American Museum Novitates 1007:4-8. Mayr, E. 1942. Systematics and the Origin of Species. Columbia University Press, New York. Mayr. E. 1949. Notes on the birds of Nort hern Melanesia, 2. American Museum Novitates 1417:5-7.

PAGE 160

146 Mayr, E. 1963. Animal Species and E volution. Belknap Press, Cambridge, MA. Mayr. E., and M. Guilleard. 1951. New speci es and subspecies of birds from the highlands of New Guinea. American Museum Novitates 1524:2-4. Mayr, E., and J. Diamond. 2001. The birds of Northern Melanesia. Oxford University Press, New York. McNab, B. K. 1994a. Energy conservation and th e evolution of flightlessness in birds. Americam Naturalist 144:628-642. McNab, B.K. 1994b. Resource use and the survival of land and freshwater vertebrates on oceanic islands. Americam Naturalist 144:643-660. McNab, B. K. 2002. The physiological ecology of vertebrates: a view from energetics. Cornell Univ. Press, Ithaca, NY. McNab, B. K., and H. I. Ellis. 2006. Flight less rails endemic to islands have lower energy expenditures and clutch sizes than flighted rails on islands and continents. Comparative Biochemistry and Physiology 144: xx-xx. Merilä, J., M. Björklund, and A. J. Baker. 1997. Historical demogr aphy and present day population structure of the greenfinch, Carduelis chloris – an analysis of mtDNA control -region sequenc es. Evolution 51:946-956. Nei, M. 1987. Molecular Evolutionary Genetics . Columbia University Press, New York. Olson, S. L. 1973a. Evolution of the rails of th e South Atlantic Islands (Aves: Rallidae). Smithsonian Contributions to Zoolology, No. 152. Smiths onian Institution Press, Washington, D.C. Olson, S. L. 1973b. A classification of the Rallidae. Wilson Bulletin 85:381-416. Olson, S. L., and H. F. James. 1991. Descripti on of thirty-two new sp ecies of birds from the Hawaiian Islands: Part I. Non-Passeri formes. Ornithological Monographs No. 46. American Ornithologists’ Union, Washington, D.C. Omland, K. E., J. Marzluff, W. Boarman, C. L. Tarr, and R. C. Fleischer. 2000. Cryptic genetic variation and paraphyly in ravens . Proceedings of the Royal Society of London, Series B 267:2475-2482. Paxinos, E. E., H. F. James, S. L. Olson, M. D. Sorenson, J. Jackson, and R. C. Fleischer. 2002. mtDNA from fossils reveals a radia tion of Hawaiian geese recently diverged from the Canada Goose (Branta canadensi s). Proceedings of the National Academy of Sciences, USA 99:1399-1404.

PAGE 161

147 Pearce, R. L., J. J. Wood, Y. Artukhin, T. P. Birt, M. Damus, and V. L. Friesen. 2002. Mitochondrial DNA suggests high gene fl ow in Ancient Murrelets. Condor 104, 84-91. Peters, J. L. 1934. Check-list of birds of the world, Vol. 2. Museum of Comparative Zoolgy, Cambridge, MA. Pimm, S. L., M. P. Moulton, and L. J. Jus tice. 1994. Bird extinctions in the central Pacific. Philosophical Transactions of the Royal Society of London, Series B 344:27-33. Posada, D., Crandall, K. A., 1998. Modeltes t: testing the model of DNA substitution. Bioinformatics 14:817-818. Pregill, G. K. 1993. Fossil lizards from th e late Quaternary of ‘Eua, Tonga. Pacific Science 47:101-114. Pregill, G. K. and D. W. Steadman 2004. South Pacific iguanas: human impacts and a new species. Journal of Herpetology 38:15-21. Quinn, T. W., 1992. The genetic legacy of Mother Goose – phylogeographic patterns of lesser snow goose Chen caerulescens maternal lineages. Molecular Ecology 1:105117. Remsen, J. V., Jr. and T. A. Parker, III. 1990. Seasonal distribution of the azure gallinule ( Porphyrula flavirostris ), with comments on vagrancy in rails and gallinules. Wilson Bulletin 102:380-399. Ricklefs, R., and G. W. Cox. 1972. Taxon cycles in the West Indian avifauna. American Naturalist 106:195-219. Ripley, S. D. 1977. Rails of the world. Godine, Boston, MA. Roff, D. A. 1990. The evolution of flight lessness in insects. Ecological Mongraphs 60:389-421. Roff, D. A. 1994. The evolution of flightle ssness: is history important? Evolutionary Ecology 8:639-657. Rogers, A. R., and H. Harpending. 1992. Population growth makes waves in the distribution of pairwise genetic differences. Molecu lar Biology and Evolution 9:552-569. Rolett, B. V. 1998. Hanamiai: prehistory, colonization, and cultural change in the Marquesas Islands (East Polynesia). Yale University Publications in Anthropology, No. 81.

PAGE 162

148 Rozas, J., J. C. Sanchez-DelBarrio, X. Messeguer, and R. Rozas. 2003. DnaSP, DNA sequence polymorphism analyses by the coalescent and other methods. Bioinformatics 19:2496-2497. Salzburger, W., J. Martens, and C. Sturmbau er. 2002. Paraphyly of the Blue Tit (Parus caeruleus) suggested from cytochrome b sequences. Molecular Phylogenetics and Evolution 24:19-25. Sambrook, J., E. F. Fritsch, and T. Mania tis. 1989. Molecular cloning: a laboratory manual, 2nd Edition. Cold Spring Harbor Labor atory, Cold Spring Harbor, NY. Sato, A., C. Oh’Uigin, F. Figueroa, P. R. Gr ant, B. R. Grant, H. Tichy, and J. Klein. 1999. Phylogeny of Darwin’s finche s as revealed by mtDNA sequences. Proceedings of the National Academy of Sciences, USA 96:5101-5106. Schluter, D. 2000. The Ecology of Adaptive Radi ation. Oxford University Press, Oxford. Schodde, R., De Naurois, R., 1982. Patterns of variation and dispersal in the Buffbanded Rail ( Gallirallus philippensis ) in the south-west Pacifi c, with description of a new subspecies. Notornis 29:131-142. Shields, G. F., and A. C. Wilson. 1987. Ca libration of mitochondrial DNA evolution in geese. Journal of Mol ecular Evolu tion 24:212-217. Sinoto, Y. H. 1966. A tentative prehistoric cu ltural sequence in th e northern Marquesas Islands, French Polynesia. Journal of the Polynesian Society 75:287-303. Sinoto, Y. H. 1970. An archaeologically base d assessment of the Marquesas Islands as a dispersal center in East Polynesia. Pacific Anthropology Records 11:105-132. Sinoto, Y. H. 1975. Report on the preliminary ex cavation of an early habitation site on Huahine, Society Islands. Journal de la Société des Océanistes 31:143-186. Sinoto, Y. H. 1979. Excavations on Huahine, Fr ench Polynesia. Pacific Studies 3:1-40. Sinoto, Y. H. 1983. Archaeological excavations of the Vaito ’tia and Fa ’ahia sites on Huahine Island, French Polynesia. Nationa l Geographic Society Research Reports 15:583-599. Slikas, B., I. B. Jones, S. R. Derricks on, and R. C. Fleischer. 2000. Phylogenetic relationships of Micronesian white-eye s based on mitochondrial sequence data. Auk 117:355-365. Slikas, B., S. L. Olson, and R. C. Fleisc her. 2002. Rapid, independent evolution of flightlessness in four specie s of Pacific Island rails (Ral lidae): an analysis based on mitochondrial sequence data. J ournal of Avian Biology 33:5-14.

PAGE 163

149 Sorenson, M. D., J. C. Ast, D. E. Dimcheff, T. Yuri, and D. P. Mindell. 1999. Primers for a PCR-based approach to mitochondria l genome sequencing in birds and other vertebrates. Molecular Phyloge netics and Evolution 12:105-114. Spriggs, M., and A. Anderson. 1993. Late colo nization of East Polynesia. Antiquity 67:200-217. Spriggs, M., and S. Beford. 2001. Arapus: A Lap ita site at Mangaasi in central Vanuatu? Pages 93-103 in Terra Australis 17: The archaeo logy of Lapita dispersal in Oceania: Papers from the fourth Lapita Conference, June 2000, Canberra, Australia. Steadman, D. W. 1986. Two new species of rails (Aves: Rallidae) from Mangaia, Southern Cook Islands. Pacific Science 40:27-43. Steadman, D. W. 1987. Two new species of rails (Aves: Rallidae) from Mangaia, Southern Cook Islands. Pacific Science 40:38-54. Steadman, D. W. 1988. A new species of Porphyrio (Aves: Rallidae) from archaeological sites in the Marquesas Islands. Proceedings of the Biological Society of Washington 101:162-170. Steadman, D. W. 1989a. Extinction of birds in Eastern Polynesia: A review of the record, and comparisons with other Pacific island groups. Journal of Archaeological Science 16:177-205. Steadman, D. W. 1989b. A new sp ecies of starling (Sturnidae, Aplonis ) from an archaeological site on Huahine, So ciety Islands. Notornis 36:161-169. Steadman, D. W. 1992a. New species of Gallicolumba and Macropygia (Aves: Columbidae) from archaeological sites in Polynesia. Los Angeles County Museum of Natural History, Science Series 36:329-348. Steadman, D. W. 1992b. Extinct and extirpat ed birds from Rota, Mariana Islands. Micronesica 25:71-84. Steadman, D. W. 1993. Biogeogr aphy of Tongan birds before and after human impact. Proceedings of the National Academy of Sciences USA 90:818-822. Steadman, D. W. 1995. Prehisto ric extinctions of Pacific is land birds: biodiversity meets zooarchaeology. Science 267:1123-1131. Steadman, D. W. 1999. The prehistory of ve rtebrates, especially birds, on Tinian, Aguiguan, and Rota, Northern Mari anas Islands. Micronesica 31:319-345. Steadman, D. W. 2002. A new species of gull (Laridae: Larus ) from an archaeological site on Huahine, Society Islands. Pro ceedings of the Biological Society of Washington 115:1-17.

PAGE 164

150 Steadman, D.W. 2006. Extinction and biogeography of tropical Pacific birds. University of Chicago Press, Chicago, IL. Steadman, D. W., and M. Intoh. 1994. Bioge ography and prehistoric exploitation of birds from Fais Island, Yap State, Federate d States of Micronesia. Pacific Science 48:116-135. Steadman, D. W., and P. V. Kirch. 1990. Pr ehistoric extinction of birds on Mangaia, Cook Islands, Polynesia. Proceedings of the National Academy of Sciences USA 87:9605-9609. Steadman, D. W., and P. V. Kirch. 1998. Bioge ography and prehistoric exploitation of birds from the Mussau Islands, Papua New Guinea. Emu 98:13-21. Steadman, D. W., and S. L. Olson. 1985. Bird remains from an archaeological site on Henderson Island, South Pacific: man-cau sed extinctions on an "uninhabited" island. Proceedings of the Nationa l Academy of Sciences, USA 82:6191-6195. Steadman, D. W., and D. S. Pahlavan. 1992. Extinction and biogeography of birds on Huahine, Society Islands, French Polynesia. Geoarchaeology 7:449-483. Steadman, D. W., and B. Rolett. 1996. A chronostratigraphic an alysis of landbird extinction on Tahuata, Marquesas Islands . Journal of Archaeological Science 23:81-94. Steadman, D. W., P. J. White and J. Allen. 1999. Prehistoric bird s from New Ireland, Papua New Guinea: extinction on a large Melanesian island. Pr oceedings of the Natural Academy of Sciences, USA. Steadman, D. W., G. K. Pregill, and D. V. Burley. 2002a. Ra pid prehistoric extinction of iguanas and birds in Polynesia. Proceedi ngs of the National A cademy of Sciences USA. 99:3673-3677. Steadman, D. W., A. Plourde, and D. V. Burley. 2002b. Prehistoric butchery and consumption of birds in the Kingdom of Tonga, South Pacific. Journal of Archaeological Science 29:571-584. Steadman, D. W., T. H. Worthy, A. J. Ande rson, and R. Walter. 2000. New species and records of birds from prehistoric sites on Niue, Southwest Pacific. Wilson Bulletin 112:165-186. Suggs, R. 1961. Archaeology of Nuku Hiva, Marquesas Islands, French Polynesia. Anthropological Papers of the American Museum of Natural History, No. 49. Swofford, D. L., 2003. PAUP*. Phylogenetic Analysis Using Parsimony (*and Other Methods). Version 4. Sinauer Associates, Sunderland, MA.

PAGE 165

151 Taylor, P. B. 1998. Rails: a guide to the rails , crakes, gallinules and coots of the world. Yale University Press, New Haven, CT. Trewick, S. A. 1997. Flightlessness a nd phylogeny amongst endemic rails (Aves: Rallidae) of the New Zealand region. Philo sophical Transactions of the Royal Society of London, Series B 352:429-446. Warren, B., E. Bermingham, R. C. K. Bowie, R. P. Prys-Jones, and C. Thébaud. 2003. Molecular phylogeography reve als island colonization hist ory and diversification of western Indian Ocean sunbirds ( Nectarinia : Nectariniidae). Molecular Phylogenetics and Evolution 29:67-85. White, C. M. N., and M. D. Bruce. 1986. The Birds of Wallacea (Sulawesi, The Moluccas & Lesser Sunda Islands, Indone sia). British Ornithologists’ Union, London. Wickler, S., and M. Spriggs. 1988. Pleist ocene human occupation of the Solomon Islands, Melanesia. Antiquity 62:703-706. Worthy, T. H. 2004. The fossil rails (Aves: Rall idae) of Fiji with descriptions of a new genus and species. Journal of the R oyal Society of New Zealand 34:295-314. Worthy, T. H., and R. N. Holdaway. 2002. The lost world of the moa. Indiana University Press, Bloomington, IN. Wragg, G. M. 1995. The fossil birds of He nderson Island, Pitcairn Group: natural turnover and human impact, a synopsis. Bi ological Journal of the Linnean Society 56:405-414.

PAGE 166

152 BIOGRAPHICAL SKETCH Jeremy J. Kirchman grew up in Bourbonna is, Illinois, and graduated from Bishop McNamara High School in nearby Kanakakee. He attended Illinois Wesleyan University in Bloomington, IL, where he became inte rested in evolutionary biology and in ornithology under the guidance of R. Given Harp er, John Faaborg, and Thomas Griffiths. Jeremy received a B.A. (with research honors) in biology in 1994. He received his M.S. in zoology from Louisiana State Universi ty, Baton Rouge, in 1997 after completing a thesis on population genetics of Cave Swa llows under the guidance of Frederick H. Sheldon. From 1998 until 2001, he lived in Chica go, IL, working at the Field Museum of Natural History and teaching biology at St . GregoryÂ’s High School. Since 2001 he has studied the evolution of flight lessness in island birds at the University of Florida, under the guidance of David W. St eadman, and graduated in Augus t 2006 receiving the degree of Ph.D. Jeremy plans to continue his resear ch on bird evolution as Curator of Birds at the New York State Museum, Albany, NY.