Genus Loxops (Drepanididae)

FUNCTIONAL ANATOMY AND ADAPTIVE EVOLUTION OF THE FEEDING APPARATUS IN THE HAWAIIAN HONEYCREEPER GENUS LOXOPS (DREPANIDIDAE)

LAWRENCE P. RICHARDS

AND

WALTER J. BOCK

ORNITHOLOGICAL MONOGRAPHS NO. 15

PUBLISHED BY

THE AMERICAN ORNITHOLOGISTS' UNION

1973 FUNCTIONAL ANATOMY AND ADAPTIVE EVOLUTION OF THE FEEDING APPARATUS IN THE HAWAIIAN HONEYCREEPER GENUS LOXOPS (DREPANIDIDAE)

LAWRENCE P. RICHARDS

AND

WALTER J. BOCK

ORNITHOLOGICAL MONOGRAPHS NO. 15

PUBLISHED BY

THE AMERICAN ORNITHOLOGISTS' UNION

1973 ORNITHOLOGICAL MONOGRAPHS

This series,published by the American Ornithologists'Union, has been establishedfor major paperstoo long for inclusionin the Union's journal, The Auk. Publicationhas been made possiblethrough the generosityof Mrs. Carll Tucker and the Marcia Brady Tucker Foundation,Inc. Correspondenceconcerning manuscripts for publication in the series should be addressedto the Editor, Dr. John William Hardy, Department of Natural Science, The Florida State Museum, University of Florida, Gainesville, Florida 32611. Copiesof OrnithologicalMonographs may be orderedfrom the Treasurer of the AOU, Burt L. Monroe, Jr., Box 23447, Anchorage,Kentucky 40223. (See price list on inside back cover.)

OrnithologicalMonographs, No. 15, x + 173 pp. Editor-in-chief, John William Hardy SpecialAssociate Editor for this issue, Richard L. Zusi, U.S. National Museum, Washington,D.C.

Issued November 1, 1973 Price $6.00 prepaid ($4.75 to AOU Members) Library of CongressCatalogue Card Number 73-88386 Printed by the Allen Press,Inc., Lawrence,Kansas 66044 TABLE OF CONTENTS

PREFACE ...... ix

INTRODUCTION ...... 1

ACKNOWLEDGMENTS ...... 7

METHODS AND MATERIALS ...... 9 FIELD STUDIES ...... 9 DISSECTION AND DRAWING ...... 10 PLATES ...... 1 1 KINETICS ...... 12 RELATIVE SIZE DIFFERENCES OF THE JAW MUSCLES ...... 12

TYPES OF FOOD AND FEEDING METHODS ...... 14 HAWAIIAMAKIHI, Loxoes tIRE,rstIRE,rs ...... 14 Typesof foodtaken ...... 14 Methodsof feeding...... 18 GREATERAMAKIHI, LoxoPs SAGITTIROSTRIS ...... 20 HAWAIICREEPER, LoxoPs MACULATAMANA ...... 20 Typesof foodtaken ...... 20 Methods of feeding...... 21 MAUI CREEPER,Loxoes M,•CVL.•t.••V•WtO•Vl ...... 22 Typesof foodtaken ...... 22 Methodsof feeding...... 22 HAWAII AKEPA,Loxoes coccI•v•.• coccI•v•.• ...... 23 Typesof foodtaken ...... 23 Methodsof feeding...... 23

SUMMARY OF FOODS AND METHODS OF FEEDING ...... 26 L. v. vtn•vs ...... 26 L. $AGITTIROSTRIS ...... 26 L. M. M.•N.• ...... 26 L. M. NEWTONI ...... 26 L. C. COCCINE.• ...... 27

iii RHAMPHOTHECAE OF THE BEAK ...... 27 DESCRIPTION ...... 27 FUNCTIONAL INTERPRETATION ...... 30

CRANIAL OSTEOLOGY ...... 35 INTRODUCTION ...... 35 SKULL OF n. •'. VIRE:VS...... 35 CRANIAL KINESIS ...... 44 GAPINC ADAPTATIONS ...... 48 COMPARISON OF THE SKULL IN LOXOPS ...... 48 CRANIAL ASYMMETRIES IN L. ½. ½O½½INEA...... 52

JAW MUSCULATURE ...... 53 INTRODUCTION ...... 53 DESCRIPTION OF INDIVIDUAL MUSCLES ...... 54 A) M. protractorpterygoidei et quadrati...... 54 M. protractorpterygoidei sensu stricto ...... 54 2) M. protractorquadraft ...... 55 B) M. adductor mandibulae extemus ...... 56 M. adductor mandibulae externus rostralis ...... 56 a) M. adductor mandibulae extemus rostralis temporalis...... 56 b) M. adductormandibulae extemus rostralis lateralis ...... 57 c) M. adductor mandibulae extemus rostralis medialis ...... 58 2) M. adductor mandibulae externus ventralis ...... 58 3) M. adductor mandibulae externus caudalis ...... 59 a) M. adductormandibulae extemus caudalis "a" _...... 59 b) M. adductormandibulae extemus caudalis "b" _...... 60 C) M. adductormandibulae posterior ...... 61 M. adductor mandibulae intemus ...... 62 D) M. pseudotemporalissuperficialis ...... 62 E) M. pseudotemporalisprofundus ...... 63 F) M. pterygoideus...... 63 l) M. pterygoideusventralis lateralis ...... 64 2) M. pterygoideusventralis medialis ...... 66 3) M. pterygoideusdorsalis lateralis ...... 67 4) M. pterygoideusdorsalis medialis ...... 69 5) M. pterygoideusretractor ...... 71 G) M. depressormandibulae ...... 73 SUMMARY OF COMPARISON OF JAW MUSCLES ...... 74

FUNCTIONAL ANALYSIS OF JAW MECHANISMS ...... 78

THE ASYMMETRICAL JAW APPARATUS OF L. COCCINEA ...... 81

THE TONGUE APPARATUS ...... 94 INTRODUCTION ...... 94

THE CORNEOUS TONGUE ...... 94

OSTEOLOGY OF THE TONGUE ...... 96

MUSCULATURE OF THE TONGUE ...... 98 A) M. mylohyoideus...... 99 B) M. ceratohyoideus ...... 99 C) M. stylohyoideus...... 100 D) M. serpihyoideus...... 100 E) M. branchiomandibularis...... 101 F) M. genioglossus...... 101 G) M. ceratoglossus...... 102 H) M. hypoglossusanterior ...... 102 I) M. hypoglossusobliquus ...... 103 J) M. tracheohyoideus...... 103 K) M. thyreohyoideus...... 104 FUNCTIONAL INTERACTIONS OF THE TONGUE MUSCULATURE ...... 104 FUNCTIONAL AND ADAPTIVE SIGNIFICANCES OF THE TONGUE ...... 108

DISCUSSION ...... 111 FEEDING ADAPTATIONS ...... 111 A) Loxops virens virens ...... 113 B) Loxops sagittirostris ...... 114 C) Loxops maculata maria ...... 115 D) Loxops maculata newtoni ...... 117 E) Loxops coccinea coccinea ...... 118 ADAPTATION AND EXTINCTION ...... 121

EVOLUTIONARY I-HSTORY OF THE GENUS LOXOPS ...... 122

SUMMARY ...... 128

LITERATURE CITED ...... 129

APPENDIX I-•GLOSSARY OF ABBREVIATIONS ...... 134 ILLUSTRATIONS

Plate 1. Lateral view of the bill of Loxops ...... 139 2. Dorsal and ventral view of the bill of Loxops ...... 141 3. Skull of Loxops v. virens ...... 143 4. Skull of Loxops m. mana ...... 145 5. Skull of Loxops m. newtoni...... 147 6. Skull of Loxops c. coccinea...... 149 7. Details of the skull of L. v. virens ...... 151 8. Detailsof the quadrateand mandibleof Loxops...... 153 9. Detailsof the quadrateand mandible of L. c. coccinea (right-billedindividual) ...... 155 10. Jaw muscles of L. v. virens ...... 157

11. Jaw muscles ofL. v. virens ...... 158

12. Jaw muscles of L. m. mana ...... 159

13. Jaw muscles of L. m. mana ...... 160

14. Jaw muscles of L. m. newtoni ...... 161

15. Jaw muscles of L. m. newtoni ...... 162

16. Jaw musclesof L. c. coccinea(right-billed individual) ...... 163 17. Jaw musclesof L. c. coccinea(right-billed individual) ...... 164 18. Jaw musclesof L. c. coccineaand L. sagittirostris...... 165 19. Corneoustongue of Loxopsin dorsaland lateral views (L. v. virens,L. m. manaand L. n. newtoni) ...... 166 20. Corneoustongue of Loxopsin dorsaland lateral views (L. c. coccineaand L. sagittirostris)...... 167 21. Tongueskeleton of Loxopsin ventralview ...... 168 22. Tongueskeleton and muscles of L. sagittirostris...... 169 23. Tonguemuscles ofL. v. virens...... 170 24. Tonguemuscles of L. m. mana...... 171 25. Tonguemuscles of L. m. newtoni...... 172 26. Tonguemuscles of L. c. coccinea...... 173

vi Figure 1. Distributionof the speciesand subspeciesof Loxops in the Hawaiian Islands ...... 4

2. Device used for measurement of skull kinesis ...... 13

3. Food sourcesof Loxops and other drepanidids...... 16 4. Skullof L. v. virensin obliqueview ...... 36 5. Skull of L. v. virens in lateral view ...... 37

6. Skull of L. v. virens in ventral view ...... 38

7. Schematic model of L. v. virens to show movements ...... 46

8. Schematicdrawings showing leaf-bud openingby L. c. coccinea ...... 85

9. Schematicdrawings showing movement of bill tips of L. c. coccinea in closed-bill method ...... 87

10. Schematicdrawings showing movement of bill tips of L. c. coccineain opened-billmethod ...... 90 11. Schematicdrawings showing opening of koa seed-pods by L. c. coccinea...... 91 12. Tongueskeleton of L. v. virens ...... 97 13. Functionsof tonguemuscles of L. v. virens...... 105 14. Dendrogram of Loxops ...... 123

vii

PREFACE

This studywas originallyconceived and plannedby LawrenceP. Richards as an inquiry into the functionalmorphology and evolutionof the cranial morphologyof the Hawaiian Honeycreepers,an aspect of the adaptive radiation of these birds that is still largely unknown. The field studies were conducted in 1951-52, and the initial dissectionsstarted in 1952. Soon afterwards,it was apparentthat the original plan to cover the entire Drepanididaewas not feasible at that time so that the study was restricted to the genus Loxops. The resulting thesis "Functional anatomy of the head region of the Hawaiian Honeycreepergenus Loxops (Aves, Drepani- idae)" was presentedin partial fulfillmentfor the Ph.D. degree,University of Illinois in June 1957. Lack of support and encouragementand, more importantly, lack of suitable publicationpossibilities precluded its publication at that time. Moreover, the generaldecline of interestin studieson passerinejaw and tongue musculaturein the second half of the 1950's made this period not propitiousfor such a study. These factors and sub- sequentdevelopment of interest by Richards in other areas of vertebrate biology causedabandonment of the project for several years. In the fall of 1961, Professor Hobart M. Smith showed Richard's thesis to Walter J. Bock, who had just joined the staff of the Department of Zoology, University of Illinois, expressinghis hope that it could be published without further delay. After initial contacts were made, Bock urged Richards to ready his thesis for publication. Further discussions and some work in the summer of 1963 followed, but a series of field trips and changesin positionby both Richardsand Bock delayed any significantwork. Moreover, a number of papers on passerinejaw and tongue musclesand on functional analysesof cranial features necessitated a rather completereview of the thesisand a reworkingof severalsections. The task of reviewingthe pertinentstudies that appearedin the decade since1957 was formidablein itself, but was renderedimpossible by the constantdevelopment of new ideas and the modification(some still unpublished)of earlierstatements, and by the fact that someof the essential studies(on the redescript-ionof the jaw and tonguemuscles) by Bock and his studentswere still unpublished.Largely becauseof the latter fact, we decidedearly in 1968 to work togetheron the final preparationof this study for publication,so as to bring togetherthe specialknowledge of Richardson the naturalhistory and morphologyof the HawaiianHoney- creepersand the specialknowledge of Bock on the functionalmorphology of the passerinejaw and tongueapparatuses. This project occupiedmost of the summerof 1968. Becausethe historyof eachauthor in this project is quite different and becauseit becamea joint venturelate in its development, a few additionalcomments on responsibilitiesmust be given. The responsibilityand all credit for field observations,obtaining of specimens,dissections of the jaw and the tongue musclesand original descriptionsand interpretationsbelong to Richards. The systemsof identification and terminologyfor the musclesare basedupon studiesby Bock, but it should be noted that few discrepanciesexisted betweenthe muscle identificationsand terminologiesused by Richardsin his thesisand those advocatedby Bock. Responsibilityfor the entire paper, functionalinter- pretationsand conclusionson the adaptive significancesof these cranial featuresand on the evolutionof the group is sharedequally as we reviewed, discussedand rewrote the entire manuscript for publication. Hence we look upon this paper as a truly cooperativeundertaking in spite of the separateorigins of the informationand ideas used in reachingthe inter- pretationsand conclusionspresented herein. Submitted for publication, April 1970 Final Revision, December 1972 INTRODUCTION

The family of oscinebirds endemic to the HawaiianIslands, the Hawaiian Honeycreepersor Drepanididae,is one of the most spectacularand geo- graphicallycompact examplesof adaptive radiation. Almost the entire range of feeding methodsand bill structureknown in passerinebirds is found within the Drepanididae. This radiation is especiallyinteresting becauseit has occurredat the low taxonomiclevel of the family and pre- sumablywithin a short span of geologicaltime. Although details still remain to be clarified, many evolutionaryphenomena are well illustrated by this groupof birds--alas, that of extinctionis all too well demonstrated by lost membersof the Drepanididae(see Greenway,1967). Consequently, the Hawaiian Honeycreepershave long held the interest of ornithologists and evolutionists,and have been the subjectof severalmonographs (Wilson and Evans, 1890-1899; Rothschild, 1893-1900; Wallace, 1880, 1902; Perkins, 1901, 1903; Henshaw, 1902; Munro, 1944; Amadon, 1950; Baldwin, 1953). Apparently, this family was too poorly known in the middle of the last century to have any role in the initial developmentof evolutionaryideas; Darwin seemsnot to have mentionedthem at all and Wallace treats them very briefly in the early editionsof his "IslandsLife." Amadon's (1950) study is of special importance,not only becauseit is the most recent monograph of the family, but mainly because it is the first (and to date, the only) review of the evolutionaryhistory and classification of the Drepanididae using the ideas generatedby recent advances in the New Systematicsand the SyntheticTheory of Evolution. He draws together all of the available information on these birds as the foundation for his conclusionson their phylogeny,patterns of speciation,nature and sourcesof selectionforces, and systematics.His monographis, and quite rightly so, the most authoritativetreatment of the Drepanididae and is the source for most discussionsof this family in general evolutionarytexts. In similar fashion, Baldwin'sstudy is the only analysisof the ecologyof these birds using modern approachesto organism-environmentinteractions. The role of introduced diseasesin the widespread and rapid extinction of the HawaiianHoneycreepers has been discussed elegantly by Warner (1968); his ideas are essentialto our conclusionson the adaptivenessof the several speciesof Loxops. Since fossils of these birds have not yet been discovered, and are not likely ever to be found, the time of original colonizationof the Hawaiian archipelagoby the form, or forms, ancestralto the present-daydrepanidids is not known. Nor can the likely ancestralform be ascertainedexcept by comparativeanalysis and inferencefrom living birds. Stearns (1946: 2, 85) indicatesthat the larger Hawaiian Islands were above sea level in 2 ORNITHOLOGICAL MONOGRAPHS NO. 15

Tertiary times, perhaps as late as the end of the Pliocene epoch of that period. The time of colonizationof the Hawaiian Islands by the ancestral stock cannot be dated more exactly than in the Tertiary, but it seems doubtful that colonizationtook place near either extreme possibilityof as early as seventy-fivemillions of years ago (in the PaleoceneEpoch) or as late as the Pleistocene,one million years ago (Knopf, 1949: 8). Zim- merman (1948: 44) considersthat land areascapable of supportingforests have been in existencein these islands since late Pliocene times, some five millionsof yearsago. Carson,Hardy, Spiethand Stone (1970) presentan excellentsummary of the geologicalhistory of the Hawaiian Islands based upon recentfindings of global plate tectonics. Whatever the actual time of colonization,a number of potential empty ecologicalhabitats for passerinebirds were presentin the Hawaiian Islands. The original habitatsmay have been the same as those found today or they may have differed somewhat. In any event, some of these habitats were filled by the ancestraldrepanidids (along with the few endemic, probablylater arriving meliphagids,turdids and muscicapids)to the partial or completelimits of the adaptationsand preadaptationsof these birds. Subsequentspeciation and phyletic evolutioninto genera and subfamilies (assumingthat the Drepanididaeare monophyletic) took place with re- peatedreinvasion of the variousislands and with competitionfor food as a main selection force (Amadon, 1950: 235, 246; 1947: 63, 66). The combined actions of adaptive radiation and extinction have pre- sumablyobliterated traces of the ancestralstock of the Drepanididaeand of its early radiation. One cannot simply assumewith any justification that any of the extant forms are close representativesof the ancestral stock. No real argumentexists today that the Drepanididaearose from the New World nine-primariedcomplex of oscine birds. Beyond that, most ideas on the possibleancestors of this family may be grouped into two main theories. The earlier one, based largely on the structureand distribution of the tubular tongue, is that the Drepanididaearose from Neo- tropical Coerebidaeor Thraupidae. Becauseof the vaguenessof the Coerebidaeand becauseof the close similarity of at least some of these birds and the Thraupidae,we make no distinctionbetween choice of either of thesefamilies as possibledrepanidid ancestors in our discussion.Gadow (1891, 1899) was the first worker to advocate this relationshipstrongly with supportinganatomical data. Amadon (1950: 231-233), Beecher (1951b: 283, 285; 1953: 312-313) and Baldwin (1953: 386-388) acceptedGadow's conclusion and suppliedadditional support for it, much of it being evidencefrom jaw musclepatterns provided by Beecher'scom- parativeanatomical studies of the oscinebirds. The secondand later notion, originallyadvocated by Sushkin(1929) on the basis of skull and horny 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 3 palate morphology,is that the Drepanididaeevolved from the cardueline finches of the Fringillidae,or from a group directly ancestralto the carduelines(including the genusFringilla). Carduelinerelationships of the HawaiianHoneycreepers has beensupported by Bock (1960b: 477-478) on the basisof jaw muscles(in disagreementwith Beecher'sconclusion), Tordoff (personalcommunication) on the basisof generalnesting habits, and Sibley 1970: 104--105on the basisof comparativestudy of several protein systems.For the purposesof this paper, we accept tentatively the carduelinetheory for the ancestorsof the Drepanididae. The jaw musclesof Loxopssupport this theorybetter than the thraupidtheory, but a soundcomparative study must be undertakenbefore more definite conclusionscan be offered. Moreover, we emphasizethat the major conclusionsof the presentstudy depend to a veryminor degree on accepting the carduelinetheory of drepanididancestry. Amadon(1950), whoseclassification we follow,divides the Drepanididae into two subfamilies,the Drepanidinaeand the Psittirostrinae.The correct name for this family is the Drepanididae,not the Drepaniidae,following thefavorable decision by theInternational Commission on ZoologicalNomen- clature (Anon., 1961) on the proposalby Amadon (1960) to place the name DrepanididaeGadow, 1891 on the Official List of Family-Group Names. The Drep.anidinaeconsists of five genera (Himatione, Palmeria, Ciridops,Vestiaria, and Drepanis) and the Psittirostrinaeof four (Loxops, Hemignathus,Pseudonestor, and Psittirostra). Amadon (1950: 164-168) merged four genera of earlier workers (Viridonia,Chlorodrepanis, Paroreomyza, and Loxops) into a singlegenus Loxops, recognizingthree subgenera,Viridonia, Paroreomyzaand Loxops. Each of thesesubgenera are widely distributedthroughout the main islands of the Hawaiianarchipelago (Fig. 1). While the final draft of this manuscriptwas being prepared, volume 14 of "Peters'Check-list," containing the Drepanididaewas published.Greenway (1968), in treatingthis family, dividedAmadon's Loxops into threegenera corresponding to the subgenera recognizedby Amadon; unfortunately,no reasonswere given for this change. Becauseour study excludedfeatures of the externalmorphology and many aspectsof their behaviorand life-history,and becausewe did not includeall membersof the family, we are unableto evaluatethe merits of Amadon'ssystem versus that of Greenway. Discussionsbetween Bock and Greenwayof the reasonsunderlying Greenway's treatment of the Loxops group as opposedto Amadon's could be summarizedas differencesin opinionon the width of genericlimits in the Drepanododae.Our preference is for broadergeneric limits, but even so, we lack comparativeevidence on which to base any argumentincluding the difficult questionon the relationshipbetween the membersof Loxopsand the membersof Hemig- 4 ORNITHOLOGICAL MONOGRAPHS NO. 15 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 5 nathus. However,because we prefer broader genericlimits, and more importantlybecause of the practicalreasons of the easeof stylein dealing with one, not three genera,we continueto follow (without prejudice) Amadon'sgeneric treatment of the genusLoxops. The subgenusViridonia containsthree species,the Amakihis,Loxops virens; the Anianiau,Loxops parva (Kauai); and the GreaterAmakahi, Loxopssagittirostris (Hawaii, extinct). The last two are monotypicspecies. L. virens, however, contains four races which inhabit six islands: L. v. virens(Hawaii), L. v. wilsoni(Maui, Molokai, Lanai), L. v. chIoris(Oahu) and L. v. stejnegeri(Kauai); stejnegeriis quite different in bill structure from other membersof L. virens. The subgenusParoreomyza contains onespecies, Loxops maculata, the Creepers,which is composedof six sub- speciesinhabiting six islands: L. m. mana (Hawaii), L. m. newtoni (Maui), L. m. /lammea (Molokai, extinct), L. m. montana (Lanai), L. m. maculata (Oahu), and L. m. bairdi (Kauai). The subgenusLoxops con- tainsone species,the Akepas,Loxops coccinea, which has four raceson four islands: L. c. coccinea(Hawaii), L. c. ochracea(Maui), L. c. ru/a (Oahu, extinct) and L. c. caeruleirostris(Kauai). The generaof Hawaiian Honeycreepersdiffer widely with respectto shapeand size of thebill andtongue, and presumably in associatedanatomical features.In many cases,as mightbe expected,direct correlationscan be made betweenjaw and tonguemorphology in the severalgenera and the uses (biologicalroles) to which thesestructures are put by the birds in obtaining food. However, the details of these correlationsand similar intragenericcorrelations have not yet been described; such conclusions dependupon a combinationof precisefield observations,morphological dissectionsand, if possible,experimental studies. Considerablefield observationson the habitats,food and feeding methods of Hawaiian Honeycreepersare available,most of them done by a few workerswho had the opportunityto reside on the Hawaiian Islands for severalyears. Many of theseobservations cannot be repeatedtoday be- causeseveral species of drepanididsare extinctor existin smallerpopulations than they did at the time these earlier workers were active. The more importantwritings on the natural historyof the Hawaiian Honeycreepers include Perkins (1893, 1895, 1901, 1903), Munro (1944), Henshaw (1902), Bryan (1908) and Baldwin (1953). Anatomical investigationsof the drepanididshave been scanty, far fewer than the availablefield observationsand far fewer than permittedby long- available materials. In truth most of these anatomical studies should be classedas descriptiveexternal morphology. The earliest and still most extensivework is that of Hans Gadow whose results were publishedas supplementsto Wilson and Evans' great monograph(1891 and 1899: 6 ORNITHOLOGICAL MONOGRAPHS NO. 15

219-249). Most of the remainingwork (Rothschild, 1893-1900: pls. 82 and 83; Hartert, 1893: xx; Lucas, 1897: fig. 3b; Gardner, 1925: 27-28, and pl. 3, fig. 19, pl. 13, fig. 140; Sushkin,1929: 379-381; and Munro, 1944: 100, 119) dealswith the structureof the tongue,rhamphotheca (includingthe horny palate) and skull structure. Amadon (1950: 213-229) has a sectionon comparativeanatomy in which he discussesand figuresthose aspects of drepanididmorphology described by earlierworkers and commentsupon the systematicrelevance of these features. Finally, Beecher(1951b: 283, 285, fig. 5; 1953: 310, 312-313, fig. 15) hypoth- esizeson the relationshipsof the Drepanididaeto the Thraupidaevia the medium of jaw musculaturepattern, horny palate, tongue architectureand other features. A detailedinvestigation of the structuraland behavioraladaptations for feedingin this highly interestingexample of adaptiveradiation has been long overdue.Not only is the amountof publishedwork sadly insufficient for any theorizingon the possiblerelationships and evolutionof this family, but much of it was done many years ago when our understandingof functionalmorphology, in general, and of the avian jaw and tongue apparatus,in particular,was inadequatelyknown. Moreover, recent developments in many fundamental conceptsof comparative morphologyhave revitalizedthe entire field and have provided a far sounderbasis for comparative anatomical-evolutionarystudies. We proposeto undertakean ex- tensiveinvestigation of the adaptive radiation of the feeding apparatusin the Drepanididaebased upon the recentlydeveloped concepts of evolutionary morphology. As a start into this project,we choseto studythe functionalmorphology and evolutionof the feeding apparatusin the genusLoxops, concentrating upon the following taxa: L. sagittirostris(in part only), L. virens virens, L. maculatamana, newtoni,and L. coccineacoccinea. The rhamphothecal, skeletal and muscularfeatures of the jaws and tongue apparatuswill be describedand comparedin detail. Types of food taken and methodsof feedingwill be correlatedas far as possiblewith the morphologicalfeatures of the jaws and tongueand with their functionalinterpretations. Finally, we hopeto speculateon the evolutionof thesefeeding adaptations and of the taxa within Loxops. The genusLoxops is particularlyadvantageous for this initial study. It has been long consideredas the most primitivegroup in the Psittirostrinae and dose to, if not, the mostprimitive stock in the whole family. More- over,members of thisgenus show considerable variation in their morphology and feeding methodswhich include: (a) nectarivorous-insectivorousleaf gleaners; (b) insectivorousleaf and bark gleaners; (c) bark and other 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 7 creviceprobers; and (d) cross-billedinsectivorous leaf-bud and bean-pod openers. The speciesof Loxops have widespreadsympatry with almost every major island possessingraces of the three commonspecies (virens, maculata and coccinea). These speciesare still sufficientlycommon to permit field observationson their feeding methods. Moreover, several practicalreasons determined the choiceof this genusand of the particular speciesstudied. Some specimensof L. virens and maculatawere already availableto Richardsat the outset of the study, and the other forms were sufficientlycommon to permit taking of additionalspecimens. Members of L. virens, maculata and coccinea could be observedon Maui and Hawaii, islandson which living accomodationswere readily availableto Richards as well as possessingsufficient roads allowing travel to the interior of the islandswhere the birds are found. Moreover, a search for L. sagittirostris could be undertakenon Hawaii. Sufficientfunds and living accomodations on Kauai were not available to Richards to permit inclusionof L. parva in this study.

ACKNOWLEDGMENTS

The unusualdevelopment of this study with one author (Bock) joining the projectlong after completionof the originalresearch and writing of the thesis,necessitates separation of the acknowledgements.Each part of these acknowledgementswill be signedand referenceof the first personpronouns, be they singularor plural, should be readily apparent. So many people have aided me in my field work and researchin so many ways that it will be difficult to acknowledgeall of them for the thanks they are due. I am indebtedto many friends and relativesin the Hawaiian Islandswho so generouslyaided and showedme hospitalityand gave me accessto their homes, mountain cabins, ranch lands and roads, vehiclesand saddlehorses. Included among these people are my grand- mother,Isabelle Jones, my parents,Robert and CatherineThompson, my uncle and aunt, Howard and Helen Farrar, my friend William Harkins of Maul and a number of the cattlemen of the Kona coast and Mauna Kea, Hawaii: R. L. Hind, Sr. and Jr:, Norman Greenwell, SherwoodGreenwell, William Thompson,Herbert Shipman,Hartwell Carter, and Roger Williams. I wish to thank Colfin Lennox, Donald Smith and the other members of the Board of Commissionersof Agricultureand Forestryof the Territory of Hawaii for appointingme honorarybiologist and issuingpermits for the collectionof drepanididsin the forestreserves; Robert Hiatt of the University of Hawaii, for giving me equipment and suppliesfor the preservationof specimens;members of the staff of the Hawaii National Park Servicefor giving me accessto cabins and other facilities at Kilauea and Haleakala; 8 ORNITHOLOGICAL MONOGRAPHS NO. 15

EdwinH. Bryan,Jr., Curatorof Collectionsat the BishopMuseum Honolulu, for loan of specimens,Alden H. Miller, directorof the Museumof Vertebrate Zoologyat Berkeley,California, for accessto drepanididskeletal specimens and for generouslyputting at my disposalphotographic equipment and supplies.To my wife, ChristinaM. Richards,and my goodfriend, Robert H. S. Glaserof SacramentoJunior College, California, I am gratefulfor aid in the translationof parts of German publicationson avian anatomy. Paul H. Baldwinof ColoradoAgricultural and MechanicalCollege, gave me much valuableadvice, information and encouragementbefore and during the conductingof my field work; furthermore,he put at my disposalsixteen alcoholicspecimens of Drepanididae,four of which were usedin this study. William Beecherof the ChicagoNatural History Museumalso passedon to me useful advice and information and answeredfor me many questions put to him in correspondence.I wish to thank my advisor,Dr. Hobart M. Smith, and former advisors,Drs. Alden H. Miller and Harvey I. Fisher, for their patience,encouragement and helpful advice. This dissertation(Rich- ards, 1957) was written in partial fulfillment of the requirementsfor the doctor of philosophydegree at the University of Illinois, Urbana.-•L. P. Richards. We wish to thank Mrs. Frances Jewel who devoted much skill and care in assembling,correcting and labeling the originalfigures drawn by Richards andused in histhesis into the final platesused in thisstudy, and in drawing a numberof new platesand text figuresthat becamenecessary during the final phasesof this study. We would also like to thank Miss SuzanneBudd who completedmuch of the labelingand undertookthe long and arduous task of final correctionsand additionsto the figures. We would like to thank Drs. H. Morioka, R. S. Hikida, J. L. Cracraft and C. R. Shear for their many helpful commentsand suggestionsgiven in the courseof many discussionsduring the summerof 1968 and thereafter. The final manuscript was typed by Mrs. Julia Cracraft who transformedour handwrittenscrawl into a clean typescript wonderfully free of mistakes. Dr. Dean Amadon made the facilitiesof the Departmentof Ornithology,American Museum of Natural History available to us and provided us with much valuable adviceand suggestionsfrom his extensiveknowledge of the Drepanididae for which we are most grateful. Supportfor the completionof this project and for readyingthe manuscriptand figures for publicationcame from a grant to W. J. Bock from the National ScienceFoundation (N.S.F.-G.B. 6909X) for studies on the feeding apparatusof the New World nine- primariedoscines. We wish to thank the National ScienceFoundation for agreeingto changesin the use of thesegrant funds which was the critical factor permittingthe completionand publicationof this study.-•L. P. Richards and W. J. Bock. 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 9

METHODS AND MATERIALS Field Studies: Becausethe goal of this study was to correlatesome of the anatomicalstructures of the headregion with use of the bill and tongue in feeding,not only were specimensof the Drepanididaeneeded but alsoas many recordedobservations as possibleon the methodsof feeding. Even thoughsuch men as Perkins,Munro, Henshawand Bryan recordedmuch valuableinformation on feedingbehavior, they undoubtedlydid not write with the thoughtin mindthat their observationswould be usedsubsequently in studiesof functionalanatomy. Therefore,it was importantthat the feedingof thesebirds be observedpersonally under natural conditions and the minutiaeof useof the bill and tonguerecorded as fully as possiblefrom the standpointof one interestedin the muscularand skeletalapparatuses involved. BetweenAugust 11, 1950, and January17, 1951, a total of 59 days were spentby LawrenceRichards in the forestsof Hawaii, Maul and Oahu, collectingand observingdrepanidids. Prior and subsequentto this period, five otherdays were spentfor the samepurposes. During these periods, 94 pagesof speciesaccounts on drepanididswere written, of which 35 were on Loxops, and 48 drepanididswere collected,of which 23 were Loxops.Paul H. Baldwingave Richards seven additional alcoholic specimens of this genus,and the BishopMuseum loaned two incompletealcoholic specimensof sagittirostris,on whichonly the occipitalregions of the crania remained. Of these32 specimensof Loxops, 25 were studied: L. sagittirostris(2 alcoholics),L. v. virens(8 alcoholics),L. m. mana (3 alcoholics, 1 skeleton),L. m. newtoni (5 alcoholics,2 skeletons),L. c. coccinea(3 alcoholics,1 skeleton). In additionto these specimens,27 skeletonsof L. v. virens and one of L. m. mana were available from the Museum of VertebrateZoology. Admittedly,the numberof specimensof each form availablefor dissectionand studyis pitifully small,and thesesmall numbers of specimenshad to be usedas the basisfor statements,suppositions, and conclusionsconcerning the wholeliving populationof each speciesor race studied. We realize that this is not necessarilyscientifically sound but feel that thisstudy and its conclusionsare justifiedin spiteof thesesmall samples becausethe Board of Commissionersof the Departmentof Forestryand Agriculturefor the Territory of Hawaii allowedunder specialpermit only three spedmensof only certain speciesor subspeciesof drepanididsto be collected;because some of the formsof drepanididscollected are practically extinct; becauseit sometimeswas exceedingly difficult to locatethe highly localizedand sometimesseasonally varying rangesof some of the forms; and becauseno one had ever attemptedsuch a functionalanatomical study on thisfamily nor had (and may neverhave again in futureyears) as highly 10 ORNITHOLOGICAL MONOGRAPHS NO. 15 a representativecollection of specimensof forms in this genus. All con- clusionsmade concerningany particularrace or speciesstudied have been madeguardedly and with the reservationdue them because of the smallsize of the sampleof availablespecimens. This fact was kept in mind at all times. It can be surmisedthat, for a particularsubspecies, sexual dimorphism, age, and individualvariation would not play importantroles in the morphologyof thoseparts of the skeletonand the importantdivisions of the jaw and tongue musculaturestudied, with one exception. This one ex- ceptionis the rare Hawaii Akepa, Loxopsc. coccinea,in which age and individualfeeding behavioral differences might, and probably do, make qute a bit of differencein the ontogeneticdevelopment of its asymmetrical rhamphothecae,jaw skeletonand jaw musculature.The study of this form was basedmainly on adult males. The specimenswere preserved in the field in one of two ways. They were eitherskinned, eviscerated and driedto be cleanedsubsequently by dermestid larvae as skeletalspecimens, or they were preservedas "alcoholics."The latter methodinvolved injection of the abdominalcavity, usually within a few minutesafter deathwith 10 per cent formalin,and later wettingthe feathersand skin with soap and water. The specimenswere then placed in a fixing and preservedsolution composed of 90 parts by volume of 70 per cent ethyl alcohol, five parts full strengthformalin and five parts glycerin; the solutionwas changedonce. The specimenswere kept in the preservingsolution until the end of the field trip and were transferredto 70 per cent ethyl alcoholin the laboratory. Dissectionand Drawing: In the laboratorythe heads of the alcoholic specimenswere skinnedpartially and the attachmentsof the extrinsictongue musculatureseparated from the mandibleand skull, the floor of the buccal cavitysevered from its lateralwalls and from the esophagus,and the glossal and laryngealapparatuses removed. The glossaland attachedlaryngeal apparatusand head were pinnedto the wax bottomof a smallplastic dis- sectingpan, partiallyfilled with 70 per cent ethyl alcohol,and the tongue and jaw musculatureswere then dissectedunder a 20 power binoculardis- secting microscope. Drawingsof the tongueswere made to scalefrom the specimens.Drawings of the tonguemusculature were madeon penciltracings of ink drawingsof glossalskeletons. Drawings of the jaw muscleswere made on penciltracings of ink drawingsof the skulls. The ink drawings of the skulls were made in the following manner: Photographswere made of the bestpreserved specimen of a skull for every form studiedexcept sagittirostris, for whichno skullwas available. For these photographsthe mandibleswere gluedto the skullsin the closedposition. Every skull was photographedin three views-•lateral,left front oblique 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS II

(i.e., looking posteriorlyand ventrally into the left orbit from the left front), and ventral. Enlarged prints were made from these negativesso that cranialheight in the lateral and obliqueviews always was kept constant for all species.Cranial width, on the otherhand, in the ventralviews varied proportionatelywith the constantcranial height in the four forms studied. (It was assumedthat, in a singlegenus or family of birds of narrowlylimited range in body size of the speciescomprising the group, the cranial height would be proportional,or isometric,to the body size. This assumption appearsto be reasonablefor the genusLoxops whose member species are uniform in height of the brain case; however,it remainsto be verified for the whole family.) Thus photographsof the skullsof all speciesin the genuswere then of differing magnifications,allowing differencesin shapes and relative sizesof the bills and other, possiblyallometric, structureson the skulls to become more evident. From these photographsink tracingswere made of the three views of the skull, upon which the jaw muscleswere drawn. In the drawingsof the glossalskeletons the length of the fused basihyale plus urohyale was kept constantin all species; therefore, magnifications of the drawings of the glossal skeletonsof the four forms are not the same. It was reasonedthat this length would probably remain most stable in evolutionarylengthening or shorteningof the tongue and that it would be isometricallyproportional to body size throughoutthe family. The tongue muscleswere drawn on these tracingsof the tongue skeleton. The drawings of the tongues are at different magnificationsbut are related to the constantlength of basihyaleplus urohyalein that they were drawn at a magnificationtwice that of the correspondingdrawings of the glossalapparatus, except for the tongue of L. m. newtoni which is three timesthe magnificationof the drawingof the glossalapparatus of that form. All dissections,descriptions and preparationsof the figuresof the rhamphothecae,jaw skeletonand muscles,and glossalskeleton and musculaturewere done, with the following exceptions,by Lawrence P. Richards between 1951 and 1956. The sknllsof the severaltaxa studiedwere jointly described from specimensby Walter J. Bock and LawrenceP. Richardsin light of new knowledgeof avian cranial kinesispublished since 1957. Severaladdi- tional drawingsof the rhamphothecaeand of the skull were preparedby Mrs. Frances Jewel. Plates: To preserveclarity in the figures of the skeletalelements, the pertinentstructures were labeledon a separatetext figure (Figs. 4, 5, 6 and 12); comparisonof this figure with thosein the platesshould pose no problems. Both the jaw and the tongue musculatureare sufficientlycom- plex that the plates of each specieshad to be labeled fully; however, obviousmuscles, such as the m. depressormandibulae, were not indicated 12 ORNITHOLOGICAL MONOGRAPHS NO. 15 in all plates. The abbreviationsused in the text figuresand platesand those usedin the text are summarizedin the glossary. Unlessotherwise specified, the specimensused for illustratingthe skull and the glossalmechanism and the outlinesof theseskeletal elements when drawing the musculatureare as follows: L. v. virens, juv. 9, no. MVZ 118, 763; L. m. mana, ad. •, no. MVZ 118, 823; L. m. newtoni, juv.(?) •, no. MVZ 122, 615; L. c. coccinea,right-billed, ad. •, no. MVZ 122, 613; and L. sagittirostris,only occipitalregion of skull, age and sex unknown, no number, BPBM. In the descriptionsof the plates of the jaw or the tonguemusculature, the museumcatalog numbers or the field catalog numbers (of Paul H. Baldwin or Lawrence P. Richards) refer to the specimensused for the dissectionand drawingsof the soft parts; the age and/or sex of the specimensare statedwhen known. The magnifications are indicatedby an "x" followedby a numbergiving the degreeof magnifi- cation. Abbreviations used are MVZ, Museum of Vertebrate Zoology, University of California, Berkeley; BPBM, B.P. Bishop Museum, Hono- lulu, Hawaii (these specimensdo not have catalog numbers); LPR, collected by Lawrence P. Richards and not as yet repositedin a museum; PHB, collectedby Paul H. Baldwin, ColoradoState University,Ft. Collins, and not as yet repositedin a museum. Kinetics: In order to measure the kinetics of the skulls (see section under anatomyon osteologyof jaws and skull), a machinewas designedand built (Fig. 2); detailson the constructionand operationof this measuringdevice may be obtainedfrom Richards (1957). Relative size differencesof the jaw muscles: An initial working hypothesis was made that muscle size is, in general, a rough index of muscle strengthamong homologousmuscles. On the basis of this assumption, Richardsjudged the relative size differencesof the jaw musclesin the four taxa of Loxops studied; these are summarizedin Table 3. The two sides of coccineawere consideredseparately, giving a total of five classesin this comparison.L. sagittirostriswas excludedfrom this comparisonbecause only its m. depressormandibulae was available. This judgementwas done by assigningrating numbers, designating relative size, to almostevery muscle part for all taxa. The number"1" was givento the speciesor race (or side of coccinea)having the largestmuscle or part of a muscleamong the series of homologousmuscles in the severaltaxa under study; number "2" was aSSignedto the taxon with the secondlargest muscle and so on. If two or more forms has a musclepart of approximatelythe same size, these two or more were given the same rating. In this systemof rating the relative sizes of homologousmuscles, no attemptwas made to distinguishquantitative differences among the forms. Furthermore,judgement of the sizeswas done visuallyand not by means 1973 RICHARDS AND BOCK: FFFDING APPARATUS IN LOXOPS 13 14 ORNITHOLOGICAL MONOGRAPHS NO. 15 of measuringlinear dimensions,volume or mass. The necessarylarge sample of specimensfor determinationof size by methodsof significantaccuracy were not available. Moreover, the attachmentsof the musclesand the lengths of their momentarms were not consideredso that the torquesproduced by thesemuscles (Bock, 1968) were not compared; thesetorques, rather than the forces,produced by the musclesare the more important measurements in comparingthe contributionof each muscleto the action of the skeletal system. The initial assumptionused in this comparisonis not valid for all comparisonsof skeletal musclesas pointed out by Gans and Bock (1965). Rather one should measure the total cross-sectional area of the muscle fibers as an index to force production,the length of the fibers as an index to displacementabilities, and the angle of pinnatenessas an index to the force and displacementcomponent along the vector directionof the muscle pull. Unfortunately,these factors are more easilydiscussed than measured, and we realize fully the shortcomingsof our comparisonsin not under- taking thesemeasurements. Size as used in thesecomparsions is a relatively valid measurementof the severalfunctional properties in a seriesof homologous muscles not too dissimilar in size and fiber arrangement such as those under considerationin this study. And the conclusionsreached on the basis of these comparisonsare relativelyrough ones that do not go beyond the assumptionsemployed.

TYPES OF FOOD AND FEEDING METHODS In this section we have not only consultedRichards' field notes but have drawn heavily upon the publishedobservations of ornithologistssuch as Robert C. L. Perkins, George C. Munro, Henry W. Henshaw,William A. Bryan and Paul H. Baldwin who have studied drepanididsover long periods and at first hand in the field. This information is somewhat scatteredin the literature, and we believe a useful purposeis served by summarizingit, in that the ideasof severalnaturalists based upon extensive observationare brought to bear on the intricaciesof feeding behavior in this genus. The original observationsreported in this chapter are those of LawrenceP. Richardswho assumesprimary responsibilityfor the material presentedherein; the pronoun 'T' refers to him.

HAWAII AMAKIHI, Loxops virens virens Types of Food Taken: Accordingto Baldwin (1953: 287) this species feeds on insectsand other arthropods,nectar, and juices of fruits in this order of importance.By and large this statementis substantiatedby the observationsof Perkins (1893, 1903), Henshaw (1902), Munro (1944) and myself. 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 15

The mostcomplete data on arthropodfoods are givenby Baldwin (1953: 314-320). He statesthat virenstakes arthropodsranging in length from 1 mm to about25 mm or more (somecaterpillars), the majorityof which fall between1.5 mm and 10 mm in length. Most of theseare soft-bodied, nonflying insects; no molluscswere found to be taken. Heteroterans, ants, roaches,beetles, larger waspsand flies, moths, adult lacewingsand land snailsseemingly were by-passedin large degreedespite ready availability. The frequencywith which identifiableitems of animal food were seen in crop and gizzardcontents he indicatedby per cent of occurrencewhich was computedby dividingthe numberof samplesin which a particular item was found by the number of samplesexamined for the bird. He sampled63 specimensof virens between 1938 and 1944, 1948 and 1949. He also recordedrecognizable food fed to young birds. His data are as follows: Homoptera(75%; Auchenorhyncha,Cicadellidae, Delphacidae, Flatidae, Psyllidae,Aphididae, Pseudococcidae, Coccidae, Diaspididae); Lepidoptera(71%; larvae of Pyraustidae(recorded by Perkins), Geo- metridae,Lycaenidae and other caterpillarsplus unidentifedmoths); Ara- neida (33%); Neuroptera(24%; larvae of Hemerobiidae);Coleoptera (11%; Proterhinidae,Curculionidae); Acarina (10%; foliage-inhabiting white mites and Tetranychidae);Corrodentia (10%; Psocidae);Heter- optera (8%; Miridae and others); Diptera (6%); Hymenoptera(3%; Apoidea); and Thysanoptera(2%). It is of interest that the Coccidae are not native to the Hawaiian Islands. Perkins (1901: 566, footnote) states ".... Nectar is undoubtedlyabsolutely necessary to the existence of... [L. virens][and otherDrepanididae] as they are constituted.... Not that nectaris ever the solefood, thougha mostimportant source of nutriment--soimportant to the adultsof somespecies that at certain seasonsno individual shot containsany trace of insect food." Perkins (1903: 409) wrote that virens takes nectar chiefly from the Ohia Lehua tree, Metrosideroscollina • (Myrtaceae), but that it also obtains nectarfrom the campanulateblossoms of the nativelobelias (Campanulaceae) and other families (Fig. 3), and from blossomsof introducedplants such as the banana, Musa sp. (Musaceae); Lantana, Lantana camara (Ver- benaceae); and canna, Canna sp. (Cannaceae). Henshaw (1902: 44) recordedthat virensvisits the Ohia, the banana, and the introducednasturtium

z In this study, this speciesof tree will be referred to either by its abbreviated vernacularname, Ohia, or by its generic name, Metrosideros,since it is by far the most common and most important speciesin the genusfound in the Hawaiian Islands (Rock, 1913; Degener, 1932). For similar reasons,the Mamani tree, Sophorachry- sophylla (Leguminosae), will be referred to by either its vernacular or generic name as will the Naio or BastardSandalwood tree, Myoporum sandwichense(Myoporaceae), and the ClimbingScrew-pine, or Ieie vine, Freycinetiaarborea (Pandanaceae). 16 ORNITHOLOGICAL MONOGRAPHS NO. 15

I 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 17

(Tropaeolaceae),a small plant with brilliantly coloredblossoms and long corollas. Bryan (1908: 162, 164) wrote that the Amakihi on Molokai, L. v. wilsoni, seldom visited lobelia blossomsbut frequently visited the blossomsof the Mountain Naupaka bush, Scaevolasp. (Goodeniaceae). Munro (1944: 100-102) saw virens take nectar from a great variety of flowers, including those of Metrosideros; lobelias; the native forest Loulu Palm, Pritchardiasp. (Palmaceae); and the ClimbingScrew-pine, or Ieie vine, Freycinetia arborea (Pandanaceae). Baldwin (1953: 311-312) frequently saw virens take nectar from Metrosideros and the Mamani tree, Sophorachrysophylla (Leguminosae). He also observedit feeding at the flowers of the native Sandalwoodtree, Santalum paniculatum (Santalaceae); the native vine, Maile, Alyxia olivaeforrnis(Apocynaceae); and the following introducedplants: Fuchsia magellanica(Onagraceae); Honey- suckle,Lonicera japonica (Caprifoliaceae); Gosmore,Hypochaeris radicata (Compositae);Air Plant,Bryophyllum calicinum (Crassulaceae). He thinks it probable that this bird also takes nectar from the blossomsof the native shrub StypheliaTameiameiae (Epacridaceae). The blossomsfrom which I saw virens take nectar most often and on many occasionswere those of Metrosiderosand Sophora.One specimenshot in an area wheremany Ohia Lehua treeswere in bloom and where many individualsof this race were visitingthese blossoms had exudingfrom its bill a clear liquid, sweet to the taste; undoubtedlythis was Ohia nectar. A specimenof the Oahu Amakihi, L. v. chloris, shot from a Koa tree (Fig. 3), Acacia Koa (Leguminosae),which was in full bloom,as were many other Koas in the same area, also had a clear, sweet-tastingliquid drippingfrom its bill. It appearedto havebeen feeding from the Koa blossoms;presumably this liquid was Koa nectar, althoughMetrosideros trees also grow in the same locality. The latter, however,are not mentionedin my field notes as beingin bloomon this date. On one occasionI saw virenstwice dip its bill into a blossomof the Asiatic Thimbleberrybush, Rubus rosaefolius (Rosaceae). Perkins (1903: 409-410) reportedvirens going to the groundfor the berriesof the introducedPoha plant, Physalisperuviana (Solanaceae). He wrotethat the morepowerfully jaw-muscled Kauai Amakihi,L. v. stejnegeri,

Figure 3. Food sourcesof Loxops and other drepanidids showing some flowers visited for nectar and a koa seedpod used by L. coccinea. Identification of plants are: A) Acacia Koa (Leguminosae)/3 flower; B) Acacia Koa (Leguminosae),• flower; C) Metrosideroscollina (Myrtaceae); D) Acacia Koa (Leguminosae)seed pod; E) Cyanea Grimesiana (Lobeliaceae); F) Rollandia purpurelli/olia (Lobeliaceae); G) Rollandia sessili/olia (Lobeliaceae); H) Clermontia coerulea (Lobeliaceae); I) Cler- montia persici/olia (Lobeliaceae); J) Rollandia lanceolata (Lobeliaceae); K) Cyanea Baldwinii (Lobeliaceae). Most redrawn from Degener (1932). 18 ORNITHOLOGICAL MONOGRAPHS NO. 15 feeds on the poisonousberries of the Akia bush, Wikstroemiasp. (Thy- meleaceae). Baldwin (1953: 312 saw virens pierce the fruits and take juiceof the exoticJerusalem Cherry, Solanum pseudocapsicum (Solanaceae). Once I saw this bird peckingat the fruits of the Thimbleberry; however,it may have been obtainingsmall insectsfrom them. At another time I saw virensin a Clermontiasp. bush (Campanulaceae)which was loadedwith the small ripe orange fruits. The bird was not actually seen to eat the fruit, but therewere what appearedto be bits of orangefruit pulp adhering to its bill. l•lethods of Feeding: The earlier authorsreferred to above failed to record,at leastin their publishedworks, the minutiaeof methodsof feeding, a knowledgeof which is so necessaryfor a study of this sort. Concerningthe obtainingof animal food, these observersfound virens hunting for insectsmostly among the leaves, both dead and living, and blossomsand along the twigs and smaller branches of the forest trees, especiallyMetrosideros (Perkins, 1893: 108; 1903: 409; Munro, 1944: 102; Henshaw, 1902: 44; Baldwin, 1953: 314, 321). Perkins (1903: 409) and Henshaw (1902: 44) saw virens also obtaining insects less frequentlyfrom the large limbs and trunks of trees and from shrubberyand ferns, and Henshaw (loc. cit.) also observedit hunting over the surfaces of the broad leavesof the banana. Bryan (1908: 162, 164) wrote that the subspeciesfrom Molokai, L. v. wilsoni, searchedfor insectsmainly in the foliage of Metrosiderosbut also in low shrubs,especially the leaves and blossomsof the Mountain Naupaka, Scaevola sp. (Goodeniaceae). In my own field notes on L. virens, mostly concerningL. v. virens, are recordedabout twice as many instancesof what appearedto be searchings for insectsamongst leaves and twigs than upon larger branchesand trunks of trees. The former were recordedfour times from Acacia Koa (including twice for L. v. chloris), twice from Sophora, once from the Kolea tree, Myrsine Lessertiana (Myrsinaceae), and once from the Bastard Sandal- wood tree, or Naio, Myoporum sandwichense(Myoporaceae). The latter were all observedto take place on the limbs and trunks of Sophora. While feedingeither amongthe smallerplant structures(i.e., blossoms, leaves or twigs) or on the larger structures(i.e., limbs and trunks), for either nectar or insects, the birds are able to maneuver their bodies into almostany imaginableposition--right side up or upside down with the axis of the body eithervertical or horizontal;the positionleast observed was that in which the body's axis was vertical and with the head pointing downward. This maneuverabilityis no doubt an adaptationwhich allows this form to place its body rapidly in the most satisfactoryposition for immediate use of its jaws and tongue mechanismsin the taking of animal and plant foods. 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 19

While feedingon insectsamongst the twigs and leaves,the neck is bent and stretchedso that the bill can be thrust between the leaves or probed into massesof hairqike lichens. On one occasion,when a bird was probing amongSophora leaves and lichensgrowing on the twigs,the slendertongue wasplainly seen through the openedbill. I think the tonguein this instance was being used as a moistenedbrush for the capturing of tiny insects, althoughI couldnot actuallysee the prey. On three occasionsI observed virens capture larger insects or larvae by prying with the bill amongst lichens growing on limbs or plucking them from the bark of Sophora branchesor trunks. Theselarger insectswere graspedstrongly in the bill. They were killed either by tappingthem againsta branchand then eaten or by holdingthem down with a foot and peckingand pulling at them, while the head was simultaneouslybeing worked strenuouslyfrom side to side. Concerningthe obtainingof nectar not much has been written on the use and movementsof the bill and tonguewhile feedingfrom Metrosideros and Sophora blossoms--themore extensivelyutilized sourcesof nectar for L. virens. Baldwin (1953: 311) wrote as follows concerningvirens feeding from these blossoms: "At such times it seemed that nectar was being taken, judging by the manner in which the bird reached into the interior of the flowers and swallowedwithout wigglingthe mandible or submalarfeathers as would be expectedhad insectsbeen eaten." On many occasions I observed this same form take nectar from these blossoms. When feeding upon the typically leguminoseSophora blossoms,the bird plungesits bill and whole face into the blossomand keeps it there for aboutfive seconds,or the bill may be inserteddown to the nectarythrough a slit puncturedin the side of .the corolla by the proddingsof the bills of previousvisitors. When feeding from the multi-stamenedblossoms of the cymes (10-20 blossomsper cyme) of Metrosideros,virens dips the tip of its bill into the shallow,exposed nectaries through the long, slender stamenseither from above the nectary or from the side. A bird may visit a dozen or so cymeswithout pausingand spend several secondsat every cyme. I never saw any movementof the bill or tonguewhile watchingthis bird take nectar; however,twice, while I was observingthe Iiwi, Vestiaria coccinea,feed from Metrosideros and Sophora blossoms,the mandible and rostrum were seen to move apart and smack together several times while the bird was feedingfrom a singlecyme. I think that duringthis gaping the tongue was protruded into the nectary of the blossom,but I could not actually see the tongue. Perkins (1895: 127; 1903: 409) and Munro (1944: 102) wrote that virensobtains nectar from large lobelia blossomsby piercingwith the bill the base of the corollas to reach the nectaries; the smaller lobelia 20 ORNITHOLOGICAL MONOGRAPHS NO. 15 blossomsare not pierced according to Perkins. Henshaw (supra. cit.) observedthis samebehavior with respectto nasturtiumblossoms.

GREATERAMAKIHI, Lo;cops sagittirostris To my knowledgethe only naturalistswho ever saw this speciesalive in its natural habitat were Henry C. Palmer (Walter Rothschild'sfield collector) who discoveredit in 1892, R. C. L. Perkins who independently found it in 1895, H. W. Henshaw who observedit on several occasionsin the middle 1890's, and a personnamed A.M. Walcott whosename appears as the collectoron the tag of a BishopMuseum alcoholic specimen taken in 1901. Both Baldwin and I (Richards and Baldwin, 1953: 222) attempted to locate this speciesin the late 1940's and in 1950 but without success. Perkins (1903: 412-413) wrote that the forest crickets of the genus Paratrigonidiurnformed a large part of the stomach contents of eight specimensof this bird which he dissected;he wrote as follows: "Someof thesecrickets are speciallyattached to the Ieie [,Freycinetia arborea,which growsattached to the trunks of Metrosiderostrees] and somelive beneaththe loosebark of large Ohias,while othersare only found on certain ferns, which on more than one occasionI notice... [Loxops sagittirostris]visiting, no doubt in search of these insects. Caterpillarsand spiderswere also taken from the bird's stomachas well as a common carabid beetle, which lives at the base of the Ieie leaves,where these are closelyattached to the stem. For obtaining the latter and the crickets [Paratrigonidiumfreycinetiae according to Zimmerman (1948: 160)] which live in the same situation, the strong beak of . . . [Loxopssagittirostris] is well adapted. Once only I saw one feeding at the flower of the Ohia, and although I was unable to procure this for examination, I have little doubt that it was feeding on nectar, sinceits tongueis still unchangedfrom the form exhibited by that of the most persistentnectar-eaters. As in some [other] species of [Drepanididae]the habit of feeding on nectar probably survives as a rare occurrence,the typicalform of the tonguebeing fully perserved, sinceit assistsin obtainingsome of theseinsects which form a large part of the bird's food, and not becausenectar is of much importance as an article of diet." Henshaw (1902: 43) wrote that sagittirostris"seems to live chiefly upon insectswhich it gleansfrom the foliage of the ohias, to which tree it it seemsto confine its attention chiefly."

HAWAII CREEPER,Loxops maculata mana Types ofFoodTaken: Perkins(1895: 122; 1901: 570; 1903: 416) reportedthat L. rnaculatais almostentirely insectivorous, feeding chiefly on exposedcaterpillars, spiders and small moths; larger moths, myriapods, slugsand beetlesare also taken. Perkins (1901: 570), Henshaw (1902: 47) and I never saw mana take nectar. Henshaw (loc. cit.) reported this 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 21 form to feed mostlyupon beetlelarvae. Munro (1944: 106) found insects, "grubs"and insecteggs in the stomachsof specimensof this race. On five differentoccasions I observedit to eat unidentifiedinsect and lepidopterous larvae and adult insects. l•Iethods of Feeding: Perkins (1895: 122; 1903: 415-416) observed the severalsubspecies of maculatahunting their arthropodprey on the trunksand branchesof foresttrees and bushesand sometimesgoing down to the basesof the trunksand, rarely, even to the ground. He reported them climbingeither up or down the trunksand on the upper and lower surfacesof horizontallimbs. He notedthat the speciessearches dead wood and beneathtree bark for caterpillars.Those exposed on foliageare also caught. Henshaw(1902: 47) reportedthat the Hawaii subspeciescreeps alongtree trunksand branchestaking insectsfrom the intersticesof bark, mossesand lichens. Perkins (1903: 415) noted this same race to be partial to large Koa trees and also to Myoporum and Sophora trees. I foundthat, out of seventeenobservations of feedingon the part of mana, eightwere on Acacia Koa, five on Sophora,two on Metrosideros,one on a deadMyoporum, and oneon an Olapatree, Cheirodendronsp. (Araliaceae). I noticedin all casesthat the birdswere creeping on the branchesor trunks. L. m. mana seemsto be just as adept at creepingright side up or upside down alonghorizontal branches as it is at creepingstraight up or straight down on verticalbranches and trunks. My field notesindicate almostthe samenumber of observationsmade on the birdsin all four of thesecategories. They werealso observed to creepin a roughspiral around vertical branches or trunks. While creepingin any positionthe bird's body is held closeto the bark and its legsclose to its belly; the bird advancesjerkfly, aboutone- half to one inch everyfew seconds,by either steppingforward, one foot at a time, or by hoppingforward, both feet movingat the sametime. The wingsare flicked now and then, presumablyin the maintenanceof balance duringadvancement or in the eventof slippageof the claws. This creeping ability appearsto me to be a highly evolvedbehavioral adaptation which allowsthis species to positionits body so that its jaw and tonguemechanisms can be made use of rapidly and efficiently. While the animal creeps,it continuouslypeers from side to side inter- rupting this only when it probes,pecks and pries with the bill into crevices in and under the edgesof sheetsof bark or into moundsof hair-like lichen or into moss. Out of twenty-five descriptionsin my field notes twelve of theseprobings were at the bark of branchesand trunks, sevenwere into hair-like lichens, two onto exposedtwigs, two into moss on branches,one in rotten wood, and one underneatha flat lichen growingon a branch. I sawone ply so hard at Koa bark that piecesof bark aboutthree quarters of an inch acrosswere falling from the site. 22 ORNITHOLOGICAL MONOGRAPHS NO. 15

Perkins (1895: 122; 1903: 416) mentionedthat when maculatacatches large moths, it holds them down with its claws and tears off the wings before eating them. Bryan (1908: 165) wrote of the Molokai Creeper, L. m. flammea, that sometimes"moths are taken of such size that . . . [the birds] are compelledto hold them under their feet and pull them to piecesso as to devourthem piecemeal."He noted that this form seldom, if ever, takesinsects on the wing. I, likewise,never saw the Hawaii Creeper, mana, either perchedor in flight, catch any insectson the wing. Twice I observedmarta capture large insectlarvae and beat them againsta limb beforeeating them; one was an inch-longcaterpillar which was removed from the bark of a Koa limb only after vigorouspecking. On otheroccasions smallerinsects were capturedfrom the bark or lichensand immediately devoured.

1VIAUICREEPER, Loxops maculata newtoni Types of Food Taken: The mainlyarthropod diet of the speciesnoted by Perkins(supra, loc. cit.) wouldapply to thisrace also. Henshaw(1902: 50) sawthis creeper feed its youngsmall green caterpillars. In additionto this animalfood, on rare occasionsMetrosideros nectar is alsotaken by the Maui Creeper(Perkins, supra, loc. cit.; Henshaw,1902: 50). The only other Creeperswhich take nectar,and only rarely from Metrosideros,are the LanaiCreeper, L. m. montana,(Perkins, supra, loc. cit.) and the Kauai Creeper,L. m. bairdi, (Munro, 1944: 105). ßIethotla of Feeding: Again,the remarksof Perkins(supra, loc. cit.) on the creepingand feedingbehavior of the specieswould apply to the Maui form. He noted that the Maui form aboundsin large Koa trees, but that it is also found in forests devoid of this tree. Henshaw (1902: 49-50) observednewtoni take small green caterpillarsfrom Koa and Metrosideros. He noted that this form hunts not only on tree trunks and branchesbut alsospends much time in the undergrowthand "not rarely descendseven to the groundin its huntingexcursions." He went so far as to say that it is "far more an inhabitant of the scrub than the trees." I observed this sub- speciesonly on two differentdays, but the first one I saw was in the three- to four-footheather-like Pukeawe shrub, Cyathodessp. (Epacrida- ceae), about 100 feet from the edge of a low forest. On a number of occasionsduring these two days I saw newtonifeeding along the small branches,which were covered with flat lichens, and among the twigs and leavesof the followingtrees: Metrosideros;Cheirodendron sp.; the Alani, Pelea sp., (Rutaceae),and Myrsinesp. I presumedthat the birds were huntinginsects, for theirheads were constantly moving, the billsbeing probed under lichensand amongthe leaves. The creepingability of this race seems as goodas that of the Hawaii Creeper,but it was notedthat the Maui race 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 23 spentmuch time among the leaves of the trees, whereasI never observed the Hawaii Creeperin this situation.

HAWAII AKEPA,Loxops coccineacoccinea Types of Food Taken: Perkins (1895: 121; 1903: 418-419) reportedthat coccineafeeds mostly upon caterpillars and the smallerspiders, the formerconsisting both of the exposedspan-worms or loopersand also smaller forms concealed inside of leaf buds or between leaves fastened together by the larvae. Henshaw (1902: 58, 62) observedthe form from Hawaiiand the Maui Akepato feedmostly upon soft, small insects, caterpillarsand very small spiders.Munro (1944: 109-110, 112) wrote that Akepastake scaleinsects, that the Hawaii race feedsmostly upon insects, caterpillarsand spidersand that the Kauai subspecies,L. c. caeruleirostris, takes caterpillars,spider eggs, coccids and other insects.Perkins (loc. cit.) observedL. c. ochraceaand L. c. caeruleirostristaking nectar on rare occasionsfrom Metrosiderosblossoms; this was provenby the fact that nectarwas dripping from the bills of freshlyshot specimens. I never observed coccinea to take nectar. Methodsof Feeding: Accordingto Perkins(1895: 121; 1901: 569; 1903: 418-419) the speciesseeks its arthropodprey chieflyfrom the foliageand leaf buds,and more rarely on the branches,of certainforest treeswhere wood-feeding larvae are obtained.L. c. coccineais partialto Koa forestsbut alsooccurs in forestdevoid of this acacia(1903: 417). In the Koas,which act as hostplants to manyspecies of Lepidoptera,the Hawaiianrace obtains its foodmainly amongst the Koa phyllodes,but in forestof Ohia it searchesmainly the terminalleaf budsof this tree for the lepidopteranlarvae concealed within the buds.In the dryerupland forests it huntsnot only on the Koasbut alsoin the smallerNaio; Aalii, Dodonaea sp. (Sapindaceae);and other lessertrees (1903: 417-418). Henshaw (1902: 58-59), also,noted that this race is partialto theKoa forests,where it searchesfor its foodamong the phyllodes,small outer twigs and flowers of this tree, but that its food is also taken on the Naio, Myoporum,and Mamani,Sophora. I foundthe Hawaii Akepa feedingonly in AcaciaKoa and Metrosideros.Observations on the feedingof twelveof thesebirds were made in a forestof mixedKoa, Ohia, Naio, Mamani and other trees. Seven werefeeding in theOhias and five in theKoas; all appearedto be searching for insects. Different individuals of these twelve birds were seen to search Ohialeaves four times,Ohia leaf budsthree times, the longgreen crescent- shapedKoa phyllodesfive timesand the dry podsof the Koa five times. Thesecrude data may aid in givingan idea of the food niche of this race. I noted that while huntingin thesesituations the birds could maneuver theirbodies into almostany imaginableposition with respectto the vertical 24 ORNITHOLOGICAL MONOGRAPHS NO. 15 and horizontal; this of courseallows the bird to place its body in the mostsatisfactory position for usingits jaw and tonguemechanisms in the captureof its prey--be it exposedor concealed.Taking exposedinsects from the Koa phyllodesand Ohia leavesundoubtedly is the simpler and easiermethod. To do this the bird merely has to perch on an Ohia twig or climb amongstthe Koa phyllodesand pick off the caterpillarsand spiders from the basesand surfacesof the leavesor phyllodes.However, to capture insectsor larvae living insidethe Ohia leaf budsor dry Koa pods (Fig. 3) more strenuouseffort must be exerted. On one occasionI observed (field notes) an Hawaii Akepa feedingin an Ohia tree "climb aboutnear the ends of the twigs,bending its neckback, & prayinginto the Ohia buds.... It kept this up for about 4 min." At anothertime I observed(field notes) one of thesebirds in an Ohia, and it "moved from one bunch of leaf buds to the next, poking into the buds from above with its short bill. The neck was often bent so that the chin restedalmost on the anterior part of the neck." I observed(field notes) on anotheroccasion an Akepa, feedingabove me in a Koa aboutforty feet tall, catchonto a dry pod hangingvertically downward. "This it clungto with both legswhile it attemptedto openthe bottom of the pod with its bill. A crackingnoise could be heard. The bird & pod swingaround crazily during this operation.At times the bird's head was doubledup almost under its breast." At another time (field notes) an Akepa "was seen to hang upside down on a Koa pod & pry at it with the bill." The Akepas, like the Holarctic Crossbill,Loxia curvirostra,have asymmetrical mandibularand maxillary (or rostral) rhamphothecaewhich cross at their tips. Parts of the skeletonand some of the musclesof the jaw apparatusare also asymmetricallyarranged. It was the opinion of Perkins (1903: 418-420) that the mandibularrhamphothecal asymmetry was of definite use to the speciesin some of its feeding habits: "The abnormal structureof the mandible[rhamphotheca] is dearly connectedwith the habit of seekingfood in the doselyimbricated buds of someof the foresttrees." He adds:

"The essentialuse of the distorted mandible of . . . [Loxops coccinea] is without the least doubt for the extractionof insectsliving hidden in the leaf-budsof certain forest trees. These buds may not inaptly be comparedto the pine conesfrom which... [Loxia curvirostra]procures its food, althoughtheir much softer substanceby no means requires the more powerful implementsof the crossbill. As has been already mentionedthe bill of Loxops [coccinea]is also useful in opening out the koa phyllodes,when fastenedtogether by certain caterpillars,or by somespiders (of whichit is extremelyfond) whichthus conceal their nests." 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 25

Here Perkinsneatly hit upon what we think is the essentialuse to which the asynunetricalrhamphothecae are put, but we do not think he was aware of the skeletaland muscularasymmetries or understoodthe methodby which the whole muscular,skeletal and horny bill apparatusis used. Even the avian anatomist Gadow did not mention (1899: 246) any skeletalor muscularasymmetries, or the methodby which such an asym- metricalapparatus could be used. However,he did recognizea parallelism betweenthe Akepa and the crossbillsand, concerningthe former,mentioned ( loc. cit. ) "the under jaws of these... birds are not symmetrical the distal half of the under jaw is twistedeither to the right or to the left.... There is not the slightestdoubt that this asymmetryis acquiredindividually by their twistingopen husksor seeds,or cracksof bark, in searchof their food." He wasof courseguessing when he usedthe terms"husks or seeds,or cracks of bark"; however,the principleof twistingprobably does apply, but on the other hand we think he (Gadow, 1891: 495-496) incorrectlyunder- stood the method by which the crossbillsopen pine cones and was not cognizantof the principlesinvolved in the jaw apparatusemployed by either the crossbillsor the Akepa. Henshaw(1902: 58-59) appreciatedthe parallelismbetween the Akepa and the crossbillsin respectto the asymmetryof the mandibularrhamphothecae and that the asymmetryin coccineaundoubtedly somehow is correlated with its feeding habits. Although he had observedthirty or forty individualsof the Akepa feeding,he did not visualizethe analogybetween an Ohia leaf bud or Koa pod and a pine cone, and therefore,could not offer any explanationfor the use of this rhamphothecalasymmetry in the feedinghabits of the Akepa. He was further misledby Gadow'serroneous mention of "husks and seeds,or crack of bark," for he knew the species well enoughto realize that it is not granivorousand that it doesnot hunt its prey by twisting open crevicesin the bark. He was seeminglyunaware of any internal anatomicalasymmetries and stated that he did not know of any opinionsof Perkins on the matter. Munro (1944: 109) suggestedthat the crossedmandibular sheath is useful for removingscale insects from leaves,for he found scale insectsin the stomachsof some specimens.It may be used this way; however, since Coccidae are not native to the Hawaiian Islands (Baldwin, 1952: 315), it seemsimpossible that the presenceof theseinsects for only 100 years or less in the niche of the Akepa could have acted as a selectiveforce for such drastic changes(from the presumedancestral condtion) in feeding behaviorand concomitantontogenetically variable morphological asymmetry. Furthermore, the removal of scale insectsfrom a leaf or stem, even by a 26 ORNITHOLOGICAL MONOGRAPHS NO. 15 twistingmethod, would not involveexerting enough force to developasym- metricallythe jaw musculatureand skeletonof an individualbird in its lifetime. Concerningthe use of the tonguein the feedingof this species,Perkins (1903: 419) has this to say: "The thin, long, honey-suckingform of tongueis fully preservedin all [theraces], in spiteof the smallattention paid to... [nectar],but with the distortedmandible it is obviouslya very efficienthelp in procuring the larvae which feed in the terminal buds of trees as well as those whichlive betweenleaves fastened together, which I have seenthem extracting."

SUMMARY OF FOODS AND METHODS OF FEEDING L. v. virens: The orderof preferenceof food for this form seemsto be soft-bodiedinsects and spiders,nectar, and juicesof fruits and fruit pulp. The arthropodprey is huntedmostly among leaves, twigs, blossoms and smallerbarnches and, to a lesserextent, upon larger branches and trunksof Metrosideros,Acacia Koa and $ophora. The larger insectsare grasped with the bill. Possiblyvery small onesare brushedup with the tongue. Nectaris takenmainly from Metrosiderosand $ophoraand sometimesfrom the campanulateblossoms of the nativelobelias (Campanulaceae) and from otherplants. This species hunts its preymore amongst the leavesand twigs and lessupon the largebranches and trunksof treesthan the otherspecies of Loxopsexcept for coccineawhich spends almost all of its time hunting amongstthe leaves.L. v. virenstakes much more nectar than any of the other speciesof Loxops. L. sagittirostris:The GreaterAmakihi fed mostlyupon certaincrickets and a carabid beetle which live at the bases of the stout leaves of Freycinetia. Caterpillarsand spiders were also taken and these possibly from the foliage of Metrosiderosand ferns. Large prey undoubtedlywas graspedwith the bill. The tongueprobably was alsoused for capturingsmaller insects in crevicesat the basesof Freycinetialeaves, and on rare occasionsthe tongue wasprobably used in takingnectar from Metrosiderosblossoms. L. m. mana: The Hawaii Creeper is almost completelyinsectivorous, feedingmostly on exposedcaterpillars, spiders, and moths, although myriapods,slugs and beetlesare alsotaken. The bird creepsover the larger branches and trunks of Acacia Koa, Myoporum and Metrosideros trees probingand prying at andalso under bark, lichens and moss to obtainthese animals. The larger mothsand larvae are held down with the daws and torn apart with the bill. L. m. newtoni: The animal food of the Maui Creeper is practicallythe same as that of the Hawaii race. On rare occasions newtoni also takes nectar 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 27 from Metrosiderosblossoms. Like marta it hunts its prey by creepingover the large branchesand trunks of treessuch as Acacia Koa and Metrosideros, but it also spendsmuch time searchingthe leavesof trees and underbrush which mana does not do. L. c. coccinea: The Akepas, like the Creepers,are almostentirely insectivorous,feeding mostly upon soft, small insects,caterpillars and small spidersobtained chiefly from the foliage of Acacia Koa and Metrosideros. Only rarely do they hunt their prey on the branchesof the trees. Ohia leaf buds and dry Koa pods containingcaterpillars, and Koa phyllodessewn togetherby spidersor lepidopterouslarvae, are openedby these crossbilled forms in a manner probablyvery similar to that used by Loxia in opening pine cones. It is probablethat the tongueof the Akepa is used to remove larvae from these openedplant structures.Although the Maui and Kauai Akepastake nectarfrom Ohia blossomson rare occasions,the Hawaii race, as far as I know, has never been observed to do so.

RHAMPHOTHECAE OF THE BEAK

Deseription: The horny sheaths,or rhamphothecaeof the maxilla and the mandibleof the severalspecies of Loxopsdiffer from one anothermostly in generalexternal shape and curvature,and in the relativesize of the opercula coveringthe nostrils(Pls. 1 and 2). Differencesin the structureof the floor and roof of the mouthwithin the bill properalso occur. Many of thesefeatures can be seen in the excellent drawings of the bills of Loxops and other drepanididsfound in Rothschild'smonograph (1893-1900: pl. 82, figs. 8, 11, 15, 18, 25). Examinationof alcoholicspecimens (except for sagittirostris)and of studyskins of all speciesand many subspeciesof Loxopsrevealed that the rhamphothecaeof both the maxilla and the mandible of virens are more decurvedthan those of the other forms. Curvature of all parts of the rhamphothecaeis judged relative to the longitudinalaxis along the length of the jugal bar. In lateral view of the bill, this curvatureis obviousboth in the dorsalsurface and ventral edge of the maxillary rhamphothecaand in the dorsal edge and ventral surface of the mandibular rhamphotheca. Closer comparisonof some of the taxa in the subgenusViridonia shows someinteresting cases of characterdisplacement. Sympatry exists between v. stejnegeri and parva on Kauai, and between v. virens and sagittirostris on Hawaii; in each case,the bills of thesesympatric forms have become markedlydifferent. L. parva is the smallestspecies in the genusand in bill shapeis muchlike v. virensexcept that its bill is somewhatstraighter. Pre- sumablyit was an early geographicrepresentative of the virensgroup on Kauai (Amadon, 1950: 165). With subsequentreinvasion of this island 28 ORNITHOLOGICAL MONOGRAPHS NO. 15 by anotherpopulation of Loxopsvirens, competition between the ancestorof parva and the morerecently arriving progenitor of stejnegeriwas presumably responsiblefor divergent evolution of their bills, resulting in a smaller straighterbill in parva and a larger,strongly decurved bill in stejnegeri.This latter form has the most stronglydecurved bill in the genus and in size is second only to sagittirostris.These forms differ markedly in feeding habits (Amadon, 1950: 245-246), parva being very similar to other races of virens and virens ste]negeribeing specializedfor obtaining insectsfrom beneathbark (a creeper-likefeeding habit, but presumablydifferent from that utilized by maculata). The interactionsbetween the speciesof Loxops on Kauai is extremelyinteresting because of the double invasionin the subgenusViridonia and the double developmentof creeper habits in two speciesof different subgenera;it deservesto be studiedfurther in greater detail. In similarfashion, v. virensmay have been the originalform on Hawaii with a secondpopulation of virens,the progenitorof sagittirostris,invading this island at a later time. Competitionbetween the ancestorsof v. virens and of sagittirostriswas again responsiblefor the divergentevolution of their bills, resultingin the very large, straight bill in sagittirostriswhich is the largestspecies in the genuspossessing the most robust bill. From the little we know about the feeding habits of this species,we can correlate the differencein bill morphologybetween it and v. virens with differences in their feeding habits. Thus, membersof the subgenusViridonia displaytwo excellentexamples of the classicalpattern of insularreinvasion with subsequentdivergence of the sympatricforms into different niche relationshipswith respectto their feeding habits and the resultant divergenceof their morphologicalfeatures (see also Amadon, 1950: 246-247; Bock, 1970). Membersof the subgenusParoreomyza have a straighterand more rug- gedlyconstructed bill than virens,and alsodisplay considerable geographic variation. The Malokai subspecies,maculata flammea has the most decurved bill, one that is almost as curved as that of virens but stouter. In degrees of curvature, newtoni comes next; its bill is so much straighter that it can hardly be called decurvedat all. The bill of mana is even straighterstill with the tips of both rhamphothecaeextending straight out to end in an extremely acute angle. Moreover, its bill is more ruggedly constructed than that of either virens or newtoni. Its strongbuild is especiallynoticeable in the greaterheight and width of the hornyridge betweenthe dorsalnasal operculaeof mana. The cline of increasingstraighter, stronger bills in the races of L. maculata from Molokai to Hawaii should be noted, although it is not known whether this representsa pattern of gene flow or is strictly coincidental. 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 29

The symmetryof the bill in thesespecies is shownin their dorsalaspect (P1. 2). Some differencesexist in the attenuation of the bill, those of v. virens and parva are thinner than the others, but the differences are considerablyless than differencesin curvature as seen in lateral view. The bill of coccinea is shortest of all and the most finch-like or siskin- like in appearanceexcept for a definitecurvature of the dorsalsurface of the rostal rhamphotheca.Some geographicvariation in size existsas that shownbetween c. coccineaand c. caeruleirostris.The tips of both the maxillaryand mandibularrhamphothecae are asymmetricalas seenin dorsal view. Becauseof these asymmetriesin the rhamphothecaeand those in severalskeletal structures(Pls. 6 and 9) and jaw muscles(Pls. 16, 17, and 18), individualsof coccineacan be referred to as either "right-billed" or "left-billed." A right-billed individual is one in which the mandibular rhamphothecaprotrudes or crossesto the right from beneaththe maxillary rhamphothecawhich crossesto the left (e.g., c. coccinea',PI. 2f and g; and c. caeruleirostris,P1. 2i and j). A left-billed bird is one in which the oppositerelationships exist (e.g., c. coccinea,P1. 2h). In a right-billed individual,the right, or cutting, edge of the tip of the mandibularhorny sheathis wider, flatter, and usually higher than the left edge. The left, or cutting,edge of the tip of the rostralrhamphotheca of a right-billedbird is likewisewider, flatter, but lower than the right. Thesecutting edges are, hence, knife-like and are able to slice through plant material, as will be discussedbelow. In a few specimens,both mandibularand maxillarycutting edgesare gougedout where they had been worn away through use (e.g., PI. 2i and j). Most of the descriptionof the cranial elementsand jaw musclesand discussionof the jaw muscleswill be for right-billedbirds. In the four speciesfor which alcoholic specimenswere available, and probablyalso in sagittirostris,the edgesof the maxillaryrhamphotheca are knife-likeand overlapthe similarlysharp edges of the mandibularrhamphotheca when the bill is closed,except at the tips of the crossedbill of coccinea. A medial trough or slot existson the ventral surfaceof the maxilla in the four forms for which specimenswere availableand presumablyalso in sagittirostris.This slot is of an invertedU-shape in crosssection and is narrower at the tip of the rostrum than at its posteriorend. From about the middle of the maxilla, two parallel ridgesextend posteriorly on either side formingthe lateral walls of the trough. Theseparallel ridges are most evident in newtoni and virens and less evident in mana and coccinea. In newtonia low, sharpmedian ridge extendsthe length of the trough to an anteriorly-pointing,wedge-shaped mound of tissuein the posteriorend of the trough. These structures(i.e., the median ridge and wedge-shaped 30 ORNITHOLOGICAL MONOGRAPHS NO. 15 mound) exist in the other forms studied but to a much lesser degree of development.Beecher (1953: 312-313; fig. 15) describesthe structure of the horny palatein Psittirostriscantans which is similarto that found in Loxops. The tongue of all forms fits into this dorsal, maxillary slot, the closestfit being in virens and newtoni. On the dorsalsurface of the mandibularpart of the bill there is another troughor slot which is also U-shapedin crosssection. In virensthe dorsal edgesof this troughoverhang its lumen. In newtonithe walls are practically vertical with an almostimperceptible amount of overhang. In mana and coccineathe walls of this troughare lesssteep (i.e., they are neitherover- hanging nor vertical but diverge laterally). In the two creepersstudied there is a low, sharp median ridge in the floor of this mandibularslot; it is more pronouncedin the Maui race. It is barely perceptiblein virens and coccinea. When the tongueis depressed,this ridge fits into a narrow, median slit in the horny undercoatingof the tongueproper. All of the forms examinedhave both dorsal and ventral nasal opercula. The dorsaloperculum is a flap of corneousskin more or lesscovering the openingof the externalnaris from above. The ventral operculumis a similar flap of skin,but smallerin area,which is overlappedby the dorsaloperculum and which extendspartway into the openingof the externalnaris from the ventralmargin of thisopening (Gadow, 1891 and 1899: 223,224, 229, 230, 231,234, 245-247, figs. 11, 17, 36, 37, 42, 49; Amadon,1950: 226-227). Two bilaterallyplaced preconchae hang within the vestibuleof the nasal cavity (Stresemann,1927-1934: 120-121, cited in Amadon, 1950: 227). The figures in Rothschild'smonograph (loc. cit.) indicate in sagittirostris the presenceof a rather large dorsal operculumif not also a ventral one. Furthermore,he states (op. cit.: 107) that the "nostrilsare protectedby an upper operculum.... "We found the dorsal and ventral operculain virens larger than in the other forms examined. Those of newtoni are only slightlysmaller, and those of mana and coccineaare developedstill less,those of coccineabeing the smallest.In virensand newtonithe ventral opercnlahide from view the cartilaginousbilateral preconchaehanging in the vestibuleof the nasal cavity; in the mana and coccineathese structures can be seenin lateral aspect. Functional interpretation: The structure of the horny parts of the bill may be correlatedto someextent with the type of food taken by the bird and the methodsby which it is obtained. These correlationsbetween the rhamphothecalmorphology and feedingobservations are largely hypotheses to be testedby further observations,not proven facts. The knife-like edgesof both rhamphothecaeare not only useful to the speciesof this primarily insectivorousgenus for graspingand killing of their 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 31 prey, but the closefit of the one againstthe other while the bill is closed also servesto keep desiccationof the interior of the mouth at a minimum. The decurvedbill of virensis probablyan adaptationfor obtainingnectar from curvedlobelia blossoms and secondarilyfrom Metrosideros,Sophora and other blossoms.Although bill shapein virens presumablyevolved in con- nectionwith feedingon nectar from lobella blossoms,this speciesnow obtains most of its nectar from the shallow flowers of Metrosideros, Sophora and other plants. It is Perkins' hypothesis(1903: 384-385) that the nectar-feedingdre- panididsobtained their principal supply of food from the blossomsof the specializedand ecologicallyrestricted lobelias (Campanulaceae)before the relatively recent invasionof the Hawaiian Islands by the South Pacific Metrosiderostree (we might add Sophoraand perhapsothers). The nectar- feedingspecies of the Drepanidinaeoriginally fed on nectarfrom the deep, curved blossomsof lobelias; their decurvedbills are an adaptation for feedingon theseflowers. Although sourcesof nectar are abundantin Metro- sideros and Sophora blossoms,observations of earlier naturalists (Baldwin, 1953: 310; Perkins, 1903: 401,422-424) suggestthat Drepanis, Vestiaria, Hemignathusobscurus and Hemignathusprocerus preferred the nectar of lobelias over that of Metrosideros. L. virenshas been observedto take nectarfrom lobeliasalthough it feeds more frequentlyfrom Metrosiderosand Sophora. This speciesis often consideredto be the most primitive member of the genusLoxops which is, in turn, regardedas the mostprimitive genusin the Psittirostrinaeand by some as the most primitive genusin the entire family (Amadon, 1950: 231). If virens evolved from the Drepanidinae,it would have a decurvedbill inherited from these forms as an adaptation for feeding on deep, curved lobelia blossoms.If Loxops is the primitive genusin the family from which both the Drepanidinaeand the Psittirostrinaeevolved, then it would have fed originallyon lobelia blossomsif Perkins'hypothesis is correct,and the bill would have becomedecurved as an adaptationfor feedingon thesedeep curved flowers. Whatever the evolution of Loxops has been, its decurved bill correspondsin shapeto the curved campanulateblossoms of lobelias,and a straightbill would be of significantlyless advantageto a bird feedingon these flowers. The decurvedbill of the virenswould thus be preadaptedto taking nectar from the shallownectarines of the long-stamenedMetrosideros cymes, from Sophorablossoms and from flowers of other more recently invading trees. The curved bill of a short-billed form like virens would be more suitable for obtainingnectar from shallow Metrosiderosnectaries than would a straightone, becausethe bird could obtain nectarby shovingthe tip of its 32 ORNITHOLOGICAL MONOGRAPHS NO. 15 curvedbill amongthe stamensfrom the sideof the cymerather than down fromthe top. Thetip of thedecurved bill wouldthen dip intothe nectaries withouthaving the long uprightstamens of the blossomspoke into the bird'seyes, which of necessityshould be kept opento locatethe nectaries. A straight-billedspecies would have to plungeits wholeface from above into the massof stamensin orderto reachthe nectarieseffectively. Nor would it be ableto reachefficiently all nectariesof a cymefrom the sidebecause the tip of the bill is not bent downwardsallowing easy entry into those nectariestoward the centerof the cyme. Richardshas observedboth virens and Himationes. sanguineaobtaining nectar from Metrosiderosblossoms by shovingthe bill amongthe stamensfrom the side of the cyme,but, on the other hand, he has seen them also plungethe whole bill and face into the stamensfrom abovethe cyme. Althoughthe decurvedbill of virensprobably did not evolveas a primary adaptationfor insect-feeding,it is presumablyreasonably suited for those feedinghabits. Many insectivorousbirds have slightlydecurved bills, and it is not obviouswhether a slightlydecurved bill or a straightbill is better suitedfor insectfeeding. One advantageto a slightlydecurved bill is that the tips are anteriorand ventral to the eye so that the bird may be able to seeits prey betweenthe tips easierthan in a straightbilled species. The straightbills of the creepers,mana and newtoni and of sagittirostris areprobably adaptations to theirinsectivorous habits. Many of the arthropods eatenby thesebirds are exposedon limbs or foliage. Othersliving beneath bark, lichens,moss, or at the basesof leavesare exposedby the actionsof the birds. Someof theseanimals move rapidly and must be snatchedup quickly. The problemlies in whethera straightbill is better adaptedthan a curved one for quick capture of insects. It seemsreasonable to assume that a straightbill is better becauseonce the tip of a straightbill is aimed accuratelyat an insect,the axis of the bill can be more easily maintained than can the axis of a curved bill during the subsequentrapid forward extensionof the neck, openingof the bill and seizureof the prey. Certainly an extremely curved-billedform would have to execute separate, more difficultcompensatory movements of the mandibleand rostrum,and perhaps of the head, in order that the tips of the bill arrive accuratelyon the target at the end of a rapid thrustingand snatchingmaneuver. But the difference in the curvature of virens and of marta, newtoni and sagittirostrisis not extreme. And it may be questionedwhether a slightlydecurved bill may not be more advantageousthan a straightone as mentionedabove. In any case, it is not definite that the straight bill of these insectivorousforms of Loxops is an adaptationfor quick captureof their prey. A straightbill is a strongerstructure than a curvedone for certaintypes 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 33 of forcesacting upon it, but it is not inherentlystronger than curved ones with respectto all forces(see Bock, 1966, for a discussionof bill shapesand forces acting on them). The almost completelyinsectivorous mana has a very straightbill whichis effectivelyused as a pry and wedgeunder bark and lichensor to graspactively and tear off piecesof bark. Theseactions would place forceson the tip of the bill, directedposteriorly or laterally. A straightbill would be strongerwith respectto such forces; as seen in the straightbill of woodpeckers(Bock, 1966). L. m. newtonifeeds more amongthe foliageand even rarely takesnectar; its bill is more decurved and less robust than that of mana as might be expected becauseof less posteriorlyand laterallydirected forces on the tips of the bill, and because of the advantagesof a curved bill in nectar feeding. L. sagittirostriscaptures much of its prey in the deep.crevices at the basesof Freycinetialeaves. Judgingfrom the attenuated,wedge-like shape of the bill and the relativelylarge size of the mandibulardepressor muscle of this species,we presumethat this bird first forces its bill betweenthe base of the leaf and the stalk of the plant and then gapeswidely which involvesstrong depressionof the mandible. The forces on the mandible would be partly posteriorlydirected ones on its tip when the bill is wedged into the spacebetween the leaf baseand stem, and partly dorsoventralones when the bird gapes. Moreover, a decurvedbill may not be well suited for gapingbecause the curved dorsalsurface of the rostrum would tend to slip and pushthe bill backwardsout of the creviceas the bill openedinstead of the bill remainingin positionand wideningthe size of the hole. Once the bird has widenedthe spacebetween the leaf and stem and disturbed this area by its movements,the cricketsand beetlesliving there may scramble about, thus placing themselvesin positionsmore vulnerableto the now gapedbeak of the predatorthan if they remainedin a resting,or "crouched," position deep in the crevice. The short, siskin-likebill of coccineais no doubt an adaptationallowing greater leverage in the opening of leaf buds and Koa pods. The short, relatively wide bill is well suited to resist the laterally directedforces on the tips of the mandibleand rostrumwhen they are rotatedin the leaf buds and Koa pods,as well as the ventrodorsalforces on the tips when the bird gapesand spreadsthe cut material apart to exposeinsects within the leaf bud or Koa pod. The cutting edgesof the two rhamphothecae(the right edgeof the mandibularand left of the rostralof a right-billedindividual), being wider and flatter, serve as better cutting edgesthan those of the oppositesides. Furthermore,being wider, they possessa greaterpotential of viable horny material which can be expendedin wear before the actual bonytissue of the dentaryand premaxillary bones is exposed.The mandibular 34 ORNITHOLOGICAL MONOGRAPHS NO. 15 cutting edge is kept sharp by its rubbing againstthe rostral rhamphothecal edge directly dorsal to it, and the rostal cutting edge is sharpenedby its grinding against the mandibular rhamphothecaledge directly beneath it. This processis analogousto the method involved in the maintenanceof sharpedges on the mandibulartusks of pigs in which the flat surfaceof the blunt dorsal tusk is ground against the sharp, beveled, ventral one. That the mandibularcutting edge is usuallyhigher than the oppositeone and that the rostral cuttingedge is lower than the oppositeone of coccinea suggestsan adaptationwhich facilitates cutting of leaf tissue or springing apart halves of dry Koa pods during strong adductionof the lower jaw and simultaneousdepression of the upper jaw. Becausethe cutting edges are situatedin this way they tend to guide the tips of the rhamphothecae towardsthe frontal plane rather than away from it during closingof the bill, allowingthe musclesinvolved to perform at maximumefficiency. Fur- thermore, during depressionof the mandible and simultaneouselevation of the maxilla, these edgeswould not be involved in cutting; insteadthe rhamphothecalsurfaces immediately ventral to the mandibularcutting edge and immediatelydorsal to the maxillary cutting edge would be pushed broadsideinto the bud tissue or against the pod margin and would act, therefore, as spreaders. The rostal and mandibular slots or troughs, found in the bills of the speciesof Loxops studied, serve to house and guide the tongue. It also seemsplausible that in the forms which feed on nectar, virens, sagittirostrisand newtoni (occasionally),these slots, when moistenedwith saliva and nectar, would act as a sealedtube facilitatingthe conductionof nectar or water up the tube-like or trough-liketongue. It should be noted that the closestfit betweenslots and tongueis to be found in virensand newtoni, both of which are partly nectarivorons. The low medianridge in the roof of the maxillary slot of newtoniand in the floor of the mandibular slot of both creeperspossibly aid further in guidingthe tongueby meansof the narrowmedian slot alongthe ventral surface of the anterior end of the tongue. The anteriorly-pointingwedge-shaped mound of tissue in the posterior part of the rostraltrough, most noticeable in the newtonibut alsopresent in the other forms, may act as a "bumper" againstwhich the tongue is held snuglyby muscularcontraction when not in use. The dorsal and ventral nasal opercula,which are better developedin the nectar-feedingvirens, newtoni and presumablysagittirostris, together with the nasalpreconchae, undoubtedly function as baffleswhich help to keep pollen grainsfrom enteringand pluggingthe vestibuleof the nasal cavity and may be regarded as adaptationsfor nectar-feeding(Amadon, 1950: 227). 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 35

CRANIAL OSTEOLOGY

INTRODUCTION

The skull and mandible of L. v. virens will be described in detail first, after which the crania of the four taxa will be compared. The general descriptionwill emphasizethe features of the skull characteristicof the genusLoxops, or at least of the family Drepanididae. Special attentionis given to thosecranial featuresessential for understandinga descriptionof the jaw musculatureand the functionalproperties of the feedingapparatus. Becausea detailed descriptionand terminologyof the passefineskull is not available, decisionon the names to be used for many of the finer featuresof the skull wasdifficult. We followedas far as possiblethe general terminology being used in the recent literature without strict regard for priority of the terms or exact homologyof the featureswith thosein other vertebrategroups (see Bock, 1960b: 367 footnote). Some arbitrary deci- sionshad to be made, and in theseinstances we chosethose names allowing greatestclarity in understandingthe functional propertiesof the skull. Comparisonof the skulls will stressthe differencesobservable among the four taxa under consideration. These differences will be correlated with the jaw musculatureand observationson feeding behavior as the basis on which cranial adaptationsfor feedingcan be analyzed. Particularattention will be givento thosemorphological features involved in cranial kinesisin the descriptionof both the skull and the jaw muscles (see Bock, 1964, for a review of avian cranialkinesis). It is our opinion that the morphology,function and adaptationof the feeding apparatusin birdscan be understoodonly with respectto the propertyof cranialkinesis.

SKULL OF L. v. virens The generalappearance of the skull of virens (Pls. 3, 7, 8, and Figs. 4, 5, 6) is that of a typicalsmall passerine bird of insectivorousor omnivorous feedinghabits. It is not dissimilarto the skull expectedin a very thin- billed finch; the general appearanceof the sktdl in coccineais reminiscent of that in a small unspecializedfinch or a heavy-billedinsect-eater. No unique or specializedpasserine features are found in the skull of Loxops. The bonyrostrum (r) (upper jaw, u j, or maxilla,mx) is about80 per cent of the length of the braincase(b c) plus the orbit (o). Its longitudinal axis (along the bony tomium,b t) bendsdownward at an angle of almost 45ø relative to the longitudinalaxis of the braincase(along the jugal bar or the sphenoidalrostrum). The plane of the occipitalplate (o p) is angled about 60 ø above the longitudinal axis of the braincase. The braincase is large, about one-third the total length of the skull with its height slightly greater than its length. The length of the orbit 36 ORNITHOLOGICAL MONOGRAPHS NO. 15

sor stf f

n-f iow

ioc pop pt s eop --rap d

r (uj• mx)\ z t zp ep mf pt h ma (I j)

Figure 4. Skull of L. v. virens in oblique view labelled as a guide to identification of the osteological features in Plates 3-6. is about two-thirdsthe length of the braincasewith its height about equal to its length. In ventral view, the maximum braincasewidth is slightly less than half the skull length; the skull with its bill tapers graduallyand evenly from its maximum width acrossthe quadratesto the tip of the rostrum. The cranial featuresof Loxops can be grouped into the four major mechanicalunits (Gans, 1969)--braincase, upper jaw, bony palate (including quadrateand jugal bar) and mandible--that comprisethe avian kinetic skull (Bock, 1964: 4; and ms.) with a prokinetic naso[rontalhinge (n-f h) formedby a thin flexiblesheet of bone. This hingeis very broad, being over 80 per cent of the maximumwidth of the base of the upper jaw and about one-fourth of the maximum width of the skull. The braincase(b c), with its anterior extensioncomprised of the supraorbital rims (s o r; frontal region), interorbitalsepturn (i o s) and sphenoidalrostrum (s r), formsa singlerigid unit. The slightlywinged and laterally notchedectethmoid plate (e p) is large (PI. 8) and separatesthe orbit (o) almostcompletely from the antorbitalspace (a o s). Its ecteth- 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 37

op e sorjef ios ppsf n-f h, tf

aos• sf Ib• db ben p •ab r(u j l mx)--- om

ph mf rap ma (I j) a c p pt Figure 5. Skull of Loxops v. virens in lateral view labelled as a guide to identification of the osteological features in Plates 3-6. moid /oramen (e f) has a singleposterior opening and two (sometimes one) anterioropenings as a resultof divisionof the ectethmoidcanal. The broad externoventraledge of the ectethmoidlies on the dorsal edge of the jugal bar. No trace of the lacrimal bone (1) remains in the adult skull (although a small vestigemay exist but is lost during cleaning). The interorbitalwidth (i o w) acrossthe frontal region lying between the supraorbitalrims is narrow, being less than one-fourth the maximum width of the braincase. Its dorsal surface just posterior to the nasofrontal hinge is concave, but becomesflat toward the braincase roof. Most of the ventral half of the interorbital septumis ossifiedbut a large unossifiedwindow leading into the cranial interior exists on each side at the medioposterodorsalcorner of the orbit. The sphenoidalrostrum, forming the hollow, beam-likeventral edge of the interorbitalseptum, is swollen along its entire length with the anterior half of this swelling extending slightlyfurther dorsally than the posteriorhalf. The largeprotractor pterygoid scar (p p s) lying betweenthe optic /oramen (o f) and the sphenoidal rostrum reduces the height of the swelling in the posterior half of the sphenoidalrostrum. The pterygoidfoot and palatinehasp articulate against the anterior half of the swollen sphenoidalrostrum in a sliding joint. The braincaseis bulbousand quite smooth; muscle scars and the limits 38 ORNITHOLOGICAL MONOGRAPHS NO. 15

Ibp

)t f ppb• mpp btp tpp sr

imp pm op mp pimp b pb emp r p pt pt

Figure 6. Skull of L. v. virens in ventral view labelled as a guide to identification of osteological features in Plates 3-6. of fossaeare scarcelyevident. The postorbitalprocess (p o p) is a small nubbin,scarcely distinct from the rest of the postorbitalrim (p o r). The zygomaticprocess (z p) is the most distinctprocess on the lateral surface of the braincase,but it is only of moderatesize in comparisonwith that in many other passerinebirds. It is slightlyflattened anteroposteriorly and hasa sharplateral edge. A smallzygomatic tubercle (z t) is locatedposterior to the zygomaticprocess near the ventroposteriorcorner of the suprameatic fossa. A small, indistincttemporal fossa (t f) lies betweenthe postorbital and zygomaticprocesses, and a slightlylarger but still indistinctsubtemporal fossa(s f) lies betweenthe zygomaticprocess and the suprameaticprocesses (s p) plus its seeminglyanterior extension, the zygomatictubercle. One or two intraorbital cristae (i o c) are found on a bulbous pseudotemporal swelling(p t s) of the postorbitalwall just roedialto the zygomaticprocess; a depressionexists on the roedialside of the swelling,corresponding in generalshape to the orbital processof the quadrate. The auditorybulla (a b) (Dilger, 1956; Bock, 1960a: 38) is not swollen,but its anterior edgeextends far forwardso that the externalauditory meatus (e a m) is somewhatnarrowed. Only a narrow notch separatesthe exoccipitalprocess 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 39

(e o p) and lateral basitemporalprocess of the auditorybulla. Two low ridge-likecristae are located posteriorto, and convergeat the postero- ventralcorner of, the exoccipitalprocess; they boundthe peripheryof the areaof originof the m. depressormandibulae. The exoccipitalprocess forms the ventrolateralportion of the occipitalplate (o p); only its ventralmost tip may be distinctfrom the rest of the occipitalplate. The basitemporal plate (b t p) is slightlybulbous with a distinctlateral basitemporalprocess (1 bt p) extendingventrally from its lateral corners. The distal end of the lateral processof the basitemporalplate bears two small facets (P1. 7). An obvious,more ventral facet faces ventrolaterallyand a less obvious (hidden by the quadrate) more anterodorsalone facesanterolaterally. The former facet functionsas a base or prop for a tiny pneumatizedtubular bone (this structure may possibly be a cylinder of collagenousfibrous connectivetissue) extendingbetween the anteroventralcorner of the bony externalauditory meatus and the pneumaticforamen in the dorsalsurface of the internal mandibularprocess. The latter facet may abut againstthe medial edge of the lower half of the otic processof the quadrate and thus may function as a medial brace or guide for the quadrateas it rocks back and forth, but the exact relationshipsof thesebones are difficult to determine in dried skulls and must be checkedfurther before the bracing function of this processis definitely established. The rostrum, maxilla, or bony upper law, (r, mx, or u j) is slender and curvesgradually downward from its base to its tip. The cuttingedge of the bonytomium (b t) projectsslightly below the ventral surfaceof the rostrum, startingabruptly at the junctionbetween the jugal bar and rostrum. The bony externalnaris (b e n) is large, in length about one-thirdthe rostral length,oval, and almostcompletely unossified; neither the membranesin the external naris nor the medial nasal septurn (n s) are ossifiedexcept for a bit around the edgesof the septum. The lateral bar (1 b) lying between the external naris and the antorbital space is narrow but relatively stout as is the dorsalbar (d b) lying alongthe dorsaledge of the maxilla above the external naris. The jugal bar and prepalatinebar join the upper jaw at the same level as does the maxillopalatine,the base of which is hidden (when viewed ventrally but not dorsally) by the prepalatinebar. The palatine processo! the premaxilla (p p pm) is completelyfused with the prepalatinebar (P1. 7). A slight expansionof the base of the prepalatine bar existsand appearsas an early rudimentarystage in the evolution(or as a late vestigialstage in the loss) of the lateral [lange condition (1 f c) of the palatineprocess of the premaxilla(Bock, 1960b: 380-381). Only the anterior two-thirds of the rostrum floor are completelyossified; the area betweenthe basesof the prepalatinebars (floor of the nasal cavity) is unossified. 40 ORNITHOLOGICAL MONOGRAPHS NO. 15

Lying at the junctionbetween the braincase,bony palate and mandible, the quadrate (q) has a central role in all kinetic movements(P1. 8). Hence the exact detailsof its articulationswith surroundingbones are most important. All processesmay be consideredas extensionsfrom a central mass the body of the quadrate(b q). The otic processof the quadrate (ot p q) extendsposterodorsally to articulate with the braincaseat its ventrolateralcorner below the suprameaticprocess and anterior to the externalauditory meatus. The quadrate-squamosalarticulation is elongated with its longitudinalaxis inclinedanterolaterad at an angle of about 75 ø to the longitudinalaxis of the skull; the quadraterotates about this articulation as it rocks back and forth during kinetic cranial movement. This articulation is two-headed with a very shallow saddle between the lateral condyle of the otic process(1 cot p) and the medial condyle of the otic process(m cot p); no clear break appearsto exist betweenthe articulating surfacesof these condyles. The lateral condyle of the otic processlies anterodorsalto the medial condyleof this process. The suprameaticprocess lies posteriorto the lateralcondyle of the otic processand apparentlyassists in its support. A large, sharply defined fossa coversmost of the postero- ventral surface of the otic process. At the level of the ventral end of this fossa, the medial edge of the otic processmay rest against the more dorsalfacet at the tip of the lateral basitemporalprocess. If thesestructures do not touch in the living bird, they approacheach other very closely. The orbital processo[ the quadrate(ob p q) projectsanterodorsomedially from the medial surface of the quadrate body; it is long and slender with a taperedtip. A shortridge-like rugosity o[ theOrbital process o[ thequadrate (rob p) existson the dorsolateraledge of the distal tip of the process. A small tubercleo[ the orbital processo[ the quadrate (t ob p) lies on its dorsolateral surface about one-third of the distance from the base of the process.The distal half of the orbital processlies closeto the bulbous swellingof the postorbitalwall on which the intraorbitalcristae are found. An elongateddepression exists on the medial edge of the pseudotemporal swellingwhich correspondsapproximately to the orbital processin size and shape. This swellingmay act as a bony stop that limits dorsoposterior movementof the orbitalprocess; hence anterodorsal rotation of the quadrate and elevationof the upper jaw might be restricted. The groovemay be a phenotypicmodification during the life of an individual bird resulting from frequent,intermittent pressure of the orbital processof the quadrate on the postorbitalwall. The mandibularprocess of the quadrate(m p q) extendsventrally on the medial side of the bone, while the ]ugal processof the quadrate (j p q) extendslateroventrally from the lateral side of the bone. A distinct saddle separatesthese processes;this is most evident 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 41 in posteriorview. The mandibularprocess bears the large knob4ike medial condyleof the quadrate(m c q) on its ventral tip and the smallerpterygoid condyleof the quadrate (pt p q) on its medial surface. The lateral tip of the jugal processhas a cup-like socketinto which the jugal bar articulates and bearsthe lateral condyleof the quadrate (1 c q) on its ventral surface. The posteriorcondyle of the quadrate (p c q) extendsback from the lateral condyle along the posterioredge of the jugal processand ends in the posterioredge of the saddle betweenthe two ventral processes.The lateral condyle is considerablysmaller and lies further dorsally than the large knob-like medial condyle. The posteriorcondyle exists mainly as an extensionof the lateral condyle. The posteriorend of the jugal bar (j b), which is slightlycompressed laterally along most of its length, has a distinct lateral bend around the distal tip of the jugal processof the quadrate. Its posteriortip then bendsmedially at a right angle and articulatesinto the concavityon the lateral face of the jugal processof the quadratein a ball and socketjoint. The jugal bar is thin and relativelystraight along most of its length. At its junction with the posterolateralbase of the rostrum, it is slightlyflattened dorsoventrally. The pterygoid (pt) is a nearly straightbar of bone from the quadrate to the palatine and sphenoidalrostrum (Pls. 7, 8); it meets the sphenoidal rostrum at an angle of about 45 ø. The pterygoidhead (pt h) is expanded slightly and articulateswith the roundedpterygoid condyle of the quadrate. On the dorsalsurface of the pterygoidhead is the stout protractor pterygoid tubercle (p pt t). It is slightly curved so that its free tip points toward the origin of the m. protractor pterygoidei et quadraft at the protractor pterygoid scar. The anterior end of the pterygoid is expanded into a dorsolateralpterygoid foot (pt f) and a ventromedialretractor processof the pterygoid(r p pt); theseexpansions curve around the swollensphenoidal rostrum; the retractor processesof the paired pterygoidsmeet midventrally. Each retractor processhas on its posterior surface a small, low depression which serves as part of the origin of the m. pterygoideusretractor. The pterygoidfoot articulateswith the palatine hasp along a flexible suture line. The palatine (p) is, morphologicallyand functionally,a bony bar extending from the pterygoid to the posteroventralbase of the upper jaw, with several processesjutting out in different directions; the bony strut transmits force between the pterygoid and the rostrum, and the several processesserve as sites of muscle attachment and supports for closely associatedstructures (P1. 7). Its elongated dorsal extension forms half of the palatinehasp (p h). The right and left halvesof the palatinehasp embracethe sphenoidalrostrum, meeting midventrally beneath it, and articulate with it in a long slidinghemicylindrical diarthrosis. From the ventral 42 ORNITHOLOGICAL MONOGRAPHS NO. 15 edge of the palatine hasp, the palatine extendsventrally as the plate-like wall of the palatinetrough (w p t). The paired walls of the palatinetrough meet along the midline of the skull below the sphenoidalrostrum to form the hemicylindricalpalatine trough (p t) which opensventrally. The wall of the palatinetrough includesthe mediopalatineprocess posteriorly and the interpalatineprocess anteriorly. In cross-section,the palatine hasp and palatinetrough form a figure 8 with its baseand top cut off. The palatine trough containsthe posteriornasal canal (p n c) that extendsfrom the internal nares to open ventrally on the roof of the pharynx. The ventral edgesof the palatinetrough support the edgeof the openingof the posterior nasal canal into the pharynx. The posteriorhalf of the wall of the palatine trough forms the mediopalatineprocess (m p p) which extends beneath the retractor processof the pterygoid (i.e., partially covers the retractor processin ventralview). The free posteriortip of the mediopalatineprocess is frequentlybroken during preparation of the skeletonso that the process appearsshorter than its actual length. The interpalatineprocess (i p p) extendsforward from the anteroventral tip of the wall of the palatinetrough and pointstoward the maxillopalatine. The interpalatineprocess and maxillopalatineare connectedby a collagenous fibered band which supportsthe anterior end of the posteriornasal canal. The palatine blade (p b) extendsventrolaterally from the lateral surface of the palatine trough. As comparedto its degreeof developmentin other oscines(e.g., Corvus), it is relatively narrow anteroposteriorlybut stout. The elongated,thin transpalatineprocess (t p p) extendsback from the lateraledge of the palatineblade; the distaltip of this processis slightly bifurcated.The stoutprepalatine bar (p p b) connectsthe palatineblade to the base of the upper jaw. A narrow lateral flange existsat the anterior end of the prepalatinebar. Althoughthe palatineappears to be thin, fragile and ramified,it is essentiallya stout,straight bar of bone extendingfrom the pterygoid foot to the base of the rostrum. The bilateralpair of pterygoidfeet and the two halvesof the palatine hasp graspthe sphenoidalrostrum in a long slidingarticulation. The bones on the two sidesmeet or almostmeet on the midventralline, but are no- where fused together. The vomer (v) extendsanteriad from the antero- ventraledges of the bilateralhalves of the palatinehasp as a pair of thin bars that terminatein a broad triangularanteromedial plate (P1. 7). The maxillopalatine (m p) arises from the dorsal surface of the rostral floor at its junctionwith the jugal bar, and passesdorsally above the prepalatine bar. The pedicelof the maxillopalatine(pm p) extendsmedioposteriorly at an angle of about 45 ø to the longitudinalaxis of the skull. The terminal plate of the maxillopalatine(pl m p) is an elongated,narrow pneumatized 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 43 tube with laterallyflattened edges, and lies ventradto the medialplate of the vomer. The mandibleor lower jaw (ma, 1 j) is thin and decurved,following the curvatureof the upper jaw; the total angle of curvatureis at least 45 ø (P1. 3). Both rami are fused togetherat their anteriortip to form the mandibularsymphysis (m s) that is about one-fourth the total length of the mandible.A low, but distinctbony mandibularboss (m b) is presenton the dorsolateralrim of the ramusjust posterior to the mandibularsymphysis. This boss fits into the space medial to the posterior end of the bony tomial ridge of the upper jaw so that it abuts and pressesagainst the ventral end of the lateral bar of the upper jaw. This arrangementof the mandibularboss and the tomial ridge reinforcedby the lateral bar forms the bony supportfor what may be cutting and/or crushingrhampothecial surfacescorrelated with insect-eating. The mandibular foramen (m f) is small and lies directly below the low coronoid processes.It lies in the posteriorend of a large externaland a large internalmandibular fossa (e m f and i m f), each of which extendsforward to the level of the mandibular boss. The ventral edge of the internal mandibularfossa is boundedby the medial mandibular ledge (m m 1). A pair of low anteriorand posteriorcoronoid processes (a c p andp c p) extendfrom the dorsaledge of the ramus abovethe mandibularfossa (P1. 8). On the medialsurface of the ramusjust anteriorto the internalmandibular processlie the tubercleof the posteriormandibular adductor (t p m a) and the tubercleof the pseudotemporalis(t pt), both of which are small and indistinctin virens. The tubercleof the posteriormandibular adductor is a tiny roundedtubercle and in height is about twice the width of the anterior coronoidprocess, both measuredwhen viewingthe ramus directlyfrom above. It lies immediatelylateral and slightlydorsal to the anteriormargin of the groovein the dorsalsurface of the internalmandibular process, into which the medialcondyle of the quadratearticulates. The tubercleof the pseudotemporalisis of about the same size, but its anterodorsalsurface is flat. It lies slightlyanteroventral and also medialto the tubercleof the posterior mandibular adductor. The internalmandibular process (i m p) extendsdorsally more than mediallyand is thinwith a slightlyexpanded tip that is laterallycompressed and anteriorlybeaked. Each side of this tip containsa musclescar. A large pneumaticforamen opens on its lateral surfacejust distal to the articularsurface at the baseof the process.The area or shelfat the comer formed by the mandibularramus and internal processis the internomandibularflange (i m f); it is best seenin ventral aspect,and its ventral surface functions as the insertion for certain musculature. The dorsal 44 ORNITHOLOGICAL MONOGRAPHS NO. 15 surfaceof this flange has a deep elongatedarticular groove (a g) limited medially by the internal mandibularprocess and laterally by the mandibular ramus; this grooveforms the mandibulararticulating surface for the large roedial condyleof the quadrate (P1. 8). On the laterodorsal edge of the ramus oppositethe internal mandibular processis the elongated low external mandibularprocess (e m p). Its dorsal surfaceforms the articulatingsurface for the lateral condyleand the small posteriorcondyle of the quadrate. The ridge-like external mandibular processfits into the deep saddlebetween the roedialand lateral condylesof the quadrate. The quadrate-mandibulararticulation is, hence, one that would permit con- siderableanteroposterior sliding in addition to dorsoventralrotation of the mandiblebut only limited lateral rotation of the mandible. The mandible terminatesin a long retroarticularprocess (r a p) which appearsin dorsal view as a posteriorextension of the externalprocess. An obvious,knobby, dorsallyprojecting tubercle on the dorsaledge of its distal tip functionsas part of the insertionfor the m. depressormandibulae. A shallowdepression on the roedial surface of the retroarticularprocess is continuouswith the articulargroove for the roedialcondyle of the quadrate. Transverselyacross the dorsaledge of the retroarticularprocess and immediatelyposterior to the externalmandibular process is a tiny retroarticularnotch (r a n) into which the small posteriorcondyle of the quadrateslips when the mandible, either in its dosed or its depressedposition, is pushedforward. This notch may serveas a bony stop or supportfor the mandible when it is depressed. One or two sesamoidbones (s b) are found in the internal jugoman- dibular ligamentwhere it passesalong the posterioredge of the quadrate- mandibular articulation (see Fiedler, 1951:268-270 for a review of these sesamoids).Moreover, a shorttubular bone or cylinderof heavycollagenous fiber extendsdorsally from the pneumaticforamen in the internalmandibular process. It runs along the posteriorsurface of the otic processof the quadrate,lying betweenthe quadrateand the lateral basitemporalprocess. This tube is sometimes still attached to the mandible or to the base of the brain case.

CRANIAL KINESIS Basic to understandingthe form, function and adaptive significanceof the craniumand jaw musculatureof Loxops, a clear comprehensionof the mechanismof avian cranial kinesis,particularly of the passerineskull, is necessary(Fig. 7). This mechanismhas been describedamong others by Lakjer (1926: 96-101), Moller (1931: 138-147), Engels (1940: 364- 365), Beecher (1951a: 412-415; 1962) and has been reviewed recently by Bock (1964; and ms). We refer the reader to Bock (1964 and ms) for a discussionof the basic propertiesof cranial kinesis. 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 45

Cranial kinesis has adaptive significancein both dynamic and static conditions(Bock, 1966); in Loxops the basic significanceof kinesisapears to be associatedmainly with movementof the upper jaw. The number of degreesof arc throughwhich the tip of the maxilla movesfrom the retracted to the protracted position is referred to as the degree of kinetics of the skull. In Loxops, this degree of kinetics in a cleaned,water-soaked skull appears to be about 27 ø. Certainly such movement could be put to good use by all forms both during protractionand retractionof the rostrum while gapingor graspingwith the bill during feeding. In virens, kinesis is important while presumablygaping in campanulateblossoms, foliage, moundsof lichen or in the pulp of fruit. The creepernewtoni probably gapes in lichens and foliage to expose insects, while its relative mana usesits kinetic upper jaw presumablyto gape with the bill beneathsheets of bark or in lichensand to grasppieces of bark or large strugglinginsects. Similarly,coccinea may gape while wideningand cuttingholes in leaf buds or springingopen Koa pods, and may clamp down on the leafy material while removingwads cut from leaf buds. The higher degree of flexibility in the nasofrontalhinge (34 ø) and the relatively long bill of newtoni are perhapscorrelated with its almostpurely insectivorous diet and its huntingin foliagerather than on bark. It can be surmisedthat the majority of arthropods living in foliage dependupon rapid scramblingmovements or aerial flight by jumping, flying or suspensionby silk threads for escapefrom predatorsrather than hiding beneath tough sheetsof bark as do many of the insectsliving on trunks or large branchesof trees. To capture such nimble, foliage-inhabitingprey, a bird must be able to expose and seize them quickly before they can escape. To hunt in a densemaze of over- lapping, flexible and obscuringleaves, a bird would be better adapted if it could spreadthese leaves widely apart by gaping its bill than if it could spread them apart to only a lesser extent with little movement of the maxilla. The bird could thus see the prey more easily, and, moveover,its beak wouldbe widely openedand ready to grab the prey quickly. Further- more, movementof both jaws during openingthe bill would decreasecon- siderablyshifts in orientationof the line of sight with respect to the longitudinal axis of the bill between the two jaws (Moller, 1931: 146; Bock, 1964: 28) which is of particular value to a bird peering quickly betweenthe tips of its jaws for insects. The fact that the bill of newtoni is relatively longer than that of mana, for example, gives the former a further advantagefor spreadingleaves widely apart, etc., becausethe same degreeof kinetics for both a long-billed and a short-billedform will allow the tip of the rostrumduring protractionto travel througha greaterdistance in the former than in the latter. 46 ORNITHOLOGICAL MONOGRAPHS NO. 15

A •o 20

• •:•.•!:::::'

bill .,,/

x•s opened bill B

Figure 7. Schematic model of L. v. virens to show movements of the maxilla and the mandible during opening and closing of the bill. A) Solid lines indicate the position of the jaws in closedposition while the dashedlines and stippling showsthem in the fully opened position. The solid lines connectingthe heavy dots indicate the longitudinal axis of the quadrate, jugal bar and beak in the closedposition. The long dashed line indicates the longitudinal axis of the beak when the jaws are opened. The heavy dots indicate: (1) the squamosalarticulation of the quadrate; (2) the quadrate- mandibular articulation in the closed (2c) and opened (2o) positions; and (3) the posteroventralcomer of the maxilla in the closed (3c) and opened (3o) positions. B) Force vectors of the two muscles, m dm and m ppq, that serve to open the jaws; the heavy dotted line indicatesthe part of a force vector lying behind a bone. - 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 47

amp

Mpsp x M pt

p$ $

D (-...... ome

pt r

C) Force vectors of the three muscles, m a m p, m ps p and m pt, that serve to close the jaws without being able to retract the whole jaw apparatus relative to the brain case. D) Force vectors of the three muscles, marne; mpss and mptr, that serve to close the jaws and retract the whole jaw apparatusrelative to the brain case. 48 ORNITHOLOGICAL MONOGRAPHS NO. 15

GAPING ADAPTATIONS In additionto the importanceof cranial kinesisfor gaping,modifications of several other features of the head are probably correlatedwith this feedingmethod. One of these osteologicaladaptations is in the shapeof the ectethmoidplate. In Loxops and many other passefinebirds, the ectethmoidflares out laterally at its dorsaland ventral ends. Betweenthese dorsaland ventralalae is an in-cutsaddle or notchas illustratedby Beecher (1953: 275, fig. 1 ) who refersto this conditionas the "winged"ectethmoid plate. This ectethmoidnotch is well developedin starlings (Sturnidae) and meadowlarksand blackbirds(Icteridae) in which the jugal bars bend medially (are pinched in) at their junction with the base of the upper jaw. Although it has been reported that most birds have little or no capabilityof rotatingthe eyeball which fits tightly into its socket (Walls, 1940: 307), the European Starling, $turnus vulgaris,and meadowlarks, Sturnella spp., can turn their eyes forward while the birds are gaping amonggrass roots, soil, etc. That thesebirds look forward for their food items betweenthe upper and lower jaws as they are gaped apart is well describedby Lorenz (1949). Forward rotation of the eyes to bring the central foveae into positionfor forward binocularvision would be of great advantageto such gapinginsectivorous birds as has been demonstratedby Oehme (1962). Beecher (1951a: 423-424; 1953: 275) also describes this mechanism for meadowlarks and other birds. In Loxops the interectethmoidwidth measuredbetween the troughsof the notches is narrowest in newtoni. It is somewhat less narrow in mana and widest in virens and coccinea. These conditions fit the observations that manaand newtoni,especially the former,feed more on highlyactive arthropod prey which must be exposedquickly by gapingmovements and snatched-up before it has a chance to escape. The narrownessof the interectethmoid width would allow a greater uninhibited view of the prey through the gaped beak.

COMPARISONOF THE SKULL IN Loxops Certain peculiar differencesin shapesand relative sizes of parts of the skullsand jaws exist in each form which allow relatively easy differentiation betweenthem. The summaryof thesedifferences in Table 1 taken together with the figures of the skulls (Plates 3 to 9) will aid in comprehending thesedifferences which are more apparentin the actual specimens.Because the crania of these four forms are all about the same height (11 to 12 mm.), direct comparisonscan be made between these congeneric taxa without the need of compensatingfor any effectsof relative size. The maxillary, premaxillary, nasal, and dentary bones serve as the foundationsfor the rhamphothecae;in curvature and relative length the 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 49

TABLE 1 COMPARISON OF OSTEOLOGYOF SKULLS AND MANDIBLES: GROSS FEATURES

Structure or Feature L. v. virens L. m. mana L. m. newtoni L. c. coccinea

General Shape: 1. Mandible: a. Length long shorter long shortest b. Intercondylar width wide wide narrow wide c. Curvature along decurved straight slightly straight dorsal edges decurved 2. Rostrum: a. Length long shorter long shortest b. Curvature decurved straight slightly decurved decurved 3. Cranium: a. Bulbousness not bulbous bulbous not bulbous not bulbous b. Mean height 11.3 (25) 11.8 (2) 11.6 (2) 10.9 (1) (mm.p Relative Strength frail robust most frail most robust

Relative Sizes: 1. Interorbital width wide wide narrow widest 2. Interorbitalsepturn thin, or small large solid;no with small perforation perforation perforation performation 3. Nasofrontal hinge: 26.7ø (4) 25.0ø (2) 34.4ø (1) ca. 25 ø (1) kinetics •, • 4. Sphenoidalrostrum thick thick thin thickest thickness 5. Nasal arch: thickness wide wider narrow widest 6. Pterygoid:thickness thin thick thinnest thick 7. Jugal bar: thickness thin thin thin thinnest 8. Prepalatinebar: width narrowest wide narrow wide 9. Palatine hasp: ventral divided divided divided divided at surface posterior end only 10. Mandibular foramen: medium medium large small size to large 11. Mandibular ramus: thick thicker thin thicker thickness

Interectethmoidal Width: 3 widest narrow narrowest wide

:tNumber in parenthesesis sample size. a Arithmetic means of degrees of kinetics. a As seen from in front and measured from troughs of notches.

underlyingbones parallel the rhamphothecae,for which the functionsand adaptationshave already been discussed. The intercondylardistance between the mandibularrami in newtoniis definitelynarrower than that of the other forms,thus giving to thisrace a relativelynarrower and more acutelypointed bill. The cranial width of all forms exceptmana is about the same. The 50 ORNITHOLOGICAL MONOGRAPHS NO. 15 craniumof the latter, however,is noticeablymore bulbousthan in the rest; its functionalsignificance is not known. In some way this bulbousness may be correlatedwith obtaining prey by strenuousprying into bark crevices. An extremelybulbous cranium also occursin the psittirostrine, Hemignathuswilsoni, which poundsand prys in dead wood and bark in searchof its insectprey much as doesa woodpecker.It is interestingto note that Burt (1930: 471, 472, 474, 521) found that those woodpeckers which obtain most of their food by activelypecking in wood have relatively larger brain cavities than those which pluck their prey from surfacesof branchesor from other sites. He offersno explanationfor this phenomenon. The skull of coccineaappears the strongestin structureand that of newtonithe frailest. The skull of mana appearsslightly frailer than that of coccinea but sturdier than that of virens. These differences (see Table 1 ) are most evidentin the interorbitalseptum (checked in as many specimens as possible); degreeof kineticsof the upper jaw, and thus flexibility of the nasofrontalhinge; amount of divisionventrally betweenthe two halves of the palatine hasp; thicknessof the sphenoidalrostrum, nasal arch between the external nares, mandibular rami, pterygoidsand prepalatine bars; and size of mandibularforamen. Seeminglyincongruous is the fact that the jugalbars of coccineaare the smallestin diameterof the four forms. Thesemay not be relativelysmaller than the jugal bars in other forms of Loxops becausecoccinea is the smallestof the four forms tmder study, and its cranial height is also slightlylower than that of the others. It seemsreasonable that these aspectsof general sturdinessor frailty of the skull are of adaptationalsignificance in feedingmethods. Certainly, coccineaand mana wouldneed a skeletalstructure of strongerarchitecture for strenuousopening of closelyimbricated leaf buds or tightly closed bean podsand for pryingin crevicesof hard bark, than would virensand newtoni for obtaininginsects from foliage and lichens or nectar from blossoms. The skulls also exhibit differencesin sizesand shapesof sites of muscle attachment.These are not so obviousas the generaldifferences in shapes, relative sizes,degrees of sturdinessof skull architectureor flexibility of the skulls describedabove; however,they are real and discernibleand have functionalsignificance. These are seen as differencesin size and shapeof a numberof small bony processesand tubercleson the cranium,palate and jaws; as variationsin sizes,shapes, and depthsof musclescars and fossae; and as variations in sizes of surface areas for attachment of certain muscles. Many of thesedifferences are discerniblein Plates 3-9 showingthe skulls, and they are tabulated in Table 2. Becausetheir functional significances are closelycorrelated with the jaw muscles,they will be discussedin the followingsection on the anatomyand functionof thesemuscles. 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 51

TABLE 2 COMPARISON OF OSTEOLOGYOF SKULLS AND MANDIBLES: MINUTE FEATURES

Structure L.v. virens L. m. mana L. m. newtoni L. c. coccinea x intmand. flange: width narrow wide narrower; left: widest bulges right: wide ventrally pal. blade: lateral area• narrower broad narrowest left: broadest right: broad prepal.bar: width narrowest wide narrow left: widest right: wide proc. art. ext.: width wide8 wide narrow left: wide right: narrow proc. art. int.: length shortest shorter long shorter proc. corn. ant.: size low; thin low; thin low; thick low; thicker proc. corn. post.: size high; thin high; thicker low; thick low; thicker proc. intraorb.: size long; thin long, thicker shortest; medial: shorter; thin thick. lateral: abbreviated proc. orb. quad.: size long; thick; shorter; shortest; long; thick; narrow thick; thin; wide narrower narrow proc. postorb.: truncation truncated truncated sharply truncated pointed proc. postpal.: length long; short;wide longest; short;wide and lateral area narrow narrow proc. retroart.: length• long shorter shortest long proc. ptery. ret.: size and medium; large; most small; large; degreeof concavity concave concave convex convex proc. transpal.: length longest; shorter; long; left: shortest; and diameter slender thicker slender thickest right: shorter; thicker proc. zyg. dors.: size long; narrow long; wide shortest; left: shorter; thicker thicker narrow wide; thicker thin right: long; wider; thickest proc. zyg. vent.: size large sm•l small;thin smallest protr. pter. scar: size long; shallow long; deep short; longest; to deep shallow deepest rug. proc. orb. quad.: size small absent absent left: larger right: large tub. add. post.: size larger? large smaller smaller tub. proc. orb. quad.: size large small large smallest tub. protr. pref.: size long; shorter; long; thin longest; thinner thick thinner

Right-billed specimens. Not shown well in Plates 3-6. Not shown well in Plate 3. Longest and deep in L. sagittirostris. 52 ORNITHOLOGICAL MONOGRAPHS NO. 15

CRANIAL ASYMMETRY IN L. c. coccinea As would be expected,the asymmetricallycrossed rhamphotheceal tips of coccineaare reflectedfurther in bilateralasymmetry in its skull, but far fewer featuresare affectedthan one might anticipate. The bony maxilla and mandible are symmetricalas are the braincase(including ectethmoid plates,supraorbital rims and basitemporalplate), jugal bars and palate exceptfor small details. The only differencesin the palate appear to be a straighter,less laterally flaring transpalatineprocess (in a right-billedbird which will be describedherein) and a slightlylarger pterygoidfoot on the left side. The temporaland subtemporalfossae are largeron the right side of the braincasewith a sharplydefined dorsoposterior border and a larger zygomaticprocess separating them. These differencesare correlatedwith bilateral differencesin jaw musclesas describedbelow. The major cranial asymmetriesoccur in the quadrate-mandibularartic- ulations (P1. 9) which are associatedwith lateral rotation of the mandible as the bill tips slidelaterally in a slicingmovement. On the right side of the skull, the medial condyle of the quadrate is slightly larger so that the pterygoidcondyle appears to be shorter. The saddlebetween the mandibular and jugal processesof the quadrateis slightlydeeper with a sharplydefined ridge along its lateral edge. This ridge forms the medial edge of the small lateral condyleof the quadrate. A large articular surfacelines the lateral wall of the saddlebelow the lateral condyle; this articular facet extendsto the bottom of the saddleand backwardto a distinct,and large for Loxops, posteriorcondyle of the quadrate.On the correspondingside of the mandible, the internal processhas an extendedtip, and the internomandibularflange appearsa bit broader. Its articular grooveis broader and less distinctwith the articular facet restrictedto the center of the groove. Most notable is the very small (low) external processof the mandible, and the elongated articularsurface along the dorsaledge of the ramus (correspondingto the large articular surface in the saddle of the quadrate) that extends onto the retroarticularprocess and endsin a distinctridge. Becausethis articular surfacecovers the base of the retroarticularprocess, this processappears to be shorterthan that on the left side, but the two retroarticularprocesses are of equal length. On the left sideof the skull, the medialcondyle of the quadrateis shorter, the lateral condyle is broader and flatter and the posterior condyle is almostabsent. No distinctarticular surfaceis presenton the saddlebetween the jugal and mandibularprocesses. On the mandible,the externalprocess is large with an extensivearticular facet (for the large lateral condyleof the quadrate) coveringits entire dorsal surface. The articular groove is narrowerand better definedwith an articularfacet extendingalong its entire length. The retroarticularprocess has a more definite concavity on its 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 53 medialsurface. The posteriormandibular adductor tubercle and the pseudo- temporalistubercle are both larger on the left side of the mandible.

JAW MUSCULATURE

INTRODUCTION The currentlyaccepted system of identificationsand homologiesof avian jaw muscleswith those of other vertebrateshas its foundationsin the now classicstudy of Lakjer (1926) and subsequentworks of Hofer (1950), Fiedler (1951) and Starck (1959 and others). In this study we have followedthe systemof identificationand terminologyfor passefinejaw musclesand their parts advocatedby Bock (ms). This systemfollows that of Fiedler (1951) very closelyand hencestems from the earlierworks of Lakjer (1926), Moller (1930, 1931) and Hofer (1950). No attempt will be madeto summarizethe synonymiesof the namesused by these authorsand otherssuch as Beecher(1951a, 1951b, 1953) and Bowman (1961); the readeris referredto Bock's(ms) studyfor this information. The functionalproperties (Fig. 7) of the individualjaw muscles,sets of thesemuscles, the entire musculature and the jaw apparatuswere analyzed and correlatedwith the field observationsof feedingbehavior made by Richards.All conclusionsreached in this phaseof the studyare specu- lationsand mustbe treatedas hypothesesto be tested,not as demonstrated facts. Discussionsof the actionsof thesemuscles follow the generalanalysis of Bockand Morioka (ms) andinclude consideration of fiber arrangement, whetherparallel-fibered or pinnate (see Gans and Bock, 1965), of the position of the musclesrelative to the articulations in the skeletal level system,whether one-joint or two-joint (see Bock, 1968), and most importantly,of the consequencesof cranial kinesisin birds (reviewed by Bock,1964, 1966). Theseseveral factors underlying the functional properties of the jaw musclesare closelyinterrelated with one anotherand must be consideredsimultaneously as demonstratedfor the m. pseudotemporalis superficialisand m. pseudotemp.oralisprofundus (Bock, 1967; Bock and Shear, ms a). The innervationsof the jaw muscleswere not included in the dissections madeby Richardsas he foundthat he couldhomologize the musclesobserved in Loxopswith thosedescribed in otherpassefine birds by Lakjer,Moller, Fiedlerand mostof thoseillustrated by Beecher.Moreover, Bock (ms) also concludedthat the homologiesof the jaw musclesand their major parts in passefinebirds could be establishedwithout considering their innervations. The jaw ligamentswere examinedonly superficiallyduring the original dissectionand will be excludedfrom the description.However, the absence 54 ORNITHOLOGICAL MONOGRAPHS NO. 15 of the postorbitalligament as a functional structure (it may be present as an extremelythin vestige) was ascertainedspecially in dissectionsmade in the summer of 1963 by Richards and Bock. Loxops agrees with the carduelineand other finches in lacking this ligament and hence possesses uncoupledcranial kinesis (Bock, 1964). The followingdescription and comparisonof the jaw musclesis basedupon dissectionof the four taxa of Loxops under consideration.All statements referringto left or right sidesof coccineaare basedupon observationsmade on right-billedspecimens. The oppositeconfiguration would be found in left-billed specimens.The figures illustratingthe jaw musclesof each speciesare groupedtogether into compositeplates; thusthe musculatureof virens is illustrated on Plates 10 and 11, of mana on Plates 12 and 13, of newtoni on Plates 14 and 15, of coccineaon Plates 16, 17 and 18, and of sagittirostrison Plate 18. Referenceto thesefigures will not be repeated in the descriptionsof individualmuscles as this would be overly redundant and of little use to the reader becausethe plates illustrate the whole musculature, not individual muscles.

DESCRIPTION OF INDIVIDUAL MUSCLES

A) M. protractor pterygoidei et quadrati (m. protr. pter. quad., m p p q): The protractor of the bony palate is frequently divided into two distinct muscles,the m. protractor pterygoidei (sensu stricto) that inserts on the pterygoid and the m. protractor quadrati that inserts on the quadrate. However, separation of these parts is nebulous as we were never certain whether the cleft between these muscles made by an insertedprobe during dissectionwas a natural divisionor an artificial extension of the slight gap at the pterygoid-quadratearticulation in the insertionof this muscle. Becausethe two parts of the protractor musclewere recorded and describedseparately during dissection,they will be so discussedherein without any commitmentas to whetherthey constitutediscrete parts. The m. protr. pter. quad. is a blade-shapedmuscle lying along the ventral edge of the posterior wall of the orbit. Hence it is one of the deepest jaw muscles and can be observedclearly only when most of the other muscleshave been removed. The fibers are unipirmate and insert on a tendon running along the ventrolateral edge of the muscle. Those fibers inserting on the protractor tubercle of the pterygoid are designatedas the m. protr. pter. s. str. and those that pass posteriorly to insert on the quadrate as the m. protr. quad. 1) M. protractor pterygoideisensu stricto (m. protr. pter. s. str.; m p p): This is the largestportion of the protractor of the palate and the only part visible in a lateral or oblique view without removal of the overlying muscles. Origin: A long fleshy origin from the sides and base of the sphenoidalrostrum and from an oblong depression,the protractor pterygoid scar on the interorbital septurnanteroventral to the optic foramen and dorsal to the palatinehasp. In coccinea, also from the ventral rim of the optic foramen. Insertion: By meansof a heavy tendon running along the ventrolateraledge of the muscle to insert on the protractor tubercle of the pterygoid. Comparison: The m. protr. pter. quad. is a unipinnatemuscle with many relatively 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 55 short fibers that run ventrolaterally from their origin on the brain case to the tendon of insertion. This muscleis relativelydeep which beliesits appearanceas a small and consequentlyweak muscle when only its superficial aspect is examined. The fibers of the medial part of this muscle,those of the m. protr. pter. s. str., are shorter and insert onto the tendon of insertion at a smaller angle. The fibers in the lateral end of the muscle,the m. protr. quad., are longer and insert onto this tendon at a large angle if at all. Most of theselateral fibers attach directly to. the quadrate,onto the mandibular process,so that the m. protr. quad. is closer to a parallel-fibered musclethan a unipinnate muscle. This muscle is shorter, thicker and more truncated at its anterior end in newtoni than in the other forms of Loxops, but may still be stronger than that of virens. A faint line of demarcation was seen in ventral view between this muscle and the m. protr. quad. in newtoni, more distinct than in the other forms. The m. protr. pter. s. str. is larger and stronger in appearancein mana than in either virens or newtoni, and almost approaches the maximum size of this muscle seen in coccinea in which the tendon of insertion is extremely strong. Moreover, in coccinea, this muscle is larger by an estimated 5 to I0 per cent on the left side than on the right (not visible in the figures). The differences in size and shape of the protractor pterygoid scars and tubercles are summarized in Table 3. The tubercle is longest and thinnest in coccinea and shortest but thickest in mana, being intermediate in newtoni and virens. The scar is deepestin coccinea,shortest, widest and shallow in newtoni, and intermediatein mana and virens. Thus, great length but not necessarilythickness of the tubercle and greater length and depth of the scar seem to correlate with larger size of the muscle. Action: Bilateral contractionof this musclepulls (protracts) the pterygoidsforward and medially and rocks the quadrates forward and upward, pivoting them around their squamosalarticulations. Becausethe m. protr. pter. quad. inserts on the medio- posterior side of the pterygoid and quadrate, and because of the orientation of the axis of the quadrate-squamosalhinge with respect to the longitudinal axis of the skull, the head of the pterygoid rotates slightly medially around the pterygoid-palatine articulation as this bone slides forward. Forward motion of the pterygoid foot is transmittedto the palatine hasp and henceto the posteroventralbase of the rostrum via the prepalatine bars. The force of the m. protr. pter. quad. pushes on the postero- ventral base of the rostrum and hence rotates it upwards (elevates or protracts it) around the nasofrontal hinge. Anterior rotation of the quadrate also carries the mandible forward. The m. protr. pter. quad. is the prime protractor of the palate and with probable assistancefrom the m. depr. mand.: the amount of force contributedby the m. depr. mand. to protraction is difficult to estimate. 2) M. protractor quadrati (m. protr. quad.; m p q): A short muscle extending between the base of the sphenoidalrostrum and quadrate, that lies deep below other jaw muscles and is visible only when most other muscles are removed (as in the deep dissectionof the ventral view). Origin: Fleshy from the lateral part of the sphenoidal rostrum, posterior to the origin of the m. protr. pter. s. str., and from the ventrolateral surface of the braincase below the trigeminal foramen. Insertion: Fleshy to the medial surface of the base of the orbital processof the quadrate, but mainly to the mandibular processof the quadrate adjacent the pterygoid condyle. The main part of the insertion may be weakly tendinous. 56 ORNITHOLOGICAL MONOGRAPHS NO. 15

Comparison: The m. protr. quad. is mainly parallel-fiberedwith longer fibers than found in the m. protr. pter. s. str. No apparentdifferences could be detectedamong the forms dissected. Action: Seeabove under m. protr. pter. s. str. B) M. adductor mandibulae externus (m. add. mand. ext.): This large dorsal, superficial adductor of the mandible may be divided into three parts that are reasonably distinct from one another although some merging of fibers occurs between these parts. These subdivisionsof the m. add. mand. ext.--the rostralis, ventralis and caudalis•have somewhat different actions and are differentially developed in the several speciesof Loxops. 1) M. adductor mandibulae externus rostralis (m. add. mand. ext. rost.; m a m e r): The large m. add. mand. ext. rost. may be subdivided further into three smaller portions--the temporalis, lateralis and mediails. In spite of the somewhat greater degree of merging between these subunits, and the awkward terminology for these muscles, these parts can be recognized as discrete features, homologized in different passerinebirds, and exhibit differential development in various groups of perchingbirds. The tendonsof insertionof the three parts join just before or at their insertion on the mandible. a) M. adductor mandibulae externus rostralis temporalis (m. add. mand. ext. rost. temp.; m a m e r t): This portion of the m. add. mand. ext. rost. occupies a central position in this muscle, lying between the medialis and lateralis portions and dorsal to the mm. add. mand. ext. ventralis and caudalis. Origin: Fleshy from the whole of the temporal fossa dorsal to the zygomatic process and the m. add. mand. ext. caudalis, from the medioanterior surface of the zygomatic process,from the tip and orbital margin of the postorbital process,from tendons arising from the tips of these processes,and in mana and coccinea from the anterodorsal edge of the m. add. mand. ext. caudalis "b". Insertion: By a central tendon starting with the fibers originatingin the temporal fossa, passing deep beneath the postorbital process and running along the medial edge of the muscle to insert on the anterior coronoid process. Part of the tendon lies close to the lateral edge of the m. add. mand. ext. rost. medialis. The fibers of the m. add. mand. ext. rost. temp. insert on the dorsal, ventral and medial sides of the central tendon depending upon their site of origin. Comparison: Basically the m. add. mand. ext. rost. temp. is a bipinnate muscle with numerous short fibers; however, the somewhat bent courso of the muscle and complex origin of the fibers result in a more complicated appearance. It is a very strong muscle as indicated by the large number of fibers contained within it, but can displace only over a limited distance as indicated by its short fibers. In regard to relative size, the rank from largest to smallest goes from coccinea right side, left side, newtoni, mana, virens; the muscle on the right side of coccinea is by far the largest in the series. It is only slightly larger in mana than in virens. This muscle inserts on the mandible farthest forward in coccinea, thus attaching farthest away from the center of rotation of the mandible at its quadrate articulation which results in the longest moment arm of the muscle force vector in this species. The large force development of the m. add. mand. ext. rost. temp. and its long moment arm means that its torque development is the largest in this species. L. m. newtoni has the second farthest insertion and hence moment arm, and virens and mana are similar, having the shortest insertion of this muscle. The postorbital process is short and truncated in all forms studied except in newtoni in which it is sharply pointed; length of this process may be correlated 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 57 with the size of the m. add. mand.ext. rost. temp. but the relationshipis not clear. In all taxa, the anteriorcoronoid process is low in height. Action: Bilateralcontraction of the m. add. mand.ext. rost. temp. simultaneously adducts and retracts the mandible. Retraction of the mandible results in a backward rocking of the quadrateabout its squamosalarticulation and hence retractionof the palate and depressionof the rostrum. Hence this muscle can close the bill via movementof both the mandibleand rostrum,not only by raisingthe mandible. Unilateral contraction of the muscle would adduct and retract the mandible on that side more than on the oppositeside, and would still lower the upper jaw. Moreover,unilateral contraction would twist the mandible(rotate it laterally) toward that side of the head, therebybringing about a crossingof the tip of the mandible under the tip of the rostrum toward the side at which contractiontakes place. In the asymmetricallymuscled coccinea, bilateral contractionof the m. add. mand. ext. rost. temp. would result in an action similar to unilateral contraction in the other symmetricalforms of this genus,i.e., a crossingof the tip of the mandible beneath the rostrum toward the side of the strongeradductor. b) M. adductor mandibulae externus rostralis lateralis (m. add. mand. ext. rost. lat.; m a m e r 1): This portion of the m. add. mand. ext. rost. is the lateralmostpart of the whole jaw musculatureand often projectsbeyond the jugal bar which lies along the ventral edgeof the m. add. mand. ext. rost. lat. Although it lies on the lateral surfaceof the m. add. mand. ext. ventralisand is frequently included as part of that muscle, the fibers of the m. add. mand. ext. rost. lat. are sharply distinct from those of the ventralis. Origin: Tendinousalong the entire length of the lateral edge of the zygomatic processand from a tendon arisingfrom the distal tip of this process,from the lateral surface (its surface aponeurosis) of the m. add. mand. ext. vent., and from the anterior surface of the m. add. mand. ext. caud. "b" near the zygomafic process. Insertion: By a tendon to the laterai surface of the mandible anterior to the insertionof the m. add. mand. ext. rost. temp. and dorsal to the anterior tip of the insertion of the m. add. mand. ext. vent. The tendon of the m. add mand. ext. rost. lat. may join that of the m. add. mand. ext. rost. temp. just before their attachment onto the mandible. Comparison: Basically the m. add. mand. ext. rost. lat. is a unipinnate muscle with relatively short fibers. In newtoni, it tends towards a bipinnate condition. This part of the rostralis is decidedly weaker than the m. add. mand. ext. rost. temp. since the total number of contained muscle fibers is far smaller. In relative size from largest to smallest, and presumablywith larger size reflecting greater force, the order is coccinea, right side, left side, newtoni, virens, and mana. Further, this muscle inserts farthest forward on the mandible in coccinea, thereby having the longest moment arm which, in combination with the greatest strength, would produce the largest torque. L. m. newtoni has the next farthest insertion followed by virens and mana which are similar with the insertion closest to the articulation. The torques produced by this muscle in these forms would be pro- gressivelysmaller. The size and shape of the zygomatic process varies among the taxa of Loxops studied, giving further indication of the strength of some of the mandibular adductors attaching to it. It is sturdiest in build in coccinea (right side more so than the left side) and in mana, less sturdy in virens and highly abbreviatedin newtoni. Hence, the build of the zygomaticprocess does not correlate simply with the relative sizes of the muscle. 58 ORNITHOLOGICAL MONOGRAPHS NO. 15

Action: The action of the m. add. mand. ext. rost. lat. is similar to that described for the temporalisportion. c) M. adductor rnandibularis externus rostralis rnedialis (m. add. mand. ex. rost. med.; m a m e r m): This portion of the m. add. mand. ext. rost. is the medial-mostpart of the external adductor,lying just lateral to the m. pseudot. superf. and m. pseudot. prof. It can be seen only when the musclesare viewed obliquely through the orbit or when most of the lateral parts of the m. add. mand. ext. are removed. Origin: Fleshy and tendinousfrom the lateral edge of the posterior wall of the orbit and medial edge of the temporal fossa medial to the origins of the m. add. mand. ext. rost. temp. and m. add. mand. ext. vent. The origin of this muscle varies slightly in the taxa of Loxops as follows: i) virens and rnana: Fleshy and deep (medial) to the origin of the m. add. mand. ext. rost. temp., from the lateral edge of the posterior orbital wall, the medial (ventral) edge of the temporal fossa, the anterodorsalsurface of the zygomatic processand the roedial surface of the origin of the m. add. mand. ext. rost. temp., and tendinous from several tendons arising from these areas of origin especially on the medial side of the muscle. ii) newtoni: Same as in virens and, in addition from part of the anterior edge of the m. add. mand. ext. caud. "b" dorsal to the origin of the m. add. mand. ext. rost. lat. iii) coccinea: Same as in virens and in addition from the anterior edge of the fascia forming the edge of the m. add. mand. ext. vent. Insertion: Tendinous to the anterior coronoid processjust posterior to the tendon of insertionof the m. add. mand.ext. rost.temp. In rnana,also p•6sterior to the anterior coronoid process along the dorsal edge of the mandibular ramus in a laterally fleshy slip medial to the belly of the m. add. mand. ext. vent. and anterior to the insertion of the m. add. mand. ext. caud. "b". Comparison: The m. add. mand. ext. rost. med. is a unipinnate to bipinnate muscle with lateral and medial fibers inserting on a central tendon. The muscle contains a very large number of quite short fibers; hence it is a very strong muscle comparable in strength to the m. add. mand. rost. temp. although it appear smaller than this latter muscle. In relative size from largest to smallest, the order is coccinea right side, left side, newtoni, rnana and virens. The relative sizes and shapesof the zygomatic process and the anterior coronoid process have been discussed above under the mm. add. mand. ext. rost. temp. and lat. The relative sizesof the muscledo not correspond perfectly with the relative sizes of the zygomatic processas, for example, in newtoni. Action: The m. add. mand. ext. rost. med. has an action similar to that of the m. add. mand. ext. rost. temp. 2) M. adductor rnandibularisexternus ventralis (m. add. mand. ext. vent.; marne v): This large fan-shaped muscle is located in the anterolateral part of the cheek region and lies lateral and ventral to most of the m. add. mand. ext. rost. and anterior to the m. add. mand. ext. caud., but its dorsolateral surface is covered by the superficial m. add. mand. ext. rost. lat. The ventral half of the m. add. mand. ext. vent. spreads out to cover a large area of the lateral surface of the ramus around the mandibular fossa. Although most of this muscle is sharply distinct from surrounding parts of the m. add. mand. ext., some merging of fibers may occur with 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 59 the m. add. mand. ext. rost. lat. (anterior), m. add. mand. ext. rost. med. (a few, roedial) and m. add. mand. ext. rost. caud. (a few, posterior). Origin: The origin of this muscle varies somewhat in the taxa studied, as follows: i) virens: By a heavy flat tendon from the ventromedial edge and tip of the zygomatic process. ii) mana: By a heavy tendon from the anteromedial surface tip and distal part of the lateral edge (beneath the origin of the m. add. mand. ext. rost. lat.) of the zygomatic process. iii) newtoni: By a rather narrow flatfish tendon from the anteroventral tip of the zygomatic process. iv) coccinea: Strongly tendinous (stronger than in other forms) from the anteromedial surface and tip of the zygomatic process,and partially fleshy from the anterior margin of this process. Insertion: This, too, varies among the several forms studied, as follows: i) virens: In a fleshy manner, coveringthe mandibularforamen, the lateral surface of the mandible surrounding this foramen and on the dorsolateral edge of the mandible between the insertions of the m. add. mand. ext. rost. med. and the m. add. mand. ext. caud. "b". ii) mana and newtoni: Same as in virens, except not on the dorsolateral edge of the mandible. iii) coccinea: Same as in virens, but, in addition, the part originating from the anteromedial surface of the zygomatic process inserts between the two coronoid processes,between the insertions of the m. add. mand. ext. rost. med. and the m. add. mand. ext. caud. "b". Comparison: The fibers of this fan-shaped muscle are unipinnate, but relatively long so that the total number of fibers is small compared to the two other parts of the m. add. mand. ext. It is definitely the weakest portion of this external adductor. In relative size from the largest to the smallest, the ranking is newtoni, coccinea right side, left side, mana, virens. Relative size, however, may not give an accurate indication of strengthfor this muscle becauseits origin in newtoni seemsweakest even though its area of insertion appearslargest. The m. add. mand. ext. vent. is probably the strongestamong these forms on the right side of coccinea which is indicated by the size of the zygomatic processin this form. Action: The action of the m. add. mand ext. vent. is very similar to that of the m. add. mand. ext. rost. reed. 3) M. adductor mandibulae externus caudalis (m. add. mand. ext. caud.; m a m e c): This portion of the externaladductor is locatedin the rear part of the cheek region posteriorto the mm. add. mand. ext. rost. and vent. and anterior to the externalauditory meatusand the m. depressormandibulae. It can be divided into two parts, "a" and "b", which are not always easily separableso that the divisionof theseparts is purely arbitrary in some cases.Part "a" lies anterodorsally to part "b". This separationof the m. add. mand. ext. caud. into two parts is very reminiscent of that found in the cardueline finches (Fiedler, 1951). These two parts will be describedseparately. a) M. adductor mandibuIaris externus caudaiis "a" (m. add. mand. ext. caud. "a"; m a m e c a): This more ventral and posterior part of the m. add. mand. ext. caud. extends from the ventral part of the subtemporal area to the mandible. Origin: Fleshy from most of the ventral subtemporal fossa, suprameatic process and from the anterior surface of the otic process of the quadrate. Tendinous from the ventral zygomatic tubercle and suprameatic process. In newtoni also by a deeper 60 ORNITHOLOGICAL MONOGRAPHS NO. 15 slender head from the anterior part of the skull base medial and ventral to the ventral zygomatic tubercle. Insertion: On the dorsal rim of the mandibularramus posteriorto the insertion of the m. add. mand. ext. rost., but varies somewhatamong the forms studied as follows: i) virens: By a stout tendon on the posterior coronoid process posterior to the insertion of the m. add. man& ext. cand. "b". ii) maria: By a short, thick, partially fleshy tendon to the posterior coronoid process,posterior and lateral to the insertion of the m. add. mand. cand. "b". Also, by a small, fleshy slip over the lateral surface of the mandible across the posteriorcomer of the mandibularforamen and on the surfaceof the m. pseudot. prof. iii) newtoni: By two tendonsto the posterior coronoid processjust behind the insertion of the m. add. mand. caud. "b"; the deeper, slender head (mentioned above) inserts posterior to the larger one. iv) coccinea: Semitendinouslyon the posterior coronoid process immediatelyposterior to the insertion of the m. add. mand. post., and by a narrow fanlike slip which insertsdirectly over (lateral to) the mandibularforamen and that part of the m. pseudot.prof. adjacentmedially. The central tendon of m. add. mand. caud. "a" dissipatesitself in this slip. Comparison: The m. add. mand. ext. caud. "a" is a simple bipinnatemuscle with the fibers insertingon a central tendon. The fibers appear to be somewhatlonger than other pinnate dorsal adductorsof the mandible. In relative size from largest to. smallest,the ranking is virens, coccinearight side, maria, coccinealeft side, and newtoni. In the size of the ventral zygomafictubercle, virens is largest, rnana and newtoni are smaller, but of equal size, and coccineais the smallest. The posterior coronoid process of rnana is tall and thick; of virens, tall and thinnest; of coccinea, low and thick; and of newtoni, low and thinner. The variation of these processesmay reflect in part the relative sizes of this muscle in the four taxa. Action: Bilateral contraction of this muscle would adduct the mandible and also depressthe rostrum via retraction of the mandible, which rotates the quadrates posteriorly about their squamosal articulations and hence retracts the bony palate. Only the portion of the m. add. man& ext. caud. "a" that originatesfrom the braincase can retract the mandible; those fibers arising from the otic process of the quadrate are unable to retract the mandible. Unilateral contraction of this muscle would probably assist in swinging the tip of the mandible toward the side of contraction. In the asymmetrical coccinea, the action brought about by bilateral contraction of the m. add. mand. caud. would simulate the consequencesof unilateral contraction in the symmetrical forms. b) M. adductor mandibulae externus caudalis "b" (m. add. mand. ext. caud. "b"; mame c b): This more dorsal and anterior portion of the m. add. mand. ext. caud. extends from the dorsal part of the subtemporal fossa to the dorsal rim, being bounded anterodorsallyby subdivisionsof the m. add. mand. ext. rost., especially by the temp. portion of that muscle. The anterior fibers of this muscle may merge slightly with fibers of the m. add. mand. ext. rost. and perhaps with some of the m. add. mand. ext. rost. lat. Origin: Fleshy from the posteroventralsurface of the zygomaticprocess, from the dorsal part of the subtemporalfossa and side of the cranium to a point dorsal and posterior to the ventral zygomatic tubercle. The posterior end of the m. add. mand. 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 61 ext. caud. "b" at its origin is coveredby the anterior tip of the origin of the m. depr. mand. "b" in virens, mana and on the left side of coccinea, but not in newtoni. On the right side of coccinea,the anterior tips of the m. depr. mand. "a" and "b" are covered by the posterior tip of the m. add. mand. ext. caud. "b". Insertion: Tendinously (partly fleshy in newtoni) on the posterior coronoid process just anterior and slightly medial to the m. add. mand. ext. caud. "a" and posteriorto the fleshy slip of the m. add. mand. ext. vent. (virens and coccinea) or of the m. add. mand. ext. rost. med. (mana; no such slip exists in newtoni). In mana this insertionis by a short, flat tendon into a tiny dorsoventrallydirected groove on the lateral surfaceon the posterior coronoid process. Comparison: The m. add. mand. ext. caud. "b" is a bipinnate muscle (possibly only unipinnate in mana) with a large number of short fibers inserting onto a strong central tendon. The numerous short fibers make this a very strong muscle for its mass,similar to the m. add. mand. ext. rost. temp. or med., or to the m. pseudot.superf. In relative size from largest to smallest, the ranking is coccinea right side, left side, mana, newtoni,virens. Note the heavy triangle of fascia on the lateral surfaceof this muscleon the fight side of coccinea. The musclefibers bulge out above this fascia, as can be seen in Plate 16. The posteroventralsurface of the zygomaticprocess is largest in coccinea(right side), second in mana, third in virens and coccinea (left side), and smallest in newtoni. This series correspondsrather well with the relative sizes of the muscle in these taxa. The differences in details of the posterior coronoid process are described above under the m. add. mand. ext. caud. "a"; these size differences probably are related in part to differences in relative sizes of the muscle. Action: The action of part "b" of the m. add. mand. ext. caud. is similar to that of part "a", but it exerts a greater pull on the mandible because of its larger number of fibers and slightly longer moment arm. And because all parts of this muscle originate from the braincase, it retracts the mandible more than does the m. add. mand. ext. caud. "a". C) M. adductor mandibulae posterior (m. add. mand. post.; m a m p): This muscle is located deep in the posteroventralcheek region so that only a small part of the muscle at its insertion can be seen without removal of the overlying m. add. mand. ext. It extends between the quadrate and the mandible, lying posteriorly and medially to the m. add. mand. ext. and laterally to the m. pseudot. superf. Origin: Fleshy from the lateral surface of the proximal part of the orbital process of the quadrate posterior to the origin of the m. pseudot.prof., from the body of the quadrate, and tendinous from the tubercle of the orbital process of the quadrate. Insertion: Fleshy to the dorsal edge and the medial and lateral surfaces of the mandibular ramus, posterior to the posterior coronoid process and the insertion of the m. pseudot. superf., and anterior to the external and internal processesof the mandible. That part of the insertion on the medial surface of the mandible includes the tubercle of the posterior mandibular adductor. Comparison: The m. add. mand. post. is bipinnate in Loxops, an unusual condition in passerine birds but which may also exist in cardueline finches (Fiedler, 1951: 261). The larger number of fibers increases the force developed by the muscle. This muscle is symmetrical in coccinea. The m. add. mand. post. is largest in virens and mana, intermediate in size in newtoni and smallest in coccinea. The surface area of the base of the orbital process of the quadrate appears to be about the same size in all forms. The tubercle of the posterior mandibular adductor is slightly larger in virens and mana than in the other 62 ORNITHOLOGICAL MONOGRAPHS NO. 15 two forms, thus correspondingwith muscle size. The tubercle of the orbital process of the quadrate is largest in virens, and newtoni, smaller in mana and much smaller in coccinea,reflecting size differences in this muscle. Action: Bilateral contraction of this muscle adducts the mandible and depresses the upper jaw, but does not retract the mandible relative to the brain case. Its main function may be protection of the jaw articulation against strong lateral forces on the mandible. M. adductor mandibulae internus (m. add. mand. int.): The remaining mandibular adductorsare sometimesgrouped together as a single muscle under this name. We include mention of this term only for ease of referenceto other studies,but do not advocate its continued use becauseit is not practical in functional discussions(three or four quite distinctactions occur in the includedmuscles) nor is it certain that all of the included parts have a common evolutionaryorigin or are homologouswith the m. add. mand. int. of reptiles. D) M. pseudotemporalissuperficialis (m. pseudot. superf.; m ps s): The com- plexly pinnatem. pseudot.superf. is locatedin the posteroventralpart of the orbit, lying betweenthe m. pseudot.prof. and the mm. add. mand. ext. and post. If the overlying m. add. mand. ext. has not been removed, the m. pseudot.superf. can be seen only in an oblique view into the orbit. In lateral view after removal of the m. add. mand. ext., most of the m. pseudot. superf. is visible except for its ventral end which is covered by the m. add. mand. post. The anterior slip of this muscle is similar to that seen in the cardueline finches (Fietiler, 1951: 251-252; Beecher, 1953: 310-312, who illustrates and discussesthis anterior slip of the m. pseudot. superf.). Origin: Fleshy from the posterior wall of the orbit around the intraorbital cristae, and tendinous from these cristae. Insertion: By one or two laterally compressed slips---one in coccinea, one or two in virens, and two in mana and newtoni. The anterior slip is present in one of the four specimensof virens dissected.The posterior slip is always present in all forms and insertsmainly tendinouslyon the tubercle of the pseudotemporalison the medial surface of the mandibular ramus. The anterior slip inserts mainly in a fleshy manner on the medial side of the dorsal edge of the mandible from the anterior to the posterior coronoid process(in mana and newtoni) or just on the medial edge of the posterior coronoid process (virens). Furthermore, in newtoni a posterior part of the anterior slip inserts on a supernumerary tubercle of the pseudotemporalislocated immediately anterior to the main tubercle. Comparison: The m. pseudot. superf. is a tripartite muscle with each part either unipinnateor bipinnate. The individual muscle fibers are very short with the total number of fibers in the whole muscle being extremely large. Hence, this muscle developsgreat force but shortensvery little, if at all, during its contraction. In size, the range from largest to smallest is: coccinea, both sides, and newtoni are similar, followed by mana and then virens. The anterior slip is most obvious in newtoni, and in this form it inserts further anterior than in mana and virens. The m. pseudot. superf. is symmetrical in coccinea. The large size of this muscle in newtoni seems incongruous when considered with the fact that the intraorbital cristae are smallest in this taxon. These cristae are large and thick in mana and coccinea, although in the latter the lateral processis sometimesmuch abbreviated. In virens, they are longer and narrower. Although these cristae in these three forms relate well to the relative sizes of the m. pseudot. superf., other factors may influence the 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 63 development of these bony plates. The tubercle of the pseudotemporalisis about the same size in all forms; newtoni has an accessory one. Action: Bilateral contraction aids in adduction of the mandible. Because of the vertical orientation of the pull (force vector) of this muscle with respect to the mandible and quadrate,the m. pseudot.superr. may contributelittle to retraction of the mandible and quadrate. However, the anterior slip of this muscle may retract the mandible somewhat. E) M. pseudotemporalispro[undus (m. pseudot.prof.; m ps p): This bulky muscle is located in the posteroventralpart of the orbit medial to the m. add. mand. ext. and the m. pseudot.superf. and laterodorsalto the m. pter. dors. It is visible only in an oblique view into the orbit or when the other dorsal mandibular adductors are removed. The m. pseudot.prof. extendsfrom the orbital processof the quadrate to the medial surface of the mandible. Origin: Fleshy from the lateroventraland medioventralsurfaces of the distal three-fourthsof the orbital processof the quadrate. The distal-mostpart of the origin is from the rugosityof the orbital process.Some fibers arise from a tendon attaching to this rugosity. Insertion: Fleshy to the medial surface of the mandibular ramus, along the medial ledge of the mandibledorsal to the insertionof the m. pter. dors. lat., along the medial side of the dorsal edge of the ramus and from the dorsal and ventral margins of the mandibular foramen and surroundingbony surface. The insertion covers the mandibular foramen, and extends back to the anterior edge of the internomandibular flange. Comparison: The m. pseudot. prof. is a bulky, essentially parallel-fibered muscle with long fibers and consequentlyrelatively few fibers for the bulk of the muscle. The dorsal portion of the muscle is partly pinnate or fan-shaped with some of the fibers arising from a tendon attaching to the tip of the orbital process of the quadrate. This muscle is asymmetrical in coccinea, being slightly larger on the left side. In relative size, the rank order from largest to smallest is coccinea, left side, right side, rnana, virens, newtoni. It appearslongest in virens but is stronger in its appearance in coccinea and rnana. The muscle is shortest in newtoni. Its insertion extends relatively farther anterior in coccinea and virens. The orbital processof the quadrate, correspondingwell with the relative size and length of the muscle, is long, thick and wide in coccinea; thick and narrower in virens; shorter than that of virens in rnana; and shortest, thinner and narrowest in newtoni. The rugosity on the orbital process is largest on the left side of coccinea, slightly smaller on its right side, and small in virens. It is absent in rnana and newtoni. Its size appears to correlate somewhat with probable differencesin force except for its incongruousabsence in mana. Action: Bilateral contraction of the m. pseudot. prof. elevates the mandible and lowers the rostrum by pulling the orbital process of the quadrate ventral (i.e., rotating the quadrate ventrally about its mandibular hinge) with the force trans- mitted to the base of the upper jaw through the bony palate and jugal bars. The m. pseudot.prof. probablyalso aids in keepingthe bill closedin the restingposition. Its relatively low force development is compensatedfor by its long moment arm. F) M. pterygoideus(m. pter.; m pt): The m. pter. constitutesthe ventral adductor of the mandible and depressor of the rostrum. It lies more or less in a frontal plane, arising from the palatine and pterygoid and inserts on the mandible except for a small portion that inserts on the basitemporal plate. The m. pter. is a large muscle, either the largest or the second largest following the m. add. mand. ext., and is 64 ORNITHOLOGICAL MONOGRAPHS NO. 15 divided into five major parts--the ventralis lateralis, ventralis medialis, dorsalis lateralis, dorsalis medialis and retractor. The m. pterygoideusdorsalis medialis is divided further into an anterior and a posteriorpart. A few small subdivisionsof the m. pterygoideusventralis medialis can be recognized. Some of these portions of the m. pterygoideusare sharply separatedfrom other parts and readily recognized,while the separation between other subunits such as between the m. pterygoideus ventralis lateralis and medialisare difficult to ascertainwith certainty. An initial division of the muscle into a dorsal and a ventral portion is not justified (see Bock, ms; Morloka, ms). Identification of the parts of the m. pter. has been badly confused in the earlier literature on passerine jaw muscle. We follow the scheme advocated by Bock (ms) which is very close to that used by Richards in his thesis. The whole m. pterygoideus,as well as its constituent parts, is asymmetrical in coccinea, being larger on the left side. It should be noted that the ventral adductor-- the m. pterygoideus---islargest on the left side of the head in a right-billed individual of coccineawhile the asymmetricaldorsal adductorsare stronger on the right side of the head. 1) M. pterygoideusventralis lateralis (m. pter. vent. lat.; m pt v 1): The lateral and ventral part of the m. pter. that arisesfrom the transpalatineprocess and prepalatine bar and extends to the mandible. It cannot be seen in a dorsal oblique view into the orbit unless the m. pter. dor. lat. is removed, but is fully visible in a ventral view of the jaw musculature. Origin: Partly fleshy, but mainly tendinous from the ventral and lateral surfaces of the transpalatine process and the prepalatine bar. It is closely pressed to the ventral surface of the m. pter. dor. lat., and indeed a common aponeurosisof origin separates these two muscles for some distance behind the distal tip of the transpalatine process. In newtoni the origin is weakly tendinousonly along the distal half of the transpalatine process. The same is true in coccinea on the left side but with some fleshy insertion; its right side is strongly tendinous. Insertion: Fleshy to the ventrolateral, and to a slight extent the ventromedialsurface and ventral edge of the mandible from a point approximatelybelow the anterior edge of the mandibular foramen almost to the base of the retroarticular process; those fibers inserting on the lateral surface of the mandible curve around its ventral edge and may be seenin a lateral view of the head. Also fleshy to the anterior part of the insertion of the m. depr. mand. "a" located on the medial side of the retroarticular process and from the ventral surface of the internomandibular flange. Tendinous on a sheet of fascia comprising the ventrolateral edge of the m. pter. vent. med., and tendinous on a bony tubercle near the posteromedial corner of the confluence of the internal mandibular and retroarticularprocess. The following variation in the insertion has been noted: i) newtoni: Fleshy to the ventromedial surface of the mandible from a point approximately below the midpoint of the mandibular foramen to the anteflor side of the base of the internal mandibular process. Also fleshy to the ventral surface of the proximal two thirds of the internal mandibular process and to the internomandibular flange. ii) coccinea: The insertion on the left side is the same as in the general description and, in addition, fleshy from most of the ventral edge of the retroarticular process. Also, on the ventrolateral edge of the mandible, fleshy to as far forward as two lengths of the mandibular foramen beyond the anterior end of this foramen. Moreover, about 75 per cent of the anteflor end of the insertion 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 65 of the m. depr. mand. "a" medial to the retroarticular process is overlapped by that of the m. pter. vent. lat. The insertion on the right side is the same as described for the left side, but less of the ventral edge of the retroarticular process is covered by the muscle and the insertion on the ventrolateral edge of the mandible does not extend so far forward. Less of the insertion of the m. depr. mand. "a" is overlapped. The tendinous part of the insertion on the tubercle at the confluence of the internal mandibular and retroarticular processesis weaker than on the left side. Comparison: Although the m. pter. vent. lat. appears as a parallel fibered muscle, the fibers are probably unipinnate throughout most of the length of the muscle. They are relatively long, however, compared to other pinnate jaw muscles. This part is one of the strongest segments, second only to the m. pter. dors. lat., of the m. pterygoideusand contributesmost to closing the bill. The demarcation between this muscle and the m. pter. vent. reed. is difficult to fix with certainty, and it is not absolutelydefinite whether the division used in the dissectionsand descriptionsof the musculature in Loxops correspondsexactly to those advocated for Corvus (Bock, ms). We have attempted to be consistentwithin Loxops even if our decision as to what constitutes the division between these muscle parts may be partly arbitrary. In relative size, the ranking from largest to smallest is coccinea, left side, right side, rnana, virens, newtoni. Of this series, the muscle on the left side of coccinea is by far the largest, being an estimated 85 to 100 per cent larger than even the second largest in the series, which is the same muscle on the right side of this species. The greater area of overlap in coccinea onto the lateral surface of the mandible, the retroarticular process and the insertion of the m. depr. mand. "a" should be noted. This muscle in newtoni is probably by far the weakest, as it does not overlap even onto the ventral surface of the mandible. The ventral surface of the internomandibular flange is flat or concave and widest in coccinea and rnana, of intermediate area in virens and narrowest and even convex in newtoni. The flange on the left side of coccinea perhaps has a greater surface area and is more concave and rugose than that on the right. The size and shape of this flange correspondswell with the relative sizes of the muscle. The transpalatine process is shortest and thickest on the left side of coccinea, a little longer and slenderer on its right side, slightly longer and slenderer in rnana, and longest and slenderest in newtoni and especially so in virens. A short, thick transpalatine process is a more efficient point of origin for the stronger muscles in this series. The prepalatine bar is widest in coccinea and rnana, slightly narrower in newtoni and narrowest in virens. It is widest of all on the left side of coccinea. This situation in general reflects the relative sizes of this muscle in the four taxa as well as the relative sizesand forces of the setsof musclesacting through the palate to elevate and depress the rostrum. Action: The main function of this muscle is retraction of the palate, relative to the mandible (not necessarilyto the braincase), and hence depressionof the upper jaw. Unilateral contraction of this muscle aids in crossing of the mandible under the rostrum in a direction away from the side on which contraction takes place (toward the contralateral side). This probably takes place to a slight degree in the symmetricalforms of Loxops. Crossingof the tips of the bill is possiblebecause the roedial condyle of the quadrate can experiencea certain amount of fore-and-aft movement and a slight amount of pivoting in the groove into which it fits on the dorsal surface of the internomandibular flange. In coccinea lateral rotation of the 66 ORNITHOLOGICAL MONOGRAPHS NO. 15 mandible and crossing of the bill tips (by way of bilateral contraction) is the usual rather than the unusual action. The quadrate condyles, mandibular grooves and other articulations are so modified that during bilateral contraction of the asymmetrical m. pterygoideus,the stronger muscleson the left side cause the whole mandible to swing toward the right, pivoting on the articulation between the mediai condyle of the quadrate and the corresponding mandibular groove on the left side of the lower jaw. Thus the tip of the mandible crosses toward the right and the right ramus "backs up" as it were, its internomandibularflange sliding beneath the medial condyle of the right quadrate. 2) M. pterygoideusventralis medialis (m. pter. vent. med.; m pt v m): The medioventralpart of the m. pterygoideusconsists usually of two and sometimesthree (virens) parts: a main belly adjacent to the m. pter. vent. lat. and lateral to the palatine salivary gland; a more medial part located dorsal to the posterior end of the palatine salivary gland and termed the m. pter. vent. med. "fan" because of its resemblanceto a fan; and, in some specimensof virens, an anteromedialpart termed the m. pter. vent. med. "scapularis" because of its resemblanceto a mammalian scapula. The m. pter. vent. med. "fan" probably is homologouswith a small part of the m. pter. vent. med. described but not named by Fiedler (1951: 251, 255), and the superficial fibers of this muscle described by Bock (ms) which attach on the fascia of the m. pter. retr. and on the connectivetissue around the openingsof the eustachiantubes. The m. pter. vent. med. '"scap." may be homologouswith the muscle part described by Moller (1930: 688 1931: 128) as the m. pter. vent. internus. This part of the m. pter. runs from the transpalatine and medinpalatine processes of the palatine to the mandible; the divisions of it are separated only in virens. Only a small part of the medial portion is visible in an oblique view through the orbit in virens; in this and the other taxa, it can be seenfully only in a ventral view of the musculature. Origin: The origins of the several subdivisionsof the m. pter. vent. med. will be describedseparately, as follows: Main belly: Partly tendinous from the ventromedial edge of the transpalatine process,anteriorly to a point adjacentto the prepalatineblade, from the ventromedial surfaceof the originatingend of the m. pter. dors. lat. and from the lateral surface and posteriorend of the palatine salivary gland. In newtoni, the thinned-outmedial part of the main belly originatesweakly from the lateral surfaceof the anterior base of the mediopalatine process. "Fan"-like part: Fleshy from connectivetissue around the oral opening of the eustachian tubes, from the ventral surface of the posterior end of the m. pter. retr., and from the dorsal surface and posterior end of the palatine salivary gland. "Scapular"part of virens: By three small fleshy slips, togetherin a group, on the lateral surface of the posterior end of the palatine hasp immediately anterior to the pterygoidfoot and origin of the m. pter. dors. med. ant., and dorsal to the palatine salivary gland. Visible through the orbital in an oblique view and barely in a lateral view. Insertion: The insertion of the three subdivisionswill be described separately, as follows: Main belly: Fleshy to the ventral surface of the distal (medial) quarter to two- thirds of the internal mandibular processand to the dorsomedial surface of the m. pter. vent. lat. "Fan"-like part: By a tiny cordlike tendon to the medial tip of the internal man- 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 67

dibular processin virens. In other forms by such a tendon and also to a thinned-out medial extenson of the main belly of the m. pter. vent. med. "Scapular" part of virens: Fleshy to the ventral surface of the m. pter. dors. med. ant. and the dorsal surface of the palatine salivary gland. (Note: Plate 11A is erroneous in that it shows the "scapular" part ventral to the anterolateral end of the m. pter. retr. and dorsal to the m. pter. dors. med. ant.; it should be ventral to the m. pter. dors. reed. ant.) Comparison: The main belly of the m. pter. vent. med. is probably unipinnate throughout most of its length, with relatively long fibers. Its size is comparable to that of the m. pter. vent. lat. in mana, newtoni, and coccinea, but is much smaller in virens. The main belly of this muscle is largest on the left side of coccinea where it is an estimated 50 per cent larger than on the right side. The right main belly of coccineaand main belly of tnana are about equal and rank secondin size. The muscle is smaller in virens and smallest in newtoni. The relative sizes and shapes of the transpalatineprocess, which is short and thick in coccineaand tnana and long and slenderin the other two forms, may reflect the relative strengthof the main belly of the m. pter. vent. med. The fan-like part is about the same size throughout,being a relatively thin muscle. In all forms, except virens, it blends laterally with the thin medial portion of the main belly of the m. pter. vent. reed. This is especiallytrue in newtoni. Action: The action of the main belly of the m. pter. vent. med. is very similar to that describedabove for the m. pter. vent. lat. Contractionof the m. pter. vent. med. "fan" may act to openthe commonaperture of the eustachiantubes in the roof of the pharynxby pulling posterolateralon the surroundingconnective tissue. Thus it may functionto equalizeair pressureson both sidesof the ear drum duringchanges in altitudein flight. Relaxationof the fan-like part would allow this orifice to close since its lips do not normally appear to hold it open. The automatic closure of this opening would be advantageousduring feeding in that it would preventfood particlesor fluids from enteringthe eustachiantubes. 3) M. pterygoideusdorsalis lateralis (m. pter. dors.lat.; m pt d 1): This part of the m. pterygoideusconsists of the large dorsolateralportion running from the transpalatineprocess and palatine blade to the medial surface of the mandibular ramus. No fibers of this part attachesto the pterygoid; these fibers are all part of the m. pter. dots.reed. The m. pter. dots.lat. can be seen,in part, in an oblique view throughthe orbit, or ventrally after removalof the ventral portionsof the m. pter. Origin: Slightlydifferent in the four taxa studied,as follows: i) virens: Fleshy from the dorsal surface of the transpalatineprocess and palatineblade, from the aponeurosison the dorsolateralsurface of the originating end of the m. pter. vent. lat., and of the dorsomedialsurface of the originatingend of the m. pter. vent. med. (Note: These three origins are separableby careful probing. See note under the descriptionof the origin of the m. pter. vent. lat.) ii) tnana: Fleshy from the dorsal surfaceof the transpalatineprocess and the palatine blade, from the aponeurosison the dorsolateralsurface of the originating end of the m. pter. vent. lat. and from the side of the palatine betweenthe mediopalatineprocess and palatine blade, dorsal to the ventral head of origin of the m. pter. retr. iii) newtoni: Fleshy from the dorsal surface of the transpalatine processand palatine blade, from the ventral surfaceof the palatine blade and from 68 ORNITHOLOGICAL MONOGRAPHS NO. 15

the side of the palatine between the mediopalatine process and the palatine blade, ventral to the anteriormost part of the origin of the m. pter. dots. med. post. iv) coccinea: On the left side, fleshy from the dorsal surface of the transpalatineprocess and palatine blade, from the ventral surface of the palatine blade to the dorsolateral and dorsomedial surfacesof the aponeurosisof the originating end of the m. pter. vent. lat., from much of the dorsal surface of the prepalatine bar and from the mediopalatine process anterior to the ventral head of origin of the m. pter. retr. On the right side the origin is less extensive than on the left; it does not extend as far medially nor posteriorly on the ventral surface of the palatine blade, and not as far anteriorly on the prepalatine bar. Insertion: As for the origin, the insertion of the m. pter. dors. lat. varies slightly in the four taxa of Loxops, as follows: i) virens: Fleshy to the roedial surface of the mandibular ramus, dorsal to the insertion of the m. pter. vent. lat. and ventral to the insertion of the m. pseudot.prof., along the roedial ledge of the mandible anteriorly from a point oppositethe anterior margin of the mandibular foramen and posteriorly to the anterolateral corner of the internomandibularflange immediately in front of the insertion of the m. pter. dots. reed. ant. Tendinous along the roedial mandibular ledge by an aponeurosis on the ventral surface of this muscle. ii) mana: Same as in virens and, in addition, tendinouslyby a tendon comprisingthe posteromedialborder of this muscle and attaching to a bony tubercle dorsal to the anterolateral corner of the internomandibular flange. iii) newtoni: Thinly tendinous to the anterior edge and dorsal surface of the internomandibular flange. iv) coccinea: On the left side, the same as in virens and, in addition, strongly fleshy from the dorsal corner of the anterior end of the internomandibular flange. Also strongly tendinous on a tendon comprising the posteromedial edge of this muscle and attaching onto the anterior edge of the internomandibular flange. On the right side, the insertion is about the same as found on the left side but not extended as far anteriorly. Comparison: The m. pter. dors. lat. is probably unipinnatethroughout most of its length, with relatively long fibers. Its fibers run ventrally-dorsallyfrom origin to in- sertionwhile those of the m. pter. vent. lat. run dorsally-ventrally. Together, the fibers of the two musclessimulate a bipinnate muscle. In size, it is equal to or usually con- siderably larger than the m. pter. vent. lat. and is usually the largest and presumably strongestpart of the m. pterygoideus. It contributesgreatly to closing the bill. This muscle part is asymmetrical in coccinea, being largest on the left side. In relative size from largest to smallest, the rank order is coccinea left side, right side, mana, newtoni, virens. In virens it appears as though the roedial part of the origin of the m. pter. dots. lat. has been crowded out by the palatine salivary gland and has left no vestige. It should be noted that the posteromedial edge of the m. pter. dots. lat. and the anterolateral edge of the m. pter. dors. reed. ant. do not meet along a common boundary in virens and newtoni but are separatedby a wide gap. This is a common condition in many thin-billed passerine birds and shows clearly the separation between the m. pter. dots. lat. and m. pter. dots. med. (Fiediet, 1951:259; Bock, ms). In mana and coccinea, the edges of these muscles come into close contact with one another giving the impression of a single bipinnate muscle. However, the tendon on the postero- roedial edge of the m. pter. dors. lat. and the tendon on the anterolateraledge of the 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 69 m. pter. dors. med. ant. are always clearly separated from each other and a probe can be inserted between them without creating an artificial tear. This is the common condition in heavy-billed birds such as finches (Fielder, 1951) and crows (Bock, ms). In coccinea, the m. pter. dors. lat. is larger on the left than on the right. This is obvious not only in the differences in the extent of the origins and insertions, but also in the width of the belly of the muscle and in the size of its originating end. On the left side, the origin on the dorsal surface of the palatine hasp is robust; on the right side, it is band-like. On the ventral surface of the palatine blade, and on the side of the palatine proximal to the mediopalatine process, and medial to the transpalatine process,the origin on the left side is a bulging ridge of muscle; on the right, the muscle is flat at its origin. The palatine blade is broadest on the left side of coccinea, slightly less so on the right, a little lessbroad in mana than on the right side of coccinea,narrower than this in virens, and narrowest in newtoni. These differences are more noticeable in dorsal view than in ventral. This situation parallels rather well the relative size of the m. pter. dors. lat. Action: The action of the m. pter. dors. lat. is very similar in general to that described above for the m. pter. vent. lat. This muscle is a very strong depressorof the upper jaw becauseof its large size and contained number of fibers. Indeed, most of the force for retracting the palate relative to the mandible and hence depressing the rostrumcomes from the contractionof thesetwo lateral parts of the m. pterygoideus. 4) M. pterygoideusdorsalis medialis (m. pter. dors. med.; m pt d m): This dorsomedialpart of the m. pterygoideusis divided into two quite distinct divisions,one originating from the anterolateral surface of the pterygoid and the other from the posteromedialsurface and pterygoid head. They will be distinguishedas the anterior (m. pter. dors. med. ant.; m ptd m a) and the posterior (m. pter. dors. med. post.; m ptdmp) divisions of this muscle part, but will be described together under the same heading. The m. pter. dors. med. lies in the medioventral comer of the orbit, running from the pterygoidto the internal mandibularprocess. It can be seen partly in an oblique view into the orbit and ventrally after removal of the underlyingsuper- ficial layer of ventral pterygoid muscles. Origin: The origin of these parts of the m. pter. dors. med. differ slightly in the four taxa, as follows: i) virens: The m. pter. dors. med. ant. has a fleshy origin along the anterolateralsurface of the anterior two-thirds of the pterygoid up to the joint between the pterygoidfoot and palatine hasp. The origin of the m. pter. dors. med. post. is fleshy from the posterior third of the anterolateral surface of the pterygoid, around the ventral surface of the pterygoid head, from the anterior half of the ventral surface and the whole of the medial surface of the pterygoid, and from the dorsal part of the pterygoid foot betweenthe origins of the m. pter. retr. (dorsally) and the m. pter. dors. med. ant. (ventrally). The anterior part of the origin is not forked about the pterygoid at the pterygoid foot, but the muscle passesdorsal to this bone and into the orbit to be "squeezed"between the originating heads of the m. pter. retr. and m. pter. dors. med. ant. ii) mana: The fleshy origin of the m. pter. dors. med. ant. covers the anterolateral surface of the anterior two-thirds of the pterygoid, extending from the anteriormost origin of the m. pter. dors. med. post. to a small half moon-shaped depressionat the base of the mediopalatineprocess immediately ventral to the pterygoid foot. The m. pter. dors. med. post. has a fleshy origin from the posterior third of the anterolateral surface of the pterygoid, around the pterygoid head, and from the 70 ORNITHOLOGICAL MONOGRAPHS NO. 15 whole of the dorsal and medial surfacesof the pterygoid. The anterior end of the muscle is forked around (dorsal and ventral) the pterygoid foot. The ventral head originatesfrom a small depressionat the base of the mediopalatineprocess, deep to and hidden by the origin of the m. pter. dors. med. ant. The dorsal head originates from the pterygoidfoot immediatelyventral to the dorsal head of the m. pter. retr. and dorsal to the originating head of the m. pter. dors. med. ant. iii) newtoni: The fleshy origin of the m. pter. dors. med. ant. extends from the anteriorthree-fourths of the anterolateralsurface of the pterygoidanterior to the tip of the origin of the m. pter. dors. med. post. The anteriorend is compressed betweenthe two halvesof the m. pter. dors. med. post. as the latter forks aroundthe pterygoid. It is not visiblethrough the orbit in lateral view. The m. pter. dors. med. post. has a fleshy origin from the posterior fourth of the anterolateral surface of the pterygoid,around the head of the pterygoid,and from the ventral and posteromedial surfaceof the pterygoid.The anteriorend of the muscleis forkedaround the pterygoid with the ventralhead originating from the lateral surfaceof the mediopalatineprocess anterodorsalto the origin of the m. pter. retr. and ventral to the pterygoidfoot and the head of the dorsalfork. The dorsalhead originatesfrom the whole surfaceof the pterygoidfoot ventral to the origin of the m. protr. pter. quad. and dorsal to the origin of the ventral head of the fork. Both headsare visible in side view through the orbit. iv) coccinea: The origin of the m. pter. dors. med. ant. is a fleshy attachment from the anterolateral surface of the anterior two-thirds of the pterygoid anterior to the tip of the m. pter. dors. med. post., lateral side of the mediopalatine processand ventral part of the palatine hasp just ventral to the origin of the m. pter. dors. med. post. The m. pter. dors. med. post. possessesa fleshy origin from the posterior third of the anterolateral surface of the pterygoid, around the pterygoid head, and from the whole length of the ventral and posteromedial surfacesof the pterygoid. The anterior end is forked around the pterygoid with a narrow ventral head originating from the mediopalatineprocess just dorsal to the origin of the ventral head of the m. pter. retr.; it is not visible in side view. The dorsal head extends over (dorsal to) the anterior end of the pterygoid to originate from the pterygoid foot and from a small depressionon the lateral surface of the palatine hasp ventral to the origin of the dorsal head of the m. pter. retr. and dorsal to the origin of the m. pter. dors. med. ant. Insertion: The insertion of the m. pter. dors. med. on the mandible varies slightly in the four taxa of Loxops, as follows: i) virens: The m. pter. dors. med. ant. has a fleshy insertion on the dorsal surface and anterior edge of the anteromedial corner of the internomandibular flange and a tendinousattachment to a small tubercleprojecting medially immediately anterior to the internomandibular flange; this tendon runs along the anterolateral edge of the muscle. The fleshy attachment of the m. pter. dors. med. post. inserts on the anteromedial edge of the dorsal surface of the internomandibularflange posterior to the insertion of the m. pter. dors. med. ant., and from the distal three-fourths of the anterodorsal edge of the internal mandibular process dorsal to the insertion of the main belly of the m. pter. vent. med. ii) mana: The fleshy insertion of the m. pter. dors. med. ant. covers the dorsal surface and anterior edge of the anterolateral corner of the internomandibular flange; the remaining fibers insert tendinously posterior to the insertion of the tendon of the m. pter. dors. lat.; the tendon of insertionruns along the anterolateraledge of the musclevery closeto the tendon of the m. pter. dors. lat. The m. pter. dors. med. post. has a fleshy insertion on the anteromedial edge of the dorsal surface of the internomandibularflange posteriorto the insertionof the main belly of the m. pter. 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 71 vent. med. Also, a very small part insertson the medial side of the pterygoidcondyle of the quadrate. iii) newtoni: The entire insertion of the m. pter. dors. reed. ant. is fleshy on the anteromedialedge and end of the dorsal surfaceof the internomandibular flange. The fleshy insertionof the m. pter. dors. reed. post. coversthe distal seven- eighthsof the anteromedialedge of the internal mandibularprocess, with an especially strongcomponent on the distal tip of this processdorsal to the origin of the main belly of the m. pter. vent. med. and m. pref. vent. med. "fan." iv) coccinea: The strong fleshy insertion of the m. pter. dors. reed. ant. covers the dorsal surface of the anterior end of the internomandibular flange immediatelyposterior to the insertionof the m. pter. dors. lat. The m. pter. dors. reed. post. has a fleshy insertion on the anteromedial edge of the dorsal surface of the internomandibularflange posteriorto the insertion of the m. pter. dors. reed. ant., with a strong portion from the distal three-fourths of the dorsal surface of the internal mandibular process;on the left side, also from the posterior edge of the tip of this process. Comparison: Both parts of the m. pter. dors. reed. are mainly unipinnate for most of their length; the posteriorpart of the m. pter. dors. reed. is fan-shapedwhere it spreadsbeneath the pterygoid head. The fibers are moderately long in comparison with other pinnate jaw musclesand are moderately numerous. The two parts of the m. pter. dors. reed. are relatively small musclesand presumablydevelop little force. In relative size, the rank order of the m. pter. dors. reed. ant. from largest to smallest is coccinea,left side (slightly larger than right), right side, rirens, newtoni, mana. Little if any difference in size exists among the four taxa studied in the fan-like posterior part in the m. pter. dots. med. post.; it may be slightly larger in newtoni than in other forms. For the rest of this muscle, the rank order from largest to smallest is newtoni, coccinea left side, smallest and similar in coccinea right side, mana, rirens. The internal mandibular processis longestin newtoni, shorter in mana and coccinea, and shortest in rirens. The greater length in newtoni may correlate in part with the larger size of the m. pter. dors. reed. in this taxon. Action: It is doubtful from their positionsand attachmentsthat the two parts of the m. pter. dors. med. contribute to adducting the mandible or depressingthe upper jaw as do the other parts of the m. pterygoideus. Their importance appearsto lie in rotating the pterygoid around its articulation with the quadrate (or controlling this rotation) during kinetic movement. The m. pref. dors. med. ant. can rotate the pterygoid foot anterolaterallyaround the pterygoid-quadratearticulate, while the m. pter. dots. med. post. can rotate the pterygoidfoot posteromediallyabout this articulation. The latter movementis essentialin depressionof the rostrumespecially in its extreme depressionwhen the palate must be retracted back beyond its resting position when the bill is closed. Anterolateral rotation would occur when the rostrum is elevated, and it is possiblethat this part of the m. pter. dors. reed. acts during openingof the bill. The posteriorfan-shaped part of the m. pter. dots. med. post. may serve to protect the pterygoid-quadratejoint againstexcessive tension; no ligamentsappear to be present around this articulation. 5) M. pterygoideusretractor (m. pter. retr.; m pt r): This small part of the m. pterygoideusruns from the pterygoidfoot and mediopalatineprocess to the basitemporal plate; it is the only part of this muscle to insert on the braincase and hence functionsquite differently from the rest of the m. pterygoideus.A small bit of its anterior end may be visible in an oblique view through the orbit; it is visible in ventral view when the covering parts of the m. pter. vent. med. are removed. 72 ORNITHOLOGICAL MONOGRAPHS NO. 15

Origin: The origin of this retractor portion varies slightly in the four taxa of Loxops studied, as follows: i) virens: The anterior end of the m. pter. retr. is forked around the anterior end of the pterygoid. The fleshy ventral head originates from the ventrolateral surface of the mediopalatineprocess ventral to the pterygoid foot and deep to the originatingend of the m. pter. vent. med. "scap." The fleshy dorsal head originates from the posterior surface of the anterior third of the pterygoid, from the retractor processof the pterygoidfoot, and from the lateral surfaceof the pterygoidfoot ventral to the origin of the m. protr. pter. quad. and dorsal to the origin of the m. pter. dors. med. post. ii) mana: The anterior end of the muscle is forked around the anterior end of the pterygoid with two fleshy heads of origin. The ventral head arises from the ventrolateral surface of the mediopalatineprocess as far anterior as the posterior margin of the palatine blade, and from the retractor process;it is not visible in lateral view becauseit is obstructedfrom view by the origins of the m. pter. dors. lat. and m. pter. dors. med. ant. The dorsal head originatesfrom the distal end of the pterygoidfoot ventral to the origin of the m. protr. pter. quad. and dorsal to the origin of the dorsal head of the m. pter. dors. med. post. iii) newtoni: The originating head of this muscle is not forked around the pterygoid; it originates tendinouslyfrom the mediopalatineprocess and retractor process(dorsalmost part of originating head); a small fleshy head originatesfrom the connectivetissue walls of the opening of the eustachiantubes. iv) coccinea: The anterior end of this muscle is forked around the anterior end of the pterygoid with two fleshy heads of origin. The ventral head arises from the ventral edge of the mediopalatineprocess and from the retractor processjust ventral to a small ventral head of the m. pter. dors. med. post. (the latter not visible in lateral view) and immediatelyposterior to the originating head of the m. pter. dors. lat. The ventral head of the m. pter. retr. is visible in side view on the right but not on the left side. The dorsal head of the fork arises from the pterygoid foot and in slight degreefrom the posteriorpart of the palatine hasp ventral to the origin of the m. protr. pter. quad. and dorsal to the dorsal head of the m. pter. dors. med. post. Insertion: Fleshy onto the base of the brain case (basitemporalplate) in a triangular depressionto the side of the sphenoidalrostrum immediately anterior to the anterior margin of the insertingcervical muscles,the m. rectus captisventralis, and immediately ventral to part of the belly of the m. protr. pter. quad. Comparison: The fibers of the m. pter. retr. are parallel to one another and are relatively long for the mass of the muscle. This muscle may be slightly asymmetrical in coccinea, perhaps larger on the left side. The relative size of this muscle in rank order from largest to smallest is mana, coccinea, both sides similar, virens, newtoni. It is interesting to note that this muscle in newtoni is noticeablymuch smaller than in the other forms, but on the other hand the m. pter. dors. med. post. is much larger in this creeperthan in the others and may compensatefor the smallnessof the m. pter. retr. The retractor processat the posteroventraltip of the pterygoid foot is largest and most deeply concavein mana. It ranks secondin size, but is concave in coccinea. It is third in size and concave in virens, and is smallest and convex in newtoni. Larger size and greater concavity possibly can be associatedwith larger size of the muscle. Action: The major function of this muscle is to retract the palatine relative to the brain case,and hencedepress the upper jaw. This is the only part of the m. pterygoideus that can retract the whole jaw apparatusrelative to the brain case. It may also function to keep the bill closedby maintaininga retractingforce on the palate. 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 73

G) M. depressormandibulae (m. depr. mand.; m d m): The m. depr. mand. extendsfrom the occipitalregion of the cranium to. the retroarticularand internal processes of the mandible. It is clearly visible in lateral and ventral view. The m. depr. mand. is stronglyasymmetrical in coccinea,being larger on the right. This is the only jaw muscle that could be studied in sagittirostris. Origin: The origins of the three parts of the m. depr. mand. will be discussed separately, as follows: Part "a": Partly anterior and deep to part "b." Fleshy from the ventral part of the occipital area of the cranium, from most of the exoccipital process,and from the anteromedial surface of part "b." Tendinous from the anterodorsaledge of the auditory bullae; this tendon comprisespart of the anterior edge of part "a." Part "b": Broadly fleshy from the side of the cranium, from the posterolateralsur- face of part "a," and from the anterolateral surface of part "c." Part "c": Fleshy from the posteromedial surface of part "b," from a narrow strip of the occipital region of the skull posterior to part "b" and anterior to the insertion of part of the m. complexusof the cervical musculature,and from the surface of the inserting end of this neck muscle. Insertion: Again, the descriptionsof the three parts will be separate, as follows: Part "a": Fleshy on the lateral surface, dorsal edge and medial surface (not on this surface in sagittirostris) of most of the length of the retroarticular process anterior to the insertion of part "b," and on the posterior edge of the internal mandibular process (not known if the insertion of part "a" in sagittirostris includes this process, but it presumably does). Tendinously by a tendon forming the anterior edge of part "a" to the anterolateralmost part of the insertion. Part "b": Tendinous to the dorsolateral and dorsomedial sides and top of the tip of the retroarticular processanterior to the insertion of part "c" and posterior to that of part "a." Part "c": Tendinous to posteromedial tip of the retroarticular process. In sagittirostris, the same and, in addition, fleshy to most of the medial surface of this process. Comparison: The fibers of part "a" are relatively short and arranged in a series of pinnate patterns;this part is essentiallya complex bipinnate muscle. Parts "b" and "c" are long, parallel-fibered muscles. The differences in fiber length in these several parts of the m. depr. mand. are probably correlated with the differences in the distance these fibers must shorten; those of parts "b" and "c" must shorten much more than those of part "a." Becauseit containsa large number of fibers, part "a" developsmuch force, perhaps as much as parts "b" and "c" combined. But it inserts much closer to the center of rotation of the mandible and hence has a shorter moment arm than either parts "b" or "c." Hence, these last two parts may develop a greater torque than "a" and contribute more to opening the jaws. The rank order of relative size of the whole muscle, from largest to smallest, is sagittirostris, mana, coccinea right side, virens, coccinea left side, newtoni. Although sagittirostris is a larger species than the other forms of Loxops studied and has a mandibular depressoran estimated 50 to 75 per cent larger in absolute size than in the other forms, this muscle is still relatively larger by some 25 to 50 per cent. Part of this large size is probably due to the relatively larger part "c" in sagittirostris. The m. depr. mand. on the right side of coccineais possibly 30 to 40 per cent larger on this side than on the left. This asymmetrical bulkinessis more noticeable in ventral than in lateral view. Furthermore, the lateral surface of the muscle on the right side is covered with an extremely tough casing of fascia not found on the left side, and the whole dorsal margin of the muscle on the right side bulges laterally dorsal to this aponeurosis. This robust, longitudinal ridge of muscle is not found on the left. 74 ORNITHOLOGICAL MONOGRAPHS NO. 15

Part "b" in coccinea,mana, and sagittirostrisappears subdividedby vertical raphes to form portions similar in shape, superficially only, to the narrow, sickle-like part "c." Thus, part "b" is subdividedinto three parts in mana--a large, broad, anterior sectionand two slenderposterior ones•into two parts of about equal size in sagittirostris, and into six slender parts in coccinea. (Note: Plates 16 and 17 do not show the small raphe between the two most anterior subdivisions.) Such subdivision of a muscle may increaseits force, especiallyif the muscle becomespinnate with more included fibers as seen in the third and fourth subdivisions (counting back from the anterior edge) on the left side of coccinea; these two subdivisions are unipinnate. Pinnatehess could not be found, however, in the larger right muscle of coccinea, nor in mana or sagittirostris. The retroarticular process is probably relatively longest in sagittirostris, although this is a suppositionbecause this processis incomplete in all specimens. But even in the incomplete specimens,it is long and dorsoventrally deep. It is long and straight in coccinea and long and dorsally hooked in virens, but is somewhat shorter in mana and short and blunt in newtoni. These observations correlate rather well with the relative sizes of the whole muscle. Action: Bilateral contraction of the whole muscle brings about depressionof the mandible. At the same time, the resultant force vector at the mandible-quadrate articulation may act to rotate the quadrate anterodorsally about its squamosal hinge and raise the upper jaw (see Bock, 1968; Bock and Morioka, ms), but this cannot be stated definitely without analyzing all of the forces acting on the mandible. Unilateral contraction in symmetrically-muscledbirds probably allows, during depression of the mandible, a certain amount of crossing of the mandible under the rostrum in a direction toward the side on which contractiontakes place (the ipsilateral side). The m. depr. mand. would draw the retroarticular process mediafly which would rotate the mandible on the roedial condyle of the quadrate, and hence the tip of the mandible would move out laterally beyond the tip of the rostrum on that side. In the asymmetrical coccinea, however, bilateral contraction of the m. depr. mand. depressesthe mandible, of course, but at the same time (in a right-billed individual) the more robust caudal parts ("b" and "c") on the right side force the retroarticular processmore strongly toward the midline. Since the joint between the roedial condyle of the quadrate and groove on the mandible is a pivot joint, the mandible can pivot on the joint on one side and slide along on the joint on the other side, so that the anterior part of the mandible rotates toward the side of more forceful contraction (the right) and its tip crossesto the right under the tip of the rostrum as the mandible is depressed.

SUMMARY OF COMPARISONS OF JAW MUSCLES With reservations,as noted in the paragraphsabove in the sectionon materialsand methods, the datapresented in Table 3 may aid in summarizing the lengthydescriptions and comparisonsof the jaw muscles.In this table, the musclesare rankedonly in relativesize with the largestmuscle given the rank of "1"; no quantitativevalues are assignedto the differencesin any of theserankings. Larger musclesare assumedto be stronger,i.e., develop a greater maximumforce. The only valid comparisonsthat can be made in thesetables are horizontalones within the homologousmuscle or possibly within a set of musclesin the severaltaxa. Comparisonsmust not be made 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 75

TABLE 3 COMPARISONOF RELATIVE SIZES OF JAW MUSCLESOF Loxops•

L. c. COCCDlea 3 L.v. L.m. L.m. Jaw Muscle s *,irens mana newtoni Left Right Weak 4 Strong• m.p. p. ss 5• 3 4 I 2 2 1 m. p.q. 3 3 3 3 3 3 3 m. d.m. 3 I 5 4 2 4 2 m. a.m. e. r.t. 5 3 4 2 I 2 1 m. a.m. e. r. 1. 4 5 3 2 I 2 1 m. a.m. e. r.m. 5 4 3 2 I 2 1 m. a. m.e.v. 5 4 I 3 2 3 2 m. a.m. e. c. "a" 1 3 5 4 2 4 2 m. a.m. e. c. "b" 5 3 4 2 I 2 1 m. a.m.p. I 1 3 4 4 4 4 m. ps. s. 5 4 1 I 1 1 1 m. ps. p. 4 3 5 1 2 2 1 m. pt. v. 1. 4 3 5 I 2 2 1 m. pt. v.m. 4 2 5 1 2 2 1 m. pt. d. 1. 5 3 4 I 2 2 1 m. pt. d. m.a. 3 5 4 1 2 2 1 m. pt. d. m.p. 3 3 1 2 3 3 2 m. pt. r. 4 I 5 2 2 2 2

Totals 8 Whole musculature 69(5) 54(3) 65(4) 37(2) 35(1) 44(2) 29(1) Adductors (all) 50(5) 46(3) 48(4) 27(2) 26(I) 33(2) 20(I) Dorsal adductors7 35(5) 30(4) 29(3) 21(2) 15(1) 22(2) 14(1) Jaw openers 11(4) 7(I) 12(5) 8(2) 8(2) 8(2) 8(2) M. pterygoideus 23(4) 17(3) 25(5) 8(1) 13(2) 13(2) 8(I) Protractors 8(5) 6(3) 7(4) 4(1) 5(2) 5(2) 4(I)

x Comparisons can be made only across the horizontal rows, not up and down the vertical columns. • See text or appendix I for abbreviations. a Right-billed individuals. •The set of weaker muscles regardless of the morphological side. Includes the right m.p. p. ss, m.p.q., m. ps.p. and all parts of the m. pt., the left m.d.m. and all parts of the m.a.m.e. The m. a.m. p., m. ps. s. and m. pt. r. are symmetrical. • The set of stronger muscles regardless of the morphological side. Includes the opposite muscles from those listed under footnote 4. • See text (pp. 74-5) for explanation of index; lowest number indicates the strongest muscle. 7 Dorsal adductors are the mandibular adductors minus the m. pterygoideus. s The smallest totals indicate the strongest set of muscles with the ranking of the species given in parentheses. Two rankings are given for L. c. coccinea, one for the morphological sides and one for the strong-weak composite sides. verticallyalong the columnsbetween different muscles or differentsets of musclesbecause the samerank, e.g., 1, in severaldifferent muscles does not imply equal size or force development.The muscleson the two sidesof coccineawere arrangedin two ways. In the pairedcolumns marked "right" and "left," the ratingsfor thesemuscles in a right-billedindividual are given; these would be reversed for a left-billed bird. In the columns marked "weak" and "strong,"the muscleswere arrangedwith respectto the size, and hence strength,of the muscleswithout regard to the morphologicalside of the animal. Thus for a right-billedindividual, the hypertrophiedmuscles on the 76 ORNITHOLOGICAL MONOGRAPHS NO. 15 right sideof the headwould be the m. depr. mand. and m. add. mand. ext., and thoseon the left side of the head would be the m. pseudot.prof., the m. pter.and the m. prot. pter. (slightly);these would be reversedin a left-billed bird. A comparativeexamination of thesetwo systemsfor expressingthe asymmetryof the musculaturein coccineashould reveal clearly the morphological and functionalarrangement of thesemuscles in providingstrength to the mandibleand upperjaw for resistingthe opposinglateral forces acting on their tips whenthe bird is workingon a leaf bud or Koa pod. The data summarizedin Table 3 suggestthat, among the four forms thoroughlystudied, coccineahas jaw muscleswhich are relatively larger, and presumablystronger, than thoseof the otherthree taxa. Secondlargest is mana, followedby newtoniand virenswhich are about equal, although newtonimay have slightlylarger jaw musclesof the two. The strongside of coccineais considerablylarger than the weak side of this specieswhich is still larger and presumablystronger than mana, the next larger species. The rankingsfor the openersof the bill, the m. protr. pter. quad. and m. depr.mand., is mana havingthe largestmuscles, followed by coccineaboth sides, virens, and newtoni with little differencebetween the overall size of the musculaturein the first two speciesand in the last two. The protractors of the rostrumshow the sameapproximate order with coccineahaving the largestmuscles with little differencebetween the other species.The m. depr. mand.is largestby far in sagittirostris,followed by marta,coccinea right side, virens,coccinea left side, and newtoni. However, this musclemay be stronger than its relativelysmall size would indicate in coccinea,because part "b" is subdividedinto six sections,four of which are pinnateon the weak side. Becausethe m. depr.mand. also protracts the rostrum,its relativesize should be includedin any considerationsof rostral protractors.Basically, sagittirostrisand mana havethe strongestopeners of the bill, followedclosely or equaledby coccinea,with thoseof virensand newtonibeing aboutequal in size and weaker than in the first three species.This ranking agreeswell with the feedinghabits of thesespecies. The rankingsof the completeset of mandibularadductors starting with the largestis coccinearight side,left side,mana, newtoni,and virens. The considerablystronger mandibular adductors in coccineaas comparedto the other forms of Loxopsis readily apparentwhen examinedon the simple basisof right-leftmorphological sides. The magnitudeof the bilateralasymmetry and the great strengthof these adductorson the "strong"side is obviousin the summaryof the weak and strongsides of the jaw musculature in coccinea. The difference in the total mandibular adductors between the weak and strongsides in this speciesindicates the size of the lateral componentof forceon the mandible,and to a lesserextent on the maxilla,when the bird dosesits jaws. The directionof this lateral force componentis 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 77 toward the side of the bill possessingthe lateral cuttingedge. The rankings of the dorsal adductorsalone and of the m. pterygoideusshows the same general sequencefrom the largest in coccineato intermediatesize in the creepersmana and newtonito the smallestin virens. Some variationsexist such as newtonibeing slightlylarger than mana in its dorsal adductorsand virens being slightly larger than newtoni in its m. pterygoideus. These variations are of little importance becauseof the small size of the differences compared to the crude system of comparisons.The comparisons betweenthe bilateral sides of the dorsal adductorsand m. pterygoideus in coccineais especiallyinteresting. Again, separationof the musculature into weak and strong sides reflects the same large asymmetryindicated by the completeset of mandibularadductors. The comparisonof these musclesby right versusleft morphologicalsides clearly demonstratesthe hypertrophyof the dorsal adductorson the right side and of the m. ptery- goideuson the left side in a right-billed individual. Thus the dorsal ad- ductorsplace a laterally directedforce component(to the right) on the right ramus and the m. pterygoideusplaces a medially directedforce component (to the right) on the left ramus. Moreover, the m. pterygoideus placesa force componentto the left on the palate and hence the maxilla. Theselaterally directed components are in the directionof the lateral cutting edgesof the mandibleand maxilla respectively. The rankingsof the adductormuscles and the m. pterygoideusagrees with the generalfood habits of these species.The weakestset of jaw closersis found in virenswhich is a twig and leaf gleaner,taking smaller,soft-bodied insects. Moreover it feeds more on nectar than do the other forms. Both feedinghabits require a small amountof muscularforce for closingthe jaws. The creepers,newtoni and mana are mostlyinsectivorous, taking their food from the surfaceof and from crevicesin the bark of larger branchesand trunks of trees. Although newtoni takes some nectar, this is a minor part of its diet. Insectseaten by thesecreepers are slightlylarger than thosetaken by virens,and the methodof captureemployed by thesecreepers may require somewhatmore force but the magnitudeof these differencesin the strength of bill-closingin virens,newtoni and mana appearto be small. The spread of the values for theseforms for the total adductors(46-50), for the dorsal adductors(29-35) and for the m. pterygoideus(17-25) is in generalsmall when comparedto the differencebetween the strongestset of thesemuscles in thesethree forms and the weakestmorphological side in coccinea. These are namely 46 in mana to 27 in coccineafor the total adductors,29 in newtoni to 21 in coccinea for the dorsal adductors and 17 in mana to 13 in coccineafor the m. pterygoideus.These differences are greaterif the average for the two sides in coccinea is used instead of the weakest side. Maximum size and strength of the jaw closing musclesin coccineaare 78 ORNITHOLOGICAL MONOGRAPHS NO. 15 correlatedwith the feedinghabits of this species.Use of the bill to open leaf buds and Koa seedpods would require large adductingforces during variousphases of openingthese plant structures(see below). Moreoverthe asymmetryof thesemuscles agrees exactly with the asymmetricalplacement of lateral cutting edgeson the mandible and maxilla and the presumed asymmetricallateral forcesexerted on both jaws when the bird is attempting to openthese plant structuresto obtainits insectprey. It shouldbe emphasizedthat the considerablylarger adductormuscles in coccineaare not correlatedwith feedingon larger insectprey (the insectprey of thesespecies appearto be similarin size) but with a radicallydifferent method of obtaining food. Althoughinformation about the sizeof the adductormuscles in sagittirostris is not available,we suspectthat this speciespossessed the largestmandibular adductorsin Loxops, even larger than thosein coccinea,judging from the sizeof its skull. Evolutionaryincrease in size and strengthof thesemuscles in sagittirostrishas been, mostlikely, associatedmore with a changein prey items (larger sizedinsects) than with its particularfeeding habits of probing into the baseof leaves. The observedlarge m. depr. mand. and presumed large m. protr. pter. quad. of sagittirostriswould be correlatedwith its probing methodof feeding.

FUNCTIONAL INTERPRETATION OF THE SKELETAL AND MUSCULAR MECHANISM OF THE JAWS The morphologyof the skeletalelements of the jaw apparatusand of the musculaturethat move and strengthenthese bones will be correlated,as far as possible,with the observationson feedinghabits of thesebirds. We are clearly cognizantof the lack of essentialinformation on the exact movements of the jaws during feeding, on the exact forces, musculatureand otherwise,acting on the jaw apparatus,on the exactfood preferencesof each speciesand other equallyimportant factors, so that theseconclusions are offeredonly as hypothesesfor furtherconsideration and testing. In spiteof their inadequatebasis, these conclusions form a valuablebasis for further studyand we offer them withoutapology. The functionsof the jaw muscles are summarized in Table 4. The crude data on relative muscle size summarized in Table 3 does correlate with the generalarchitecture of the skull and with the typesof food taken and the general feeding methods. A more detailed comparisonof the structureof the jaw musculature,size of the muscles,degree of pinnation, sitesof attachment,together with the structureof the skull is requiredto ascertainwhether these morphological features are adaptationsfor the different feeding methodsobserved in these species. As noted above in the sectionon cranial osteology,coccinea has the 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 79

TABLE 4 FUNCTIONS OF THE JAW MUSCLES t

Function

Retraction Right Special Pro- Whole jaw lateral Muscle• Abduction Adduction traction Palate s apparatusa pull • m.p.p. ss x m.p. q. x m. d.m. x x m. a.m. e. r. t. x right m. a.m. e. r. 1. x right m. a.m. e. r. m. x right m. a.m.e.v. x right m. a.m. e. c. "a" x .95 right m. a.m. e. c. "b" x right m. a.m.p. x m. ps. s. x m. ps. p. x x left m. pt.v. 1. x? 7 x left m. pt.v.m. x? 7 x left m. pt. d. 1. x? 7 x left m. pt. d. m. a. X $ m. pt. d. m.p. m. pt. r.

• See text for complete statements of functions of these muscles in Loxops. a See text or appendix 1 for abbreviations. a Muscles running from the braincase to the mandible or palate can retract the whole jaw apparatus, while those running from one part of the jaw apparatus to another pull the mandible forward while retracting the palate with the whole jaw apparatus not moving relative to the skull. • For fight-billed L. c. coccinea, combined action of these muscles of the right or left sides pulls the mandible to the right. *The ability of this muscle to retract the whole jaw apparatus or only the palate depends upon whether it originates from the braincase or quadrate or both. • Depending upon the direction of its force vector relative to the quadrate, but probably it does not contribute to retracting either the whole jaw apparatus or the palate. 7 Adduction of the mandible by these muscles depends upon the position of their force vectors relative to the center of rotation of the mandible. Probably contributesslightly to mandibular adduction by these muscles. s Rotates the pterygoid about its quadrate articulation during protraction and retraction of the palate. sturdiestskull followed by mana, then virens and last newtoni, which has the frailest skull of all. These observations seem to correlate rather well with thoseon the overall relative size of the jaw musculatureof theseforms, the only possiblediscrepancy being that newtoniranks third rather than fourth in musculaturesize. However, the difference in musculaturebetween virens and newtoniis probablyinsignificantly small. What is importantis that the specieswhich use their jaws least strenuouslyin obtainingfood virens and newtoni--are the oneswith frailer skullsand weakerjaw musculature. The slightlylarger jaw musclesin newtonicompared to virensin spite of the somewhatfrailer skull in the former may be associatedwith the possibility that newtoniworks more strenuouslyat bark and lichensto obtain some of its food. 80 ORNITHOLOGICAL MONOGRAPHS NO. 15

Similarcorrelations can be madefor the large mandibulardepressors and rostral protractorsin coccineaand mana which probe and gape stronglyin their feedingoperations on tight leaf buds and Koa pods, and on bark and lichens,respectively. These activities require more force in openingthe jaws, and thesespecies have the largestand presumablystrongest m. depr. mand. and m. protr. pter. quad. The specieswhich gape mostly in foliageand blossomsand to a much lesserdegree than othersbetween pieces of bark and lichen--virensand newtoni--requireless force to opentheir jaws and hence have smaller,weaker muscles to openthe jaws. L. sagittirostrisobtains much of its food by gapingin the crevicesat the basesof stout Freycinetialeaves which would involveconsiderable gaping force. This specieshas the largest m. depr. mand. observedin the genusand presumablyhas a relativelyhuge m. protr. pter. quad. Captureof insectprey would be relativelysimilar in the four taxa as this action involvesrapid movementof both jaws againstlittle resistanceuntil the prey item is graspedbetween the jaws. The sizeof insectand other prey is similar in the four forms of Loxops and should require approximately equalamounts of adductorforce for capture(closure of the jaws) and sub- sequentcrushing. It is not possibleto distinguishbetween those adductor musclesprimarily responsiblefor closingthe jaws and those that provide mainly crushingforces on insectsand other food objectsheld in the mouth. The basic size of adductor musclesin Loxops would correlate with this minimumrequirement of closingforce. Gleaningof insectsfrom the surfaceof foliageand small twigsby virens requireslittle crushingforce and the small size of the mandibularadductors in this speciesis an adaptationto the food habitsof this species. Graspingand pullingbits of bark and lichensby the two creepers,mana and newtoni,as they probefor insectson the bark surfaceand in the crevices of large branchesand tree trunks would require somewhatmore adductor force. The slightlylarger total adductormusculature and sturdier skull in theseforms appearto be adaptationsfor their probingmethods of feeding. The extra amountof adductorforce requiredby the probersover that needed by the surfacegleaner virens is smallas indicatedby the minor increasein the adductormuscles in the creeperscompared to thosein virens. Considerablymore adductorforce would be neededby coccineafor opening leaf buds and Koa pods which is reflected in the sturdier skull and larger adductormuscles (including the m. pterygoideus)in this species.In- creaseddevelopment of the mandibular adductor in coccinea,as shown in Table 3, providesforce over and above that requiredto captureand crush its insectprey. A considerablepart of this increasein adductorforce arose becauseof the requirementfor a laterally directed force component(see below). However, at least some of the increase in mandibular adductors in 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 81 coccineadeveloped because of the need for greaterbilateral force for closing the jawsduring opening plant structuresto exposeits insectprey. Nectar feedingin virens,coccinea and newtoni (rarely) doesnot require any specialdevelopment of the jaw muscles.Presumably the actionsof the jaw musclesrequired for this feeding method fall with the capabilitiesof thesemuscles for the other feedingmethods employed by each species.

THE ASYMMETRICAL JAW APPARATUS OF L. coccinea In the sectionon the methodsof feeding of the coccineawe suggested that this speciesprobably opens the imbricated leaf buds of Metrosideros by using methodsalmost analogousand very similar to the methodsused by Holarcticcrossbills of the genusLoxia in openingpine conesto obtain pine nuts. A number of natural historians and anatomistshave investigatedthe method of feeding and the jaw and tongue mechanismsof the crossbills. Among thosewhose observations on the feedingbehavior have been published are Townson (17997 in Yarrell, 1829:462-463), Gadow (1891:495- 496), Duerst (1909:284-287), B/Sker(1922:88-89), and Robbins (1932: 161-164); it is our opinion that the accountsof Robbins and Duerst are the mostcomplete and logical, and probablythe most accurate. Many of the asymmetricalstructures in the crossbill (Huber, 1933) are similarlyso in coccinea.Some of the structureshave alreadybeen described in the sectionson the rhamphothecaand the jaw musculature.Others have not. Before describingthese, it may be instructiveto outline briefly the asymmetriesalready mentioned. Descriptions are for right-billedspecimens as are all discussionson possiblefeeding methods. The tips of both rhamphothecaeare asymmetrical;the bony skeletalelements deep to the horny parts are not. The mandibularrhamphotheca is bent or crossedto the right under the rostral rhamphotheca,which is bent or crossedover the former to the left (see Plate 2). The right or cutting edgeof the tip of the mandibularsheath is wider and flatter (thoughit may alsobe gougedout throughuse) and usuallyhigher than the left edge. The left or cutting edge of the tip of the maxillary rhamphothecais likewise wider and flatter (it too may be gougedout), but in contrastto the mandibular rhamphotheca,the left edgeis lower than the right edge. Severaljaw musclesare asymmetricalwith the muscleon one side of the head hypertrophiedbeyond the expectedsize of that musclein coccinea. In addition,the fiber arrangementhas modifiedin some of these musclesto provideincreased force. In listingthe hypertrophiedjaw muscles,the numbers in parenthesesare estimatesin per cent of the amountof increasein sizeof the hypertrophiedmuscle against the unhypertrophiedmuscles of the opposite side and the asterisk indicates those muscles believed to be the 82 ORNITHOLOGICAL MONOGRAPHS NO. 15 moreimportant for providinglateral force. The musclesenlarged on the right sideof the headinclude all partsof the m. add. mand.ext.* (30-40) and m. depr.mand. "b"* (30-40). The muscleshypertrophied on the left are the m. protr. pter. s. str. (5-10), m. pseudot.prof. (slight), m. pter. vent. lat.* (85-100), m. pter. vent. med.* (50), m. pter. dors.lat.* (60-70), m. pter. dors.med. ant. (slight), m. pter. dors. med. post.* (20-30), and m. pter. retr. (slight). Those muscleswhose pull is in or closeto the plane of rotation of the mandibleabout its quadratehinges, e.g., the m. pseudot.supeft. and m. add. mand. post., are symmetricalin size. It could not be determinedwhether m. complexusof the back of the neck is asymmetricallydeveloped because of destructionof this area during col- lectionand dissectionof the specimensbefore realization of the importance of this musclein the feeding methodof coccinea. We suspectthis muscle to be hypertrophiedon the left as in Loxia becauseof the similar twisting actionof the head in the two genera. Certain features of the bony skeleton of the head of the coccinea are developedasymmetrically. Some of thesehave alreadybeen described briefly in the sectionon the descriptionof the musculature.On the right side the zygomaticprocess extends farther posteriorlyonto the lateral surface of the cranium and has wider anterodorsaland posteroventralsurfaces than on the left. This asymmetryreflects the hypertrophyof mm. add. mand. ext. rost. temp., add. mand. ext. rost. med., and add. mand. ext. vent. on the right side of the head. The sturdierbuild of the prepalatinebar and transpalatineprocess and the larger and more rugosesurface area of the ventral surfaceof the internomandibularflange on the left sideof the headcorrelate well with the enlargedleft m. pter. vent.lat. And the largeleft palatineblade correspondswell with m. pter. dors. lat. which is larger on this side than on the right. The quadratomandibularjoints (Plate 9) of coccineaare similarin basic designand functionto thoseof the crossbillsalthough not quite as highly perfectedas thoseof the latter. In coccineathe joint on the right is much looserthan the one on the left and allowsa lot more fore and aft slidingof the medialcondyle of the quadratein the trough-likedepression in the dorsal surfaceof the internomandibularflange. The left joint actspretty much like a ball and socketjoint althoughnot perfectlyso, for a small amount of anteroposteriormovement is permitted. The reasonsfor such movements can be seenin the bony structuresof thesejoints, namely, the internomandibularflange, mandibular and jugal processesof the quadrate,and external processof the mandible. The trough or slot in the dorsal surfaceof the right internomandibular flange is wide and open, providing a relatively large space in which the longerright medialcondyle of the quadrateon the mandibularprocess can 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 83 slide backwardsand forwards. The anterior end of this trough is wider than the same area in the left slot. The ventral surfacesof the medial condyle are smooth, rounded articular surfaces. The articular surface of the right side extendsfarther posteriorlythan the left one, providing a longer surface for maintenanceof articulation with the wider anterior part of the slot in the right internomandibularflange during retraction of the right mandibular ramus at the time of closure of the bill. The trough in the left internomandibular flange is narrower than that in the right; this is especiallynoticeable at the anterior end of the slot and may representthe incipientstage of the evolution of a shortened,eventually round, socket in this flange similar to that of Loxia. The shorterand rounderleft medial condyleof the quadrate fits more snuglyinto its trough than does that on the right. Little anteroposteriormovement is permittedby this tight fit, but pivotingof the mandible to the right is allowed, along with the normal dorsoventralmovements. The jugal processesof the quadrateand the externalmandibular processes are also asymmetrical.The ventral surfaceof the right jugal processof the quadrateis more deeplygrooved than that of the left. This greatlyfacilitates fore and aft movementof the externalmandibular process over the narrow, flattened dorsal edge of the right mandibularramus in this region, and aids in preventingdisarticulation. The articulatingsurface of the lateral condyle of the left quadrateon the jugal processis much flatter than that of the right and, therefore,facilitates rotary pivoting of this processon the wide, flat left externalmandibular process. This left mandibularprocess projects laterally from the mandibularramus like a shelf and is muchmore extensive in area than the one on the right. It is our opinionthat the left external mandibularprocess is used in main- tainingthe articulationbetween the lateral condyleof the left quadrateand the left ramus during retraction of the mandible on the right and lateral rotation of the mandibletowards the right during both openingand closing of the bill. The left external mandibularprocess is placed so that, at the time of depressionof the rostrum, adduction of the mandible, and rotation of the mandibletowards the right, the lateral condyleof the left quadrate maintains its contact with the flat, dorsal articulation surface of the left externalmandibular process, thus aiding in insuringsmooth articulation. In order that this coincidenceof articulatingsurfaces can take place, it is no- ticeableto the eyethat the left quadratemust be in its restingposition because thesearticulating surfaces do not meetwhen the left quadrateis in an extreme retractedposition. This is probablybrought about on the left throughthe strongcontractions of the hypertrophiedpterygoid muscles which place the left quadratein the restingposition by overpoweringthe tendencyof the small left m. add. mand. ext. and the symmetricalm. pseudot.superf. to retract the mandible and hence the left quadrate. Simultaneouslyon the 84 ORNITHOLOGICAL MONOGRAPHS NO. 15 right sideof the head,the hypertrophiedm. add. mand. ext. aids in retracting the mandibleand quadrateon the right, and rotating the mandible to the right. Thus, on the left side of the head, mandibularramus and quadrate are in approximatelythe restingposition, and on the right, the mandibular ramusand quadrateare in the retractedposition, with smooth,strong articulation preservedon both sides. Such a maneuvercould take place probably only in birds with an asymmetricaljaw apparatussuch as is possessedby Loxia and Loxops coccinea. The left externalmandibular process is placed, also, so that at the time of protractionof the rostrum (when the quadrateis swungforward and when the mandibleis depressed,retracted on the right, and swungto the right) the left jugal processof the quadrateagain perfectlycoincides with it, thus maintainingsmooth articulation. If the left external mandibular processwere not situatedas it is, there would be no articulationat this time betweenit and the left jugal processof the quadrate,and at sucha time, the chancesof disarticulationbetween mandible and quadrate would be great. In the sectionon methodsof feeding,we outlinedthe presumedbasic methodused by coccineato openOhia leaf budsand Koa pods. This de- scriptionis hypotheticalbecause close, accurate observations were impossible to make in the field. It is basedupon superficialobservations of feeding behavioron the part of Perkins (1903:418-420) and Richards,and upon a carefulstudy of the anatomyof the jaw apparatusincluding considerations of the feasiblefunctional properties of thesestructures. The severalmethods to be described were reached on the basis of a careful reconsideration of all availableevidence by Richardsand Bock at the time the final draft of this manuscriptwas readiedfor publication;although the basic ideas are the samethese methods differ somewhatfrom thoseadvocated earlier by Richards (1957) in his thesis. In thesedescriptions, we shall emphasizeimportant aspectsthat shouldbe watchedfor wheneverthe feedingbehavior of coccinea can be observedby other ornithologists.The diagramsin Text Figures8, 9, 10 and 11 are semi-schematicattempts to show the relationshipsbetween the positionsand movementsof the bird and the tips of its jawswith respect to the leaf bud or Koa pod.

Figure 8. Schematicdrawings showing the position of L. c. coccinea relative to a leaf bud while opening it and several possiblemethods of inserting the bill into the bud. A) Closedbill method with bill insertedinto far side (from body) of leaf bud, this being the most probable approach. B) Closed bill method with bill inserted into near side of leaf bud. C) Openedbill method with bill insertedinto far side of leaf bud. D) Openedbill method with bill insertedinto near side of leaf bud; the ap- proachesshown in B and D are probably used rarely if at all. 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 85 86 ORNITHOLOGICAL MONOGRAPHS NO. 15

Threebasic methods will be considered,two for openingleaf budsand one for Koa pods;we have excludedminor variationsof these methods. When workingon an Ohia leaf bud, the bird would graspthe twig firmly with its feet just below the bud; the bird's body and many of its movements were usuallyobscured by the large leavesjust below the bud. The bird was observedto bend its head downwardso that its chin seemedto rest against its neck. Presumablythe bird may hold the leaf bud againstits body (Fig. 8) so that the forceexerted by the tip of the bill on the leaf bud as the tip pene- tratesinto the bud is resistedby force exertedon the oppositeside of the bud by the body of the bird. If the bud was not bracedagainst the body of the bird, it would move when the bill is inserted into it and thus reduce the effectiveforce of penetration. The positionof the bud relative to the body of the bird is an importantpoint and shouldbe noted by field observers. The secondpoint is whetherthe bird bendsits head over the tip of the bud and insertsits bill into it on the sideopposite to the body of the bird or whetherthe bird insertsits bill on the sameside as its body (Fig. 8). We haveassumed that the bird bendsover the bud and penetratesthe bud on the oppositeside. This has two clear advantages.First, the force of the bill on the bud is resistedby the force of the body on the bud which would not be possibleif the bill is insertedon the same side as the body; otherwisethe force of the bill would tend to push the bud away from the bird (bend it on its stem) and thus reducethe force of penetration.Second, the curvature of the jaws correspondscloser to the curvatureof the leaf scaleswith the bill either closedor openedat the time of penetration.If the bird inserts its bill into the bud on the sameside as the body, then the curvatureof the jaws do not coincideto the curvatureof the leaf scales. Consequently,the bird wouldrequire more forcewhen insinuatingits bill, and the bill probably could not be inserted into the bud as far as the bases of the leaf scales. It is possiblethat the bird insinuatesits bill into the apex of the bud from above. The shapeof the jaws correspondsreasonably well to the curvatureof the distal leaf scales,but the bud would probably bend somewhatunder the force of penetration. It is possiblethat the bird attacksthe bud directlyfrom the side and forces its bill throughthe surfacesof the leaf scalesrather than insinuatingit between adjacentscales. However, we doubt that this occursas a regular method and shall not consider it further. The field observershould note the side of the bud penetratedby the bill as the bird startsto open it, to determineif the bird approachesthe bud from above,on the sameside as the body, or on the oppositeside, and, if possibleto ascertainwhether the bird insinuatesits bill between adjacent scalesor pushesit throughthe flat surfaceof leaf scales. We assumethat the bird insertsits bill betweenleaf scaleson the oppositeside of the bud. 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 87

Figure 9. Cross-sectionthrough leaf bud and tips of the maxilla (D) and mandible (V) of a right bill L. c. coccinea with its bill inserted in a leaf bud as shown in Figure 8A to show the sequenceof movementsof the bill in opening the leaf bud. The series of movements are: A) Insertion of bill into leaf bud; B) Rotation of head and bill 90 ø to the right; C) Lateral movement of mandible to the right; D) Gaping of bill; E) Rotation of gaped bill to the left to the original inserted position; F) And continued rotation 90 ø to the left of the original inserted position. As a result of these movements the leaf bud is damaged and a wad of material is cut and may be removed to exposed hidden insects. 88 ORNITHOLOGICAL MONOGRAPHS NO. 15

Further,the field observershould note, if possible,whether the bill is closed or gapedslightly at the time of penetration.Both methodsare feasibleand we shall describethe probable sequenceof actions associatedwith each type of penetration. In the closed-billmethod (Figs. 8 and 9), the bird would insinuateits closedbill betweentwo adjacentleaf scaleswith the frontalplane of the bill heldparallel to theflat surfacesof theleaves; the tip of bill is slightlyflattened dorsoventrallywhich reduces the force requiredto insertit into the leaf bud. The two adjacentscales are forced apart as the bill is pushedtoward the axisof the bud. Whenthe bill haspenetrated fully, severalactions take place probably,but not necessarily,in the followingsequence. Also, the bird may repeat one or more of theseactions, again not necessarilyin the order described. The first step (Fig. 9) may be a rotation of the head and bill approximately 90 ø to the right, that is, oppositeto the directionof the final cutting motion. Some,but relativelylittle resistancefrom the leafy material will be met by the bill. The lateral cuttingedges of the jaws wouldpush the adjacent leaf scalesfurther apart and may have cut or torn the surfacesof the leaves. The cuttingedges of the jaws are perpendicularto the surfacesof the leaves and henceare in positionto penetrateand cut them. Moreover, the left m. complexusof the neck is stretchedprior to the turning of the head toward the left during the cuttingaction which is soon to follow. Such stretching permitsthe muscleto remain in the high force portion of the tension-length curve when it contractsand shortensto rotate the head during the cutting phaseof this sequence.This rotationof the head to the right is important and shouldbe notedby field naturalists. A lateral movementof eachjaw in oppositedirections forcing their lateral cutting edgesinto the proximal and distal leavesmay follow next. This actionmay take placeprior to gaping,simultaneously with gaping,or following it. The extentand significanceof lateralmovement of the jaws is difficult to ascertain.In any case,the maximumlateral displacement of the rhamphothecaltips is smallas comparedto their movementduring gaping. A gaping action would follow in which the mandible and maxilla are openedwith an asymmetricallateral movementof the jaws if this has not occurredalready. The lateral rhamphothecalcutting edges possibly scour, cut into or otherwisedamage the adjacentsurfaces of the proximaland distal bud leaves. The jaws would now be in final positionprior to the rotary cuttingaction. Gapingis essentialfor increasingthe sizeof the wad of leafy material to be cut from the bud. The head and bill would now twist to the left by action of the m. com- plexusand other cervicalmuscles. The cuttingedges of both jaws, having 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 89 been placedperpendicular to the surfacesof adjacentleaves, damage and turn back, or cut through the bud leaves. The cut presumablycontinues until the head has rotated through approximately180 ø , and the mandible and maxilla reachingthe positionoccupied by the other at the start of the rotary cut. During this rotary motion, a large asymmetricallateral force would act on the edge of each jaw--a force directed toward the right acts on the maxilla and one directed toward the left acts on the mandible. These forces must be resistedor counteractedby laterally directedmuscular forces. The asymmetricaljaw musclescan provide the necessarylateral forces. The hyptertrophieddorsal adductorson the right side of the head provide a lateral force componenton the right mandibularramus directedto the right as do the hypertrophiedleft m. pseudot.prof. and m. pter. on the left mandibularramus. The hypertrophiedleft m. pseudot,prof. and m. pter. provide a lateral force componenton the maxilla via the bony palate directedto the left. The asymmetricalmuscular forces prevent rotation of the mandible to the left and reducestress on the nasofrontalhinge of the maxilla.Although the maxilla cannot rotate laterally, the lateral force on its tip would place a lateral torque on the maxilla about the nasofrontalhinge; hence the bony material of the hinge would be stressed,and this stresscan be reducedby the oppositeasymmetrical muscular force of the m. pseudot.prof. and m. pter. via the bony palate. Finally, the jaws would closeon the wad of cut and damagedleaves which is graspedfirmly and pulled out of theleaf bud by withdrawingthe head. The wad is then discarded,and any caterpillarsor other insectsexposed insidethe bud are seizedquickly and eaten. A subtle but important point concerningthe asymmetricallydeveloped jaw musclesmust be made. The laterally directedforce componentof these musclescan rotate the mandible to the right. We have assumedthat such rotation occursbecause of the structureof the quadrate-mandibularhinges and becauseslight lateral rotation of the mandiblewould allow the jaws to absorbexcessively large shocksthat may disarticulatethe mandible if it was rigid in the horizontal (frontal) plane. Further, we have assumedthat the mandible is crossedtoward the right beneath the maxilla before the start of rotary cutting;its tip may be forced back to the left as the bill cuts through the leaf bud. However, lateral movement of the mandible is not essential for the methodof cuttingwads in leaf buds. And asymmetricaldevelopment of the jaw musclesand associatedskeletal elementscan occur without the possibilityof lateral movement of the mandible. The asymmetricallateral forces producedby these musclescould serve only as a counterbalanceto the asymmetricallateral external forces acting on the bill as it cuts through 90 ORNITHOLOGICAL MONOGRAPHS NO. 15

Figure I0. Cross-sectionthrough leaf bud and tips of the maxilla (D) and mandible (V) of a right-bill L. c. coccinea with its bill inserted in a leaf bud as shown in Figure 8C. In this method the bird uses the following sequence. A) The opened bill is inserted with the result that several leaves separate the two jaws. B) The bill is rotated about 90 ø to the left until a wad of material is cut and can be removed to expose hidden insects. the leaf bud. Consequently,while we believe that crossingof the mandible to the right occursas a normal part of cuttingwads out of leaf buds, this is not an absolutelyessential part of this feeding method. A secondpossible method of cutting wads from leaf buds is the opened- billed methodin which the bird would gape prior to insinuationof the bill into the leaf bud. In this case, the maxilla and mandible would be inserted between different leaf scalesso that several leaves lie between the two jaws (Fig. 10). When the jaws have penetratedfully into the bud, they are already gaped and displacedlaterally. In this case,rotation of the head to the right probablydoes not occuras thiswould force the blunt edgeof the jaws against the bud leaves. Rather the head would be rotated only to the left with the cuttingedges of the jaws cuttingthrough leaf material. Presumablythe head must be rotated only 90 ø or slightly more to cut a wad from the bud; part of the circumferenceof the wad is formed by the gap.between different sets of adjacentleaves which has been enlargedwhen the jaws are insinuatedinto the bud. The wad is then grasped,pulled out and any exposedinsects seized and eaten. During the cuttingof wadsby either method,the bird may gape, cut laterally and graspthe wad repeatedlyto furtherweaken the bud. It is unlikely that the bird cuts a neat wad from the leaf bud every time. No matter what variationsexist in the cutting methods,a twisting of the head seemsto be

Figure 11. Schematicdrawings showing A) the position of L. c. coccinearelative to a koa seed pod while prying apart the two valves and B to F) the sequenceof movements of the jaws during this action showing the edges of the seed pod valves and bill tips in section. In this sequencethe bird: B) Inserts the tip of its bill between the free edges of the valves; C) Rotates the mandible to the right, and/or moves its head and bill from side to side; D) Rotates its entire head and bill 90 ø to the right; E) Gapes its jaws with the bill held in the rotated position; F) Rotates its gaped bill 90 ø to the left back to its original inserted position. The result of this action is to spring open the seed pod and expose any hidden insects. 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 91

B 92 ORNITHOLOGICAL MONOGRAPHS NO. 15 an essentialmovement. Were this not to occur, the functional and evolutionary basesfor the asymmetricaldevelopment of jaw muscleswould be difficult, if not impossibleto explain. Little evidenceor reasoningexists on which any decisioncan be made aboutthe frequencywith which the closed-billor opened-billmethod is used or whether one or the other is not used at all. The size of the wads cut is about the samealthough their shapesdiffer slightly. Insinuationof the bill may be easier and require less force in the closed-billmethod than in the opened-billmethod. But the opened-billmethod requiresfewer movements after the bill has penetratedinto the leaf bud and perhapsrequires less ex- penditure of energy than the closed-bill method. It seemsfeasible to assume that the bird can generatesufficient force for all stepsin each method,and no step in either methodappears to require considerablymore force than the maximum force requiredin the other method. At this time we can con- dude only that eachmethod is possibleand that field observationsare needed for further resolution. Openingthe long, slender,thin Koa pods can be accomplishedby similar movementsof the jaws and head. The bird clingsto the pods and insertsits bill betweenthe slightly openedslit betweenthe two shellsof the pod (Fig. 11 ). In this case, the bird clings to the pod so that the force required to wedge the bill betweenthe pod halvesis resistedby transmissionof force via the feet and hind limbs of the bird. The bird's head is bent downward in a positionsimilar to that usedin cuttingleaf buds. The first stepis forcingthe bill into the slit betweenthe halvesof the pod. The bill is probablyheld so that its frontal plane is parallel to the edgesof the slit, allowingthe slightlyflattened tip of the bill to be wedgedbetween the pod halves (Fig. 11 ). Presumablythe bill is insertedonly a slight distance,probably only enoughto allow the tips to obtaina hold on the margins of the pod walls. The secondstep would be lateral movementof the mandible without any gapinguntil the tips of the jaws are crossedto a maximumamount. During this movementthe lateral cuttingedges would cut any material connecting the edgesof the two pod shells. Possiblythe bird also swingsits head from side to side to cut throughmore connectingmaterial. With the tips of the bill crossedmaximally, the bird would rotate its head about 90 ø to the left so that the smooth curved surfacesof the jaws slide alongthe edgesof the pod. Rotation of the jaws would force the pod halves apart as the lateral distanceof the crossedmandibular tips is greaterthan their vertical dimension. The gougesin the lateral cutting edges of the jaws probably hook on the edge of the pod shells. Presumablythe initial twistingof the mandibleis not sufficientto. spring openthe pod. The bird would then gapewith the jaws held in the crossed 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 93 position.As the jawsopen, they would slide along the edgesof the pod; the gougeswould maintain the positionof the jawsand prevent them from slippingout of the pods. With the jaws held opened,the bird wouldtwist its headback to the originalposition, thereby spreading the halves of the podfurther open. The bird mayhave to repeatthe entiresequence at severalpoints along the pod beforeit is sufficientlyopen or springsopen by elasticrecoil of its drying wallsto exposeinsects that may be livinginside the pod. The insectswould then be seizedquickly and eaten. In openingKoa pods,lateral crossing of the tipsof the jawsis an important component.Actual movement of the mandibleis involvedto increase the distancebetween the two cuttingedges and hencethe distancethe pod halvescan be forcedapart during the initialrotation of the bill. Oncemaxi- mumcrossing of the jawsis reached,the musclesmust continue to provide lateralforce to holdthe jawsin theircrossed position as the bill turnsagainst the resistanceof the pod halves.Consequently, when openingKoa pods, both lateral movementand large lateral muscleforce componentto counter externalforces appear to be essential.The gougesobserved in the rhamphothecaeof the cuttingedges are presumablythe resultof wear as the bird gapesits bill in the semi-openedpod. The gougesare adaptationsin that theymaintain the positionof the jawsand prevent their slipping out of the pod while the bill is gaped. A comparisonof the methodsused to cut wadsfrom leaf budsand to openKoa podssuggests that the morphologicalfeatures of the headin coccineaare adaptationsfor both typesof feeding.Most interestingis that neitherfeeding method involves structural modifications that are disadvan- tageousfor theother method. The asymmetryof the jaw musclesin coccinea wouldpermit an automaticsidewise cutting by the rhamphothecaltips or asymmetricalresistance of forceby thesetips whenthe musclescontract. Comparedto a bird withsymmetrical jaw muscles,like the creepers,coccinea has a tremendousadvantage at suchtimes when theselateral movements or resistanceto force are usedin feeding. Presumablythe creeperscan use theirjaws asymmetrically on occasionwhen feeding. In orderto bringabout a sidewisecutting and spreadingmovement of the jawssimulating coccinea, the creeperswould have to relaxpart of the mm. add.mand. ext. and depr. mand."b" and "c" on the left andthe mm. pter. vent. and dors.on the right (that is, the bird wouldhave to contractthe musclesdifferentially and not achieve maximum force of these muscles on one side of the head). Thus, the creeperswould not be usingtheir musclesfully, whilecoccinea could counteractits musclesmaximally and still achievelateral movementsor lateral force development.Further, coccineacan achievesymmetrical actionsof itsjaws by not contractingthe hypertrophied muscles fully andhence 94 ORNITHOLOGICAL MONOGRAPHS NO. 15 developingequal force on eachside without having to sacrificeany force that could be attained. The asymmetriesof the head and the functioningof the asymmetrical jaw apparatusin the crossbillsof the genusLoxia and in coccineaare similar in that theyboth depend upon a very similarquadrate-mandibular articulation and similarasymmetries in the individualmuscles. In both cases,the mandiblecan be rotatedlaterally on the crossedside, but not the other, and thehead is rotatedduring cutting action. In bothcases, strong and opposite lateral muscularforce componentsare neededto resistthe asymmetrical lateralforces on the tips of bothjaws. And in both cases,the bird "drills" its bill into the plant material.The major differencesare in the shapeof the crossedrhamphothecal tips whichnecessitate a differentsequence of movementsin the two forms. Consequently,the cross-billedcondition and asymmetricalmorphology of the head in the generaLoxia and Loxopsand associatedfeeding methods are not homologous,having evolved independently from differentancestors with symmetricalcranial features. But it is quite possiblethat they are closeparallel developments dependent upon the pos- sessionof a similarquadrate-mandibular hinge in the commonancestor that permittedlateral swinging of the mandible.

THE TONGUE APPARATUS

INTRODUCTION The morphologyof the tongue,or hyoid apparatusas it is often called, will be describedin three sections,the corneouspart of the tongue,the skeletonof the tongueand the tonguemusculature, which are followedby a considerationof the functionalproperties and adaptationsof the tongue. We preferthe termstongue skeleton and the tonguemusculature rather than hyoid skeletonand hyoid musculaturefor severalreasons. First, skeletal elements from the first branchial arch are included in addition to elements from the hyomandibulararch; the "hyoid"horns are formedfrom partsof the first branchial arch. Second, musclesderived from mandibular arch muscles,branchial muscle plates and the hypobranchialspinal muscle series are includedin the set of musclesoften called the hyoidmuscles in addition to thosefrom the hyomandibulararch. And lastly,the wholemorphological complexis associatedmainly or only with the mechanismof the tongue; hencein analogyto the jaw apparatus,the tongueapparatus is a more apt term than the hyoid apparatus.

THE CORNEOUS TONGUE The corneoustongue might be calledthe tongueproper and is that part visiblein an examinationof the externalmorphology of a bird. It consists 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 95 of a resilientcorneous sheath supported from within by the paraglossalbones of the hyoid skeleton.The sheath,presumably continuous with the stratum corneumof the epidermisof the skin, is boundby fibrousconnective tissue to thesebones. Other than somesmall blood vessels, nerves and glands,and perhapssome skeletal muscle, the paraglossal bones and surrounding corneous sheathconstitute the major histologicalcomponents of the tongue (Moller, 1931:121, Fig. 10). In the genusLoxops the tonguesare of two generalshapes (Pls. 19 and 20). Thosespecies in the subgenusViridonia (virensand sagittirostris)have tongueswhich are long, slender, somewhatdecurved and tubular. The tonguesof the forms in the subgenusParoreomyza (mana and newtoni) are shorter,broader, rather straight, and shallowly troughqike rather than tubular. The tongueof coccineaof the subgenusLoxops is structurallysomewhat intermediatebetween the tonguesof the two other subgenera.It is broad in its posteriorhalf in a mannersimilar to the tonguesof Paroreomyza,but it is tubular in its anterior half like those of Viridonia. It is not as decurved as the tongueof virensbut is more so than thoseof Paroreomyza. In decurved,tubular tonguessuch as thoseof Viridonia the lateral edges are curled upwardsand medially. As the tonguenarrows anteriorly,the edgeof one sidecurls over that of the otherside. Theseedges are rather thick proximally,but distallythey are thin and transparentand somewhat shreddedinto laciniae(fringes) so that they have the appearanceof the edgesof the vane of a partially unpreenedfeather. At the tip of the tongue these laciniae are long, slender,and finely bristleqikeand sometimeseven forked at their tips, e.g., in sagittirostris,so that the tip of the tonguelooks like a minute brush. Such a tongue no doubt makes a good conduit for nectar, water, and probablyjuice from fruits in the partially nectarivorous forms such as virens and probably also sagittirostris.The brush-liketip possiblyallows thesebirds to take advantageof the physicalprinciple of capillary action when they ingestthese liquid foods. Amadon (1950:223) notedthat the driedtubular tongues of drepaniidscollected fifty yearsearlier "whenplaced in alcohol. . . still suck up the fluid by capillaryattraction." The decurvedshape of the tongueprobably is an adaptation,concomitant with the decurvedbill, to the taking of nectarfrom the curvedblossoms of the endemiclobelias. It was notedin the sectionon methodsof feedingthat the Maui and Kauai races of coccinea also take nectar on rare occasions. This species,too, hasa tubularand decurved tongue, although the tubularpart is relativelyshorter than in Viridonia. It was Perkins' opinion that these tubulartongues of virens,sagittirostris and coccineaare of useto the species in procuring small insectsfrom small crevicesbetween leaves or in lichens, at the basesof Freycinetialeaves, or in Metrosiderosleaf buds,respectively. Althoughthe fringed tip of the tonguesin thesespecies would be useful in 96 ORNITHOLOGICAL MONOGRAPHS NO. 15 pickingup insectsespecially if it is coveredwith sticky mucus,the tubular configurationof the tonguewould not possessany value in insectfeeding; we presumethat Perkins was alludingto the fringed tip of the tongue in his statement. In the shallow, trough-like tonguesof Paroreornyza,the edges curl upward and inward only near the tip, and only here are laciniae and bristles present. The edgesdo not overlap to form a tube. The tip of the tongue is stiff and forked. It may be that this forked type of tongue is better adapted to picking up insects,perhaps larger ones, than the brush-tipped, tubulartongues of Viridonia, and that the tonguesof the latter are primarily adaptedto taking nectar and secondarilyadapted for ensnaringtiny insects. Nevertheless,the trough-liketongues of newtoni, and also of the Lanai and Kauai forms, must be capable of conductingnectar into the mouth, as we have no reasons to doubt the observations of Perkins and Munro of these forms taking nectar. It would seemthat the backwardly-projectingspines at the proximal ends of all of thesetongues and on the roof of the mouth are adaptationscommon to many passefinebirds, aiding in the retention of highly active arthropod prey in the mouth.

OSTEOLOGY OF THE TONGUE The tongueskeleton (Pls. 21 and 22; Fig. 12) which supportsthe corneous tongue and its musculatureis derived from the combinedhyoid (second) and third visceralarches of primitivepiscine vertebrates. We follow the terminology advocatedin Bock and Shear (ms. b) which follows closelythat of Engels (1938). The tongue apparatuslies for the most part ventral to the skull between the mandibularrami exceptfor the epibranchialiawhich curl upward around the back of the skull in a slot or pocket in the superficialfascia outsidethe musculature,posterior to m. depr. mand. "c" and lateral to m. complexus of the neck. The apparatusis attachedto the mandible and skull by five paired muscles. Anteriorly the m. genioglossusattaches the tongue to the mandibularsymphysis; laterally the m. branchiomandibularisand m. mylo- hyoideusattach it to the medial surfaceof the mandibular ramus; and posteriorly the m. serpihyoideusconnects it to the basal part of the cranium, and the m. stylohyoideusattaches it to the retroarticularprocess of the mandible. Furthermore,at the distal tip of the ceratobranchialia,tufts of fascia attachedto the surface of m. branchiomandibularisblend in with the superficial fasciaand the thin periostialsheath which coversthe whole skull, thus attachingthe hyoid skeletonto the skull in this area, too. The whole apparatusis flexible by reason of its three setsof joints and the thin, flexible bones comprisingthe horns. Anteriorly, the paraglossal- 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 97

pg-bh h - tmcgimhgo bhPg

/cc

eb-

Figure 12. Tongue skeleton of L. v. virens in ventral view labelled as a guide to identification of the osteologicalfeatures in Plates 21 and 22. basihyalhinge allowsthe two paraglossaliato be depressedas a unit. They pivot on the lateral and anteriorsurfaces of the tip of the basihyalewhen the tendon of m. ceratoglossus,inserting on the tubercle on the lateral surface of the paraglossale,is under tension. This same hinge allows the two paraglossaliato be turned as a unit to either side. It is thus closeto a uni- versaljoint. The anteriorends of the pairedceratobranchialia articulate with the posterolateralcorners of the basihyalein ball and socketjoints. These joints allow the ceratobranchialiato move upward, downwardand mediad, and laterad. Betweenthe ceratobranchialeand the epibranchialeis a single joint allowingthe latter to bend dorsad. The urohyaleis fused onto the posteriortip of the basihyale. As statedin the sectionon methodsand materials,the combinedlengths of basihyaleplus urohyalewere held constantin the drawingsof the tongue skeletonsfor all formsbecause it was reasonedthat the length of this composite bone probably would changeless during evolutionthan the lengths of the paraglossaleor the hyoid horn (ceratobranchialeand epibranchiale). The drawingsmade in this manner point out two features (see Pls. 21 and 22). The first is that the epibranchialeis relativelylonger in virensand sagittirostristhan in any of the other forms. The secondis that the para- glossaliaof the two creepers(mana and newtoni) are relativelylonger than in the other species. 98 ORNITHOLOGICAL MONOGRAPHS NO. 15

The first feature appearsto be found then in the birds with long, tubular tongues;it may be notedthat the epibranchialiaof coccinea,a specieswhose tongueis tubular in its distal half, are shorter than those of the subgenus Viridonia but are no shorterthan thoseof mana and are longer than those of newtoni. The extra lengthof thesebones, especially in Viridonia,perhaps is correlatedwith nectarfeeding since it would allow the tongueto be pro- tractedfurther (a longerhorn allowsa longerm. branchiomandibulariswhich allows further shorteningand hence protraction of the tongue). Greater protractionis a useful adaptationto birds which take, or whoseprogenitors took, nectarfrom lobeliablossoms with deep corollas. It may also be useful in capturingsmall insects from deepcrevices. The fact that the paraglossalia are relatively shorter in these subgenera,Viridonia and Loxops, than in Paroreomyzaperhaps also may be correlatedwith presentday or ancestral nectarivorousfeeding behavior. Short paraglossaliawould give less support, and thus more flexibility, to the attenuated,resilient distal end of a tubular tongue,hence perhaps facilitating the insinuationof such a tongue into the farthestcomers of a lobelia nectary. The secondfeature, long paraglossalia,is found in the subgenusParoreo- myza which has a relativelyshorter, open, forked tongue,presumably more useful in snaringinsects than a brush-tippedtubular one. These relatively longerbones would give to this tonguemore supportand hencestiffness, a featureseemingly advantageous to the captureof someof its prey with the tongue. The lateral surface of the anterior end of the ceratobranchiale has an elongated,shallow concavity in all speciesof Loxops. A similar, but more pronouncedcavity is found in the ceratobranchialeof carduelinesand in someother groupsof finches (Eber, 1956). Other differencesin shapesand sizes of tuberclesand muscle scars are discussedin the sectionon the tongue musculature.

MUSCULATURE OF THE TONGUE The studiesof Moller (1931:116-120, 131-134), Edgeworth (1935:58- 60, 97, 107, 109, 158, 203-205, 274-288), Engels (1938:642-649), and Bock and Shear (ms.) were usedas guidesto the identificationand analysis of the tongue musculature. The nomenclatureof Bock and Shear (ms. b) hasbeen followed; their systemfollows that of Engels(1938) and of George (1962) quite closelywhich in turn is basedlargely on the studiesof Gadow (1891-1893). Arrangementof musclesinto groupsis basedupon the system of Engelswho classifiedthese muscles into setsaccording to their ontogenetic developmentas shownby Edgeworth(1935). The descriptionsof the tongue musclesuse the same systemas for the jaw muscleswith the abbreviations 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 99 givenin parenthesesfollowing the name of the muscle. Note that the name m. thyreohyoideushas been retained and usedeven though a thyroidcartilage does not exist in birds. Innervationsof the tonguemuscles were not traced in this study because the homologiesbetween the tonguemuscles in Loxops and thosein other passerinebirds could be establishedeasily on other comparativemorpho- logicalcriteria. The generalpattern and identificationof tongue musclesare presented in text Figure 13. Details of the skeletalelements and musculatureof the tongueare illustratedin Plate 22 for sagittirostris;Plate 23 for virens;Plate 24 for mana; Plate 25 for newtoni; and Plate 26 for coccinea. As for the jaw muscles,reference to theseplates will not be repeatedin the descriptions of the individual musclesas it would be overly redundant and of little advantageto the reader. A) M. mylohyoideus (m. mylohy.; m m h): A transverse,thin, sheet-like muscle layer that extends between the mandibular rami with a raphe along the median line between the two halves. It lies immediately deep to the skin of the throat region except laterally where it is covered by the m. branchiomand. It is thin and translucent posteriorly where it covers (lies ventral to) the medioanterior tip of the m. serpihy. Some variation may exist at this point in that the m. serpihy.may cover (lie ventral to) the medioposteriortip of the m. mylohy. Origin: Fleshy along a long, thin line from the medial surface of the mandible, deep (dorsal) to the origin of the m. branchiomand.,from a point near the mandibular symphysisposteriorly to the area of origin of the m. branchiomand. (posterior head). Insertion: On a roedial raphe betweenthe two halves of the muscle. Comparison: The m. mylohy. is a parallel-fibered muscular sheet. It is very thin and containsrelatively few fibers (one shouldnot be fooled by the apparentlarge size of this muscle when viewed from below); hence it is a rather weak muscle. Its fibers are quite long, their effective length being the distance between the mandibular rami or twice the length of the muscle from origin to insertion. In both coccinea and newtoni, this muscle extends further posteriorly and superficially to the ventral surface of the m. serpihy. than in virens and tnana. It appears to be a thicker sheetin coccineaand mana than in the other two forms. Unfortunately, it was destroyedin the availablespecimens of sagittirostris. Action: Contraction of the two halves of this muscle would aid in constricting the lumen of the buccal cavity by elevating the floor of the mouth, and thereby providing a firm base along which the tongue can move back and forth. B) M. ceratohyoideus (m. ceratohy.; m ch): A ribbon-like muscle with the two halvesforming a "V" with its apex pointing anteriad. It lies deep (dorsal) to the m. serpihy. Origin: Fleshy from a tiny concavityin the roedialsurface of the distal tip of the ceratobranchiale,immediately anterior to its articulationwith the epibranchiale. The origin of the m. ceratohy. is encasedby the m. branchiomand. and is covered dorsally by the originating end of the m. ceratoglos. Insertion: Each half of the muscle inserts on the median line with its contralateral mate. This line of insertion is attached to the roedial line of the dorsal surface of the m. serpihy. and of the m. mylohy. 100 ORNITHOLOGICAL MONOGRAPHS NO. 15

Comparison: The m. ceratohy. is a parallel-fibered, strap-like muscle. Again, the two halves constitutemorphologically and functionally a single muscle more than forming two separate muscles. The m. ceratohy. is a thin muscle, hence it contains few fibers and is relatively weak. Its fibers are long, extending the length of the muscle as required for the relatively great displacementundergone as it contracts. Little difference exists in the structure of this muscle among the forms studied. Action: Contraction of the two halves of this muscle would prevent the two hyoid horns from separatingwhen the tongue is protruded by the pull of the m. branchiomand.; its medially directed force component opposesthe laterally directed force component of the m. branchiomand. C) M. stylohyoideus (m. styloh.; mst h): A long, thin, flat, tapering muscle that extends from the retroarticular process to the basihyale. It lies anterior to the m. serpihy. and superficial to the m. branchiomand.,with its posteromedialedge bound to the anterolateraledge of the m. serpihy. At its origin, fibers of the m. stylohy. may blend with fibers of the inserting end of the m. depr. mand. "b" and "c." Origin: By a tough aponeurosisover the lateral surface and ventral edge of the retroarticular processand a flat tendon immediately anterior to the insertion of the m. depr. mand. In sagittirostris,rnana, and newtoni, the inserting end of the m. depr. mand. is also covered by an aponeurosis. Insertion: Fleshy on a small area on the lateral surface of the basihyale near its dorsal edge and halfway between its anterior and posterior tips; this insertion lies anterior to the insertion of the m. tracheohy. and dorsal to the insertion of the m. hypoglos.obl. Comparison: The m. stylohy. is a strap-like, parallel-fibered muscle with quite long fibers. Becauseof the small total number of fibers, the m. stylohy. is a relatively weak muscle, but one that can be stretched over a long distance and can shorten a relatively large absolute length. This muscle is approximately the same in all forms except for the more extensive origins noted for sagittirostris, mana and newtoni. Action: Bilateral contraction of this muscle would aid in retracting the tongue by pulling backward on the basihyale, this being its major function. The m. stylohy. is the major retractor of the tongue. If the tongue is held in place by the m. branchiomand., the m. stylohy. would aid in enlarging the lumen of the buccal cavity by depressing the floor of the mouth. Unilateral contraction would deflect the corneous tongue to the ipsilateral side when the paraglossal-basihyalarticulation is stabilized. D) M. serpihyoideus(m. serpihy.; ms h): A broad, flat, ribbon-like muscle, forming with its mate a "V" with the apex pointing forward. It lies posterior to the m. mylohy. with its anterior end dorsal to the posteriortip of the m. mylohy. The tips of these muscleslie in close contact and their fibers may be blended with one another. The anterolateral edge of the m. serpihy. is strongly appressedto the posteromedial edge of the m. stylohy. Origin: By a small, fleshy head from the base of the cranium immediately medial to the retroarticular process. Although it was not clearly observed, this muscle presumably originatesfrom the anterior edge of the occipital plate (exoccipital process) closeto the lateral processof the basitemporalplate. Insertion: On a medial raphe with its mate; this raphe attaches to that of the m. ceratohy. Comparison: The m. serpihy. is a strap-like, parallel-fibered muscle with relatively long fibers. Its small physiologicalcross-section indicates that this muscle can develop only little force, while the long fibers allow the muscle to stretch and shorten over a great distance. 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 101

Little describable variation exists in this muscle throughout the series of forms. Action: The paired mm. serpihy. forms a loop around the pair of hyoid horns; contraction of this muscle would maintain the correct position of the hyoid horns as the tongue is protracted and retracted. In particular, this loop of the mm. serpihy. would prevent the hyoid horns from spreadinglaterally and ventrally as they are pulled forward by the m. branchiomand. Bilateral contraction of this muscle would aid in constricting the lumen of the buccal cavity by pulling the ceratobranchialia dorsally and by elevating the posterior part of the floor of the mouth. E) M. branchiomandibularis (m. branchiomand.; m b m): This largest of all tongue muscles is entwined about the epibranchiale and distal part of the ceratobranchiale from which it extends to the mandibular ramus. It lies dorsal to the mm. serpihy. and stylohy., and ventral to the m. mylohy. at the origin of the latter muscle from the mandibular ramus. The m. branchiomand. is divided into two parts, an anterior and a posterior portion, over most of its length; these parts have distinct origins which are separated by a clear gap in several forms, notably coccinea. Origin: By two large fleshy heads from the medial surface of the mandibular ramus ventral in position to the origin of the m. mylohy. The origin extends posteriorly from a point approximately adjacent to the beginning of the rhamphotheca to a point adjacent to the insertion of the m. add. mand. ext. vent. In newtoni, the m. branchiomand. post. is attached also by a thin, strap-like aponeurosisunder the ventral edge of the mandibular ramus. In mana and coccinea, this aponeurosis extends even further to cover that part of the inserting end of the m. pter. vent. lat. attached to the ventral edge of the mandibular ramus. The originating portion of this muscle was destroyed in sagittirostris. Insertion: After entwining around the epibranchiale, both parts of this muscle insert along its distal end up to the free tip of the epibranchiale. Comparison: Although partly distorted by its twisting around the hyoid horn, the m. branchiomand. is essentially a parallel-fibered muscle. It is a large muscle in total mass, but much of this mass is dependent upon the length of the fibers which are perhaps the longest in any of the tongue muscles. However, the number of fibers in this muscle is large and may well be the largest of any of the tongue muscles. The length of the hyoid horns is directly related to the length of the m. branchiomand., and hence to the distance that the tongue can be protruded from the mouth. The m. branchiomand. is larger and stronger in mana and coccinea than in virens and newtoni. Action: Bilateral contraction of this muscle would protract the tongue by causing the muscle-encasedepibranchiale to slide forward in the slippery pocket of superficial fascia located posterior to the m. depr. mand. "c" and lateral to the m. complexus of the neck. Unilateral contraction would bring about protraction of the whole tongue apparatus toward the contralateral side. F) M. genioglossus(m. genioglos.;m g g): A thin, stringy muscle extending from the mandibular symphysis posterior underneath the mucosa of the mouth and dorsal to the m. mylohy. This muscle divides near its posterior end into two slips of insertion, an anterior one and a posterior one. The m. genioglos. ant. is medial to the m. genioglos. post. Origin: Fleshy by a small head from the posterior surface of the mandibular symphysis just lateral to the midline. Insertion: The m. genioglos.post. appears to insert in a fleshy manner underneath the oral mucosajust medial to the internal and medial mandibular salivary(?) glands. Possibly and quite likely, this muscle continues to the tongue skeleton and inserts on the mucosaanterior to the cricoid cartilage. The m. genioglos.ant. inserts in a fleshy 102 ORNITHOLOGICAL MONOGRAPHS NO. 15 manner on the dorsal surface of the posterolateraledge of the paraglossaleposterior to the paraglossal-basihyaljoint. Comparison: The m. genioglos.is a parallel-fiberedmuscle with few, long fibers;hence it is a weak muscle that can shorten over a great distance. The relative size of the m. genioglos.post. from largest to smallestis virens, rnana, newtoni and sagittirostris (equal), and coccinea. The shriveled appearance of this muscle in sagittirostrismay be due to shrinkage of the specimen which was obtained in 1901. The range in relative size from largest to smallest of the m. genioglos.ant. is rnana, newtoni and coccinea (all equal), virens, and sagittirostris. Again, this muscle in the last species may have been shrunken. Action: The action of this very thin muscle is hard to deduce because of its presumed small force development. One suggestionadvanced by Bock and Shear (ms) is that this pair of musclesacts as a set of guides along which the tongue slips as it is protruded. G) M. ceratoglossus(m. ceratoglos.;m c g): This is a long muscleextending from the body of the ceratobranchiale to the paraglossale with a tough, cord-like tendon extending from a large aponeurosis on the belly of the muscle. The belly of the muscle is shaped like the hull of a dugout canoe lying on its side with the cavity opening toward the medial line and the paper-thin "bottom" facing anterolaterally. The ceratobranchiale is encased by the "gunwales" and "bottom" of the canoe. Origin: Fleshy from the dorsal surface of the ceratobranchialefrom its distal tip to its proximal end, and from the lateral and dorsal surfaces of the joint between the ceratobranchiale and basihyale. The ventral "gunwale" attached along the medial edge of the ceratobranchiale, and in coccinea, the dorsal "gunwale" also attached to the anterior third of the medial edge of the ceratobranchiale. Insertion: By a strong tendon to the bony tubercle on the ventrolateral side of the paraglossaleanterior to the paraglossal-basihyalhinge. Comparison: The m. ceratoglos.is an unipinnate muscle with relatively short fibers. Although it is a small muscle in total mass, the m. ceratoglos. contains a large number of fibers so that it may equal or exceed the strength of the m. branchiomand. Certainly it is one of the strongestmembers of the tongue muscles. This muscle shows little variation throughout the forms with the exceptions of a stronger origin in coccinea and being longer and narrower in sagittirostris. The last exception, however, may be due to shrinkage in the specimen. The tubercles of insertion are relatively largest in virens and mana, intermediate in size in sagittirostrisand coccinea, and smallest in newtoni. These differences may not be significant. Action: Bilateral contraction of this muscle brings about depressionof the tongue since its insertion lies in front of the paraglossal-basihyaljoint. Unilateral contraction would cause the tongue to be depressed on the ipsilateral side. This muscle need shorten only slightly to accomplishits action--hence its short fibers. But it may have to depressthe tongue against resistanceand therefore must develop considerableforce --hence the large physiological cross-section. H) M. hypoglossusanterior (m. hypoglos. ant.; mhga): This muscle per se is absent in all forms of Loxops dissectedby Richards; however, there appear to be some remaining nonmuscular vestiges. In virens, sagittirostris and mana, there are more or lesspaired bundlesof whitish fibers on the undersideof the tongue deep to the corneous layer and to a layer of cartilage-like material. These bundles originate between the ventral edges of the paraglossalia anterior to the paraglossal-basihyaljoint and extend anteriorly almost to the tip of the corneous tongue. Skeletal muscle striations were not visible in this tissue when viewed under a compound microscope. These bundles 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 103 appear to be composedof fibrous connectivetissue that may representthe connective tissue stroma of the almost completely vestigial m. hypoglos. ant. The vestigial nature of the m. hypoglos. ant. in Loxops agrees with the condition of this muscle in finches (George, 1962:31) and in the cardueline finch Hesperiphona (Bock and Shear, ms), and in the drepanidine Ciridops (Bock, 1972). I) M. hypoglossusobliquus (m. hypoglos. obl.; mhgo): This is a short, chunIcy muscle extending from each side of the basihyale to the underside of the posterior end of the paraglossale. Its fibers are transverse,running at right angles to the longitudinal axis of the tongue. Origin: Fleshy from the whole of the flat, lateral surface of the basihyale ventral to the insertionsof the mm. stylohy. and thyreohy., and from the anteroventral surface of the bulge in the basihyalejust anterior to its joint with the ceratobranchiale.Anteriorly, the ventral edge of the basihyal is crossedby a thin layer of transversemuscle fibers extending from one muscle to its mate. Insertion: Fleshy into a cup-like (in some forms) depressionin the ventral surface of the posterior end of the paraglossale. Comparison: The m. hypoglos.obl. is a parallel-fibered musclewith a rather large number of fibers that are short for a tongue muscle. Its total number of fibers is relatively large and probably exceededonly by the m. branchiomand.and m. ceratoglos. There is little describablevariation in size and shape of this muscle throughout the series of forms dissected. The muscle is chunkier in virens and mana, and slenderer in sagittirostris, coccinea and newtoni. The lateral surface of the basihyale is relatively largest in area in sagittirostrisand virens,is intermediatein area in mana, and is smallestin area in coccineaand newtoni. The area of insertionof this muscleis a cup-like, oblong depressionin sagittirostrg• and virens, is more or less flat and oblong in coccineaand mana, and is long, narrow and flat in newtoni. These differencesin the skeletonare probablynot significant. Action: Bilateral contraction of this muscle depressesthe posterior ends of the paraglossalia,thus elevating the anterior end of the tongue. Unilateral contraction would probably elevate and turn the tip of the tongue toward the ipsilateral side. J) M. tracheohyoideus• (m. tracheohy.; mtr h): This musclelies along the ventrolateral surface of the trachea just lateral to the m. tracheolateralis. Its insertion on the cricoid cartilage lies between and is partly obscuredby the two heads of origin

z The medial muscleslying along the trachea and extending from the crlcoid cartrilage to the basihyaleposed considerabledifficulties to Richards during his dissectionsand the writing of his original thesis description. These difficulties involved the m. tracheolateralis, the m. tracheohyoideus and the m. thyreohyoideus, and arose because of vaguenessin the description of these muscles by Engels (1938). Description and discussion of these muscles in Richards' thesis (1957: 181-183) were only partly correct. Fortunately the figures of the tongue musclesdrawn by Richards from his dissections are excellent and allow conclusiveidentification of every muscle based upon comparison with the tongue muscles of other passerine birds (Bock and Shear, ms. b; Bock, 1972). Descriptionsof these musclespresented herein are based partly upon information gleaned directly from the thesis and from Richards dissectionnotes, and partly upon comparative inference from our knowledge of these muscles in other passerine birds. The m. tracheolateralis(m tr 1) is usually grouped with the syringeal muscles and was not included in the dissectionsof the tongue muscles, but it is shown in the illustrations. It lies along the ventral surface of the trachea, medial to the m. tracheohyoideus and the origin of the ventral head of the m. thyreohyoideus. It appears similar in all respectsto the m. tracheolateralisin other passerinebirds (Bock and Shear, ms. b). 104 ORNITHOLOGICAL MONOGRAPHS NO. 15 of the m. thyreohyoideus. Some fibers of the m. tracheohyoideusmay continue into the dorsal head of origin of the thyreohyoideus,as is typical for many passerinebirds, but this possibility could not be ascertained from the dissection notes. Origin: The origin of the m. tracheohy.is posteriorto the cut by which the head was severedfrom the body in preparation for dissection. In all passerinebirds, and presumably also Loxops, this muscle originates from the sternum as the m. sterno- hyoideus with an intermediate attachment to the skin of the neck close to the midpoint of the muscle (Bock and Shear, ms. b). Insertion: The major part of this muscle inserts in a fleshy manner on the ventrolateral surfaceof the cricoidcartilage between the two headsof origin of the m. thyreohyoideus. Presumablya few fibers (5% of the total number) continue into the dorsal head of the m. thyreohyoideusand insert with this muscle onto the basihyale. Comparison: The m. tracheohy. is a strap-like, parallel-fibered muscle with few, very long fibers. It is, hence, relatively weak but would be able to shorten over a long distance. Little or no variation exists in this muscle between the forms studied. Action: Bilateral contractionwould aid in retracting the tongue by pulling posteriorly on the cricoid cartilage. It would also retract the anterior end of the trachea. K) M. thyreohyoideus(m. thyreohy.; m th h): A flat muscle extending from the cricoid cartilage,dorsal to the proximal end of the ceratobranchiale,to the basihyale. Origin: From the cricoid cartilage by two separate heads. The large ventral head arises from the ventral surface of this cartilage between the insertions of the m. tracheohy. and the m. tracheol. The small dorsal head arises from the lateral edge of the cricoid cartilagedorsal to the insertionof the m. tracheohy.;this head receives a small number of fibers from the m. tracheohy. Insertion: Fleshy onto the lateral surface of the basihyaleposterior to the insertion of the m. stylohy. and deep (medial) to the body of the m. hypoglos. obl. Comparison: The m. thyreohy. is a strap-like, parallel-fibered muscle, having about the same number of fibers as the m. tracheohy., but the fibers are much shorter, being only the length of the gap between the cricoid cartilage and the basihyale (midpoint of each element). Little or no variation exists between the forms studied. Action: Bilateral contractionretracts the basihyaleand hence the corneoustongue with respect to the cricoid cartilage. Unilateral contraction could bend the corneous tongue slightly toward the ipsilateral side. The main function of this muscle is apparently to retract the tongue and thus close the gap between the tongue and larynx (anterior end of the trachea) that may widen as the tongue is protruded.

FUNCTIONAL INTERACTIONS OF THE TONGUE MUSCULATURE The above descriptionsof the tongue musclesand their actionsdemon- stratesthe large number of potentialsynergistic and antagonisticactions of thesemuscles regulating normal tongue movements (Fig. 13 and Table 5).

Figure 13. Schematicdrawings of the skull, tongue and tongue musclesof L. v. virens in ventral view as a guide to the morphologyand functional propertiesof the tongue muscles. A) The two muscles that serve as slings to maintain the proper positionof the tongue apparatusrelative to the jaws. B) The four musclesthat serve to protract and retract the tongue (indicated by the arrows) relative to the skull. C) The four muscles that serve to move parts of the tongue relative to one another (indicatedby the arrows) but not to the skull. D) The two musclesin lateral view that can raise (mhg o) or lower (m c g) the anterior part of the corneous tongue relative to the basihyaleand the rest of the tongue apparatus. 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 105

Msh

•/M trh Mbm• 106 ORNITHOLOGICAL MONOGRAPHS NO. 15

TABLE 5 FUNCTIONS OF THE TONGUE MUSCLES •

Functions

Lateral Tilt Modify Protract Retract rotation corneous Support oral Muscle s tongue tongue tongue tongue tongue cavity Special m.m.h. x x(c) 3 m. c. h. m. st. h. x(w) • x x(e) 3 m. s.h. x x(c) • m.b.m. x x x(e) • m. gg. m.c.g x x(d? II1ohgo Oo x x(u) 5 m. tr. h. x(w) 4 m. th. h. x(c t) •

See text for complete statements of functions of these muscles in Loxops. See text or appendix 1 for abbreviations. Either constricts (c) or expands (e) lumen of mouth; the m. st. h. acting together with m. b. m. Either retracts the whole tongue apparatus (w) or only the corneous tongue (c t). Either tilts the corneous tongue up (u) or down (d) about the basihyale.

Consideringthe rapidity and subtle variation in the contractionof these muscles,the delicatedegree of neededcontrol, and the small size of the whole tongueapparatus, it is unlikely that sufficientlyaccurate studies of the functionsand biologicalroles of these muscleswill be made in the immediatefuture. We would like to outline only the broadestactions of the tongueapparatus and of the underlyingmuscular control. The functionalproperties of the tonguemuscles, either acting singly or in sets,can be comprehendedwith greaterease if the musclesare arranged into three groups(Fig. 13). The first set comprisesthe musclesextending from one side of the skull to the other and forming a ventral sling for the tongue;these include the m. mylohy.and the m. serpihy. The secondgroup comprisesthe extrinsictongue muscles that extendfrom the skull or other part of the bodyto the tongueapparatus (including the headof the trachea) and can move the whole tonguerelative to surroundingfeatures; these include the m. branchiomand.,m. stylohy.,m. genioglos.and m. tracheohy. The third set comprisesthe intrinsictongue muscles that extend from one part of the tongueapparatus to anotherand hencecan only moveparts of the tonguerelative to each other;these include m. ceratohy.,m. ceratoglos., m. hypoglos.obl., and m. thyreohy. The other intrinsicmuscle in the pas- serinetongue apparatus, the m. hypoglos.ant., is absentin Loxops. When consideringdiverse actions of the tonguemuscles, it is necessaryto distinguishbetween movement of the whole tonguerelative to surrounding structuresand movementof parts of the tongue apparatusrelative to one another. This is done most easily by examiningmovement of the center of 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 107 massof the whole tongueand movementof the separatecenters of mass of its individualparts. Only the extrinsicmuscles can move the center of mass of the entire tongueapparatus relative to the skull, be it protraction,retraction or lateral bending. Intrinsic musclescan move the corneoustongue, for example,in varied ways, but cannotshift the entire tonguerelative to the head. Unlessthis distinctionis kept in mind, incorrectfunctions are easily ascribedto individualmuscles because of the complexityof the whole system. The followingmovements and other actionsof the tongueapparatus are consideredmainly with respectto thoseof the corneoustongue. All movementsof the corneoustongue will be describedas the movementof its anterior tip relativeto other features. Theseactions are summarizedin Table 5. a) Protractionof the tongueis achievedby bilateralcontraction of the m. branchiomand.,which is the only tongueprotractor; we doubt that the m. genioglos.contributes any force to protraction. The latter muscleaids protractionby providinga set of guidesalong which the tongue"runs" as it movesforward. Possiblythis musclealso reducesany slack in the mucosa of the mouth floor and the tonguebase during protraction. Unilateralcontraction of the m. branchiomand.would protrudethe tongue, pull it toward the ipsilateralside and rotate it toward the contralateralside of the head. This rotationis probablya minor componentof the total side- ward movementof the corneoustongue. b) Retractionof the tongueis achievedby bilateral contraction of the m. stylohy., an extrinsic musclewhich is the main tongue retractor. Unilateral contractionof the m. stylohy.would retract the tongue,pull it to the ipsilateralside and probablyrotate it toward the contralateralside. Con- traction of the m. tracheohy.would retract the head of the trachea, and in combinationwith the m. thyreohy.would retract the tongue;the two muscles would act as an extrinsicmuscle. Contractionof the m. thyreohy. alone would retract the corneoustongue toward the head of the trachea,i.e., reduce the gap betweenthese structures. c) Lateral rotationof the tongueis under the controlof several muscles;this rotation may be of the entire tongue apparatusor only of the corneoustongue via movementof the supportingparaglossalia and in some casesalso of the basihyale.Unilateral contractionof the m. branchiomand. and of the m. stylohy.would rotate the whole tongue toward the contralateral side, but quite likely these extrinsicmuscles have a minor role in lateral movementof the corneoustongue. Such movementsare probably under the control of severalintrinsic musclesfor the most part. The m. thyreohy.could rotate the corneoustongue toward the ipsilatera!side by unilateral contractionwithout elevation or depressionof the tongue tip. Likewise,unilateral contraction of eitherthe m. ceratog!os.or the m. hypoglos. obl. will rotate the tonguetoward the ipsilateralside but with depressionor 108 ORNITHOLOGICAL MONOGRAPHS NO. 15 elevation,respectively, of the tonguetip. Simultaneousunilateral contraction of thesemuscles will rotate the corneoustongue toward the ipsilateralside in a horizontalplane. Varying the forcesproduced by these muscleswould alter the degreeof elevationor depressionof the tonguetip accompanying lateral rotation. d) The wholetongue apparatus may be raisedsomewhat while it is beingprotracted by the m. branchiomand.Elevation of the tonguetip is achievedby bilateralcontraction of the m. hypoglos.obl. e) Depressionof the whole tonguewould occur while it is being retractedby bilateral contractionof the m. stylohy. Combined action of the m. tracheohy.and the m. thyreohy.may alsodepress the tongue. Pas- sive depressionwould be causedby the weight of food and other objects held within the mouthcavity. The tip of the corneoustongue is depressed relativeto the rest of the tongueby bilateralcontraction of the m. ceratoglos. f) The corneoustongue (and its foundationof the paraglossalia) can be stabilizedin the sagittalplane by simultaneousbilateral contraction of the m. ceratoglos.and the m. hypoglos.obl.; thisaction fixes the basihyal- paraglossalioint. Stabilityof the basihyalemay be achievedby contraction of the m. ceratohy.and possiblyof the m. serpihy.which would hold the posterior(free) endof theurohyale against the ventralsurface of the tracheal head. This actionmight be aidedby the m. thyreohy. Fixation of the basihyale might be essentialfor fine movementsof the tonguetip which appear to be controlledmainly by the m. ceratoglos.and the m. hypoglos.obl. which move the paraglossaliarelative to the basihyale. g) The floor of the buccalcavity can be depressedactively, therebyenlarging the lumenof the mouth,by bilateralcontraction of the m. stylohy.when it retractsthe tongueand possiblyby contractionof the m. branchiomand.which would force the hyoid horns againstthe mouth floor; however,such action is probablylimited in extent and weak. Strongsimul- tanconsbilateral contractionof the m. branchiomand.and the m. stylohy. would presumablyarch the hyoid hornsstrongly against the buccalfloor as the main active method of enlargingthe mouth cavity. Passivedepression of the buccalfloor would be achievedby the weightof food or other objects held in the mouth. h) The buccalfloor can be raised, decreasingthe lurnen of the mouthby bilateralcontraction of the m. mylohy.and the m. serpihy. These transversemuscles form slingsbeneath the whole tongue apparatus,sup- portingit or raisingit alongwith the floor of the mouth.

FUNCTIONAL AND ADAPTIVE SIGNIFICANCE OF THE TONGUE Although the avian bill is frequentlyregarded as the primary food- obtainingorgan, the tongueis equallyimportant especially in the final stages 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 109 of handlingor conveyingfood particlesinto the mouth and esophagus.In Loxops, this is especiallytrue when the bird is ingestingliquid foods, including water, nectar or fruit juices, or is capturing and ingestingsmall arthropod prey from crevicesin plant structures. We shall considerthe probablemethod of taking up nectarfrom a lobelia blossomby virenssince such a maneuverprobably would make use of all of the functionsof the tonguemusculature just described.Capture of small arthropodspresumably wouldinvolve most of the samemovements of the tonguebut not the utiliza- tion of capillary action of liquids in the terminal bristlesof the tongue or the creation of a partial vacuum in the buccal cavity. The bird approachesa lobelia blossomand probesits closedbill through the slit in the upper surfaceof the corolla base, and then forces it through the tissuesof the baseof the staminalcolumn into the partly coveredspacious nectary. The bill is opened--probablyjust enoughto emit the tongue--and the tip of the tongueis protrudedout of the mouthinto the nectaryby means of bilateralcontraction of the m. branchiomand.The positionof the brush tip of the tonguewithin the nectarycan be minutelyand accuratelyadjusted by the bird to insure that nectar from all parts of the nectary is taken up and that the nectaryis thus emptiedefficiently at the first visit to that par- ticularblossom. These movements are achievedby the complexinteractions of the tonguemuscle functions described above which need not be repeated here. At this time, the tip of the tongueis swishingaround in the nectarand is thoroughlywetted with nectar and possiblyalso with saliva. The mech- anismby which nectarmoves through the tubular tongueand into the mouth is still unclear. Several workers (Gadow, 1883:69; Amadon, 1950:223) have suggestedthat capillary action allows nectar to ascendthe tubular tongue;the bristlesof the brush-liketip and the laciniae of the tube may be essentialto the capillarymechanism. Water will rise to the posteriorposterior end of a tubular drepanididtongue when it is held verticallyand its tip insertedinto the fluid. But capillary action alone appearsto be too slow a processfor the bird to rely upon exclusivelyfor the efficient,rapid obtaining of nectar. Presumablythe nectar is either suckedinto the mouth through the tubulartongue in muchthe sameway that a personsucks fluids through a straw, or is drawn into the mouth by a rapid seriesof protractionsand retractionsof the tongueanalogous to the lappingof water by a cat or dog. Suckingactions would dependupon the creationof a partial vacuumin the buccalcavity. The bird couldcreate a vacuumby depressingthe mouthfloor, but a partial vacuumwould be formed only if all openingsinto the mouth were sealed. Unfortunately,no mechanisms,either a flap or a muscular value, are known which could seal the internal nares. Nor is it known how the spacebetween the corneoustongue and the horny bill can be closed. 110 ORNITHOLOGICAL MONOGRAPHS NO. 15

Anotherpossible mechanism may existto createthe necessarypartial vacuum. A slightconcavity is often seen in the flat dorsal surfaceat the posterior end of the corneoustongue just behind the posterioropening of the tubular portion. The lateral edgesof this flat area are supportedby the posterior processesof the paraglossalia.A vacuumcould be formed if the bird pressed this flat surfaceof the corneoustongue against the roof of the mouth and increasedthe sizeof the centraldepression. Enlargement of this cavitycould be achievedby elevatingthe posteriorprocesses of the paraglossaliarelative to the basihyalewhich is a function of the M. ceratoglos.Possibly, the basihyale could be lowered,but the necessarymuscular action is difficult to com- prehend. Lateral spreadingof the posteriorends of the paraglossaliawould also serveto increasethe size of the central cavity. With the creationof a vacuumin this small space,nectar would be pulled up through the tubular tongueand would fill this cavity. Releaseof the vacuum by depressionof the posteriorprocesses of the paraglossaliavia contractionof the M. hypoglos. obl. would allow the nectar to flow into the mouth cavity and be swallowed. This mechanismdoes not requiresealing the entire mouthcavity and yet may provideenough suction to empty a singlenectary of its nectar. Nectar could be obtained by repeated protractions--retractionsof the corneoustongue which would draw a small amount of nectar into the mouth in eachcycle. Sucha mechanismwould not, however,depend upon a tubular tongueand might be more effectivewith a different configurationof the corneoustongue. Available knowledgeof the structureof the tongue and surroundingfeatures is inadequateto permit further analysisof the mech- anismswhereby these birds conveynectar into their mouth cavity. Equally little is known about other nectarivorouspasserine birds. Nectar collectedin the mouth cavity is presumablyswallowed by tilting the head backward as a bird does when swallowingwater. The bird may collectnectar from severalflowers before swallowing the accumulatedamount. It shouldbe rememberedthat all of thesecomplicated actions take place extremelyrapidly. Probablythese actions account for the observationsby Baldwin (1953:311 ), mentionedabove, of swallowingmovements in virens, of the smackingtogether of the rhamphothecaof Vestiariawhile feedingfrom Ohia blossomsnoted by Richards,and of the rapid dartingin and out of the tonguefrom the bill in Drepanis [unerea while taking water from wet moss as noted by Perkins (1895: 126-127). In summary,then, it can be saidthat the corneoustongue in the members of this genusare of two general types with that of the subgenusLoxops being intermediatein shape. In Viridonia, of which virens at least is nectarfeeding,the tongueis long, slender,decurved and tubular with a thickly tufted tip and laciniatededges to the halvesof the tube. In the insect-eating creepersof the subgenusParoreomyza, it is shorter,broader, straighter and 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 1 II shallowlytrough-like. The tip is forked and bristly,and laciniaeare present only near the tip. In the skeleton,the epibranchialiaare relativelylonger in Viridonia than in Paroreomyzaand more or less intermediatein length in the subgenusLoxops; the m. branchiomand.would be correspondinglylonger in Viridonia. This feature gives "added reach" to the tongue apparatusso that the corneoustongue can be protrudedfurther in Viridonia. The paraglossaliaare shorter in Viridonia and Loxops than in Paroreomyza,thus givingless support but more flexibilityto the corneoustongue. The tongue of the creepers(Paroreomyza), being used to capture larger insectsfrom crevices,would have to be more rigid with a firmer basal support. These differencesin the shapeof the corneoustongue and relative sizes of segmentsof the tongueskeleton probably are correlated,for reasonsdis- cussedabove, with presumedancestral nectarivorous feeding behavior or continuedpartial nectar-feedingin the subgeneraViridonia and Loxops (and in the ancestorof the entiregenus), and with insectivorousfeeding behavior in Paroreomyza.The tonguemusculature appears to vary little throughout the genus;presumably it is adaptedto either type of feeding behavior as the neededmotions of the tongueare similar and requiredforces are much the same. The major differencesmay be in the length of the m. branchiomand. and possiblyin the m. stylohy. The major evolutionarychanges, therefore,were in the corneoustongue, the tongueskeleton and the feeding behaviorof the severalspecies.

DISCUSSION

FEEDING ADAPTATIONS In the severalprevious chapters, a comparativesummary of the functional propertiesand of plausibleadaptive significances was given for each of the severalmorphological systems contributing to the feeding apparatus. The disjointednature of thesesummaries does not provide a sufficientbasis for understandingthe adaptationof the entire feedingapparatus for the feeding habits in each speciesof Loxops, and most importantly,for comprehending the adaptivehistory of the genusLoxops. The latter is essentialfor any future analysisof the evolutionand classificationof the Drepanididae,including considerationsof its origin from some mainland group within the New World nine-primariedoscines. Feeding adaptations: The morphologicaladaptations for feeding observedin the severaltaxa of Loxops cannotbe comprehendedonly in terms of the food objectsof thesebirds becauseall forms feed upon insectsand occasionallyupon nectar. Although some size differenceexists between the insectprey of the severalspecies of Loxops, thesedifferences cannot account for the obviousmorphological differences between these birds. Of far greater 112 ORNITHOLOGICAL MONOGRAPHS NO. 15 importanceare the locationof the insectprey in the environmentand the variousmethods used to obtain theseinsects by the severaltaxa of Loxops investigatedin this study. These feedingmethods may be regardedas behavioral adaptationsfor obtaininginsects under different circumstances(surface of leaves,crevices in bark, within leaf buds and koa seed pods, and behind the basesof Freycinetialeaves), and the morphologicaladaptations evolvedin conjunctionwith the differentfeeding methods. In the terminology advocatedby Bock and yon Wahlert (1965), the selectionforces associated with the evolutionof feedingadaptations in Loxopsdepend upon the insect prey plus their exactposition in the environment.Feeding on the samein- sects found on surfaces of leaves, in bark crevices and within leaf buds or koa pods would result in different selectionforces. The feedingbehaviors usedby the severalspecies of Loxopsto obtain insectsfound in theseseveral parts of the environmentwould constitutethe biologicalroles of the feeding apparatus.These biological roles depend upon the forms and functionsof the featuresof the feedingapparatus; in this studywe are concernedwith the morphologicalfeatures of the feeding apparatus. Hence the adaptationof thesemorphological features of the feedingapparatus depends upon the in- teraction(synerg) between the selectionforce (the insectprey and its location in the environment) of the environmentand the biological roles (feeding methods)of the birds. If any of thesefactors are modified,then the adaptations would be subjectto modification. Even if the birds feed upon the samespecies of insects,but upon individualanimals found in a differentpart of the environment(e.g., the insectsfound in crevicesin the bark of trees rather than on the surfaceof leaves), then the selectionforce would change and could result in the evolutionof new morphologicaladaptations. In this case,the evolutionarychange would be discussedin termsof the birdsusing a new sourceof food. If the insectprey and its locationin the environment remain the same, but with the birds modifying their feeding methods,thus establishinga new relationship(synerg) with the environmentalfactor, this circumstancecould result in different selectionforces acting on the birds. For example,we couldimagine that insteadof prying open leaf buds or koa seedpods by directthrusts of the bill, the bird might gapeand twist its bill to force openplant structuresand exposehidden insects. Such evolutionary changesare often discussedin terms of adaptationsfor greater efficiency. The environmentalfactors have not changed,but the animal has modified the way in whichit interactswith the sameenvironmental factor and in doing so, often modifiesthe selectionforce acting upon it. New morphological adaptationscould result in responseto thesenew selectionforces. The morphologicaladaptations for feedingin the severaltaxa of Loxops providemany excellentexamples supporting the conceptsof evolutionary adaptationand associatednotions advocated by Bock and yon Wahlert 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 113

(1965). Therefore,particular attentionis given to exact location of food sourcesand feedingmethods used to obtain food and to correlationof the morphologicalfeatures to thesefactors. As far as possible,the individual morphologicalfeatures will be discussedas part of a larger integratedfunc- tional unit. A) Loxopsvirens virens: The Amakihi possessesthe slenderestbill of all speciesin the genus,and in many ways representsthe most generalized member of Loxops sinceit has the broadestrange of feedinghabits in the genus. Moreover, it may well representthe most primitive speciesin the genus. The followingremarks are basedupon the HawaiianAmakihi (L. v. virens) but probablyhold true for all other subspeciesexcept for the Kauai Island form, L. v. stejnegeri,which has a larger and stronglydecurved bill, and almostcertainly has a differentmethod of feeding. However, the following analysisof the adaptationof the feeding apparatusin L. v. virens possiblyholds true for the Kauai Island L. parva. The externalmorphology of the bill of this speciesis very similarto that seenin mostpopulations of L. virens. The Hawaiian Amakihi feeds mainly on insectsand small arthropods, preferringsoft-bodied, flightless forms between1.5 and 10 mm in length. Larger, heavy-bodiedinsects are not taken. The bird hunts for the insects amongleaves, blossoms, twigs and smallerbranches, and lessfrequently (less than one-third) on larger branchesand trunks of trees. Insectsare taken mainlyfrom the surfaceof leavesby gleaning,but alsofrom betweenleaves and massesof lichensby probing. All parts of leavesand twigs are visited with the body held in all possiblepositions although some positions, such as verticallywith headdown, were observedinfrequently. Nectar is the second item of food but is taken far less frequentlythan insects. The flat broad blossomsof $ophora and Metrosiderosare visitedmost frequentlywith the birds approachingthe flower from the side to avoid the long stamens.The long tubularlobelia flowers are piercednear the baseof the corollasto reach the nectaries. Becauseof the rapid movementsof the birds while feeding, no useful observationscould be made on actions of the tongue. The last and least importantfood item is the juice of fruits; so infrequentlyis this food item taken that it is presumedto lack a significantinfluence on the evolution of feedingadaptations in this species. The bill of virensis of mediumlength for the genus,but slenderand most decurved. The bony jaws are correspondinglyfrail as is the skull. In all featuresof the cranialosteology, virens is eitherthe mostlightly constructed memberof the genusor only slightlysturdier than newtoni. The light con- structionof the skull is closelyparalleled by its small jaw muscles.For most individual jaw musclesor sets of jaw muscles,virens possesseseither the weakest or second weakest muscles. 114 ORNITHOLOGICAL MONOGRAPHS NO. 15

The tongue of virens is long, slender,somewhat decurved and tubular with the edgesand tip shreddedinto a fringe. The tonguefits into a well developedroedial slot boundedby two parallelridges in the horny palate of the upper jaw. The epibranchialiaare relativelylong in this specieswith a longm. branchiomandibularisand probably a longm. stylohyoideus(although relativelengths of this musclewere difficult to ascertain). The paraglossalia are short,allowing a more flexibletongue. The slenderbill with light skull and weak jaw musclesare all adaptations for gleaningsmall insectsfrom the surface of leaves. No great amount of muscularforce is required to capture the insects,hence small musclesare sufficient,and the skullneed not be strengthened.Indeed, it shouldbe noted that larger musclesand stouterbones than requiredwould be a disadvantage for a bird feedingon small insectsbecause of increasedweight and increased metabolicenergy required to maintain and operatethese larger structures. A slightlydecurved bill as comparedto a straightbill may or may not be an advantagefor leaf-gleaners,but it is not a disadvantage.The terminal fringeof the tonguemay be advantageousin brushingup small insects. The slightdecurvature of the bill is a definiteadaptation for nectar-feeding. The bird can probe into shallowblossoms from the side and therebyavoid the long stamensthat would interfere with the bird if it approachedthese flowers from above. Tubular flowers are pierced from the side near their basesand the tongueinserted between the tips of the gapedjaws. Piercing the thin floral tissuesdoes not require great force, and the strengthof the jaws and their musculatureis sufficientfor thesepiercing and gapingactions. The slightlydecurved bill permitsthe bird to reach the nectar in nectaries openedat the top while probinginto the flower from the side. The length of the bill and of the tongue are likewise advantagesfor nectar-feedingas are the tubular nature of the tongueand its terminal fringe, and the roedial slotin the hornypalate. Long epibranchialiaand branchiomandibularismus- clespermit the tongueto be protrudedfrom the mouth more than in other membersof the genusand hencemore readily reach the furthestnectaries. The shortparaglossalia allow greaterflexibility of the tongue. Thus, the shapeof the bill and tongue,the lightly-builtskull and weak muscles,long tubular fringed tonguewith long epibranchialiaand branchiomandibularismuscles are adaptationsfor gleaningsmall insects from the surfacesof leavesand twigs and for obtainingnectar from broad, flat blossoms of Metrosideros and tubular lobella flowers. B) Loxops sagittirostris:The food, feeding habits and morphologyof the Greater Amakihi are poorly known becauseof the extinctionof this speciesaround 1900, but sufficientinformation is availableto allow some conclusionson the adaptationsof its feedingapparatus. The main food items of the Greater Amakihi are larger insectssuch as 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 115 forestcrickets of the genusParatrigonidium, caterpillars, spiders and a carabid beetle. Most of these arthropodsare taken from their hiding placesat the base of Ieie leaves, beneath loose bark of large Ohia trees, ferns and similar hiding places. Their prey is obtainedby strongprobing and prying with the bill. A few observationswere made of this speciesfeeding at blossomsof the Ohia, so that it probablyobtained some nectar in its feeding; however,nectar was probably a minor food item. The long stout bill of sagittirostrisis adaptedto the larger insectsthat form a major part of its food and to the strongprobing and pryingrequired to securethese insectsfrom their hiding placesat the basesof leaves. Its straightbill is advantageousin feeding methodsthat result in large forces actingbackwards on the tip of the bill. The only jaw muscle that could be describedin this genus is the m. depressormandibulae which was much larger than that in other members of the genus. A correspondinglylong retroarticularprocess exists in this species.These featuresare presumablyadaptations for strong gaping, a feedingmethod that would be expectedin a bird that obtainsinsects from the areabetween leaf basesand stem,beneath loose pieces of bark and other similar places. The long, slender,decurved tubular tonguewith a terminalfringe that was observedin sagittirostrisis presumablyan adaptationfor nectar feeding. So are the long epibranchialiaand branchiomandibularismuscles and shortparaglossalia. These agree with the few observationsof the Greater Amakihi feedingat blossomsand suggestthat nectarmay havebeen a moreimportant part of its diet than suspected.Possibly sagittirostris relied on nectar in its total diet to the same extent as did virens. Thus,the long, stout,straight bill, large m. depressormandibulae and long retroarticularprocess of the Greater Amakihi are adaptationsfor obtaining larger insectsfrom crevicesat the basesof leavesand etc. by strongprobing, prying, and presumablygaping. The long tubular tonguewith its terminal fringe,long epibranchialiaand branchiomandibularismuscles and shortparaglossalia are adaptationsfor feedingon nectarin a mannersimilar to that used by virens. C) Loxops maculatamarta: The Hawaiian Creeper feeds mostly on arthropodsand other invertebrates,taking caterpillars,spiders and small moths,with larger moths,slugs, and beetlessometimes eaten. These prey itemsare taken on the trunks and large branchesof foresttrees, either from the surfaceor generallyfrom crevicesin the bark by vigorousprobing, pecking and pryingwith the bill. The bird climbsup and downthe trunk and creeps on the upper and lower surfacesof horizontallimbs, searchingdead wood andbeneath the bark for caterpillarsand otherinsects. The HawaiianCreeper thus fills the ecologicalposition of nuthatches(Sitta) and even more closely, 116 ORNITHOLOGICAL MONOGRAPHS NO. 15 that of the Black and White Warbler (Mniotilta varia, Parulidae) in the Hawaiian forests. Few insectsare taken in flight. The Hawaiian Creeper apparentlynever feedson nectar. The skull and jaws of mana are stoutlydeveloped with the rhamphotheca of both jaws extendingstraight out to end in an extremelyacute angle. The ruggedconstruction of the bill is shownby the greaterheight and width of the horny ridge betweenthe nasal operculae. The straightbill is clearly adaptedfor large posteriorlydirected forces acting on the tip of the bill. Not only are the bonesof the skull stout, but the braincaseis bulbous,a conditionsimilar to that seenin woodpeckersand other peckingbirds. The bulbouscranium may serveto protectthe brain and/or provide a broad base for the bill whichwould be advantageousin withstandinglateral forcesacting on the tip of the bill. Certainlynot all blowswould be placeddirectly on the bark but would glanceoff, resultingin lateral forcesacting on the bill tips. Moreover,prying actionsof the bill would place stronglateral forces on the bill tip which could be resistedbest if the bill had a broad base at the quadratearticulation and at the nasal-frontalhinge. A bulbouscranium would provide the desiredwidth. The jaw musclesof mana are somewhatstronger than thoseof newtoni and virens,but are still considerablyweaker than thoseof coccinea.Although individualmuscles in mana are the smallestor the largestin the genus,no subsetof jaw musclesin thisspecies is exceptionallywell or poorlydeveloped. Rather the entire musculatureis uniformly better developedin comparison to that of virensto providethe force it requiresin vigorousprobing and prying for insectprey. The short, broad and rather straighttongue with its stiff and forked tip is clearlyadapted for insectfeeding. The long paraglossaliaprovides support for the tonguerequired in strongprobing for insects.Although the epibranchialiaare short,associated with reducedrequirement to probe long distances,the branchiomandibularismuscules are larger than in other members of the genus,which is an adaptationfor the larger force neededto probe stronglywith the tongueand to spearinsects with it. The shallowtrough and only slightlycurled edges are associatedwith the lack of nectarfeeding (hencelack of any needfor a tubular tongue) and with the need for a stiff flattenedtongue that can be insertedinto crevicesfor insects. Thus the strong, straight bill, stout skull with its bulbous braincase, strongerjaw muscles,strong straight tongue with its stiff, forked tip and strongbranchiomandibularis muscles are all adaptationsfor uncoveringand capturinginsects hidden in crevicesin the bark of trunksand largerbranches. The absenceof nectarin the diet of the HawaiianCreeper precludes the need for compromisein morphologicaladaptations for nectar feedingand for insectfeeding by probing,especially in the structureof the tongue. 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 117

D) Loxopsmaculata newtoni: The food and feedinghabits of the Maui Creeper differ little from those of the Hawaiian Creeper; indeed these taxa are far more similar to one another than are any of the other speciesof Loxopsincluded in this study. A few importantdifferences do exist between thesecreepers. The Maui Creeperspends much time amongleaves of trees and lichensalthough it can creepas well as the Hawaiian Creeper,which was never observedby Richardsto feed among leaves. It appearsto be more of an inhabitantof scruband undergrowth,although it still spendsmuch time on the trunks and branches. The food items are almost the same as those taken by marta;no reportsexist suggesting that newtonitakes smaller insects. On rare occasions,the Maui Creepertakes nectar from blossomsof Merto- sideros,a habit shared with the Lanai Creeper, montana',and the Kauai Creeper,bairdi; however,nectar is a very minor item in the diet of all of thesecreepers. As for the Hawaiian Creeper,the bill of newtoniis straight,but it is more decurved and weaker than the bill of marta. Associated with its weak bill is its frail skull, the frailest of thosestudied, and smallerjaw muscles(compared to mana), which are about the same size and strengthas those in virens. Presumably,this creepercannot probe and pry as vigorouslyas mana and spendsmore time gleaningits arthropodprey from the surfaceof leaves and off the bark than probingin crevices. If membersof this race are able to obtain sufficientfood by gleaningand weak probing,then the frail skull and weak jaw musclesare an adaptationin that they can be lighter in weight and require less energy. Unfortunately,we do not have sufficientobserva- tionson the size of the insectprey taken by newtonicompared to that eaten by rnanaor of the exact feedingmethods, both of which would be essential to judge the adaptivenessof the frail skull and weak muscles. The slightly curved bill may be associatedwith greater gleaning rather than vigorousprobing and prying. The short, straight tongue with its stiff, forked tip and lack of curled edgesare adaptationsfor insect eating as in rnana. As in the Hawaiian Creeper, the epibranchialiaare shorter and the paraglossalialonger, for in- creasedsupport of the tongue,and the m. branchiomandibularisstrong, but smaller in newtoni than in rnana. A medial slot with parallel lateral ridges is found in the horny palate; the tonguefits into this groove which is an adaptationfor the rare taking of nectar. This arrangementwould permit formation of a tube for transportationof nectar without compromisingthe adaptationsof the tonguefor insect feeding. Thus the straight,pointed bill and straighttongue with a stiff tip of the Maui Creeperare adaptationsfor capturinginsects by probingand prying. The frail skull and weak jaw musclesappear to be adaptationsfor weak probing,greater gleaning of insectsfrom leavesand bark and taking smaller 118 ORNITHOLOGICAL MONOGRAPHS NO. 15 insects.The medialgroove of the hornypalate is an adaptationfor feeding on nectar without modificationof the tongue,which would have reducedits degreeof adaptationfor insect-feedingby probing. The major aspectsin the adaptationof the feedingapparatus of newtoniare associatedwith weaker probingand greateramount of surfacegleaning in its feedingmethods. E) Loxops c. coccinea: The Hawaiian Akepa, which may be quite ac- curatelycalled the Hawaiiancrossbill, is the mostspecialized member of the genuswith mostof the morphologicaladaptations of the feedingapparatus closelyassociated with the twistingmotions of its head and beak during feeding. The food of the Hawaiian Akepa is mainly caterpillars,other insectsand small spiders.These arthropods are taken either from the surfaceof leaves and bark, or from within leaf buds, leavesfastened together by the larval insectsor koa seedpods. Richardsobserved the Akepa feedingonly in Ohias (Metrosideros)and koas (Acacia Koa), and other workersreport that this speciesis verypartial to thesetrees. The birdswere observedfeeding chiefly in the foliageand leaf budsor on the koa podsand more rarely on branches. While feeding,the bird held its body in almostevery imagineableposition and couldbe observedwith its head bent so that the bird's chin restedagainst the anteriorpart of its neck and upperbreast. Unfortunatelydetails of its feedingmethods could not be observed. The Akepa has beenreported to feed on nectarrarely, althoughthis was not observedby Richards. Becauseof the observedmorphology of the feedingapparatus, we presume that the feedingmethods used by coccineainvolve a twistingmotion of the headand beakwhile the tips of bothjaws are insertedinto the plant material, either in leaf buds or koa seedpods. The essentialconsequence of these twistingactions is a pair of laterally directedforces acting in oppositedirec- tionson the tips of the upperand lowerjaws. However,we muststress that the detailsof the feedingmethods presumably used by the Akepa as presented above are deductionsbased upon the morphologyof the feeding apparatusand are not basedupon observedfacts. The major justification that may be offeredis that the entirehypothesis is plausibleand internally consistent. The Akepais a smallspecies with a shortbill resemblingthat seen in many smallunspecialized finches such as siskinswith no unusualfeature noticeable in lateral view. However, in dorsal or ventral views, the tips of the two rhamphothecacan be seento crossone another, the degreeof crossingbeing minoras comparedto that seenin crossbills(Loxia). Both right-billedand left-billedindividuals are found; our descriptionswill be for the more common right-billedindividuals. The lateral edgeof both the maxillaryand mandibularrhamphothecae on the crossedside forms a sharpcutting edge while 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 119 the oppositeedge is rounded. Thesecutting edges of the lower and upper jaws are forcedagainst the plant tissueas the bird twistsits head;the sharp edgeis an adaptationto increasethe force with which the rhamphothecae cut into the plant tissue. In overallshape, the jaws are short and relativelywide, providinggreater leveragefor the jaw musclesand strengtheningthe bill againstthe strong lateral forcesacting on the tips of the mandible. The skull of coccineais short and is the strongestof those in all the taxa studiedas are its jaw muscles;both the strengthof the bone and of the large musclesare direct adaptationsto the strongforces acting on the jaws. However,the mostimportant aspect of theseforces is that they act laterally and asymmetricallyon the tips of the jaws. A force directedtoward the bird'sleft actson the right lateral edgeof the mandibulartip in a right-billed bird, while a force directed toward the bird's right acts on the left lateral edge of the maxillary tip; theseforces are equal in magnitude. Hence, the mandibleis pushedtoward the left and the maxilla toward the right. The forcesdeveloped by the jaw musclesmust opposethese external forces and their consequences. Most of the braincase,jaws and bony palate are symmetricalin coccinea with no hint of the asymmetricalrhamphothecae reflected in the bony jaws. Somedifferences exist in the sizeand shapeof the temporaland subtemporal fossa,in the zygomaticprocess between them and in the transpalatineprocess; these differencesare associatedwith asymmetriesin the jaw muscles. The mostimportant asymmetries exist in the configurationof the jaw articulations. Thesefeatures are adaptationsto permit a slightlateral rotationof the mandible to the right with the mandibularhalves of the jaw articulationsliding and rotatinglaterally around the quadratecondyles. A slightdegree of lateral rotationis an adaptationin the suggestedmethod for openingkoa pods in which actual movementof the mandibleis an advantageand in all feeding methodsas a means of reducing undesirablestrong shocks. Reduction of theseshocks can be achievedby small lateral rotation of the mandible which could lengthenimpact time. A mandiblerigid in terms of lateral rotation would be subjectedto shorterimpacts and larger shocks. The jaw musclesof the Akepa are not only larger, both absolutelyand relatively, than in the other speciesof Loxops includedin this study, but asymmetricallydeveloped. The degreeof asymmetryin the musclesis con- siderablygreater than in the rhamphothecaor in the skull and is readily apparenteven in a quick superficialexamination of the head. Moreover, the pattern of asymmetryin the jaw musclesis a clear adaptationto the pattern of externalforces placed on the tips of both jaws. Examinationof the relative sizes of the muscleson the right and left sides of the head revealsthat those muscleswith a direct dorsal pull on the mandiblesthe 120 ORNITHOLOGICAL MONOGRAPHS NO. 15 m. pseudotemporalissuperficialis and m. adductormandibulae posterior--are symmetrical. Those dorsal muscleswhich have a laterally directed force componentpulling the mandibleto the right while adductingit--most of the parts of the m. adductormandibulae externus--are larger on the right side of the head. Lastly, ventralmuscles such as the m. pseudotemporalispro- fundusand mostof the m. pterygoideuswhich have a mediallydirected force componentpulling the mandibleto the right while adductingit are larger on the left side of the head. Hence all of the enlargedjaw muscleshave hori- zontalforce components,which would rotate the tip of the mandiblelaterally to the right or would resista stronglateral force actingtoward the left on the right edge of the mandibulartip. Moreover, the retractorsof the palate are on the left side of the head and would result in a strongerposteriorly directedforce at the left ventroposteriorcorner of the upperjaw. This larger force is in a proper positionto counteracta lateral force actingto the right on the left edgeof the maxilla and henceto reducepart of the stressplaced on the nasal-frontalhinge. Lastly, it shouldbe emphasizedthat the asymmetrical jaw musclespose no problem when the Akepa wishes to open or close its jaws when no external lateral forces are acting on the jaws. At these times, the bird needs only to contract its musclesdifferentially so as to developequal amountsof force on both sidesof the head. The morphologyof the tongue in the Akepa does not show any clear adaptationsto the feedingmethods of this bird on insectshidden within leaf budsor koa seedpods with the possibleexception of its length and frilled tip which may be used to ensnaresmall insects. Rather, the length of the tongue,tubular configuration,frilled tip and large m. branchiomandibularis are all adaptationsfor nectar feeding. The importanceof these adaptations are difficult to assess.Possibly they may be holdoversfrom an early stage in the evolutionof the genuswhen nectar was an essentialpart of the diet and maintainedbecause the Akepa still feeds occasionallyupon nectar and becausethese nectar adaptedfeatures of the tonguemorphology do not conflict with the particular insectivorousfeeding methods of this species.Yet it is possible,and even probable, that nectar-feedingand the associated adaptationsof the tongue are essentialfor the survival of this species. Even if nectar is taken relatively rarely, it may provide an important source of food that could be used when the insectson which the Akepa feeds are in short supply. Although the Akepa has the advantageover other members of the Drepanididaein utilizing a food sourcethat is unavailableto them, it may have a disadvantageif its food supplyvaries greatly during the year. Both leaf buds of Metrosiderosand seed pods of Acacia Koa would be in variable abundanceduring the year, althoughprobably reaching their peak supplyat differenttimes. The availabilityof anotherfood source--nectar-- and especiallyone that existsat a different time from either insectsin leaf 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 121 buds or seed pods would be a definite advantageespecially if the needed adaptationsfor nectarfeeding and insecteating do not conflict. All of these conditionsexist in coccinea.(It may be noted, in passing,that thesecondi- tions do not hold for the creeperswhich feed on a sourceof insectsthat would be relativelyconstant during the entire year and which necessitates particular adaptationsof the tongue that conflict directly with adaptations required for nectar feeding. Hence the creepershad to abandonor were able to abandonnectar in their evolutionfor feeding on insectsin bark crevicesbecause this food supplywas dependableduring the entire year.) Thus, the short bill, with crossedrhamphothecal tips having opposing sharp lateral edges,the asymmetricaljaw articulationsand jaw musculature of the Akepa are adaptationsassociated with stronglateral forceson the jaw tips and slightlateral rotation of the mandibleinvolved in capturinginsects hidden in leaf buds and koa pods. The stout skull and strongjaw muscles are adaptationsfor the large forcesneeded for thesefeeding methods. More- over, it is reasonableto predict that associatedadaptations exist in the cervical musclesof this speciesbecause of the twistingof the head relative to the body. Featuresof the tonguesuch as its tufted tip, tubular anterior half and length are adaptationsfor nectarfeeding.

ADAPTATION AND EXTINCTION Becauseof the widespreadextinction of many taxa of drepanidids,the questionof the degree of adaptationof these birds to their environmentis an importantquestion. Adaptationof a speciesto its environmentinvolves all featuresof the organismand its entire habitat. Extinction can occur if the adaptiverelationship between the organismand its environmentbreaks down for only one facet of its entire niche, if this factor happensto be essentialfor the survival of the individual or successfulreproduction. In consideringthe adaptationsfor feedingin Loxops,we are confidentthat all of the known taxa in this genusare well adaptedto their environmentand that decreasein their populationsize or their extinctionwas not a result of a failure in any aspectsof the adaptationof thesebirds to the food gathering part of their niche. Rather we feel that reductionin the numbersor extinction of the severaltaxa of Loxops was due to a lack of adaptationof these birds to a new factor in their environment,namely the appearanceof several diseases(bird pox and malaria), the result, in part, of the introductionby man into the HawaiianIslands of the necessaryinsect vectors (Warner, 1968). The birds were unable to adapt to this new factor of their environmentand becamereduced in numbersor extinct in spite of their continuedadaptation to probablyall other parts of their niche includingfood gathering.Although reductionor extinctionof any living speciesof organismsis a tragedy,careful analysisof the factorspresumably responsible for the demiseof each species 122 ORNITHOLOGICAL MONOGRAPHS NO. 15 is importantfor a clarificationof the importantbut little understoodevolu- tionary mechanismof extinction.

EVOLUTIONARYHISTORY OF THE GENUS Loxops Any considerationsof the origin of and the probableevolutionary trends in the genusLoxops dependgreatly upon an understandingof the evolutionaryhistory of the entirefamily. Sucha discussionis not possibleat this time becauseof our still inadequateknowledge of the relationshipsand evolution within the New World nine-primariedoscines and of the possibleorigin of the Hawaiianhoneycreepers from this complex. Even if we had a thorough knowledgeof the morphologyand evolutionof the feedingapparatus in the Drepanididae,the necessarycomparative information from the nine-primaried complexis not available. Yet, it seemsreasonable to offer somethoughts on the possibleevolutionary trends in Loxopsso that we may presentour ideas on the origin and evolutionof the Drepanididae(Fig. 14). This presenta- tion will be brief because our ideas are based more on intuition than factual evidence. Ever sincethe family Drepanididaewas describedby Gadow, thesebirds were believedto be relatedto the New World nine-primariedoscines. Gadow ( 1891, 1899) regardedthe Coerebidaeas the ancestralstock of the Hawaiian honeycreepers.Most workershave acceptedthis idea or the relativelyminor modificationthat the Hawaiian honeycreepersarose from thin-billed members of the Thraupidae (Amadon, 1950: 231-233, 253; Baldwin, 1953: 386-388; Beecher, 1953: 312-313, 324). Much of the evidence cited as supportfor this conclusionmust be re-evaluatedcarefully, but to do this would lead too far from the purposeof this discussion.However, one bit of evidenceis most importantand cannot simply be brushedaside, namely the presenceof a tubular, nectar-adaptedtongue in most membersof the family includingthose speciesthat apparentlynever take nectar; Amadon (1950) quite rightly stressesthis point. If the Drepanididaeevolved from nectar-feedingancestors (e.g., Coerebidae),then a tubulartongue would be a primitive feature of the family (present in the original stock of birds colonizingthe Hawaiian Islands) and hencemay still be presentin species that no longerdrink nectar. If the Drepanididaeevolved from someother nine-primariedstock, such as finches,then it is almostmandatory to postulate an early nectar-feedingstage in their evolutionaryhistory with the evolution of a tubular tongueand to arguethat all known membersof the family de- scendedfrom this nectar-adaptedform possessinga tubular tongue. Hence the tubular tonguewould be a primitive feature for the known membersof the Drepanididae,but not an ancestralfeature inherited from its mainland ancestors. The secondmajor conclusionon the origin of the Drepanididaestems from 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 123

•œ.parva œ.vlren$•• lœ. $a•ittirostri $ • • œ.maculata

• / ,,•to Hemignathu$ toother •/ J' andother Drepanidinae• •j9------•Ancestrol œoxop$Psittirostrinae H/ma#one•-- ' ( viren$- like)

? / Drepanidinae•Psittirostrinae C/r/dop$-Iike (with tubular tongue) I Originofthe Drepanididae C/Yidop$-Iike (without tubular tongue) I HawaiianIslandscolonized Cardueline ancestor (Fr/ng///a - l i ke) Figure 14. Dendrogramsummarizing our ideas of the origin and evolution of the Drepanididaeand of the genus Loxops. Only the portion involving Loxops is supported by the featuresanalyzed in this study. Arrangementof the specieswithin Loxops is for convenienceonly and does not necessarilyreflect an evolutionaryorder.

Sushkin(1929) who suggestedthat thesebirds evolvedfrom the cardueline finches. This idea has attractedless attention in the past, but it is becoming more widely accepted.Although we shall baseour discussionon the conclusionthat the Hawaiian honeycreepersevolved from carduelinefinches or from their immediateancestors, we wish to emphasizethe dangersof its wholeheartedacceptance. Much of the citedsupporting evidence for thisidea 124 ORNITHOLOGICAL MONOGRAPHS NO. 15 is really no better than the evidencesupporting the Coerebidae-Thraupidae originsof the Drepanididae. Uncritical acceptanceof the carduelineorigin of the drepanididshas the sameserious disadvantages as had the generalac- ceptanceof coerebid-thraupidorigins of the Drepanididae. Sushkin's(1929) argumentswere basedlargely upon the close resem- blance of the heavy-billed finchlike Psittirostra to the cardueline finches. While it is true that Psittirostrais very similarto carduelinefinches in cranial osteologyand jaw musculature,this resemblancecould be due to convergence just as many points in the morphologyof nectar feedingdrepanidids and coerebidscould be due to convergence.However, other evidencedoes support the theory of carduelineorigins of the drepanididswhich we would like to summarize(see Sibley, 1970, for evidencefrom the egg-whiteproteins). The skull and jaw musclesof Psittirostra(Beecher, 1953: 311-313) are verysimilar to thoseof carduelinefinches including the presenceof the lateral flange conditionof the palatineprocess of the premaxillary (Bock, 1960b: 455). Moreover, even the thin-billed Loxops showscardueline-like features. A slightlateral flaring of the palatineprocess of the premaxillacan be seen in coccinea,and an enlargedsubtemporal fossa is presentin all speciesof Loxops. The postorbitalligament is vestigialin Loxops as in carduelines, while it is stronglydeveloped in coerebids(Bock and Morony, unpublished observations). The m. adductor mandibulae externus caudalis is divided into two distinctparts, the anterodorsalone expandinginto the enlarged subtemporalfossa. The m. pseudotemporalissuperficialis has an anteriorslip that is well developedin Psittirostrabut is alsopresent in Loxopsin a much lessdeveloped condition. Both muscleconditions are characteristicof cardueline finches(including Fringilla) but not coerebidsand most tanagers. It shouldbe noted that sometanagers such as Tanagraand Chlorophoniahave an anterior slip of the m. pseudotemporalissuperficialis as describedby Beecher(1951b: 278-279) but obscuredin his major work (Beecher, 1953: 310-311) and verified by Bock (unpublishedobservations), so that the commonpossession of this featureby carduelinesand drepanididsis not such strongevidence as it had once appeared(Bock, 1960b). The positionof the two parts of the m. branchiomandibularisin Loxops is reminiscentof that foundin finches,not coerebids.The configurationof the quadrate-mandibular hingein coccineais very similarto that in Loxia, althoughthis couldbe the result of convergence. The plumageof drepanididsis much like that found in many carduelines and also that found in many tanagers. However, we would like to note the especiallyclose similarity betweenthe plumage of the drepanididgenus Ciridopsand the carduelinegenus Leucosticte. The latter genusmay be a primitive componentof the carduelines,being far closerto Fringilla than to the more advancedgenera. 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 125

Unfortunately,little is known about the breedinghabits of the Hawaiian honeycreepers.H. B. Tordoff (personalcommunication) has informed us that his observationsof captivePsittirostra cantans breeding in his aviaries indicatedthat thesehoneycreepers are extremelysimilar to carduelinefinches in courtshipand nestingbehavior. Lastly, the chancesfor any group of songbirdsto be successfulcolonizers of remoteoceanic islands would increaseif they were birds that wandered in flocks,preferably erratically, over long distances,and bred in the region to which they had wandered(Bock, 1960b: 477; Mayr, 1965). Cardueline finchespossess these attributes. The coerebidsand thraupidsare generally sedentarytropical birds, and as suchare lessprobable colonizers of a remote oceanicisland. (This conclusionis not in disagreementwith Moynihan's conclusion,1968, that the "Coerebini"are goodcolonizers of new areasand habitatsand have spreadthroughout the islandsof the Caribbean. He has not consideredthe problem of crossinglarge water gaps which is the issue of primary concernhere.) Acceptingthe carduelinefinches as the most likely ancestralstock for the Drepanididaedoes not automaticallysolve the questionof the probablean- cestralgenus among the honeycreepers.Indeed, the latter questionbecomes more difficult. Most likely the typical advancedcardueline finches, such as Spinusor Carpodacus,did not give rise to the drepanidids.It is far more probablethat the honeycreepersarose from either the primitivemembers of the carduelinae,similar to the present-dayFringilla and Leucosticteor the birds ancestralto the carduelinegroup. On the basisof this notion and on the basisof the closesimilarity in plumagebetween Leucosticte and Ciridops anna, we believe that Ciridops may well be the closestpresent-day representativeof the primitive stockof the Drepanididae. This conclusionis con- trary to Amadon's (1950: 530) conclusionthat Ciridops is an advanced member of the Drepanidinae. Amadon'splacement of Ciridops is based uponhis conclusionthat the Hawaiianhoneycreepers evolved from coerebids or thin-billedtanagers. Unfortunately little is knownabout the morphology andhabits of thispeculiar and extinct honeycreeper so that it will be difficult, if not impossible,to obtainmuch support for our conclusion.The morphology of the tongueapparatus has beendescribed recently by Bock (1972). If the drepanididsarose from primitivecardueline finches and if Ciridops is representativeof the ancestralstock of honeycreepers,then somespecula- tionscan be offeredon the early evolutionaryhistory of the Drepanididae. Carduelinesof genera,such as Fringilla, feed on small seedsand on insects (Eber, 1956) with considerableseasonal variation in their diet. The small seedson whichthese birds feed (grassesand beech)may be relativelyrare in the habitatsavailable in Hawaiian forests. Hence, after their invasionof theHawaiian Islands, the originalcolonizers may have been restricted initially 126 ORNITHOLOGICAL MONOGRAPHS NO. 15 to insects.Exploitation of otherfood sourceswould be of definiteadvantage to thesebirds especiallyif no other avian specieswere usingthese other food items. If Hawaiianforests at the time of the drepanididinvasion were at all similar to present-dayHawaiian forests,then nectar was in abundant supplyat least during certaintimes of the year. We postulatethat the ancestral drepanididsincluded nectar in their diet and that evolution of a tubular, fringed tonguewas one of the earliest, if not the first, important adaptationin the feedingapparatus of this family. Further, it is reasonable to suggestthat all known drepanididsdescended from birds with a tubular tongue (Fig. 14). Once a tubular tongue evolved in the honeycreepers, further specializationin this directioninvolves relatively simple evolutionary changesleading to Himatione and other membersof the subfamilyDrepa- nidinae,which includes the major nectar-feeding members of the Drepanididae. Evolution from a partly insectivorous,partly nectarivorousCiridops-like bird to present-dayLoxops involves relatively minor changes. Possibly Loxops arosedirectly from a Ciridops-likeancestor or possiblyfrom a Himatione- like ancestor.Amadon (1950:231 ) regardedHimatione as the most primitive member of the Drepanidinaeand Loxops as the most primitive genus in the Psittirostrinae;we concurexcept for the placementof Ciridops. Fur- ther, he emphasizesthe closenessof Himatione and Loxops. On the basis of Amadon's conclusionand our argument (above) we concludethat the ancestorof Loxops was a thin-billed bird that fed partly upon nectar and partly upon small insectsobtained by gleaning. Evolution in Loxops and other generaof the subfamilyPsittirostrinae has beenin the directionof in- creasingspecialization for insect- and seed-feeding. Loxopsvirens is probablycloser to the ancestralstock of the genusLoxops than the otherknown members of the genus.It was,as it is now, a generalized nectar-feederand insect-gleanerand was able to spreadto most of the major islands. Certainlysome differentiation occurred on someof the islandswith double invasionstaking place from time to time, althoughthe location of thesedouble invasions is unknown. At presenttwo pairs of suchdouble in- vasionscan still be seen,namely involving parva and v. stejnegerion Kauai and v. virens and sagittirostrison Hawaii. Neither of the divergentforms, ste]negeriand sagittirostris,spread subsequently to other Hawaiian islands. However, the creepers,maculata, and the cross-bills,coccinea, which pre- sumablyarose from virens-likeancestors via doubleinvasions and specializa- tion for different methodsof obtaininginsects and reducingthe importance of nectarin theirdiet, were both able to spreadto mostof the major Hawaiian islands. It is not possibleto date the relativetimes of thesedouble invasions and subsequentspreads with any accuracy.However, on the basisof the degree of morphologicalchanges involved, we would suggestthat coccineaarose 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 127 from the virens-likeancestor first, followedby rnacuIataand last the twin double invasionsof v. virens-sagittirostrisand v. ste]negeri-parva.The result of this patternof doubleinvasions and subsequentspread of the differen- tiated form is a finer divisionof the availableinsect food sourceby several speciesof Loxopsand the presenceof three of four speciesof this genuson the largeislands of Kauai, Oaku, Maui and Hawaii (Fig. 1 ). Moreover,this patternof doubleinvasions and spreadprobably extends without a break to the evolutionof the genusHemignathus which representsanother direction of specializationfor insectprey (see Bock, 1970). Evolutionarychanges in the morphologyof the feeding apparatusthat occurredduring the evolutionof v. stejnegeri,sagittirostris, macuIata and coccineafrom virenscan be appreciatedby a comparisonof the morphologicaladaptations for feedinghabits as givenabove; these comparisons need not be repatedhere. We would like to emphasizethat thesechanges in the morphologyof the feedingapparatus can be arrangedin a seriesof gradually increasingmodifications. L. v. stejnegeriis consideredto be a geographic representativeof virens,differing in the size and shapeof the bill and pre- sumablyosteological and myologicalfeatures. L. sagittirostrisis basicallya heavy-billedvirens. L. maculatais a straight-billedvirens which creepson large branchesand trunks and probesfor insectsmore than virens. More- over, the severalsubspecies of macuIatadiffer by becomingincreasingly more creeper-likewith strongerbills, as shown by the differencebetween newtoniand mana. The greatestdifference is betweenvirens and mana. Hence, we believe that the availableevidence supports the conclusionthat the evolution of the various members of Loxops from a virens-like ancestor took placeby a smoothseries of gradualevolutionary changes. The rate of thesechanges presumably varied greatly in the severalphyletic lines and at different times in each of the lineages. The Drepanididaerepresent among living vertebrates a remarkableexample of adaptiveradiation in a shortspan of geologicaltime. As a part of this radiation,the genusLoxops provides a clear exampleof the beginningof suchan adaptiveradiation and illustratesthe basicevolutionary mechanisms involvedin adaptivemorphological changes in the feedingapparatus. Al- thoughconsiderable insight into the originand evolution of the Drepanididae hasbeen gained by thisanalysis of Loxops,these ideas must be verifiedand furtheredby studyof the wholefamily. Unfortunatelya completelysatis- fying studyis no longerpossible because of the extinctionof many important taxa in the Drepanididaeand becauseof widespreaddestruction of the originalHawaiian environmentat the handsof Europeanand Asian man. Enoughoriginal environment is still exanton the HawaiianIslands to permit valid conclusionson the adaptiveinterrelationships between the drepanidids and their habitat (see Lack, 1965), and populationsof most speciesof 128 ORNITHOLOGICAL MONOGRAPHS NO. 15 honeycreepersstill survive,but remain in dangerof extinctionthrough dis- ease (Warner, 1968). Hopefullythe mostsignificant result of this studyof the adaptationin the feedingapparatus of Loxopsis to providea stimulus for additional studieson the comparativebiology and evolution of the Hawaiianhoneycreepers and the eventualunderstanding of the classicalex- ampleof adaptiveradiation they provide.

SUMMARY

1. The genusLoxops was chosenfor this initial studyof the adaptation and evolutionof the feedingapparatus in the Drepanididaebecause it is a relativelyprimitive genus and becausemembers of the three subgeneraare still extant on the sameisland, and up to four species,including the now extinct sagittirostris,live on the island of Hawaii which was most suitable for field studies. 2. The food and feedinghabits of eachspecies could be deducedreason- ably well on the basis of informationin the literature and field work by Richards. L. v. virensfeeds on small arthropodsobtained by gleaningand upon nectar of Metrosiderosand lobelias. L. sagittirostrisfeeds on crickets and other large insectsobtained by forceful probinginto crevicessuch as the basesof Ieie leavesand less upon nectar. L. m. martafeeds upon insects obtainedby creepingalong trunks and largebranches and probingforcefully into crevices,but doesnot take nectar. L. m. newtonifeeds upon insects obtainedby creepingand weak probingor gleaning;some nectar is taken. L. c. coccinea feeds on insects concealed in Metrosideros leaf buds and in koa seedpods which are capturedby first openingthese plant structuresby twistingmovements of the head and bill; nectaris taken occasionally. 3. The rhamphothecaof the bill of virensis thin, of mediumlength and slightlydecurved, of sagittirostris,heavy, long and straight,of mana, heavy, of mediumlength and straight,of newtoni,thin, of mediumlength and very slightlydecurved, and of coccinea,stout, short and finch-likewith the tips slightlycrossed. In a right-billedbird, the tip of the mandiblecrosses to the right and the tip of the maxilla crossesto the left; left-billedbirds are the opposite.The lateraledges of boththe mandibleand maxillarytip are sharp on the crossed side. 4. The skull of virens is described in detail and that of each taxon is comparedwith it. Somenoteworthy features are the retroarticularprocess, a well developedpalatal hasp, a reducedlateral flange conditionof the palatineprocess of the premaxillary,and a well developedsubtemporal fossa. The asymmetricalfeatures in the skull of coccineaare described;the most importantof theseare the quadrate-mandibulararticulation and surrounding bones. 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 129

5. The jaw musclesare describedand comparedin detail in the several taxa of Loxopsunder study. Of interestare the structureof the m. pseudotemporalissuperficialis and the m. adductormandibulae externus caudalis which are similar to these musclesin carduelinefinches. Relative strength of the muscleswere estimatedusing a crudeindex. The functionalproperties of the jaw musclesin Loxops and their correlationswith feedinghabits are discussedincluding the correlationsbetween the osteologicalfeatures of the skull, myologicalfeatures and feeding methods. The form and functional propertiesof the asymmetricaljaw musclesof coccineaare describedand comparedto thoseof Loxia. 6. The corneoustongue is long, tubular and with a frilled tip in virens, sagittirostrisand coccinea,and shorter,flat and with a stiff forked tip in mana and newtoni. The hyoid horns are longer in those specieswith a tubular tonguewhile the paraglossaliaare longer in those specieswith a flat tongue. The tonguemuscles are describedand comparedin the several taxa, including sagittirostris,of Loxops with relatively little interspecific variation found. The functional properties of the tongue apparatus are discussedand their adaptivesignificance postulated. 7. The morphologicaladaptations in the feedingapparatus for the food and feedingmethods in each speciesare discussed,the entire feedingappa- ratusbeing considered as a singleunit. Evolutionarychange leading to the particularadaptations in the feedingapparatus of each speciesis correlated mainly with modificationsin their methods of obtaining food with little differenceexisting in the actual insectprey and other food items in virens, newtoni,mana and coccinea.Only in sagittirostris,does evolutionary change in its feedingapparatus appear to be largelyassociated with a changein food items, to larger sized insects,and only to a lesser extent with changein feeding habits. Each speciesis believed to be well adapted to the food gatheringpart of its niche; reductionin population size or extinction of severaltaxa has resultedfrom a failure to adapt to other changesin the habitat of thesespecies. The origin and evolutionof Loxopsare discussed. It is suggestedthat the evolutionof the severalspeices of Loxops with their different feeding methodshas resultedfrom a pattern of double invasion, subsequentdivergence and spread of the divergentform throughoutthe Hawaiian Islands.

LITERATURE CITED

ANON. 1961. Drepanididae and Drepanidae (Aves and Insecta); addition to the official list. Opinion 610. Bull. Zool. Nom., 18: 267-269. AMADOr, D. 1947. Ecology and the evolution of some Hawaiian birds. Evolution, 1: 63-68. AM•t)ON, D. 1950. The Hawaiian Honeycreepers (Aves, Drepaniidae). Bull. Amer. Mus. Nat. Hist., 95: 151-262. 130 ORNITHOLOGICAL MONOGRAPHS NO. 15

AMADON,D. 1960. Proposalsconcerning homonymous family-group names based on the genericnames Drepana Schrank, 1802 (Class Insecta, Order Lepidoptera), and Drepanis Temminck, 1820 (Class Aves): Requestfor the use of the plenary powers to suppressthe generic name Drepanis Brisson, 1760. Z.N.(S.) 901. Bull. Zool. Nom., 17: 220-223. BALDWIN, P. H. 1953. Annual cycle, environment and evolution in the Hawaiian Honeycreepers (Aves: Drepaniidae). Univ. California Publ. Zool., 52: 285-398. BEECriER, W. J. 1951a. Adaptations for food-getting in the American blackbirds. Auk, 68: 411-440. BEECriER,W.J. 1951b. Convergencein the Coerebidae. Wilson Bull., 63: 274-287. BEECriER,W.J. 1953. A phylogeny of the Oscines. Auk, 70: 270•333. BEEC}IER,W. J. 1962. The bio-mechanicsof the bird skull. Bull. Chicago Acad. Sci., 11: 10-33. Bo½•:, W.J. 1960a. Secondary articulation of the avian mandible. Auk, 77: 19-55. Boc•:, W.J. 1960b. The palatine process of the premaxilla in the Passeres. Bull. Mus. Comp. Zool., 122: 361-488. Boc•:, W.J. 1964. Kinetics of the avian skull. J. Morph., 114: 1-42. Bo½•:,W.J. 1966. An approachto the functional analysisof bill shape. Auk, 83: 10-51. Boc•:, W.J. 1967. The consequencesof pinnatenessof the M. pseudotemporalis superficialis in passerine birds. Amer. Zool., 7: 775-776. Boc•:, W.J. 1968. Mechanicsof one- and two-joint muscles. Amer. Mus. Novitates, no. 2319, 45 pp. Bo½•:,W.J. 1970. Microevolutionarysequences as a fundamental conceptin macro- evolutionary models. Evolution, 24: 704-722. BocK, W.J. 1972. Morphology of the tongue apparatus of Ciridops anna (Drep- anididae). Ibis, 114: 61-78. Boc•:, W.J. ms. A redescriptionof passefinejaw muscles. Boc•:, W. J., •D H. Momo•rA. ms. Functionalanalysis of passefinejaw muscles. Boc•:, W. J., Am) C. R. SHEAR. ms a. Pinnatenessof the M. pseudotemporalissuperficialis in Passer. Boc•:, W. J., ANDC. R. SI-mAR. ms b. A redescriptionof passefinetongue muscles. Boc•:, W. J., •D G. YON Wa_m•ERT. 1965. Adaptation and the form-function complex. Evolution, 19: 269-299. B•Jr•R, H. 1922. Die Bedeutungder 12berkreuzungder Schnabelspitzenbei der Gat- tung Loxia. Biol. Zbl., 42: 87-93. BOWMAN,R.I. 1961. Morphological differentiation and adaptation in the Galapagos Finches. Univ. California Publ. Zool., 58: vii q- 302. BRYn, W.A. 1908. Some birds of Molokai. Occas. Papers Bishop Mus., 4: 133- 176. BURT, W. H. 1930. Adaptive modifications in the woodpeckers. Univ. California Publ. Zool., 32: 455-524. C•SON, H. L., D. E. HAm)y, H. T. Sl,mr}I ANDW. S. STONE. 1970. The evolutionary biology of the Hawaiian Drosophilidae. Pp. 437-543 in Essays in evolution and genetics,Supp. to Evolutionarybiology (W. Steereand M. Hecht, eds.). Appleton- Century-Crofts,New York, pp. xx q- 594. DEGENER, O. 1932. Flora Hawaiiensis or new illustrated flora of the Hawaiian Is- lands. Privately printed, Honolulu, Books 1-5 (Books 1-4 published 1932-40, reprinted 1946, book 5 published 1946-1957, looseleaf.) DILGER,W. C. 1956. Relationshipsof the thrush genera Catharus and Hylocichla. Syst. Zool., 5: 174-182. 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 131

DUERST,U. 1909. PathologischeDifformation als gattungs-,art- and rassenbildender Factor. I. Mechanische, anatomischeund experimentelle Studien uber die Morpho- logie des Sch•idelsyon Angehfrigen der Gattung Loxia. Mitt. naturf. Ges. Bern, 1909: 281-303. EBER,G. 1956. VergleichendeUntersuchungen tiber die Ern•ihrung eininger Finken- v6gel. BioL Abhand., No. 13/14: 1-60. EDGEWORTh,F.H. 1935. The cranial musclesof vertebrates. CambridgeUniv. Press, London, viii q- 493 pp. E•qGErS,W.L. 1938. Tongue musculatureof passerinebirds. Auk, 55: 642-650. ENGELS,W.L. 1940. Structuraladaptations in thrashers(Mimidae: genusToxostoma) with commentson interspecific relationships. Univ. California Publ. Zool., 42: 341-400. FIEDrER, W. 1951. Beitr•ige zur Morphologie der Kiefermuskulatur der Oscines. Zool. Jb., Abt., Anat., 71: 145-288. G/•DoW, H. 1883. On the suctorial apparatusof the Tenuirostres. Proc. Zool. Soc. London, 1883: 62-69. G/•DOW,H. 1891-1893. V6gel. I. Anatomischer Theil. Bronn's Klassen und Ord- nungen des Thier-Reichs, C. F. Winter Verlagshandlung, Leipzig, vol. 6, pt. 4, 1008 pp. G/•DoW, H. 1891. Remarks on the structure of certain Hawaiian birds, with reference to their systematicposition. Pp. 219-241 in Aves Hawaiienses:the birds of the SandwichIslands (S. B. Wilson and A. H. Evans). R. H. Porter, London, xxvii q- 257 pp. G/•DoW, H. 1899. Further remarks on the relationships of the Drepanididae. pp. 243-249 in Ibid. GANS, C. 1969. Functional componentsversus mechanical units in descriptive morphology. J. Morph., 128: 365-368. GANS, C., AND W. J. BoCK. 1965. The functional significanceof muscle architecture --a theoretical analysis. Ergebn. Anat., 38: 115-142. GARDNER,L. L. 1925. The adaptive modifications and the taxonomic value of the tongue in birds. Proc. U.S. Natl. Mus., no. 2591, 67: 1-49. GEORGE,W.G. 1962. The classification of the Olive Warbler, Peucedramus taeniatus. Amer. Mus. Novitates, no. 2103, 41 pp. GREENWAY,J. C., JR. 1967. Extinct and vanishingbirds of the world. 2nd edition, Dover Publications, New York, xvi q- 520 pp. GREENWAY,J. C., JR. 1968. Family Drepanididae.Pp. 93-103 in Check-listof birds of the world, (R. A. Paynter, Jr., ed.). Harvard Univ. Press,Cambridge, vol. 14, x q- 433 pp. HARTERT,E. 1893. Note by E. Hartert. In The avifauna of Laysan and the neighbor- ing islands:with a completehistory to date of the birds of the Hawaiian possessions (W. Rothschild). Parts 1-3. R.H. Porter, London, xx, xiv q- 320 pp. HENSHAW,H.W. 1902. Birds of the Hawaiian Islands being a complete list of the birds of the Hawaiian possessionswith notes on their habits. Thomas G. Thrum, Honolulu, 146 pp. HOFER,H. 1950. Zur Morphologie der Kiefermuskulaturder V6gel. Zool. Jb., Abt. Anat., 70: 427-556. HUBER,W. 1933. Untersuchungentiber die Genese der Asymmetrien am Kopf von Loxia curvirostra. Gegenbaurs Jb., 71: 571-588. KNOPF, A. 1949. Time in earth history. Pp. 1-9 in Genetics, paleontology and evolution. (G. L. Jepsen,E. E. Mayr, and G. G. Simpson,eds.). Princeton Univer- sity Press, Princeton, xiv q- 474 pp. 132 ORNITHOLOGICAL MONOGRAPHS NO. 15

LACK, D. 1965. Evolutionary ecology. Ecol., 53: 237-245. LAK.IER,T. 1926. Studien tiber die trigeminus-versorgteKaumuskulatur der Saurop- siden. C.A. Reitzel, Copenhagen, 154 pp. Lom•Nz, K. 1949. •ber die Beziehungenzwischen Kopform und Zirkelbewegungbei Sturniden und Ikteriden. Pp. 153-157 in Ornithologie als biologischeWissenschaft (E. Mayr and E. Schuz, eds.). Carl Winter UniversitS/tsverlag,Heidelberg, xii q- 291 pp. LucAs, F.A. 1897. The tongues of birds. Rep. U.S. Natl. Mus., 1895: 1001-1019. MA¾•, E. 1965. The nature of colonizations in birds. Pp. 29-47 in The geneticsof colonizing species(H. G. Baker and G. L. Stebbins,eds.). Academic Press, New York, xv q- 588 pp. MOLI•ER, W. 1930. 13ber die Schnabel- und Zungenmechanik bltitenbesuchender Vfigel. I. Ein Beitrag zur Biologie des Blumenvogels. Biol. gen., 6: 651-726. MOL•ER, W. 1931. ldern. II. Idib., 7: 99-154. MOVaOKA,H. ms. Morphology and relationshipsof the Turdidae, Mimidae and Tro- glodytidae. UnpublishedPh.D. thesis,University of Illinois, Urbana. MOYNIHAN,M. 1968. The "Coerebini": A group of marginal areas, habitats and habits. Amer. Nat., 102: 573-581. Munro, G. C. 1944. Birds of Hawaii. Tongg Publ. Co., Honolulu, 189 pp. (reprinted, 1960, Charles E. Tuttle Co., Rutland and Tokyo, 192 pp.) OEHME, H. 1962. Das Auge von Mauersegler, Star und Amsel. J. Omith., 103: 187-212. PEVaZINS,R. C.L. 1893. Notes on collecting in Kona, Hawaii. Ibis: 101-112. PERKINS,R. C.L. 1895. Notes on some Hawaiian birds. Ibis: 117-129. PERKINS,R. C.L. 1901. An introductionto the study of the Drepanididae,a family of birds peculiar to the Hawaiian Islands. Ibis: 562-585. PERKINS,R. C.L. 1903. Vertebrata.pp. 365-466 in Fauna Hawaiiensisor the zoology of the Sandwich(Hawaiian) Isles. CambridgeUniv. Press,Cambridge, vol. 1 (4). RICHXP,•s,L. P. 1957. Functional anatomy of the head region of the Hawaiian Honeycreeper genus Loxops (Aves, Drepaniidae). Unpublished Ph.D. thesis, University of Illinois, Urbana. RICI-•RI•S, L. P., AN• P. H. BArr)W•N. 1953. Recent records of some Hawaiian Honeycreepers. Condor, 55: 221-222. ROBBINS,C.A. 1932. The advantage of crossed mandibles: a note on the American Crossbill. Auk, 49: 159-165. RocK, J.F. 1913. The indigenoustrees of the Hawaiian Islands. Privately printed, Honolulu, vii + 518 pp. ROTHSCHILD,W. 1893--1900. The avifauna of Laysan and the neighboringislands: with a completehistory to date of the birds of the Hawaiian possessions.Parts 1-3. R. H. Porter, London, xx, xiv, 320 pp. SIBrEY, C. G. 1970. A comparative study of the egg-white proteins of passerine birds. Bull. Peabody Mus. Nat. Hist., 32: vi q- 131 pp. STARCK,D. 1959. Neure Ergebnisseder vergleichendenAnatomie und ihre Bedeutung ftir die Taxonomieerl•iutert an der Trigeminus-Musculaturder Vfigel. J. Ornith., 100: 47-59. STEAmqS,H.T. 1946. Geology of the Hawaiian Islands. Terr. Hawaii, Div. Hydrog- raphy Bull., 8: 1-106. ST•ESEMAmq,E. 1927-1934. Sauropsida:Aves. In Handbuch der Zoologie. (W. G Kukenthaland T. Krumbach,eds.). Walter de Gruyter, Berlin, vol. 7, pt. 2, xii q- 899 pp. 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 133

SUSHKIN,P. P. 1929. On the systematicposition of the Drepanidae. Verhandl. VI Internatl. Ore. Congr., in 1926, Copenhagen: 379-381. TowNsoN, R. (?). 1799 (?). [Traits and (?)] Observationsin natural history and physiology. London, 232 pp. WALLACre,A.R. 1880. Island life. Macmillan and Co., London, xvii q- 526 pp. WALLACE, A.R. 1902. Island life. 3rd edition, Macmillan and Co., London, xx q- 563 pp. WALLS,G.L. 1942. The vertebrate eye and its adaptive radiation. Cranbrook Inst. Sci., Bull., 19: xv q- 785 pp. WARNER,R.E. 1968. The role of introduced diseasesin the extinction of the endemic Hawaiian avifauna. Condor, 70: 101-120. WILSON, S. B., AND A. H. EVANS. 1890-1899. Aves Hawaiienses: the birds of the Sandwich Islands. R.H. Porter, London, xxvii q- 257 pp. YnaRELL, W. 1829. On the structure of the beak and its muscles in the Crossbill (Loxia curvirostra). Zool. Journ., 4: 459-465. ZIMMERMAN, E. C. 1948. Insects of Hawaii--a manual of the insects of the Ha- waiian Islands, including an enumeration of the speciesand notes on their origin, distribution, hosts, parasites, etc. Vol. 1, Introduction. Univ. Hawaii Press, Hono- lulu, xxi q- 206 pp. 134 ORNITHOLOGICAL MONOGRAPHS NO. 15

APPENDIX I GLOSSARY OF ABBREVIATIONS Skull

Braincase

ab auditory bulla aos antoribal space bc braincase btp basitemporal plate earn external auditory meatus ef ectethmoid foramen eop exoccipital plate ep ectethmoid plate ioc intraorbital cristae ios interorbital septurn iow interorbital width jb jugal bar 1 lacrimal lbtp lateral basipterygoidprocess n-f h nasal-frontal hinge o orbit of optic foramen op occipital plate pop postorbital process por postorbital rims pps protractor pterygoid scar pts pseudotemporalswelling sf subtemporalfossa sor supraorbital rim sp suprameaticprocess sr sphenoidalrostrum tf temporal fossa zp zygomatic process zt zygomatic tubercle

Rostrum ben bony external naris bt bony tomlure db dorsal bar lb lateral bar lfc lateral flangecondition of the palatineprocess of the premaxilla mx maxilla ns nasal septurn ppm palatineprocess of the premaxilla r rostrum uj upper jaw Quadrate bq body of the quadrate jb jugal bar JPq jugal processof the quadrate lcotp lateral condyle of the otic process lcq lateral condyle of the quadrate 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 135

APPENDIX I (Continued)

mcotp medial condyle of the otic process mcq medial condyle of the quadrate mpq mandibular processof the quadrate obpq orbital processof the quadrate otpq otic processof the quadrate pcq posterior condyle of the quadrate ptcq pterygoid condyleof the quadrate q quadrate robp rugosity of the orbital process rob p tubercle of the orbital process Pterygoid p ptt protractor pterygoid tubercle pt pterygoid pt f pterygoid foot pt h pterygoid head rppt retractor processof the pterygoid Palatine ipp interpalatineprocess mpp mediopalatine process P palatine pb palatine blade ph palatine hasp pnc posterior nasal canal ppb prepalatine bar pt palatine trough tpp transpalatine process wpt wall of the palatine trough Vomer and maxillopalatine mp maxillopalatine pl mp plate of the maxillopalatine pmp pedicle of the maxillopalatine v vomer Mandible acp anterior coronoid process ag articular groove emf external mandibular fossa emp external mandibular process imf intemal mandibular fossa imp internal mandibular process im f intemomandibular flange lower jaw ma mandible mb mandibular boss mf mandibular foramen mml medial mandibular ledge ms mandibular symphysis pcp posterior comoid process ran retroarticular notch rap retroarticular process sb sesamoid bone 136 ORNITHOLOGICAL MONOGRAPHS NO. 15

APPENDIX I (Continued)

tpma tubercle of the posteriormandibular adductor tpt tubercleof the pseudotemporalls Tongue Skeleton bh basihyale cb ceratobranchiale cc cricoid cartilage eb epibranchiale i hg o insertion hypoglossusobliquus Pg paraglossale pg-bh h paraglossal-basihyalhinge tr tracheal rings tmcg tubercle of the ceratoglossus uh urohyale Jaw Muscles mameca m. adductor mandibulae externus caudalis "a" mamecb m. adductor mandibulae externus caudalis "b" mamerl m. adductor mandibulae externus rostralis lateralis

mamerm m. adductor mandibulae externus rostralis mediaIls mamert m. adductor mandibulae externus rostralis temporalis mamev m. adduuctor mandibulae extemus ventralis mamp m. adductor mandibulae posterior mdm m. depressormandibulae mppq m. protractor pterygoideiet quadrati (mppss) (m. protractorpterygoidei sensu stricto) (mpq) (m. protractor quadrati) mpsp m. pseudotemporalisprofundus mpss m. pseudotemporalissuperficialis m pt m. pterygoideus mptdl m. pterygoideusdorsalis lateralis mptdma m. pterygoideusdorsalis medialis anterior mptdmp m. pterygoideusdorsalis medialis posterior mpt r m. pterygoideusretractor mptvl m. pterygoideusventralis lateralis mptvm m. pterygoideusventralis medialis mptvmfan m. pterygoideusventralis medialis fan rn pt v rn scap m. pterygoideusventralis medialis scapularis Tongue Muscles mbm m. branchiomandibularis mcg m. ceratoglossus mch m. ceratohyoideus mgg m. genioglossus mhg a m. hypoglossusanterior mhg o m. hypoglossusobliquus mmh m. mylohyoideus msh m. serpihyoideus msth m. stylohyoideus mth h m. thyreohyoideus mtrh m. tracheohyoideus mtrl m. tracheolateralis 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 137

PLATES 138 ORNITHOLOGICAL MONOGRAPHS NO. 15

Plate I. Lateral view of the bill of Loxops. A) L. v. virens (A.M.N.H. 453127). B) L. v. stejnegeri(A.M.N.H. 453194). C) L. m. mana (A.M.N.H. 453289). D) L. m. newtoni (A.M.N.H. 453356). E) L. m. flammea (A.M.N.H. 453296). F) L. c. coccinea (A.M.N.H. 453403). G) L. c. caeruleirostris (A.M.N.H. 193452). H) L. parva (A.M.N.H. 453216). I) L sagittirostris(A.M.N.H. 453236). 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 139 140 ORNITHOLOGICAL MONOGRAPHS NO. 15

Plate 2. Dorsal and ventral view of the bill of Loxops. A) L. v. virens, dorsal (A.M.N.H. 453127). B) L. v. stejnegeri, dorsal (A.M.N.H. 453194). C) L. m. mana, dorsal (A.M.N.H. 453289). D) L. m. newtoni, dorsal (A.M.N.H. 453356). E) L. parva, dorsal (A.M.N.H. 453216). F) L. c. coccinea, dorsal (A.M.N.H. 453403). G) L. c. coccinea, ventral (A.M.N.H. 453403). H) L. c. coccinea, ventral (A.M.N.H. 168614). I) L. c. caeruleirostris, dorsal (A.M.N.H. 193452). J) L. c. caeruleirostris, ventral (A.M.N.H. 193452). K) L. sagittirostris, dorsal (A.M.N.H. 453231). 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 141

HrI'"-

PLATE 2 142 ORNITHOLOGICAL MONOGRAPHS . NO. 15

Plate 3. Skull of Loxops v. virens. A) Oblique view. B) Lateral view. C) Ventral view. 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 143

P•TE 3 144 ORNITHOLOGICAL MONOGRAPHS NO. 15

Plate 4. Skull of Loxops m. mana. A) Oblique view. B) Lateral view. C) Ventral view. 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 145

PLATE 4 146 ORNITHOLOGICAL MONOGRAPHS NO. 15

Plate 5. Skull of Loxops m. newtoni. A) Oblique view. B) Lateral view. C) Ventral view. 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 147

PLATE 148 ORNITHOLOGICAL MONOGRAPHS NO. 15

Plate 6. Skull of Loxops c. coccinea (right-billed individual). A) Oblique view. B) Lateral view. C) Ventral view. 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 149

PLATE 6 150 ORNITHOLOGICAL MONOGRAPHS NO. 15

Plate 7. Details of the skull of L. v. virens. A) Ventral view of skull. B) Lateral view of ectethmoidplate. C) Lateral view of palatine. D) Ventral view of palatine. E) Cross-sectionof palatine hasp and trough. 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 151

•bt p

Gm ef,, ep

c ,wpt pb prnp • oo lO$ lb-

mpp

D ppb

I f c p pt p m I

PLATE 7 152 ORNITHOLOGICAL MONOGRAPHS NO. 15

Plate 8. Details of the quadrate and mandible of Loxops. A) Ventral view of quadrate and surroundingbones of L. v. virens. B) Lateral view of quadrate and surrounding bones of L. v. virens. C) Dorsal view of posterior end of the mandible of L. v. virens. D) Posterior view of quadrate of L. v. virens. E) Posterior view of quadrate of L. m. mana. F) Posterior view of quadrate of L. m. newtoni. 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 153

ptc mcq ,pop z p /z t A pcqIcq B

r ppf s'.'t•-•.j••iPq

'lcøtppt f/ toby/

Ibtr •:•' p pt t q otpq c otp

c cp D

--pcp

e ill --i mf

imp

ran ag

rap

E F •pptt•

PLATE 8 154 ORNITHOLOGICAL MONOGRAPHS NO. 15

Plate 9. Details of quadrate and mandible of L. c. coccinea (right-billed individual) to show the asymmetriesin these skeletal features. A) Ventral view of posterior end of skull. B) Posterior view of quadrates. C & D) Dorsal view of posterior end of mandibular rami spacedto correspondto the position of the quadratesin B. 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 155

Left Right

c D

Left Right

PLATE 9 156 ORNITHOLOGICAL MONOGRAPHS NO. 15

Plate 10. Jaw muscles of L. v. virens. A) Oblique view. B) Lateral view of superficial muscles. C) Lateral view of deeper muscles. 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 157

PPq

rn pt ps m ptdm p• mptdma mdm mamert mptvm mamecb gland mameca mptv -mamp

mamev amerl mamerm ps p

mpsp B mamerl mamev mdm mamec mamp mameca

mptdl mptvm d ill c mpsp mpss mptvl mamp PLATE 10 158 ORNITHOLOGICAL MONOGRAPHS NO. 15

mptdma mptdl

A •\m ptdmp .:.•"'•,... mptr

m pt v m scap m ptv pt v m fan

•msh

Plate 11. Jaw musclesof L. v. virens. A) Ventral view of m. pterygoideus;ventral layer removed on left side of head. B) Ventral view of some tongue musclesand more dorsal jaw muscles.

Plate 12. Jaw muscles of L. m. mana. A) Oblique view. B) Lateral view of superficial muscles. C) Lateral view of deeper muscles. 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 159

A mptdmp • mptd m a

m ptd

raps p

ameca amecb 1laineft m amev

mareerin

m psp mamecb mameca ma mp

ptdma

mptd m ps p mpss •amp 160 ORNITHOLOGICAL MONOGRAPHS NO. 15

m ptd m m ptr mptdm a/ mptd

m pt d m m ptvm

m pt v I

mptv m fan

Plate 13. Jaw musclesof L. m. mana. A) Ventral view of m. pterygoideus;ventral layer removed on left side of head. B) Ventral view of some tongue muscles and more dorsal jaw muscles.

Plate 14. Jaw musclesof L. m. newtoni. A) Oblique view. B) Lateral view of superficial muscles. C) Lateral view of deeper muscles. 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 161

mptdm mptdma

m ptd

rn ps amert mamec •amerl

mamer B mame ameca namp

mptdm

mptdma

mptdl•

C mpsp mpss mamp 162 ORNITHOLOGICAL MONOGRAPHS NO. 15

mptdmp mptd m a,•

m ptvm

,,mpp q mp$$••• B

Plate 15. Jaw muscles of L. m. newtoni. A) Ventral view of m. pterygoideus; ventrallayer removedon left sideof head. B) Ventralview of sometongue muscles and more dorsal jaw muscles.

Plate 16. Jaw musclesof L. c. coccinea(right-billed individual). A) Lateral view of superficialmuscles of the right side. B) Lateralview of superficialmuscles of the left side. C) Lateral view of deeper musclesof the left side. 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 163

•Ila meca

m pt m ptdm m ptdm mptdl B

mpsp mdm

amev mameca ptvl

/mpsp mdm 164 ORNITHOLOGICAL MONOGRAPHS NO. 15

m pt d rn p• ,m a m e . mptvl• '•'/•• m mp•vmd m mp td ma• • ti' • • m ptr

mpt d I' mpt v I / '•• mpt vm/ '••a me-m d m

d m mptd m pt r B

mptdl

m ptdml• mdm

rn pss d 113

Ppq

mpsp mdm 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 165

mptr• amert m ptd mdm

mamecb

amerl

mamev npsp •ptdma ptdmp •/,/• md!•:•

Plate 18. Jaw muscles of L. c. coccinea and L. sagittirostris. A) Oblique view of the left-side of L. c. coccinea (right-billed individual). B) Lateral view of the hind end of the head of L. sagittirostris showing the m. depressor mandibulae.

Plate 17. Jaw muscles of L. c. coccinea (right-billed individual). A) Ventral view of the M. pterygoideus and some surrounding muscles (left-side of head uppermost). B) Ventral view of dorsal layer of M. pterygoideus (ventral layer removed on both sides). C) Ventral view of some tongue musclesand more dorsal jaw muscles, 166 ORNITHOLOGICAL MONOGRAPHS NO. 15

Plate 19. Corneoustongue of Loxops in dorsal and lateral views. A) L. v. virens. B) L. m. mana. C) L. m. newtoni. 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 167

Plate 20. Corneoustongue of Loxops in dorsal and lateral views. A) L. c. coccinea. B) L. sagittirostris. 168 ORNITHOLOGICAL MONOGRAPHS NO. 15

Plate 21. Tongue skeleton of Loxops in ventral view. A) L. v. virens. B) L. m. mana. C) L. m. newtoni. D) L. c. coccinea. 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 169

m hgo

mthh

,m trh

st h rncg ,mbrn o

mbmp

mtrl brn

sh Plate 22. Tongue skeleton and musclesof L. sagittirostris. A) Tongue skeleton in ventral view. B) Tongue musclesin ventral view; superficialmuscles on right side, deeper muscles on left side. 170 ORNITHOLOGICAL MONOGRAPHS NO. 15

msh•

Ill Ill

mcg m ch

m tr I

m trh

/ m gg rn •

mbm

ch m hgo cg

•mtrl

mtrh

mth h

Plate 23. Tongue musclesof L. v. virens in ventral view. A) Superficial muscles. B) Deeper muscles. 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 171

msh

Amm h• mmcgch

m trl

m trh

mgg

mbm

mhg o• ch B

tr I

tr h

mcg

Plate24. Tonguemuscles of L. m. manain ventralview. A) Superficialmuscles. B) Deeper muscles. 172 ORNITHOLOGICAL MONOGRAPHS NO. 15

st h

mch

tr h m trl

m s h

mbmp

m gg mbm

B mch mtrh

mtr! m hg o

th h Plate 25. Tonguemuscles of L. m. newtoniin ventralview. A) Superficial muscles. B) Deeper muscles. 1973 RICHARDS AND BOCK: FEEDING APPARATUS IN LOXOPS 173

A ch

mbma

mcg mbmp mgg m sfh mbm

rn fh

ch B

--mtrl

mtrh mhgo

Plate 26. Tongue musclesof L. c. coccineain ventral view. A) Superficial muscles. B) Deeper muscles. ORNITHOLOGICAL MONOGRAPHS

No. 1. A Distributional Studyof the Birds of British llonduras, by StephenM. Rus- sell. 195 pp., 2 color plates. 1964. Price $4.50 ($3.60 to AOU members). No. 2. A ComparativeStudy of SomeSocial Communication Patterns in the Pele- caniformes, by Gerard Frederick van Tets. 88 pp., 4 text figures. 1965. Price $2.00 ($1.60 to AOU members). No. 3. The Birds of Kentucky, by RobertM. Mengel. Cloth bound,xiv q- 581 pp., 4 color platesplus text figuresand vignettes. 1965. Price $10.00 ($8.00 to AOU members). No. 4. Evolution of SomeArctic Gulls (Larus): an Experimental Study of Isolating Mechanisms,by Neal Griffith Smith. 99 pp., 62 text figures. 1966. Price $2.50 ($2.00 to AOU members). No. 5. A ComparativeLife-history Study of Four Speciesof Woodpeckers,by Louise de Kiriline Lawrence. 156 pp., 33 text figures. 1967. Price $3.75 ($3.00 to AOU members). No. 6. Adaptationsfor Locomotionand Feedingin the Anhinga and the Double- crested Cormorant, by Oscar T. Owre. 138 pp., 56 text figures. 1967. Price $3.50 ($2.80 to AOU members). No. 7. A Distributional Survey of the Birds of Honduras, by Burt L. Monroe, Jr. 458 pp., 28 text figures,2 color plates. 1968. Price $9.00 ($7.20 to AOU members). No. 8. An Approach to the Study of EcologicalRelationships among Grassland Birds, by John A. Wiens. 93 pp., 30 text figures. 1969. Price $2.50 ($2.00 to AOU members). No. 9. Mating Systems,Sexual Dimorphism, and the Role of Male North American PassefineBirds in the Nesting Cycle, by Jared Verner and Mary F. Willson. 76 pp. 1969. Price $2.50 ($2.00 to AOU members). No. 10. The Behaviorof SpottedAntbirds, by EdwinO. Willis, vi + 162 pp., 3 color plates,27 text figures. 1972. Price $6.00 ($4.75 to AOU members). No. 11. Behavior,Mimetic Songsand SongDialects, and Relationshipsof the Para- sitic Indigobirds (l//dua) of Africa, by Robert B. Payne,vi + 333 pp., 2 color plates,50 text figures,40 audiospectrographs.1973. Price $8.00 ($6 40 to AOU members). No. 12. Intra-island Variation in the MascareneWhite-eye Zosteropsborbonica, by Frank B. Gill, vi + 66 pp., 1 color plate, 31 text figures. 1973. Price $2.00 ($1.60 to AOU members). No. 13. Evolutionary Trends in the Neotropical Ovenbirds and Woodhewers,by Alan Feduccia,iv + 69 pp., 20 text figures. 1973. Price $2.00 ($1.60 to AOU members). No. 14. A Symposiumon the House Sparrow (Passerdomestlcus) and European Tree Sparrow (P. montanus)in North America, S. CharlesKendeigh, chair- man. vi + 121 pp., 25 text figures. 1973. Price $3.50 ($2.80 to AOU members). No. 15. Functional Anatomy and Adaptive Evolution of the Feeding Apparatus in the Hawaiian HoneycreeperGenus œoxops (Drepanididae), by LawrenceP. Richards and Walter J. Bock. vii + 173 pp., 14 text figures + 26 plates. 1973. Price $6.00 ($4.75 to AOU members). Like all other AOU publications,Ornithological Monographs are shippedpre- paid. Make checks payable to "The American Ornithologists'Union." For the convenienceof thosewho wish to maintain completesets of OrnithologicalMono- grapsand to receivenew numbersimmediately upon issue,standing orders will be accepted. Order from: Butt L. Monroe, Jr., Treasurer, American Ornithologists' Union, Box 2a447,Anchorage, Kentucky 402•a