Phylogenetic and ecological aspects of cooperative breeding in the bee-eaters (Aves: Meropidae)

Item Type text; Dissertation-Reproduction (electronic)

Authors Burt, Donald Brent, 1965-

Publisher The University of Arizona.

Rights Copyright © is held by the author. Digital access to this material is made possible by the University Libraries, University of Arizona. Further transmission, reproduction or presentation (such as public display or performance) of protected items is prohibited except with permission of the author.

Download date 27/09/2021 02:11:00

Link to Item http://hdl.handle.net/10150/282167 INFORMATION TO USERS

This manuscript has been reproduced from the microfilm master. UMI films the text directly from the original or copy submitted. Thus, some thesis and dissertation copies are in typewriter face, while others may be from any type of computer printer.

The quali^ of this reproduction is dependent upon the quality of the copy submitted. Broken or indistinct print, colored or poor quality illustrations and photographs, print bleedthrough, substandard margins, and improper alignment can adversely affect reproduction.

In the unlikely event that the author did not send UMI a complete manuscript and there are missing pages, these will be noted. Also, if unauthorized copyright material had to be removed, a note will indicate the deletion.

Oversize materials (e.g., maps, drawings, charts) are reproduced by sectioning the original, beginning at the upper left-hand comer and continuing from left to right in equal sections with small overlaps. Each original is also photographed in one exposure and is included in reduced form at the back of the book.

Photographs included in the original manuscript have been reproduced xerographically in this copy. Higher quality 6" x 9" black and white photographic prints are available for any photographs or illustrations appearing in this copy for an additional charge. Contact UMI directly to order. UMI A Bell & Howell Information Company 300 North Zedj Road, Ann Aibor MI 48106-1346 USA 313^61-4700 800/521-0600

PHYLOGENETTC AND ECOLOGICAL ASPECTS OF COOPERATIVE BREEDING IN THE BEE-EATERS (AVES: MEROPIDAE)

by Donald Brent Bxirt

A Dissertation Submitted to the Faculty of the DEPARTMENT OF ECOLOGY AND EVOLUTIONARY BIOLOGY In Partial Fvdfillment of the Requirements For the Degree of DOCTOR OF PHILOSOPHY In the Graduate College THE UNIVERSITY OF ARIZONA

1996 DMI Nxjinber; 9713401

UMI Microform 9713401 Copyright 1997, by UMI Company. All rights reserved.

This microform edition is protected against unauthorized copying under Title 17, United States Code.

UMI 300 North Zeeb Road Ann Arbor, MI 48103 2

THE UNIVERSITY OF ARIZONA GRADUATE COLLEGE

As members of the Final Examination Committee, we certify that we have

read the dissertation prepared by tVinalH Rr-^nt-. Rnr-h

entitled Phyloqenetic and Ecological Aspects; of O^ioperative

Breeding in the Bee-eaters (Aves: Meropida^K

and recommend that it be accepted as fulfilling the dissertation requirement for the Degree of of Philosophy

Date ' ~

i!.. 7/zz/9e Date

Date

Final approval and acceptance of this dissertation is contingent upon the candidate's submission of the final copy of the dissertation to the Graduate College.

I hereby certify that I have read this dissertation prepared under my direction and recommend that it be accepted as fulfilling the dissertation requirement.

DissertaG^n Director Date ' Wayne I-Iaddison 3

STATEMENT BY AUTHOR

This dissertation has been submitted in partial fulfillment of requirements for an advanced degree at The University of Arizona and is deposited in the University Library to be made available to borrowers under rules of the Library.

Brief quotations from this dissertation are allowable without special permission, provided that accurate acknowledgment of source is made. Requests for permission for extended quotation from or reproduction of this manuscript in whole or in part may be granted by the head of the major department or the Dean of the Graduate College when in his or her judgment the proposed use of the material is in the interests of scholarship. In all other instances, however, permission must be obtained from the author.

SIGNED: / 4

ACKNOWLEDGMENTS

I thank my committee members: Judie Broiistein, David Maddison, Nancy Moran, Dan Papaj, and the chair, Wayne Maddison, for their advise and close attention to my dissertation research. I received considerable advise on bee-eater biology, behavioral phylogenetics, and avian systematica from people outside the University of Arizona including Alan de Queiroz, Stephen Emlen, Hilary Fry, Brad Livezey and Doug Siegel-Causey. Groups deserving special mention included the Phylogeny Discussion Group and the entire crew in Wayne Maddison's lab, especially Greta Binford, Marshal Hedin, Susan Masta and Shelley McMahan. Patrick O'Grady and Joana Silva were also always encouraging and supportive. Very special thanks go out to Wade Leitner for getting me out into the field so that I did not forget what birds look like without cotton eyes. Crucial therapy was provided by the departmental basketball and ultimate Frisbee groups and my many dancing partners, particularly Sarah Mesnick. Consistent and considerable financial support was provided by the Department of Ecology and Evolutionary Biology, the Silliman Memorial fund, and particularly the Research Training Group in the Analysis of Biological Diversification (Fellowships and research grants) and the National Science Foimdation, (Dissertation Improvement Award). Field Work in Kenya was made possible with the assistance of Terry Repp and Kim Score from the University of Arizona, and Titus Imboma, Peter Njoroge, Edward Wayaki and the remaining staff of National Museums of Kenya. In Thailand, Kim Burt worked very hard in the field. I was also assisted by Dr. Schwann of the Royal Forest Department, and Dr. Sawat, Mr. Som Yot, and Mr. Boriphan, of the National Watershed Management Division. Dr. Jamroon and Ms. Piyathip of the National Marine Parks Division were also very accommodating. The following Museums were gracious in either loaning specimens or allowing me access to their collections: the University of Kansas Museum of Natural History, the National Museum of Natural History, and the British Museimi of Natural History. Finally, this degree and the work behind it would not have been possible without the ever present support of my family. Kim Burt provided a great deal of support in so many ways for which I will always be appreciative. DEDICATION

This work is dedicated to my anchors in life, the people who are always ready with their love and support; my family. 6

TABLE OF CONTENTS LIST OF HGURES 8 LIST OF TABLES 10 ABSTRACT 11 CHAPTER ONE- INTRODUCTION TO THE EVOLUTION AND MAINTENANCE OF COOPERATIVE BREEDING IN BIRDS 13 FOUR BY FOUR; LEVELS OF ANALYSIS AND MAJOR QUESTIONS 15 WHY DO HELPERS DELAY DISPERSAL? 26 Selective advantages 26 Proximate mec±ianisms 26 OntQggny 27 WHY DON'T HELPERS BREED INDEPENDENTLY? 28 Selective advantages, helper viewpoint 28 Potential conflict of interest with breeders 30 Proximate mechanisms 31 Ontogeny 31 WHY DO INDIVIDUALS HELP? 33 Selective advantages 33 Proximate mechanisms 34 Ontogeny 35 DO HELPERS REALLY HELP? 36 HISTORICAL PATTERNS 37 Phylogenetic constraints 38 SUMMARY 38 CHAPTER TWO- PHYLOGENETIC RELATIONSHIPS OF THE BEE-EATERS (AVES: MEROPIDAE) 40 METHODS 43 Plumage characters 43 Phvlogenetic analyses 46 RESULTS 51 DISCUSSION 59 Comparison of primary and secondary analyses 59 Biogeographic patterns 60 Comparison with earlier systematic study 64 Conclusions 69 CHAPTER THREE- BREEDING BIOLOGY AND SOCIALITY OF BEE-EATERS IN THAILAND 70 STUDY SITES AND METHODS 71 RESULTS 73 Little green bee-eaters. 73 Colony structure 73 Breeding chronology 73 Predators and anti-predatory behavior 73 7

TABLE OF CONTENTS - Continued Evidence of cooperative breeding. 74 Blue-tailed bee-eaters. 75 Colony structure 75 Breeding chronology 75 Evidence of cooperative breeding and sociality 75 Bay-headed and blue-bearded bee-eaters. 76 DISCUSSION 78 Little green bee-eaters in India and Thailand 78 Social systems of poorly studied spedes 79 Significance of these data in comparative studies 80 CHAPTER FOUR- BEHAVIORAL MALLEABILITY VERSUS PHYLOGENETIC INERTLA. AND COOPERATIVE BREEDING IN BEE- EATERS 83 METHODS 89 RESULTS 93 Patterns of evolution in breeding systems 93 Correspondence of other traits to breeding system evolution 99 Social systems 99 Nesting substrate requirements 102 Habitat utilization 104 Wing shape, migration, and foraging mode 106 Diet specialization 113 DISCUSSION 115 Evolution of breeding systems relative to other traits 115 Behavioral malleability versus phylogenetic inertia 119 Need for detailed ecological data on all species 122 Summary 124 CHAPTER FIVE- PHYLOGENETIC STRUCTURE INDICES: WHAT DO THEY TELL US AND HOW SHOULD THEY BE USED? 126 Expectations of a preferred index 128 METHODS 133 RESULTS 134 Matrices failure rate 134 Behavior of CI. RI. and RC 136 Behavior of gi 138 Behavior of average nonparametric bootstrap values 140 DISCUSSION 142 CONCLUSIONS 146 APPENDIX A 148 APPENDIX B 157 APPENDIX C 172 WORKS CITED 173 8

LIST OFnCURES FIGURE 2.1, Consensus tree from primary analyses of races 52 FIGURE 2.2, Four trees from primary analyses of species 53 FIGURE 2.3, Trees from primary an^yses with standardized character weights 54 FIGU^ 2.4, Trees from secondary analyses where character weights equaled one 56 FIGURE 2.5, Trees from successive approximation character weighting analyses 57 FIGURE 2.6, Trees from secondary analyses with standardized character weights 58 FIGURE 2.7, Biogeographic patterns reconstructed on primary analysis frees where characters weights equaled one 62 FIGURE 2.8, Biogeographic patterns reconstructed on primary analysis trees where characters weights were standardized 63 FIGURE 2.9, Fry's diagram of a cross section of the top of a phylogenetic tree 67 FIGURE 2.10, Estimated free from diagram given by Fry (1969) 68 FIGURE 4.1, Cautious coding reconstructions of breeding system evolution 95 FIGURE 4.2, Moderate coding reconstruction of breeding system evolution 97 FIGURE 4.3, Liberal coding reconstructions of breeding system evolution 98 FIGURE 4.4, Reconstructioris of evolutionary patterns of social systems 101 FIGURE 4.5, Reconstructions of evolutionary patterns of nesting substrate requirements 103 FIGURE 4.6, Reconstructions of evolutionary patterns of habitat utilization 105 FIGURE 4.7, Reconstructioiis of wing shape evolution 109 FIGURE 4.8, Reconstructions of foraging mode evolution 110 FIGURE 4.9, Reconstructior\s of migratory evolution Ill FIGURE 4.10, Reconstructions of evolutionary patterns of diet specialization 114 FIGURE 5.1, Preferred characteristics of an index of phylogenetic confidence 130 FIGURE 5.2, Distribution of type n error rates 135 FIGURE 5.3, Behavior of CI, RI, and RC indices to the addition of random characters to core symmetrical free matrices 137 FIGURE 5.4, Behavior of the gi tree length skewness index to the addition of random characters to core symmetrical tree matrices 139 FIGURE 5.5, Behavior of the average nonparamefric bootstrap index to the 9

LIST OF HGURES - Continued

addition of random characters to core symmetrical tree matrices 141 FIGURE 5.6, General characteristics of phylogenetic confidence indices seen in this study 143 10

LIST OF TABLES TABLE 1.1, Bird families with cooperative breeding species 16 TABLE 2.1, Bee-eater classifications 42 TABLE 2.2, Phylogenetic analysis options used in this study 49 11

ABSTRACT Cooperative breeding (CB) is found in a wide diversity of avian lineages and can be explained at: several levels of analysis. After a brief introduction to the theory explaining CB, I take an historical approach to examine CB evolution in the bee-eaters (Family Meropidae). Parsimony analyses of plumage color and shape characters yielded a number of phylogenetic hypotheses. The best supported phylogenies are six fully resolved trees from three analyses and a strict consensus tree from another analysis. These trees are used to examine the possible patterns of evolution in CB and how transitions correspond to transitions in other ecological and behavioral traits. Bee-eaters were also studied in Thailand. Little green bee-eaters, Merops orientalis, breed cooperatively and predation pressure may be high in this species. Blue-tailed bee-eaters, M. philippinus, breed cooperatively in dense colonies and show signs of potential extra-pair copulation and intraspecific brood parasitism. Observations of the bay-headed bee-eater, M. leschenaulti, and the blue-bearded bee-eater, Nyctyomis athertoni, docimient CB in the former and support non-CB designation for the latter. Cooperative breeding is either primitive in bee-eaters or evolved early in the family. Reversals to non-CB occurred in one to three lineages. Transitions in breeding systems are not generally correlated with transitions in nesting requirements, habitat utilization, migratory behavior, or diet. Evidence suggests correlated evolution between CB and both foraging mode (weak evidence) and social systems (stronger support). This study does not support any single hypothesis for the adaptive basis of CB across the family. Social system evolutionary patterns do suggest the importance of kin selection in several lineages. Lack of change in breeding systems, given diverse ecological and 12

ABSTRACT - Continued behavioral ciroomstances, means either cooperative breeding is malleable (selectively advantageous in a variety of ecological conditions) or represents phylogenetic inertia. A final analysis demonstrates that phylogenetic confidence indices fail to express the degree to which characters in a matrix are non- conflicting and congruent (for a given level of noise) and they have only limited abilities to distinguish among probabilities of analyses making type II errors. 13

CHAPTER ONE- INTRODUCTION TO THE EVOLUTION AND

MAINTENANCE OF COOPERATIVE BREEDING IN BIRDS Cooperative breeding (CB) is a reproductive systems in which certain individuals (helpers) regularly provide some degree of parental care to young that are not their own offspring. This chapter introduces the principal hypotheses concerning the evolution and maintenance of CB in birds and describes the distribution of the behavior in the Class Aves. Many of the concepts introduced here are subsequently examined in more detail with regard to the Family Meropidae in Chapters 3 and 4. The study of CB is one of the most active areas of avian social behavior research (Emlen 1984, Brown 1987, Stacey & Koenig 1990). A wide diversity of breeding systems and social systems can be found among the over 230 species of birds (see Table 1.1) that breed cooperatively and it is clear that CB has evolved independently in birds a number of times (Brown 1987; Edwards & Naeem 1993; Ligon 1993; Burt, impublished data). Cooperative breeding can be difficult to detect by casual observation and the number of CB species is sure to increase as additional behavioral studies are conducted on additional species. Therefore, it is not clear how to interpret families with no listed CB species. These families may represent either groups with truly no CB species or groups in need of further ecological study. It is clear, however, that the majority of bird species are non-CB. For example, approximately 30 of the 900 species of North American birds are CB (Brown 1987). Also, of 217 well studied bird species in South Africa, only 42 (19.4%) are CB (Du Plessis et al. 1995). Many CB species are foimd in the tropics (Brown 1987). However, less information is typically available for tropical species, and "non-cooperative families" with many tropical species may 14 very well turn up CB species (e.g., Eopsaltriidae, Megalaimidae). Certain families do, however, appear to have no CB species. Most groups composed of nectar-feeders do not appear to have CB species (i.e., Nectariiiiidae, Trochilidae; but see Meliphagidae) nor do pelagic seabird groups (e.g., Fregatidae, Procellariidae) or groups with precocial yoxmg (e.g., Megapodiidae, Scolopacidae) (Brown 1987). Variation in breeding systems across CB species includes monogamous, polyandrous, polygynous, and polygynandrous mating systems. Social systems range from solitary-breeding territorial species to those nesting in dense colonies. Social systems can also include singular breeding (one breeding female per group) or plural breeding (two or more breeding females per group, separate or joint nests) systems. Helpers may be male only or include both sexes and may be related or imrelated to yoimg. 15

FOUR BY FOUR: LEVELS OF ANALYSIS AND MAJOR QUESTIONS Researchers have taken a variety of approaches in attempting to explain the presence of this seemingly altruistic behavior. These different approaches have lead to different, and in many cases complementary, explanations for the existence of CB. These different approaches, or levels of analysis (Tinbergen 1963, Sherman 1988), can be placed in four categories: first, functional consequences or selective advantages; second, proximate mechanisms or physiological requirements; third, ontogeny; and fourth, historical origin and transition patterns. Additionally, examination of factors responsible for CB behavior requires addressing four questions (Emlen 1984,1991, Brown 1987). First, why do helpers delay dispersal? Second, why don't helpers breed, despite this delayed dispersal? Third, why do individuals help, given delayed dispersal? And finally, do helpers really help? The first three questions look at the behavior from the helper's point of view, while the last examines the possibility of benefits to the breeders due to the helper's presence. 1 will examine each of these four major questions in the light of the four levels of analysis, where appropriate, in the remainder of this chapter. 16

Struthionidae Ostrich Struthio cameliis Rheldae Casuariidae Apterygidae Tinamidae Cracidae Megapodiidae Odontophoridae Numididae Phasianidae Anhimidae Anseranatidae Magpie Goose Aneranas semipalmata Dendrocygnidae Anatidae Turnicidae Ramphastidae Collared Aracari Pteroglossus torquatus Lybiidae White-eared Barbet Buccanodon leucotis Pied Barbet Lybhis albicaiida Black-collared Barbet Lybiiis torquatus D'Amaud's Barbet Trachyphonus damaudi Red-aiid-yellow Barbet Trachyphonus erythrocephalus Megalaimidae Indicatoridae Picidae Red-cockaded Woodpecker Dendrocopus borealis Red-fronted Woodpecker Melanerpes cruentatus Acorn Woodpecker Melanerpes formicivorus Bucconidae White-fronted Nunbird Monasa morphoeus Galbulidae

TABLE l.l. Classification of bird families (Sibley and Ahlquist 1990) with list of known CB species in each family. The majority of cooperative breeding species data are from Brown 1987. Data on additonal species for the Family Meropidae are from: Bryant and Tatner 1990; Chapter 3; Douthwaite 1986; Dyer and Crick 1983; Emlen 1990, pers. com.; Filewood et al. 1978; Fry 1972; 1984; Gartshore 1984; Sridhar and Karanth 1993. 17

Upupidae AMcan Hoopoellpupa epops Phoeniculldae Green Woodhoopoe Phoeniculus purpureus Rhinopomastidae Bucorvidae Bucerotidae Bushy-crested Hombill Anorrhiniis galeritus White-crested Hombill Berenicomis comatus Ground Hombill Bucorvus leadbeateri Casqued Hombill Bycanistes subcylindricus Trogonidae Leptosomidae Coraciidae Brachypteraciidae Meropidae White-throated Bee-eater Merops albicollis European Bee-eater Merops apiaster Red-throated Bee-eater Merops bullocki White-fronted Bee-eater Merops bullockoides Carmine Bee-eater Merops nubicoides Rainbow Bee-eater Merops omatus Blue-throated Bee-eater Merops viridis Swallow-tailed Bee-eater Merops hirundineus Little green Bee-eater Merops orientalis Bay-headed Bee-eater Merops leschenaulti Blue-tailed Bee-eater Merops philippinus Little Bee-eater Merops piisillus Cinnamon-chested Bee-eater Merops oreobates Momotidae Todidae Puerto Rican Tody Todiis mexicanus Alcedinidae Cerylidae Pied Kingfisher Ceryle nidis Dacelonidae Blue-winged Kookaburra Dacelo leachii Laughing Kookaburra Dacelo novaegiiineae (gigas) Striped Kingfisher Halcyon chelicuti

TABLE 1.1. (Continued) 18

Forest Kingfisher Halcyon macleayii Buff-breasted Paradise-kingfisher Tanysiptera sylvia Coliidae Speckled Mousebird Colius striatus Crotophagidae Smooth-billed Ani Crotophaga ani Greater Ani Crotophaga major Groove-billed Ani Crotophaga sulcirostris Guira Guira guira Neomorphidae Opisthocomidae Hoatzin Opisthocomus hoazin Coccyzidae Centropodidae Cuculidae Psittacidae Apodidae Ashy-tailed Swift Chaetura andrei Short-tailed Swift Chaetura brachyura Chimney Swift Chaetura pelagica Vaux's Swift Chaetura vauxi Hemiprocnidae Trochilidae Musophagidae Strigidae Tytonidae Aegothelidae Podargidae Batrachostomidae Nyctibiidae Steatornlthidae Eurostopodidae Caprimulgidae Columbidae Rallidae Red-knobbed Coot Fulica cristata Moorhen Gallinula chloropus Purple Gallinule Gallinula martinica Dusky Moorhen Gallinula tenebrosa

TABLE 1.1. (Continued) Black Crake Limnocorax flavirostris Pukeko Porphyria porphyria Native Hen Galinula mortierii Mesitornithidae Eurypygidae Otididae Cariamidae Rhynochetidae Psophiidae Heliornithidae Gruidae Pteroclidae Jacanidae Rostratulldae Thinocoridae Pedionomldae Scolopacidae Charadriidae Southern Lapwing VaneUus chilensis Burhinidae Chionididae Glareoiidae Laridae Lonnberg Skua Catharacta mackormacki lonnbergi Arctic Tern Sterna paradisea Falconidae Saglttarlidae Acclpitrldae Gdapagos Hawk Buteo galapagoensis Peregrine Falcon Falco peregrinus Mississippi Kite Ictinia mississippiensis Harris' Hawk Parabuteo unicinctus Bateleur Terathopius ecaudatus Podicipedidae Homed Grebe Podiceps auritus Austalasian Grebe Tachybaptus novaehollandiae Phaethontidae

TABLE 1.1. (Continued) Sulidae Anhingidae Phalacrocoracidae Ardeidae Scopidae Phoenlcopteridae Threskiomithidae Ciconiidae Pelecanidae Fregatidae Spheniscidae Gaviidae Procellariidae Acanthisittidae Yellow-rumped Thombill Acanthiza chrysorrhoa Striated Thombill Acanthiza lineata YellowThombill Acanthiza nana Buff-rumped Thombill Acanthiza reguloides Chestnut-mmped Thombill Acanthiza uropygialis Yellow-bellied Gerygone Gerygone chrysogaster Brown Gerygone Gerygone mould White-browed Sarub-wren Sericomis frontalis Large-billed Sarub-wren Sericomis magnirostris Eurylaimidae Philepittidae Pittidae Furnariidae Rufous-fronted Thombird Phacellodomus rufifrons Formicariidae Conopophagidae Rhinocryptidae Thamnophilidae Tyrannidae Purple-throated Fruit-crow Querula purpurata White-bearded Flycatcher Conopias inomata Rufous-margined Flycatcher Myiozetetes cayanensis Climacteridae Menuridae

TABLE 1.1. (Continued) Ptilonorhynchidae Maluridae Superb Blue Wren Malunis cyaneus Red-winged ¥airy-wrenMalurus elegans Variegated Fairy-wren Malurus lamherti White-winged Fairy-wren Malurus leucopterus Blue-breasted Fairy-wren Malurus pulcherrimus Splendid Fairy-wren Malurus splendens Weebill Smicromis brevirostris Meliphagidae Little Wattlebird Anthochaera chrysoptera Rufous-throated Honeyeater Conopophilia rufogularis Yellow-tufted Honeyeater Meliphaga melanops White-plumed Honeyeater Meliphaga penicillatus White-throated Honeyeater Melithreptus albogularis Brown-headed Honeyeater Melithreptus brevirostris White-naped Honeyeater Melithreptus lunatus New Holland Honeyeater Phylidonyris novaehoUandiae Striped Honeyeater Plectorhyncha lanceolata Yellow-throated Miner Manorina flavigula Noisy Miner Manorina melanocephala Bell Miner Manorina melanophrys Little Friarbird Philemon citreogularis Pardalotldae Striated Pardalote Pardalotus striatus Eopsaltriidae Irenidae Orthonychidae Pomatostomidae Laniidae Yellow-billed Shrike Corvinella corvina Magpie Shrike Corvinella melanoleuca White-crowned Shrike Eurocephalus anguitimens Gray-backed Fiscal Shrike Laniiis exicubitorius Straight-crested Helmet-shrike Prionops plumata Retz's Redbilled Shrike Prionops retzii Chestnut-fronted Shrike Prionops scopifrons Vireonidae

TABLE 1.1. (Continued) Corvidae White-spotted Wattle-eye Platysteira tonsa Qiestnut-cap Flycatcher Erythrocerus mccalli Blue Rycatcher Trochocerciis lonicauda Crested Shrike-tit Falcuncubis frontatus Cinnamon Quail-thrush Cinclosoma cinnamomeum Logrurmer Orthonyx temminckii Chabert Vanga Leptopterus chabert Black Drongo Dicninis adsimilis Black-faced Woodswallow Artamus cinereus Dusky Woodswallow Artamus cyanopterus White-breasted Wood-swallow Artamus leucorhynchus Little Wood-swallow Artamus minor Ground Cuckoo-shrike Coracina maxima Florida Scrub-jay Aphelocoma coerulescens Mexican Jay Aphelocoma ultramarina White-throated magpie-jay Calocitta formosa Black-throated magpie-jay Calositta formosa colliei American Crow Corvus brachyrhynchos Northwestern Crow Corvus caurinus Beechey Jay Cyanocorax beecheii Tufted Jay Cyanocorax dickeyi Bushy-crested Jay Cyanocorax melanocyanea San Bias Jay Cyanocorax sanblasiana Green Jay Cyanocorax y. yncas Yucatan Jay Cyanocorax yucatanica Azvire-winged Magpie Cyanopica cyana Pinyon Jay Gymnorhinus cyanocephaliis Brown Jay Psilorhinus morio Piapiac Ptilostomus afer Stresemarm's Bush Crow Zavattariomis stresemanni White-winged Chough Corcorax melanorhamphus Pied Butcherbird Cracticus nigrogularis Grey Butcherbird Cracticus torquatus Australian Magpie Gymnorhina tibicen Apostlebird Struthidea cinerea Callaeatidae Picathartidae Bombycillidae Cinclidae

TABLE 1.1. (Continued) 23

Muscicapidae Pale Flycather Bradomis pallidus Eastern Yellow Robin Eopsaltria australis Hooded Robin Petroica cucullata White-breasted Robin Erithacus georgiana Western Yellow Robin Erithacus griseogularis European Robin Erithacus rubecula Abyssinian Black Wheatear Oenanthe lugens Rufous Roclqumper Chaetops frenatus Forest Flycatcher Fraseria ocreata Abyssinian Slaty Flycatcher Melaenomis chocolatinus Anteater Qiat Myrmecocichla aethiops Sturnidae Black-capped Donacobius Donacobius atricapillus Galapagos Mockingbird Nesomimus macdonaldi Galapagos Mockingbird Nesomimus parvulus Galapagos Mockingbird Nesomimus trifasciatus Yellow-billed Oxpecker Buphagus africanus Sturnidae Red-billed Oxpecker Buphagus erythrorhynchus King Glossy Starling Cosmopsarus regius Cape Glossy Starling Lamprotomis nitens Pied Starling Spreo bicolor Fischer's Starling Spreo fischeri Chestnut-bellied Starling Spreo pulcher Golden-breasted Starling Spreo regius Superb Starling Spreo superbus Sittidae Varied SiteUaDaphoenositta chrysoptera Brown-headed Nuthatch Sitta pusilla Pygmy Nuthatch Sitta pygmaea Certhiidae Fasciated Wren Campylorhynchus fasciatus Bicolored Wren Campylorhynchus griseus Spotted Wren Campylorhynchus jocosus Gray-barred Wren Campylorhynchus megalopterus Stripe-backed Wren Campylorhynchus nuchalis Band-backed Wren Campylorhynchus zonatus Banded Wren Thryothorus pleurostictus Red-browed Treecreeper Climacteris erythrops

TABLE 1.1. (Continued) Black-tailed Treecreeper Climacteris melanura Brown Treecreeper Climacteris picumnus Rufous Treecreeper Climacteris rufa Parldae Long-tailed Tit Aegithalos caiidatiis Tufted Titmouse Panis bicolor Black Tit Parus niger Bushtit Psaltriparus minimus Aegithalidae Himndinidae Bam Swallow Hirundo nistica Regulidae Pycnonotidae Spotted Greenbul Ixonotus guttatus V^te-tailed Greenbul Thescelocichla leucopleura Cisticolidae Zosteropidae Seychelles White eye Zosterops modesta Sylviidae Seychelles Bush Warbler Acrocephalus sechellensis Tit HyhsLPholidomis rushiae Short-tailed Bush Warbler Cettia squameiceps Green-backed Eremomela Eremomela pusilla Diisky-faced Warbler Eremomela scotops Yellow-eyed Babbler Chnjsomma sinensis Hall's Babbler Pomatostomus halli Rufous Babbler Pomatostomus isidori White-browed Babbler Pomatostomus superciliosus Grey-crowned Babbler Pomatostomus temporalis White-headed Babbler Turdoides affinis Common Babbler Turdoides caudatus Striated BahhlerTurdoides earlei Arrowmarked Babbler Turdoides jardinei Large Grey Babbler Turdoides malcolmi Bla^-lored Babbler Turdoides melanops Brown Babbler Turdoides plebejus Blackcap Babbler Turdoides reinwardii Arabian Babbler Turdoides squamiceps Jungle Babbler Turdoides striatus Formosan Yuhina Yiihina brunneiceps

TABLE l.l. (Continued) Alaudidae Nectariniidae Meianocharitidae Paramythiidae Passeridae Alpine Accentor Prunella collaris Dimnock Prunnela modularis Cape Wagtail Motacilla capensis Hovise Sparrow Passer domesticiis White-browed Sparrow-weaver Plocepasser mahali Sociable Weaver Philetairiis sociiis Grey-capped Social Weaver Pseudonigrita amaudi Fringillidae Speckled Tanager Tangara chrysophrys Plain-colored Tanager Tangara inomata Golden-masked Tanager Tangara larvata Mexican Tanager Tangara mexicana Stripe-headed Sparrow Aimophila ruficauda Northern Cardinal Cardinalis cardinalis Bobolink Dolichonyx oryzivonis Gray-headed Bvinting Emberiza fiicata Medium Ground-finch Geospiza fortis Cactus Ground-finch Geospiza scandens Bay-winged Cowbird Molothrus badiiis Austral Blackbird Curaeus curaeus Bolivian Blackbird Oreospar bolivianus Brown-and-yellow Marshbird Pseudoleistes virescens

TABLE 1.1. (Continued) 26

WHY DO HELPERS DELAY DISPERSAL? Selective advantages The majority of hypotheses for why helpers remain on their natal territory focus on habitat usage and possible ecological constraints. Constraints can work through intrinsic and extrinsic means. The habitat saturation hypothesis states that suitable breeding habitat slots are filled and young individuals are forced to delay dispersal due to these extrinsic factors (Selander 1964, Brown 1974). The marginal habitat model builds on this concept by stating that an additional constraint is the lack of habitats of marginal quality. A lack of marginal habitats eliminates the possibility of individuals dispersing into the role of non-breeding floaters (Koenig and Pitelka 1981, Emlen 1982, Woolfenden & Fitzpatrick 1984). The benefits-of-philopatry (BOP) model stresses the importance of intrapopulational variation in territory quality (Stacey and Ligon 1987,1991). According to the BOP, individuals bom in high quality natal territories have intrinsic reasons for choosing delayed dispersal and remain at home until they can iriherit the natal territory or occupy another nearby territory of equal quality. These individuals may also avoid increased chances of mortality associated with lower quality habitats by remaining at home. Additional explanations for delayed dispersal confound questions of dispersal and breeding. For example, an excess of males can be seen in several CB populations and this sex ratio bias has been suggested as a possible cause for both delayed dispersal and breeding (Emlen et al. 1986). This bias would not, however, explain why individuals do not disperse to become floaters. Proximate mechanisms 27

There are also endocrine-based reasons why individuals may delay dispersal. If a male is not able to maintain a siifficient testosterone and aggression level it may not be possible to establish and defend a territory. Given this proximate-mechanistic constraint, it may be selectively advantageous (i.e., different level of analysis) to remain in the natal territory if this option is allowed by the resident breeders. This is especially likely if territory residence is closely tied to survival probability or if dominance hierarchies more severely restrict access to food resources for individuals outside the natal group than those within the natal group. Additionally, if females are unable to demonstrate ability to reproduce, they may be rejected by territorial males and forced to remain on their natal territory for the same survival probability reasons listed above. Ontogeny The skill hypothesis suggests that individuals must first leam the necessary skills to be able to disperse and attempt independent breeding (Brown 1987), especially in harsh, variable environments (Emlen 1982). In many cases, cooperative species show delayed morphological development and it is possible that attainment of necessary skills shows parallel development. 28

WHY DON'T HELPERS BREED INDEPENDENTLY? Selective advantages, helper viewpoint Given that individuals delay dispersal, why do they not simply attempt independent breeding on their natal territory? Actually, in some cases individuals do breed, albeit sometimes surreptitiously, while serving simultaneously as helpers. In plural CB groups a helper can also be a paired breeder. However, in many cases the only option available to helpers is extra- pair copulation (EPC). The most important consideration in such cases is the relatedness of the breeders in the helper's social group or the proximity and access he/she may have to neighboring social groups. This premise assumes individuals avoid inbreeding, which appears typical for most CB species (Brown 1987, but see Rowley and Russell 1990 for coimterexample). Below, I further examine the potential for male and female helpers to gain direct fitness benefits via EPC and how this potential varies with different social systems. Male helpers in singular CB groups are usually sons of the breeding female unless the breeding male (usually their father) has re-mated. Pair bonds in many cooperative breeders are for life and breeder mortality is tj^ically low (Brown 1987). Therefore, in this case the male helper has limited chances for EPC within the social group. However, if the species is colonial he may have greater access to neighboring breeding females. Singular CB can include monogamous and polyandrous mating systems, which provide few EPC opportimities for male helpers. However, in plural CB groups the male will have at least one breeding female in his social group who is more distantiy related than his mother and the opportunities for EPC are greater. Plural CB can include monogamous, polygynous and polygynandrous mating systems. Again, opportimities for EPC 29

are potentially even better if the species is colonial. In summary, the male helper's opportxmities for EPC probably increase through the following social systems in this order: singular territorial, plural territorial, singular colonial and plural colonial species. Evidence of EPC by male helpers has been recorded in

several species (pukeko, Porphyrio porphyrio [Craig and Jamieson 1990];

dunnocks. Prunella modularis [Davies 1990]; stripe-backed wren, Campylorhynchus

nuchalis [Rabenold 1990]; splendid fairy-wren, Malurus splendens [Rowley and

Russell 1990]; Arabian babbler, Tiirdoides squamiceps [Zahavi 1990]) and is suspected in others (acom woodpecker, Melenerpes formicivorus, [Koerug and

Stacey 1990]; red-cockaded woodpecker, Picoides borealis [Walters 1990]). Females will generally have fewer options for benefiting from EPC while serving as helpers. In addition to the genetic constraints mentioned above for the males, females must have means of depositing their eggs. Female helpers that receive EPC from the breeding males only have the options of sneaking eggs into the nest(s) of the breeding female(s) in their group or into those of breeders outside their social group. In many CB species, helpers are not permitted near the nest imtil after incubation has started and sneaking eggs into the nest would be difficult. In singular breeders there is only one nest and one female breeder. Sneaking eggs into nests under these circumstances should be readily detected by researchers and possibly, but not necessarily, by the breeding female, vmless helper females also remove eggs of the breeder. In plural breeders, breeding females may nest separately or jointly. It is within the confusion of joint nesting species that female helpers may be most prone to soliciting EPC and deriving

direct fitness gains (e.g. acom woodpecker, Melenerpes formicivorus, [Koenig and

Stacey 1990]; Arabian babbler, Turdoides squamiceps [Zahavi 1990]). 30

Genetic conflicts between different helpers could arise relative to EPC attempts if helpers are not siblings. A helper gaining EPC with the parent of another helper is diluting the genetic relatedness of the clutch relative to what it would be if both of the latter helper's parents contributed all the genes. The conditional opportuiuties described above for individuals to obtain EPC while helping may be an additional variable that is considered when deciding whether to delay both dispersal and independent breeding. By this route, the benefits of EPC may then set the stage for selection on alloparental behavior (see 'Why do individuals help' section below). Alternatively, helpers may delay dispersal and breeding for other reasons and choose to participate in EPC as a means of 'making the best of a bad job'. Potential conflict of interest with breeders Can breeders gain through the breeding of helpers? The perspectives from breeders of each sex on the costs and benefits of breeding helpers can be very different. Males may derive benefits from EPC with female helpers at the expense of female breeders if these eggs end up parasitizing the female breeder's nest. Females may gain from EPC with male helpers at the expense of the male's certainty of paternity. Breeders must consider these factors when weighing the costs of potentially losing certainty of maternity/paternity versus loosing the alloparental care behavior helpers can provide. Both sexes may benefit from EPC, but under what conditior\s might the costs of EPC make breeders refuse the presence of helpers? Alternatively, when might this cost be accepted due to the importance of alloparental behavior for breeders to achieve any reproductive success? If all helpers are closely related to the breeders, the point is moot. If helpers are unrelated to breeders and the costs 31 for breeders of EPC are high, then helpers should be tolerated only in desperate conditior\s. Such conditions, however, could exist. For example the threat of predation or cost of food provisioning may be so high that breeders are unable to raise any youmg without the assistance of helpers (e.g. white-winged choughs,

Corcorax melanorhamphos [Heinsohn 1991]). Proximate mechanisms Delayed breeding may also be due to several physiological factors. First, an individual may be sexually immature and lack the ability to produce the appropriate levels of hormones to reproduce (Koenig & Pitelka 1981, Reyer et al. 1986). Even if helpers are otherwise physically mature, breeders may prevent them from becoming reproductively functional through aggression. Corticosterone is secreted from the adrenal glands in response to chronic stress and can result in reproductive iiihibition. Therefore, breeders may 'psychologically castrate' the helpers by way of constant aggressive attacks (Brown 1978, Reyer et al. 1986). Alternatively, individuals may be reproductively competent but imable to gain access to mates. For example breeding males may closely guard their mates against helper male attempts to copulate (Mumme et al. 1983, Emlen & Wrege 1986, Reyer et al. 1986). A final possibility is that helpers may 'choose' to keep hormone levels low so that they may remain in the social unit and still avoid such deleterious factors as inbreeding and increased social turbulence (Reyer et al. 1986, Wingfield 1990, Schoech et al. 1991). Ontogeny Delayed development of skills, in a similar manner as mentioned for reasons to delay dispersal, may prevent individuals from breeding. The skills typically referred to in this context are foraging and provisioning behavior, but 32 could include any parental behavior such as territory or nest defense (Brown 1987). 33

WHY DO INDIVIDUALS HELP? Selective advantages If individuals delay dispersal and apparently delay breeding, why then should they help? Several adaptive explanations for helping behavior exist (reviewed by Emlen 1991). Helping may increase survivorship imtil an individual can become a breeder. For example, by helping to raise more individuals, the group size increases and an individual may be safer from the threat of predation. Helping may also be a 'pajonent' to breeders for the privilege to remain on the high quality territory. Helping may increase the probability of future breeding. If helping results in an increase in group size, the group may then outcompete neighboring smaller groups for territorial space. At some point a helper individual may be able to "bud off a section of this habitat as a breeding territory (e.g., Florida scrub-jay,

Aphelocoma coerulescens [Woolfenden and Fitzpatrick 1984]). The dominance hierarchy of non-breeders and the possibility of inheritance of the original territory may also be tied to who provides helping behavior. Helpers that assist in raising younger individuals may form coalitions of these individuals that can then assist in fights for breeding vacancies (e.g., green woodhoopoes, Phoeniciilus purpureus [Ligon and Ligon 1990]). Helpers can sometimes increase chances of future mate acquisition by forming social bonds with the opposite sex breeder(s) in the group via helping behavior (e.g., pied kingfisher, Ceryle rudis [Reyer 1990]). Helping can also improve the success of future independent breeding. While helping, individuals may develop critical, specialized skills necessary for successful breeding. Helpers may also increase the chance of gaining the aid of the individuals whom they helped (reciprocal altruism, Trivers 1971) once they 34 become breeders (e.g., white-fronted bee-eaters, Merops bullockoides [Emlen 1990]; white-winged choughs, Corcorax melanorhamphos [Heinsohn 1991]). Finally, helping may increase the production of non-descendent kin. To receive an increase in the indirect component of fitness a helper must be closely related to the breeders and significantly increase the fitness of these breeders in comparison to what they could produce on their own or with fewer helpers. The best examples of kin selection benefits of helping behavior come from white- fronted bee-eaters, Merops bullockoides (Emlen 1990) and pied kingfishers Ceryle rudis (Reyer 1990). Proximate mechanisms What role could hormonal control play in alloparental care behavior? Jamieson (1986,1989) has argued that the act of providing food to begging nestlings, one component of helping behavior, is a non-adaptive consequence of parental behavior. The stimuli associated with begging young are so closely linked to selection for parental behavior that selection against individuals who allofeed would also result in decreased parental care behavior. Note that this hypothesis includes both cases where the helper is non-breeding or when the helper is also a breeder in the social unit. If the helper is also a breeder, restraining resource allocation only to one's own offspring might negatively impact parental care behavior in general. The hormones progesterone and prolactin have been identified as having a significant positive influence on parental care behavior (Silver 1978, Vleck et al. 1991). Helper individuals may have high levels of progesterone or prolactin and may therefore respond to begging yoimg and allofeed whether the behavior is adaptive or not. However, I suggest that a high level of progesterone, even 35 higher than that of breeders, may be selectively advantageous if it increases the individuals indirect component of fitness. Ontogeny Again, regarding Jamieson's (1986,1989) non-adaptive arguments, allofeeding by helpers can he seen simply as an early ontogenetic shift to a behavior that will be needed by these individuals when they become breeders. This feeding behavior is so closely linked to selection for parental behavior that selection against individuals who allofeed would also result in eventually decreased parental care behavior. 36

DO HELPERS REALLY HELP? What are the possible gains that accrue to the breeders by allowing individuals to help? Three major categories of helper benefit have been proposed concerning the adaptive basis of this behavior (Brown 1987, Emlen 1991). First, helpers may increase the production of offspring. Helpers may do so by providing increased, or more consistent, supplies of food to young. Helpers may also protect the nest from predators. Second, helpers may allow breeders to attempt more clutches per season. This extra clutch may be possible because, by feeding young and aiding in territory defense, helpers reduce the breeders' responsibilities to the previous clutch. Third, helpers may increase the life span and possibly the lifetime reproductive success of breeders. Ample evidence is found supporting each category for a variety of species (Brown 1987, Emlen 1991, Stacey & Koenig 1990). 37

HISTORICAL PATTERNS Historical approaches to the study of CB examine evolutionary patterns of transitions between cooperative and non-cooperative systems and how these changes correspond to other behavioral and ecological traits such as those discussed in the previous sections. Historical approaches have shown that CB evolved early in the diversification of many bird lineages, that it is not evenly represented across genera of passeriform birds, and that the ecological and social conditions which led to the evolution of the behavior in a common ancestor are quite different from those conditions in many of the descendant species (Peterson and Burt 1992, Edwards & Naeem 1993, Ligon 1993, Chapter 4). In many cases, when the behavior retains significant phylogenetic information, it is possible that phylogenetic inertia and phylogenetic constraints play important roles in patterns of change between social systems (Chapter 4). Phylogenetic constraints are defined as "any result or component of the phylogenetic history of a lineage that prevents an anticipated course of evolution in that lineage" (McKitrick 1993). The two key components in this defirution concern the expectation of certain evolutionary patterns occurring, and mechanisms restricting these evolutionary changes. Phylogenetic inertia, on the other hand, is invoked in circumstances where an anticipated evolutionary course is not seen, but no mechanism, other than time, is restricting evolution. In this case, not enough time has passed for selection to operate and produce expected results. Phylogenetic inertia has been suggested for the presence of CB in certain

Aphelocoma jays (Peterson and Burt 1992). Ecological constraint clearly play a role in maintaining CB in the Florida scrub-jay {A. coerulescens) but not in cooperative 38 populations of Western scrub-jays {A. califomica). It is possible that selection has simply not been important for reversing this behavior to the non-cooperative state. Similarly, even if CB is currentiy a selectively advantageous trait in a population, its present-day function may be different from the original adaptation (Gould and Lewontin 1979). Phylogenetic constraints There are some components of CB that may be functionally cot\strained from evolutionary transitions. One example might be that suggested by Jamieson (1986,1989) discussed above. If selection against allofeeding behavior by helpers is closely tied to the parental care behavior of these individuals when they do subsequently breed, then allofeeding may be constrained from changing. In this case, selection to prevent allofeeding is too costly to long term reproductive success. Unpublished data I am currently analyzing also suggest the potential for phylogenetic constraints. If transitioris are vmconsfrained, then no fransition bias shovild be seen among CB and non-CB states. I reconstructed CB patterns on the avian free of Sibley and Ahlquist (1990) and found that CB has evolved at least 13 times in the class with only 5 reversals to the non-CB state. While this pattern does not indicate a significant deviation from an unbiased fransition pattern, the frend is reverse that expected by some authors (Edwards and Naeem 1994). SUMMARY Cooperative breeding has been studied from a several points of view relative to the adaptive, physiological, developmental, and historical bases of the behavioral system. Cooperative breeding is a complex behavior, and the existence of each component of the behavior (delayed dispersal, delayed 39 breeding, helping) may be explained at different levels of analysis (functional, mechanistic, developmental, phylogenetic). Cooperative breeding has evolved a number of times and probably for a number of different reasons in different groups. In Chapters 3 and 4,1 discuss several of the points introduced in this chapter. In particular, I examine the relative roles that ecology and history have played in shaping the patterns of evolution in CB systems seen in the bee-eaters (Family Meropidae). 40

CHAPTER TWO- PHYLOGENETIC RELATIONSHIPS OF THE

BEE-EATERS (AVES: MEROPIDAE) The species in the family Meropidae, commonly referred to as the bee- eaters, are a group of 26 species of brightly colored coraciiform birds (Sibley and Monroe 1990). They are distributed throughout the paleotropics and southern Eurasia. Bee-eaters display a great range of diversity in several ecological and behavioral traits, including breeding systems, migratory habits, foraging behaviors, and nesting preferences (Fry 1969,1984, Fry and Fry 1992; see chapter 4). These traits could be subjects of fascinating and productive comparative studies. However, to make a valid comparative study on any of these traits, we must first know the phylogenetic relationships within the group. Few systematic studies of the family have been completed previously (Table 2.1). Boetticher (1951) used, primarily, plumage color and shape characters to derive a classification for the family. Fry (1969) used behavioral, ecological and plumage shape/color characters to derive a phenetic classification and phylogeny. Fry (1984) later modified the placement of certain lineages using information from geographic distributions of races and further behavioral observations. The major areas of disagreement between these two researchers are the placement of four forest species. Fry's Merops gularis and M. muelleri are placed in a separate genus in Boetticher's classification. Boetticher also Ivimps

Fry's Merops breweri and Meropogonforsteni into Nyctiomis {Nyctyomis).

Additionally, Fry 1969 places the forest dwelling African M. breweri well within

African savanna Merops species while Boetticher places this species with Asian forest species. Fry later revised his views, agreeing that M. breweri's closest relative may be Meropogon forsteni (Fry 1984:203). The classification of Sibley and 41

Monroe (1990) agrees with that of Fry (1984) except the former grants Merops philippiniis (M. siiperciliosiis philippinus in Fry (1984)) and M. niibicoides (M. nubicus nubicoides in Fry (1984)) full species status. Relationships within bee-eaters from previous systematic work were based primarily on each authors intuitive feel for the characters considered. The following study is the first to reconstruct the phylogenetic relationships within the family using explicit, cladistic methods. In this study, I recoristruct the evolutionary relatioriships among the species in the family Meropidae using a data set based on variation in plumage color, pattern and shape. A variety of maximum parsimony-based analyses, given different assumptions concerrung character structuring and weighting, are used to examine the robustness of the resultant trees. I then examine the biogeographic distribution of each species and make recommendations, given the current evidence, concerning the best phylogenetic hypotheses for this family. Finally, I compare the results of my analyses to a previous phenetic study of this group. Boetticher 1951 Classification Fry 1984 Classification Nyctyomis (Nyctiomis) amictus Nyctyomis amicta athertoni athertoni (Meropogon) forsteni Meropogon forsteni (Bombylonax) breweri Merops breweri Meropiscus gularis gularis muelleri muelleri Merops (Melittophagus) pusillus pusillus lafresnayei oreobates variegatus variegatus (Dicrocerciis) hirundineus hirundineus (Coccolarynx) revoilii revoilii bullocki bullocki bullockoides bullockoides (Melittias) erythrocephalus leschenaulti (Merops) apiaster apiaster (Aerops) boehmi boehmi albicollis albicollis (Phlothrus) orientalis orientalis bicolor viridis (Blepharomerops) superciliosus superciliosus omatus omatus (Tephraerops) malimbicus malimbicus (Melittotheres) nubicus nubicus persicus

TABLE 2.1. Two classifications of bee-eaters. Names in parentheses indicate subgeneric designations. 43

METHODS Plumage characters A quick glance at a series of bee-eater specimeixs suggests several plumage regions that retain information on species relationship. However, each region typically has a large array of colors when examined across all species and races. Coding characters in a manner that captures all the information available in these plumage regions can be accomplished in two ways. First, each plumage region can be represented as a single character with a large number of color character states. Second, each plumage region could be broken into several characters, each representing presence or absence of a particular color. If a region is represented by a single multistate character, additional information can then be incorporated in each character using character-specific step matrices (Appendix A). Character step matrices allow easier transitions between only the most similarly colored character states relative to all other colored character states (e.g. orangish-red to red easier than either is to blue). Such step matrices were used in the primary analysis of this study. No attempt was made to construct more detailed step matrices for relationships between colors of more distant spectral affinity. Therefore, although complex, these step matrices are actually quite conservative, while still utilizing much of the information available in each multistate plumage character. Three basic levels of transition rules were used in construction of step matrices. Each level assumes increasingly difficult genetic transitions between colors. The easiest level of transitions was that between green and green with blue tips, chestnut and red chestnut, and dark greeriish-brown and light greenish-brown. The next level of transitions linked all greens together and to 44 greens mixed with other colors. Also in this transition level were similar niles for blues, oranges, and reds. The final, most difficult, level of transitions were those between more distant colors. The exact weighting of steps in each matrix was determined by the number of classes represented in that matrix. Rules for transitions between character states for characters composed of complex color patterns were based on the number of color components shared between states (e.g. see gorget. Appendix A). Again, a second, alternative method of extracting the same information is to break each multistate plumage character into multiple, individual characters. This latter method was not used in this study because it would artificially inflate the number of independent characters in each data matrix. That is, certain characters would contain character states that could not exist with any character state in certain other characters. An example illustrates how this problem could occur. The step matrix for the forecrown color in the race data matrix has three basic trarisition classes (Appendix A). The first transition class represents slight changes among species with green in the forehead. In some cases the basic green forehead is only slightly modified by the addition of faint blue tips and this slight difference is indicated by a cost of only a single step. Other transitions from the basic green involve more complex changes such as the addition of other colors (e.g., orange, brown) throughout the forehead region. These more complex changes result in an increased cost of two steps. Finally, some species have either a chestnut or reddish-chestnut forehead, probably indicative of these species' close relatior\ships, and therefore transitions between these two colors cost only a single step in the matrix. All other transitions are between colors of more distant 45 spectral affinity and therefore cost three steps. Now, while it would be possible to represent these basic color changes in the forehead as different characters, (e.g., "basic green or not", "green with other color or not", "chestnut or not"), the characters cannot be independent. That is, if the forehead is chestnut, it cannot be green. While the independence of characters in most phylogenetic analyses is typically assumed and not tested, one certainly should not knowingly use non- independent characters (Farris 1983). The nature of the characters used in this study provided one additional analysis problem in this study. Bootstrap analyses are reliable estimates of the information content of data matrices and the probabilities of each node in resvdtant reconstructions representing those in the true phylogeny (Felsenstein 1988, Hillis and Bull 1993). Bootstrapping measures the variation present in the characters of a data matrix and how representative this subset of characters is to the larger distribution of all possible characters. Multiple bootstrap pseudoreplicate matrices are created by sampling characters with replacement from the original data matrix. Use of this techniques assimies independence of characters. As explained above, a great deal of information is contained in each of the large multistate characters (and its relevant step matrix) used in this study. The amount of information lost during the creation of each bootstrap replicate by leaving out any single, large, multistate character is, therefore, much greater than that lost in more conventional characters. The result of such analyses are highly conservative bootstrap consensus trees with very little resolution. This problem can be corrected if each multistate plumage character is broken into multiple characters. However, we then have the problem of non-independence discussed 46 above. Therefore, regardless of how characters are coded in this study, use of the bootstrap technique is not appropriate. Phylogenetic analyses Thirty plumage regions were scored for color, pattern and shape using museum study skins and spirit spedmeiis. Detailed structural components of feathers consistently associated with carotenoid pigments suggests homology of these colors (Fry 1969). Multiple individuals of each sex where examined in each of the 51 geographically defined race recognized by Fry and Fry (1992). Only plumage characters that were invariant within races were retained in analyses. Races identified as distinct by Fry and Fry (1992), but were not identified as distinct entities using the plumage characters, were combined {Nyctyomis athertoni athertoni = N. a. brevicaiidata, Merops hirundineus hirundineus = M. h. furcatus, M. pusillus meridionalis = M. p. argutus, M. bullockoides bullockoides = M. b. randorum, M. orientalis cyanophrys = M. o. mjdanus, M. superciliosiis superciliosus =

M. s. altemans). Twenty six species are recognized (Sibley and Monroe 1990). Sexual dichromatism is rare in bee-eaters, but in cases where it does occur, analyses were conducted in two ways. First, sexes were used as separate "taxa" in phylogenetic analysis. Second, the dichromatic characters were coded as poljnnorphic for that species or race. Maximvim parsimony phylogenetic analyses were conducted using PAUP (version 3.1.1, Swofford 1993). Two matrices were used in the study: one with species as terminal taxa and one with races as terminal taxa (Appendix B). Due to the number of taxa in each matrix, an exhaustive tree search was not possible and heuristic searches were conducted. All most parsimonious trees (MPTs) were retained. To increase the chance that the most parsimonious solution 47 would be found, 100 random addition sequence searches were completed with TBR branch swapping in each analysis. Starting trees for branch-swapping were derived by stepwise addition of taxa. The number of TBR islands represented in the MPT set and the number of replicates in which they were found is reported to indicate thoroughness of searches (Maddison 1991). A range of additional analysis options were used in this study on both race and species data matrices (Table 2.2). The primary analyses in this study used characters structured with character-specific step matrices. In this first set of analyses, the weight of all characters was one. This analysis extracts the maximimi amount of information from each character and is expected to produce the most accurate phylogeny or phylogenies. The next most iriformation-rich analysis used the "scale" option in PAUP to standardize each character's relative weight. This approach controls for the fact that characters with larger numbers of character states will have a greater influence on the tree's reconstruction by requiring more steps than smaller characters. The option works by down- weighting large characters so that all characters have a standardized ii\fluence on tree reconstruction. The result of using this option, however, is a reduction of influence by the most information rich, large characters. Additional analyses were completed to test the robustness of conclusions from the primary analysis. Secondary analyses examined all characters unordered. Analyses of unordered characters were conducted both with character weightings equal to one and also with characters scaled to have standardized weights. One additional analysis was conducted with characters initially having weights of one, with results subsequently reanalyzed with successive approximation character weighting until resultant trees were stable. 48

The maximum value of the rescaled consistenq'^ index was used to reweight characters after each cycle. Decay-index analyses were conducted for specific taxa and dades to assess the degrees of confidence in their relative placements in each tree. These analyses were conducted using the exact same analysis options and search criteria listed above; however, topological constraints appropriate to the question at hand were maintained in each search. Comparisons of tree lengths between original and constrained searches point to the degree of support of the taxon's or clade's original placement. 49

Characters Structured with Step Matrices • characters' weight equals one • characters scaled to have standardized weight

Characters Unordered • characters' weight equals one • characters scaled to have standardized weight • characters first without weighting assumptions and then reanalyzed with successive approximation character weighting xmtil resultant trees were stable

TABLE 2.2. Combinations of character weighting and ordering options used in phylogenetic analyses for both species and race data matrices. 50

Relationships among the families of the order Coraciiformes are unclear and times of divergence among the families are quite long (Cracraft 1981, Sibley and Ahlquist 1990). The choice and utility of outgroup taxa among these families are therefore questionable, particularly when using plumage characters. I instead rooted the trees between the two species of the likely basal genus

Nyctyomis and the remaining bee-eater lineages. Evidence for the basal placement of this genus is seen in the DNA-DNA hybridization data of Sibley and Ahlquist (1990: Figure 359). Resultant trees from each phylogenetic analysis were examined for clades robust to the range of different options used across analyses. Certain analyses, however, had a priori expectations concerning their accuracy. Analyses using step matrices, using more of the information in each character, were expected to provide more accurate estimates of phylogeny relative to unordered analyses. Additionally, analyses using races as terminal taxa were expected to jaeld better phylogenies concerning relationships at the species level than would those using the species matrix, due to the latter's high number of polymorphic characters (Wiens 1995). Comparisons of trees' fit to biogeographic patterns were examined by mapping geographic distributions on each phylogenetic hypothesis using MacClade (version 3.0, Maddison and Maddison 1992). 51

RESULTS Primary arialyses used characters with weights equal to one and structured with step matrices. When races were used as terminal taxa, 2016 MPTs were found (Figure 2.1). This single TBR island of trees was found in 73 of 100 replicates. Four MPTs were reconstructed in the analysis using the species data matrix (Figure 2.2). This latter set of trees was composed of two pairs of very similar trees representing two TBR islands. Islands were found in 30 and six of 100 replicates. When characters were scaled to have standardized weights and structured with step matrices, nine MPTs were found using the race matrix (one TBR island found in 11 replicates) and one MPT using the species matrix (found in 30 replicates) (Figure 2.3). Note that the strict consensus tree of the nine MPTs in figure 2.3A collapses into a single species level tree if the monophyly of M. gularis is assimied. N. amicta N. athertoni Meropogoti forsleni M. breweri M. muelleri muelleri M. muelleri mentalis M. gularis gularis M. gularis australis M. pusillus pusillus M. pusillus meridionalis M. pusillus cyanostictus M. pusillus ocularis M. variegatus variegatus M. variegatus loringi M. variegatus bangweoloensis M. variegatus lafresnayii M. oreobates M. bullocki bullocki M. bullocki jfrenatus M. bullockoides M. malimbicus M. nubicus nubicus M. nubicoides M. revoilii M. albicollis M. hirundineus hirundineus M. hirundineus chrysolaimus M. hirundineus heuglini M. ornatus 5 VI M. persicus persicus M. persicus chrysocercus M. superciliosus superciliosus M. philippinus M. ahiaster A1. Doehmi M. orientalis ferrugeiceps M. viridis viridis M. viridis americanus M, leschenaulti leschenauiti M, leschenaulti quinlicolor M. leschenaulti aiidamaiiensis M. orientalis orientalis M. orientalis beludschicus M. orientalis viridissimus M. orientalis cleopatra M, orientalis cyanophrys FIGURE 2.2 Four most parsimonious trees firom primary analysis of the spedes data matrix. Qiaracter weights equaled one and characters were structured with step matrices. N. amicia N. athcrtoni Nyctyomis amlcia jWer^p^on. forsleni NyclyomI altwrtoni M. muelleri muelleri Mempogon loistenl M. muelleri rtieiflalis M. mlari? gutaris Merops tjrewerl M. ^jlans auslrfflis M. rmundjii Iiirwiwnejis Merops muelleri M. pirunqineus ciirysochryspla laimus m u> Merops gulatis M. htrundtneus heugmini M. boehmt Merops hlrundlneus M. orfenla/fs pnertlalif M. onentahs beludscnicus CO fO Merops boehml M. orienlalis cvamphrys Merops orlenlalls A1. onentahs viridissimus M. orientaUs cleopatra Merops perslcus M. orifnlahs kxruseiceps M. v(r(d{s vindfs Merops superclllosus M. vindis americanus M. leschemulii lesclfenaulti Merops aplaster M. leschenaulU qumticolor Merops phlllpplnus M. leschenaulU anaamanensis M' ornatus, Merops virldls . m/t/mtbiciis.mafimmcus, . iwbicus iwbicus Merops leschenaulU . nub{coides Merops ornatus . perstcus persicus . persicu^ chrusocercus Merops mallmblcus . supfrctliosus superctliosus pluhppinus Merops nubicus M. apiasfef ^ fS" S! Merops nublcoldes albicf^lis Merops alblcollls Merops revollil

Merops bullockoldes memtonalts Merops bullocU !lus ocularis pusi(lus cyanosljctus Merops pusillus varlegatus variegatus var/egatus bapgweolocnsis Merops varlegatus variegatus loringi Merops oreobales variegatus lamsnauu oreobates 55

Subsequent analyses were conducted to test the robustness of the results of the primary analysis to changes in analysis options. Analyses of unordered characters with weights equal to one yielded 8472 MPTs from the race matrix (one TBR island fovmd in 38 replicates) and 12 MPTs from the species matrix (one TBR island fotmd in 40 replicates) (Figure 2.4). After subsequent analysis of each data matrix with successive approximation character weighting, 675 MPTs were reconstructed using the race matrix (one TBR island foimd in 55 replicates of the last iteration) and three MPTs from the species matrix (one TBR island found in 47 replicates of the last iteration) (Figure 2.5). Analyses of imordered characters with weights standardized resulted in 27 MPTs from the race matrix (one TBR island found in 38 replicates) and one MPT from the species matrix (found in seven replicates) (Figure 2.6). 56

FIGURE 2.4. Results from analyses in which characters were unordered with weights equal to one. (A) Strict consensus tree of 8472 MPTs from race matrix analysis. (B) Strict consensus tree of 12 MPTs from species matrix analysis. N. amicla N. alherloni Nycti/ornis amicla M. hiriiiidineus hinmdlneus M. Iiirumlineiis ehrysalaimus Nyclyorinis alherloni M. hirumlineus heuglitil M. leschenaulli lescliemulli Merofs liiruniiineiis M. leschenaulli aiidamanensis Meraps leschemulti M. leschenaulli i]ulnlicolor M. apiasler Merops opiaster M. orienlalis arienlalis M. orienlalis viridissimus Meraps boehmi M. orienlalis cleopaira Meraps arienlalis M. orienlalis cyanaphrys M. orienlalis beludschicus Meraps viridis M. orienlalis /errugeiceps M. viridis viridis Meraps arnaltts M. viridis americanus Merops philippimis M. onialus M. philippinus Merops persicus M, poehmi M. persicus persicus Merops superciliasus M. persicus chrysacercus M. superciliosus superciliasus Meraps albicallis M, albicollis M, revoilii Merops revoilii M. malimbicus Merops malimbicus M, nubicus nubicus M. nubicoides Meraps nuhicus M. bullackoides Merops nubicoides M. bullocki bullacki M. bullackifrenalus Merops bullackoides M, pusillus pusillus M. pusillus meridlanalis Merops bullacki M. pusillus cyanosliclus Merops pusillus M. pusillus ocularis M. variegalus variegalus Merops mriegatus M. variegalus banguvoloensis M. variegalus loringi Meraps areobates M. variegalus lafresnayii M, areabales Merapoganfarsleni Merapogon JbrslenI Merops miielleri M. muelleri muelleri M. muelleri meiilalis Merops gularis M. gularis gularis M. gularis auslralis Merops breweri M. breweri FIGURE 2.6. Results from analyses in which characters were unordered with weights standardized. (A) Strict consensus tree of 27 MPTs from race matrix. (B) Single MPT from species matrix. 59

DISCUSSION When examined across all analyses several consistent patterns can be seen. First, certain clades do stay relatively consistent across the shallow portions of the trees (i.e., Mero-pogon forsteni - Merops gularis; M. viridis - M. leschenaulti; M. malimbicus - M. nubicoides; M. siiperciliosus - M. philippinus; M. pusillus - M. oreobates). Additionally, certain lineages, with rare exceptions, group together in sequential evolutionary patterns coming off the main backbone of the Merops dade (i.e., M. bullocki - M. bullockoides; M. revoilii - M. albicollis). The second main pattern seen is grouping of races into spedes dusters, if not always clades, for all species except M. orientalis. This latter species has one race, M. orientalis femigeiceps, that does not groups with the other races in the spedes in one reconstruction (Figure 2.1). Finally, Meropogon forsteni is placed within the

Merops dade in all trees. However, comparing trees across analyses shows considerable imcertainty relative to the placement of deep branches of the tree. Comparison of primary and secondary analyses Trees derived from primary analyses in which characters had weights of one differ from each other relative to rearrangements of certain deep branches.

In particular, M. viridis and M. leschenaulti lineages are placed within a small dade with M. hirundineus and M. omatus in two trees in Figure 2.2 but show a deeper branching pattern in the other two trees in Figure 2.2 and also the tree in

Figure 2.1. Additionally, M. orientalis lineages are followed by the subsequent evolution of a cluster of species (M. superciliosus, M. philippinus, and M. persicus) at the base of the remaining portion of the Merops dade in Figure 2.2, but possibly not in Figure 2.1. 60

The next set of analyses had characters scaled to standardized weights. Resulting trees from these latter analyses are very similar to each other (compare Figure 2.3A and B), but differ substantially from trees from the analysis in which characters have weights equal to one in regard to deep branching patterns. For

example compare the placement of M. orientalis in Figures 2.1 and 2.2 to that in Figure 2.3. Secondary analyses utilize unordered characters. For these analyses, the trees from both race and species matrices show a great deal of congruence given the same analyses options (e.g. compare Figure 2.4A to B, 2.5A to B and 2.6A to B). These trees do differ substantially from those of primary analyses, particularly in deep branching patterns. Trees from primary analyses (Figures 2.1 - 2.3) are reconstructed using more of the information contained in each multistate character. Therefore, I suggest that these trees are the best hypotheses given current data. In the following sections, I will compare the biogeographic patterns of each of these primary analysis trees and how each compares to previous systematic work on the family. Biogeographic patterns Biogeographic patterns are similar for each of trees from the primary analysis (Figures 2.7 and 2.8). Breeding distributions were taken from Fry (1984). Most trees show the origin of the family as either southeast Asia or Africa with most of the diversification of the family occurring in Africa. In all trees,

Meropogon forsteni shows a pattern of invasion of Indonesia from African ancestors. Other invasions of non-African areas are concentrated in one subclade

of Merops in one tree (Figure 2.8B), but tend to be spread over the tree in other 61 reconstructions. Figure 2.7A shows more uncertainty in biogeographic patterns due two deep polytomies. Assuming an African origin in this figure would show fairly simultaneous start of diversification within Africa, with invasion of non- African areas. However it is also possible to show early diversification in Asia with a later invasion of Africa in this tree. 62

If* i 1 *3 O & S3 S i|^ Jli 5 fi? • y* '3 ^ >3S V*«A W ^ Ci S= •"c "55^=? saai- ^1 I S*5^v» W (A -5 ^ ^ K u ill =.s \*3 *J3 3V L-5 t •S.lfc •^i•s^S' I ^ ^ M *• ^ *«» *« o 'S w> "*wt ^ ^>3 ii> "S42 -ffSa Sb 53 3S S3 •52 42 .2 43 s s s 5 3 *S 5 5 >SS5£o^^^llil §lll111 3SSS:Q>C.S.SO&&>*AS^sa iill-l i i i.ll-i|lll il ^ SSSSSSSSSSS555555555555 5555555 55555 D B 0 I I

Biogeography unordered I I SEAsia I I Indonesia Bgsa India HH Africa i^H Middle East Eurasia Australia polymorphic I I equivocal W> (A S C V» (A >«• i •«. S .S v> 1 •2 tg .S tn Is ^ .o C S? s: &0 s -S 1 .u3 •?• 'rr a s cs •H•5 1 "§ •JS «_» cs il^ v» wi ~ a •3 S *« *§ ,c ^ .S *? ? Q 5 So A a. § "5 S 2 s 1c *5 •5; oSt 2 I- 1•6 5-5(A <« (A

FIGURE 2.7. Reconstructions of biogeographic patterns on trees from primary analysis in which character weights equaled one and characters were structured with step matrices. (A) Strict consensus tree of races. (B) Representative species tree. Biogeographic patterns are similar for each of the four species trees (see Figure 2.2). aN.amlcIa O Ni/ctyonus aniicia taN. alhcrloni ta Meropogon forsleni Q Nyctyornis allierloiii m M. hmivrl •Meropogon forsleni m M, miielleri muelleri •M. muelleri menlalis •A4ero;)s breweri mM.gularisgularls mM.gubirlsauslriilis MMerops imielleri m M. hirundineus hirumlineus UMerops gularis m M. hirundineus chrysokimus m M. hirundineus heuglini Merops hinmdineus mM. boehmi B M. orienlalis orienlalis Merops boelinii oM. orienlalis beludschicus m M. orienlalis cyanophrys 10 Merops orienlalis mM, orienlalis viridissimus \0 Merops persiciis m M. orienlalis cleopaira taM. orienlalis frrrugelceps U Merops superciliosus 13 M. viridis viridis aM, liridis amerimnus \\3 Merops apiaster aM.lesclienaulli leschenaulli \0 Merops philippinus aM. leschenaulliijuinlicolor dM. leschenaulli andamanensis \Q Merops viridis eaM.omalus M. malimbicus BMerops leschemulli m M. nubicus nuMcus Merops onuitus M. nubicoides o M. fiersicus persicus Merops malimbicus M. ferslcus chrysocercus M. superciliosus superciliosus Merops mibiciis M. philippinus M. apiasler Merops nubicoides mM. albicollis Merops albicollis M. revoilii mM, bullockoides Merops revoilii •M. bullocki bullockl Merops bullockoides = M. bullockifrenalus •m M. pusillus pusillus Merops biillocki mM. pusillus meridionalis mM. pusillus ocularis Merops pusilliis m M. pusillus cyanosliclus •M. varlegalus variegalus Merops variegaliis mM. variegalus bangweoloensis Merops oreobates m M. variegalus loringi mM. tarlegalus lafresnayii mM. oreobtiles 64

Comparison with earlier systematic study The only serious attempt to reconstruct explicit phylogenetic relationships within the family Meropidae prior to this study was that of Fry (1969). In that study, a combination of structural, ecological and behavioral characters were used to derive a phenetic based hj^othesis of species relatior\ships. This hypothesis is represented in the diagram in figure 2.9, which represents a "cross- section through the top of a phylogenetic tree" (Fry 1969, pg. 587). I have translated this information into an estimate of a tree (Figure 2.10). Comparisons between the preferred race and species matrix trees in this study to the Fry tree show many similar groupings, but not necessarily as monophyletic clades, in the shallow portions of the tree: N. amicta-N. athertoni;

M. giilaris-M. muelleri; M. bullocki-M. bullockoides; M. pusillus-M. variegatus-M. oreobates; M. superciliosus-M. philippinus-M. persicus; and M. malimbiciis-M. nubicus-M. nubicoides. The sister-group relationship of M. boehmi and M. orientalis agrees with each tree from this study except for that in figure 2.1 in which M. boehmi is the sister taxon to only one M. orientalis race. The M. viridis-M. leschemulti-M. apiaster clade in Fry's tree agrees completely with trees in figure 2.2 only. All other trees in this study show a sister-group relationship between

M. viridis and M. leshenaulti but no consistent relationship of these species to M. apiaster.

The branching of Meropogon forsteni early in the Fry tree is quite different from the pattern seen in this study. How strong is the support for the positioning of this species within Merops lineages, as seen here? In subsequent decay-index analyses, Meropogon was forced to the base of the tree, after the

Nyctyomis lineage and before the Merops clade, resulting in longer trees of 65 varying extent. In the primary analyses in which characters are weighted one, trees are 9.1 percent longer using the race matrix (826 steps versus 901 steps) and 1.3 percent longer using the species matrix (761+ steps versus 771+ steps. Pluses indicate uncertainty of exact length due to the presence of more than two character states in several characters). In primary analyses in which characters had standardized weights, trees are 1.0 percent longer vising the race matrix (87594 steps versus 88488 steps) and 1.3 percent longer using the species matrix (83087+ steps versus 84224+ steps). The level of support indicated by these differences in tree lengths regarding the inclusion of Meropogon forsteni within

Merops is uncertain. However, the placement of the Indonesian Meropogon forsteni well within an African clade seems biogeographically improbable.

Branching of this species between the Asian/Indonesian Nyctyomis and primarily African Merops seems to be more intuitive and would suggest and Asian origin for the family. Additionally, differing rib numbers and structure are reported for this species, setting it apart from both Merops and Nyctyomis. (Fry 1969,1984). I did not examine these characters personally and they were therefore not vised in my analyses. Obviously, further phylogenetic data, particularly DNA sequence data, are needed to examine this question further.

The branching pattern of M. breweri in the Fry tree also differs from those seen in this study. In this study, the forest dwelling M. breweri is consistently associated in some maimer with other forest species: M. muelleri, M. gidaris and

Meropogon forsteni. Fry's phylogeny, however, has this species nested within savaivna species. As mentioned in the introduction. Fry (1984:203) revised his views, stating that M. breweri's closest relative may be Meropogon forsteni. This sister-group relationship is seen in several trees in this study (possibly figure 2.1, 66 two trees in Figure 2.2, and in Figure 2.3A and B). Whether this sister-group relationship in Fry's revised assessment would pull M. breweri outside of Merops, or place Meropogon forsteni within Merops is unclear. 67

African Endemics

FIGURE 2.9. Fry's diagram of a cross section of the top of a phylogenetic tree. Circles describe distance relationships with respect to Merops pusillus (6). Bottom to top of figure represents general advancement of characters from plesiomorphic states. Dashed-line circles represent superspecies groups. Shaded areas equal forest, unshaded equals savanna. Black dots = Nyctyomis amicta and N. athertoru; hatched dot = Meropogon forsteni; numbers = Merops: (1) gularis, (2) muelleri, (3) buUockoides, (4) buUocki, (5) hirundineus, (6) pvisillus, (7) variegatus, (8) oreobates, (9) breweri, (10) revoilii, (11) malimbicos, (12) nubicus, (13) albicoUis, (14) boehmi, Q5) orientalis, (16) omatus, (17) superciliosiis, (18) philippinus, (19) viridis, (20) leschenaulti, (21) apiaster. Redrawn from Fry (1969). Nyctyomis amicta Nyctyorni athertoni Meropogon forsteni -^Merops gularis toMerops muelleri ^Merops bullocki oiMerops buUockoides foMerops breweri (nMerops hirundineus oMerops pusillus •^Merops variegatus coMerops oreobates o Merops revoilii ^Merops boehmi ^Merops orientalis ^Merops albicollis Merops persicus Merops superciliosus mMerops philippinus m Merops ornatus Merops viridis ^Merops leschenaidti ^Merops apiaster :! Merops malimbicus Merops nubicus Merops nubicoides 69

Conclusions Givai the available evidence and the maximum parsimony analysis presented here, two basic sets of phylogenetic hypotheses seem most probable for the bee-eaters. The first set is represented by six completely resolved species trees (Figures 2.2 and 2.3). The remaining set of trees is represented as a strict corisensvis tree and is a more conservative estimate of race relationships (Figure 2.1). Biogeographic patterns show either a South-east Asian or African origin for the family, with most of the diversification in the family occurring in Africa. Consistent groupings of many taxa are apparent when trees from each analysis are compared to each other and to the previous work of Fry (1969). Much uncertainty remains, however, concerning placement of certain taxa such as

Meropogon forsteni and Merops ferewen and the pattern of divergence among deep branches. Analysis of slowly evolving DNA sequences might provide greater insight to help resolve these conflicts. 70

CHAPTER THREE- BREEDING BIOLOGY AND SOCLVLITY OF

BEE-EATERS IN THAILAND The bee-eaters (Aves: Meropidae) are a relatively small clade of 26 species that show considerable diversity in their social and breeding behaviors. For several species, however, data on these behaviors are scant or completely lacking. As part of a broader study of the evolution of cooperative breeding in this group, this paper describes aspects of the social structure and breeding biology of four species breeding in Thailand.

The little green bee-eater (Merops orientalis) is easily the most morphologically variable species in the family with six to eight races (Fry 1984, Chapter 2 this dissertation). Whether this species also shows behavioral variation across its wide geographic range is not known. To address this question, behavioral data from Thailand are compared to a similar study conducted in India (Sridhar and Karanth 1993). Concerning the blue-tailed bee- eater (M. philippinus), little information was previously available on its breeding biology except that it sometimes nested in colonies. Here I show that this species breeds cooperatively and has a complex social system similar to other colonial bee-eaters. I also describe behavioral inferences made from opportunistic observations of the bay-headed bee-eater (M. leschenaulti) and the blue-bearded bee-eater {Nyctyomis athertoni). 71

STUDY SITES AND METHODS I studied the little green bee-eater at Khao Sam Roi Yot National Park (Prachuap BQuri Khan Province) from 16-22 March;1 -10,22,23 April; and 3-5 May 1996 (110 obs. hrs). The study area is an open deciduous woodland habitat just southwest of the park headquarters. Surroimding areas include open, dry marshlands, prawn farming ponds, mangrove bordered streams and open scrub. The blue-tailed (59 obs. hrs.) and bay-headed bee-eaters (11 obs. hrs.) were studied on the banks of the Huai Sai Yai, approximately 15 km. west-northwest of Ban Nadee (Prachinburi Province) from 12,13 March; and 25 - 29 April 1996. The river is surroimded by secondary-growth deciduous forest and agricultural fields. I observed nesting blue-bearded bee-eaters at Khao Yai National Park (Nakhon Ratchasima Province) on 30 April 1996 (4.5 obs. hrs.) in a montane evergreen rainforest. I captured little green and blue-tailed bee-eaters using mist nets and marked individuals with unique combinations of non-toxic paints to allow recognition of individuals. Behavioral observations of blue-tailed bee-eaters were made from within a blind. I estimated colony population size in the blue- tailed bee-eaters by repeated censuses of the number of marked and unmarked individuals at the colony. I assumed that the percentage of marked birds not seen in each census was a reasonable estimate of the percentage of unmarked birds also away from the colony at that time. Focal-individual observations of little green bee-eater individuals were collected to gauge the extent to which individuals remained within view of their nesting cavities in anti-predator guarding behavior. These observatioris were conducted by watching one individual continuously for 30 minutes noting whether it remained within the 72 vegetation immediately surroimding the nest cavity ("guarding trees"), whether it was facing the nest cavity, the frequency of perch changes, and the behaviors associated with perch changes (e.g. flycatching, pursxiit of predators), and the presence or absence of another bee-eater in a guarding tree. Other non-trivial behaviors such as copulation and courtship feeding were also recorded during focal observations. Determination of "breeder" or "helper" status of individuals was made using observations of dominance interactions, courtship feedings, and copulation frequency. Average values for each data category are reported with standard error values. Species were considered to have cooperative breeding systems if at least one nest had three or more individuals bringing food to nestlings or fledglings. 73

RESULTS Little green bee-eaters

Colony structure- Little green bee-eaters breed in a loose colony. A loose colony is defined here as one having nests separated by short distances but still being clustered within the habitat patch, with home ranges of colony individuals highly overlapping. Seven nests were found dug into slightly sloped or flat ground in the main study area. Several additional "false" nest holes were dispersed among those finally chosen for breeding. Three additional active nests were foimd just outside the main study area. Average distance between adjacent nests was 19 m.

Breeding chronology- Breeding is quite synchronous. Excavation of nest cavities was in its final stages or complete by mid-March. Courtship feedings and copulations were seen frequently during mid-March and early April. Typically, males fed only one female and females were fed only by a single male; however, in two groups a bachelor/helper male was seen to feed the breeding female once. One group was feeding two fledglings on 3 May. On the same date, two nests contained nestlings near fledging age, while two others had less developed nestlings. Dragonflies appeared to make up a high proportion of food items fed to nestlings and fledglings. The remairiing two nests failed before fledging yovmg, possibly due to nest predation.

Predators and anti-predatory behavior- One potential nest predator in the area is the butterfly lizard {Leiolepis belliana), which is abundant in the area. Bee- eaters directed diving attacks at these lizards on more than 25 separate occasions. This anti-predator behavior was also directed less frequently towards dogs and snakes. Other species of birds and lizards did not elicit this form of anti-predator 74 behavior even when very near the nest. During the presumed periods of egg la3^g and incubation, individuals spent a significant portion of their time in vegetation that gave them clear views of the nest cavity. Five individuals were observed in 20,30-minute observation periods from 5-10 April to examine the frequency of this guarding behavior. Individuals spent an average of 55% (± 8%) of their time in guarding frees, with at least two individuals guarding 45% (± 8%) of the time. Nests were left unguarded for only 18% (± 6%) of the time. Males are much more active while guarding than are females, making an average of 24.8 (± 4.0) movements/obs. period versus 9.3 (± 3.1) for females. Most of these movements were due to flycatching behavior (17.3 (± 3.8) sallies/obs. period, male; 4.7 (± 1.7) female).

Evidence of cooperative breeding- Little green bee-eaters breed cooperatively in Thailand. Helpers were associated with at least four of the seven nests (57%) on the main study area. At one other nest, the breeding male disappeared during the study and ariother individual was seen shortly afterwards bringing food to this nest. It is unclear whether this individual was a helper or a replacement mate, and therefore the percentage of groups with helpers could be as high as 71%. Two cooperatively breeding groups in the main study area and possibly another group observed on a single occasion outside the study area each contained two helpers. All other cooperative groups had a single helper. I saw helpers associated with groups only rarely before the incubation stage of nesting was well vmderway. In fact, before egg-laying was sxispected to be complete, breeding males frequently chased potential helpers from the area of both the nest and breeding female. Helper duties included mobbing predators, feeding nestlings and fledglings and occasionally the breeding female, and possibly 75

incubation. Helpers confaibuted as much or more of the food items brought to nestlings. Observations at one nest showed the helper bringing food items to the nest in seven of twelve occasions. Blue-tailed bee-eaters

Colony structure- The blue-tailed bee-eaters in this study bred iri a dense colony in a sandy river bank. In the center of the colony, a 130 m^ area contained 49 nest cavities, 16 - 19 of which were active. The activity of three cavities was questionable because individuals were seen only to perch in the cavity entrance on a few occasions. People from the local village claim the colony has bred in this general location for at least 10 years and the remaining cavities were either nests from previous breeding seasons or false nests dug this season. Eight additional active nests were seen scattered on the fringes of the main colony. Estimates of colony population size ranged from 40 - 70 individuals.

Breeding chronology- Stages of the breeding cycle are fairly asynchronous in this species. Excavation of nests dominated colony activity in mid-March but was also seen in late April at two nests. Birds at most nests made irifrequent nest visits (avg. 0.65 ± 0.11 visits/hr., 11 nests), suggesting most individuals were incubating in late April. Two groups, however, were apparentiy feeding recently hatched nestiings in late April (avg. 2.25 ± 0.33 visits/hr.). The visitation rates of these two groups are significantly different (f = 4.9360, fo.ooi,ll = 4.437). At least one female was still laying eggs at that time. Certain breeders were therefore imsynchronized by at least the length of the incubation period, a minimum of 24 - 26 days. Evidence of cooperative breeding and sociality- Like little green bee-eaters, blue-tailed bee-eaters breed cooperatively in Thailand. Four nests had three or 76

more individuals delivering food items to the nest. The presence of multiple unmarked individuals at four additional nests suggests that they also had helpers. Interactions among individuals in the colony were frequent and suggest a complex social system, similar to that of other colorual bee-eaters. In particiilar, intraspedfic brood parasitism and forced extra-pair copulations, behaviors difficult to document, may be common in this population. Concerning the first, three individuals were seen to enter at least two nest cavities and five individuals perched at the entrance of either two or three cavities each. Certain individuals vigilantly defended the immediate area around their nest by frequently displacing interloping individuals not in its group. Also, a cracked bee-eater egg with a small puncture was foimd on the ground outside a series of nest cavities from which it had apparently been ejected. With regard to the possibility of forced extra-pair copulations, several individuals (probably male) spent considerable time perched on the colony cliff face and were seen on two occasions to pursue an individual (probably female) exiting its nest cavity and flying from the colony. Bay-headed and blue-bearded bee-eaters I observed bay-headed bee-eater nests at the same site at which blue-tailed bee-eaters were studied. One nest was located on the edge of the blue-tailed colony, while two others were located along the river bank 45 m upsfream. One nest was only 1 m from a little green bee-eater nest. This species is cooperatively breeding in Thailand. At least one nest had three individuals bringing food to nestlings in late April. These individuals repeatedly perched outside the nest cavity, waiting in queue to feed nestlings. 77

A Blue-bearded bee-eater nest was found along the orc±iid waterfall trail in Khao Yai National Park. The nest cavity was dug 1 m high in the side of a small pit. On 30 April, two individuals were seen frequently bringing food to nestlings probably near fledging. There was no evidence of cooperative breeding at this nest. 78

DISCUSSION Little green bee-eaters in India and Thailand In most respects the breeding biology of little green bee-eaters in India (M. orientalis orientalis, Sridhar and Karanth 1993) and Thailand (M. o. femigiceps) are similar. However, small differences are apparent. In both areas the species breeds cooperatively in loose colonies. In India 20 - 57% of groups had helpers over three years of study, with an average of 38%. Cooperative breeding is fotind in 57 - 71% of groups in Thailand. The proportions of nests with helpers between these populations are not, however, statistically different, regardless of which of the two percentages is used for the Thailand population (x\ = 0.241 or 1.885, X^o.05,1 = 3.841). In India only a single helper was seen in attendance at each cooperative nest, while in this study two helpers were seen in two of four cooperative groups. Helpers typically arrive after the start of incubation. Indeed, in this study the behavior of breeding males at early stages of breeding indicates that helpers may be a threat with regard to both extra-pair copulation and intraspecific brood parasitism. Outside individuals bringing food to or simply perching near breeding females were displaced by breeding males, as were individuals who approached the nest cavity. Observations in India of one helper's preference for feeding one particular fledgling suggest the potential for mixed broods in this population as well. Two alloparental duties provided by helpers in many cooperative breeding species include feeding yoimg and protection of young from predators (Brown 1987). Helpers in this study delivered a significant portion of the food items to nestlings, as is the case in India. Bee-eaters also spend a significant 79

portion of their time in guarding trees, almost always facing their nest cavity. Frequent diving attacks were directed towards butterfly lizards and other potential predators when they approached the area of the nest cavity. In India, nests with helpers showed no predation, while those without suffered 20% predation (Sridhar and Karanth 1993). Social systems of poorly studied species Previously, very little information was available on the remaining species studied here. Blue-tailed bee-eaters were known to nest both solitarily and colonially. Colonies commonly contain 10 - 30 active nests, but can have hundreds (Fry and Fry 1992). The colony studied here was a typical small colony. This study is the first to document that blue-tailed bee-eaters breed cooperatively. This species shows additional similarities to other colorual cooperative breeding bee-eaters: red-throated (M. bullocki), white-fronted (M. bullockoides), and European bee-eaters (M. apiaster). Colony sizes in these species range from 40 individuals in the blue-tailed bee-eater to over a thousand in the European bee-eater (Fry 1984). Regardless of colony size, interactions between individuals both within their own and among different breeding groups can be complex (Fry 1972, Emlen and Wrege 1986, Jones et al. 1991, Fry and Fry 1992). In each species, males guard their mates from attempts at extra-pair copulation by other males, while at the same time attempting to gain extra-pair copulations themselves. Individuals guard their nests agair\st parasitic eggs from other colony members. Brood parasitism obviously lowers the fitness of the host female, but could also hurt the male's reproductive success imless these parasitic eggs are the product of his extra-pair copulations ("quasi-parasitism", Wrege and Emlen 1987). Biochemical studies on the effective rate of extra-pair copulations 80

and brood parasitism show that behavioral studies alone can lead to large underestimates of their occurrence (Wrege and Emlen 1987). In other words, if there is any evidence of these behaviors, then they may be qxiite common. The related behaviors seen in this study therefore are strongly suggestive that extra- pair copulation and intraspedfic brood parasitism occur in blue-tailed bee-eaters. Breeding synchronization is important to the feasibility of these two behaviors. If breeding is synchronous, fertile females become predictable resources for male extra-pair copulation efforts, and nests in the early stages of incubation become predictable resources for female brood parasitism efforts (Emlen and Wrege 1986). Blue-tailed bee-eaters lack close breeding synchroriization, so the success of these alternative reproductive tactics may be low in this species. Concerning the bay-headed bee-eater, an unpublished study cited by Sridhar and Karanth (1993) claims cooperative breeding in India. Blue-bearded bee-eaters apparently breed orily in solitary pairs (Fry 1984, Fry and Fry 1992). Observations described here further confirm these patterns in these species. Significance of these data in comparative studies The frequency and importance of extra-pair copulation and intra-specific brood parasitism in the little green and blue-tailed bee-eaters await more detailed study. However, the behavioral parallels with other bee-eaters are striking. If these behaviors prove to be common, then they represent both an increased cost for breeders of colonial breeding and a potential alternative reproductive strategy for breeders and helpers (although rarely do helper males take part in extra-pair copulations in the white-fronted bee-eater, Emlen and Wrege 1986). Documenting the existence of these behaviors in both loose and dense colonies and with significant differences in colony sizes suggests one of two evolutionary 81 patterns: adaptive flexibility in a variety of sodal conditions or phylogenetic inertia (latter possibility, sensu McKitrick 1993). If the behaviors serve as viable alternative reproductive tactics, then they should increase the lifetime reproductive success of certain individuals, and selection would work to maintain the behaviors in the population. However, given that these species are closely related, these behaviors in certain species may have little if any effect on individual fitness and may represent a holdover of past selection. The lack of synchronous breeding in species such as the blue-tailed bee-eater may seriously reduce the effectiveness of extra-pair copulation and intraspecific brood parasitism. If there is no cost to the behaviors, particularly for helpers or "interrupted" breeders (e.g. whose egg-laying in their own nest is disrupted) who may be making the "best of a bad job", the behaviors will not be purged. In the absence of selection, the status quo remains for a period of time. It is now known that 13 of the 26 species of bee-eaters breed cooperatively. We also know more about the extent of geographic variation in breeding behavior in the little green bee-eater. In both Thailand and India this species breeds cooperatively in loose colonies. Helpers in both populations arrive typically after incubation has begun, deliver food to nestlings, and may provide increased protection against nest predators. However, the number of helpers per nest does differ between populations. In India only a single helper is seen at cooperative nests, while in Thailand nests may have multiple helpers. These basic behavioral and natural history data have allowed me to conduct more complete comparative studies presented in the next chapter. These phylogenetically explicit comparative studies are essential for an imderstanding 82 of the patterns of behavioral evolution in this group and the ecological forces that have molded these patterns. 83

CHAPTER FOUR- BEHAVIORA.L MALLEABILITY VERSUS

PHYLOGENETIC INERTLV AND COOPERATIVE BREEDING IN BEE-

EATERS. The study of cooperative breeding is one of the most active areas of avian behavioral research (Emlen 1984,1991; Brown 1987; Heinsohn et al. 1990; Stacey & Koenig 1990). Cooperative breeding (CB) is a catch-all term that includes a number of different social and breeding systems that evolved independently in a number of bird lineages (Brown 1987, Edwards & Naeem 1993, Ligon 1993, Burt unpublished data). All of these breeding systems do, however, share one intriguing behavioral characteristic: certain individuals (helpers) provide some degree of parental care to young that are not their own. Different researchers have taken a variety of approaches in attempting to explain this seemingly altruistic behavior. These different approaches have led to different, and in many cases complementary, explanations for the existence of CB. These different explanations, referred to as different levels of analysis by other authors (Tinbergen 1963, Sherman 1988), can be placed in four categories: (1) functional consequences or selective advantages, (2) proximate mechaiiisms or physiological requirements, (3) ontogeny of parental care behaviors, or (4) historical origin and transition patterns. The first category includes the majority of CB studies in the literature. Using this approach, researchers examine how the fitness of individuals that help at some time in their life compares to the fitness of individuals that instead attempt alternative tactics such as immediate breeding or floating among territories or in marginal habitats. A concept fundamental to many of these studies is the importance of ecological cor\straints such as territory availability 84

for young birds entering breeding status and how these territories vary in quality. These environmental factors have been shown on theoretical and empirical grounds to be important for maintaining CB in the current envirorunent of populations of several species (Emlen 1982a, Woolfenden and Fitzpatrick 1984, Stacey & Koenig 1990, Koerug et al. 1992). Examples of explanations of CB due to proximate mechanisms or physiological requirements (the second category) include studies of hormonal control of parental care behavior (Vleck et al. 1991) and reproductive inhibition (Mays et al. 1991, Schoech et al. 1991, Poiani & Fletcher 1994) of helpers. An example of the third category includes Jamieson's (1989) suggestion that at least the provisioning of young, a major component of helping behavior, is a stimulus response by extranumerary adults to the sound of begging yoimg in the territory. Allofeeding by helpers is simply an early ontogenetic shift to a behavior that will be needed by these individuals when they become breeders. The final approach uses the observed diversity of breeding behaviors among extant taxa and a phylogenetic hypothesis of the group to infer evolutionary patterns of origin and subsequent change in breeding systems. Historical approaches to the study of cooperative breeding have shown that CB has evolved early in the diversification of many bird lineages, it is not evenly represented across genera of passeriform birds, and that the ecological and social conditions that led to the evolution of the behavior in a common ancestor may be qmte different from the conditions in many of the descendant species (Russell 1989, Peterson and Burt 1992, Edwards & Naeem 1993, Ligon 1993). In this study I use a phylogenetic approach, in combination with information from the other levels of analysis, to examine the evolution of CB in 85 the bee-eaters (Family Meropidae). Bee-eaters are a group of 26 species of brightly colored coraciiform birds. They are distributed throughout the paleotropics and southern Eurasia and display varying degrees of diversity in certain behavioral and ecological traits with important theoretical ties to the adaptive significance and evolution of CB (Fry 1984, Fry and Fry 1992). Examples of such traits that are potentially relevant in regard to the adaptive nature of cooperative breeding include: social system, nesting requirements, habitat utilization, migratory behavior, and foraging behavior. Below, I briefly discuss the importance of each of these traits. The adaptive basis of CB varies across social systems due to the different ecological and behavioral contexts found in each system. Breeding habitats probably are not saturated in most non-territorial, colonial species, but may be limiting in solitary breeders (Brown 1987). Therefore, if habitat saturation is an important determinant of CB in this group, solitary species should breed cooperatively more often than colonial species. Alternatively, the importance of kin selection and reciprocity to the evolution of CB systems is more likely in colorual species due to the more frequent and repeated interactioris individuals have with kin and non-kin (Trivers 1971, Brown 1987; Emlen and Wrege 1988, 1989; Siegel-Causey and Kharitonov 1990; Lessells et al. 1994). Cooperative breeding and coloniality may also be linked by the increased potential for parental harassment due to the close spacing of nests. In this behavior, breeders disrupt the breeding attempts of their progeny to increase the chance that these latter individuals will become helpers at their parents' nest (Emlen 1982b, Emlen and Wrege 1992). 86

One major positive effect helpers serve in many CB species is the reduction of predation on nests (Woolfenden and Fitzpatrick 1984, Austad and Rabenold 1985, Dawson and Marman 1991, Emlen 1991, Sridhar and Karanth 1993, Komdevir 1994, Webber and Brown 1994, Qiapter 3). The choice of nest placement may also lead to different levels of predation (Fry 1984). All coraciiform birds are cavity nesters (Ligon 1993); however, in bee-eaters considerable variation occurs among and within species concerning which specific nesting substrates are used. All bee-eaters dig cavities in the ground; however, the substrate can range from flat ground to tall cliff faces. Nests on flat ground and low banks are more accessible to reptilian and mammalian predators than are cliff nests (Fry 1984). CUff-nesting species such as the red-throated bee- eater, M. bullocki, and the white-fronted bee-eater, M. bullockoides, have relatively few nests depredated (<10 % nests; Fry 1984, Emlen and Wrege 1991, Wrege and Emlen 1991). One might expect species nesting in flat groimd or low banks, which are more prone to predation, to have CB systems more commonly than cliff-nesting species. Habitat utilization patterns relate to CB systems in several ways. First, as mentioned above, ecological constraints play a key role in the maintenance of CB in many species. In many CB species, constraints are related to habitat-use specialization and saturation of breeding habitats (Brown 1987, Koenig et al. 1992, Burt 1996). Species restricted to using one habitat type are expected to be CB more often than habitat generalists. In this study, the number of habitat types a species utilizes is used as an index to its habitat specialist/generalist nature. A second, related point concerns the predictability and seasonality of breeding habitats relative to CB. Stable, predictable environments may result in 87 demographic conditions (long-lived birds, high fecundity, and a buildup of potential breeders) that lead to habitat saturation in sedentary, territorial species (Brown 1974, Emlen 1982a). Under similar reasoning, CB may be more common in areas with low seasonality in food supply (Ford et al. 1988). However, CB also may be associated with sigriificant seasonality in food resources, as long as resoxirces are predictable during the breeding season (Du Plessis et al. 1995). Arid areas in Africa and Australia frequently have unpredictable rainfall with resultant tmpredictability in food resources. Resultant demographic variation among years is great and habitat saturation is not at issue. Instead, increased costs of reproduction in harsh years, especially for young birds, are thought to influence their choice in becoming helpers by making success at independent breeding unlikely (Fry 1972,1977; Dow 1980; Ford et al. 1988; Emlen 1982a; Emlen and Wrege 1991; Wrege and Emlen 1991). Species in which environmental unpredictability plays a key role in their CB systems are also predicted to show more variation across years in the percentage of groups with helpers (Emlen 1982a, Du Plessis et al. 1995). Most CB species are non-migratory and permanently territorial (Brown 1974, Brown 1987). These traits are likely to lead to habitat saturation due to the continuous occupation of available breeding habitats by breeders with the higher survivorships typical of non-migratory species. Finally, foraging behavior and diet may also be related to CB. In species in which young must leam complex foraging skills, yoimg birds may not have the necessary foraging skills to breed successfully, helping to explain delayed breeding in these individuals ("skills hypothesis": Brown 1987, Heinshohn 1991). Complexity of foraging skills may be related to the degree of diet specialization. 88

Bee-eaters are insectivores, with many species, as their name implies, specializing on Hymenoptera. Complex foraging skills can include the detection, acquisition and maruptdation of food. Two main foraging modes are foimd within the family relative to prey detection and acquisition: short distance hawking from open perches, or continuous flight/long sally foraging. Hawking species return to a perch after capturing hymenopteran prey and proceed to alternately beat the head of the prey against the perch and rub the abdomen of the stinging insect on the perch. The latter action results in ejection of venom from the prey and occasionally the loss of the stinger. Species that consume prey without returning to the perch are thought to maiupulate prey in their bill somehow while in flight to achieve the same detoxification results, although exactly how they do so is unknown (Fry 1984). Both foraging modes are examined in this study for correlated changes in breeding systems in the light of the potential differences in skills associated with each mode and the degree of diet specialization of each species. In this study I use a number of likely phylogenetic hj^otheses for the group to examine the possible patten\s of evolution in breeding systems in the bee-eaters. I then look for correlated evolutionary patterns in the behavioral and ecological fraits mentioned above. I conclude that cooperative breeding evolved early in the diversification of the bee-eaters. Additionally, CB has been resistant to reversals to the non-CB state, despite coiisiderable evolutionary change in the behavioral and ecological traits in which correlated changes were expected a priori. I then examine these findings with regard to the possibilities of behavioral malleability and phylogenetic inertia relative to CB. 89

METHODS I determined behavioral and ecological character states using the literature and personal observations of bee-eaters in Kenya and Thailand. Citations for data on each character for each species are listed in Appendix C. MacClade (version 3.0, Maddison and Maddison 1992) was used to reconstruct character evolution patterns using maximum parsimony. Characters analyzed are breeding system (CB and non-CB) and five traits which may be evolutionarily related to breeding systems: social system, nesting requirements, habitat utilization, migratory behavior, and foraging behavior. Character states for the diet specialization character were ordered. Other character were imordered. The phylogenetic hypotheses serving as the basis for these comparative studies were taken from the maximum parsimony analyses of plumage characters in Chapter two. The phylogenetic analyses in Chapter Two provided several possible phylogenies for the bee-eaters. These possible trees break down into two sets, each from the primary analyses in Chapter Two, and represent the best justified phylogenetic hypotheses. First, a set of six equally likely, completely resolved phylogenies was used. This set was derived from three separate analyses where characters were structiired with character specific step matrices. Two analyses used species as terminal taxa with characters weighted one or all characters scaled to have standardized weights. The third analysis used races as terminal taxa with characters scaled to have standardized weights (Chapter Two). This set of six fully-resolved trees allows more detailed examination of character evolution patterns and the robustness of patterns among trees. Second, a strict consensus tree from one phylogenetic analysis that produced a number of most- 90 parsimonioias trees is used as a conservative estimate of each character evolution pattern. This consensus tree is the result of an analysis with races used as terminal taxa and in which characters were structured with character specific step matrices and weighted one (Chapter Two). This latter tree is referred to as the "conservative tree" in subsequent sections. These two sets of trees were used to examine each of the character evolution patterns below. The nature of many of the characters examined here presented three difficulties of which the reader should be aware. First, data were xmavailable, or of varjdng quality, for some behavioral and ecological characters for some bee- eater species. Because of this problem, I chose to analyze the evolution of breeding systems (cooperative or non-cooperative breeding) at three levels of "data confidence". One level of confidence allowed designation of breeding system for species only when researchers had doamiented three or more individuals (i.e. at least one helper) bringing food to a nest or had spent cor\siderable time observing nests of apparently non-cooperative species with no evidence of helper activity (subsequently referred to as the "cautious breeding system coding". Figure 4.1). A second level allows consideration of additional species as cooperative breeders given evidence of three or more adult birds in close association around a single nest hole during the breeding season (subsequently referred to as the "moderate breeding system coding". Figure 4.2). The final, and most liberal, character codings for the breeding system character designate species as non-cooperative when limited observations of nests have shown no evidence of helper activity in certain species (subsequently referred to as the "liberal breeding system coding". Figure 4.3). This latter categorization is the most prone to error due to the low percentage of nests with helpers in most 91 cooperative bee-eater species. Consideration of breeding system evolutionary patterns using each character coding scheme, however, does allow examination of the robustness of conclusions given the level of vmcertainty in the breeding systems of certain species. Note that when data for certain species are lacking, evolutionary patterns are "filled in", given available data, by MacClade using parsimony. The second difficulty, regarding the nature of the characters examined, concerns the necessity to make generalizations in categorizing character states. For example, fitting species into breeding habitat categories requires obvious generalizations. This problem is diminished by attempting to find common descriptions and contexts of use for each character. For example, when information on each species could be gathered from a single source in which terminology was corisistent across species, this source was preferred. A third difficulty in this study concerns the question of geographic variation. For most species, breeding systems and related behaviors have not been studied in different parts of their range and the extent of behavioral geographic variation is unknown. However, a few species studied in different areas have shown little geographic variation in basic components of their breeding systems (see discussion). If geographic variation in a trait is known for certain species, the species was coded as polymorphic. The concentrated-changes test option of MacClade was used to examine whether changes in breeding system states are concentrated on branches corresponding to changes in other ecological and behavioral traits (Maddison 1990). The test, as used here, examines the probability that a given number of gains and losses of CB (the dependent character) are concentrated on branches 92 distinguished by change in another ecological or behavioral trait (the independent character). The null model for the test is that changes are spread randomly among branches of the tree. This test is currently restricted to binary data and was therefore not used in all behavioral and ecological comparisons. Decay-index analyses were conducted to assess the degrees of confidence in the relative placement of one particular clade {Merops pusillus - M. variegatus) in each tree. These analyses were conducted using the same analysis options and search criteria as those used to derive the original phylogenies in Qiapter Two. However, a topological constraint appropriate to the question at hand was maintained in each search. Comparisons of tree lengths between original and constrained searches point to the degree of support of the clade's original placement. 93

RESULTS Patterns of evolution in breeding systems Regardless of which of the likely phytogenies or three breeding system codings is used, the emergent pattern is an early origin for CB in the family. When the cautious breeding system coding is used, the conservative tree and four of the six fully-resolved trees show one basic pattern (Figure 4.1). The evolution of CB occurred either after the divergence of Nyctyomis from the remaining species in the family or possibly was the primitive state for the family.

These trees also show a reversal to the non-cooperative state once within Merops and possibly one time at the base of the Nyctyomis lineage. Referring to additional outgroups could possibly help resolve this problem concerning the ambiguity of the basal state. However, relationships among the families of the Order Coraciiformes are imclear and times of divergence among the families are quite long (Cracraft 1981, Sibley and Ahlquist 1990). Additionally, there is a near even split between families in this order with and without CB representatives

(see Chapter One, Table 1.1). Note that the polymorphic state of M. piisillus requires at least one additional change between breeding system states within that lineage in all the trees examined in this study. The two remaining fully- resolved trees show more uncertainty in the evolutionary patterr\s of breeding systems. Recoristruction of deep branches as equivocal indicates seven possible patterns of evolution. One possibility in these trees is the same as one pattern seen in the other four resolved trees: the basal state of the family was CB with reversals to non-CB once in Nyctyomis and once within Merops. On the other side of the spectnmi, non-CB may be primitive for the family with evolution of CB occurring once in M. oreobates and once at the base of the M. himndineiis - M. 94

bullocki clade. Several intermediate reconstructions are also equally parsimonious. One question related to the confidence in the patterns of breeding system evolution observed above relates to the placement of the non-CB M. pusilliis - M. variegatus clade. If this clade is forced to the base of the Merops clade, reconstructions would show a single origin of CB with no reversals to Non-CB. Decay index analyses of trees derived from species data matrices (see Chapter Two) show that trees with this topology are 1.3% (761 versus 771 steps and 83087 versus 84198 steps) longer than the most parsimonious trees. Exact interpretation of the significance of these differences is unclear. However, the support for trees with this topological constraint do not make breeding system patterns obviously doubtful. Breeding System, Confirmed I I Non-cooperative Breeding •• Cooperative Breeding r*T

^|1 . 52 II ^ „ I 5.1,.a js2 -js».^C g S !!-Sa 5 8 = :a = £S ?•§ S § pa 1.5.6 tS iiiiiiiiiiii-ii-riiiiiiiiiii • • ••••• • ••• • •• ••

FIGURE 4.1. Cautious coding reconstructions of breeding sjretem evolutionary patterns. Character codings consider states from species only with confirmed breeding systems. (A) representative fully-resolved trees each showing one of two basic patterns. (B) Conservative tree recor^truction. 96

Coding an additional two species as CB (moderate breeding system coding) reduces the number of likely patterns. The same basic pattem is seen regardless of which tree is examined (Figure 4.2). This pattem is the same as that seen in four of six fully-resolved tree and the conservative tree in the above analysis. Using the liberal breeding system coding, which may possibly err towards a bias for non-CB species, shows similar patterns to those above but with more reversals to non-CB (Figure 4.3). Each of the six fully-resolved trees shows the familiar patterns of either the evolution of CB after the divergence of Nyctyomis from the remaining lineages or for CB to be the primitive state for the family. Each of these six trees then shows evidence for three independent reversals to non-CB in the Merops clade and possibly at the base of Nyctyomis. The trichotomy at the base of the Merops clade in the conservative tree makes the state of breeding systems uncertain from the base of the tree through the branch leading towards the M. boehmi - M. leschenaulti clade. Cooperative breeding may have been primitive for the family, evolved at the base of the Nyctyomis - Merops split, or evolved at the base of the major Merops subclade and again in the M. viridis - M. leschenaulti clade. This tree also shows two to four reversals to non- CB. In summary, CB either evolved early or was the primitive state for the family. Subsequent reversal to non-CB occurred in one to three lineages, excluding the intraspecific breeding system variation in the little bee-eater. 97

tt

.j2 .*'1

•• ••I

Breeding System, Estimated Non-cooperative Breeding Cooperative Breeding Polymorphic Equivocal

FIGURE 4.2. Moderate coding recoristructions of breeding system evolutionary patterns. Two additional species are considered(marked with arrows). Reconstructions show the same basic pattern on each tree, except for one deep branch that is equivocal when using the conservative tree. Breeding System, Err to non-CB I I Non2. S ^ w sja .V?o .s 5 C S .2 V - Wi ? = e= . P s 11 gj-i C S ScS.* -^'^ e a£ 5 c "3 5 '5 ^fi Is. Sspcp^^appiS.£.S.IS.S.S.S.S.S." i-i-l-lp p p p IIIS.& a. ^555S55S5S5S 55s:

FIGURE 4.3. Liberal coding reconstructions of breeding system evolutionary patterns. Two additional non-CB species are coded (marked with arrows). (A) Reconstructions show the pattern on each full-resolved tree is the same. (B) Reconstruction on the conservative tree shows a nvunber of equally likely patterris. 99

Correspondence of other traits to breeding system evolution

Social systems- Bee-eaters show a range of social systems, including solitary, territorial breeding pairs; loose aggregations of breeding groups dispersed over a few acres (loose colonies); and densely packed colorual nests. The reconstruction of social systems on each tree shows two major patterns (Figure 4.4). The majority of trees show solitary nesting on deep branches with subsequent independent evolution of loose and dense coloniaiity and then one reversal to solitary breeding. These reconstructions require 17 evolutionary steps, with the majority of changes occurring within polymorphic taxa. In fact, many species run the gamut of social systems, suggesting considerable flexibility of this trait. The other major pattern seen on two trees required 18 steps. This pattern shows considerable uncertainty on several branches relative to the timing of evolution of solitary and dense colonial nesting. At first glance, bee-eater species appear not to shift between CB and non-CB systems coincident with changes in social systems in the family. However, if M. revoilii and M. boehmi are non-CB (liberal breeding system coding), losses of CB appear correlated with lineages that show solitary social systems only. Characters were recoded as either "solitary only" or "other social system" and concentrated-changes tests were conducted on the resiiltant reconstructions. Results of these tests show that given either zero gair\s or one gain and four or three losses in CB, the probability of observing all of these transitions by chance in "solitary only" lineages ranges from 0.008 - 0.02. Given a more conservative recoristruction of CB (moderate breeding system coding), with either zero gairis or one gain and one or two losses in CB, the probability of observing transitions in "solitary only" lineages ranges from 0.083 - 0.19. Opposing this possible "non-CB /solitary breeding" trend is the likely evolutionary origin of CB in a solitary bee-eater lineage. 101

— C! § J I .t c g 1 Q 8 S s 5 8^8 "S I |:g I ^ 5 g:.s •§ ."1 ll 11 c E gj-fe s 5 a jj - •i:a -SiS i- 5 X i6. o- s S i- S. 8. l-i-i-i- 1 •g J r s s g e 2 1

••••••••I ••••••••••••

Sodal Structure Solitary or a few nests LooselyColonial

Colonial •• •• J I c: I to Equivocal il i -a "S g ij S 2 wl

FIGURE 4.4. Reconstructions of evolutionary patterns of social systems on full-resolved trees that represent the two major patterns observed. Pattern on the conservative tree is very similar to that of the top tree. Taxa with closed boxes have confirmed non-cooperative populations. Taxa in dash- line boxes have limited evidence suggesting non-cooperative breeding. 102

Nesting substrate requirements- The reconstruction of nesting substrates on each tree shows two major patterns, each showing a considerable number of eqmvocal branches (Figiire 4.5). Most species are pol5anorphic for this trait and many are known to nest in all three substrate t5rpes (i.e., cliff banks, low banks, and flat ground). Resulting reconstructions show either 28 or 29 evolutionary steps in this apparently very plastic trait. The considerable ecological diversity in this trait has apparently had little impact on transitions between breeding systems in bee-eaters. 103

Nesting Substrate I I Cliff banks Low banks •• Flat ground L_J Equivocal

S « I I 2 V) c J I ^ S "i .2 I I w .a| jl V) S A I' d I « 1- = •§ S g ^ 'f? s :J' S •Sik. iiiil ,sg S-fe ^-5 Sc :§g Js S,= e lllllllls &-== 5 U. •a. «l g e 60 S|-' W) tft ll a. ^ S! 1 ' s- s e s S^ ^e «j|811 I ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^1 I s^ stj i 1 55 5 5 51; 5555555555551 '5 5 2 2 ••••••••••••••••••••••••••

FIGURE 4.5. Reconstructions of evolutionary pattern of nesting substrate requirements on full-resolved trees representing the two major patterns observed. Pattern on the conservative tree is very similar to that of the top tree. Taxa with closed boxes have confirmed non-cooperative populations. Taxa in dash-line boxes have limited evidence suggesting non-cooperative breeding. 104

Habitat utilization- Due to the extreme degree of polymorphism seen in this character, the resviltant reconstructions should be considered in a skeptical light. However, if we accept them as a possibility for the moment, we see two evolutionary patterns emerge (Figure 4.6). The first pattern shows the root of the tree as either forest or open savanna woodland with subsequent divergence of these two habitats into separate lineages at the divergence of Nyctyomis from

Merops. The second pattern is similar to the first, differing only by the invasion of open savanna woodland by a large Merops subclade occurring after divergence from the forest lineages. Therefore, despite a great deal of intraspecific variation in habitat utilization, two habitat types may have had central roles in bee-eater evolution. The first evolutionary pattern of habitat utilization shows concordant origin of CB and invasion of open savarma woodland in the majority of breeding system reconstructions (e.g. Figures 4.1A top, 4.2,4.3A). However, reversals to non-CB do not correspond to changes in habitat utilization, except possibly in the

M. revoilii lineage. For example, invasion of forest habitat is not related to changes in breeding systems. The second habitat utilization pattern does not show any correspondence to the evolution of CB. In fact, most reconstructions of CB show this behavior evolving within forest lineages, before the invasion of savarma woodland habitats, but persisting in the majority of lineages within this latter habitat. 105

§ I I „ I I ._ e ^w» !§^ -5? S 1 Qw> ^411-5 e ^ -s E i l-fe i g o^ ^ '•3 s "s .S ^ "I. S. - S s milW wlwi k. i-k. ti.t lllllll 1 •§ III!^ ^ V ^ V 2 2. S 52'5 S 52 5555555

•!••• ••••••••I ••••••••• B IQ Q Q Q Q Q QI

Environment Forest Grassland K/V/Vi Marshes •• Cultivated areas Open savanna woodland Semi-desert Equivocal § -• VI I I. fc c i| I - s .2 tft !5 «A ^ e -c £ « I sf .c^ .L9 ' s S 5 p cl g c bs c ^ 111 §j»a fc o S .fij o 5 e c ^«A lllll|llll^ ^ ^ gj|_d _v ^ _sj — III 11 555553555555555555 55

_ _iamumaammmmmmuaaa

FIGURE 4.6. Reconstructions of evolutionary pattern of habitat utilization on full-resolved trees representing the two major patterns observed. Pattern on the conservative tree is very similar to that of the top tree. Taxa with closed boxes have confirmed non-cooperative populations. Taxa in dash-Kne boxes have limited evidence suggesting non-cooperative breeding. 106

Wing shape, migration, and foraging mode- The next two behavioral characters must be examined together because they may be functionally related through wing shape. In fact, we have a "chicken and the egg" type question. Bee-eater wing shapes can be divided roughly into two wing shapes: long and pointed, or short and roimded. Long, pointed wings are generally characteristic of long-distance migratory species, while sedentary and short-distance migrants are more likely to have short, rounded wings. However, irisectivores that forage in continuous flight or with long sallies also have long, pointed wings, while short-range, hawking species have more maneuverable, short, rounded wings. The question then is: did wing shape evolve first in response to migration or foraging technique? Secondly, were different states of the secondary behavior then preadaptations due to selection on the first behavior? Understanding this issue may seem off the point relative to the issue of the evolution of breeding systems; however, each behavior may effect the evolution of CB. Therefore, it is important to understand whether migration behavior has influenced the evolution of foraging behavior or vice versa, because with an answer, we can judge the relative influence that transitions in each behavior might have on the evolution of CB. Reconstructions show that bee-eater wings were short and roimded primitively with two possible patterns of subsequent evolution (Figure 4.7). One pattern shows the evolution of long, pointed wings within Merops after the divergence of M. orientalis. Subsequent reversal to short, round wings occurred three to four times with additional evolution of long, pointed wings possibly once. The second pattern has long pointed wings evolving within one Merops subclade with reversal to short, roimded wings occurring twice. The 107 reconstruction on the conservative tree shows cor\siderable uncertainty. Short, round wings may have evolved once, with a number of independent changes to long, pointed wings. Alternatively, long and pointed wings may have arisen once with several reversals to short, round wings. Patterns of evolution in foraging mode show a remarkable convergence with wing shape evolution (Figure 4.8). In fact, the only deviation from expected patterns (i.e. long, pointed wings and continuous flight or long sally foraging, versus short, rotmd wings and short-range hawking) is found in M. leschenaulti which has relatively long, pointed wings and frequently forages using both techniques. Results from concentrated-changes tests show probabilities of less than or equal to 0.0001 that foraging mode and wing shape evolution would be so highly congruent by chance. Given the close correspondence of transitions between states in both wing shape and foraging mode, it is clear that a close functional relationship exists between these two traits. Reconstructions of migration behavior are more variable among phylogenies examined and show varying degrees of correspondence with transitions in wing shape (Figure 4.9). If equivocal branches are assiimed to be migratory, concentrated-changes test probabilities for correlated evolution of these two traits are 0.0027,0.081, and 0.13 for different reconstructions. If equivocal branches are assumed to be sedentary, concentrated-changes test probabilities are 0.65,0.80,0.85 for different reconstructions, showing little or no correspondence between wing shape and migration behavior evolution. Certain patterns here show species or groups of species that revert to sedentary behavior but still retain long, pointed wings (Figure 4.9A & B). Other reconstructions show migration behavior evolving at least two times in the context of long. pointed wings, but other lineages that also have long, pointed wings remain sedentary (Figure 4.9C & D). Patterns reconstructed on the conservative tree similarly intermediate in the degree of correlation of these two traits. 109

i-s i fesHS 6BJSSgg5S.r-S. ®"3|£ -B 4 rllllilllllllli-llllll :s I i ••••••DBBBBr---"-'

Wing Shape I I Rounded L43ng pointed t—^ Equivocal

c I 5 .§1 L 1 i i ^ «| ! I iisi B .fi I - i s 61J 11 l|i I i 11.1.1^ I III!

FIGURE 4.7. Reconstructions of wing shape evolution on (A) full-resolved trees representing the two major patterns observed, and (B) the conservative tree. Taxa with closed boxes have confirmed non-cooperative populations. Taxa in dash-line boxes have limited evidence suggesting non-cooperative breeding. 110

llll

ForagingModc I ' Hawidng •• Long sallies and continuous flight Polymorphic \ I Equivocal

FIGURE 4.8. Reconstructions of foraging mode evolution on (A) full-resolved trees representing the two major patterns observed, and (B) the conservative tree. Qiaracter evolution patterns for this trait are very similar to those for wing shape. Taxa with closed boxes have confirmed non-cooperative populations. Taxa in dash- line boxes have limited evidence suggesting non-cooperative breeding. Migration • I Sedentary •• Migratory CZQ Polymorphic Equivocal

•H "p ^ SI I'S ^ 2 A .1 c w *"1 1^ HI'S aP ^ IKI'C I i - •§ u. -^1 I "1 IIII 11 i11 i111 1 It1 i ^ il l-i i "I 11111ll I i i ll S i-i[i I A^llk H^^^^ ilk5^^ i^^i^ •••••••••••••••••••••••••a •••••••••••••••••nBBBBngnn

FIGURE 4.9. Reconstructions of migratory evolution on (A - C) full-resolved trees representing the four patterns, and (D) the conservative tree. Taxa with dosed boxes have confirmed non-cooperative populations. Taxa in dash-line boxes have limited evidence suggesting non-cooperative breeding. 112

The origin of long, pointed wings is coincident with continuoiis flight and long sally foraging methods in all reconstructions and shows congruence with the evolution of migration in a majority of reconstructions. Patterns of subsequent evolution show a clear functional tie between wing shape and foraging mode. Subsequent evolutionary patterns between wing shape and migration behavior are highly variable across reconstructions and the closeness of possible correlated evolutionary changes is uncertain. The relative degree to which each behavior influences the other through constrained wing shape evolution is, therefore, unclear. Regardless of the uncertainty in the above pattenis, neither migratory behavior nor foraging mode have a great impact on transitions between CB and non-CB systems in most reconstructions. However, if equivocal branches are assumed to represent continuous flight foraging, either zero gains or one gain and either three or four losses of CB correspond to zero gains and two losses in long sally and continuous flight foraging. The probability of such concerted evolution occurring by chance is 0.02. In other words, this recoristruction suggests a tie between the evolution of non-CB and short-distance hawking. All other reconstructioris of foraging mode show no significant correspondence of evolutionary patterns with breeding system evolution (range of probabilities: 0.21 - 0.85). None of the reconstructions of migration behavior show a significant correspondence to the evolution of breeding system patterns (range of probabilities: 0.21 - 0.89). In most reconstructions of CB, this behavior did not evolved on branches distinguished by shifts in migratory or foraging behaviors. Additionally, subsequent transitioris between migration and foraging behaviors do not regularly correspond to changes in breeding system states. 113

Diet specialization- Figure 4.10 shows the evolutionary plasticity in degree of diet specialization in bee-eaters. Most species show a majority of Hymenoptera in the diet. However, percentages change frequently across lineages. Additionally, some species show considerable geographic and seasonal range in the proportion of Hymenoptera in the diet. Temporal variation in diet may actually be quite common, especially as several species are known to take advantage of temporary swarms of insects (Fry 1984). Diet patterns show no consistent changes correlating with the evolution of breeding systems. 114

.-I I a i A

i-l

Proportion of Hymenoptera in Diet I 1 25-49% •• 50-74%(Majority) IBI 75%orinore u*»ti Polymorphic Equivocal I U cs .*2 si \£ .s .2 -i al |1 s 1 S? J: •? «.i'~ 5 P -S "i-.a •s a "5 ^y-is is ct -5? ir ,v» (A 5 k. •a c lliii IWl: ^ oC s oc c ® Si. wj ^ S TJ §, 11^ iJi-lgi ^ §, _k ^ ttit U ^ u ~ ^ WB-—-w 2J5 § 5 5 SiS,'5 555555555S 5.S'S^1 5 ID

FIGURE 4.10. Reconstructions of evolutionary patterns of diet specialization on full-resolved trees showing the observed diversity. Taxa with closed boxes have confirmed non-cooperative populations. Taxa in dash-line boxes have limited evidence suggesting non-cooperative breeding. 115

DISCUSSION The importance of historical studies of animal behavior has been recognized for some time:

"Every living organism is a system produced by a process with a past history, and every life phenomenon of an orgaiusm is fundamentally only open to understanding after rationalistic and causalistic retrospective tracing of the process of its phylogenetic emergence. Nowadays, this fact is self-evident to any biological thinker." (Lorenz 1954, see Lorenz 1970). Recent studies demonstrate that behavioral characters can retain significant levels of historical information and may not be as plastic as intuition might lead one to believe (de Queiroz and Wimberger 1993, Paterson et al. 1995, Wimberger and de Queiroz 1996). Consistent with this idea, I have shown here that CB is not prone to frequent evolutionary change despite considerable change in other behavioral and ecological traits. Cooperative breeding either evolved before the diversification within

Merops or was the primitive state for the family. If CB evolved within the family after divergence from Nyctyomis, reversals to the non-CB state occurred from one to three times, depending on which breeding system truly represents M. revoilii and M. boehmi. At least one additional transition also occurred within M. pusillus. If CB is primitive for the family, an additional reversal must have occurred at the base of the Nyctyomis lineage. Evolution of breeding systems relative to other traits Evidence is mixed regarding the correlated evolutionary patterns between breeding systems and ecological and behavioral traits that are hypothesized to determine the selectively advantageous basis of CB. Two traits show some correspondence of character state traiisitions with those in breeding systems. 116

One possible pattern is the loss of CB correlated with lineages that show only solitary social systems. This correlation is supported only if M. revoilii and

M. boehmi are assumed CB . Given a more conservative reconstruction of CB, this correlation is insignificant. However, assimiing for a moment that the correlation is real, what are the possible functional relationship between non-CB and solitary social systems versus CB and some form of colonial breeding? In most non- territorial, colonial species, the role of saturation of breeding habitats is minimal at best (Brown 1987). The fact that territorial, solitary species are the very ones that are typically non-CB in this group reinforces the idea that habitat saturation is not a relevant issue regarding CB in the bee-eaters. Coloniality can lead to an increased probability of seeing CB behavior for several reasons. Colonial individuals are in contact with more conspecifics and more frequently than are individuals of solitary species. Therefore, colonial living provides more opportunities for parents to harass their offspring into serving as helper. Parental harassment does appear to be an important tactic for recruiting helpers in White-fronted bee-eaters, M. bullockoides (Emlen 1982b, Emlen and Wrege 1992). Additionally, colonial living generally provides greater potential for an individual to serve as a helper at a relative's nest if its own nest attempt fails (Brown 1987). In fact, this pattern of rapid transition to helping after breeding failure is seen in white-fronted bee-eaters (Emlen and Wrege 1988,1989) and

European bee-eaters, M. apiaster (Lessells et al. 1994). European bee-eater brothers may even nest closer to each other than expected at random, which facilitates this transition (Lessells et al. 1994). One additional ecological trait that may show some correspondence with patterns of breeding system evolution is foraging mode. One reconstruction 117 shows possible ties between the evolution of non-CB and short-distance hawking, and long distance sallies/continuous flight foraging and CB. All bee- eater species must master a series of complex foraging skills including the detection, acquisition and manipulation of hymenopteran prey. In fact, young birds are highly dependent on adults for the first few weeks of their life (Fry 1984). Of the two foraging modes fovmd in the family, continuous flight and long sally foraging may be the most difficult to master. Species foraging in this mode typically detect prey at a greater distance and possibly leam a unique, yet undetermined, method of devenoming hymenopteran prey (Fry 1984). It is possible that one major fitness effect of helpers is increased fledgling survival during the resultant extended weaning period. Also, if yearlings are still unable to master foraging skills to the extent of allowing successful breeding, this shortcoming might explain one component of the adaptive nature of delayed breeding in CB species. However, this latter situation is tmlikely, as most bee- eaters appear to be as proficient as adults by seven to eight weeks of age (Fry 1984). All other reconstructioris of foraging mode show little or no correspondence to the evolution of breeding systems and the likelihood of any functional tie between the two is uncertain. Several other ecological and behavioral traits show no evidence of correlated evolution relative to breeding systems. Preference in nesting substrates has apparently little, if any, effect on transitions between breeding systems in bee-eaters. However, given that predation threats are greater in flat groimd and low-bank nesters, the antipredator role of helpers may still play a crucial role in maintaiiiing this behavior in these popvilatior\s (Fry 1984, Sridhar and Karanth 1993, Chapter 3). Other selective forces must then be invoked to 118

explain the persistence of CB in cliff-nesting species where predation rates are low (Fry 1984, Emlen and Wrege 1991, Wrege and Emlen 1991). Colonial species are often cliff nesters, and predation rates may be particularly low due to passive defense associated with colorual living (Siegel-Causey and Kharitonov 1990). Helpers in this situation would provide little added benefit. As discussed above, evidence from colonial and solitary social system correspondence to CB suggests that saturation of breeding habitats does not occur. Additional evidence to reject the habitat saturation hypothesis for CB evolution in bee-eaters comes from evolutionary patterns of habitat utilization and migration behavior. Cooperative breeding birds in which habitat saturation is key in maintaining the behavior are typically habitat specialists (Brown 1987, Koenig et al. 1992, Burt 1996). Most bee-eaters are habitat generalists, clearly showing corisiderable flexibility in habitat utilization patterns. Cooperative breeding might also be related to envirorunental variability and predictability (Fry 1972,1977; Brown 1974; Dow 1980; Emlen 1982a; Ford et al. 1988; Du Plessis et al. 1995). In fact, environmental uncertainty has been cited as an important factor in maintaining CB in white-fronted (Emlen et al. 1991, Wrege and Emlen

1991) and white-throated bee-eaters (M. albicollis, Dyer 1983). However, the range of habitats in which CB and non-CB species breed is so broad that examining differences in envirorunental variability and predictability carmot explain CB across all CB bee-eater species. Migration behavior might be related to breeding system evolution because permanently territorial species are more likely to show demographics leading to saturation of breeding habitats. However, bee-eaters do not show any hint of habitat saturation. In fact, reversals 119 to non-CB occur only in sedentary lineages. Otherwise, lineages appear to retain CB behavior regardless of whether they are migratory or sedentary. The final behavioral trait with potential ties to CB, diet specialization, shows no correlated evolutionary patterns. All bee-eaters are insectivores and all specialize, to different degrees, on Hymenoptera. The skills involved in manipulating such potentially dangerous prey and the length of time needed to leam such skills would appear to be a perfect test of the skills hypothesis (Brown 1987, Heinshohn 1991). The skills hypothesis states that if yoimg birds attempt to breed, they may not have adequate foraging skills to allow successful provisioning of their nestlings. These birds may therefore have higher lifetime reproductive success if they show early reproductive restraint. Unfortimately, the fact that all bee-eaters are specialists on Hymenoptera precludes the possibility of comparing shifts in this behavior relative to CB and non-CB species. One additional point suggests that diet specialization on Hymenoptera cannot by itself explain CB in bee-eaters. Jacamars (Aves: Galbvdidae) are the New World equivalent to bee-eaters in many ways. In particular, their diets are very similar. This convergence on such a specialized diet does not, however, reflect convergence in breeding systems, as all 18 species of jacamar are thought to breed non-cooperatively (Fry 1970). Behavioral malleability versus phylogenetic inertia The results of this study show that transitions between CB and non-CB systems do not match evolutionary changes in several traits proposed to be correlated with CB in other groups. These results can be explained in two ways. First, the lack of change in breeding system states may reflect the malleability of the basic CB system to fit a variety of social and ecological circumstances. The 120

fact that the diversity observed in each behavioral and ecological trait is not linked across traits may be important. For example, helpers may be crucial for successful nesting in populations nesting on flat ground due to the increased predation pressure these populations face (Sridhar and Karanth 1993, Chapter 3). Other populations may be relatively free from predation pressure, and the key role helpers play to increase nesting success may be an increased and more stable provisioning rate for nestlings (Emlen and Wrege 1991). In short, CB may have evolved in one set of ecological circumstances and the behavioral suite was then co-opted for use by species invading other environments. Phylogenetic inertia is a second possible explanation for why CB systems remain in lineages despite shifts in ecological situations that are expected on theoretical grounds to lead to reversal to non-CB. The role of phylogeny in determining contemporary behavioral states can be described in several ways, depending on our knowledge of the relevant underlying processes and the strength of these processes in molding evolutionary patterns. Phylogenetic constraints are defined as "any result or component of the phylogenetic history of a lineage that prevents an anticipated course of evolution in that lineage" (McKitrick 1993). The two key components in this definition concern the expectation of certain evolutionary patterns occurring, and mechanisms restricting these evolutionary changes. Phylogenetic inertia, on the other hand, is invoked in circumstances where an anticipated evolutionary course is not seen, but no mechanism, other than time, is restricting evolution. In this case, not enough time has passed for selection to operate and produce expected results. Of course, phylogenetic constraint can be mistaken for 121 phylogenetic inertia due to a lack of discovery of mechanisms preventing the development of expected patterns. Mechanisms that would prevent correspondence of evolutionary patterns between CB other traits examined here are not known. The failure to see expected patterns of evolution might therefore represent phylogenetic inertia. However, is it possible that some of these expected patterns are really red herrings? After all, cooperative breeding has evolved in the class Aves multiple times and the adaptive factors for the behavior do not have to be the same in all lineages. Predictions regarding the role of habitat utilization and migration behavior relative to habitat saturation may in fact be irrelevant in bee-eaters. Habitat saturation has never been cited as a probable determinant of CB in any bee-eater species. The concept of habitat saturation is however the most widespread explanation for CB in avian species (Brown 1987, Emlen 1991) and was therefore an obvious trait to examine in this study. Certaiivly, if a pattern became evident between CB and aspects related to habitat saturation, the point would be made to focus future research on identifying potential ecological constraints. Other traits failing to match predicted patterns of evolution do have evidence supporting their potential importance in maintaining CB in some bee- eater species. Helpers are important for defending nests against predators in the little green bee-eater, M. orientalis (Sridhar and Karanth 1993, Chapter 3). The role of environmental variability and unpredictability has been shown to influence the ability of young individuals to breed successfully in harsh years. In harsh years they have a better chance of increasing their inclusive fitness by helping close relatives by increasing the continuity and rate of nest provisioning (Emlen and Wrege 1991). In these cases, phylogenetic inertia may be relevant. 122

One example of possible phylogenetic constraint has been proposed for cooperative breeding for birds in general and may be pertinent here. Given that individuals delay dispersal and breeding for certain reasoi\s, they may serve as helpers regardless of the adaptive basis of the behavior. It has been proposed that the provisioriing of young is merely a stimulus response by individuals to the soimd of begging yoimg in the territory Jamieson and Craig 1987, Jamieson 1989). Allofeeding by helpers is simply an early ontogenetic shift to a behavior that will be needed when these individuals become breeders. Selection to prevent this provisiorung behavior before becoming a breeder may not be possible without negatively affecting this behavior later in life. If helping is due to such phylogenetic constraints, ecological studies might document adaptive reasons for delayed dispersal and delayed breeding, but fail to show how helpers increase reproductive success at others' nests. Need for detailed ecological data on all species At this point the reader may wonder why variables such as levels of predation or time to foraging competence are not directly examined for correlations with the evolution of CB. Data on these traits are currently available for so few species as to make these comparisor\s impossible. In fact, given adequate ecological data on each species, additional character evolution patterr\s should be examined. In particular, specific components of CB systems might provide more information regarding the relative roles that adaptation and phylogenetic inertia play in the evolution of CB in this group. Crucial comparative data include the degree of interspecific variation in sex of helpers, proportion of nest with helpers, number of helpers per nest, age of helpers, range and importance of helper duties, frequency of reciprocity between roles for 123 helpers and breeders, population sex ratio, and incidence of inhraspedfic brood parasitism and extra-pair copulation. The evolution of CB in bee-eaters probably involved the complex interaction of several ecological and behavioral traits. Data on specific components of CB such as those listed above will allow more detailed tracing of evolutionary patterns in CB. In fact, such tracings might reveal serial switches in different adaptive explanations for CB in different lineages living in different ecological circumstances. However, if evolutionary patterns in subcomponents of CB behavior prove to be resistant to change, then phylogenetic inertia must be considered a serious possibility. Understanding the degree of intraspecific geographic variation in breeding systems, ecology and associated behaviors is also crucial to further research on the evolution of CB in bee-eaters. Few species have data available on breeding biology in different parts of their range. The littie green bee-eater shows few differences in major components of its breeding biology in India (Sridhar and Karanth 1993) and in Thailand (Chapter 3). On the other hand, littie bee-eaters are CB in Somalia, Tanzania, and Botswana but probably non-CB in Nigeria and Malawi (Douthwaite 1986). Unfortunately, no additional data are available for this species in regard to geographic variation in major breeding behavior components. Due to the apparentiy widespread occurrence of the behavior in the family, I predict that future studies will not show much geographic variation in CB systems. Where should we place priorities for collecting additional data on bee- eater breeding biology? Clearly, conducting detailed behavioral ecology studies on each species and each geographically distinct race would take decades to complete. I suggest four levels of research priorities. First, the basic breeding 124 biology of species with imknown or uncertain breeding systems should be documented. Results from this research may then require a reanalysis of factors examined in this study. Resultant patterns from this reanalysis may then suggest a subset of obvious sister-groups (one CB lineage and one non-CB lineage) for which more detailed ecological and behavioral data would be particularly informative. The third priority is to survey geographic variation in the basic breeding biology of certain widespread bee-eaters that have significant variation in other ecological traits. Finally, if variation in CB behavior is found, detailed ecological and population-level historical studies may then be reqviired. Sununary Cooperative breeding is either primitive in bee-eaters or evolved early in the diversification of the family. Subsequent reversal to non-CB occurred in one to three lineages, excluding the intraspecific breeding system variation in the little bee-eater {Merops pusillus). Transitions in breeding system states are not generally correlated with the evolutionary transitions in nesting requirements, habitat utilization, migratory behavior, or diet. Limited evidence suggests correlated evolution both between CB and continuous flight/long sally foraging, and between non-CB and short-distance hawking. Extended time required to master foraging skills may partially explain helping in continuous flight/long sally foraging species. Stronger evidence supports correlated evolutionary patterns between non-cooperative breeding and lineages that only breed in solitary social systems. Patterns observed in this study, however, do not support any single hypothesis for the adaptive basis of CB across the entire family. Evidence from patterns of social system evolution do support the probable importance of kin selection in several lineages. 125

Lack of change in breeding systems, given the great diversity of ecological and behavioral circiimstances, means one of two things. Cooperative breeding may be a malleable system, selectively advantageous in a variety of ecological conditions. Alternatively, CB may represent an example of phylogenetic inertia in this family of birds. Disentangling these two explanations will require more detailed comparative study. These studies cannot be conducted until both broader and more detailed ecological studies are conducted. 126

CHAPTER FIVE- PHYLOGENETIC STRUCTURE INDICES: WHAT DO

THEY TELL US AND HOW SHOULD THEY BE USED? In recent years, biologists have become increasingly aware of the need to integrate phylogenetic information with comparative studies of biodiversity and evolutionary processes (Ridley 1983, Felsenstein 1985a, Donoghue 1989, Brooks and McLennan 1991, Harvey and Pagel 1991). Additionally, increased concerns have been raised regarding how to best assess the quality of data underlying phylogenetic hypotheses (Fitch 1984, Felsenstein 1985b, Hillis 1991, Huelsenbeck 1991). Behaviorists, developmental biologists, ecologists, and physiologists, as well as phylogenetic biologists, require an accurate index to the quality and degree of structure in phylogenetic data sets that form the basis of their comparative studies. A variety of indices are available that claim to assess structure in phylogenetic data, but the levels of confidence associated with different index values are poorly known. This paper examines the levels of confidence associated with values of several of the more readily available and frequently reported phylogenetic struchore indices: the consistency index (CI, Kluge and Farris 1969), retention index (RI, Farris 1989), rescaled consistency index (RC, Farris 1989), skew statistics of tree-length distributions (gi, Le Quesne 1989, Hillis 1991, Huelsenbeck 1991), and the nonparametric bootstrap (Felsenstein 1985c). The ensemble or tree CI, RI, and RC are calculated as: n ^WiMi Tree CI = ^WiSi 1=1 127

^ wiGi —^ WiSi

TreeRI= fl n WiGi - ^ WiMi 1=1 1=1

TreeRC= (Tree CIXTree RI) where wi is the weight applied to character i. Mi is the minimum number of steps that character i could possibly show on any tree, Sf is the number of steps reconstructed for character i for a given tree, and Gf is the greatest or maximum number of steps that character i could possibly show on any tree. The statistic is a measure of the skewness of a distribution and is defined as the third central moment divided by the cube of the standard deviation:

ns Where Tf is the tree length of tree i, n is the number of trees and s is the standard deviation of tree lengths. For a perfectly symmetrical tree-length distribution = 0, while a left skewed distribution has a gi < 0 and a right skewed distribution has a >0. The bootstrap is a measure of the variation present in the characters of a data matrix and how representative this set of characters is to the larger distribution of all possible characters. Multiple bootstrap pseudoreplicate matrices are created by sampling characters with replacement from the original data matrix. A consensus tree of all most parsimonious trees from all pseudoreplicates is then constructed showing the percentage of pseudoreplicate trees that showed the reconstruction of each clade in the coi\sensus tree. 128

This study examines the ability of these indices to indicate appropriate level of confidence in phylogenetic hypotheses derived from data matrices with different levels of homoplasy. Initially core data matrices were used that had no homoplasy. These data matrices then had increasingly larger blocks of random data ("noise") added. More specifically, this study tests the ability of each index to meet two related goals of a confidence index. First, for a fixed noise level, does each index indicate the relative structure of characters in a data matrix. In other words, does the index indicate the level of homoplasy in the matrix and are homoplastic characters correlated so that they strongly contradict homologous characters. Second, how well does each index indicate the probability that the true tree is included within the set of most parsimoruous trees (MPTs). Only indices that assess phylogenetic confidence for the tree as a whole are considered, not those that assess confidence in specific nodes (e.g., decay index). Maximum parsimony is the algorithm used in all reconstructions and indices derived from distance matrices are not examined. Expectations of a preferred index There are two preferred characteristics or goals of a phylogenetic confidence index that I test in this study. First, a confidence index should express the degree to which the characters of the data matrix are non-conflicting and congruent. This point concerns the relative level of homoplasy in a matrix and how strongly it contradicts homologous characters. If patterns in homoplastic characters are correlated, they will have a stronger impact contradicting evolutionary patterns of homologous characters. Second, a confidence index should express the appropriate degree of confidence that a group of MPTs contains the true tree. An index that meets these goals can be used by a 129 researcher to assess the degree of confidence to place in their data matrix quality and the probability of reconstructing the true tree using this matrix. Under the conditions of this study, one might hope an index to behave as shown in Figure 5.1. For a particular class of matrices (those with a certain number of taxa, initial node support and tree shape) the solid line would indicate the average value of a confidence index for data matrices that successfully reconstruct the true phylogeny with different levels of random data. The broken line would indicate the average value of a confidence index for data matrices that fail to reconstruct the true phylogeny at different levels of random data. Of course, in a perfect world the top line would always be as high as possible and the bottom line would always be as low as possible. However, the curves depicted here are probably more likely, with it becoming increasingly difficult to successfully predict if the true tree is found with larger numbers of random characters. 130

O) Metric values for matrices finding the true tree

o o Metric values for matrices c ' failing to fmding the true tree 0) ^ -O O

V fl)E .ru b 4^ 3 Q> CO C (0 0) » O) S o

o

Few > Many Random Characters or "Noise"

FIGURE 5.1. Preferred characteristics of a phylogeiietic confidence index. The solid line indicates the mean index value for data matrices that successfully reconstruct the true phylogeny. The broken line indicates the mean value of that index for data matrices that fail to recoristruct the true phylogeny. Tv^o crucial points concerning these curves are the iiutial distance between curves (*) and the steepness of the central portion of the solid curve (**). 131

There are two importarit feahires of the curves in Figure 5.1 that relate directly to the two preferred characteristics or goals of a phylogenetic confidence index mentioned previously. The first characteristic concenis the distance between the solid and broken lines. Our goal here is for an index to maximize this distance between ciirves in the left portion of the x axis. This distance can be viewed as a measure of the congruence or non-coriflicting nature among characters. A great distance between lines indicates a relatively high number of homologoias characters, and the stronger the correlation among homoplastic characters must be to prevent the reconstruction of the true tree. With few random characters in matrices, one hopes that parsimony would normally reconstruct the true tree and that a confidence index would have a value reflecting this high confidence. A few matrices may fail to reconstruct the true tree and should have very low index values. These latter matrices fail with only a few random characters because, just by chance, their homoplastic characters are correlated, overwhelming homologous characters. With an increasing number of random characters, the relative number of non-conflicting and congruent apomorphic characters in the matrices will decrease and the distance between the two lines should decrease. As the number of random characters increases, parsimony increasingly fails to find the true tree. The second point concerning the shapes of the curves in Figure 5.1 and relates to the ability of an index to track increasingly lower corifidence levels with increasingly lower index values. The second goal for a index is that the solid line should drop in concert with lower probabilities of reconstructing the true tree. In other words, we are concerned with the steepness of the solid line. It is preferred that the solid line be steepest in the central 132 portion of the curve. Such a pattern maximizes the difference in index values for matrices that are usually successful and those that usually fail to reconstruct the true tree. In cases where we have a large amount of random data, we will only reconstruct the true tree successfully in a few cases and a good structure index should indicate this with a low value. 133

METHODS Blocks of 5 to 50 random binary characters were added in 5-character increments, with 20 replicates at each random-character level, to two core data matrices. Core data matrices specified either a symmetrical or a pectinate tree topology. A total of 202 matrices were created. Each core matrix was composed of seven taxa and 10 non-conflicting binary characters with two characters supporting each node in the most parsimonioiis phylogenetic tree. These trees reconstructed from each core matrix are subsequently referred to as the "true" trees. MacClade 3.0 (Maddison and Maddison, 1992) was used to generate the random binary characters with an equal probability of a 0 or a 1 being placed in any cell in the random block. This simulates a model of evolution in which characters are evolving at such a rapid rate as to eliminate any phylogenetic information, although it is possible that some randomly generated characters might support original "true" nodes by chance alone. Phylogenetic estimates were made using the exhaustive search option in

PAUP 3.0S4-1 (Swofford 1991) to calculate all of the 945 possible trees for each data matrix. For each matrix the following data were recorded: CI excluding uninformative characters, RI, RC, gi, average nonparametric bootstrap value, number of MPTs and their tree length, number of steps to the next MPT set, and whether the true tree was included in the MPT set. Average nonparametric bootstrap values were calculated by averaging the values at the 4 nodes of a majority-rule consensus tree of 100 nonparametric bootstrap replicates for each data matrix. I then examined how each index varied in its ability to indicate confidence in finding the correct tree at different probability levels with increasing levels of random characters. 134

RESULTS Matrices failure rate Core matrices of both tree topologies were fairly robust to the effects of adding a large number of random characters. Figure 5.2 shows the percentage of replicates at each random data level which failed to reconstruct the true symmetrical tree. For symmetrical trees it took 2.5 times as many random characters as core characters (25:10) before there was an even chance of getting a true or a false tree and 2 times as many for pectinate trees (20:10, not shown). A type n error in this case is the risk of accepting an incorrect phylogeny as the true phylogeny. To have confidence in not making a t)^e n error at levels approximating p values of 0.05 for symmetrical trees and 0.1 for pectinate trees it is appropriate to examine index values from matrices with 10 random characters. 135

Symmetrical Trees

1.0T

Number of Random Characters

FIGURE 5.2. Distribution of type n error rates when different levels of random data are added to core symmetrical tree matrices. Distribution for pectinate tree matrices is similar. 136

Behavior of CI. RI. and RC The behaviors of CI, RI, and RC with the addition of random characters were similar for both sjnnmetrical (Figure 5.3) and pectinate trees. Remember that the first goal is for an index to maximize the distance between index values of matrices that successfully reconstruct the true tree and those that fail to do so, for a given level of noise, in the left portion of the x axis (e.g. solid vs. dashed lines). However, in this case, there were no significant differences between index values of matrices that successfully reconstruct the true tree and those that fail to do so. The rescaled consistency index had the steepest and deepest decline in the solid line. This decline maximizes differences in index values of matrices that always reconstruct the true tree and those that fail to do so at a p value of .05 for symmetrical trees and 0.10 for pectinate trees. Therefore, RC values above 0.4 are a fairly conservative confidence measures for both tree types. Unfortimately, the curve then leveled off, eliminating the possibility of differentiating among lower levels of confidence such as between 0.20 to 0.80. The second goal for an index, indication of the appropriate probability of a matrix reconstructing the true tree, is therefore not met for type 11 error rates greater than 20%. In other words, RC does not differentiate between probabilities of a matrix reconstructing the wrong tree between 20% or 80% of the time. 137

Symmetrical Trees a WHEN OGRRECT a WHEN WRONG n WHEN CORRECT 0.8- n WHEN WRONG HC WHEN CORRECT RC WHEN WRONG

Number of Random Characters

FIGURE 5.3. Behavior of CI, RI, and RC indices to the addition of random characters to core symmetrical tree matrices. Numbers above the x axis represent the proportion of the 20 replicate data matrices at each level of random characters that failed to support the true tree. These numbers represent estimates of the type II error rate. Standard error bars shown for RC values only. 138

Behavior ofgi For the gi statistic, the more negative the value the more structured the data. While it appears that there are some differences between index values of matrices that successfully reconstruct the true tree and those that fail to do so, for a given level of noise, there are no consistent significant differences. This lack of difference between successful and failed matrices again indicates that this index does not measure the degree of character congruence within matrices (i.e., goal one fails). Values above the dashed line in Figiire 5.4 indicate data matrices not different from random at p=0.05, according to studies of Hill is and Huelsenbeck (1992). This line intersects the line of matrices correctly reconstructing the true symmetrical tree at the 10 random character level. Again this random character level has failure rates of 5% for symmetrical trees and 10% for pectinate trees. Therefore, values below the dashed line are fairly conservative confidence measures for each tree type. The slope of the curve may allow further resolution of slightly lower levels of confidence (goal two) with p values of 0.20 for symmetrical trees at the level of 15 random characters, before the line more or less levels off. This increased power of resolution is not seen for pectinate trees. The distribution of points at different levels of random characters is more variable for both symmetrical and pectinate topologies when compared to the previoias three indices. 139

Symmetrical Trees 0.0

(0.2) -

(0.4) -

U) (0.6) - Significantly ^ Structured Data

(0.8) - G1 WHEN CORRECT

(1.0)- --O G1 WHEN WRONG

05 .20 .25 80 .75 .75 .80 (1.2) 0 10 20 30 40 50 Number of Random Characters FIGURE 5.4. Behavior of the gj tree length skewness index to the addition of random characters to core symmetrical tree matrices. See Figure 5.3 for explanation of the numbers above the x axis. 140

Behavior of average nonparametric bootstrap values Average noriparametric bootstrap values above 52 for symmetrical trees arid 60 for pectinate trees are indicative of confidence probabilities of 0.05 (Fig. 5) and 0.10 respectively. Again, while it appears that there are some differences between index values of matrices that successfully reconstruct the true tree and those that fail to do so, for a given level of noise, there are no consistent sigriificant differences, indicating that this index does not measure the degree of character congruence within matrices (goal one). The slope of the curve again allows resolution of slightly lower levels of confidence (goal two) with p values of 0.20 (symmetrical trees) and 0.30 (pectinate trees) at the level of 15 random characters, before the line levels off. 141

Symmetrical Trees 100 -I

BOOTSmAP WHEN CORRECT BOOTSTRAP WHEN WRONG

Number of Random Characters

FIGURE 5.5. Behavior of average nonparametric bootstrap indices to the addition of random characters to core symmetrical tree matrices. See Figure 5.3 for explanation of the numbers above the x axis. 142

DISCUSSION Given the results of this study, how do the structure indices examined here match up to the expectations of a preferred index (Figiure 5.6 versus Figure 5.1)? Relative to the first goal, none of the indices examined here can express the degree to which characters of the matrix are non-conflicting and congruent within a given level of noise (i.e. distance between lines is small in the left region of the x axis of figure 5.6). A good congruence index would indicate when the relative levels of homologous characters to conflicting, correlated homoplastic characters were such that the reconstruction of the true tree was likely to fail. Relative to the second goal, the indices examined here show only limited abilities to distinguish among different probabilities of matrices making type n errors. The solid line is steep over only a very narrow range of random character values and then becomes flat. Each index can therefore effectively identify matrices in which the true phylogeny is contained in the MPT set with error rates between 5 and 10% but then are typically not able to differentiate among higher error rates. The same index value may be seen for error rates ranging from 20 to 80%. These analyses under these conditions suggest that, relative to goal two, RC is a more powerful indicator of phylogenetic confidence than RI, which in turn is more powerful than CI. The behaviors of the gi and average nonparametric bootstrap statistics are more variable at increasing levels of random characters when compared to the previous three indices. (0 0) O a> Metric values for matrices S o finding the true tree

lUletric values for matrices failing to finding the true tree 3 o

Few Many Random Characters or "Noise"

FIGURE 5.6. General characteristics of phylogenetic confidence indices seen in this study. When compared to figure one, it is clear that many of the preferred characteristics of a phylogenetic structure index are not met (see text). 144

An interesting side note to this study is the finding that maximum parsimony is quite robust to the addition of a large number of characters evolving at a maximal rate. These data confirm the theoretical arguments of Farris (1983) who claimed that parsimony does not depend on rarity of homoplasy. Correct reconstruction of trees is not simply a matter of rate variation among characters. The important point is how everUy the informative ("slow") characters in a matrix are spread among branches of the tree. If a index can be found that identifies the level of character compatibility and conflict in a data matrix, we may then be able to look deeper into these matrices and attempt a posteriori, yet objective character weightings with subsequent phylogenetic analysis (Farris 1969, Felsenstein 1981). Unfortunately, measuring this relative level of character congruence is specifically where all of the indices examined here fail. The generality of these results to other classes of matrices (those with different number of taxa, node support and tree shape) is not clear. Of course, it is not possible for researchers to know the amoimt of good core support or the number of random characters they have in their data sets. It is therefore crucial to understand how phylogenetic structure indices behave under a variety of conditions. Future similar studies should look at the robustness of the results seen here relative to changes in levels of initial support, number of taxa, and character types with increasing levels of random data. After such studies, we might have reliable gmdes to the appropriate values associated with currently utilized confidence indices and how they may be modified to improve their performance. 145

If an index could be produced that meets the two goals outlined above, it could be used in two main ways. First, the index could be used to examine the quality of data in data sets. If a data matrix was found to probably contain several homoplastic characters, index values of individual characters could be examined tintil homoplastic characters were identified. These characters could then be either filtered out of the matrices or down weighted to have minimal influence on tree recor^truction. The CI, RI and RC are used in this latter maimer by successive approximation character weighting methods (e.g., in PAUP, Swofford 1991,1993). Second, this index could identify the probability of the true tree being represented in MPT set. This last point could be particularly useful in comparisons of phylogenies reconstructed from different data sets. 146

CONCLUSIONS The work presented here is the most thorough examination of the evolution of cooperative breeding (CB) in any bird family. One general ecological influence on CB evolution is supported by this study. The importance of social interactions to CB evolution in this group suggests the importance of kin selection. Interestingly, the only other species in which kin selection appears to be the primary influence on CB evolution is another coraciifonn species: the pied kingfisher, Ceryle rudis. Additional research on other CB coraciiform birds may point to a single origin of CB due to kin selection. Other ecological and behavioral influences, however, when considered individually, had very little impact on the evolution of CB in the bee-eaters. The pattern of early evolution of CB within the family, with few subsequent transitions, despite considerable ecological and behavioral diversity, is consistent with patterns seen for other CB bird groups (Peterson and Burt 1992, Edwards and Naeem 1993). Whether this lack of reversals to non-CB is due to the malleable nature of CB systems or that it represents a phylogenetically influenced time-lag is not known. This question can only be resolved by further historical analyses of various subcomponents of CB systems. These studies can not, however, be completed until detailed ecological data are available for all species in question. In the end we find ourselves in a quandary. Comparative studies at the family and higher phylogenetic levels can provide evidence on broad categories of traits ir\fluencing CB evolution. These studies require only basic knowledge of the breeding systems of each species examined and ecological studies like those of Chapter Three are sufficient for the time being. However, these studies miss 147 much of the detailed influences involved in the evolution of this very complex behavioral system. More detailed comparative studies of CB subcomponents, however, require detailed ecological data on each species. When studies are conducted at the family level, we find that these data are simply not available for more than a few species. Because of these difficulties, I recommend a combined approach. First, higher level comparative studies can identify general traits influencing CB evolution. Subsequently, sister-groups can be identified and detailed ecological comparisons of subcomponents of CB behavior can be examined to leam more relative to the specific evolutionary mechaiusms involved in the evolution of breeding systems. An additional problem in historical studies such as that presented here is the lack of confidence in the phylogeny for the group in question. This point is not as critical many times while examining the influence of general ecological traits on CB evolution. Conclusions of such studies tend to be more robust to slight variations in a number of possible phylogenetic hypotheses. More detailed historical studies however will require confidence in fewer or preferably a single, well supported, phylogeny. Concerning confidence in phylogenetic reconstruction, the last study in this dissertation conveys mixed information. The ability of many currentiy used phylogenetic confidence indices to express confidence in the level of support for phylogenetic recor^structions is weak. However, it appears that parsimony analyses are fairly robust to relatively high levels of homoplasy in data matrices. Therefore, if character support for tree branching patterns is evenly spread throughout the tree, researchers should feel fairly confident in their phylogenetic hypotheses. 148

APPENDIX A

CHARACTER STEP MATRICES ^ a ^ X)^

Forecrown

lilac[•TT > I 3 I 3 I 3 I 3 I 3 green with blue tips 3 "W deep purple ~3 3" black ^"T light blue

white dark brown chestnut

rea cnestnut greenish blue

Hindcrown

nrangp . • 3 3 3 3 3 3 3 3 3 3 3 3 3 i;: 3 3 green 3 • 3 3 3 m 3 3 3 3 ,Za- a- 3 3 3 2- '2' 3 deep purple & brown 33 3 * 3 3 3 3 3 3 3 3 3 3 3 X 3 3 3 black 3 3 5 • 3 3 3 3 3 3 3 3 3 3 3 3 3 3 light blue 3 3 5 3 § 3 3 3 3 3 3 3 ix >4: 3 3 3 3 green with blue tips 3 1 3 3 3 • 3 3 3 3 ••2-; 3 3 3 2.-,, •-2-: 3 orange 2 3 3 3 3 3 4 3 3 3 3 3 3 3 3 .i:.- 3 3 dark brown 3 3 3 3 3 3 3 • 3 3 3 3 3 3 3 3 3 3 chestnut 3 3 3 3 3 3 3 3 • 3 3 3 3 3 3 3 3 m gray 3 3 3 3 3 3 3 3 3 • 3 3 3 3 3 3 3 3 pied 3 2 3 3 3 1 3 3 3 3 • "if! 3 3 3 'A- •-.ir- 3 dark greenish-brown 3 2 3 3 3 2 3 3 3 3 2 * 3 3 3 3 greenish-blue 3 3 3 3 ± 5 3 3 3 3 3 3 • ;4? 3 3 3 3 blue 3 3 3 3 2 3 3 3 3 5 3 3 2 • 3 3 3 3 purple 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 green with orange 2 2 3 3 3 2 2 3 3 3 2 2 3 3 3 * 2 3 light greenish-brown 3 2 3 3 3 2 3 3 3 3 2 1 3 3 3 2 • 3 red chestnut 3 3 3 3 3 3 3 3 1 3 3 3 3 3 3 3 3 • 149

APPENDIX A - Continued

uX -

green * i 2 2 2 T 2 2 2 2 2 2 light blue 1 • 1 tla 1 2 2 2 2 2 white 2 2 m 2 m 2 2 2 2 2 2 2 2 small blue 2 1 2 • sdi 2 2 2 2 2 2 white & blue i 1 1 1 * !ilX i i 2 2 2 2 ds very little blue 1 1 1 1 1 • 2 J 2 2 2 2 green & lilac 1 2 2 2 2 2 • 2 2 2 2 2 2 deep purple i i i i i i i • 2 2 2 2 2 black 2 2 2 2 2 2 2 2 * 2 2 2 2 chestnut 1 1 1 i 1 i 2 J 2 • 2 2 red chestnut 2 2 2 2 2 2 2 2 2 1 • 2 2 gray 2 2 2 2 2 2 2 2 2 2 2 • 2 greenish-blue 2 1 2 1 1 1 2 2 2 2 2 2 *

Infradiium

green «' 2 -2 2 2' "1~ •5" •2" 2 2 -2- 2" light blue 1 * 2 ,sls ^le 2 2 J 2 2 2 -1.. white 1 1 • 2 2 1 2 2 vt; 2 2 light blue & white 2 1 1 • i 2 i 2 2 2 .i- very little blue 1 1 2 1 « I 2 2 2 2 2 -i green& red 1 2 2 2 2 • 2 2 2 2 2 deep purple 2 2 2 2 2 2 • 2 2 2 2 2 black 2 2 2 2 2 2 i • 2 2 2 2 yellow 2 2 2 2 2 2 2 2 # wl''- 2 2 white & yellow 2 2 1 2 2 i 2 2 1 • 2 2 red 2 2 2 2 2 1 2 2 2 • 2 greenish-blue 2 1 2 1 1 i 2 2 2 2 2 • .4: .cT ve K>° J" Nape A J-

green • 3 3" 3 3 3 3 3 3 3 w m T" T- dark brown 3 • 3 3 T 3 3 3 3 3 3 3 3 3 3 black li 3 * 3 5 3 3 3 3 3 3 3 3 3 3 deep purple s 3 3 • 5 3 3 3 3 3 3 3 3 3 3 green with blue 1 3 3 3 • 3 >1:; 3 3 3 r;lv A- 3 3 orange 3 3 3 3 3 • 3 3 3 3 3 3 3 orange-green 2 3 3 3 i 2 * 3 3 3 c2,-.' 3 3 chestnut 3 3 3 3 3 3 3 • 3 3 V 3 3 3 ry gray 3 3 3 3 3 3 3 3 • 3 3 3 3 3 red i 3 3 3 3 3 3 3 3 • 3 3 3 3 green with blue over orange i 3 3 3 2 2 1 3 3 3 * 3 3 dark greenish-brown 2 3 3 3 2 3 2 3 3 3 2 • 3 3 light greenish-brown 2 3 3 3 2 3 2 3 3 3 2 1 * 3 3 red chestnut 3 3 3 3 3 3 3 1 3 3 3 3 3 • 3 orangish red 3 3 3 3 3 3 3 3 3 i 3 3 3 3 * APPENDIX A - Continued

Mantle

green • 3 3 3 3 42£ 3 3 chestnut 3 S 3 3 red 5 3 « 3 3 5 3 /i* black 3 3 3 • 3 3 3 3 3 green with blue tips 1 3 3 3 • 3 3 3 gray 3 3 3 3 3 * 3 3 3 light greenish-brown 2 3 3 3 i 3 • 3 3 red chestnut 3 1 3 3 3 3 3 f 3 orangish-red 3 3 2 3 3 3 3 3 • •OJ" :> Back

green • 3 3 3" 3 3 green with blue tips 1 f 3 3 3 3 3 ?S- 3 chestnut 3 3 • 3 5 3 3 3 black 3 3 3 • 3 3 3 3 3 yellow & chestnut 3 3 2 3 • 3 3 3 3 gray,slight red 3 3 3 3 3 * 3 -i. red 3 3 3 3 3 2 • 3 i; light greenish-brown 2 2 3 3 3 3 3 • 3 orangish-red 3 3 3 3 3 2 2 3 •

^ vs- Rump / green • 3 3 i 3 3 3 chestnut-light blue 3 • •i' ¥3 3 3 i- 3 blue 3 2 • 3 3 3 i-. 3 green with blue tips 1 3 3 * i. 3 3 3 yellow-green 2 3 3 2 • 3 3 3 gray,slight red 3 3 3 3 3 • 3 3 light blue 3 2 2 3 3 3 • 3 blue-purple 3 3 3 3 3 3 3 *

J"

UpperTaU Coverts

green • 3 3 IS 3 3 blue 3 • ••2-1 3 3 3 light blue 3 2 • 3 3 3 green with blue tips 1 3 3 * 3 3 grajfishred 3 3 3 3 • 3 deep purple 3 3 3 3 3 • APPENDIX A - Continued

•is

3-S-- Inner Rectrices, Dorsal -«V>

green • 2 i 2 2 2 '4S 2 2 light blue, white tip 2 • 2 jis m 2 2 2 2 2 2 2 dark blue to black i i • i i i 1 2 1 i blue 2 1 1 • 2 2 2 2 2 2 2 blue-green, white tip 1 1 2 2 • 2 2 2 2 2 reddish black 2 2 2 2 2 • 2 2 2 2 2 iits purple i i i i i i • i i 1 i i blue-green 1 2 2 2 1 2 2 • 2 2 2 blue-green, black tip 1 2 2 2 1 2 2 1 • m 2 2 green, black tip 1 2 2 2 2 2 2 2 1 • 2 2 blue, black tip 2 i 1 1 i 1 i i i i • 2 red, black tip 2 2 2 2 2 1 2 2 2 2 2 •

Outer Rectrices, Ventral

yellow to black • 2 2 2 2 2 2 2 2 yellow 1 • 2 2 2 2 2 2 2 2 orange-brown to brown (m to 1) 2 2 • 2 2 2 2 2 2 2 darker light brown i i i • i 2 i 1 t. 2 black 2 2 2 2 • 2 2 2 2 2 brown, white tips i i i i • i 1. i i orange-brown, black, white (p to d) 2 2 2 2 2 2 • 2 2 2 brown i i i 1 i 1 i • ,1- 1 light brown 2 2 2 1 2 2 2 1 • 1 light brown, dark tip 2 i 2 2 i i 2 2 1 •

^ •x?' Rectrices Rachis, Ventral ^ ^ ^ -<9^ ^ ^

yellow • i 2 2 2 2 2 off-white 2 • 2 2 2 2 orange-brown 2 2 • i 2 2 2 tan to black 2 2 2 • 2 2 off-white to brown 2 1 2 2 • 2 2 tan 2 2 2 1 2 • a-. brown 2 2 2 2 2 1 • 152

APPENDIX A - Continued

blue then red • 2 2 2 2 •i 2 2 2 1 chestnut 2 * 2 2 2 2 2 2 2 2 2 black then purple 1 2 • 2 2 1 2 2 2 2 2 black 2 2 1 • 2 2 2 2 2 2 2 2 blue 1 i i i • 2 2 2 2 2 2 red 2 2 2 i 2 * 2 2 2 2 2 2 yellow 2 2 2 2 2 2 • 2 2 2 2 white 2 2 2 2 2 2 2 • 2 2 4* 2 bluish-gree 2 2 2 2 2 2 2 2 * ate 2 2 green 2 2 2 2 2 2 2 2 1 « 2 2 yellow & white 2 2 2 2 2 2 1 1 2 2 t 2 greenish-blue 1 2 2 2 1 2 2 2 2 2 2 •

Throat ..^•€

red • 2 2 2 2 2 2 2 2 2 2 blue, green sides 2 • 2 2 2 2 2 2 5^ 2 deep purple 2 2 * 2 2 2 2 2 2 2 2 black 2 2 2 • 2 2 2 2 2 2 2 yellow 2 2 2 2 • 2 2 2 2 2 2 white 1 2 2 i 2 • 2 2 2 2 2 bluish-green 2 2 2 2 2 2 * 2 2 2 chestnut 2 2 2 2 2 2 2 • 2 2 2 blue 2 1 2 2 2 2 2 2 • 2 -l.; green 2 2 2 2 2 2 1 2 2 • 2 greenish blue 2 1 2 2 2 2 2 2 1 2 * 1^ i- .N. xi^- . ^VV/V fit-' >sr v>®'' c*' sjr ^

Gorget ^ y ss" y y «ir ^^• ^ none • 2 2 2 2 2 2 2 2 2 2 2 2 2 chestnut 2 • 2 2 1.^ 2 2 2 2 2 2 2 2 2 It. blue, dk. blue 2 2 * 2 2 .11 2 2 2 2 2 IL blue, black, chestnut 2 2 2 • 2 .-1;^ 2 2 2 i-i .1. il:r 2 chestnut, black 2 1 2 2 • 2 2 a-'- •-1« 2 2 2 2 2 white. It. blue, dk. blue, chestnut 2 2 1 1 2 • 2 2 2 • t. -li 2 It. blue, black. It. blue 2 2 2 2 2 2 * 2 2 2 2 2 2 1 black 2 2 2 2 1 2 2 • 2 2 2 2 2 2 chestnut, black, blue 2 2 2 2 1 2 2 2 • 2 2 2 2 2 white. It. blue, black, chestnut 2 2 2 1 2 1 2 2 2 • •'•k rV 2 It blue, dk. blue, black, chestnut 2 2 1 1 2 1 2 2 2 1 * -Ilr ;ly 2 white. It blue, dk. blue-black, chestnut 2 2 1 1 2 1 i 2 2 1 1 * i. 2 It blue, dk. blue, chestnut 2 2 1 1 2 1 2 2 2 1 1 1 • 2 green-blue, black, green-blue 2 2 2 2 2 2 1 2 2 5 2 2 2 • 153

APPENDIX A - Continued

A ^

Breast

red, green sides T 4 n BEF1 T PH T greenish-blue 4 • 4 4 4 4 4 4 4 4 4 1 4 1 4 deep purple 4 • 4 4 4 4 T" "5" 4 4 1 4 1 4 yellow-orange 4 4 4 "W 4 4 4 H 4 4 4 4 4 black & irrd. blue T T T T 1" 4 4 4 T" 4 i orange-yellow-green 4 4 4 2 4 4 m B 4 4 14 light green 2 4 4 4 4 IMBi 4 4 r? ereen T 4 4 T 4 2 • 4 4 orange-brown T T" 1 T" T T f T" T T red 4 T" 4 4 4 4 4 4 • 4 4 blue, green sides 1 4 4 4 4 "3~ T" 4 4 • 4 black & irrd. blue with red 4 4 4 4 T" 4 4 4 4 4 ~9 green with blue 2 4 4 T "3" 2 1 T "3" 4 yellowish green 2 4 4 4 4 2 2 4 4 3 4

Flank ^ J'.

green * 3 3 3 3 3 3 3 3 3 green & yellow 2 * 3 3 5 3 3 3 3 3 i,. 3 brown-gray 3 3 • 3 3 3 3 3 3 3 3 3 3 yellow-orange S 3 3 • 5 r'i" •ji. 3 3 3 3 3 3 blue 3 3 3 3 * 3 3 3 3 3 3 3 orange-yellow-green 3 3 3 2 3 • fh 3 3 3 3 J 3 orange-brown 3 3 3 2 3 2 • 3 3 3 3 3 3 greenish-blue 3 3 3 3 2 3 3 • 3 3 3 3 3 red 3 3 3 3 3 3 3 3 * 3 3 •fZ-: 3 ind. blue & black 3 3 3 3 3 3 3 3 3 • 3 3 3 light green 2 2 3 3 3 3 3 3 3 3 • 3 ;2;: orange-red 3 3 3 3 3 3 3 3 2 3 3 • 3 green with blue 1 2 3 3 3 3 3 3 3 3 2 3 * 'f w /A

Belly

green • 2 3 3 3 3 3 3 m 3 :2f; 3 3 an green & yellow 2 • 3 3 3 5 3 3 •2>, i 3 3 brown-gray 3 3 * 3 3 3 3 3 3 3 3 3 3 3 buff 3 3 3 4 3 3 =i-j 3 3 3 3 •i.' 3 3 blue 3 3 3 3 • 3 3 3 3 3 •2' 3 greenish-blue 5 3 3 3 2 • 3 3 3 3 3 -2.- 3 yellow-green 3 3 3 2 3 3 • 3 3 3 3 '.2: 3 3 greenish-white 3 3 3 3 3 3 3 * 3 3 3 3 3 3 light green-blue 2 2 3 3 2 2 3 3 • 3 3 3 2. • Z.' red 3 3 3 3 3 3 3 3 3 • 3 3 3 i light green 2 2 3 3 3 3 3 3 3 3 * 3 3 ;2- orange-yellow-green 3 3 3 2 3 3 2 3 3 3 3 « 3 3 irrd. blue 3 3 3 3 2 2 3 3 2 3 3 3 • 3 green with blue 1 2 5 3 3 3 3 3 2 3 2 3 3 • APPENDIX A - Continued

^ < UndeitaU Coverts v5-'^^

light green |9* |3|3|3|3|3|33 3 3 3 3 3 3 3 3 3 3 yeUow T" "y* "5"i 3"3 "5~3 "T3 X3 3 5 3 3 3 3 burnt orange, green tips 133 I 3 i• • I 3 |^| 3 3 3 3 3 3 3 light blue 3 3 3 • 3 3 3 3 3 iHigi buff 3 3 2 3 * 3 3 3 3 'il'i 3 3 blue-purple S i i 2 3 • 3 3 3 3 ?iE blue 3 3 3 2 3 2 • 3 3 3 3 i23 I22S reddish-gray 3 3 3 3 3 3 3 • 3 3 3 3 3 irrd. blue 3 3 3 3 3 3 3 3 « 3 3 3 3 i 3 2 i 2 3 3 3 3 • 3 3 3 green with blue 2 3 3 3 3 3 3 3 3 3 • 3 3 greenish blue 3 3 3 1 3 2 2 3 3 3 3 • *> light green-blue 3 3 3 1 3 2 2 3 3 J 3 2 •

oSb ^ Forehead

light blue * 3 3 3 3 3 3 3 3 3 5 3 2 yellow & white 3 • 3 3 3 3 -4;i 3 3 3 3 3 3 3 3 3 deep purple-blue 3 3 • 3 3 3 3 3 3 3 3 3 3 3 3 3 black 3 3 3 * 3 3 3 3 3 3 3 3 3 3 3 3 green with blue tips 3 3 3 3 * 3 3 3 3 3 iis 3 3 3 green 3 3 3 3 1 3 h 3 3 3 3 ;4.r 3 3 3 white 3 2 3 3 3 3 • 3 3 3 3 3 3 3 3 pied 3 3 3 3 ti i 3 * 3 3 3 3 3 3 3 gray 3 3 3 3 3 3 3 3 * 3 3 3 3 3 3 3 chestnut 3 3 3 3 3 3 3 3 3 • 3 3 3 3 1 3 Ught blue & green 2 3 3 3 3 3 3 3 3 3 • -.2.- 3 3 3 2 little light blue 2 3 3 3 3 3 3 3 3 3 i • 3 .1 3 i green with orange 3 3 3 3 2 2 3 2 3 3 3 3 * 3 3 3 white & light blue 2 3 3 3 3 3 i 3 3 3 3 i 3 # 3 3 red chestnut 3 3 3 3 3 3 3 3 3 1 3 3 3 3 • 3 greenish blue 2 3 3 3 3 3 3 3 3 3 i i 3 3 3 *

Lore »• rNr

lilac & red • i 1 2 green 2 • 2 2 black 2 2 * 1- black & chestnut 2 2 1 • APPENDIX A - Continued

^<0 <

Scapulars

green • 3 3 3 3 3 3 3 3 green with blue 1 3 T" "5" "3~ 3 "T" 3 chestnut 3 3 "W T 3 "5" T" 3 T" 3 chestnut-orange 3 5 T" % T" 3 "3" "3~ 3 "3~ 3 black 3 5 3 3 • 3 3 3 3 3 3 green with orange 2 2 "3~ 3 "3~ * "3~ "3" 3 "3~ 3 greenish brown 3 5 "3~ 3 T 3 T" 3 T" 3 yellow-orange 3 3 "3~ 3 T 3 "3" T" 3 yellow-orange with green 3 3 3 3 3 3 3 2 m 3 3 gray,slight red 3 5 "3~ 3 "3" 3 "3" "3~ 3

red 3 3 T 3 T" 3 ~ 3 2 *

/X Marginal Coverts

green with blue tips « HA 3 3 3 3 green 1 • 3 3 3 3 chestnut 3 3 • 3 3 3 3 3 black 3 3 3 « 3 3 3 3 light greenish-brown ± 2 3 3 • A 3 3 dark greenish-brown 2 2 3 3 1 • 3 3 gray 3 3 3 3 3 3 * 3 red 3 3 3 3 3 3 3 *

Lesser Coverts

green with blue tips • m 3 3 :2:- M 3 3 green 1 • 3 3 •i'- 4. 3 3 chestnut 3 3 * 3 3 3 :2-i: 3 3 black 3 3 3 • 3 3 3 3 3 light greenish-brown 2 2 3 3 • gfABaR 3 3 dark greenish-brown 2 i 3 3 1 • -i.; 3 3 green with chestnut 2 2 2 3 2 2 * 3 3 gray 3 3 3 3 3 3 3 • 3 red 3 3 3 3 3 3 3 3 • APPENDIX A - Continued e, j''" V>> 0 ^ Median & Greater Coverts

green with blue tips • 3 3 as ppl i 3 green 1 m "3~ "3" T

chestnut T 3 T" 3 13 13 lac?! 3 1" black 3 3 3 • 3 3 3 1 3 3 3 green with orange 2 2 T" T" "3"

light greenish-brown T" 2 T" T ~ dark greenish-brown 1" 1 T T" 2 11 [•pziel 3 green with chestnut 2 2 2 3 2 1 2 1 2 [• 1 3 3

gray 3 3 3 3 3 13 1 3 1 3 !• 3 red 1" i T T 3 1 3 1 3 1 3 1 3 T

Tertials

* 2 2 2 2 1 • 1 5 2 1 2 2 • 2 2 2 biacK i i 4 • i i gray 2 2 2 2 « 2 red with blue tips i 5 1 i •

black white & blue white

red red & blue red & white blue & chestnut blue blue & green w ^ rhp«:tniif

1 dark brown 1 1 • APPENDIX B SPECIES DATA MATRIX

Species Character Number 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Nyctyarnis amicta 0 0 1 5 5 0 0 0 0 0 0 0 0 0 0 Nyclyorni atherloni 0 4 1 0 0 0 4 1 0 4 1 3 3 0 2 Meropogon forsteni 2 2 2 7 6 1 1 2 0 0 0 0 0 0 2 Merops breweri 3 3 3 6 7 2 2 2 0 0 1 0 0 0 2 Merops muelleri 0 4 4D 14 7 2 3 2 0 1 2 14 5 26 0 Merops giilaris OB 3 3 18 7 2 2 2 0 3 3 7 2 2 0 Merops hirundineus 05 5 1 01 3 3 0 2 1 0 0 27 12 17 2 Merops pusillus 04 1 5 138 8 5 4 2 1 4 1 3 3 0 12 Merops variegatus 04 1 5 18 9 4 4 2 1 4 1 3 3 0 2 Merops oreobales B 1 5 1 9 4 4 2 1 4 1 3 3 0 0 Merops buUocki 04 1 5 018 014 67 5 2 1 4 1 3 3 0 1 Merops bullockoides 6 6 6 1 2 8 5 2 1 4 1 3 1 4 2 Merops revoilii 7 A A 1 2 4 A 2 1 4 1 7 2 4 1 Merops albicollis 6 7 7 2 2 4 6 2 1 4 1 3 2 8 0 Merops boehmi 9 8 8 9 14 9 4 2 1 4 1 3 3 8 2 Merops orientalis 059C 458D 18E 01 1 OA 067 2 1 014 01 03 03 A 2 Merops persicus D 4 15 4 3 B 04 2 1 04 01 03 03 A 1 Merops superciliosus 6 B B 2 2 C B 2 1 6 1 3 3 0 1 Merops philippinus A C F 1 1 9 C 2 1 6 2 2 78 1 Merops orfialus 4 1 0 0 1 D 5A 2 1 4 1 2 2 9 2 Merops viridis 9 8 8 9 1 A 7 2 1 1 1 2 2 3 1 Merops leschenaulli 9 8 8 9 8 5 7 23 12 1 0 2 2 0347 1 Merops apiaster 16 1 8 4 3 E 7 2 1 1 4 5 3 4 1 Merops inaliwbicus 8 9 9 A 2 8 8 2 1 5 5 6 4 B 2 Merops nubicus E E C B B F 9 2 1 2 6 2 2 5 2 Merops mtbicoides E E C B A 6 D 2 1 7 8 2 2 5 2

ui I APPENDIX B - Continued

SPECIES DATA MATRIX

Species Character Number 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 Nyctyornis amicta 0 0 0 0 0 0 0 0 A 0 0 0 0 0 1 Nyctyorni athertoni 1 0 1 4 1 0 A 1 1 1 0 0 0 0 1 Meropogon forsteni 2 2 0 2 2 0 2 2 2 2 1 1 1 1 0 Merops breiueri 2 2 0 3 3 1 3 3 C 9 1 1 1 1 1 Merops muelleri 4 3 23 5 3 0 2 4 4 6 0 2 2 2 2 Merops gtilaris 4 3 4 5 0 0 4B 9 B 8 01 3 3 3 3 Merops hirundineus 5 6 1 6 4 2 7 7 5 3 2 1 1 1 1 Merops pitsilliis 6 3 1 6 4 39A 5 5 C 4 3 0 0 0 1 Merops variegatus 6 3 1 6 4 SBC 5 36 6D 4 3 0 0 0 0 Merops oreobales 6 3 1 6 4 A 3 3 C 9 0 0 0 0 1 Merops bullocki 7 6 1 5 0 0 8 6 3 5 0 1 1 1 1 Merops bullockoides 7 6 1 7 0 0 8 6 3 5 0 0 0 0 1 Merops revoilii 8 5 1 7 5 0 8 6 3 3 0 0 0 0 1 Merops albicoUis 9 1 5 7 5 6 6 A 7 3 1 0 0 4 1 Merops boehmi 9 1 1 1 7 0 7 0 0 3 1 0 0 0 1 Merops orientalis 9 1 1 489 689 6D 7C 06C 5D 3AB 1 0 0 0 1 Merops persicus 8 5 1 6 7 0 C C D A 1 0 0 0 1 Merops superciliosus 8 5 6 A 7 0 7 C D A 1 5 5 6 1 Merops philippinus 8 1 6 6 7 0 7 C D 3 1 4 4 5 1 Merops ornatus 7 4 1 6 4 8 7 0 58 3 1 0 0 0 1 Merops viridis 7 1 1 48 68 0 C c 5 3 1 0 0 0 1 Merops leschenaulli 7 1 0 6 4 47 D 1 8 3C 0 1 1 1 1 Merops apiaster 8 5 78 6 4 7 1 7 5 3 1 1 26 27 0 Merops rnalimbicus 7 5 9 7 0 0 9 8 9 7 1 6 7 8 4 Merops mtbicus 3 5 A B A 0 9 B 9 3 1 7 8 9 5 Merops mibicoides 3 5 A 5 0 0 9 B 9 3 1 7 8 9 5 APPENDIX B - Continued CHARACTERS AND CHARACTER STATES (SPECIES MATRIX) Forehead: 6-orange 0- light blue 7- dark brown 1- yeUow and white 8- chestnut 2- deep blue-purple 9-gray 3- black a. pied 4- green with blue tips b. dark greenish-brown 5-green c. greenish-blue 6-white d. blue 7- pied e. purple 8-gray f. green and orange 9- chestnut g. light greenish brown a- light blue and green b- little light blue Superdlium c- green with orange 0. green d- white and light blue 1. light blue e- greenish blue 2. white 3. small blue Forecrown 4. white and blue 0- Lilac 5. green and lilac 1- green with blue tips 6. black 2- deep purple 7. deep purple 3- blade 8. very little blue 4- light blue 9. chestnut 5- green a. gray 6- white b. greenish-blue 7- dark brown 8- chestnut Infradlium 9-gray 0. green a. pied 1. light blue b. dark greenish-brown 2. white c. light greenish brown 3. light blue and white d. green with orange 4. very little blue e. greenish blue 5. green and red 6. deep purple Hindcrown 7. black 0- orange and green 8. yellow 1- green 9. white and yellow 2- deep purple and brown a. red 3-black b. greenish blue 4-blue 5- green with blue tips APPENDIX B - Continued CHARACTERS AND CHARACTER STATES (SPECIES MATRIX) Cheek 1. black 0. green 2. chestnut 1. deep purple 2. black Mantle 3. white and blue 0. green 4. white 1. chestnut 5. yellow 2. red 6. red 3. black 7. red and blue 4. green, blue tips 8. red and white 5. gray 9. blue and chestnut 6. light greenish-brown a. blue 7. orangish red b. blue and green c. white, yellow and chestnut Back d. yellow and blue 0. green e. white and yellow 1. green, blue tips f. greenish blue 2. chestnut 3. black Nape 4. yellow and chestnut 0. green 5. gray, slight red 1. dark brown 6. red 2. black 7. light greenish-brown 3. deep purple 8. orangish red 4. green with blue 5. orange Rump 6. orange-green 0. green 7. chestnut 1. chestnut-light blue 8. gray 2. blue 9. red 3. green, blue tips a. green with blue over orange 4. blue-purple b. dark greenish brown 5. yellow-green c. light greenish-brown 6. gray, slight red d. orangish red 7. light blue Lore Upper tail coverts 0. lilac and red 0. green 1. green 1. blue 2. black 2. light blue 3. black and chestnut 3. green, blue tips 4. grayish red Eyes tripe 5. deep purple 0. none 161

APPENDIX B - Continued CHARACTERS AND CHARACTER STATES (SPECIES MATRIX) First rectrix. dorsal 7. yellow-orange 0. green 8. yellow-orange with green 1. light blue, white tip 9. gray, slight red 2. dark blue to black . red 3. blue 4. blue-green Chi^ 5. red to black tip 0. blue then red 6. purple 1. chestnut 7. blue-green, white tip 2. black then purple 8. blue-green, black tip 3. black 9. blue, black tip 4. blue 10. green, black tip 5. red 11. reddish black . yellow 7. white First rectrix rachis. dorsal 8. blueish-green 0. black 9. green 1. brown a. yellow and white 2. dark brown b. greenish blue Outer rectrices. ventral Throat 0. prox. yellow, distal black 0. red 1. yellow 1. blue, green sides 2. orange-brown to brown (m to L) 2. deep purple 3. darker light brown 3. black 4. black 4. yellow 5. brown, white tips 5. white 6 p to d, orange-brown, black, white? 6. blueish-green 7. brown 7. chestnut 8. light brown 8. blue 9. light brown, dark tip 9. green . greenish blue Outer rectrix rachis. ventral 0. yellow gprggt 1. off-white 0. none 2. orange-brown 1. chestnut 3. tan to black distcdly 2. light blue, dark blue 4. off-white to brown distally 3. light blue, black, chestnut 5. tan 4. chestnut, black 6. Brown 5. white, light blue, dark blue, chestnut . light blue, black, light blue Scapulars 7. black 0. green 8. chestnut, black, blue 1. green with blue 9. white, light blue, black, chestnut 2. chestnut a. light blue, dark blue, black, chestnut 3. chestnut-orange b. white, light blue, dark blue-black, 4. black chestnut 5. green with orange c. light blue, dark blue, chestnut 6. greeriish brown d. green-blue, black, green-blue APPENDIX B - Continued CHARACTERS AND CHARACTER STATES (SPECIES MATRIX) Breast d. green with blue 0. red, green sides 1. greenish-blue Undertail coverts and vent 2. deep purple 0. light green 3. yeUow-orange 1. yellow 4. black and irrid. blue 2. burnt orange, green tips 5. orange-yellow-green 3. light blue 6. light green 4. buff 7. green 5. blue-purple 8. orange-brown 6. blue 9. red 7. redish-gray a. blue, green sides 8. irrid. blue b. black and irrid. blue with red 9. orange, green tips c. green with blue a. green with blue d. yellowish green b. greenish blue c. light green-blue Flank 0. green Tail shape 1. green and yellow 0. Square 2. brown-gray 1. streamer 3. yellow-orange 2. forked 4. blue 3. slight fork 5. orange-yellow-green 6. orange-brown Marginal coverts 7. greenish-blue 0. green, blue tips 8. red 1. green 9. irrid. blue and black 2. chestnut a. light green 3. black b. orange-red 4. light greenish-brown c. green with blue 5. dark greenish-brown 6. gray Bellv 7. red 0. green 1. green and yellow Lesser coverts 2. brown-gray 0. green, blue tips 3. buff 1. green 4. blue 2. chestnut 5. greenish-blue 3. black 6. yellow-green 4. light greenish-brown 7. greenish-white 5. dark greenish-brown 8. light green-blue 6. green with chestnut 9. red 7. gray a. light green 8. red b. irrid. blue c. orange-yellow-green APPENDIX B - Continued CHARACTERS AND CHARACTER STATES (SPECIES MATRIX) Median and greater coverts 0. green, blue tips 1. green 2. chestnut 3. black 4. green with orange 5. light greenish-brown 6. dark greenish-brown 7. green with chestnut 8. gray 9. red Tertials 0. green 1. green, blue tips 2. chestnut 3. black 4. gray 5. red, blue tips APPENDIX B - Continued RACE DATA MATRIX Race Character Number 12345678910 Nyctyomis amicta amicta 0 0 1 6 5 0 0 0 0 0 Nyctyomis athertoni athertoni 0 4 1 0 0 0 4 1 0 4 Meropogon forsteni forsteni 2 2 2 7 6 1 1 2 0 0 Merops breweri breweri 3 3 3 8 7 2 2 2 0 0 Merops muelleri muelleri 0 4 4D 4 7 2 3 2 0 1 Merops muelleri mentalis 0 4 E 1 7 2 3 2 0 1 Merops gularis gularis 0 3 3 1 7 2 2 2 0 3 Merops gularis australis B 3 3 5 7 2 2 2 0 3 Merops hirundineus hirundineus 5 5 1 0 3 3 0 2 1 0 Merops hirundineus chrysolaimus 0 5 1 1 3 3 0 2 1 0 Merops hirundineus heuglini 5 5 1 0 3 3 0 2 1 0 Merops pusillus pusillus 4 1 5 5 8 5 4 2 1 4 Merops pusillus meridionalis 4 1 5 3 8 5 4 2 1 4 Merops pusillus cyanostictus 0 1 5 1 8 5 4 2 1 4 Merops pusillus ocularis 4 1 5 3 8 5 4 2 1 4 Merops variegatus variegatus 4 1 5 5 9 4 4 2 1 4 Merops variegatus loringi 4 1 5 1 9 4 4 2 1 4 Merops variegatus bangweoloensis 4 1 5 5 9 4 4 2 1 4 Merops variegatus lafresnayii 0 1 5 1 9 4 4 2 1 4 Merops oreobates oreobates B 1 5 1 9 4 4 2 1 4 Merops bullocid bullocki 4 1 5 05 04 6 5 2 1 4 Merops bullockifrenatus 0 1 5 1 1 7 5 2 1 4 Merops bullockoides bullockoides 6 6 6 1 2 8 5 2 1 4 Merops revoilii revoilii 7 A A 1 2 4 A 2 1 4 Merops albicollis albicollis 6 7 7 2 2 4 6 2 1 4 Merops boehmi boehmi 9 8 8 9 1 9 4 2 1 4 Merops orientalis orientalis C C F 0 1 A 6 2 1 4 Merops orientalis viridissimus 5 5 1 0 4 0 0 2 1 0 Merops orientalis cleopatra 5 5 1 0 4 0 0 2 1 4 Merops orientalis cyanophrys 0 4 1 1 1 A 0 2 1 4 Merops orientalis beludschicus c C F 0 1 A 6 2 1 4 Merops orientalis ferrugeiceps 9 8 8 0 1 A 7 2 1 1 Merops persicus persicus D 4 5 4 3 B 4 2 1 4 Merops persicus chrysocercus D 4 1 4 3 B 0 2 1 0 Merops superciliosus superciliosus 6 B B 2 2 C B 2 1 6 Merops philippinus philrppinus A D G 1 1 9 C 2 1 6 Merops omatus omatus 4 1 0 0 1 D 5A 2 1 4 Merops viridis viridis E E H A 1 A D 2 1 7 Merops viridis americanus 9 8 8 9 1 A 7 2 1 1 Merops leschenaulti leschenaulti 9 8 8 9 8 5 7 2 1 1 Merops leschenaulti quinticolor 9 8 8 9 8 5 7 3 2 1 Merops leschenaulti andamanensis 9 8 8 9 8 5 7 3 3 1 Merops apiaster apiaster 61 1 8 4 3 E 7 2 1 1 Merops malimbiais malimbicus 8 9 9 B 2 8 8 2 1 5 Merops nubicus nubicus F F C C B F 9 2 1 2 Merops nubicoides nubicoides F F C C A 6 E 2 1 8 APPENDIX B (continued) RACE DATA MATRIX Race Character Number 11 12 13 14 15 16 17 18 19 20 Nyctyomis amicta amicta 0 0 0 0 0 0 0 0 0 0 Nyctyomis athertoni athertoni 1 3 3 0 2 1 0 1 4 1 Meropogon forsteni forsteni 0 0 0 0 2 2 2 0 2 2 Merops brewed breweri 1 0 0 0 2 2 2 0 3 3 Merops muelleri muelleri 2 1 5 2 0 4 3 2 5 3 Merops muelleri mentalis 2 7 5 6 7 4 7 3 5 3 Merops gularis gularis 3 6 2 2 0 4 3 4 5 0 Merops gularis australis 3 6 2 2 7 4 7 4 5 0 Merops hiruridineus hirundineus 0 6 2 1 2 5 6 1 6 4 Merops hirundineus dirysolaimus 0 2 1 4 7 5 7 1 6 4 Merops hirundineus heuglini 0 2 1 4 7 5 7 1 6 4 Merops pusillus pusillus 1 3 3 0 1 6 3 1 6 4 Merops pusillus meridionalis 1 3 3 0 2 6 3 1 6 4 Merops pusillus ojanostictus 1 3 3 0 7 6 7 1 6 4 Merops pusillus ocularis 1 3 3 0 7 6 7 1 6 4 Merops variegatus variegatus 1 3 3 0 2 6 3 1 6 4 Merops variegatus loringi 1 3 3 0 7 6 7 1 6 4 Merops variegatus bangweoloensis 1 3 3 0 7 6 7 1 6 4 Merops variegatus lafresnayii 1 3 3 0 7 6 7 1 6 4 Merops oreobates oreobates 1 3 3 0 0 6 3 1 6 4 Merops bullocki bullocki 1 3 3 0 1 7 6 1 5 0 Merops bullocki frenatus 1 3 3 0 7 7 7 1 5 0 Merops bullockoides bullockoides 1 3 1 7 2 7 6 I 7 0 Merops revoilii revoilii 1 6 2 7 1 8 5 1 7 5 Merops albicollis albicollis 1 3 2 8 0 9 1 7 5 Merops boehmi boehmi 1 3 3 8 2 9 1 1 1 7 Merops orientalis orientalis 1 3 3 9 2 9 1 1 4 68 Merops orientalis viridissimus 0 0 9 7 9 7 1 9 9 Merops orientalis cleopatra 1 3 3 9 7 9 7 1 9 9 Merops orientalis cyanophrys 1 3 3 9 7 9 7 1 4 1 Merops orientalis beludschicus 1 3 3 9 7 9 7 1 4 6 Merops orientalis ferrugeiceps 1 3 3 9 7 9 7 1 8 6 Merops persicus persicus 1 3 3 9 1 8 5 1 6 7 Merops persicus chrysocercus 0 0 9 1 8 7 1 6 7 Merops superciliosus superciliosus 1 3 3 0 1 8 5 A 7 Merops philippinus philippinus 7 2 2 48 1 8 1 6 7 Merops onwtus omatus 1 2 2 A 2 7 4 1 6 4 Merops viridis viridis 1 2 2 3 1 7 1 1 4 8 Merops viridis americanus 1 2 2 3 7 7 7 1 8 6 Merops leschenaulti leschenaulti 0 2 2 07 1 7 1 0 6 4 Merops leschenaulti quinticolor 0 2 2 3 7 7 7 0 6 4 Merops leschenaulti andamanensis 0 2 2 04 7 7 7 0 6 4 Merops apiaster apiaster 4 4 3 7 1 8 5 87 6 4 Merops malimbicus malimbicus 5 5 4 5 2 7 5 9 7 0 Merops nubicus nubicus 6 2 2 B 2 3 5 A B A Merops nubicoides nubicoides 8 2 2 B 2 3 5 A 5 0 APPENDIX A (continued) RACE DATA MATRIX (continued) Race Character Number 21 22 23 24 25 26 27 28 29 30 Nyctyomis amicta amicta 0 0 0 A 0 0 0 0 0 1 Nyctyomis athertoni athertoni 0 A 1 1 1 0 0 0 0 1 Meropogon forsteni forsleni 0 2 2 2 2 1 1 1 1 0 Merops breweri breweri 1 3 3 B 9 1 1 1 1 1 Merops mttelleri muelleri 0 2 4 4 6 0 2 2 2 2 Merops muelleri mentalis 0 2 4 4 6 1 2 2 2 2 Merops gularis gularis 0 4 9 C 8 0 3 3 3 3 Merops gularis australis 0 B 9 c 8 0 3 3 3 3 Merops hirundineus hirundineus 2 7 7 5 3 2 1 1 1 1 Merops hirundineus chrysolaimus 2 7 7 5 3 2 1 1 1 1 Merops hirundineus heuglini 2 7 7 5 3 2 1 1 1 1 Merops pusillus pusillus 3 5 5 B 4 3 0 0 0 1 Merops pusillus meridionalis 9 5 5 B 4 3 0 0 0 1 Merops pusillus cyanostictus A 5 5 B 4 3 0 0 0 1 Merops pusillus ocularis 9 5 5 B 4 3 0 0 0 1 Merops variegatus variegatus 5 5 5 6 4 3 0 0 0 0 Merops variegatus loringi 5 5 5 6 4 3 0 0 0 0 Merops variegatus bangweoloensis B 5 5 6 4 3 0 0 0 0 Merops variegatus lafresnayii C 5 3 B 4 3 0 0 0 0 Merops oreobates oreobates A 3 3 B 9 0 0 0 0 1 Merops bullocki bullocki 0 8 6 3 5 0 1 1 1 1 Merops bullocki frenatus 0 8 6 3 5 0 1 1 1 1 Merops bullockoides bullockoides 0 8 6 3 5 0 0 0 0 1 Merops revoilii revoilii 0 8 6 3 3 0 0 0 0 1 Merops albicollis albicollis 6 6 A 7 3 1 0 0 4 1 Merops boehnti boehmi 0 7 0 0 3 1 0 0 0 1 Merops orientalis orientalis D C C 5 B 1 0 0 0 1 Merops orientalis viridissimus D 7 0 D A 1 0 0 0 1 Merops orientalis Cleopatra D C C 5 B 1 0 0 0 1 Merops orientalis cyanophrys 6 C C 5 3 1 0 0 0 1 Merops orientalis beludschicus D c C 5 B 1 0 0 0 1 Merops orientalis ferrugeiceps D c 5 5 B 1 0 0 0 1 Merops persicus persicus 0 c C D A 1 0 0 0 1 Merops persicus chrysocercus 0 c C D A 1 0 0 0 1 Merops superciliosus superciliosus 0 7 c D A 1 5 5 6 1 Merops philippinus philippinus 0 7 c D 3 1 4 4 5 1 Merops omatus omatus 8 7 0 58 3 1 0 0 0 1 Merops xnridis viridis 0 C c 5 3 1 0 0 0 1 Merops viridis americanus 0 C c 5 3 1 0 0 0 1 Merops leschenaulti leschenaulti 4 D 1 8 C 0 1 1 1 1 Merops leschenaulti cjuinticolor 7 D 1 8 3 0 1 1 1 1 Merops leschenaulti andamanensis 4 D 1 8 C 0 1 1 1 1 Merops apiaster apiaster 7 1 7 5 3 1 1 62 72 0 Merops malimbicus malimbicus 0 9 8 9 7 1 6 7 8 4 Merops nubicus nubicus 0 9 B 9 3 1 7 8 9 5 Merops nubicoides nubicoides 0 9 B 9 3 1 7 8 9 5 APPENDIX B - Continued CHARACTERS AND CHARACTER STATES (RACE MATRIX) Forehead Hindcrown 0- light blue 0- orange and green 1- yellow and white 1-green 2- deep blue-purple 2- deep purple and brown 3-black 3-black 4- green with blue tips 4- light blue 5- green 5- green with blue tips 6-white 6-orange 7- pied 7- dcirk brown 8-gray 8- chestnut 9- chestnut 9-gray a- light blue and green a. pied b- little light blue b. dark greenish-brown c- green with orange c. greenish-blue d- white and light blue d. blue e- red chestnut e. purple f- greenish blue f. green with orange g. light greenish brown Forecrown h. red chestnut 0- Lilac 1- green with blue tips Superdlium 2- deep purple 0-green 3-black 1- light blue 4- light blue 2- white 5-green 3- small blue 6- white 4- white and blue 7- dark brown 5- very little blue 8- chestnut 6- green and lilac 9-gray 7- deep purple a. pied 8-black b. dark greenish-brown 9- chestnut c. green with orange a. red chestnut d. light greenish brown b.gray e. red chestnut c. greenish-blue f. greenish blue APPENDIX B - Continued CHARACTERS AND CHARACTER STATES (RACE MATRIX) Infr^alivm e. orangish red 0. green 1. light blue Lore 2. white 0. lilac and red 3. light blue and white 1. green 4. very little blue 2. black 5. green and red 3. black and chestnut 6. deep purple 7. black Eyestripe 8. yellow 0. none 9. white and yellow 1. black a. red 2. chestnut b. greenish blue 3. black and chestnut Cheek Mantle 0. green 0. green 1. deep purple 1. chestnut 2. black 2. red 3. white and blue 3. black 4. white 4. green, blue tips 5. yellow 5. gray 6. red 6. light greenish-brown 7. red and blue 7. red chestnut 8. red and white 8. orangish red 9. blue and chestnut a. blue Back b. blue and green 0. green c. white, yellow and chestnut 1. green, blue tips d. yeUow and blue 2. chestnut e. white and yellow 3. black f. greertish blue 4. yellow and chestnut 5. gray, slight red Nape 6. red 0. green 7. light greenish-brown 1. dcU'k brown 8. orangish red 2. black 3. deep purple Rump 4. green with blue 0. green 5. orange 1. chestnut-light blue 6. orange-green 2. blue 7. chestnut 3. green, blue tips 8. gray 4. yellow-green 9. red 5. gray, slight red a. green with blue over orange 6. light blue b. dark greenish brown 7. blue-purple c. light greenish-brown d. red chestnut APPENDIX B - Continued CHARACTERS AND CHARACTER STATES (RACE MATRIX) Upper tail coverts 4. off-white to brown distally 0. green 5. tan 1. blue 6. Brown 2. light blue 3. green, blue tips S<;apyl?T5 4. gra)Tsh red 0. green 5. deep purple 1. green with blue 2. chestnut First rectrix. dorsal 3. chestnut-orange 0. green 4. black 1. light blue, white tip 5. green with orange 2. dark blue to black 6. greenish brown 3. blue 7. yellow-orange 4. blue-green, white tip 8. yellow-orange with green 5. reddish black 9. gray, slight red 6. purple a. red 7. blue-green 8. blue-green, black tip Chin 9. green, black tip 0. Blue then red 10. blue, black tip 1. chestnut 11. red to black tip 2. black then purple 3. black First rectrix rachis, dorsal 4. blue 0. black 5. red 1. brown 6. yellow 2. dark brown 7. white 8. blueish-green Outer rectrix . ventral 9. green 0. prox. yellow, distal black a. yeUow and white 1. yellow b. greenish blue 2. orange-brown to brown m to L 3. darker light brown Throat 4. black 0. red 5. brown, white tips 1. blue, green sides 6 p to d, orange-brown, black, white? 2. deep purple 7. brown 3. black 8. light brown 4. yellow 9. light brown, dark tip 5. white 6. blueish-green Outer rectrix rachis. ventral 7. chestnut 0. yellow 8. blue 1. off-white 9. green 2. orange-brown a. greenish blue 3. tan to black distally APPENDIX B - Continued CHARACTERS AND CHARACTER STATES (RACE MATRIX) Gorget Belly 0. none 0. green 1. chestnut 1. green and yellow 2. light blue, dark blue 2. brown-gray 3. light blue, black, chestnut 3. buff 4. chestnut, black 4. blue 5. white, light blue, dark blue, chestnut 5. greenish-blue 6. light blue, black, light blue 6. yellow-green 7. black 7. greenish-white 8. chestnut, black, blue 8. light green-blue 9. white, light blue, black, chestnut 9. red a. light blue, dark blue, black, chestnut a. light green b. white, light blue, dark blue-black. b. orange-yellow-green chestnut c. irrid. blue c. light blue, dark blue, chestnut d. green with blue d. green-blue, black, green-blue Undertail coverts and vent Breast 0. light green 0. red, green sides 1. yellow 1. greenish-blue 2. burnt orange, green tips 2. deep purple 3. light blue 3. yellow-orange 4. buff 4. black and irrid. blue 5. blue-purple 5. orange-yellow-green 6. blue 6. light green 7. redish-gray 7. green 8. irrid. blue 8. orange-brown 9. orange, green tips 9. red a. green with blue a. blue, green sides b. greenish blue b. black and irrid. blue with red c. light green-blue c. green with blue d. yellowish green Tail shape 0. Square Flank 1. streamer 0. green 2. forked 1. green and yellow 3. slight fork 2. brown-gray 3. yellow-orange Marginal coverts 4. blue 0. green, blue tips 5. orange-yellow-green 1. green 6. orange-brown 2. chestnut 7. greenish-blue 3. black 8. red 4. light greenish-brown 9. irrid. blue and black 5. dark greenish-brown a. light green 6. gray b. orange-red 7. red c. green with blue 171

APPENDIX B - Continued CHARACTERS AND CHARACTER STATES (SPECIES MATRIX) Lesser coverts 0. green, blue tips 1. green 2. chestnut 3. black 4. light greenish-brown 5. dark greenish-brown 6. green with chestnut 7. gray 8. red Median and Greater coverts 0. green, blue tips 1. green 2. chestnut 3. black 4. green with orange 5. light greenish-brown 6. dark greenish-brown 7. green with chestnut 8. gray 9. red Tertials 0. green 1. green, blue tips 2. chestnut 3. black 4. gray 5. red, blue tips 172

APPENDIX C

SOURCES FOR TRACED CHARACTERS

Breeding Social Migration Foraging/ System System Nesting Habitat Behavior Diet Nyctyomis amicta 1 1 1 1 1 1 Nyctyomis athertoni 1,2 1 1 1 1 1 Meropogon forsteni 1 1,2 1 1 1 1 Merops breweri 1 1 1 1 1 1 Merops muelleri 1 1,2 1 1 1 1 Merops gularis 1 1 1 1 1 1 Merops hirundineus 1 1 1 1 1 1 Merops pusillus 3 1 1 1 1 1 Merops variegatus 4 1,4 1 1 1 1 Merops oreobates 5 1 1 1 1 1 Merops bullocki 6 1,13 1 1 1 1 Merops bullockoides 7 1,7 1 1 1 1 Merops revoilii 1 1 1 1 1 1 Merops albicollis 8 1,8 1,8 1 1 1 Merops boehmi 1 1 1 1 1 1 Merops orientalis 2,9 1,2,9 1 1 1 1 Merops persicus 1 1 1 1 1 1 Merops superciliosus 1 1 1 1 1 1 Merops philippinus 2 1,2 1 1 1 1 Merops omatus 10 1,10 1 1 1 1 Merops viridis 11 1 1 1 1 1 Merops leschenaulti 2,9 1,2 1 1 1 1 Merops apiaster 12 1 1 1 1 1 Merops malimbicus 1 1 1 1 1 1 Merops nubicus 1 1 1 1 1 1 Merops nubicoides 5 1 1 1 1 1

1= Fry 1984,2= Chapter 3,3= Douthwaite 1986,4= Gartshore 1984,5= Emlen, pers. com., 6= Fry 1972,7= Emlen 1990,8= Dyer and Crick 1983,9= Sridhar and Karanth 1993,10= Filewood et al. 1978,11= Bryant and Tatner 1990,12= Lessells and Krebs 1989,13= Fry 1977. 173

WORKS CITED

Austad, S. N., and K. N. Rabenold. 1985. Reproductive erihancement by helpers and an experimental inquiry into its mechanism in the bicolored wren {Campylorhynchus griseus). Behav. Ecol. and Sociobiol. 17:19-28.

Boetticher, H. von 1951. La systematique des guepiers. Oiseau Rev. Fr. Om. 5:194-199.

Brooks, D. R., and D. A. McLennan. 1991. Phylogeny, Ecology and Behavior: A Research Program in Comparative Biology. Univ. Chicago Press, Chicago.

Brown, J. L. 1974. Alternative routes to sociality in jays- with a theory for the evolution of altruism and communal breeding. Am. Zool. 14:6^80.

Brown, J. L. 1978. Avian communal breeding systems. Arm. Rev. Ecol. Syst. 9:123-155.

Brown, J. L. 1987. Helping and communal breeding in birds: ecology and evolution. Princeton University Press, Princeton, N.J.

Bryant, D. M., and P. Tatner. 1990. Hatching asynchrony sibling competition and siblicide in nestling birds studies of swiftlets and bee-eaters. Anim. Behav. 39:657-671.

Burt, D. B. 1996. Habitat-use patterns in cooperative and non-cooperative breeding birds: testing predictions witii western scrub-jays {Aphelocoma califomica). Wil. Bull. 108:in press.

Cracraft,J. 1981. Toward a phylogenetic classification of the recent birds of the world (Class Aves). Auk 98:681-714.

Craig, J. L., and I. G. Jamieson. 1990. Pukeko: different approaches and some different answers. Pages 387-412 in Cooperative breeding in birds: long- term studies of ecology and behavior (P. B. Stacey and W. D. Koenig, eds.). Cambridge University Press, New York.

Cvirry, R. L., and P. R. Grant. 1990. Galapagos mockingbirds: territorial cooperative breeding in a climatically variable environment. Pages 291-331 in Cooperative breeding in birds: long-term studies of ecology and behavior (P. B. Stacey and W. D. Koenig, eds.). Cambridge University Press, New York. 174

Davies, N. B. 1990. Duimocks: cooperation and conflict among males and females in a variable mating system. Pages 457-485 in Cooperative breeding in birds: long-term studies of ecology and behavior (P. B. Stacey and W. D. Koenig, eds.). Cambridge University Press, New York.

Dawson, J. W., and R. W. Mannan. 1991. Dominance hierarchies and helper contributions in Harris' hawks. Auk 108:649-660. de Queiroz, A., and P. Wimberger. 1993. The usefulness of behavior for phylogeny estimation: Levels of homoplasy in behavioral and morphological characters. Evol. 47:46-60.

Donoghue, M. J. 1989. Phylogenies and the analysis of evolutionary sequences, with examples from seed plants. Evol. 43:1137-1156.

Douthwaite, R. J. 1986. Effects of drift sprays of endosulfan, applied for tsetse fly control, on breeding little bee-eaters {Merops pusillus) in Somalia. Envir. Poll. Series A Ecol. and Biol. 41:11-22.

Dow, D. D. 1980. Commxmally breeding Australian birds with an analysis of distributional environmental factors. Emu 80:121-140.

Du Plessis, M. A., W. R. Siegfried, and A. J. Armstrong. 1995. Ecological and life- history correlates of cooperative breeding in South African birds. Oecologia 102:180-188.

Dyer, M. 1983. Effect of nest helpers on growth of red-throated bee-eaters Merops bullocki. Ostrich 54:43-46.

Dyer, M., and H. Q. P. Crick. 1983. Observations on white-throated bee-eaters Merops albicollis breeding in Nigeria. Ostrich 54:52-55.

Edwards, S. V., and S. Naeem. 1993. The phylogenetic component of cooperative breeding in perching birds. Am. Nat. 141:754-789.

Edwards, S. V., and S. Naeem. 1994. Homology and comparative methods in the study of avian cooperative breeding. Am. Nat. 143:723-733.

Emlen, S. T. 1982a. The evolution of helping I. An ecological constraints model. Am. Nat. 119:29-39. Emlen, S. T. 1982b. The evolution of helping n. The role of behavioral conflict. Am. Nat. 119:40-53. 175

Emlen, S. T. 1984. Cooperative breeding in birds and mammals. Pages 305-339 in Behavioural ecology: an evolutionary approach (J. R. Krebs and N. B. Davies, eds.). Blackwell Scientific, Oxford.

Emlen, S. T. 1990. White-fronted bee-eaters: helping in a colonially nesting species. Pages 489-526 in Cooperative breeding in birds (P. B. Stacey and W. D. Koenig, eds.). Cambridge University Press, Cambridge.

Emlen, S. T. 1991. Cooperative breeding in birds and mammals. Pages 301-337 in Behavioural ecology: an evolutionary approach (J. R. Krebs and N. B. Davies, eds.). Blackw^ Scientific, Oxford.

Emlen, S. T., and P. H. Wrege. 1986. Forced copulations and intra-specific parasitism: Two costs of social living in the white-fronted bee-eater. Ethology 71:2-29.

Emlen, S. T., and P. H. Wrege. 1988. The role of kinship in helping decisions among white-fronted bee-eaters. Behav. Ecol. and Sociobiol. 23:305-316.

Emlen, S. T., and P. H. Wrege. 1989. A test of altemate hypotheses for helping behavior in white-fronted bee-eaters of Kenya. Behav. Ecol. Sociobiol. 25:303-320.

Emlen, S. T., and P. H. Wrege. 1991. Breeding biology of white-fronted bee- eaters at Nakuru (Kenya): The influence of helpers on breeder fitness. J. Ariim. Ecol. 60:309-326.

Emlen, S. T., and P. H. Wrege. 1992. Parent-offspring conflict and the recruitment of helpers among bee-eaters. Nature 356:331-333.

Emlen, S. T., P. H. Wrege, N. J. Demong, and R. E. Hegner. 1991. Flexible growth rates in nestling white-fronted bee-eaters: a possible adaptation to short-term food shortage. Condor 93:591-597.

Farris, J. S. 1969. A successive approximations approach to character weighting. Syst. Zool. 18:374-385.

Farris, J. S. 1983. The logical basis of phylogenetic analysis. Pages 7-36 in Advances in Cladistics, Volume 2. Proceedings of the Second Meeting of the Willi Hennig Society (N. I. Platnick and V. A. Fimk, eds.). Columbia Univerisity Press, New York.

Farris, J. S. 1989. The retention index and the rescaled consistency index. Cladistics 5:417-419. 176

Felsenstein, J. 1981. A likelihood approach to character weighting and what it tells us about parsimony and compatibility. Biol. J. Liimean Soc. 16:183- 196.

Felsenstein, J. 1985a. Phylogenies and the comparative method. Am. Nat. 125:1- 15.

Felsenstein, J. 1985b. Confidence limits on phylogenies with a molecular clock. Syst. Zool. 34:152-161.

Felsenstein, J. 1985c. Confidence limits on phylogenies: An approach using the bootstrap. Evol. 39:783-791.

Felsenstein, J. 1988. Phylogenies from molecular sequences: iriference and reliability. Arm. Rev. Genet. 22:521-565.

Filewood, L. W. C., K. Hough, I. C. Morris, and D. E. Peters. 1978. Helpers at the nest of the rainbow bee-eater. Emu 78:43-44.

Fitch, W. M. 1984. Cladistics and other methods: problems, pitfalls, and potentials. Pages 221-252 in Cladisitcs: perspectives on the reconstruction of evolutionary history (T. Duncan and T. F. Stuessy, eds.). Columbia University Press, New York.

Ford, H. A., H. Bell, R. Nias, and R. Noske. 1988. The relationship between ecology and the incidence of cooperative breeding in Australian birds. Behav. Ecol. and Sociobiol. 22:239-250.

Fry, C. H. 1969. The evolution and systematics of bee-eaters (Meropidae). Ibis 111:557-592.

Fry, C. H. 1970. Convergence between jacamars and bee-eaters. Ibis 112:257-259.

Fry, C. H. 1972. The social organization of bee-eaters (Meropidae) and cooperative breeding in hot-climate birds. Ibis 114:1-14.

Fry, C. H. 1977. The evolutionary significance of co-operative breeding in birds. Pages 127-135 in Evolutionary Ecology (B. Stonehouse and C. Perrins, eds.). Macmillan, London.

Fry, C. H. 1984. Thebee-eaters.Poyser,Calton [Staffordshire].

Fry, C. H., and K. Fry. 1992. Kingfishers, bee-eaters and rollers: a handbook. Princeton University Press, Princeton. 177

Gartshore, M. E. 1984. Notes on the nesting of two little-known species of bee- eaters in Cameroon. Malimbus 6:95-96.

Gould, S. J., and R. C. Lewontin. 1979. The spandrels of San Marcos and the Panglossian paradigm: a critique of the adaptationist programme. Proc. Royal Soc. Lond. 205:581-598.

Harvey, P. H., and M. D. Pagel. 1991. The Comparative Method in Evolutionary Biology. Oxford University Press, Oxford, U. K.

Heinsohn, R. G. 1991. Slow learning of foraging skills and extended parental care in cooperatively breeding white-winged choughs. Am. Nat. 137:864- 881.

Heinsohn, R. G., A. Cockbum, and R. A. Mulder. 1990. Avian cooperative breeding: old hypotheses and new directions. TREE 5:403-407.

Hillis^ D. M. 1991. Discriminating between phylogenetic signal and random noise in DNA sequences. Pages 278-294 in Phylogenetic Analysis of DNA Sequences (M. M. Miyamoto and J. Cracraft, eds.). Oxford Univ. Press, New York.

Hillis, D. M., and J. J. Bull. 1993. An empirical test of bootstrapping as a method for assessing confidence in phylogenetic analysis. Syst. Biol. 42:182-192.

Hillis, D. M., and J. P. Huelsenbeck. 1992. Signal, noise, and reliability in molecular phylogenetic analyses. J. Hered. 83:189-195.

Huelsenbeck, J. P. 1991. Tree-length distribution skewness: an indicator of phylogenetic information. Syst. Zool. 40:257-270.

Jamieson, I. G. 1986. The functional approach to behavior: is it useful? Am. Nat. 127:195-208.

Jamieson, I. G. 1989. Behavioral heterochrony and the evolution of birds' helping at the nest: An unselected consequence of communal breeding? Am. Nat. 133:394-406.

Jamieson, I. G., and J. L. Craig. 1987. Critique of helping behaviour in birds: a departure from functional explanations. Perspect. Ethol. 7:79-98.

Jones, C. S., C. M. Lessels, and J. R. Krebs. 1991. Helpers-at-the-nest in European bee-eaters (Merops apiaster): a genetic analysis. Pages 169-192 in DNA fingerprinting approaches and applications (T. Burke, G. Dolf, A. J. Jeffreys & R. Wolff, eds.). Birkhauser Verlag, Basel Switzerland. 178

Kim, J. 1993. Improving the acairaqr of phylogenetic estimation by combining different methods. Syst. Biol. 42:331-340.

BQuge, A. G., and J. S. Farris. 1969. Quantitative phyletics and the evolution of anurans. Syst. Zool. 18:1-32.

Koenig, W. D., and F. A. Pitelka. 1981. Ecological factors and kin selection in the evolution of cooperative breeding in birds. Pages 261-280 in Natural selection and social behavior: recent research and new theory (R. D. Alexander and D. W. Tinkle, eds.). Qiiron Press, New York.

Koenig, W. D., and P. B. Stacey. 1990. Acom woodpeckers: group living ad food storage under contrasting ecological conditions. Pages 415-453 in Cooperative breeding in birds: long-term studies of ecology and behavior (P. B. Stacey and W. D. Koenig, eds.). Cambridge University Press, New York.

Koenig, W. D., F. A. Pitelka, W. J. Carmen, R. L. Mumme, and M. T. Stanback. 1992. The evolution of delayed dispersal in cooperative breeders. Quart. Rev. Biol. 67:111-150.

Komdeur, J. 1994. Experimental evidence for helping and hindering by previous offspring in the cooperative-breeding Seychelles warbler Acrocephalus sechellensis. Behav. Ecol. and Sodobiol. 34:175-186.

Le Quesne, W. J. 1989. Frequency distributions of lengths of possible networks from a data matrix. Cladistics 5:395-407.

Lessells, C. M., and J. R. Krebs. 1989. Age and breeding performance of european bee-eaters. Auk 106:375-382.

Lessells, C. M., M. L. Avery, and J. R. Krebs. 1994. Nonrandom dispersal of kin: Why do European bee-eater {Merops apiaster) brothers nest close together? Behav. Ecol. 5:105-113.

Ligon, J. D. 1993. The role of phylogenetic history in the evolution of contemporary avian mating and parental care systems. Cur. Omith. 10:1- 46.

Ligon, J. D., and S. H. Ligon. 1990. Green woodhoopoes: life history traits and sociality. Pages 33-65 in Cooperative breeding in birds: long-term studies of ecology and behavior (P. B. Stacey and W. D. Koenig, eds.). Cambridge LFniversity Press, New York. 179

Lorenz, K. 1970. Studies in animal and human behavior. Harvard Univ. Press, Cambridge.

Maddison, D. R. 1991. The discovery and importance of multiple islands of most-parsimoruous trees. Syst. Zool. 40:315-328.

Maddison, W. P. 1990. A method for testing the correlated evolution of two binary characters: are gains or losses concentrated on certain branches of a phylogenetic tree? Evol. 44:539-557.

Maddison, W. P., and D. R. Maddison. 1992. MacClade: Analysis of Phylogeny and Character Evolution (Version 3.0). Sunderland, Massachusetts, Sinauer Associates.

Mays, N. A., C. M. Vleck, and J. Dawson. 1991. Plasma luteinizing hormone, steroid hormones, behavioral role, and nest stage in cooperatively breeding Harris' Hawks {Parabuteo unicinctus). Auk 108:619-637.

McKitrick, M. C. 1993. Phylogenetic coristraint in evolutionary theory: has it any explanatory power? Ann. Rev. Ecol. Syst. 24:307-330.

Mvimme, R. L., W. D. Koenig, and F. A. Pitelka. 1983. Mate guarding in the acom woodpecker: within-group reproductive competition in a cooperative breeder. Anim. Behav. 31:1094-1106.

Paterson, A. M., G. P. Wallis, and R. D. Gray. 1995. Penguins, petrels, and parsimony: does cladistic analysis of behavior reflect seabird phylogeny. Evol. 49:974-989.

Peterson, A. T., and D. B. Burt. 1992. Phylogenetic history of social evolution and habitat use in the Aphelocoma jays. Ariim. Behav. 44:859-866.

Poiaiu, A., and T. Fletcher. 1994. Plasma-levels of androgens and gonadal development of breeders and helpers in the bell miner [Manorina melanophrys). Behav. Ecol. and Sociobiol. 34:31-41.

Rabenold, K. N. 1990. Campylorhynchus wrer\s: the ecology of delayed dispersal and cooperation in the Venezuelan savanna. Pages 159-196 in Cooperative breeding in birds: long-term studies of ecology and behavior (P. B. Stacey and W. D. Koenig, eds.). Cambridge University Press, New York.

Reyer, H. U., J. P. Dittami, and M. R. Hall. 1986. Avian helpers at the nest: are they psychologically castrated? Ethology 71:216-228. 180

Ridley, M. 1983. The Explanation of Organic Diversity. Oxford University Press, New York.

Rowley, I., and E. Russell. 1990. Splendid fairy-wrens: demonstrating the importance of longevity. Pages 3-30 in Cooperative breeding in birds: long-term studies of ecology and behavior (P. B. Stacey and W. D. Koenig, eds.). Cambridge University Press, New York.

Russell, E. M. 1989. Co-operative breeeding - a Gondwanan perspective. Emu 89:61-62.

Schoech, S. J., R. L. Mumme, and M. C. Moore. 1991. Reproductive endocrinology and mechariisms of breeding inhibition in cooperatively breeding Florida Scrub Jays {Aphelocoma c. coerulescens). Condor 93:354- 364.

Selander, R. K. 1964. Spedation in wrens of the genus Campylorhynchus. Univ. Cal. Pub. Zool. 74:1-224.

Sherman, P. W. 1988. The levels of analysis. Anim. Behav. 36:616-618.

Sibley, C. G., and J. E. Ahlquist. 1990. Phylogeny and classification of birds: A study in molecular evolution. Yale University Press, New Haven.

Sibley, C. G., and Monroe, B. L. 1990. Distribution and taxonomy of the birds of the world. New Haven, Yale University Press.

Siegel-Causey, D., and S. P. Kharitonov. 1990. The evolution of coloniality. Cur. Omith. 7:285-330.

Silver, R. 1978. The parental behavior of ring doves. Am. Sci. 66:209-215.

Sridhar, S., and K. P. Karanth. 1993. Helpers in cooperatively breeding small green bee-eaters {Merops orientalis). Cur. Sd. 65:489-490.

Stacey, P. B. and W. D. Koenig (eds.) 1990. Cooperative breeding in birds: long- term studies of ecology and behavior. Cambridge University Press, New York.

Stacey, P. B., and D. J. Ligon. 1987. Territory quality and dispersal options in the acom woodpecker, and a challenge to the habitat saturation model of cooperative breeding. Am. Nat. 130:654-676. 181

Stacey, P. B., and D. J. Ligon. 1991. The benefits-of-philopatry hypothesis for the evolution of cooperative breeding: variation in territory quality and group size effects. Am. Nat. 137:831-846.

Strahl, S. D., and A. Schmitz. 1990. Hoatizins: cooperative breeding in a folivorous neotropical bird. Pages 133-155 in Cooperative breeding in birds: long-term studies of ecology and behavior (P. B. Stacey and W. D. Koenig, eds.). Cambridge University Press, New York.

Swofford, D. L. 1993. PAUP: Phylogenetic Analysis Using Parsimony (Version 3.1.1). Washington, D. C., Smithsonian Ir\stitution.

Tinbergen, N. 1963. On aims and methods of ethology. Z. Terpsyhol 20:410-433.

Trivers, R. L. 1971. The evolution of reciprocal altruism. Q. Rev. Biol. 46:35-57.

Vleck, C. M., N. A. Mays, J. W. Dawson, and A. R. Goldsmith. 1991. Hormonal correlates of parental and helping behavior in cooperatively breeding Harris' Hawl^ {Parabuteo unicinctus). Auk 108:638-648.

Walters, J. R. 1990. Red-cockaded woodpeckers: a 'primitive' cooperative breeder. Pages 69-101 in Cooperative breeding in birds: long-term studies of ecology and behavior (P. B. Stacey and W. D. Koerug, eds.). Cambridge Uruversity Press, New York.

Webber, T., and J. L. Brown. 1994. Natural history of the vmicolored jay in Chiapas, Mexico. Proc. West. Fotmd. Vert. Zool. 5:135-160.

Wiens, J. J. 1995. Polymorphic characters in phylogenetic systematics. Syst. Biol. 44:482-500.

Wingfield, J. C. 1990. Interrelationships of androgeris, aggression, and mating systems. Pages 187-205 in Endocrinology of Birds, Molectdar to Behavioral (M. Wada, S. Ishii and C. G. Scanes, eds.). Springer-Verlag, New York.

Wingfield, J. C., R. E. Hegner, and D. M. Lewis. 1991. Circulating levels of luteinizing hormone and steroid hormones in relation to social status in the cooperatively breeding white-browed sparrow weaver. Plocepasser mahali. J. Zool. (Lond) 225:43-58.

Woolfenden, G. E., and J. W. Fitzpatrick. 1984. The Florida Scrub Jay. Princeton University Press, Princeton.

Wrege, P. H., and S. T. Emlen. 1987. Biochemical determination of parental uncertainty in white-fronted bee-eaters. Behav. Ecol. Sociobiol. 20:153-160. 182

Wrege, P. H., and S. T. Emlen. 1991. Breeding seasonality and reproductive success of white-fronted bee-eaters in Kenya. Auk 108:673-^87.

Zahavi, A. 1990. Arabian babblers: the quest for social status in a cooperative breeder. Pages 105-130 in Cooperative breeding in birds: long-term studies of ecology and behavior (P. B. Stacey and W. D. Koenig, eds.). Cambridge University Press, New York.