Ornithological Monographs

No. 55

Obligate Army-ant-following : A Study of Ecology,Spatial Movement Patterns, and Behavior in Amazonian

SUSAN K. WILLSON

PUBLISHED BY THE AMERICAN ORNITHOLOGISTS' UNION OBLIGATE ARMY-ANT-FOLLOWING BIRDS: A STUDY OF ECOLOGY, SPATIAL MOVEMENT PATTERNS, AND BEHAVIOR IN AMAZONIAN PERU ORNITHOLOGICAL MONOGRAPHS

Editor:John Faaborg

224 Tucker Hall Division of BiologicalSciences Universityof Missouri Columbia, Missouri 65211

ManagingEditor: Bradley R. Plummer Proof Editors: Mark C. Penrose, Richard D. Earles

AOU Publications Office 622 ScienceEngineering Departmentof BiologicalSciences University of Arkansas Fayetteville,Arkansas 72701

The OrnithologicalMonographs series, published by the American Ornithologists' Union,has been established for majorpapers too long for inclusionin the Union'sjournal, The Auk.

Copiesof OrnithologicalMonographs are availablefrom ButeoBooks, 3130 Laurel Road, Shipman,VA 22971. Price of OrnithologicalMonographs no. 55:$10.00 ($9.00 for AOU members).Add $4.00for handlingand shippingcharges in U.S., and $5.00for shipping to other countries.Make checkspayable to ButeoBooks.

Author of this issue, Susan K. Willson.

Library of CongressControl Number 2004105806

Printedby CadmusCommunications, Ephrata, PA 17522

Issued5 May 2004

OrnithologicalMonographs, No. 55 x + 67 pp.

Copyright¸ by theAmerican Ornithologists' Union, 2004

ISBN: 0-943610-60-5

Cover: Assemblage of ant-following birds: (1) Dendrocinclamerula, (2) Rhegmatorhinamelanosticta, (3) Phlegopsisnigromaculata, (4) Myrmecizafortis, and (5) Gymnopithyssalvini. (Ink sketchby KirstenCarlson.) OBLIGATE ARMY-ANT-FOLLOWING BIRDS: A STUDY OF ECOLOGY, SPATIAL MOVEMENT PATTERNS, AND BEHAVIOR IN AMAZONIAN PERU

BY

SUSAN K. WILLSON

Divisionof BiologicalSciences, University of Missouri, 105 Tucker Hall, Columbia,Missouri 65211, USA

ORNITHOLOGICAL MONOGRAPHS NO. 55

PUBLISHED BY THE AMERICAN ORNITHOLOGISTS' UNION WASHINGTON, D.C. 2004 TABLE OF CONTENTS

Listsof Tablesand Figures ...... vii ABSTRACT ...... 1 1. INTRODUCTION ...... 3 2. BACKGROUND: SOME NATURAL HISTORY OF THE BIRDS AND ARMY ANTS 5 THE BIRDS ...... 5 THE ARMY ANTS ...... 8 THE STUDY SITE ...... 10 3. RESOURCE USE AND COEXISTENCE IN FIVE OBLIGATE ANT-FOLLOWING SPECIES ...... 11 METHODS ...... 12 FIELD METHODS ...... 12 STATISTICAL METHODS ...... 14 METHODSFOR SPECIFIC QUESTIONS ...... 14 RESULTS ...... 18 ARMY ANT COLONY DENSITY ...... 18 ADULT AVlAN POPULATION DENSITIES ...... 18 DOMINANCE RELATIONSHIPS AMONG BIRD SPECIES ...... ][8 AVIAN HOME-V•NGE SIZE ...... ][9 DISCUSSION ...... 20 ECOLOGICAL RELATIONSHIPS AMONG BIRD SPECIES ...... 20 AVIAN HOME RANGES AND POPULATIONS ...... 29 SEASONALITY AND MOVEMENT PATTERNS ...... 31 CONCLUSION ...... 32 CONSERVATION CONCERNS ...... 32 4. SURVIVAL RATES AND POPULATION DYNAMICS IN OBLIGATE ARMY ANT FOLLOWERS ...... 35 METHODS ...... 36 SURVIVAL ESTIMATES ...... 36 RESULTS ...... 37 SURVIVAL AND RECRUITMENT RATES ...... 37 POPULATION DYNAMICS AND TERRITORIALITY ...... 41 Discussion ...... 41 5. NESTING AND REPRODUCTION IN A GUILD OF OBLIGATE ARMY ANT FOLLOWERS IN AMAZONIAN PERU ...... 45 RESULTS ...... 45 SALVINI ...... 45 RHEGMATORHINA MELANOSTICTA ...... 47 PHLEGOPSISNIGROMACULATA ...... 49 DENDROCINCLA MERULA ...... 50 DISCUSSION ...... 51 6. CONCLUSIONS AND FUTURE QUESTIONS ...... 55 SUMMARY OF COEXISTENCE PATTERNS ...... 55 FUTUREQUESTIONS ...... 56 CONSERVATIONAPPLICATIONS ...... 58 ACKNOWLEDGMENTS ...... 59 LITERATURE CITED ...... 61 APPENDIX ...... 67 LIST OF TABLES

2.1. Morphologicalmeasurements of five bird species ...... 6 3.1. Colony densitymeasurements of army ant Ecitonburchelli ...... 18 3.2. Densityestimates for abovegroundforaging swarms of army ant Labidus praedator...... 19 3.3. Adult populationdensities and relative abundancesof five bird speciesover threeyears ...... 19 3.4. Rate-of-displacementmatrix for five bird species...... 20 3.5. Mean home-rangesizes of five bird species...... 20 3.6. Estimatednumber of Ecitonburchelli and Labiduspraedator swarms on an average bird species'home range, and percentageof time Ecitonburchelli colonies would not be available ...... 23 3.7. Frequencyand relativeuse of Ecitonburchelli and Labiduspraedator swarms by individually radiotrackedbirds ...... 23 3.8. Relativeuse of eachant speciesby five bird species...... 24 3.9. Mean displacementrate per minute per individual bird; results for Ecitonburchelli and Labiduspraedator swarms over threeyears ...... 25 3.10.Displacement rate with 95% confidenceintervals at Ecitonburchelli swarms ... 25 3.11.Displacements per minuteper bird for threeyears, grouped by birds per meter of antswarmfront ...... 25 3.12.Predictions and resultsof four hypothesesregarding avian use of spacealong Ecitonburchelli swarm fronts ...... 27 3.13. Avian use of total foraging spaceavailable at Ecitonburchelli and Labiduspraedator swarms over threeyears ...... 28 3.14.Comparison of home-rangeand populationestimates for five bird speciesbetween work presentedhere and a 1990study at the samesite .... 31 4.1.Adultsurvival estimatesfor four of the five bird species...... 38 4.2. Recruitmentestimates over two breedingseasons for five bird species...... 40 5.1. Nest charactersfor threeof the five bird species...... 46 5.2. Measurementsof nestlingbirds for threeof the five bird species...... 47 5.3. Nestingdates and nestoutcomes for three of the five bird species ...... 48 5.4. Mate switchingin pairsof obligateant-following birds ...... 53 6.1. Summaryof key ecologicaland behavioraldifferences amongfive bird species...... 56

LIST OF FIGURES

2.1. Ecitonburchelli army ant bivouac...... 9 3.1. Probabilityof an Ecitonburchelli colony foraging, given the number of colonies in an area ...... 16 3.2.Mean adulthome-range size per yearfor five bird speciesfor threeyears ...... 21 3.3. Estimatedmeters of antswarmfront utilized by eachof the five bird speciesin 1998and 2000-2001 ...... 26 3.4. Precipitationby monthat CochaCashu Biological Station, southeasternPeru, from 1998through 2001 ...... 32 3.5. Use of Ecitonburchelli and Labiduspraedator swarms by five bird species ...... 33 4.1. Recruitmentrate per 100ha per year for five bird species...... 40 4.2. Estimatedadults per 100ha for five bird species,over threeyears ...... 41 4.3. Selectedhome-range areas of Gymnopithyssalvini, nigromaculata,and Myrmecizafortis,2000-2001 field season...... 42 5.1. Gymnopithyssalvini nest with two eggsin low stump ...... 47 5.2. Rhegmatorhinamelanosticta nest with two eggsin leaf-sheath cavity of live palm ...... 49 5.3. Phlegopsisnigromaculata nestlings with larval botflies...... 51

APPENDIX

Map of the trail systemat CochaCashu Biological Station, Peru ...... 67 From the Editor

This is the secondOrnithological Monograph distributed to all membersof the American Ornithologists'Union. Whereas Ornithological Monograph No. 54 narrowlyfocused on the demogra- phy of a singlespecies, with contributionsfrom multipleco-authors, this monograph by SusanK. Willson attemptsto describethe ecologicalinteractions involved in the coexistenceof five species that spendnearly all their livesforaging at the front of antswarms.To appreciatethe many mecha- nisms involved in how these birds coexist, we also need to know a fair amount about the ants that providethe foraginglocation used by the birds. This monographincludes much of the researchdone by its authorfor her doctoraldissertation. Without somethinglike this monographseries, the dissertationwould have beenseparated into five or six unitsthat likely would haveappeared in five or six differentjournals. The ant ecology that is so importantto her story would likely have been in a journal ornithologistsrarely read. Anyonewho wantedto seethe whole storywould havehad to roundup thosearticles from an ar- ray of locationsand put thepackage together. Instead, with thismonograph, we canread the whole story abouthow the variousecological factors interact to suggesthow five obligateant-followers cancoexist in a singlePeruvian rainforest. We hope this monographencourages other graduatestudents to considertelling the story of their own researchin a singlelocation when thereis a longer,more complexstory than canbe told in piecemealfashion. We alsorecognize the downsideof publishinga singlemonograph rather than five or six separatepublications; we hope that departmentchairs, deans, and provostswill recognizethat a monographought to countfor morethan onepublication when considerationsof hiringand tenureare involved! Reviewingand editinga dissertationthat becomesa monographinvolves the help of many peo- ple. SusanK. Willsoh'sdoctoral committee included John Terborgh, Bette Loiselle, Frank Thompson III, RexCocroff, and myself;all of theseindividuals made numerous comments on developmentof the researchand early drafts of its results.Edwin O. Willisand PhillipStouffer were kind enoughto spenda largeamount of theirtime makingadditional comments on the monographicform of Sue's dissertation.Kimberly Smith, Bradley Plummer, Mark Penrose,and RichardEarles of the AOU PublicationsOffice were exceedingly helpful as we pusheddeadlines. Dr. JohnDavid, Chair of the Divisionof BiologicalSciences of the Universityof Missouri-Columbia,provided funds to support the colorplate. We thank all of theseindividuals for helpingto make OrnithologicalMonograph No. 55 a compellingexamination of someof the mostinteresting birds of the New Worldtropics.

JohnFaaborg COLORPLATE: From left to right,beginning upper left: Rhegmatorhina melanosticta (adult; photo copyright 2004,C. E. T. Paine);Gymnopithys salvini (female); Myrmecizafortis (female); Rhegmatorhma melanosticta (female); Dendrocinclamerula (adult); Eciton burchelli army ants; Myrmeciza fortis (male); Phlegopsis nigrornaculata (adult); Gymnopithyssalvini (male). OrnithologicalMonographs Volume (2004),No. 55, 1-67

OBLIGATE ARMY-ANT-FOLLOWING BIRDS: A STUDY OF ECOLOGY, SPATIAL MOVEMENT PATTERNS, AND BEHAVIOR IN AMAZONIAN PERU

SUSAN K. WILLSON 1 Divisionof BiologicalSciences, LiniversiO/ of Missoari,105 Tucker Hall, Columbia,Missouri 65211, LISA

ABSTRACT.-Five species of obligateant-following birds--Phlegopsis nigromaculata, Myrmeciza fortis,Rhegmatorhina melanosticta, Gymnopithys salvini (Thamnophilidae), and Dendrocinclamer- ula (Dendrocolaptidae)--andtwo speciesof army ants(Eciton burchelli and Labiduspraedator) werestudied in AmazonianPeru over five years. Here, I exploreaspects of speciescoexistence in thesefive ecologicallysimilar birds through analyses of their populationecology, resource use,behavior, and spatialmovement patterns. Mean home-rangesize of eachantbird species was reducedthrough reliance on the unpre- dictablebut abundant foraging opportunitiesat L. praedatorswarms. This little-known ant speciesplayed a pivotalrole in expandingthe foragingresource available to the obligateant- followingbirds, which allowed an increasein the birds' populationdensities well abovewhat wouldbe supportedsolely by thebetter-known E. burchelliarmy ants. Two of the five bird species(D. merulaand M. fortis)displayed resource selectivity among antswarmsby foragingsignificantly more with one of the two ant species.The D. merulafurther segregatedfrom the four antbirdsin its utilizationof white-lippedpeccary (Tayassupeccari) herds as a foragingresource; the peccariesact as "beaters"of arthropodprey in a mannersimilar to that of the army ants.The threeantbirds that did not preferone army ant speciesover the other(P. nigromaculata, R. melanosticta,and G. salvin0 segregatedby body mass,which may allow differentialuse of spacealong the width of an antswarmfront. That sizedifference would permit a smaller,more subordinatespecies to "fit" alongthe front of a swarmthat was already "full" to a differentbird species. Populationdynamics of thebirds were not stableover five years of datacollection, and total populationof obligateant-followers declined by almosthalf over the courseof the study.It is suggestedthat periodicpopulation fluctuations are a normaloccurrence in guildsof obligate ant-followersand may be exacerbatedby the lack of territorialityexhibited by mostof these species.Lower population density correlated with decreasedinterference competition among individuals.Population fluctuations may increase the ability of the subordinatespecies R. mela- nostictato coexistwith the larger,dominant P. nigromaculata in floodplain forest. Nest-siteselection may contributeto nichebreadth among the obligateant-followers. I pro- vide descriptionsof the nests,eggs, and nestlingsof P. nigromaculata,R. melanosticta,and G. salvini,which were undescribedat the start of the presentstqdy. Received 22 November2003, accepted3 February 2004.

RESUMEN.--$eestudiaron cinco especies de aves que siguen hormigas arrieras-- Phlegopsisnigromaculata, Myrmeciza fortis, Rhegmatorhinamelanosticta y Gymnopithyssalvini (Thamnophilidae)y Dendrocinclamerula (Dendrocolaptidae)--y dos especies de estashormi- gas (Ecitonburchelli y Labiduspraedator) en la Amazonia peruanadurante cincoaftos. En este trabajose exploran aspectos de la coexistenciade estascinco especies eco16gicamente similares por mediode analisisde suecologla poblacional, uso de recursos,comportamiento y patrones espacialesde movimiento. E1tamafio promedio del rangode hogarde cadaespecie de Thamnophilidaerue reducido por su dependenciaen las impredeciblespero abundantesoportunidades de forrajearque brindanlos ej•rcitos de L. praedator.Esta especie de hormigajug6 un papeldeterminante para expandirlos recursos disponibles para las aves, lo quepermiti6 un incrementoen la densidad poblacionalde •stasm•is all•i de losniveles que podrianmantenerse finicamente con las hor- migasde la especieE. burchelli. Dos de las cincoespecies de aves(D. merulay M. fortis)exhibieron selectividad de recursos

hE-mail:[email protected] 2 ORNITHOLOGICAL MONOGRAPHS NO. 55

entrelegiones de hormigas,forrajeando signficativamente mils junto a una de las dosespecies de hormigas.Ademfis, el trepatroncoD. merulase segreg6eco16gicamente de las otrascuatro especiesde avesen t6rminosde su utilizaci6nde manadasde huanganas(Tayassu peccari) comorecurso de forrajeo;estos animales espantan artr6podos que pueden ser depredados por lasaves en una formasimilar a comolo hacenlas hormigas arrieras. Las tres especies que no prefirieronuna especiede hormigasobre la otra (P.nigromaculata, R. melanosticta y G. salvini) sesegregaron en t6rminosde su pesocorporal, lo quepodrla perrnitir un uso diferencialdel espacioen el frentede avancede losejfircitos de horrnigas. Las diferencias de tarnafiopodrlan permitirque las especies mils pequefias y subordinadaspuedan acomodarse en el frentede un ej6rcitoque ya estarla11eno para individuos de otraespecie. Las dinfimicaspoblacionales de las avesno fueronestables a trav6sde los cincoaftos de tomade datos,y el totalde la poblaci6nde aves seguidoras de hormigas declin6 a casila mitad duranteel cursodel estudio.Se sugiereque las fluctuacionespoblacionales peri6dicas son eventosnormales en losgremios de avesque siguen hormigas obligatoriamente, y que•stas podrlanser acentuadas por la falta de territorialidadexhibida por la mayorlade lasespecies. Lostamafios poblacionales menores se correlacionaron con una competenciapor interferencia reducidaentre individuos. Las fluctuaciones poblacionales podrlan incrementar la habilidad de la especiesubordinada R. melanostictapara coexistir en el bosquedel planode inundaci6n conP. nigromaculata, que es mils grande y dominante. La selecci6nde sitiosde nidificaci6npodrla contribuira la arnplituddel nichoentre las avesseguidoras de hormigas.Aqui proveodescripciones de losnidos, huevos y pichonesde P.nigromaculata, R. melanosticta y G. salvini,los cuales no habiansido descritos al comienzode este estudio. 1. INTRODUCTION

FORCENTURIES, SCIENTISTS have beenawed by (Rettenmeyer1963, Schneirla1971, H611dobler the richnessof tropical diversity.Early natural- and Wilson 1990). Most of the 144 speciesof ists introduced the faunal and floral exuberance army ants in the subfamilyEcitoninae forage of the Amazon regionto the world throughtheir in long columns(H611dobler and Wilson 1990). speciescollections and natural-historywritings However, two species, Eciton burchelli and (von Humboldt and Bonpland1814-1829, Bates Labiduspraedator, forage in swarmsthat blanket 1862,Wallace 1889,Spruce 1908). Since the time the ground, advancingin fan-shapedraids or of Wallace, scientists have remarked on a latitu- patches.Moving over the ground surface,they dinal species-diversitygradient. High species flush leaf-litter .If the ants do not diversityin the tropicsresults from a combina- catchthem immediately,the insectsflee by es- tion of historical, climatic, and ecologicalfac- capingahead of the swarm front, providing an tors and area (Dobzhansky1950, Fischer1960, easilyaccessible food resourcefor birds. Many Connell and Orias 1964, Fedorov 1966, Pianka species of birds opportunistically use army 1966, Janzen 1967, MacArthur 1969, Connell ants as a foragingresource (Willis and Oniki 1978, Huston 1979, Ricklefs and Schluter 1993, 1978);>49 specieshave beenobserved foraging Rosenzweig 1995, Gaston 1996, Leigh 1999). at swarms in Amazonian Peru (S. K. Willson Within tropical guilds, higher species-packing unpubl. data). Somespecies, however, appear and finer niche-partitioningalong axesrelated to feed only or mainly at antswarms.Although to body size, foraging behavior, and vertical army ants offer a predictable resourcein the habitat-stratification contribute to coexistence long term, swarmsare spatiallypatchy and un- of ecologicallysimilar species (Hutchinson predictablein the short term (Sehneirla1971). 1959,MacArthur 1969,Diamond 1975,Terborgh Specialists(or "bligates") on army ants must 1980,Sherry 1984,Fitzpatrick 1985,Grant 1985, overcomethe challengesof finding and track- Marra and Remsen 1997, Rosenberg 1997). ing swarmsfrom day to day and of foragingef- Levey and Martinez del Rio (2001), for ex- fectivelyamid numerousand potentiallylarger ample, includedphysiological ability to digest competitors. fruit among their axes for avian frugivore The obligateant-following guild is ecologi- diversification;by examining why more birds cally interesting,because it appearsthat sev- do not eat fruit, they provided new insight into eral coexistingobligate species are sharingthe patternsof frugivore coexistence. same fundamental ecologicalniche (Grinnell Avian speciesdiversity reachesits peak in 1922, Gause 1934, MacArthur 1958, Hutchinson the Neotropics,where some 3,300 speciesare 1959).At somesites, up to three obligatespe- found (Karr 1989). The richest region of the des may be of equalbody mass(Terborgh et al. Neotropicsmay be in the lowland rainforestsof 1990).Additionally, diet compositionand prey southwestern Amazonia. Researchers in Manu size have been shown to completelyoverlap National Park, Peru, have recorded >540 bird among obligateant-following guild members, specieswithin ~3 km2 of the station(Terborgh countering predictions of speciescoexistence et al. 1984).At that site,245 speciesare resident basedon body size and interspecificcompeti- territory-holderson a 97-ha floodplain forest tion (Chesser 1995). Willis and Oniki (1978) plot. By superimposingterritory mapsor areas describeda seriesof concentricfeeding zones in of occupancyof individual species,Terborgh et a simple ant-following communityin Panama, al. (1990)quantified single-point (alpha) diver- in which the dominant (i.e. larger) obligate sity at >160 speciesthrough territory overlap. speciesmonopolized areas closer to flushed Diversificationand specializationin tropi- prey. Willis's numerous studies indicate that cal birds is aided by resourcesthat are avail- ant-followingspecies may segregateby perch able year-round in the tropics (Orians 1969). type at swarm fronts,using verticalperches of One such resource,unique to the tropics, are varioussizes, the ground, or horizontal perches. the army ants (Willis and Oniki 1978). Army Additionally, Willis and Oniki (1978) suggest ants forage both below ground and over leaf that obligateant-following birds segregateby litter for a wide range of invertebrate prey dominancebased on size--with larger, more 4 ORNITHOLOGICAL MONOGRAPHS NO. 55 dominant speciestaking the lowest and clos- ant-followersfirst offeredby Willis and Oniki est position to the oncoming antswarm so (1978), I examined differential use of available as to capture the largest amount of fleeing foraging resources(i.e. the army ants) at the arthropodprey (see summaryin Zimmer and community level as well as temporal variation Isler 2003). Explanationsof coexistenceare less in population-levelprocesses. I consideredthe clear for Amazonian guilds, in which members behavioral and small-scaleecological patterns frequently overlap in body mass, foraging of individual birds and army ant colonies,and behavior,or feeding preferences.Building on the broaderpatterns of species'annual survival, explanationsfor speciescoexistence in obligate populationdynamics, and recruitment. 2. BACKGROUND: SOME NATURAL HISTORY OF THE BIRDS AND ARMY ANTS

THE BIRDS from antswarms,and their rarity as a group-a "legendary" status (especiallyamong birders) OBLIGATE ANT-FOLLOWING BIRDS demonstrate that is perpetuatedin popular writing (Forsyth the evolutionof a highly specializedlifestyle, and Miyata 1984,Kricher 1997). which is intrinsicallytied to the ecology of Here, ! introduce the five study speciesand the army ants on which they dependfor food. briefly review their socialsystems, which range Obligate ant-followersseem behaviorallyinca- from solitaryindividuals to family groups.All pableof catchingprey that is not beingflushed informationnot specificallyreferenced was first by anotherorganism (see Skutch 1996 for an ac- describedin the presentstudy. countof the authoracting as a "beater"of prey The dominantobligate ant-follower at Cocha for an obligateant-following bird). Thereare no Cashu,P. nigromaculata(Black-spotted Bare-eye), publishedaccounts of an obligate ant-follower lives in family groupsof two to possiblyfour hunting and catchingprey without the service adults,plus juveniles in the postbreedingseason, of army ants or other beatersof prey. Hence, and averages46 g at CochaCashu. Cadena et al. obligateant-followers might more correctlybe (2000)first described the nest. The only published termed"obligate followers of beaters."Generally, behavioralinformation comesfrom Willis (1979), army antsprovide that service.Willis and Oniki who studiedthe spedesin .Composition of (1978)discussed "professional" and "facultative" the species'family groups is not well understood. ant-followers,but allowedthat professionals may At CochaCashu, multiple family-groups gather forageon their own, away from army ants,for at antswarms, and individuals from different up to 50%of the time.Here, ! usethe term "obli- familiesmay perchand "loaf" togetherbetween gate"for all fivestudy species (four and foragingbouts. Additionally, ! foundthat a fam- one woodcreeper).During five yearsof careful ily maysimultaneously have individuals foraging observationand intensiveuse of radiotelemetry, at differentswarms hundreds of metersapart, as no individualswere everknown to forageaway in Phaenostictusmcleannani (Ocellated ) in from beatersof prey. The woodcreeperwas Panama(Willis 1973). Willis (1979)speculated that found,during the study,to obligatelydepend on "extra males"of Pha. mcleannani,probably sons two typesof beatersto flushprey- armyants and of previousclutches, stay with the parentsfor herdsof white-lippedpeccaries (Tayassu pecari). severalyears before dispersing. At CochaCashu, For convenience,all five speciesare referredto ! confirmedthat P. nigromaculatais a cooperative hereas "obligateant-followers,"even though the breeder(see Chapter 5). woodcreeperis technicallyan "obligatefollower Althoughon averageit is slightlylarger (at 46.5 of beatersof prey"at thestudy site. g) thanP. nigromaculata,M. fortis () Few researchers have studied the natural his- is secondin the strictlinear-dominance hierarchy tory of obligate ant-following birds. The most that existsamong the obligateantbird species. No significant and well-known publicationson accountsof its natural historyhave been pub- the subjectcome from the prolific E. O. Willis, lished,though Willis (1985) describesfrequent who began his studiesin Central America and ant-following by M. fortis and related species. subsequentlytraveled the Amazon, publishing Its nest was first described in 1997 (Wilkinson an extensiveliterature on the natural history and Smith 1997). Myrmecizafortis individuals of membersof the obligateant-following guild form monogamouspairs that are highly ag- (see Willis and Oniki 1978, 1992). Much of his gressiveboth inter- and intraspecifically.The work describesthe posturingand perchuse, vo- lack of differencein massbetween M. fortisand calizations,intra- and interspecificinteractions, the dominantP. nigromaculata(Table 2.1) suggests and minutiaeof daily life at an antswarmfront. that groupsize may accountfor the subordinate Most other publishedinformation is anecdotal statusof M. fortis.Given that P. nigromaculata indi- or scattered,and obligate ant-followers have vidualsgenerally forage in groupsof threeto five, attained-becauseof their specificityof habitat it is possiblethat their sheernumber, through and diet, the difficultyof observingthem away collectiveresource-defense, allows for species 6 ORNITHOLOGICAL MONOGRAPHS NO. 55

dominanceover M. fortispairs. Terborgh (1983) describeda similar exampleof cooperatively exerteddominance between two Cebusmonkey speciesthat differ in mass.Myrmeciza fortis is notablefor its extremeintolerance of conspecif- ics. Willis (1985) noted differentpairs at ends of the sameantswarm, but observedfights be- tweenpairs. Unlike otherobligate ant-followers at CochaCashu, M. fortis pairs never shareda swarmwith conspecificpairs during the present study and held all-purpose"type A" defended territories (as described in Karr 1971). All oc- currences of more than two adults at a swarm duringthe presentstudy were of parentswith a grownoffspring. The status of M. fortis as an obligate ant- followerwas in doubtwhen the presentresearch began,because no studyof itsbehavior had been publishedpreviously and becauseof its phylo- geneticplacement apart from the otherobligate antbirds.The genusMyrmeciza has high morpho- logical variabilityand encompassesgeneralist foragers,facultative ant-followers, and a few spe- ciesbelieved to exclusivelyfollow army ants(see Zimmerand Isler2003). Through radiotelemetry, the presentstudy demonstratedthat M. fortisis an obligateant-follower that feedsexclusively at antswarms.Myrmeciza fortis also engages in the specializedbehavior of bivouac-checking(see Swartz 2001), though Willis (1982a and E. O. Willis pers.comm.) showed that facultativespe- ciesmay performbivouac-checks as well. The medium-sized R. melanosticta(Hairy- crestedAntbird) is the third obligateantbird speciesin the dominancehierarchy at Cocha Cashu, subordinateto both P. nigromaculata and M. fortis. Rhegmatorhinamelanosticta aver- aged 31.4 g and generallymaintained socially monogamouspair bonds.In the only publica- tion dealingwith this species,Willis (1969)de- tailedbehavior of the (five species),but he did not follow marked individuals and did not performlong-term observations of R. mela- nosticta.My work suggeststhat the frequencyof exclusivepair bondsmay be inverselyrelated to populationdensity. A dramaticdecrease in adult populationdensity between 1998 and 2000 corre- spondedwith changesin observedintraspedfic interactions. In the 1998 and 1999 field seasons, individualsgenerally formed loose associations and seemedto form exclusivepair bondsonly during nestingattempts. In 2000,all adultswere in tight pair bondsand were seenapart only OBLIGATE ANT-FOLLOWING BIRDS IN PERU 7 while incubating.The lackof availablemates was homeranges and regularmeetings with males demonstratedwhen an adult male begancourt- at antswarmslikely allow femalesto evaluatea ing a fledglingfemale less than one month out numberof potentialmates. of the nest;the malewas accepted by her family Althoughnot part of the presentstudy, four and beganforaging with them daily.The male facultative ant-followers at Cocha Cashu- took over provisioningof the fledgling,which Hylophylaxpoecilonota, Dendrocolaptes picumnus, had still been receivingfood from her parents. Dendrocinclafuliginosa, and Neomorphusgeof- Alternative13the observedpopulation decline froyi-deserve mentionhere (basedon previous may have eliminated"floaters" (unpaired indi- publishedaccounts of their use of antswarms). viduals) and decreasedcompetition for mates, Hylophylaxpoecilonota (Scale-backed Antbird) allowing all individuals to form pair bonds. is a small thamnophilid species (see Willis When populationdensity was high, 11 adult in- 1982c),but residesmainly on terrafirme forest dividualswere present at a largeantswarm. at CochaCashu. It was very rarely seenin the The smallestand most subordinateobligate study area (threesightings in four years). ant-followerin the presentstudy is G. salvini Dendrocolaptespicumnus (Black-banded (White-throatedAntbird). This speciesaveraged Wood-creeper)attended antswarms (see Willis 25.9g and maintainedsocially monogamous pair 1984), but its status as an obligate at Cocha bonds.Willis (1968)observed unbanded individu- Cashu was rejected by Pierpont (1986); she als of the speciesfor lessthan a week;his study observedDendroc. picumnus foraging both at is the only publishedbehavioral information for antswarmsand with understorymixed-species the species.As many as five pairs of G. salvini flocks and characterizedits affinity for army aggregated at large antswarms, and intraspe- ants as "medium." cific aggressionwas common.Willis (1967)de- Another woodcreeperspecies, D. fuliginosa scribesreversals of dominancefor G. leucaspisin (Plain-brown Woodcreeper), was character- Panama,when a pair crossesinto the centerarea ized by Willis (1966a) as a "professionalant of anotherpair. Although individual dominance follower in all areasI have studied it," though interactionswere noted in the presentstudy, I he mentionedthat it foragesaway from swarms did not observe clear reversals of dominance on occasion.At Cocha Cashu, D. fuliginosais across territorial boundaries in G. salvini. a facultative ant-follower and spendsmost of Finally, D. merula (White-chinned Wood- its time with understorymixed-species flocks. creeper), averaging 47.5 g, was subordinate From observationand radiotelemetry,Pierpont to the two largest antbirds but received and (1986) characterized the species' affinity for initiated agonisticencounters with R. melano- army antsas "low." sticta(average 31.4 g), leaving the two species Neomorphusgeoffroyi (Rufous-vented Ground- similarlypositioned in the dominancehierarchy. Cuckoo)is a specieswhose rarity and shyness Dendrocinclamerula was mainly intraspecifically precluded much observation of it. Almost aggressive,and individualssometimes lost large nothingis known aboutits ecologyor behavior portionsof foragingtime to conspecificaggres- (Willis 1982b,Ridgely and Greenfield2001). It sion. Thoseinteractions generally did not pre- was sometimessighted at antswarms,but was clude individuals from gatheringat the same also seen foraging under squirrel monkeys swarm; sometimes more than five adults were (Saimirisciureus), which act as beatersof prey presentat once.My observationsfrom Cocha as they drop arthropodsto the ground (M. B. Cashu are counter to those of Willis (1979) in Swartzpers. comm.). The namegiven to N. geof- Brazil, who reported that individuals "dispute froyiby thelocal Matsigenka people, "huangana with each other little." However, D. merulais rare piscco,"translates to "white-lipped-peccaryfol- in all Brazilian areas Willis has checked (E. O. lower."Foraging across the forestfloor for nuts, Willis pers.comm.).The undescribed nest is pre- roots,and matter,white-lipped peccary sumablytended solely by the female,as in other herds can flush arthropods;herds were regu- Dendrocinclaspecies (Willis 1972). Individuals larly seenon the studysite (Silmanet al. 2003). did not form long-termpair bonds,and females myotherinus (Black-faced caredfor fledgedyoung alone. Males seemed to Antbird) was the most common facultative contributesperm and little else,which suggests ant-follower at Cocha Cashu antswarms, and female choice for male quality. Overlapping individuals and pairs frequented swarms of 8 ORNITHOLOGICAL MONOGRAPHS NO. 55 both ant species.Specifically, I recorded Myrrno. nomadicphase, which lastsN14 days, the entire rnyotherinusat 94 of 345 observedswarms (27%) colony moves its temporary nest, or bivouac, in 2000-2001.All birds in the present study almostevery night. During the day,adult work- weredominant to Myrrno.rnyotherinus. ers forageout from the bivouacin a swarmraid Family groups of Psophialeucoptera (White- to capture arthropodsand social-insectlarvae, winged Trumpeter; Psophiidae) occasionally which they carry back to the bivouac-where foraged at antswarm fronts. Individual trum- the army ants' own brood of N300,000larvae is peterssometimes ran throughthe front to cap- waiting to be fed (Fig. 2.1). ture large fleeing arthropods (i.e. centipedes, During the stataryphase, the colonyremains scorpions,and tarantulas) and momentarily in a protectednest-site for -21 days.The single disrupted foraging by obligate ant-followers. queenlays her eggsover a period of 3-4 days, Significantdisruption of the front occurredonly and all larvaefrom the previousnomadic phase when two family groupsconverged at a swarm. pupate.Relieved of the needto feedlarvae, adult Family groupschased each other and at times workersforage sporadically for prey.At the end disturbed the forward movement of the advanc- of the stataryphase, eggs hatch into larvae,and ing army ants. However, front cohesiveness pupae emergefrom their cocoonsand join the was not affected for more than a few moments, ranks of adult workers.The army ant cycleis and obligateant-following birds did not leave thereforea 35-day period. An individual egg swarmsattended by trumpeters. mustgo throughtwo cycles(70 days)before the pupa emergesas a new,or callow,adult worker. ThE AR•¾ ANTS Callow workers are lighter in color than their older adult counterparts,and thus can be visu- The ecologyof the Neotropicalarmy ant E. ally distinguishedfrom the latteruntil their exo- burchelli(Formicidae: subfamily Ecitoninae, tribe skeletonsdarken, after approximately 2-3 days. Ecitonini)has been studied by only a handfulof During the nomadic phase, swarm raids investigators.Much of the publishedinforma- begin around dawn. Ant activity in the biv- tion on army ant naturalhistory comes from T. ouac quickly changesfrom little or no visible C. Schneirla, who dedicated his life's work from activity on the part of the workers to a sudden the 1930sto 1960sto their study (compiledin exodus from the bivouac, which can be likened Schneirla1971). Beginningin the 1960s,E. O. to water pouring from a waterfall. The ground Willis amasseddata during his study of the immediatelysurrounding the bivouacbecomes obligateantbird G. leucaspis(Bicolored Antbird) dark with ants; within 30 min, the ants choose a in Panama. Willis's data on E. burchelli were primary foragingdirection and form a column analyzed by Nigel Franks, who augmented pushingoutward in that direction.The forag- them with his own while studying army ant ing body can be thought of in metaphor as a populationecology, beginning with his doctoral tree. The base of the tree is the bivouac, and dissertation(Franks 1980). Frankshas advanced the trunk is the ever-lengtheningforaging col- the understandingof army ant spatialdynamics umn (generallyonly a few centimeterswide). through use of simulationand mathematical Toward the distal end of the column, workers models. Schneirla, Willis, and Franks did the begin a processof dendriticbranching, forward majorityor all of their work on BarroColorado and outward from the column. The branches Island in Panama.Unfortunately, lack of studies become smaller along the column until all elsewherehas led to often-repeatedassumptions branchesmerge at the swarm front (the top of about army ant population densitiesin both the "ree"), where ants (the "eaves") darken the scientificjournals and the popular press.My ground.The swarmfront variesin sizedepend- data from southeastern Peru demonstrate that ing on time of day,colony size, phase, and other populationdensities are >l.5x higher there than factors, but can reach >25 m across at Cocha on Barro Colorado Island, where densities have Cashu. The swarm front is where the ants are remainedstable at 3.2 coloniesper km2 for >50 actively searchingfor and capturing prey. As years(Willis 1967,Schneifia 1971, Franks 1982a). they move through the leaf litter, arthropods Eciton burchelliants live in colonies of 500,000 and vertebratesin their path hop, fly, or run to 2,000,000individuals, and their life cyclehas ahead of the advancingswarm. In dry leaf lit- two distinctphases (Schneirla 1971). During the ter, the noise of fleeing prey is quite audible. OBLIGATE ANT-FOLLOWING BIRDS 1N PERU 9

F•G.2.1. Ecitonburchelli army ants build nests, or bivouacs, entirely from their own bodies.The imageabove shows a no- madic bivouac-1.5 m long, under a fallen log. The image on the left is a close-upof that bivouac, with worker ants moving out from the bivouacto forage. l0 ORNITHOLOGICAL MONOGRAPHS NO. 55

In additionto preyingon groundfauna, army (1963)provided much basic natural history and antsclimb treesin their path to scoutfor nests biology.Individual L. praedatorcolonies are a of socialinsects. Rettenmeyer (1963) has shown spatially and temporally unreliable resource that social-insectlarvae, particularlythose of for obligate ant-followingbirds, becausethe other ant species,are the preferredprey of E. location and activity of abovegroundswarms burchelliand generallyaccount for the majority are unpredictable.When they do raid above of their prey intake. When swarm raids find a ground, swarms are small (averagewidth 2.5 large social-insectnest in a tree,the entire front m), movein an ""shape with frequentchanges may stopits ground raid and concentrateits ef- of direction,and candisappear underground at fortson retrievinglarvae from that nest.Adults any moment.However, L. praedatorswarms are of the raided nestssoon give up any attemptto a valuableresource for obligateant-following rebuffthe army ants,and often attemptto save birds that find them opportunistically,because themselvesby "aining" out of an E. burchelli- they are much more commonat CochaCashu filled tree onto the forest floor below. than E. burchelli swarms. Trafficalong the E. burchelliforaging column is continuouslybidirectional and connectedto ThE STUVY S•TE the bivouac.Because army antsforage forward from the bivouacat -14 m h-•, the foragingfront Field data were collected at Cocha Cashu can be >125 m from the bivouacby late after- BiologicalStation, Manu National Park, Peru noon.During the nomadicphase, a colonywill (11ø54'Sand 71ø22'W;elevation -400 m). Mean generallymove its entire bivouac-beginning annualtemperature is 23.5øC,and annual rainfall at the end of the day and workingthrough the averages2,000 mm, puttingthe sitenear the cli- night-to a new location.The ants place the maticboundary between tropical and subtropical new bivouacsomewhere along that day'sexact moistforest in the Holdridgesystem (Holdridge foragingpath by followingthe pheromonetrail. 1967,Terborgh 1990). The dry seasonlasts from At times,a colonymay foragein morethan one April to mid-September,and the rainy season direction from the bivouac, either simultane- from Septemberor Octoberthrough March. The ouslyor overthe courseof a day.That is a more station and surroundingarea lie in the flood- commonphenomenon during the stataryphase, plain of the Manu river, a meanderingwhite- but can also occur during the nomadicphase waterAmazonian tributary of the Rio Madre de when there is an obstructionin the primary Di6s. The foresthas a canopyheight of -30 m, direction(e.g. a body of water, other E. burchelli with scatteredemergent trees reaching 50 m in ant-colonypheromones, a rain shower). height.Large tree-falls are common,because the A second army ant species,L. praedator,is soilsare regularlyinundated. The forestis in a alsoan importantresource for the obligateant- constantstate of succession,new tree-falls creat- followers at Cocha Cashu. Not as well studied ing light-gapsand old tree-fallsfilling with fast- as E. burchelli because of their subterrenean growingsaplings and lianas. Interspersed within habits,from their L. praedator propensityare often to swarm called on "rain the ants,"forest the forest are lower areas that become inundated duringthe rainy season, creating a mosaicof dry floor after a downpour.They are driven to and wet patches(0.1 to 1.5 m deep) that may the surfaceby inundationof the subterranean persistuntil the following dry season.Larger cavitiesthey generallyfavor for raiding; thus, depressedareas within the study plot are fig the speciesoften exhibits a seasonalpattern swamps,formed when old oxbowlakes dried up of abovegroundswarming. E. O. Willis (pers. and filled with sediment. Those areas are char- comm.) has speculatedthat L. praedatormay acterizedby scattered,sprawling fig trees(Ficus forage more nocturnally during dry weather. trigona)and othercanopy cover, with few under- Labiduspraedator does not display the cyclic storytrees but a denseundergrowth of tall her- lifestyle of E. burchelli,and can remain at the baceousplants (Heliconia spp. and Marantaceae). samenest-site for up to five months(Schneirla A detaileddescription of the ecological,geologi- 1971).Few studiesof its ecologyhave been un- cal, and climatic characteristicsof the site can be dertaken,but Schneirla(1971) and Rettenmeyer foundin Terborgh(1983) and Gentry(1990). 3. RESOURCE USE AND SPECIES COEXISTENCE IN FIVE OBLIGATE ANT-FOLLOWING BIRD SPECIES

SPECIESTHAT OCCUPY similar ecologicalniches using average swarm width per ant species, within a habitat will compete for resources estimateddensity of each ant speciesover the that are in limited supply (MacArthur 1958, study area, and daily probability that an ant Connell 1961). Interspecificcompetition may colony will forage. Becausebirds aggregateat negativelyaffect fecundity, survival or growth the front of a swarm, across its width, that cal- of one or more competing species through culationaccurately reflects the averageamount density-dependent effects (Feinsinger 1976; of "space"available to foragingbirds per 100 Brown 1989a, b). Theoretical, experimental, ha. I comparepredicted use of availableforag- and empirical studieshave shown that for eco- ing space,based on bird populationdensity and logicallysimilar speciesto coexistin a habitat antswarm density, with observedspacing be- with limited resources, niche differentiation havior of individual birds at swarms, over three must occur (MacArthur and Levins 1964; field seasons.Individual birds may need some May 1973; Schoener1974; Brown 1975;Brown minimum amount of space along a swarm's 1989a, b). Niche differentiationmay involve width to forageeffectively among competitors. dissimilaritiesin body size, habitat use, food On a smallerspatial scale, I examinespacing be- choice,or temporalactivity patterns(Lack 1944, havior of birds alonga swarm'swidth to better Hutchinson1957, Schoener 1974, Terborgh 1983, understandwhether width representsa limiting Durant 1998,Himes 2003). In natural communi- resourcefor obligateant-following birds. ties, a heterogeneousenvironment is coupled To understand limitations on abundance of with evolutionary trade-offsamong coexisting obligateant-following birds, it was necessaryto speciesin their abilitiesto use various parts of determinesome basic yet unknown ecological the environment.Each species may havecertain parametersfor thebird and army ant studyspe- behavioraladaptations that provideadvantages cies:(1) What is the colonydensity of E. burchelh over its competitorsunder specific combina- at CochaCashu? (2) What is the densityof active tions of environmental conditions,allowing abovegroundswarms of L. praedator?(3) What is competingspecies to coexistin the same envi- the yearly adult populationdensity of eachof ronment (MacArthur and Levins 1967, Pianka the five bird speciesat CochaCashu? (4) How is 1981, Abrams 1983). Competition may not be the dominancehierarchy structured among the active during long periods of relatively high five obligateant-following bird species?And (5) resourceavailability, but episodesof low avail- what is the yearly mean home-rangesize for ability may significantlyinfluence community each of the five bird speciesin the study area? structure (Wiens 1977). With estimatesof thoseparameters, I can explore My goalsin this chapterare to explorepat- relationshipsamong antswarm availability, com- terns of coexistenceamong ant-following-guild petitionbetween bird species,and differentiation members and to determine how resource use in patternsof use of the army ant resource.My may differ among species.I focus on species- ultimategoal in thischapter is to determinehow specificutilization of two army ants(E. burchelli the five speciesof obligateant-followers differ in and L. praedator)and on the dynamicsof abun- their use of the availableforaging resource, and danceand home-rangesize. I explorethe inter- how thosedifferences potentially contribute to and intraspecificdominance hierarchies among speciescoexistence. Using results from the basic the five bird species,and discusswhere and how ecologicalquestions above, I explorethe follow- interferencecompetition is mostintense. Yearly ing questions: populationdensities of the five bird speciesand I. Doesbody sizeor relativecompetitive abil- two ant speciesare alsoexplored in depth, and ity among speciescorrelate with home-range investigated as an influence on competition size? asamong a conceptual the birds.tool I introduce related to"foraging the quantityspace" of II. Do obligateant-followers minimize home- rangesize to fit theminimum number of reliable antswarmresource available to foragingbirds food resourcesneeded for daily foraging? over a given area. Foragingspace is calculated III. Do obligate ant-following species

11 12 ORNITHOLOGICAL MONOGRAPHS NO. 55

differentiallyutilize E. burchelliand L. praedator unpubl. data), and the best methodfor trackinga swarmsand, if so,in what proportions? colonyday-to-day is to follow its emigrationcolumn IV. Are ratesof agonisticinteractions between from one bivouac to the next. Eciton burchelli colonies birds differentat L. praedatorversus E. burchelli at CochaCashu generally began moving their bivouac swarms? by lateafternoon, and it wasalmost always possible to follow the moving columnof ants to the new bivouac V. Doesbody size or competitiveability cor- beforenightfall. relatewith spaceutilization at a swarmfront? Six days eachweek, we observedall foragingE. VI. Judging from overall obligate ant- burchellicolonies for bird activity. We carried out followerdensities, are birds maximizing spatial observationsfor ~60 min per colony before noon, utilizationof availableE. burchelliand L. praeda- whenarmy antforaging is at its peak(Schneirla 1971). tor swarms? Observersstayed longer at large swarmsthat held VII. Can spaceat swarmsbe a limiting re- more birds, to assurethat all obligateant-followers sourcefor obligateant-followers? presenthad beenaccounted for. My assistantsand I often monitored >10 colonies at a time, but with four researchersand the probability that some colonies METHODS were statary and not foraging on a given day, we were able to accomplishall foraging observations FIELD METHODS beforenoon eachday. At the end of an observation, we recordeddata on antswarmcharacters, including Approximately 17 months of daily field observa- width of the swarmfront. We checkedall statarycolo- tion of army antsand ant-followingbirds was carried niesdaily for activity,and carriedout an observation out on a 161- to 277-hastudy plot from September if the ants were foraging.We mappedall foraging 1997 to April 2002. Almost all work was conducted frontsand bivouacsto the CochaCashu trail system, m the rainy season,which coincideswith the breed- usingestimates of degreesand metersfrom the near- mg seasonof obligateant-following birds. It remains est trail-marker. Observations were not carried out on unknown whether behavior of the five obligate Sundays,but all coloniesthat were likely to move to a ant-followerschanges in the dry season,when in- new bivouacon Sundaynight--nomadic colonies and sect availability and foraging opportunitieswith L. thoseat the end of the statary phase--were checked praedatorants may decrease(Willis 1967,Levings and that afternoon and followed to their new bivouac if Windsor 1982). necessary. Data collectionwas carried out by crewsof two to We found activeaboveground swarms of L. praeda- five individualsper field season.Duration of the field tot either opportunisticallyor while tracking birds seasonvaried from year to year.In 1997,the study plot with radiotelemetry.Labidus praedator swarms do not was selectedand birds were intensivelymist-netted displaythe dendriticform characteristicof E. burchelli and banded from Septemberthrough November.In (seeChapter 2); rather,they form a dense"carpet" 1998 and 1999, the seasonlasted from September of ants.Measurements included widthl and width2, throughDecember, with bandingsupplemented by which we took perpendicularlyafter observingany antswarmobservation and radiotelemetry.We contin- ongoing bird activity in the same manner as at E. ued that work from October2000 throughFebruary burchelli swarms. 2001. In 2002, I returned for a month in March and Armyant densities.--We calculatedaverage density April, mainlyto gatherdata on adultsurvival rates. of E. burchelli colonies at Cocha Cashu from known Army antfield methods.--Onarrival at the studysite densitymeasurements of bivouacsover a 50- or 100- eachyear, my assistantsand I (yearly total of four to ha area for 1998, 1999, and 2000-2001. We included a five researchers),over the courseof one to two weeks, colonyin weekly countsif its bivouacwas locatedin walked trails throughoutthe study plot and located the sampledarea. I calculateda mean densityfrom all E. burchelli colonies within the area. Trails were weekly countsover a field season.I calculateda core monitored throughout each field season,and new density,rather than the densityof the entireplot, be- colonieswere found as they enteredthe study plot. causeplot sizevaried year-to-year and samplingwas Eachcolony that was found was monitoredand fol- lessintensive on the edgesof the plot. Becauseof our lowed daily from bivouacto bivouacuntil it eitherleft intensivedaily samplingin the core area, my assis- the study area or was lostby us, or until the field sea- tants and I were aware of all E. burchellicolonies there, sonended. Loss of a colonygenerally resulted from and mean densities from each season are based on the colony'smovement into very thick, Heliconia- actualcounts of colonybivouacs rather than extrapo- dominatedherbaceous swamps, where trackingwas lations or estimates(Franks 1982a). To accountfor difficult. Army ant tracking was not carried out in time spentfinding all army ant colonieson the plot, I 1997. Ant colonies can move their bivouacs >150 m omitted the first three weeks of each field seasoh's E. per night during the nomadic phase (S. K. Willson burchellicolony counts. OBLIGATE ANT-FOLLOWING BIRDS IN PERU 13

In the 2000-2001 and 2002 seasons, I estimated vegetation,and we regularlyobserved swarms passing densityof activeaboveground L. praedatorswarms at obliquelyor perpendicularlyover trails. CochaCashu. Unlike E. burchelli,L. praedatoris subter- Avian field methods.--Mist-nettingand banding raneanas well as terrestrial,and ant-followingbirds methodologiesfollow the guidelinesof the North must searchfor activeaboveground swarms rather AmericanBanding Council (2001a, b, c). Mist-netting than trackingknown colonies.I developedthe fol- wascarried out opportunistically,with linesof 10-20 lowing techniqueto estimateL. praedatordensity as netsplaced near known E. burchellibivouacs that were availableto ant-followingbirds. All researcherson the closeto trails.Because the largestnumber of obligate project recorded their own distancewalked on trails ant-followingbirds are caughtwhile antsare passing in the study plot to the nearest25 m and recorded under nets,we strategicallyplaced mist-nets to take the number of L. praedatorswarms encountered on advantageof army ant coloniesmoving through an trails. We imposeda "15-minuterule" to mimic the area.We placednets on trails in the afternoon,near movement speed of an average E. burchelliswarm the foragingpath or bivouacof an army ant colony front (becauseL. praedatormovement speed has not At dawnthe next da.• we openedthe nets and caught been measured).If a trail was walked >15 min after birds either as they flew to the bivouacto checkits it had previouslybeen walked, that distancewas in- activity(Swartz 2001) or later as they passedwith the cludedin the day'sdistance, to allow a hypothetical army antsthrough the mist-netline. L. praedatorfront to moveacross a trail without being Mist-netswere generallyopen from dawn until countedtwice. Labiduspraedator swarm movements noon, thoughsome late-afternoon netting was done aremore S-shaped than linear; I assumedthat swarm We outfitted captured birds with two individual movement would still be forward rather than circular color bands and one numbered metal band. I used (that assumptionseems appropriate, from personal standard techniquesfor morphologicalmeasure- observationof L. praedatorswarms). My assistantsand ments:culmen length was measuredfrom anterior I coordinated our times and movements so that trails edgeof naresto bill tip, culmenwidth and depthwere near campwould not be countedmultiple timesin a measuredat anterioredge of nares,and wing chord day.Total distances walked, per personper da.• were was unflattened(North AmericanBanding Council summedto get a total distance(m) walked per week. 2001a,b, c). Monomorphicspecies (D. merulaand P The equation nigromaculata)and R. melanostictajuveniles were sexed using standard moleculargender-assignment tech- niques(Griffiths et al. 1998).I drew blood (30-50 •tL) totalL.praedator/wk ß1,000,000 = est. #swarms (3.1) by brachialvenipuncture. Extracted DNA was ampli- total m/wk * 3 m 100 ha fiedusing polymerase chain reaction (PCR), and PCR productswere separatedon a gel by electrophoresis describesthe methodfor calculatingtotal numberof to displaysex-specific banding patterns. activeL. praedatorswarms per week. I multiplied dis- During antswarmobservations (described above), tancesby 3 to getan area(m 2) covered per week.Field one or two observerspositioned themselves _>10 m trails were generally2 m wide, and the area within from the sides of the swarm front, recorded all in- 1.5 m of the centerof the trail in either direction(total dividual birds present,and took noteson behavior width 3 m) was easyto checkfor antswarmactivity and interactions between individuals. We determined while observerswalked trails at a normalpace. I cal- dominancewithin and betweenspecies by quantify- culatedestimates of the numberof activeL. praedator ing the number of times individuals displacedthe swarmsper 100ha (1,000,000m 2) eachweek of each perchsite of otherindividuals during the observation field seasonand then averaged to getan overallmean period.I calculatedpercentages of perchor ground for the season. Because the 2002 field season covered displacementsthat eachspecies exhibited intra- and lessthan four weeks, all datawere averaged to getone interspecificallyto determine dominancerelations estimate for that season. and a dominancehierarchy among the five obligate The techniquedescribed above is basedon the as- ant-followingbird species. sumptionthat trails throughout the study plot are a fair Beginningin 1998, we outfitted focal birds that representationof the study plot as a whole.That as- were caught in mist-netswith radiotransmittersto sumptionseems reasonable, because trails were cut in monitoractivity away from E. burchelliswarm fronts gridform to provideaccess to all habitatsin thevicinity We used Holohil BD-2 1.3-g transmitters, which of the CochaCashu Biological Station (see Appendix 1 weighed<5% of thebird's body weight and had a bat- for a map of the trail system).Labidus praedator swarms tery life of ~60 days.We evaluatedthree attachment were sometimesseen moving down trails, which methods:gluing to thebird's back, wing harness, and would suggestthat the trail was influencingmove- leg harness(Raim 1978, Rappole and Tipton 1991, mentdirection. However, swarms larger than the trail Thompson1994). Gluing to the back (usingeyelash width did not constraintheir width to fit onto a trail, cementor "superglue")always failed within days, but spilledover the sidesof the trail into surrounding and wing harnesseswere often quickly removedby 14 ORNITHOLOGICAL MONOGRAPHS NO. 55 the birds. Rappole and Tipton's (1991) leg-harness locationpoints from sightingsat swarms,E. burchelli method worked well, and very few birds lost their bivouacs, mist-net records, and roost sites. Unlike radiotransmitters. In fact, three individuals that were generalistinsectivores, which may have preferential not recaughtby the end of the 2001 field seasonwere feedingsites within territories,obligate ant-followers still carrying transmitters14 months later, in April movethrough their homerange following local army 2002. One M. fortis male was recaughtthen and the ant coloniesand do nothave designated foraging sites transmitterremoved; his skinshowed no signof abcess within homeranges. Observations of individual birds or irritation, thoughthe cord was coveredwith shed at swarmsmay biasthe kernelmethod downward by skin aroundhis upper legs.That skin cameoff easil.• allowingit to placeseparate contours around high-use and the bird appearedhealthy and unharmedby hav- pointswithin a home-rangearea, thereby minimizing ing carriedthe transmitterfor a prolongedperiod. The the 95%-use area. That situation could arise if a bird leg-harnessmethod also seemed to causethe birdsthe was observedat two spatially separatedE. burchelli leaststress, because placement was carried out quickly coloniesbut also, without an observer'sknowledge, and did not irritate their skin. We followed birds with utilized an L. praedatorswarm located between the active radiotransmitters for 2- to 4-h blocks before and two E. burchellicolonies. In that scenario,the bird may afternoon; we collecteddata on their activitypatterns, have utilized the area between the two contours of use includinguse of nestingsites, roost sites, and L. praeda- at the E. burchelliswarms with the samefrequency as torswarms. When birds were not directlyobserved (i.e. within the contours, but was not detected there. I did during roosting),we used triangulationfrom three not calculateminimum convexpolygon (MCP) home pointsalong a trail separatedby -50 m each(depend- ranges,which include all areasbetween points as part hugon distancefrom observerto bird). I later calculated of a homerange, because that methodgenerally needs locationsof triangulatedsites using the computerpro- >150points for accuracy(Seaman and Powell1996). gram LOCATE II (Nams 2000). I used a two-way analysisof variance(ANOVA) to test whether home-rangesize varied acrossspecies STATISTICAL METHODS by year.I alsoused one-way ANOVA testson home- range data from each speciesto examinewhether Avian population-densityestimation.--I estimated there were significantdifferences in home-rangesize adult populationdensities for eachbird specieseach acrossyears. Models did not includecovariance to ac- year, using Bowden'smodel estimation (Bowden countfor individualsobserved in more than one year, 1993) in the program NOREMARK (White 1996). becausethe number of thoseindividuals per species From that information, I calculated number of indi- was low. viduals per 100 ha and 95% confidenceintervals for eachspecies. The Bowden model estimation uses data METHODSFOR SPECIFIC QUESTIONS from (1) numberof bandedbirds, (2) numberof sight- ingsof bandedbut unidentifiedbirds, and (3) number I. Doesbody size or relativecompetitive ability among of sightingsof unbandedbirds over a seasonto esti- speciescorrelate with home-range size?--Home-range or matetotal adult bird densityper speciesin the study territory sizeis the resultof multiple factors,mostly area, plus 95% confidenceintervals. That densityis related to population and resourcedensity, defend- not a measureof "territorial"birds per area,because ability, and patchiness(Brown 1964). Agonistic in- •t includes unsettled adult floaters as well as adults teractionswithin (Willis 1967) and between species with establishedhome ranges.I treated individuals (Robinsonand Terborgh1995) also contributeto a banded within the last two weeks of each field season bird's home-rangesize. Obligate ant-followershave as unbandedto accuratelyestimate the total adult been characterizedin the literature as holding over- population. lapping, non-exclusivehome rangesrather than de- Home-rangemethods.--I calculated home ranges fended territories (Willis 1967, 1973;Willis and Oniki of birds usingthe programANIMAL MOVEMENTS 1978).If territorial defenseoccurs, it is only near the (Hoogeand Eichenlaub1997) in ARCVIEW. I present nest site, as noted by Willis (1967). The five coexist- yearlydata for kernelhome ranges with samplesizes ing speciesat Cocha Cashu provide an opportunity per individualper seasonof >_20points. Seaman et al. to test competinghypotheses about home-range size. (1999) recommendthat researchersobtain a minimum Density of army ant coloniesmay limit minimal of 30 points,because kernel home-range estimates will home-range size, but relative mass of bird species overestimatehome-range size at small sample sizes. or relative competitiveability (or both) may also be Having >30 observationsof few individualsper year,I influential. wasunable to usethat cut-off;therefore, home ranges I proposedtwo competinghypotheses. Energetic may be biasedupward for someindividuals. I used requirementssuggest that, all elsebeing equal, larger least-squarescross-validation to selectthe smoothing animalsshould have larger home ranges than smaller parameterof the fixed kernel (Seamanand Powell (Schoener1968). If body mass and energy 1996),and report95%-use areas. I includedindividual requirementsdetermine home-range size for obligate OBLIGATE ANT-FOLLOWING BIRDS IN PERU 15

ant-followingbirds, we can predict that the largest phase (H611doblerand Wilson 1990). The nomadic specieswill havethe largesthome-range size and that phase--duringwhich antsforage all day,every day, home-rangesize will decreaseas bird-speciesmass and move their bivouac almost every night--offers decreases.This hypothesispredicts a large, approxi- reliablefood resourcesfor birds. During the statary matelyequal home-range size for D. merula,M. fortis, phase,however, the colony remainsin its bivouac and P. nigromaculata,with a progressionto smaller and doesnot forageevery day.Using Franks' (1982b) home-range size in R. melanostictaand G. salvini.An estimateof 13 foragingdays per stataryphase (62% importantassumption of thisprediction is that group of statarydays), I estimatedthat an E. burchellicolony size(solitary individuals, pairs, or family groups)will has a 77.1%probability of foragingon any given day not influencehome-range size. For the purposesof The function this prediction,I assumedthat one antswarmcould adequatelyfeed any of the abovegroup sizes.That Fect•on= 1 - (1 - .77)n (3.2) assmmptionseemed reasonable, given that the yearly mean number of birds at E. burchelliswarms ranged describesthe probabilitythat at leastone E. burchelh from 5.7 to 8.9 individuals, and the lowest mean for colony is foraging within a defined area, where n = number of birds at an L. praedatorswarm was 3.4 (see the number of coloniesin an area. For example,the bird numbersin Table3.13). probabilitythat at leastone E. burchellicolony is forag- Alternatively,I hypothesizedthat the dominance ing on a givenday is 77.1%for onecolony, 94.7% for hierarchyamong species influences home-range size. two colonies,and 98.8%for three colonies.The prob- Dominant speciesmay gather at highly productive abilityequations asymptote at approximatelythree E swarms,denying subordinate species access to them. burchellicolonies, so beyond that numberit is likely If subordinatesneed to samplemore swarmsto find unprofitableto follow more colonies(Fig. 3.1). sitesat whichto forageeffectively, we canpredict that The function theywill havelarger home-range sizes than dominant species.This hypothesispredicts that the small, sub- Fa•temaaw= 1.0 - Fec•to, (3.3) ordinatespecies R. melanostictaand G. salvinihave the largestaverage home-range sizes. predicts the percentageof days an obligate ant- II. Doobligate ant-followers minimize home-range size followermust rely on L. praedatorswarms because no tofit theminimum number of reliable food resources needed E. burchellicolonies are foragingin their home-range for dailyforaging ?-- In addition to the hypotheses area.For example,if a homerange encompasses two above,density of army ant coloniesmay be a major E. burchellicolonies, with a 94.7% chance that one is factor determininghome-range size in obligateant- foragingon a given day, a bird will need to find an followingbirds. The birds may attemptto minimize alternativefood source(i.e. an L. praedatorswarm) on home-rangesize by trackingjust enough army ant 5.3%of days. coloniesto ensuredaily foragingopportunities; that The factorsthat determinehow many E. burchelh minimum may depend on body mass,competitive coloniesa foragershould track also dependon the ability,or both.I assumedthat a bird must forageeach densityof thosecolonies in the landscape.If colonies day, and modeleda predictedminimal home-range are a limiting resource,their distributionmay deter- size using estimatedarmy ant densitiesfrom Cocha mine the minimum home-rangesize of an obligate Cashuand' informat/on on army ant foragingprob- ant-follower.The equation abilitiesper day.I comparedresults of the modelwith resultsobserved for eachspecies to gain insightinto homerangeminimu m = species-specificecological strategies and to test the # colonies bird needs to track predictionsof body size and interspecificdominance (3.4) (colonydensity/100 ha) on home-rangesize. First, I modeled the number of E. burchellicolonies a bird mustkeep track of per day to ensuredaily foragingopportunities. That model as- variesdepending on the number of coloniesa bird sumedthat obligate ant-following birds generally rely simultaneouslyfollows and the densityof E. burchelh on E. burchellifor foragingopportunities, and use L. coloniesin the area.For example,if colonydensity = praedatoronly as a secondarysource (Willis and Oniki 3.2 per 100 ha, as it does on Barro Colorado Island, 1978,Skutch 1996). I basedthat assumptionon differ- Panama(Franks 1982b), an individual would need an encesin reliability of the ant species.Obligate ant- averagehome rangeof _>90.9ha to encompassthree followers monitor the mobile bivouacs of E. burchelli colonies.If a 94.7%chance of foragingper day (or two acrossspace and time,but are unableto trackcolony colonies)is sufficientfor a bird at that ant density, movementsof L. praedatorants, because that species we can predict a minimum daily home-rangesize of doesnot have conspicuous,aboveground bivouacs. 60.6ha for obligateant-followers on BarroColorado A full cyclefor E. burchellilasts 35 days:14 days Thesepercentages provide general daily probabilities in the nomadicphase and 21 days in the statary of swarm availabilityto foragingbirds. Given that 16 ORNITHOLOGICAL MONOGRAPHS NO. 55

...... • ...... 7 ...... P•-•6t,Jf.i•b

0.8

0.6

0.4

0.2

0 1 2 3 4 Number of colonies

Fit. 3.1.The probability that >1 E. burchelliarmy ant colony is foraging,given the number of coloniesin an area, asymptotesat -3 colonies.Probability values are: 1 colony(77.1%), 2 colonies(94.7%), and 3 colonies(98.8%). statary ant-coloniesmay not swarm all day (S. K. themto correctfor statisticalsampling bias. All bird Willsonpers. obs.), 77.1% may be too high an hourly specieswere followed with radiotelemetryin approxi- probabilityof foragingfor any givenbird. However, matelyequal amounts. for the purposeof the presentanalysis, the general I calculatedrelative percentageof abundanceby probabilitiesshould reflect reality for any givenobli- dividing number of sightingsof a bird speciesby gateant-follower. total sightingsof all obligateant-followers, for each III. Do obligateant-following species differentially ant species.If all bird speciesproportioned their use utilizeEciton burchelli and Labidus praedator swarms similarlybetween E. burchelliand L. praedatorswarms, and, if so, in what proportions?--Ianalyzed relative we would expectthat eachbird specieswould make abundancesof eachbird speciesat L. praedatorand up thesame relative percentage with eachant species. E. burchelliswarms to determinewhether they used If any speciesvaried from that pattern,we should the two speciesof army ants differently.Only data see a correspondingshift in relative abundancesof from the 2000-2001 field season were used, because all other species.For example,if D. merulaused E. the number of L. praedatorswarm observationsin burchelliswarms more than L. praedatorswarms, rela- previousyears was small. I calculatedtotal numberof tive percentageof D. merulashould be higherwith E. sightingsof eachbird specieswith eachspecies of ant. burchellithan with L. praedator,and all other bird I was unableto lookfor heterogeneityacross the two speciesshould have inflatedrelative abundances at ant speciesindependently for eachbird species,be- L. praedatorswarms to accountfor the absenceof D, causeour observationsof theantswarms depended on merula.My analysisshowed that D. merulaindeed findingthe ants.We were consistentlyable to find E. had lower relative abundanceat L. praedatorswarms burchelliswarms, because we followed the coloniesev- comparedwith its presenceat E. burchelliswarms, ery day;but findingan L. praedatorswarm depended but there was no correspondingincrease in three of on followingradiotracked birds, hearingbird vocal- the other four species.Because those three species lzationsat activeswarms, or occasionallyhappening showed consistent abundances, I considered them upon an active swarm.Therefore, measurements of "normal" for the purpose of comparingthe other bird-speciesoccurrence at L. praedatorswarms are bi- two species(D. merulaand M. fortis) againstthem. asedby themethod of locatingthose swarms. Because GymnopithyssaIvini had the most consistentrelative G. salviniwas the mostevenly distributed of all bird abundancewith the two ant species(23.3% and 22.6%, speciesbetween swarms of thetwo ant species,I used respectively),and I calculateda frequencyvalue for OBLIGATE ANT-FOLLOWING BIRDS IN PERU 17 eachbird species,normalized to the frequencyof G. populationparameters, and werecalculated for 1998, salvinisightings. I comparednormalized abundances 1999,and 2000-2001.For hypothesis(1), the equation using chi-squaretests (Snedecor and Cochran1989). Calculations of normalized abundance allowed me to Predictedm/bird = avg.m/bird * Ms/• (3.6) determinewhether D. merulaand M. fortisutilized the two ant speciesdifferently from the other threebird (whereM s = massof birdspecies and • = weighted species.By normalizing to onebird species,I also veri- averageof bird massbased on populationparameters fied the assumptionthat G. salvini,P. nigromaculata, of five species)provides predicted space use by each and R. melanostictawere behaving similarly. bird speciesat a swarmfront, based on mass.For hy- IV. Are ratesof agonisticinteractions between birds pothesis(2), I predicteda lineardecrease in spaceuse differentat Labiduspraedator versus Eciton burchelli correspondingto a species'position in the dominance swarms?--Icalculated displacement rates per minute hierarchy,with P. nigromaculatausing the most space per obligate ant-follower at E. burchelliswarms for eachyear. For hypothesis (3), I predictedthat D. merula 1998,1999, and 2000-2001and at L. praedatorswarms spaceuse at a swarmfront would be withinthe upper for 1999 and 2000-2001. I calculated a two-way 95% confidenceinterval of the dominantyet same- ANOVA to explorerelationships among displacement sizedspecies P. nigromaculata.In hypothesis(4), I com- rates,ant species,and year. I calculateda one-way paredresults for eachspecies against the mean amount ANOVA to test whetherrates of agonisticinteraction of spaceavailable per bird at an antswarm.That esti- differedbetween years at E. burchelliswarms. A sec- mateis from the "observedspace available per bird at ond one-wayANOVA testedthe relationship between an E. burchelliswarm" (seebelow). Predictedtrends rateof displacementsper minute and number of birds werevisually compared with resultsfrom the 1998and per unit of swarm-widthat E. burchelliswarms. In all 2000-2001field seasons--years with highand low total cases,"sample size" refersto total number of swarm densitiesof ant-followingbirds, respectively. observationsused in the analysis. VI. Judgingfrom overall obligate-ant-follower densities, V. Doesbody size or competitiveability correlate with arebirds maximizing spatial utilization of availableEciton spaceutilization at a swarmfront?--I used a linear burchelliand Labiduspraedator swarms?--I use the regression(SAS Institute 1997) to estimate meters term "foraging space" to describethe amount of of spacean individual of each speciesused on an antswarmresource available to foragingbirds overa E. burchelliswarm front in the years 1998, 1999,and givenarea; this conceptallows estimation of the ob- 2000-2001.To estimatea "metersper bird" (mpbird) ligate ant-followingbird population'smaximization parameterfor eachbird species,I fitted the number of the availablearmy ant resources.I calculatedes- of birds of each speciesat each swarm front to the timatesof foragingspace at E. burchelliand L. praeda- equation for swarms for three field seasons: 1998, 1999, and 2000-2001. I estimated mean E. burchelli swarm width antswarm width = [mpbird from field data taken at swarms in each season. Mean (Phleg) * number of Phleg] + L. praedatorswarm width wasavailable from the 2000- Impbird (M.fort) + number of 2001season only; that width was usedin calculations M. fort] + Impbird (D.mer) * (3.5) for all years. Likewise,density of abovegroundL. number of D.mer] + Impbird praedatorswarms was estimated only in the 2000-2001 (Rheg)* numberof Rheg]+ Imp- season,and that estimatewas used for all years. bird (G.sal)* numberof G.sal] Foragingspace for each army ant speciesper year was calculatedusing the averageswarm width, the I estimatedamount of spaceeach specieswas estimatedcolony (for E. burchelli)or swarm (for L. predictedto usealong an E. burchelliswarm front on praedator)density per 100ha, and a generalprobability the basis of four hypotheses.Space use at a swarm that a colonyforaged on a given day (Franks1982b). frontmay be influencedby (1) bodymass, with larger Probabilityof foragingper day is 1.0in L. praedator,be- speciesusing more spacethan smaller species;(2) causethe densitymeasurement specifically estimated dominance,with dominantspecies using more space activeaboveground swarms, not colonies.I used the than subordinatespecies; or (3) ability to maximize followingcalculations to generatethe data presented spaceby choosinga swarm with fewer competitors. in Table3.13 (lettersin parenthesescorrespond to let- Hypothesis(3) is specificto D. merula,which may be tersin thattable). I estimatedtotal foraging space avail- physiologicallyable to samplemore swarmsbecause able (i) to birds per year by summingthe estimates of its longerwing-shape and better flying ability as of foragingspace for E. burchelli(d) and L. praedator comparedwith the antbirds(S. K. Willsonpers. obs.). (h) for a given year. Mean number of obligate ant- Lastly (4, null hypothesis),space use alonga swarm followingbirds per E. burchellior L. praedatorswarm front may not be affectedby the above factors,and was estimatedfor a given swarm when bird pres- insteadmay be similar for all species.Predictions for encewas >0 (i.e. swarmswith no birds presentwere each hypothesisdepend on yearly conditionsand not includedin the estimate).I estimatedamount of 18 ORNITHOLOGICAL MONOGRAPHS NO. 55 foragingspace available per bird if all birdsused only 1971,E. O. Willis pers.comm.). However, an es- E. burchelliswarms by dividingthe total availableE. timateof 19.9swarms per 100ha over25 daysin burchelliforaging space per 100ha (d) by the total obli- March-April2002 may indicatethat thereis less gateant-following population per 100 ha (j).I performed variabilityin L. praedatordensity than is generally the samecalculations with L. praedatordata to estimate assumedto occurbetween wet and dry seasons. foragingspace available per bird if all birdsused only L. That estimate coincides with the transition into the praedatorswarms (h/j). I estimatedthe observedspacing perbird at swarmsby dividingthe meanswarm width dry season,when rainfallwas scarce.However, by themean number of obligateant-following birds per the groundmay havebeen wet enoughto affect speciesof swarm,for eachyear (b/n, f/q). To determine subterraneanL. praedatorforaging. Because most themaximal space a bird couldobtain at a swarmfront of my observationswere made during the rainy tf all obligateant-followers maximized their use of the season,the estimateof 20.7active swarms per 100 two ant species,I calculated a measureof yearlymaxi- ha is used in all further calculations. mal foragingspace. Total foraging space available per 100ha (i) wasdivided by thetotal population of obligate ADULT AVIAN POPULATION DENSITIES ant-followersper 100 ha (j). Finally,I askedhow well birds maximized their use of the swarm resources each year.I estimatedthe percentage of maximizationof total Total adult populationdensity per 100 ha for swarmresources twice: first, by dividingthe observed the five obligateant-following species varied spacingat E. burchelliswarms (o) by the maximum between45 and 86.6 individualsper year, and potentialforaging space per bird (m); and second,by declinedby ~25%each year (Table 3.3). However, dividingthe observedspacing at L. praedatorswarms (r) individualspecies generally did not showa con- by themaximum potential foraging space per bird (m). stantdecline throughout the studyperiod. Based VII. Canspace at swarmsbe a limitingresource for ob- on confidence intervals, estimated densities of P. ligateant-followers? Answering that questiondoes not nigromaculataand R. melanostictasignificantly de- requirespecific methods, but involvesa synthesisof clined between 1998 and 1999, and R. melanosticta datagathered while answeringother questions. and G. salvini declined between 1999 and 2000. Eachyear, relative abundance per speciesvaried RESULTS with changesin populationdensities; and each year,a differentspecies was relatively most com- ARMY ANT COLONY DENSITY mon on the studyplot: in 1998,P. nigromaculata; in 1999, G. salvini;and in 2000-2001, D. merula. Mean E. burchellidensity during the study period was 4.9 coloniesper 100 ha. In 1998, DOMINANCE RELATIONSHIPS AMONG BIRD SPECIES E. burchellicolony density per 100 ha averaged 4.4 coloniesover a 10-weekperiod (Table 3.1). In For speciesthat differ in size, mass largely 1999,density averaged 5.2 colonies over 13 weeks; determineda species'rank in the strict domi- in 2000-2001,density averaged 5.0 coloniesover nancehierarchy that existsamong the antbirds 15 weeks.There were no significantdifferences in (Willis and Oniki 1978).Social system, as well, colonydensity among years. appearedto affecthierarchical placement when ! estimated20.7 activeaboveground L. praeda- specieswere, on average,equal in mass. In tor swarmsper 100 ha for the 2000-2001field contrastto the antbirds,the placementof the season,which fell exclusivelywithin the rainy one woodcreeperspecies (D. merula)was less season(Table 3.2). Densitymay be lower in the clear.Although it had the highestmass of the dry season,when L. praedatorcolonies do more five species(see Table 2.1), D. merulainteracted foragingbelow groundand at night (Schneirla most with R. melanosticta,which is third in the

T,•sr• 3.1. Colony densitymeasurements of Ecitonburchelli army ants at CochaCashu Biological Station, Peru, 1998-2001.A one-way ANOVA found no differencein meansbetween years (F = 0.527, df = 2 and 35, P = 0.60).

Year Estimate SD SE Area surveyed Period 1998 4.4per 100ha 2.27 0.76 50 ha 10weeks 1999 5.2per 100ha 2.39 0.69 50 ha 13weeks 2000-2001 5.0per 100ha 1.20 0.35 100ha 15weeks Mean 4.9 per 100ha OBLIGATE ANT-FOLLOWING BIRDS IN PERU 19

TAELE3.2. Density measurementof aboveground were of D. merula(which weighs an average foraging swarmsof Labiduspraedator army ants 47.5 g)--the highestdisplacement rate toward at Cocha Cashu BiologicalStation, Peru, from D. merulaby any of the four antbird species. October2000 to February2001 (n = 126 days)and Dendrocincla merula, on the other hand, was from March to April 2002 (n = 25 days). almostexclusively intraspecifically aggressive; Month(s) Estimate SD SE only 10%of its displacementswere of antbirds, October 2001 26.1/100 ha - - whereas90% were of conspecifics.Dendrocincla November 2001 20.4/100 ha - - merula woodcreeperspreferentially cling to December 2001 14.7/100 ha - - large trunksabove advancing swarms (Willis January2001 22.1/100ha - - 1978)and are lessapt to contestperch sites with February2001 20.1/100ha - - the obligateantbirds. However, they regularly October-February fly to theground to catchfleeing prey 2001average 20.7/100ha 11.03 2.60 and will perchon the groundamong antbirds March-April 2002 19.9/100ha - - while foraging.At thebottom of the dominance hierarchyis G. salvini,which weighs 25.9 g antbird dominancehierarchy and averages16 on average;90% of its displacementswere of g smallerthan D. merula(Table 3.4). The domi- conspecifics.As with R. melanosticta,multiple nant speciesis P. nigromaculata,a 46.0-g (on pairsmay gatherat an antswarm,and same-sex average)bird that lives in family groups of aggressionbetween pairs was high. two to five individuals and is extremely ag- gressiveinterspecifically, with 69% of recorded AVIAN HOME-RANGE SIZE displacementsbeing directedat other obligate ant-followingspecies. Myrmeciza fortis is the Home-rangesize varied widely acrossspecies secondspecies in the dominancehierarchy. and over time (Table 3.5). Mean home ranges Although it averaged46.5 g, 0.5 g larger than acrossall years, using individualswith >_20 P. nigromaculata,it is subordinateto that speciesand directed only 2% TABLE3.3. Estimatednumber of individualsper 100 ha for five of displacementstoward it. Eighty obligate ant-followersin Amazonian Peru. Estimateswere calculatedusing Bowden'smodel estimationin the program percent of recordeddisplacements NOREMARK.Population data are from 1998,1999, and 2000- by M. fortis were aimed at the two 2001.Size of studyplot variedby year (162ha in 1998,277 ha in smaller antbird species.Myrmeciza 1999,271 ha in 2000-20001);CI = 95%confidence interval. fortis is extremelyintraspecifically intolerant, and more than one pair Estimate Relative never foragedtogether at the same Species per 100ha CI 100ha abundance(%) swarm. As an M. fortis individual 1998 approachesa swarm,it singsloudly, D. merula 19.8 16.0-24.1 22.9 21.0 16.7-26.5 24.2 and will retreat when answeredby P. nigromaculata 9.3 3.1-27.2 10.7 a bird other than its mate. When M. fortis R. melanosticta 16.7 14.2-19.8 19.3 two pairs found themselvesat the G. salvini 19.8 14.8-27.2 22.9 same swarm, fighting always en- Total 86.6 sued until at least one pair left the 1999 area. Myrmecizafortis was the only D. merula 15.2 13.4-17.7 22.9 obligate speciesat Cocha Cashu P.nigromaculata 13.0 11.6-14.8 19.5 that defended multipurpose ter- M. fortis 10.1 7.9-12.3 15.2 ritories (see below). Rhegmatorhina R. melanosticta 10.1 8.3-11.9 15.2 melanosticta is the third-smallest G. salvini 18.1 13.4-24.5 27.2 antbird (average 31.4 g) and the Total 66.5 third antbird in the dominance hi- 2000-2001 erarchy.Only 2% of displacements D. merula 14.0 11.1-18.1 30.6 by R. melanostictawere directedat P.nigromaculata 10.7 8.9-13.3 23.4 the two largerantbirds; unlike those M. fortis 8.5 5.9-11.8 18.6 species,it directedmost aggression R. melanosticta 4.8 3.3-6.6 10.5 (52%)at conspecifics.However, 16% G. salvini 7.7 6.6-9.2 16.9 of displacementsby R. melanosticta Total 45.7 20 ORNITHOLOGICAL MONOGRAPHS NO. 55

TABLB3.4. Ratesof displacementamong five obligateant-following bird speciesin AmazonianPeru. Rows representproportion of totaldisplacements directed at eachof thefive species(in columns)by theaggressor species(first in row). Samplesize (n) of total displacementsper speciesis shownin the lastcolumn.

P.nigrornaculata M. fortis D. rnerula R. rndanosticta G. salvini (46.0g) (46.5g) (47.5g) (31.4g) (25.9g) n P. nigromaculata 0.32 0.05 0.09 0.45 0.10 267 M. fortis 0.02 0.11 0.07 0.53 0.27 115 D. merula 0.01 0.00 0.90 0.03 0.06 463 R. rndanosticta 0.01 0.01 0.16 0.52 0.31 444 G. salvini 0.01 0.00 0.02 0.07 0.90 154 Total n = 1,443 observations within a season, were: D. merula = foragingstrategy in each of the four antbirds, 64.6ha (n = 20, SD = 33.8);P. nigromaculata= 45.9 but may havebeen less important for the wood- ha (n = 20, SD = 30.9);M. fortis= 15.4ha (n = 4, creeperD. merula. SD = 1.1); R. melanosticta= 50.8 ha (n = 24, SD = 41.5); and G. salvini= 27.5 ha (n = 21, SD = 13.2). DISCUSSION Thoseaverages, however, mask the strongeffect of yearon homeranges (ANOVA: F = 12.56,df = ECOLOGICAL RELATIONSHIPS AMONG BIRD SPECIES 2 and 86, P < 0.0001). In 1998 and 2000-2001, home-range size was similar within species; I. Doesbody size or relativecompetitive abil- but in 1999, all study speciesincreased their ity amongspecies correlate with home-range size?- averagehome-range sizes by 19-59 ha (Fig. 3.2; Therewas little supportfor the hypothesisthat M. fortishome-range size was not estimatedfor home-rangesize was dependenton body mass. 1999 because there were too few observations The three large ant-followersof equal size (P. per individual). Becauseof unequal variances nigromaculata,M. fortis, and D. merula)held strik- and large standarddeviations in D. merulaand ingly differentmean home-range sizes (Fig. 3.2). G. salvinidata, only P. nigromaculataand R. mela- (In 1999, mean home-rangesizes of all spedes nostictaestimates are stafisticallysignificant for were greatlyinflated, and no patternsbetween an increasein 1999 (ANOVA, P. nigromaculata:species were evident.See Chapter 4 for discus- F = 38.53, df = 2 and 17, P < 0.001; R. melanost- sionof 1999.)Home ranges of D. merulaaveraged icta:F = 3.14, df = 2 and 21, P = 0.06). For at least more than 3-4x as large as thoseof the equiva- one year per antbird species,mean home-range lently sizedM. fortis.Average M. fortis home- estimates were below the minimum needed range size was consistentlysmaller, not only to ensure that one of two E. burchelli colonies than thoseof the two speciesof equalmass, but would be foragingin a home-rangearea on a thanthose of all obligateant-followers, including given day (minimum = 40.8 ha, at 94.7%daily G. salvini,whose mass is only half that of M. for- foraging probability; see Equation 3.4). Those tis. Only the speciespair R. melanostictaand G. resultssuggest that foragingwith L. praedator salvinishowed a trend correlatingsmaller home- army antsmay be an importantcomponent of rangesize with a decreasein body mass.

TABLE3.5. Estimatesfor 95% kernelhome range (with samplesize and SD) for adult obligateant-following bird speciesin 1998,1999, and 2000-2001.Home rangeswere estimatedusing the ANIMAL MOVEMENTS programfor ARCVIEW.Estimates include only individuals with >20sightings per season.

Mean home range Total Overall 1998 1998 1999 1999 2000- 2000- Species (ha) n SD (n) SD (n) SD 2001(n) 2001SD P.nigromaculata 45.9 20 30.9 26.83(6) 5.74 86.01(5) 13.70 24.86(9) 13.18 M. fortis 15.4 4 1.1 15.38(1) - - - 15.50(3) 1.31 D. rnerula 64.6 20 33.8 49.42 (6) 26.84 77.78 (7) 44.51 66.59 (7) 24.56 R. rnelanosticta 50.8 24 41.5 38.00(10) 20.10 75.63(11) 52.80 29.26(3) 14.10 G. salvini 27.5 21 13.2 20.21 (5) 4.37 39.32 (4) 25.83 22.85 (12) 6.10 OBLIGATE ANT-FOLLOWING BIRDS IN PERU 21

140

120

1 oo

• 8o IQ1998 11999 [] 2000 6O

40 1I IT

2O

0 7 6 tO. I 3 5 12

D. rnerula P nigromaculata M. fortts R. rnelanosticta G. salvmi species

Fro. 3.2. Mean adult home-rangesize (95% kernel) for five speciesof obligate ant-following birds in Amazonian Peru over three years,with SD. Home rangeswere estimatedusing the ANIMAL MOVEMENTS program for ARCVIEW. Samplesizes per year are shown within eachbar; only individuals with _>20 data points are included.

There was support for the hypothesis that were not explained by either hypothesisand placementin the interspecificdominance hi- will be furtherexplored below. erarchy affectshome-range size. The greatest il. Do birdsminimize home-range size tofit the amount of interspecificaggression among the least mnnberqf food resourcesneeded?--Animals five bird specieswas directedat R. melanosticta use valuable energy stores when moving be- (Table 3.4). The two dominant species,P. nigro- tween patchy food resources(Charnov 1976, maculataand M. _fortis,displaced R. melanosticta McNair 1982, Sutherland and Parker 1985). In more than any other species.In both 1998and the caseof obligateant-following species, which 2000-2001, R. melanosticta individuals held, on generally do not defend exclusiveterritories, average,larger home rangesthan P. ,igromacu- food acquisitionand positionin the dominance lataand M.fortis individuals(Table 3.5). hierarchymay act synchronouslyto determine Although home ranges of the subordinate home-rangesize. In termsof energeticdemands, speciesG. salviniwere not larger than thoseof a minimum preferred home-rangesize may be R. melanosticta,the former speciesreceived only one that holdsa minimum numberof army ant 31% of the latter's aggression.Rhegmatorhina colonies providing some acceptableprobabil- •nelanostictais primarily intraspecificallyaggres- ity of foraging success.We have seen that the sive, and may not have much influenceon the dominancehierarchy affectshome-range size, movement patterns of G. salvi,i. Gymnopithys but how do we predictthe smallesthome-range salvini consistentlyheld larger home ranges si7ean obligateant-follower should have? Here, than those of the largest antbird, M. fortis. In I askwhether colony density of E. bnrchelharmy sum, there was little support for the hypothesis antsdetermines home-range size. That question that body size affectsrelative home-rangesize is especiallypertinent to the dominant species, and somesupport for the dominance-hierarchy P. •tigromaculata,which is not constrained in hypothesis.However, mean home-rangesizes home-rangesize by other guild members. for D. merula,which were consistentlylargest, Using a mean of 4.9 E. burchellicolonies per and M. fortis, which were consistentlysmallest, 100 ha (Table 3.1) in the minimal-home-range 22 ORNITHOLOGICAL MONOGRAPHS NO. 55 equation(Eq. 3.4), we find thatif an obligateant- merulais presumablydivided between foraging following bird dependedsolely on E. burchelli with L. praedatorants and with peccaries. and tracked two colonies,it would require a Usingdata from 2000-2001, I estimated that for minimumhome range of 41 ha at CochaCashu P.nigromaculata, an average home range of 25 ha to provide a 94.7%chance that oneof the colonies will encompass1.2 E. burchellicolonies and 5.2 L. wasforaging on a givenday. If the individualre- praedatorswarms (Table 3.6). That meansthat on quiredthe greaterassurance of oneof threecolo- any givenday, there is a 17%chance that no E. niesforaging on any givenday, or a 98.8%chance burchellicolony will be foraging.Thus, assuming of foraging,the minimumhome range would be that individual birds must forageevery day, P. 61 ha. An alternativehypothesis is that obligate nigromaculatais predictedto have foragedwith ant-followersdo not need multiple E. burchelli L. praedatoron >17%of daysin the 2000-2001sea- colonieswithin their home range. At Cocha son. The smallesthome range was consistently Cashu,a homerange of 20 ha wouldprovide an that of M. fortis.Using the samelogic as above, obligateant-follower with a 77.1% probability thatspecies was found to havehad an averageof thatthe colonyis foragingon a givenday. 0.8E. burchellicolonies within its homerange and In the 1998 and 2000-2001 seasons, the four to haveforaged with L. praedatoron >_29%of days antbirdspecies all heldaverage home-range sizes in the 2000-2001 season. less than the minimum 41 ha needed to track two Data from 1998were generallysimilar to re- E. burchellicolonies (Fig. 3.2). Only R. melanostictasuits from the 2000-2001 season(Table 3.6). In came dose to the predictedminimum for two 1999, becausehome rangesof all specieswere colonies,with an averagehome-range size of 38 much larger than in other years,expected per- ha in 1998.Average home-range size per yearfor centageof dayswhen no E. burchelliant colonies the woodcreeperD. merulawas greater than that were foragingwas much lower. Only G. salvini, needed to include three E. burchelli colonies in with a meanhome-range size of 39.3ha in 1999, theirhome range in all yearsexcept 1998. As stat- was predicted to need a food resourceother ed above,all homeranges in 1999were greatly than E. burchellion >1%of days.However, much inflated;that inflationand its probablecause are E. burchelliforaging in 1999 was aboveground further discussedbelow. These results suggest in vegetation,because the study plot was inun- that, at leastfor the antbirds,E. burchellicolony dated with rainwater--makingthose swarms densityis notthe only factor driving home-range lessavailable to foragingbirds (see below). size.Below, I examinewhether home ranges are BecauseI was unable to monitor all L. praeda- smallerthan predictedbecause of bird use of L. tor swarms in the study area as I did for E. praedatorswarms. burchelli,it is difficult to determinean average III. Do bird speciesdifferentially utilize Labidus percentageof time that each ant specieswas praedatorversus Eciton burchelli as a foraging used by the bird species. However, radio- resource?--Foreach bird species,I determined tracked birds provide evidencethat utilization the mean number of E. burchelli colonies and of L. praedatoris an important aspectof the L. praedatorswarms within the averagehome study bird species'foraging ecologies(Table range per year. Using equations3.1 and 3.2, I 3.7). Two M. fortisindividuals were trackedex- calculatedthe percentageof time a bird species tensivelyin 2000-2001(a male and a femaleon should devote to each ant speciesif E. burchelli separateterritories). While being radiotracked, is the preferred resource(Table 3.6). By calcu- the male was observedforaging 13 times at lating the percentageof days that E. burchelli temporally or spatially separatedswarms, 54% would not be available,I assumedthat obligate of which were L. praedator.The femaleM. fortis ant-followers must use L. praedatorswarms on was tracked to 19 swarms, 37% of which were thosedays. No obligateant-following antbird L. praedator.Those numbers are greaterthan the was ever seenforaging away from army ants. 29%use of L. praedatorpredicted on the basisof It is unknown what percentageof the time averagehome-range size and preferencefor E. the woodcreeperD. merulaused white-lipped burchelliarmy ants.Predicted home-range sizes peccaryherds, but I expectthat percentageto be assume that a bird can move from one side of small,because those herds are spatiallypatchy its homerange to the otherto effectivelybenefit andhighly mobileat CochaCashu. The percent- from the number of E. burchelli swarms its home age of daysthat E. burchelliis unavailableto D. range contains,and that assumptionis valid. OBLIGATE ANT-FOLLOWING BIRDS IN PERU 23

TABLE3.6. Average home-range size per bird speciesper year allowsa predictionof the number of foragingopportunities available each day with eachof the two army ant species,Eciton burchelli(E. b.) and Labiduspraedator (L. p.). The percentageof days when E. burchelliis unavailableprovides a predictionof thepercentage of daysa speciesshould attend L. praedator swarms(assuming that birds will preferentiallyuse availableE. burchellœswarms).

Average E.b. Numbera Numbera % of days % of days homerange density expected expected E.b. E.b. Species (ha) per 100ha with E.b. with L.p. available unavailable 1998 P. nigromaculata 26.83 4.4 1.2 5.6 82.9 17.1 M. fortis 15.38 4.4 0.7 3.2 64.3 35.7 D. merula 49.42 4.4 2.2 10.2 96.0 4.0 R. melanosticta 38.00 4.4 1.7 7.9 91.8 8.2 G. salvini 20.21 4.4 0.9 4.2 73.4 26.6 1999 P. nigromaculata 86.01 5.2 4.5 17.8 99.9 0.1 M. fortis - 5.2 .... D. merula 77.78 5.2 4.0 16.1 99.7 0.3 R. melanosticta 75.63 5.2 3.9 15.7 99.7 0.3 G. salvini 39.32 5.2 2.0 8.1 94.7 5.3 2000-2001 P. nigromaculata 24.86 5.0 1.2 5.2 82.9 17.1 M. fortis 15.50 5.0 0.8 3.2 71.2 28.8 D. merula 66.59 5.0 3.3 13.8 99.2 0.8 R. melanosticta 29.26 5.0 1.5 6.1 89.0 11.0 G. salvini 22.85 5.0 1.1 4.7 80.1 19.9 • Of foragingopportunities.

TASLE3.7. Frequency and relative use of Ecitonburchelli and Labidus praedator swarms by individualradiotracked birds.All individualslisted had separatehome ranges (none are matesor sharethe samehome range).

% with % with Species ID Sexa n (swarms) E. burchelli L. praedator Year P. nigromaculata g-st U 5 60.0 40.0 2000-2001 P.nigromaculata y-st M 20 80.0 20.0 2000-2001 P. nigromaculata so-b U 11 45.5 54.5 2000-2001 P. nigromaculata oo-s U 7 28.6 71.4 2000-2001 M. fortis r-sg M 14 46.2 53.8 2000-2001 M. fortis g-sy F 19 63.2 36.8 2000-2001 D. merula wp-s F 22 81.8 18.2 2000-2001 D. merula s-pr F 14 64.3 35.7 2000-2001 D. merula rg-s F 19 73.7 26.3 2000-2001 R. melanosticta gp-s F 56 76.8 23.2 2000-2001 R. melanosticta r-st M 8 87.5 12.5 2000-2001 R. melanosticta s-yr M 20 80.0 20.0 1999 G. salvini go-s M 23 43.5 56.5 2000-2001 unknown, M = male, F = female.

Individualsof all five bird specieswere observed predictionfrom home-rangeestimates (Tables attending E. burchelliswarms at the opposite 3.7 and 3.8). In P. nigromaculata,individual use extremesof their home ranges on consecutive of L. praedatorswarms ranged from 18.2% to days (S. K. Willson unpubl. data). All species 35.7%, whereas predicted use from 2000-2001 displayedsimilar patterns of L. praedatorswarm home-range data was 17%. In R. melanosticta, use (percentageof total swarms at which bird individual use of L. praedatorranged from 12.5% was observedwhile radiotracked),in that indi- to 23.2%,whereas predicted use was 11%. The vidual bird use of that ant speciesexceeded the one male G. salvini that was radiotracked used 24 ORNITHOLOGICAL MONOGRAPHS NO. 55

TABLE3.8. Relativeabundance and normalizedfrequencies of five coexistingobligate ant-following bird speciesat Labiduspraedator and Eciton burchelli antswarms in AmazonianPeru. Percentages were calculated from all adult sightingsat each species'swarms over one field seasonin 2000-2001.Frequencies are normalizedto thoseof G. salvinito correctfor inherentdifferences between E. burchelliand L. praedator swarms.Abundance = percentageof obligatepopulation at swarmsof each ant species;frequency = number of individuals at a swarm for each individual of G. salvini; n = observed number of swarms of eachant species.

L. praedator(n = 91) E. burchelli(n = 247) Abundance Frequency Abundance Frequency X2 value • df P-value b P. nigromaculata 31.40% 1.35 29.60% 1.31 0.20 1 0.657 M. fortis 19.90% 0.85 12.00% 0.53 24.36 1 <0.001 D. merula 9.30% 0.40 20.80% 0.92 67.19 1 <0.001 R. melanosticta 16.10% 0.69 15.00% 0.67 0.10 1 0.757 G. salvini 23.30% 1.00 22.60% 1.00 NA 1 NA TheX 2 values were calculated using frequency normalized to totalG. salvinisightings. valuecrit= 3.841

L. praedatorswarms 56.5% of the time; estimated burchelli swarms?--Mean rates of agonistic usebased on home-rangesize was 20%. Finally, interactions,as describedby displacementsof three individual D. merulawoodcreepers used perch sites,varied betweenyears and between L. praedatorswarms 18% to 36%of the time,even ant species.Mean displacementrate per indi- though predicteduse was <1%. In effect,those vidual at E. burchelliswarms was highest in resultsemphasize the importanceof L. praeda- 1999 (x = 0.0054) and lowest in the 2000-2001 tor as a food resourcefor obligateant-followers, season(x = 0.0018)(Table 3.9). Rate of displace- and provide evidencethat individual birds are mentsper minute per obligateant-follower was choosingto forageopportunistically at spatially dependenton year (P = 0.049),but neither spe- unpredictableL. praedatorswarms rather than cies of army ant nor speciesx year interaction temporally and spatially reliable E. burchelli was significant(1998 was not includedbecause swarms. Below, I explore whether individual there were too few L. praedatorsamples in that bird specieshad an army ant speciespreference. year). Year was highly significantwhen rates I alsoexplore how ratesof agonisticencounters of displacementwere comparedamong only E. differedbetween ant speciesand betweenyears burchelli swarms in 1998, 1999, and 2000-2001 of high and low bird-density. (P < 0.001). Confidence intervals for the three Data from 339 antswarms observed in the field seasonsshow very little overlap (Table 2000-2001 field seasonwere analyzed to de- 3.10).Not surprisingly,the relationshipbetween termine relative abundance and normalized rate of displacementsper minute and number frequency of each bird speciesat E. burchelli of birds per unit of swarm width at E. burchelli (n = 247) and L. praedator(n = 91) swarms(Table swarms was significant (P < 0.001). In contrast 3.8). Phlegopsisnigromaculata, G. salvini,and R. to 1998 and 1999, during the 2000-2001season melanostictaall used the two ant speciesin the the rate of agonisticinteraction depended on samewa3• judging from normalizedfrequencies. how many birds were presentat a swarm (Table More than thosethree antbird species, the wood- 3.11).The lessspace individual birds had along creeper D. merula favored E. burchelli(20.8% the front, the more they fought.When popula- relative abundance at E. burchelli swarms versus tion density was higher during 1998 and 1999, 9.3% at L. praedatorswarms). Myrmeciza fortis the rate of agonistic interactions was higher, used L. praedatorswarms more than the other no matter how many birds were at a particular birds (19.9% relative abundanceat L. praedator swarm front. An interpretation of those data and 12% at E. burchelli).Both D. merulaand M. is that the rate increasedto the point where fortisshowed statistically significant differences it ceasedto be dependent on the birds imme- in ant-speciesuse, compared with the otherthree diately present at a given swarm. Rather than bird species(X2: P < 0.001;Table 3.8). fighting for more foraging spaceat the swarm, IV. Are ratesof agonisticinteractions between the birds seemedto engagein generalcompeti- birdsdifferent at Labiduspraedator versus Eciton tion againstany subordinatebirds present,even OBLIGATE ANT-FOLLOWING BIRDS IN PERU 25

TABLE3.9. Averagedisplacement rate per minute per obligateant-follower at Ecitonburchelli and Labidus praedatorswarms (in bold) differsby year and by ant species."Avg. min" is the averageamount of time a swarmwas observed, "Avg. displ. obs. TM is the averagenumber of displacementsper observation,"Avg. no. obligates"is the averagenumber of obligateant-followers at a swarm,"Displ. min TMis the averagerate of displacementsper minute,and "Displ.min I obl.-1" is the averagerate of displacementsper minuteper obligateant-follower present at a swarm.

1998 1999 2000

E. burchelli E. burchelli L. praedator E. burchelli L. praedator Total displacements 549 606 14 356 44 Avg. min 79 85 53 74 53 SD min 36.6 37.3 50.0 45.4 31.9 Avg.displ. obs. i 3.50 3.67 0.70 1.50 0.58 SD displ.obs. • 5.96 6.34 1.49 3.85 1.61 Avg. no. obligates 9.41 7.81 4.05 5.69 3.28 SD obligates 5.55 5.09 1.93 3.74 2.13 Displ.min -1 0.0371 0.0392 0.0139 0.0148 0.0088 SD displ.min -1 0.0479 0.0637 0.0298 0.0334 0.0257 Displ. min-• obl? 0.0034 0.0054 0.0028 0.0018 0.0018 SD displ.min -• ob•• 0.0041 0.0139 0.0068 0.0036 0.0048 Number of swarms(n) 157 166 20 238 76

TABLE3.10. Displacements per minute per ant- and (4) birds divide swarm-frontspace evenly following bird (D min • bird •) at Ecitonburchelli among species. swarms, with 95% confidence intervals (CI). Data from three seasons were examined Overallmodel using ANOVA weightedfor unequal (1998, 1999, 2000-2001), and a linear fit was varianceswas significant (F = 12.588,df = 2 and 292, possible for two of those seasons(1998 and P < 0.001). 2000-2001).A linear fit was not biologically D min-1 meaningful for 1999, because the fit resulted Year n bird -• SD 95%CI in negative estimatesfor some species;results 1998 157 0.003448 0.00414 0.002160-0.00473 for that year are not discussedhere. For both 1999 166 0.005398 0.01386 0.004150-0.00665 the 1998 and 2000-2001 field seasons,number 2000 238 0.001759 0.00364 0.000715-0.00280 of birds of eachspecies at an E. burchelliswarm front correlated well with that swarm's width if there was plenty of spaceat that particular (r2aaj.>0.66, P < 0.001in each year). Those results swarm. indicatethat individuals of each speciesused V. Doesbody size or competitiveability correlate consistentamounts of spacewithin eachof the with spaceuse at a swarmfront?--I tested four two years. hypotheses regarding swarm-front utiliza- Estimated spaceuse ranged from 0.51 to 1.0 tion: (1) space use is directly related to bird m per speciesin 1998and from 0.73 to 1.9 m per speciesmass; (2) space use is related to rank speciesin 2000-2001 (Fig. 3.3). In 1998, space in the dominancehierarchy; (3) woodcreeper use was similar among the three speciesof spaceuse is significantlyhigher than that of the equal mass (P. nigromaculata,M. fortis, and D. antbird species,because its longer wings and merula),at 0.87, 1.0, and 0.99 m, respectively. strongerflying ability allow D. merulato sample Thoseresults provide some support for hypoth- multiple swarmswith lessenergy expenditure; esis (1) (Table 3.12). The two smaller antbirds,

TABL•3.11. Displacements per minuteper bird for threeyears, grouped by birdsper meter of antswarm front.

Birdsm • 1998 n 1999 n 2000 n All years n _<0.5 0.0012 21 0.0054 48 0.0008 104 0.0021 173 >0.5-1.0 0.0048 31 0.0047 57 0.0023 100 0.0034 188 >1.0-1.5 0.0041 39 0.0045 21 0.0025 33 0.0036 93 >1.5 0.0036 43 0.0049 28 0.0042 35 0.0041 106 26 ORNITHOLOGICAL MONOGRAPHS NO. 55

R. melanostictaand G. salvini,used less space, at and overlappedslightly with thosefor M. fortis. 0.51 and 0.69 m, respectively.In oppositionto Confidenceintervals for D. meruIaoverlapped hypotheses(1) and (2), the smallestand most slightlywith thosefor M. fortis,but were above subordinatespecies, G. saIvini,used 0.18 m more thosefor all otherbird species. spacethan R. meIanosticta.With 95%confidence Increasedfront-space per species in 2000-2001 intervals,all speciesoverlapped in swarm-front correspondedwith a drop in both intra- and in- useestimates, though D. meruIaand R. melanost- terspecificcompetitive pressures (see Table 3.9), ictaoverlapped only slightly. becausetotal densityof ant-followingbirds was In 2000-2001, all speciesused more space almosthalf what it had been in I998 (Table3.3). along swarmfronts than in 1998,and two of the Higher support for spaceuse based on body four hypotheseswere supported(Fig. 3.3 and massin 2000-2001than in 1998may be related Table3.12). Space use along a front correspond- to the populationdecline over that period. In ed with body mass (hypothesis1) in all five 1998, all antswarms held the maximal number species.Woodcreeper D. meruIaused the most of birds observedover the study period, and spacealong the front, 1.9 m--an estimateeven birdsmay not havebeen able to maximizespace greaterthan that predicted using the upper 95% use because of the sheer numbers of individu- confidence interval of the dominant antbird P. als per swarm.That also providesan explana- nigromacuIata--whichsupports hypothesis (3). tion for why D. meruIa,in 1998,was unable to The two antbirdsof equal size,P. nigromaculata"choose" swarms that offeredmore spacealong and M. fortis, used similar amountsof space the swarmfront; all swarmswere maximallyat- (1.25and 1.32m, respectively).Hypotheses for tendedin thatyear. size and dominancehierarchy were supported VI. Judgingfrom overallobligate-ant-follower for the two smallest antbirds, with R. melanost- densities,are birds maximizing spatial utilization of icta space-useestimated at 0.95 m and that of available swarms ? --With detailed information on the subordinate G. salvini estimated at 0.73 m. densitiesof obligateant-following birds, army Confidence intervals for G. salvini did not over- ant densitiesand foragingdynamics, and swarm lap with thosefor P. nigromaculataor D. merula widths, it is possibleto determinethe average

2.5

0,5

I• M. fortis D. merula R G. salvini nigromaculata rnelanosticta species

FIG.3.3. Estimatedmeters of antswarmfront usedby individualsof five obligateant-following bird species in AmazonianPeru in 1998 and 2000-2001,with 95% confidenceintervals. Yearly estimatesare from a linear regressionof swarmwidth basedon the numberof birdsof eachspecies present at E. burchelliswarms (n = 138 swarms for 1998, 227 swarms for 2000-2001). OBLIGATE ANT-FOLLOWING BIRDS IN PERU 27

TABLE3.12. Estimates of spaceuse by birdsalong Eciton burchelli swarm fronts: predictions of four hypothesesand resultsfor two field seasons.Hypotheses (1), (3), and (4) provide estimatesof metersof spacean individualbird of eachspecies is predicted to use. Hypothesis(2) predictsa ranking of relative spaceuse. Units of predictions and resultsare in meters(m) per bird and are providedfor one field seasonof high bird-populationdensity (1998) and one of low bird-populationdensity (2000-2001). Predictionsfor hypothesis(3) are basedon the upper 95% CI of the dominantbird species(P. nigromaculata). Predictions for hypothesis4 arebased on the averagemeters per bird observedacross species in eachyear. No hypothesiswas fully supportedin 1998;hypotheses (1) and (3) were supportedin 2000-2001.

1998 2000-2001 m bird -1 m bird -• Predictions Hypothesis1: Spaceuse is related to speciesmass P. nigromaculata(46.0 g) 1.20 1.72 M. fortis(46.5 g) 1.22 1.74 D. merula(47.5 g) 1.24 1.77 R. melanosticta(31.4 g) 0.82 1.17 G. salvini(25.9 g) 0.68 0.97 P. nigromaculata= M. fortis= D. merula> R. melanosticta> G. salvini Hypothesis2: Spaceuse is directly related to position in dominancehierarchy P.nigromaculata > M. fortis> D. merula= R. melanosticta> G. salvini Hypothesis3: D. merula spaceuse is significantlygreater than antbird spaceuse D. merula 1.50 1.74 D. merula= upper 95% CI of P.nigromaculata Hypothesis 4: Birds divide swarm front evenly among species All species 1.02 1.58 P. nigromaculata= M. fortis= D. merula= R. melanosticta= G. salvini Results P. nigromaculata(46.0 g) 0.87 1.25 M. fortis(46.5 g) 1.00 1.32 D. merula(47.5 g) 0.99 1.90 R. melanosticta(31.4 g) 0.51 0.95 G. salvini(25.9 g) 0.69 0.73

daily foragingspace available to theobligate ant- 104.5 m in 1999, and 96.4 m in 2000-2001 (Table followingpopulation on the studysite each year 3.13). If ant-followersmaximize spatial use of and to estimateoverall bird use of that space. L. praedatorand E. burchelliswarms, observed That estimatesimplifies community dynamics average spacing per bird at swarms should in termsof both intra- and interspecificcompe- approach the maximal estimated amount of tition, by assumingthat individual birds have foraging spaceavailable per individual. If all knowledgeof and accessto all availableswarms obligate ant-following individuals used E. they encounterand that all swarmsare equal burchelli swarms, but never used those of L. in quality. In realit)• despoticdominance both praedator,a minimum amount of spacewould within andbetween species may affectantswarm be availableper individual at E. burchelliswarm choice. Use of all available swarm resources will fronts. For example,based on thosetotals and theoreticallyfollow an idealpreemptive distribu- the population density of ant-followingbirds tion pattern(Pulliam and Danielson1991) more in 1998, individual birds could gain <1.07 m than a patternof ideal free distribution(Fretwell per individual if all birds were maximally and Lucas1970) and may affectthe actualuse of using the swarm resourcesof E. burchelliand L. spacealong swarm fronts. praedator,but _<0.36m if all birds foragedsolely Estimatedmean daily amount of foraging with E. burchelli(Table 3.13). Observed foraging space available to the obligate ant-following spaceper individual at E. burcheItiswarms was bird populationper 100 ha was 92.5 m in 1998, 1.02 m per individual, or 95.3% maximization. 28 ORNITHOLOGICAL MONOGRAPHS NO. 55

TABLE3.13. Comparison of observedspace use by ant-followingbirds at antswarms.The theoreticalmaximum basedon antavailability provides an estimateof howwell thebirds maximize their use of availableforaging space(in bold). Estimates are given for Ecitonburchelli (E.b.) and Labidus praedator (L.p.) swarms; density of L. praedatorwas measuredin 2000-2001only, and is assumedconstant across years.

1998 n 1999 n 2000-01 n a. E.b.density per 100ha 4.4 10weeks 5.2 13weeks 5.0 15weeks b. Averagewidth of E.b.swarms (m) 9.1 102 10.7 112 9.0 196 c. Chanceof foragingper day 77% a 77% a 77% a d. Spaceat E.b.per 100ha [abc](m) 30.8 - 42.8 - 34.7 - e. L.p.density per 100ha .... 20.7 15weeks f. Averagewidth of L.p.swarms (m) .... 2.98 91 g. Chanceof foragingper day 100% - 100% - 100% - h. Spaceat L.p.per 100ha [efg] (m) .... 61.7 - i. Totalforaging space per 100ha [d+h] (m) 92.5 - 104.5 - 96.4 - j. Totalbirds per 100ha 86.6 NA 66.5 NA 45.7 NA k. Spaceper bird if all at E.b.[d/j] (m/bird) 0.36 - 0.64 - 0.77 - 1. Spaceper bird if all at L.p.[h/j] (m/bird) 0.71 - 0.93 - 1.35 - m.Maximum potential spacing [i/j] (m/bird) 1.07 - 1.57 - 2.11 - n. Averagebirds per E.b.swarm 8.9 102 7.8 112 5.7 196 o. Observedspacing at E.b.lb/n] (m/bird) 1.02 - 1.37 - 1.58 - p. Maximizationof spaceat E.b. [o/m] (%) 95.3 - 87.3 - 74.9 - q. Averagebirds per L.p.swarm 5.3 6 4.1 27 3.4 91 r. Observedspacing at L.p.If/q] (m/bird) 0.56 - 0.74 - 0.88 - s. Maximization of spaceat L.p. [r/m] (%) 52.3 - 47.1 - 41.7 - (Franks 1982b)

In 1999,maximal foraging space was 1.57m. If burchellidata. First, as explained earlier, L. all birds foragedonly with E. burchelli,foraging praedatorswarm use is dependenton theobligate spacewould be reduced to 0.64 m. Observed ant-followingbirds' ability to find the swarms. spacingat E. burchelliswarms was 1.37 m, or Birds cannotefficiently use a resourceif they 87.3% maximization of available resources. In do not know where it is, and birds only find L. the 2000-2001 season, birds obtained an aver- praedatorswarms opportunistically. Second, L. age foragingspace of 1.58 m per individual at praedatorswarms are often small enough that E. burchelli swarms, or 74.9% maximization. If only one family groupwill find and forageat a birds had not foraged with L. praedator,they particularswarm during an observationperiod. would have obtained a maximum spacingof A family will have decreasedagonistic interac- only 0.77m per individual. tions,and may use a swarmtoo small for mul- Actualspacing per bird at L. praedatorswarms tiple unrelatedindividuals to forageat together was alwaysnear or below half of what it might without aggression.Data from familiesat small have beenhad the birds taken advantageof all L. praedatorswarms will lowerthe observedfor- antswarm resources available to them, and rates agingspace for birdsat L. praedatorswarms and of maximizationof L. praedatorswarms ranged producea lower estimatefor maximizationof from 41.7% to 52.3%. That is based on the as- foragingresources with that ant species. sumptionthat L. praedatordensity was the same BecauseL. praedatorswarms cannot be fol- in the other years as in 2000-2001. However, lowed by birds day-to-daywith the precision that assumptionseems reasonable, given that I that an E. burchellicolony can, utilization of estimated19.9 aboveground swarms per 100ha the former speciesis not expectedto approach for a two-monthsample in 2002(Table 3.2). the maximum possible,and percentageof use For all threeyears, the percentageof utiliza- of total swarm resources is lowered. However, tion of antswarmresources differed dramatically the estimateof swarm use in 1998 approaches when spacingbehavior at L. praedatorswarms completeutilization of swarm resources.The was calculatedversus spacingbehavior at E. 1998season had the highestdensity of obligate burchelliswarms. Two main factorshelp explain ant-followerson the studyplot, the leastnumber why the estimate using L. praedatorswarms of E. burchellicolonies (4.4 per 100 ha), and the is less useful than that obtained with the E. lowest amount of availableforaging space with OBLIGATE ANT-FOLLOWING BIRDS IN PERU 29

E. burchelli(Table 3.13). It is possiblethat birds homeranges, in at leastone year,that werebe- weremore reliant on L. praedatorin 1998to avoid low the minimum size neededto securedaily agonisticencounters at the fewer E. burchelli foragingopportunities with E. burchellialone. swarms.If birdstracked areas where L. praedator Additionally,radiotelemetry of individualbirds swarmshad beenpreviously encountered rather showsthat use of L. praedatorswarms exceeds the than locatingthose swarmsopportunistically, percentage expected if bird speciesforaged with that couldraise the percentageof useof resourc- that antonly whenE. burchelliwas unavailable. esfor that year. No previousstudy has measured the ecologi- VII. Canspace be a limitingresource for obligate calvalue of L. praedatorto obligateant-following ant-followers?-- Obligateant-following birds are birds.Willis (1966b,1972) described their useby dependenton the food resourceprovided by facultative and migrant birds as high, because foraging swarms.As army ants move over the of lackof competitors.Generalized accounts of forestfloor, arthropodsare continuallyflushed birds and army antsin the secondaryliterature aheadof the oncomingants. Paradoxicall)• that imply that L. praedatoris a relativelyunimport- "superabundant"prey has allowed the ant- ant, marginallyused resourcein obligateant- followersto evolveas highly specializedforag- followers. Data presented here demonstrate ers,but it is alsowhat keeps their populations in the contrary:although spatially and temporally check.Competition is density-dependent,and unreliable,an unoccupiedL. praedatorswarm althoughantswarms appear to be an unlimited can offer foraging opportunitiesunencum- resource,space along a swarmfront is mediated beredby aggressionfrom other obligates,and throughintense interference competition (Tables a superabundanceof prey.Foraging L. praedator 3.9-3.11).Ability to capitalizeon prey availabil- coloniesseem common enough at CochaCashu ity restson a species'requirement of a particular that obligate ant-followers can greatly ben- amountof foragingspace along the front. Space efit from using the resourceopportunistically. requirementswere shapedby both body mass Additionally,averages for the four antbirdssug- and density-dependenteffects of competition, gestthat home-range size is minimizedthrough and spaceuse per speciesincreased in the year use of L. praedatorswarms. with the leastcompetitive pressure and lowest Two of the study species,M. fortis and D. avianpopulation density (Fig. 3.3). Competition merula,deserve special mention here. Home- at swarms may be made more intense by range size of M. fortis was consistentlysmaller "bet-hedging"resulting from birds'uncertainty than those of other ant-following species. about future availabilityof food. Particularly Myrmecizafortis proved an indefatigablefol- in the rainy season,there is alwaysthe chance lower of L. praedatorand, comparedwith other that a heavydownpour will begin,which may ant-followers,spent a significantlyhigher per- precludeant-followers from foragingthe restof centageof time with that speciesthan with the day.If rainis heavyenough, army ants stop E. burchelli(Table 3.8). Density of available L. foragingand begina massmovement returning praedatorswarms determined the small territo- to thebivouac (S. K. Willsonpers. obs.). Obligate ries that M. fortispairs defended.To the extent ant-followerscannot depend on availabilityof that there is any predictabilityin L. praedator their food resource for immediate future use, foraging patterns,it ought to be most appar- and thus cannot afford to "wait" or "sit out" ent to M. fortis,given that they spendthe most a foragingbout when competitorsare already time with L. praedatorand hold the smallestter- at a swarm front. Future uncertainty and the ritories.That they defend small, exclusiveter- physiologicalneed to eatevery day mayincrease ritoriesalso suggeststhat M. fortis individuals what competitionalready exists for the limiting should know their territories better than other resourceof spacealong a swarmfront. ant-followersat CochaCashu. Data presented here seemto unravel someof the mystery of AVIAN HOME RANGES AND POPULATIONS whether M. fortis obligately or facultatively follows army ants. To assumethat E. burchelli Data presentedhere strongly suggestthat is the preferredarmy ant resourceof all obli- L. praedatorswarms play a key role in allowing gate ant-followersmight lead to the erroneous obligateantbirds to decreasehome-range size. conclusionthat M. fortisis not obligatelytied to Each of the four antbird speciesheld average army ants. 30 ORNITHOLOGICALMONOGRAPHS NO. 55

The woodcreeperD. rnerula consistently Amazonian Brazil. That hypothesisis based had home rangesfar larger than thoseof the on the assumptionthat white-lippedpeccaries four antbird species.It seemslikely that fly- playeda prominentrole in the foragingecology ing ability may at least partially accountfor of D. merulain all regionswhere their historical that. Antbirds typically fly between swarms rangesoverlapped. by makingshort horizontal "hops" of 3-10 m In a study of bird communitystructure at from sapling to sapling (S. K. Willson pers. CochaCashu, Terborgh et al. (1990) estimated obs.). Dendrocinclarnerula individuals, on the home ranges for four of the five speciesin- other hand, are powerful fliers and can rap- vestigated here (P. nigrornaculata,M. fortis, idly movehundreds of meters,typically flying R. rnelanosticta,and G. salvini). Given the ef- throughthe understoryat a height of ~2 m (S. fort involved in that large-scale community K. Willson pers. obs). In that way, they can study--carriedout over three monthsof the sample multiple antswarmswith ease, assess dry seasonin 1982 and mainly using mist-net potentialcompetitors and mates,and searchfor capturesto createhome-range sizes for those peccaryherds. Cocha Cashu Biological Station birds--sample sizes for some species esti- is uniqueamong Neotropical research areas in mateswere necessarilysmall. My resultsrefine that it hasa large,healthy population of white- Terborghet al.'s(1990) estimated home ranges lippedpeccaries (Terborgh et al. 1984)--aglob- of >25 ha for R. melanosticta and G. salvini, but ally threatenedspecies listed by the Convention disagreewith their resultsfor P. nigromaculata on InternationalTrade in EndangeredSpecies andM. fortis(Table 3.14). Using sample sizes of (CITES), having been hunted to extinction in just threeand one,respectively, they estimated many areasof the Neotropics(Emmons 1990). home-rangesize of P.nigrornaculata at 14 ha, and At Cocha Cashu, I determined that D. rnerula that of M. fortisat >50 ha. Althoughsome P. ni- woodcreeperssupplement their foraging at grornaculataindividuals had smallhome ranges, antswarmsby foraging with peccaries.More my averagefor the threeyears of data was 46 than obligateant-followers, they are obligate ha, with yearly averagesranging from 25 to 86 followersof "eaters,"and peccaryherds pro- ha. The largehome-range size calculated for M. vide that serviceas they move throughan area, fortisby Terborghet al. (1990)may be an artifact rootingthe soil and uncoveringarthropods in of their sampling.Over threeyears of intensive the leaf litter. One D. rnerula individual was observation of color-banded individuals, it is seenperched on the hindquartersof a peccary clear that mated M. fortis pairs defend small, (T. Baggalaypers. comm.), and they have also intraspecificallyexclusive territories. It is pos- been observedperching on trees aboveherds. sible that the one individual spot-mappedin Additionally, >50% of D.rnerulamist-net cap- Terborghet al.'s(1990) study had recentlylost turesin the presentstudy smelledstrongly of its mateand was searchingover a large areafor white-lippedpeccary. Given that white-lipped a new mate.In 1998,I observeda male M. fortis peccariesare absentor greatly reduced in performingsimilar movementsafter his mate numbersover much of their former range,this died. Eventuall• he settledin a new areawith a interestingspecies interaction necessarily does new female.Unfortunately, inaccurate estimates not occur in most D. rnerulapopulations. It is of extremelylarge home ranges in obligateant- unknownhow the extirpationof white-lipped followingbirds are passedthrough the litera- peccarieshas affectedD. rnerulaabundance in ture, perpetuatingsome of the misconceptions otherregions. However, the D. rnerulapopula- aboutthese species (e.g. Stutchbury and Morton tion north of Manaus, Brazil, is estimated at 3.6 2001). birds per 100 ha during peak (postbreeding) In addition to home ranges, population density(Marantz et al. 2003).During the present densities of each bird speciesstudied here study,average adult D. rneruladensity at Cocha are estimated in Terborgh et al. (1990), and I Cashuwas 5--6xhigher, not includingjuveniles. comparedmy resultswith that study's(Table Although white-lipped peccariesare listed 3.14). Populationdensities of all study species as presentnorth of Manaus,Brazil, low herd droppedsignificantly throughout the present density--resultingfrom forest fragmentation study,but estimateshere are generallymuch and hunting pressures--mayhave negatively higher than those calculatedin Terborghet affectedhistorical density of D. rnerulain central al. (1990). The lowest populationdensities of OBLIGATE ANT-FOLLOWING BIRDS IN PERU 31

TABLE3.14. A comparisonof populationestimates and home-rangesizes for five speciesof obligateant- followingbirds at CochaCashu Biological Station, Peru. The 1990estimates are from Terborghet al. (1990);the 1998-2001estimates are from the presentstudy.

1990 1998-2001

Number of Home Number of Mean individuals range individuals homerange Species per 100ha (ha) n 100per ha (ha) n D. merula 16 NA - 16.7-22.1 64.6 20 P. nigromaculata 9 14 3 11.3-18.3 45.9 20 M. fortis 2 >50 1 10.4-11.7 15.4 4 R. melanosticta 4 >25 2 5.4-12.5 50.8 24 G. salvini 1 >25 1 8.8-17.5 27.5 21 three species(D. merula,P. nigromaculata,and becomesa mosaicof wet and dry patchesin the R. melanosticta)were fairly similar to the den- rainy months(see Terborgh 1983). sitiesfound in 1982;the othertwo species(M. In the 1999 early rainy season,dry land was fortisand G. salvini)had populationlows here fragmentedinto temporary islandswithin a that were 4.3-7.7x higher than the 1982 density matrix of rainwater over much of the study estimates.Although populationsampling was plot. Standingwater had an observableeffect low in Terborghet al. (1990),the large discrep- on army ant colonies,which had to "hopscotch" anciesbetween some results may represent from dry land-massto dry land-mass.Although natural populationfluctuations in the obligate E. burchellidensity in 1999 was similar to that ant-followingbirds. Evidenceis presentedin of otheryears (Table 3.1), much foraging during Chapter 6 that overall populationfluctuations the period occurredabove ground, in shrubs, are attributableto both low juvenilerecruitment trees, and vine tangles--unsuitablefor avian and variableadult survival.More yearsof con- foragers.Labidus praedator is dependenton dry tinuousstudy will determinewhether the trend land, becausemost of its foragingis doneun- of overallpopulation decrease observed from dergroundor on the surfaceof the ground,and 1998through 2001 is part of a larger cyclical colonynests are locatedunderground. For those populationfluctuation. reasons,it wasimpossible to obtainan accurate measure of swarm density for L. praedatorin SEASONALITY AND MOVEMENT PATTERNS 1999.Although L. praedatorcolony density may have been similar to that of other years, it is Individual movement patterns, and thus probable that colonieswere heterogeneously average home-range sizes, increased mark- distributedwithin the study plot as a result of edly for all bird speciessampled in 1999. inundation. Ecitonburchelli density was not abnormally The decreasein foragingopportunities avail- low in that year, and total populationdensity able to birds at E. burchelli swarms in 1999 was of obligateant-followers was intermediatebe- evident from the increased rate of agonistic tween 1998 and 2000-2001in the study area encountersat E. burchetliswarms in that year (Table 3.3). Anomalousrainfall patternsfrom (Tables3.9 and 3.10).That increaseis notablebe- CochaCashu correspond with theexpansion in causebird populationshad decreasedfrom 1998, home-rangesize acrossbird species.The rainy whereasarmy ant colonydensity was slightly seasonbegan one month earlierin 1999than in higherin 1999than in 1998,which suggeststhat 1998 or 2000, and 1204.5 mm of rain fell from less competitionshould have occurred(Tables Septemberto December,as comparedwith 741 3.1 and 3.3). Fewerforaging opportunities may mm in 1998and 868 mm in 2000 (Fig. 3.4). The have forced obligate ant-followersto maintain period of maximal rainfall was in November largerhome ranges so as to encompassenough and December in 1999-2000, two months ear- ground-foragingswarms. lier than in 1998-1999 and 2000-2001, when In the 2000-2001 field season, mean home- maximalranfall was in Januaryand March.The range sizes of all bird specieswere similar to studyplot liesentirely within floodplainforest, those seen in 1998, before the study plot was which, with inundation and rising river levels, inundated.More yearsof study in both the dry 32 ORNITHOLOGICAL MONOGRAPHS NO. 55

500

45O

4OO

35O

300 • 1998-1999 250 •.•"•1999-2000 • ß •2000-2001 200

150

100

50

Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun

month

FiG.3.4. Monthly rainfall at the CochaCashu Biological Station, Manu NationalPark, Peru, from 1998to 2001. Graph organizationfollows the cyclicrainy and dry seasons.Precipitation patterns were similar for the rainy seasonsof 1998-1999and 2000-2001.Approximately 400 mm more rain fell from Septemberto Decemberof 1999than in 1998or 2000,though total rainfallin eachrainy seasonwas similar (2,322, 2,363, and 2,607mm). and wet seasonswould certainlyhelp to clarify relatively equal proportionsare segregatedby the connections between rainfall and movement body mass (Table 3.4), which may allow differ- patternsof obligateant-following birds. ential use of spacealong the width of a swarm front (Fig. 3.3). CONCLUSION CONSERVATION CONCERNS In summary, species coexistence among five obligate ant-followersis at least partially Although the birds in Manu National Park, explained by their differential use of the two Peru, are in no danger from forest fragmenta- army ant species,E. burchelliand L. praedator tion, obligate ant-followersin other areasof the (Tables 3.7 and 3.8). Most speciesminimized tropicshave been shown to be one of the first home rangeswell below the size required to groupsof birds to go locallyextinct once an area follow multiple E. burchellicolonies by relying has been isolated (Harper 1989, Stouffer and on unpredictable but abundant foraging op- Bierregaard1995). That seemsto be attributable, portunitiesat L. praedatorswarms (Table 3.6). By in part, to their requirementsof relatively large using both L. praedatorand E. burchelliswarms, home ranges,as comparedwith birds of similar individual birds maximized foraging space size,and alsoto their completereliance on army per bird (Table 3.13) and decreasedagonistic ants. Eciton burchelli, because of its narrow tol- interactions between individuals. Further de- erancefor changesin humidity, requiresforest creasesin potential interspecificcompetition with canopycover (Schneirla1971). If the ants were gained throughone species'specialization die out, the birds either die out as well or, as on L. praedatorarmy ants (Fig. 3.5) and another demonstratedin the Biological Dynamics of species'use of white-lipped peccaryherds. The ForestFragmentation project in Brazil, abandon three obligate antbirds that used E. burchelliin fragmentsof <100 ha for mature forest(Harper OBLIGATE ANT-FOLLOWING BIRDS IN PERU 33

35

30

25

• 2o

•15

.>_

.elO

P. nigrornaculata M. fortis D. rnerula R. rnelanosticta G. salvini species

FIG.3.5. Of fiveobligate ant-following bird speciesin AmazonianPeru, M. fortisand D. merulautilize the army antsE. burchelliand L. praedatorin significantlydifferent proportions (chi-square tests: P < 0.001).

1989,Stouffer and Bierregaard1995, Bierregaard from wet to dry season(Willis 1967),which sug- et al. 2002). geststhat relianceon L. praedatorswarms is an In Panama,flooding of the ChagresRiver to unlikely strategyon BCI. Equation3.4 predicts form the Panama Canal and Gatun Lake has led that Pha.mcleannani and G. leucaspison BCI hold to localextinctions of someobligate ant-followers largerhome ranges than obligateant-followers (Willis 1974,Robinson 1999). The dominantant- at Cocha Cashu, because of lower E. burchelh follower in Panama, Pha. mcleannani,is one of colony density on BCI; however, the island manyavian species that have gone locally extinct is still predictedto hold >750 individual Pha on the 1,600-haisland of Barro Colorado (BCI) mcleannani,assuming that conditionsapproxi- sinceit was formedin 1913.The smallerobligate mate Panama'smainland (usingthe estimateof speciesG. leucaspisis now found at a densityof 48 individualsper 100ha on the "Limbo plot", -2.5 individuals per 100 ha on BCI, compared Robinson et al. 2000a, W. D. Robinson pers with 12 individuals per 100 ha on the nearby comm.).However, that assumptionis not valid mainland,and is predictedto go locallyextinct Experimentaltests have determinedthat nest on BCI in the near future (Robinsonet al. 2000a, predationrates on BCI are significantlyhigher W. D. Robinsonpers. comm.). than relative rates on the mainland, possibly The island has had a stable population of becauseof mesopredatorrelease (Loiselle and E. burchellicolonies for >50 years, estimated Hoppes 1985, Sieving 1992). Furthermore, at 3.2 coloniesper 100 ha (Willis 1967, Franks Sieving (1992) found that BCI-extinct birds 1982b).Willis (1967) estimatedthe averageL. that are terrestrialinsectivores have relatively praedatorswarm densityover 14 monthsas 8.5 predator-safenests on the mainland;as a result, active swarmsper 100 ha (no recent estimates she argued, those speciesmay have evolved are available for BCI). However, L. praedator lower capacityto renest;unable to adaptto the swarm densityon BCI fluctuatesdramatically high nest-predationpressures on BCI (brought 34 ORNITHOLOGICAL MONOGRAPHS NO. 55 about by isolation and mesopredatorrelease), Insights from studies of bird communities they went locally exinct.Additionally, if high in Panamaand Brazil highlight the sensitivity annualpopulation fluctuations found in Cocha of obligate ant-followingbirds to fragmenta- Cashubirds may be generalized,they provide tion, isolation,and nest predation.A working anotherpotential reason for the disappearance understandingof the ecologicalneeds of this of Pha.mcleannani from BCI. As an islandpopu- specialized,vulnerable guild of birds may fa- lation, Pha. mcleannani on BCI lacks a source cilitateconservation plans that will preventthe populationfrom whichto add moreindividuals obligateant-followers from goinglocally extinct after "bad"years (Willis 1974, Leigh 1982, as larger areasof lowland rainforestare contin- Robinson 1999). ually altered and fragmentedby humans. 4. SURVIVAL RATES AND POPULATION DYNAMICS IN OBLIGATE ARMY ANT FOLLOWERS

LOWLAND TROPICAL RAINFORESTS have histori- was carriedout on a 2-ha study plot. Terborgh cally been characterizedas climaticallybenign et al. (1990) challengedwhether the scale of environments(MacArthur 1972).In comparison Karr'swork wassufficient to adequatelysample with the wide fluctuationsin temperatureand the bird community,citing primarily the fact day-length associatedwith temperate areas, that the meanhome-range size for tropicalbirds tropical lowlands display seemingly insig- is larger than Karr's study plot. nificant seasonalchanges in thosefactors. That In addition to seasonal fluctuations, the stabilityis thebackbone of manyearly "equilib- lowland tropicsundergo periodic climatic dis- rium" theoriesput forth to explainthe high spe- turbances,most commonlythrough cyclic E1 ciesdiversity in the tropics(Dobzhansky 1950, Nifio Southern Oscillation (ENSO) processes Fischer 1960, Connell and Orias 1964, Pianka (Conditet al. 1996,Wright et al. 1999,Holmgren 1966, MacArthur 1969). In theory, climatic sta- et al. 2001). Janzen (1967) discussedtropical bility promotesyear-round resourcestability, stabilityas promotingnarrow ecologicaltoler- which leadsto increasedspecialization and thus ancesto gradientsin temperatureand rainfall to increasedcoexistence of ecologicallysimilar in tropicalspecies as comparedwith temperate species(Orians 1969, Leigh 1982). A stable species.Although his argumentswere madeto resource-baseleads to population stability in explainpatterns in tropicalspecies' ranges, they consumers,whose populations are theoretically have implicationsfor populationdynamics. In mostaffected by density-dependentfactors and tropical dry forest,Faaborg (1982) found that held at or near a stablecarrying capacity (Cody populationlevels of someresident bird species 1966, MacArthur 1972, Newton 1998). dropped significantlyin responseto drought. However, lowland tropical systemsdisplay Stiles(1992) reported how an unusualdrought distinctfluctuations in someabiotic parameters; affectedsurvivorship, breeding, and body con- in particular, ecologistshave documentedthe dition in a rainforestpopulation of Phaethornis importanceof seasonalrainfall variation. The superciliosushummingbirds. In that study, transitionfrom rainy to dry seasoncoincides recoveryof age-structureand populationtook with significantdrops in leaf-litter and canopy 3-4 years. Effectsof ENSO in the southwest- arthropod abundance (Williams 1941, Janzen ern Amazon basin are small, and southeastern 1973, Willis 1976, Wolda 1978, Levings and Peru, where the present study took place, is Windsor 1982,Smythe 1974, Pearson and Derr minimally affectedbecause it is locatedjust 1986),leaf loss in deciduouscanopy trees (Leigh south of the transition node between the wetter- and Windsor 1982),and a period of famine in than-normal conditions of northern South many mammal and bird species(Leigh and America and the drier-than-normal conditions Windsor 1982,Leigh 1999).Large fluctuations of Bolivia and southwesternBrazil (Marengo in the tropical resource-baseseem at odds with 1992, Pezzi and Cavalcanti 2001, Coehlo et al. the theoreticalunderpinnings of tropicalstabil- 2002, Silman et al. 2003). In the southwestern ity; somestudies question whether the "table" Amazon, cold air-masses from far-southern- tropicsare asstable as ecologists once proposed. latitude stormsperiodically move north along Karr and his colleagues(Karr 1971,Karr et al. the Andean chain and descend into the Amazon 1982, Karr and Freemark 1983) describedhow basinin the dry season,pushing interior forest bird populationscan fluctuateseasonally with temperaturesbelow 15øCfor up to five days changingresource levels; they found that indi- (Terborgh1983). In the rainy season,destructive viduals continuallyreselect microhabitat in re- wind- and rainstormsmove through lowland sponseto gradientsin moistureand vegetation. tropical forests unpredictab13•creating new Karr and Freemark(1983) asserted that tropical gapsthe size of footballfields (Vandermeeret bird assemblagesare highly dynamicin space al. 2000). During that season,which coincides and time, leading to nonequilibriumbut still with the nesting seasonsof most insectivorous resultingin relativelypredictable communities. birds, rapid flooding can create a mosaic of Issues of scale must be considered; Karr's work dry and wet patcheswithin floodplainforest.

35 36 ORNITHOLOGICAL MONOGRAPHS NO. 55

In some species,unpredictable fluctuations in insectivores,most obligate ant-followingspe- abioticfactors may influencejuvenile and adult des hold much larger,more amorphoushome mortality, reproduction,and population and rangesthat arenot conspecificallyexclusive (see communitydynamics (Stouffer and Bierregaard Chapter3; alsoWillis 1967,1973). If territoriality 1993, 1996). is an importantproximate factor that keeps local In the southwesternAmazon basin, obligate bird populationsstable, obligate ant-followers ant-following birds are exposed to seasonal should show less rigidity in their population rainfall patterns and to irregular cold fronts dynamics. Ant-following birds aggregate at and floodplain inundation. All those factors food resources,and population fluctuations may affectdensity and availabilityof army ants, have the potential to affect rates of agonistic the ant-followers'foraging resource. In the pre- encounters between dominant and subordinate vious chapter,I askedhow the five speciesof species.Ecologically similar species may be af- obligateant-followers at CochaCashu differen- fectedby changesin populationlevels of their tially usedthe resourcesoffered by two species competitors,and periodicfluctuations in those of army ants (L. praedatorand E. burchelli).Two levelsmay contributeto maintaininghigh di- of the five bird species(D. merulaand M. fortis) versitywithin a guild. preferentiallyforaged with one or the otherof Here, I explorethe populationdynamics of a the two ant species;those preferences showed guild of five obligateant-following bird species up in the specificbehavioral ecology of each in southeasternPeru over a five-yearperiod. My bird species.Myrmeciza fortis pairs held small data on adult annual survival rates,population conspecificallydefended territories, which al- densities, food-resource densities, and rainfall lowed them exclusiveaccess to the L. praedator enabled me to assessthe effects of army ant swarmsin their territory.On the otherhand, the density(a bioticfactor) and of precipitation(an woodcreeperD. merulapreferred E. burchelli abioticfactor) on bird populationparameters. swarms. Ecologically,assessment of multiple I investigatedfour questions:(1) What are the antswarms allowed D. merula individuals to estimated adult annual survival rates and re- maintainhome rangesthat averaged5x larger cruitmentrates for the five speciesof obligate than the similarlysized M. fortis.Through linear ant-followers?(2) Do annual rainfall patterns regressionanalyses, I found that obligateant- or antswarm density correlatewith changes following speciesat CochaCashu used differ- in avian population parameters?(3) Does a ent amountsof spacewhile foraging along an strictlyterritorial ant-followingspecies display antswarm front--which facilitates inclusion at populationdynamics that differ from thoseof swarms of smaller, more subordinatespecies ant-followingspecies with lessrigid (and non- that can "fit" into and forage effectivelyin the exclusive)home ranges?(4) Can fluctuations spacesaround their larger competitors.Finall)• in population dynamicsof individual species ! found that D. merulasupplemented its forag- assistin maintainingcoexistence of the five ob- ing with army antsby followingherds of white- ligate ant-followers? lipped peccaries. All those ecologicaland behavioral differ- METHODS encesenhance the coexistenceof the five spe- ciesof obligatesat CochaCashu, but there are SURVIVAL ESTIMATES still unansweredquestions regarding species coexistence.Most Neotropicalbird speciesthat Intensive capture-and-resightdata collectedover havebeen intensively studied have stable popu- five years were used to estimateannual adult sur- lations, as well as stable territorial boundaries, vivalrates. Data were analyzed in the programMARK which oftenremain unchangedfor years(Willis (White and Burnham 1999), following the approach 1974; Munn and Terborgh 1979; Munn 1985; outlined by Lebreton et al. (1992). Cormack-Jolly- Seber(CJS) constant and time-dependentsurvival and Greenbergand Gradwohl 1986, 1997; Roper recapturemodels were fitted for all species.In addition, 1996;Jullien and Thiollay 1998;Robinson 2000; I includemodels built apriori to testspecific hypotheses Styrsky2003), and territorialityhas been hy- about survivaland recapturerates for particularspe- pothesizedas a drivingproximate force in regu- cies.A priorimodels were based on the followingob- lating populations(Greenberg and Gradwohl servations:(1) the study plot was not identicalin 1999 1986). In contrast to generalist territorial and 2000, and thus may have alteredrecapture rates; OBLIGATE ANT-FOLLOWING BIRDS IN PERU 37

(2) adultR. melanostictasightings were much lower in on the M. fortis data set, and resultsare considered 2000;and (3) capture-and-resighteffort was lower in preliminary. 2002.All survivalintervals are denotedby the yearin Annual adult population densities with 95% whichthe intervalends (e.g. 2000 denotes the interval confidence intervals were calculated using the 1999-2000).The a priorihypotheses were as follows: Bowdenmodel-estimation technique in the program (1) Survival rate (d?)is consideredconstant, and NOREMARK(White 1996;see Chapter 3 for morede- recaptureprobability (P) is differentin 2002 than in tails).Recruitment rate for eachspecies was estimated other years.This model accountsfor the shorterdura- for the 1999 and 2000 breedingseasons. I multiplied tion of the 2002field season,which may havelowered adultsurvival rate by estimatedadult population den- bird resightingrates. Model 1 is d?(.)p(2002). sityin yeart (calculatedwith Bowden'smodel estima- (2) Survival rate for the interval 1999-2000is dif- tor on field data) to estimatethe adult population,or ferent from that of other years,and recapturerate "survivors,"still alive the followingyear (year t + 1). is constant.This model formally testswhether sur- That estimatewas then compared with the estimated vival rate was different in the 1999-2000 interval as adult populationas gatheredfrom field datafor that comparedwith otherintervals, and is basedon the a year (year t + 1), calculatedwith Bowden'smodel prioriobserved drop in numberof R. melanostictaindi- estimator.The replacementrate is the Bowden'ses- viduals between the 1999 and 2000 seasons. Model 2 timated populationminus the estimatedsurvivors. is c•(2000)p(.). A replacementrate of zero equalsno recruitment;a (3) Survival rate for the interval 1999-2000is differ- negativenumber signifiesno recruitmentplus some ent from that of otheryears, and recapturerate is time adult mortalityabove that predictedby the estimated dependent.This model is similarto model2, but pre- adult survivalrate; and a positivenumber signifies re- dictsa differentrecapture rate for eachyear interval. cruitmentinto the population,either by youngof the Model 3 is d?(2000)p(t). previousyear or by adultfloaters from outside areas. (4) Survival rate is constant,and recapturerate in 2000is differentfrom that of otheryears. This model RESULTS accountsfor thefact that the studyplot was not identi- calin 1999and 2000, and thus may have altered recap- SURVIVAL AND RECRUITMENT RATES ture rates.Model 4 is d?(.)p(2000). (5) Survivalrate is constant.Recapture rate is the The bestmodels for four of the five studyspe- same between 1998 and 1999, and differs in both 2000 cies had constant annual adult survival over the and 2002. This model formally testswhether the ob- serveddrop in number of R. melanostictaindividuals five-yearperiod. Survival was higher in larger- between 1999 and 2000 is attributable to differences bodied species(Table 4.1). Dendrocinclamerula in recapturerate, with the assumptionthat 1998and survival was highest,at c• = 0.80;M. fortissur- 1999 had similar recapturerates. Recapturerate is vival wasestimated at c•= 0.72,P. nigromaculata at modeledseparately for 2002to accountfor the shorter c•= 0.63,and G. salviniat c•= 0.59.Rhegmatorhina durationof the field season.Model 5 is d?(.)p(1998= melanostictasupported a modelwith a constant 1999,2000, 2002). adult survival rate in the first, second, and (6) Survival rate is constant.Recapture rate is the fourth yearly intervals,at cb = 0.62,but survival same between 1998 and 1999, and the same between 2000 and 2002. This model accounts for the fact that droppedto c•= 0.36in the interval1999-2000. Estimated recruitmentrates varied widely the percentageof G. salvinicolor-banded individuals was much higher in 2000-2001and 2002 than in 1998 by year and species(Table 4.2 and Fig. 4.1). and 1999.Model 6 is d?(.)p(1998= 1999, 2000 = 2002). From 1998 to 1999,the total adult population Models were selected for fit on the basis of number declined by 23% (see Table 3.3), and three of parametersand deviance;the mostparsimonious species(P. nigromaculata,R. melanosticta,and model has the lowest value for Akaike's Information D. merula) showed no recruitment. Based on Criterion (Burnham and Anderson 2002). Models confidenceintervals, population estimatesfor that did not supporta priori hypotheseswere not two of thosespecies--P. nigromaculata and R. selected as candidate models to describe the data. melanosticta--showedsignificant declines in the In accordancewith White (2002),I usedthe program sameperiod, whereas the D. merulapopulation RELEASEto test overdispersionin the data, rather decline was not as clear. In the same interval, M. than relying on bootstrapgoodness-of-fit tests. Data for four of the five speciesdisplayed no overdisper- fortisshowed positive recruitment and G. salvini sion.A fifth species,M. fortis,holds small territories; showedhigh recruitment(6.4 individuals per thus,sample sizes of bandedbirds of that speciesfor 100ha; Fig. 4.1). survival estimatesare low (n = 15). To avoid overfit- From 1999 to 2000, the total populationof ting the data,only the four basicCJS models were run obligate ant-followers decreased again, from 38 ORNITHOLOGICAL MONOGRAPHS NO. 55

TABLE4.1. Model selectionresults from the programMARK for apparentadult survival(c•) and recapture rates (p) for five ant-followingbird speciesin AmazonianPeru between1997 and 2002;CI = 95% confidence interval.

Dendrocincla merula A AICc Model AIC AICc weight #Par Deviance {c•(.)p(2000,2002)} 193.005 0.00 0.319 4 18.613 {c•(.)p(t)} 193.070 0.06 0.308 5 16.525 {c•(2000)p(t)} 194.536 1.53 0.i48 6 15.804 {c•(t)p(t)} 195.384 2.38 0.097 7 14.431 {c•(t)p(.)} 195.733 2.73 0.081 5 19.187 {c•(.)p(98=99,00=02)I 198.093 5.09 0.025 3 25.823 {c•(.)p(2002)} 198.447 5.44 0.021 3 26.177 {c•(.)p(.)} 209.925 16.92 0.000 2 39.745 {c•(2000)p(.)} 211.266 18.26 0.000 3 38.996 {c•(.)p(2000)} 211.319 18.31 0.000 3 39.049 Realfunction parameters of {•b(.)p(2000,2002)}

Parameter Estimate SE LowerCI UpperCI 1:c• 0.799 0.050 0.682 0.880 2:p 0.974 0.026 0.837 0.996 3:p 0.697 0.105 0.465 0.859 4:p 0.354 0.098 0.191 0.560 Phlegopsisnigromaculata A AICc Model AICc AICc weight #Par Deviance {c•(.)p(.)} 157.271 0.00 0.250 2 15.877 {c•(.)p(98=99,00=02)} 157.593 0.32 0.213 3 14.063 {c•(2000)p(.)} 158.307 1.04 0.149 3 14.776 {c•(.)p(2002)} 158.742 1.47 0.120 3 15.211 {c•(.)p(2000)} 159.003 1.73 0.105 3 5.473 {c•(.)p(2000,2002)} 159.502 2.23 0.082 4 3.787 {c•(t)p(.)} 161.189 3.92 0.035 5 13.239 {c•(.)p(t)} 161.607 4.34 0.029 5 13.656 {d)(2000)p(t)} 163.086 5.82 0.014 6 12.848 {c•(t)p(t)} 165.380 8.11 0.004 7 12.802 Realfunction parameters of {c•(.)p(.)}

Parameter Estimate SE LowerCI UpperCI 1:c• 0.631 0.056 0.516 0.733 2:p 0.838 0.073 0.644 0.936 Myrmecizafortis A AICc

Model AIC½ AIC• weight #Par Deviance {c•(.)p(.)} 49.790 0.00 0.804 2 10.387 {c•(.)p(t)} 52.836 3.05 0.175 5 4.955 {c•(t)p(.)} 57.723 7.93 0.015 5 9.842 {c•(t)p(t)} 59.728 9.94 0.006 7 4.625 Realfunction parameters of {c•(.)p(.) PIM coding} Parameter Estimate SE LowerCI UpperC! 1:c• 0.723 0.108 0.475 0.883 2:p 0.754 0.143 0.405 0.933 OBLIGATE ANT-FOLLOWING BIRDS IN PERU 39

TABLE 4.1. Continued.

Rhegmatorhinamelanosticta A AICc Model AICc AICc weight //Par Deviance {d?(2000)p(.)} 94.889 0.00 0.491 3 6.070 {d?(.)p(.)} 96.696 1.81 0.199 2 10.070 {d?(.)p(2002)} 98.889 4.00 0.067 3 10.070 {d?(.)p(2000)} 98.889 4.00 0.067 3 10.070 {d?(.)p(98=99,00=02)} 98.889 4.00 0.067 3 10.070 {d?(t)p(.)} 99.124 4.23 0.059 5 5.702 {d?(.)p(2000,2002)} 101.153 6.26 0.021 4 10.070 {d?(2000)p(t} 101.548 6.66 0.018 6 5.710 {d?(.)p(t)} 103.492 8.60 0.007 5 10.070 {d?(t)p(t)} 104.038 9.15 0.005 7 5.702 Realfunction parameters of {d?(2000)p(.)} Parameter Estimate SE Lower CI Upper CI l:d? 0.622 0.072 0.474 0.751 2:d? 0.364 0.103 0.193 0.577 3:p 1.000 0.000 1.000 1.000 Gytnnopithyssalvini A AIC c Model AICc AICc weight //Par Deviance {d?(.)p(98=99,00=02)} 125.077 0.00 0.301 3 21.206 {d?(.)p(.)} 126.921 1.84 0.120 2 25.215 {d?(.)p(t)} 127.171 2.09 0.106 5 18.791 {d?(.)p(2000,2002)} 127.300 2.22 0.099 4 21.206 {d?(t)p(t)} 127.353 2.28 0.097 7 14.207 {d?(.)p(2002)} 127.608 2.53 0.085 3 23.737 {d?(.)p(2000)} 127.686 2.61 0.082 3 23.815 {d?(2000)p(.)} 129.027 3.95 0.042 3 25.156 {d?(t)p(.)} 129.412 4.34 0.035 5 21.032 {d?(2000)p(t)} 129.451 4.37 0.034 6 18.722 Real functionparameters of {d?(.)p(98=99,00=02)} Parameter Estimate SE Lower CI Upper CI l:d? 0.586 0.055 0.475 0.689

2:p 0.805 0.112 0.503 0.944 3:p 1.000 0.000 1.000 1.000 an estimated67 adults per 100 ha, to 46--a de- drop in populationin both the 1998-1999and cline of 33% (seeTable 3.3). However, patterns 1999-2000intervals. In 1999-2000,the drop was within speciesdiffered from thoseobserved in largelyattributable to the nearly 50% declinein 1998-1999.In contrastto its reproductivesuccess adult survival; however, some successfulrecruit- in 1998-1999,G. salviniwas the only speciesto ment occurred,which kept the populationfrom show negativerecruitment in the 1999-2000in- decliningeven further (Table4.2). The two other terval.That estimatesuggests little-to-no juvenile speciesthat showed negative recruitment in recruitment,in additionto higheradult mortality 1998-1999,P. nigromaculataand D. merula,both than predictedby normaladult survivalrates for displayedpositive recruitment in 1999-2000.In the species(Table 4.2). Confidenceintervals for all of the abovecases, except for R. melanosticta G. salvinisupport a declinein adult population in 1999-2000,there is no changein estimatesof densityover that period(Fig. 4.2). Rhegmatorhina adult survival;population declines in the other melanostictais the only speciesthat shows a speciesappear to be attributableto failure of 40 ORNITHOLOGICAL MONOGRAPHS NO. 55

TABLE4.2. Recruitmentestimates over two yearsfor five speciesof obligateant-following birds in Amazonian Peru. Multiplying the estimatedadult survivalrate by the estimatedpopulation per 100 ha providesan estimateof the adult populationstill alivethe followingyear. That estimate(e.g. "1999 estimated survivors") is then comparedwith the actualestimated population per 100 ha for that year (usingBowden population models).The replacementrate is the Bowdenestimated population (e.g. "1999 estimated population") minus the estimatedsurvivors. A replacementrate of zero equalsno recruitment,a negativenumber signifies no recruitmentplus some adult mortally, and a positivenumber signifies recruitment into the population,either by youngof theprevious year or by adultfloaters.

1998-1999 Survival 1998 estimated 1999 estimated 1999 estimated 1999 Species rate population survivors population replacementrate D. merula 0.80 19.8 15.8 15.2 -0.64 P. nigrornaculata 0.63 21.0 13.2 13.0 -0.23 M. fortis 0.72 9.3 6.7 10.1 3.40 R. rnelanosticta 0.62 16.7 10.4 10.1 -0.25 G. salvini 0.59 19.8 11.7 18.1 6.42 1999-2000 Survival 1999 estimated 2000 estimated 2000 estimated 2000 Species rate population survivors population replacementrate D. merula 0.80 15.2 12.2 14.0 1.84 P. nigrornaculata 0.63 13.0 8.2 10.7 2.51 M. fortis 0.72 10.1 7.3 8.5 1.23 R. rnelanosticta 0.36 10.1 3.6 4.8 1.16 G. salvini 0.59 18.1 10.7 7.7 -2.98

8.42

4 3.4

2.51

1.84

1.23 1.16

-0.25 -0.25 -0.64

-2

-4 Phlegopsis Myrrneciza Rhegrnatorhina Gyrnnopithys Dendrocincia nigromaculata fortis rnelanosticta saivini rnerula

F•c. 4.1. Estimatedrecruitment rate per 100 ha per year for five obligateant-following bird speciesin AmazonianPeru. Recruitment rate is adjustedfor adult annualsurvival; a rateof 0 signifiesno juvenilerecruit- ment,a negativerate signifies no juvenile recruitment and adultmortality above the averagesurvival rate, and a rate>0 signifiespositive recruitment. OBLIGATE ANT-FOLLOWING BIRDS IN PERU 41

30

25

20

15

10

D. merula P. nigromaculata M. fortis R. melanosticta G. salvini sr•ecies

F•c. 4.2. Estimatedadults per 100ha for five speciesof obligateant-following birds in AmazonianPeru, over threeyears. Error bars are 95%confidence intervals, using Bowden's model estimation. juvenilesfrom the previousbreeding season to DISCUSSION recruitinto the population. Total populationdensity of the five study POPULATION DYNAMICS AND TERRITORIALITY speciesdeclined by -25% each year, from an estimatedhigh of 87 adultsper 100ha in 1998 If territorialityis a stabilizingproximate fac- to a low of 45 adultsper 100ha in the 2000-2001 tor for populationsof tropicalbirds, species with field season. Based on confidence intervals, M. less-rigidterritorial systems can be hypothesized fortis and D. rnerulaadult population levels to showmore variation in populationdynamics. remainedstable throughout the study,P. nigro- That can be tested among the four phyloge- rnaculataand G. salviniadult populationsfell in neticallysimilar obligate antbird species, three one of the two yearly intervals,and R. rnelanost- of which hold amorphous,non-exclusive home ictaadult populationsfell in bothintervals (Fig. ranges(P. nigrornaculata,R. rnelanosticta,and G. 4.2). Adult survival estimatesfor all species salvini)and one of whichis conspecificallyter- were stablethroughout the study period, with ritorial (M. fortis)(Fig. 4.3). Basedon confidence one exception;only R. rnelanosticta'sdrop in intervals, each of the three nonterritorial antbirds populationcould be attributedto adult mortal- showedsignificant population declines (Fig. 4.2). ity, and only for the interval 1999-2000(Table Populationdensity of M. fortis, on the other 4.1). All other dropsin populationdensity are hand,remained stable over three years. Because attributableto a lack of recruitmentof young of thesmall sample sizes, those results should be from the previousbreeding season(Table 4.2 viewed as preliminary;however, they provide and Fig. 4.2). Below,I discussthe implications supportfor the hypothesisthat territorialityis a of low recruitmentfor obligateant-followers. drivingproximate force in stabilizationof popu- All data from recruitmentpresented here lationdynamics in tropicalbirds (Greenberg and were inferredfrom adult survivaland popula- Gradwohl 1986,Newton 1998). tion estimates;actual juvenile recruitmentwas 42 ORNITHOLOGICAL MONOGRAPHS NO. 55

' A

Fro. 4.3. Selectedhome-range areas for (A) G. salvini,(B) P. nigromaculata,and (C) M..fortis during 2000-2001 field seasonat Cocha Cashu BiologicalStation, Peru. Polygonsenclose all sightingsof pairs or family groups.

not measured.In the few instanceswhere juve- long-term implications of population fluctua- niles from one year were resightedin following tions in these birds, it is instructive to examine years with mates(S. K. Willson unpubl. data for ratesof recruitmentof juvenilesin what would P. nigromaculataand R. melanosticta),individuals be considered"good" years. In the period were located within a few hundred meters of 1998-1999, the smallest and most subordinate where they were banded as nestlings.That lim- ant-follower, G. salvini, showed the highest ited evidencesuggests that juveniledispersal is recruitmentand no significantchange in adult confinedto short distancesin somejuvenile ant- survival or populationdensity (Table 4.2 and followers.However, it is alsopossible that most Fig. 4.1).At a level of 19.8adults, or 10 breeding juveniles dispersed long distancesand were pairsper 100 ha, during the breedingseason of not observedagain becausethey settledoff the 1998-1999, only 6.4 new recruits entered the study plot. A study of dispersalusing radiote- population. Those 6.4 adults were presum- lemetry would provide invaluabledata toward ably birds that fledged during the 1998-1999 answeringthe questionmore directly. breedingseason--which signifies that during a To gain a better understandingof potential "good"year, in which G. salvinihad the highest OBLIGATE ANT-FOLLOWING BIRDS IN PERU 43 recruitmentof any obligateant-follower, each continueto use the areabut foragemore often pair contributed an average of 0.6 successful abovethe terrestrialleaf litter,in vegetationand young out of a clutchsize of two eggs(Skutch trees.Army ant densitythus remainsthe same, 1996).Between 1999 and 2000, no new fledglings but availabilityto ant-followingbirds changes are estimated to have recruited, and the adult considerably.Aboveground swarms flush fewer populationdeclined significantly. insectsfor birds, becauseforaging shifts from In the period 1999-2000,highest recruitment fan-shapedraids over leaf litter to column- was seenin P. nigromaculata(Fig. 4.1). Out of an raids up trees(Schneirla 1971). Although total estimatedpopulation of 13 adults per 100 ha, rainfall was similar in the three rainy seasons 2.5 new juvenileswere addedto the population. (September-Aprilin 1998-1999,1999-2000, and If we assumethat, for this cooperativebreeder, 2000-2001),the shiftforward in onsetof precipi- each breeding group consistsof three adults, tation in 1999,coupled with the suddendrop in then 4.3 breedinggroups per 100 ha eachcon- precipitationlate in the rainy season,may have tributed an averageof 0.6 successfulyoung. In disrupted normal foraging patterns in ant- the previousperiod, 1998-1999,no recruitment followingbirds in that year.Concurrently with was seenand a significantadult populationde- thosechanges in precipitationin 1999,all bird cline was evident. Thoseresults are important, speciesfor whichwe have sufficientdata (four becausethey provide evidencethat in obligate of the five) displayedlarge increasesin mean ant-followers, juvenile recruitment may be home-rangearea (seeFig. 3.2). Acrossspecies, consistentlylow, and often zero. If recruitment there was a significantincrease in home-range remainslow or negativefor consecutiveyears size for 1999 as comparedwith 1998 and 2000 in a closedpopulation, population densities will (ANOVA: F = 12.56,df = 2 and 86, P < 0.0001). decreaseas a result of normal mortality. In the It is possiblethat somebirds moved from the five ecologicallysimilar study species,popula- floodplainstudy-plot to terrafirme forestdur- tion densitiesfluctuated for almostall species, ing the 1999wet season.However, of 147 indi- and total populationdensity within the guild viduals of the five speciesthat were bandedby droppedconsecutively over three years.Those 1998, only one (a P. nigromaculatafemale) was resultssuggest that population processeswere resightedin later yearsbut not in 1999.If any not local, or recruitment should have been buff- birdsleft the plot in 1999,they did not return in ered from outsidethe plot. If we hypothesize later years.In "ordinary"years,when inunda- that density-dependenteffects are driving the tion is most prevalent in the late rainy season, system,we canpredict that resourceabundance patchyflooding and antswarmavailability may or availabilityshould have decreasedover the be a normalfactor of mortalityfor juvenileant- courseof the study (MacArthur 1972, Cormell followersthat arejust learninghow to effective- 1978, Newton 1998). ly find swarmsand forageon their own. During How do thosepopulation changes correlate thoseyears, inundation may have lesseffect on with the biotic and abioticinfluences of army adult ant-followers,which generallycomplete ant colonydensity and rainfall?Given that E. nestingby the time inundationpeaks and thus burchellicolony density did not vary signifi- have loweredenergetic demands, coupled with cantlyfrom year to year,army ant colonyden- a clearspatial map of nearbyarmy ant colonies sity in itself likely had little influenceon avian (whichyoung birds lack). Of course,dry-season population dynamics. Precipitationpeaks, on populationcrises could also have occurredin the other hand, were shifted forward in one of monthswhen I wasnot present. the three rainy seasons(September 1999-April Shiftsin precipitationlevels, lower availability 2000) and correlatetemporally with the drop of E. burchelliswarms, and subsequentincrease in adult survivalof R. melanosticta(Fig. 3.4 and in avian home-rangesize may accountfor the Table4.1). During eachrainy season,floodplain sharpdrop in adult survivalseen in R. melanost- forest at CochaCashu becomes patchily inun- ictafrom 1999to 2000.That subordinatespecies dated,creating heterogeneous "islands" of dry receivedthe mostinterspecific aggression from groundamid wet depressionareas that fill with dominant obligate ant-followers at swarms approximately0.1-1.0 m of standingwater (S. (Table 3.4). Fully 45% of P. nigromaculatadis- K. Willson pers. obs.). When floodplain for- placementsand 53% of M. fortisdisplacements est becomesinundated, E. burchelliarmy ants were directedat R. melanosticta.Additionally, R. 44 ORNITHOLOGICAL MONOGRAPHS NO. 55 melanostictaindividuals were highly aggressive find prey at terrestriallyforaging E. burchelliand intraspecifically(52% of total displacements). L. praedatorswarms, and other factorsrelated Competition for foraging positionsat swarms to nesting successand nest predation may af- with a high prey-basefor birds was extremely fect each speciessynergistically. The variable intenseduring November-December 1999. Based environmentof CochaCashu's floodplain forest on ratesof displacementsper meterper obligate may contributeto each species'having particu- antbird,there was significantlyhigher interfer- lar behavioraland ecologicaladaptations that encecompetition along E. burchelliswarm fronts allow it be more successful than others under in November-December1999 as comparedwith certainenvironmental and temporal conditions September1999 (t = 4.27,df = 165,P = 0.001).If R. (MacArthur and Levins 1967, Pianka 1981, melanostictaindividuals were repeatedlychased Abrams 1983, Ives 1995). An underappreciated away from availableswarms by dominantcom- method of diversification and coexistence of eco- petitorsduring that period of intenseinterference logicallysimilar speciesmay be their manner competition,some individuals may havestarved of tolerating fluctuationsin the environment to deathfrom lackof feedingopportunities, even (Janzen1967). By periodicallydepressing total afterincreasing their home-rangesize by a mag- populationand varying the relative abundance nitude of 2-3x (see Table 3.5). As home-range of different obligateant-following species over size increases,birds may becomeless efficient time, such factorsmay contributeto the high at finding ants and facemore competitionfrom number of speciesable to share the "bligate conspecificsthat are likewise increasingtheir army ant follower"niche at CochaCashu. home ranges. Thefloodplain forest of CochaCashu reminds Regularseasonal inundation may be one fac- us that althoughtropical climates are certainly tor affectingthe observedlack of recruitmentof "stable" when compared with the seasonal young birds, even in what appearto be "good" fluctuationsof the temperatezone, population years (Table4.2). Populationdeclines resulting stability may be unattainablein some special- from lack of recruitment were evident in three of ized tropicalguilds, and particularlyin species the five speciesover three years,and suggested that are nonterritorial.Leigh (1982) stated that in a fourth species. In long-lived sedentary "presumably,the stabler the environment,the tropical birds, recruitmentwill vary from year more specializeda speciescan be and still sur- to year,depending on bioticand abioticfactors. vive, and the more speciescan coexist."In the However, recruitmentis generallybelieved to caseof obligateant-followers, the instability of be low in tropical birds (Fogden 1972, Skutch the floodplainenvironment may be actingto in- 1985, Morton and Stutchbury2000, Robinson creasespecies diversity by periodicallydepress- et al. 2000b).Rainfall patterns, opportunities to ing somespecies' populations. 5. NESTING AND REPRODUCTION IN A GUILD OF OBLIGATE ARMY ANT FOLLOWERS IN AMAZONIAN PERU

NEw DATAON nestingand reproductionof RESULTS Neotropical antbirds (Thamnophilidae)-- and, to a lesser extent, of GYMNOPITHY$ $ALVINI (Dendrocolaptidae)--areaccumulating quickly (del Hoyo et al. 2004). However, there is still Nest.--Three nests were found from 1998 to muchto learnregarding nest structure, incuba- 2000 (Table 5.1). All nestswere locatedin ma- tion, and adult behavior within those families. ture floodplainforest, either on high groundor Many specieshave never been scientifically in low-lyingswamp. Nests were in dead palm studied,and basic reproductive parameters are stumps41-50 cm in height. All nest stumps still unknown.Antbirds and woodcreepersare were narrow in width (6-9 cm) and containedan endemicto the New World tropicsand their opencavity 5-26 cmin depth.Height of thefloor centersof diversityare locatedin the Amazon cavity varied from ground-levelto 45 cm. All Basin.The 209 antbirdspecies have a diversity cavitieswere completely open on top,providing of nest types--includingopen cup, enclosed an unobstructed view of the nest and no cover hangingpouch, open-topped cavity, and oven- from rain. Dead palmqeaf material sparsely shaped nests (Skutch 1996, del Hoyo et al. linedthe bottom of eachnest cavity (Fig. 5.1). 2004)--with nestheights ranging from ground Eggs.--Oneegg was measured,at 16.5 x 23.7 to canopy.All 52 speciesof woodcreepersare min. All observedeggs were smooth,with a assumedto be secondaryhole nesters (del Hoyo light-pinkmatte surface, covered with maroon et al. 2004). specklingconcentrated at the blunt end. An in- Within the specializedguild of obligateant- cubatingpair wasseen foraging at an antswarm following antbirdsand woodcreepers,many 256 m from its active nest. That swarm observa- species'nests remain undescribed.At the start tion was made between 0745 and 0930 hours; I of the present study in 1997 (Wilkinson and have insufficientdata to saywhether one indi- Smith1997), only one of the five studyspecies vidual arrivedand the otherimmediately left to (M. fortis)had a describednest. Myrmeciza fortis resume incubation duties. nestson theground in covereddome-nests. Few Nestlings.--Oneobserved nest was success- nestshave beenfound of the many Myrmeciza ful, fledgingtwo young.On day 10, youngwere species,but M. exsul,of CentralAmerica south banded,measured, and weighed (Table 5.2). The to Ecuador,builds a concealedcup nest"amid femalenestling was larger than her malesibling low vegetationatop [a] loosefoundation of low in weight,wing, and tarsusmeasurements, and plants"(Zimmerand Isler 2003). bill measurements were similar between the Nestsof the threeother obligate antbird spe- two. The male nestlinghad one maturebotfly cies at Cocha Cashu (P. nigromaculata,R. mela- larva on the side of his head. Nestlingswere nosticta,and G. salvini) are describedbelow. I sexuallydimorphic in color,as are adults.The have publishedpreliminary nest information male was dark gray,with brown wings,while elsewhere for R. melanosticta and G. salvini the female was primarily rufous. Parentsof (Willson2000), but hereI presentmore detailed nestlingswere observedforaging at antswarms descriptionsof nests,eggs, nestlings, and paren- <216 m from their active nest. One nest was fol- tal behaviorin thosespecies. A nestand eggsof lowed from egg stagethrough fledging; chicks P. nigromaculatawere describedfrom Colombia left the nestin the morning12 daysafter hatch- (Cadenaet al. 2000),but the hatchlingsdisap- ing (Table5.3). peared soonafter birth. Here, I describeaddi- Fledglings.--Bothof the fledglingsremained tionalnests for that species,as well as detailed with their parentsfor at least three months feedingbehavior by threeattending adults at one Immediatelyafter fledging,the chickshid in nest.We neverfound a nestof the woodcreeper densevegetation behind antswarmfronts and D. merula. However, I have included dates of werefed by their parents.On 24 November2000 nesting at Cocha Cashu and information on (9 days after fledging),one of the fledglings breedingbehavior for thatspecies. was observed 10 m from an E. burchelli swarm

45 46 ORNITHOLOGICAL MONOGRAPHS NO. 55

o o e• • o o o •

z

•. E OBLIGATE ANT-FOLLOWING BIRDS IN PERU 47

FiG.5.1. Stump containingnest and two eggsof G. salvini. front. By mid-February,fledglings were gener- ally foragingon their own but still occasionally begged food from their parents.There was no indicationthat fledglingswere divided between parents,with one parent alwaysproviding for the samefledgling; Willis (1967)described that type of division in G. leucaspisof Panama.Each fledglingG. salvinibegged and wasfed by either parent, and the family remained togetherat all times while under daily observation.Neither fledglingwas seenthe followingyear, although it is possible that they dispersed and were missedduring the shortfield-season of 2002.

RHEGMATORHINA MELANOSTICTA

Nest.--Four active nests were found from 1998 to 2000 (Table5.1). One nest site was reused a second time, presumably by the same pair, after an initial successfulbreeding (see Table 5.3).All nestswere locatedwithin maturehigh- ground forest.Three nests were locatedin verti- cal cavitiesleft by fallen leavesalong the trunk of mature Scheeleaspp. palm trees(Fig. 5.2);the other nestwas in the hollow verticalcavity of a dead sapling.Nest heightranged from 21 to 90 cm aboveground. Cavity depth rangedfrom 7 to 11 cm, and width rangedfrom 5.5 to 10.5 cm. The nestitself was a shallowbed of dry, shred- ded palm leaves,completely open at the top. Eggs.--Three eggs were measured from two separate nests. Eggs averaged approxi- mately 17 x 22 mm and displayed a smooth matte surface, light pink with longitudinal dark-maroonstreaks, interspersed with maroon 48 ORNITHOLOGICAL MONOGRAPHS NO. 55

ø • ' •. z • z • •

Z Z Z Z Z Z Z Z Z Z OBLIGATE ANT-FOLLOWING BIRDS IN PERU 49

andapproached the swarm. The adult female ag- gressivelydisplaced her mate four timesin one hour when he approachedthe fledgling. Interestingly,the nest of fledgling99Rgw-s was re-usedless than two monthsafter 99Rgw-s fledged.Because of time constraints,I was un- ableto determinethe identityof the pair attend- ing the two eggs,but a vocalizingfemale near the nestconfirmed the speciesas R. melanosticta. A related antbird of Panama (G. leucaspis)is noted to occasionallyreuse the samenest twice (Zimmer and Isler 2003). Fledgling 99Rgw-s successfullysurvived to adulthood(confirmed by resightdata in later years),so if the pair at the nest was her parents, they were renesting while caring for a dependentfledgling. That behavior has not been documented in other ant-followers. FIG.5.2. Nest and two eggsof R. melanostictain the leaf-sheathof a live Scheeleasp. palm tree. The second successfulnest (00Rgp-s) de- scribedhere fledged two young, which were first seen at an antswarm when they were 13 speckling;there was less color at each end. days out of the nest.The fledglingsremained Incubatingadults were observed_<318 m from in nearbycover and were fed by their parents. active nests. As with G. salvini,each parent provisioned both Nestlings.--At least two of the four nesting fledglings.Over the nextfew weeks,fledglings attemptswere successful,and one failed. The clumsily began foraging on their own. One fourth nestwas found on the last day of a field fledglingbegan being courted by an adultmale season, and its outcome is unknown (Table when she was less than a month out of the nest. 5.3). One nestfledged two young;another nest The adult male was acceptedby the fledglings fledgedonly one,though two nestlingswere in parents and began regularly foraging with it until day 11. Althoughboth of the latternest- them, taking over most duties of provisioning lings werefully featheredby then,the one that the fledgling. Both fledglingsremained with disappearedfrom the nestdid not survive(i.e. it their parents for at least three months,when was subsequentlynever seenwith its parents). the field seasonended in mid-February.Neither Its siblingfledged in the morning13 daysafter wasseen the followingyear, although the short hatchingand was measuredon the day before period of field days (3.5 weeks) or dispersal it left the nest (Table 5.2). Nestlings from the from the study site may have precludedtheir secondsuccessful nest were measuredone day beingobserved. beforefledging, or at -12 daysold (Table5.2). Adultswith nestlingswere observed foraging at PHLEGOPSIS NIGROMACULATA antswarms _<386 m from their active nest. Fledglings.--All fledglingshad a scalloped Nest.-- Three nests were found in the 2000- back similar in pattern to an adult female's. 2001 field season,of which two fledgedyoung Museum notes from Willis (1969) state that (Table 5.3). If all nests had been successful,nest both male and female fledglings have scal- activitywould have spanned October-February. loped backs,and males presumablygain the All nests were located in low vertical cavities unscallopedadult plumagein their next moult. with completelyopen tops:two were in stumps Fledglingshad gray eyeskinand a dark-gray of dead palm saplings,and the other was in a crest.Fledgling 99Rgw-s was first seenforaging live tree (Table5.1). Nest height aboveground at an antswarmone month after it fledged.Given rangedfrom 8 to 51 cm.Cavity depth was from that thebanded parents had beenseen regularly 10 to 15 cm, and cavitydiameter was 4.5 to 10 with that ant colonythroughout October, that cm. Nestswere composedof a few smallpieces was likely the first day the fledglingleft cover of dry palm frond, forming a looseflat lining 50 ORNITHOLOGICAL MONOGRAPHS NO. 55 insideeach cavity. Nest cavity locationsvaried cooperativebreeding among thamnophilidsto in habitat,from gapedge to swampedge to ma- speculateas to whether any of the adults in- ture high-groundforest. volved were related. Eggs.--No eggs were measured for this species.Egg colorationwas similar to that of DENDROCINCLA MERULA R. melanostictaand G. salvini:pale pink with maroonstriations. Incubating individuals were Although the nest of this woodcreeperspe- observedtraveling <360 m from theirnest to for- cies remains undescribed,I present informa- age at antswarms. tion here on breeding biology and fledgling Nestlings.--Twoof the three nestseach suc- behavior.At CochaCashu, nesting appears to cessfullyfledged two young(Table 5.3). Young beginin the middle of the rainy season,around remained in one nest for -13 days. Four nest- December.Beginning in January,three females lingswere measuredat -9 and -11 daysold, re- (radiotaggedor banded or both) brought one spectively(Table 5.2). Botflylarvae parasitized fledgling each to antswarms.No fledglings all nestlings;each nestling had oneto sevenma- were ever seenat swarmsin the early months ture larvae at the time of banding (Fig. 5.3). The of the rainy season (September-December). individualwith sevenlarvae did not showany Fledglingsappeared to be almost identical to negativesigns in its growth as comparedto its adultsin thefield, with a slightlydarker bill and siblingand successfullyreached adulthood (i.e. a bit lesswhite on the throatarea. Fledglings so- licited food from their mothers, but also from was sighted the following year). Adults with other D. merulafemales, which either ignored nestlingsin the nest were observedtraveling <161m to forageat antswarms.However, adults or pecked at them. Fledglingsoften "played" with eachother at swarmsrather than foraging. werenot locatedon manydays, and couldhave One individual would perch directly behind used L. praedatorswarms of unknown distance from the nest. anotheron a verticaltree and poke its bill into the back feathers of the other. Both would then Fledglings.--Allfledglings Were observed at fly to a new tree and reversethe "pokingorder." antswarmsafter fledging, and were still receiv- Fledglingbirds would repeatthat behaviorwith ing food from adults after 2.5 months out of up to three nonrelatedindividuals for <10 min the nest. Fledglingshad black eyeskin,which at a time, stoppingfinally to demandfood from seemedto slowly turn red over the courseof their respectivemothers. At times a fledgling the next many months:in February,older ju- would poke its mother or a nonrelatedadult, veniles, presumably from early-rainy-season which would usually fly away to a different nests (September-October),had dull reddish perch rather than peck at the fledgling.Willis eyeskin. (1978) describessimilar behavior in D. merula Helpersat the nest.--One nest was observed fledglingsin Brazil. over the entire nestling stage and confirmsfor Fledglingsstayed with their parentsfor at the first time that this speciesis a cooperative least six weeks, but began to become more breeder (Table 5.3). Over the courseof eight independentby that time. One individual, at days, we observedthe nest for -18 h and re- a month post-fledging,was seen alone at an cordedfeedings by one adult female and two antswarm, 600 m from a second swarm where adult males.At time of nesting,the female (Ps- its mother was foraging.The sameindividual by) was at leasttwo years old, one male (Pfs-fg) was observedlater that week roostingat least was at least four, and the secondmale (Ps-pg) 500 m from where its motherregularly roosted. was at least two. In 77 observedfeedings, the The fledgling roostedbetween the buttressesof femalefed 20% of the time, the older male (Pfs- an emergentFicus sp. tree,3 m from the ground. fg) fed 42% of the time, and the younger male Rather than perch on a branch, it held verti- (Ps-pg)fed 36% of the time. In 2% of the feed- cally to the smoothbark with its clawsand tail, ing, the adult'sidentity was unknown.Blood makingit virtually impossiblefor a predatorto was collectedfrom both nestlingsand all three reach.The next day,mother and fledglingwere adults, and DNA testingwill be carried out to againforaging together at a swarm. determine paternity and relatednessamong Although a known nest locationis not avail- all individuals. Too little information exists on able, I calculated the minimum distance a OBLIGATE ANT-FOLLOWING BIRDS IN PERU 51

FEe;.5.3. Phlegopsisnigromaculata nestlings with larval botfliesembedded in skin around neck and head. The first individualhad sevenlarge larval botflies at time of measurement;the second(inset) had three. female traveled from her active nest. Females radiotelemetry data taken during the present were identified as having nestlings over a study show that, at leastfor R. melanostictaand given range of days by counting backward to G. salvini,the female sits on the nest at night. probable nestling dates from the initial day That is consistent with Skutch's antbird data that a female was observed at a swarm with a from Costa Rica (Skutch 1996). In all three spe- fledgling (using information for the genus in cies, the nest is placed inside a vertical cavity Marantz et al. 2004). By examining,post-hoc, the with a completelyopen top. The cavity can be distancestraveled betweenswarms by females a rotting tree stump, a small live tree with a with active nestsand taking the midpoint be- natural cavity formedbetween branches, or the tween swarms, I found that females traveled sheathof a fallen palm frond still attachedto a a minimum of 300 m from the nest while it mature Scheeleasp. tree. Although samplesizes was active. Given that D. merula individuals are small, it seemsthat P. nigromaculalamay be are strong fliers and sometimestraveled more restricted to tree cavities and not use Scheelea than a kilometer between successivelyvisited palm sheaths,perhaps because the latter are too swarms, the actual distance a nesting female small.Rhegmatorhina melanosticta is the only spe- may fly is likely much higher. ciesfound, thus far, to nest in all three types of cavity.Further study will elucidatewhether the DISCUSSION smallerG. salvinialso uses palm-sheath cavities; in Panama,the closelyrelated G. leucaspisnests Nests of G. salvini, P. nigromaculata,and R. in palm-sheathcavities attached to the trunk of melanosttctashare many attributes, perhaps Scheeleaspp. trees(Willis 1967). becauseof the species'close phylogenetic his- Nest-site selection may contribute to niche tories. Male and female of the three species breadth among the obligate ant-followersof share incubation of eggs and nestlings, and CochaCashu, and may actto limit densitiesof the 52 ORNITHOLOGICAL MONOGRAPHS NO. 55 dominantP. nigromaculatabelow what the army after fledging one nestling.That fledgling suc- ant resourcecould support. If verticalstumps are cessfullysurvived to adulthood,and was seen a limitingresource, smaller species that cansuc- in subsequentyears. More observationsof cessfullynest in Scheeleaspp. palm sheathsmay nest attemptswill help answerthe questionof be ableto increasedensities to capitalizeon the whetherjuveniles stay with parentswhile the availabilityof thearmy ant resource. Brightsmith latter attemptto rear a secondbrood. (2000) experimentallyincreased availability of Mate fidelity and mate switching.--Thefour nest holes at Cocha Cashu to determine whether antbirdstudy speciesmaintain pair bondsover secondarynest-holers were limited by low avail- consecutivebreeding seasons.In the courseof ability of nesting substrates.He found little fieldwork,many bandedpairs were identified evidencefor that hypothesis,which suggests and observedtogether for up to three seasons. that P. nigromaculatadensities may not correlate A total of 37 mated pairs were identified:13 R. with nest-siteabundance. However, Brightsmith melanostictapairs, 9 G. salvini,9 M. fortis, and (2000)investigated availability of secondarytree- 6 P. nigromaculata(Table 5.4). Thosepairs were holes,whereas P. nigromaculatarequire open- identified by repeated observationtogether, topped stumpswith a narrow range in width mate feedings,nesting attempts,or a combi- and height.Palm stumpsare alwaysin a state nation thereof. Phlegopsisnigromaculata pairs of decay,and--unlike tree holes--generallydo were more 'difficult to determine, because of not lastfor morethan one breeding season (S. K. their propensityto forage in groups and mate Willsonpers. obs.). cooperatively,and are not representedin Table Eachspecies constructs a shallow,flat nestof 5.4. Fourteen known instances of mate switch- strippeddead palm leaves,about 5-12 cm long ing occurredover four years:seven in the R. x 1 cm wide at thebottom of the cavity.The nest melanostictapairs, four in G. salvini,two in M. is basicallya bed for the eggsand is not cup- fortis, and one in P. nigromaculata.In all but one shapedor twined together.As a rule, antbirds instance,the individual that was replacedfrom lay two eggsper nest (Skutch1985). Eggs of the pair was never seenagain and presumably all three speciesare similarly colored,with a died, which promptedthe remainingindividual hght-pink backgroundand maroon striations to seek a new mate. Data from 2003 are added to or speckling.No nestswere observedthrough Table5.4, because the only observedinstance of the entire egg stage,but one nest was observed mate switchingamong living partnersoccurred with eggs for nine days before they hatched. in that year (seeR. melanostictafemale s-yp). An Skutch (1996) gives the incubationperiod for individual may quicklyfind a new mate if its the relatedG. leucaspisas 15-16 days.Nestlings partner dies. For example, a banded G. salvini of thatspecies remained in nestsfor 12-13days, female,which was paired with a bandedmale which is a bit shorterthan Skutch's(1996) ac- for at leasttwo years,paired with a new mate countof 13-15 days. on her homerange within two weeksof her first In southeasternPeru, obligateant-followers mate's death (the radiotaggedmale was swal- seemto nest in the early to middle rainy sea- lowed by a horned toad [Ceratophryscornuta]). son. All nestswere found from Septemberto A year later, the female was again with a new January,and it seemslikely that nestingtapers male,on a differentpart of the studyplot, ~500 off after Januaryunless nesting attemptsre- m from the boundary of the home range she peatedly fail. Juvenilesof each speciesbegan shared with her first known mate. appearing at antswarmsin October, and by The antbirdfamily is largeand currentlypara- Januaryfledglings accompanied many adults. phyletic(Hackett and Rosenberg1990, Isler et al. Fledged young of all speciesdescribed here 1998, Zimmer and Isler 2003). A re-examination stayed with their parents for at least three of the thamnophilidphylogeny is underway, months. If obligate ant-followersmust stay and variousresearchers are using geneticdata, with their parents for a protractedperiod to vocal characters,and natural-historyinforma- learn the specializedbehavior of following tion in their revisions (Hackett 1993; Isler et al. army ants,it seemsreasonable that they would 1997,1999; Bates et al. 1999).Within the family not renest if the first nest were successful. Dendrocolaptidae,taxonomic con_fusion is still However, one R. melanostictapait apparently widespreadat many levels(Marantz et al. 2004). renested in the same nest less than two months Besidesadding to our limited knowledgeof the OBLIGATE ANT-FOLLOWING BIRDS IN PERU 53

reproductionand natural history of tropicalbirds, charactersfor phylogeneticstudies (Sheldon and nesting informationmay provide additional Winkler1999, Zyskowski and Prum 1999).

TABLE5.4. Obligateant-following antbird pairs maintain long-termpair bonds.Individual disappearances may representdeaths or "divorce."One divorcewas observed(see R. melanostictafemale s-yp in 2000-2001 and 2003).No otherindividuals were seenagain after leavinga mate.Mated pairsof P. nigromaculatawere difficult to discern,and are not listedhere. Bird pahtsare listedby color-bandcombinations, with males precedingfemales. Bold print signifiesa changein pair composition;"unb" indicates an unbandedbird. Data are from Cocha Cashu, Peru.

1998 1999 2000-2001 2002 2003 G. salvini yy-s,s-fg yy-s, s-oy yy-s, s-oy yy-s, bw-s yy-s, bw-s go-s,bw-s go-s,bw-s pp-s, s-pp pp-s, s-pp pp-s, s-pp rp-s, s-pp bp-s, s-pb s-yp, s-yp s-yp, s-yp s-yp, s-yp r-sr, unb s-by,y-sb unb, rb-s R. melanosticta S-WW,s-f• s-ww, s-ff pr-s, s-wp unb, s-wp by-s, yg-s by-s, yg-s by-s, gp-s by-s, s-yp s-rp• S-OW s-rp, gp-s s-yr, s-yp s-yr, s-yp s-yr, wg-s s-yr, ? s-yg,s-pb r-sr, p-sr s-py,s-gb s-py, p-sr M. fortis oh-s, r!g-s ob-s,g-sy oh-s, rr-s unb, rr-s x-s, s-bw x-s, s-bw s-rg, s-pp s-rg, s-pp s-yr, ry-s s-yy,ff-s s-yy,ff-s s-yy,unb s-bb, bb-s s-bb, unb r-sg, s-gr

6. CONCLUSIONS AND FUTURE QUESTIONS

SUMMARY OF COEXISTENCE PATTERNS biodegradeand fall apart after a breedingsea- son.Not only are the food resourcesof obligate THE PRESENTSTUDY contributes new insight ant-followers continually moving across the towardunderstanding the complexitiesof com- landscape;nest sitesare also changingseason- munity organization--amajor goal of the study ally as older palm stumpsdegrade, leaves fall of ecology.The five obligateant-following birds from Scheeleaspp. trunks, and small palms die at Cocha Cashu coexistthrough more subtle and createnew stumps. means than segregationby simple ecological Population fluctuationsdecreased interfer- factors like body size, habitat, or prey size. ence competitionamong individuals and may Two of the five species(D. merulaand M. for- increasethe ability of the subordinatespe- tis) preferredone or the other of the two army cies R. melanostictato coexistwith the larger, ant speciesas a foragingresource. The wood- dominant P. nigromaculataat Cocha Cashu. In creeperD. merulafurther segregatedfrom the termsof army ant use patterns,R. melanosticta's four antbirdsin its useof white-lippedpeccary niche spacewas not well differentiatedfrom herds--whichact as beatersof arthropodprey either that of the larger P. nigromaculataor the in a mannersimilar to that of the army ants--as smaller G. salvini.I suggestthat R. melanosticta, a foragingresource. The threeantbirds that did in particular,may rely on competitiverelease not preferone army ant speciesover the other resultingfrom populationfluctuations of the (P. nigromaculata,R. melanosticta,and G. salvini) dominantspecies, as well as fewer constraints are segregatedby body mass,which may allow on nest-site selection, to coexist in the Cocha differentialuse of spacealong a swarmfront's Cashubird assemblage. width. That would permit a smaller, more Few speciesof tropicalforest birds have been subordinatespecies to "fit" alongthe front of a intensively studied. The present research-- swarmthat was already"full" to a larger spe- focusingon a guild of highlyspecialized "atypi- cies.These factors add to descriptionsof niche cal" birds--provides important insight into differentiationby perch type and angle pub- hypothesesconcerning the stability of more lishedby Willis (1968,1969, 1978, 1979, 1985). "normal" (i.e. generalist,territorial, insectivo- The bird species'population dynamics were rous) tropicalspecies; the resultssupport the not stableover five yearsof data collection,and hypothesisthat avian territoriality contributes the total populationof obligateant-followers to the stability of tropical bird populations dropped by almosthalf over the courseof the (Greenbergand Gradwohl1986, Newton 1998). study.I suggestthat periodicpopulation fluctu- All three widely rangingobligate antbird spe- ationsare a normal occurrencein obligateant- cies(P. nigromaculata,R. melanosticta,and G. sal- following guilds. Declinesover the courseof vini)exhibited significant changes in population the studywere partially attributable to changes levelsover three years,whereas the territorial in availabilityof army antsduring forest flood- antbird species(M. fortis) displayeda constant ing, which may have decreasedadult survival population density. Adult survival rates in in R. melanosticta.Because I did not study the mostyears were constantacross species, which birds in the dry seasons,it is unknown whether suggeststhat juvenile recruitmentis the key dry-seasonevents contributed to mortality. On difference, and is either lower or more volatile BCI, loss of Pha. mcleannaniwas attributed to a in nonterritorial bird speciesthan in related long 1968drought that causeda shortbreeding territorial species. That prediction deserves seasonand lack of replacementsdue to meso- further testing in both obligate ant-followers predation(Willis 1973,1974). Declines at Cocha and other bird guilds with variable territorial Cashuwere also attributableto a low juvenile behavior,while controllingfor environmental, recruitment rate that failed to offset normal temporal,and landscapeeffects. The population adult mortality. Nest-siteavailability may con- fluctuationsof the widely ranging study spe- tribute to fluctuations in recruitment, because cies,documented here, provide a counterpoint suitablestumps and palm sheaths will generally to a number of studies that have found stable

55 56 ORNITHOLOGICAL MONOGRAPHS NO. 55 population dynamicsin territorial tropical birds. Somekey ecologicaldifferences that I foundamong the five obligate ant-following speciesare summarized inTable 6.1. Nest-site selection contributes to niche breadthamong the obligate ant-followers, and may act to constrainpopulation den- sityof thedominant species P.nigromacu- lata. Dendrocinclarnerula, a woodcreeper, presumablynests in tree holeswith a verticalopening, like other woodcreepers. The antbirdM. fortis differentiatesfrom the otherobligate antbirds in itsconstruction of terrestrialdome nests. Phylogenetics likely contributeto that differentiationby con- strainingthenest-substrate choicesof the threemore closely related antbirds P. nigro- rnaculata,R. rnelanosticta,and G. salvini,all threeof which nest in open-topped vertical stumps.However, R. rnelanostictaalso nests in Scheeleaspp. palm sheaths,which act as verticalcavities. That tree genus is common throughoutthe Neotropics;Cocha Cashu has 25 adult trees ha-• (J. Terborgh pers. comm.).High availability of palmsheaths for nestingwould remove a limiting factor onbird species that could profit by using them.Presumably, G.salvini also uses palm sheaths,ashas been demonstrated forits congenerG.leucaspis in Panama (Zimmer and Isler 2003). Scheeleapalm sheathsare small and likely unsuitablefor the domi- nantspecies P.nigromaculata. Verticalopen- toppedwood stumpsmay thus act as a key limitingfactor that constrainslocal popula- tionsof P.nigromaculata fromexpanding to usemore of the availablearmy ant resource. Constraintson the dominantspecies may providethe socially subordinate R.rnelano- stictaand G. salviniwith the nichespace they needto coexist with it.

FUTUREQUESTIONS

The work presentedhere opensthe door for new questionsabout the ecology and behavior of obligate ant-followingspecies; the interactionsof specializationand vulner- ability, tropical stability,and territoriality; andthe potential role of source-sinkdynam- ics betweencontiguous floodplain and terra firmehabitats in the Amazonbasin. All data presentedhere were collectedin the rainy OBLIGATE ANT-FOLLOWING BIRDS IN PERU 57 season,and it is not known how the army ant time. However, the question of how stump and obligateant-follower system changes in the availabilityaffects the breedingbehavior of the dry season.Data from BCI (Willis 1967;Franks specieswould be easier to answer in a short- 1982a,b) suggestthat E. burchellidensities re- term study. If "helpers at the nest"are con- mainstable through seasonal changes. However, strainedby nest availability,we might predict L. praedatorswarms may be less availableto that helpersdisperse when nest sitesare made birdsin the dry season,when antsmay forage available.Results presented here from onenest moreoften below ground or at night (Schneirla suggestthat males may breed cooperatively 1971, H611doblerand Wilson 1990). How does with a singlefemale, but too little information that affectthe territorial systemof M. fortis?At is known on P. nigrornaculatareproductive biol- the constantdensity that I estimatedduring my ogy to discountfemale helpers as well. Finally, study, an increasein territory size would force groupsof P.nigrornaculata may form for reasons pairs to have someoverlap in neighboringter- otherthan nesting constraints. Do largergroups ritories.A comparisonof M. fortis behaviorand repel smallerconspecific groups from swarms? spatialdynamics across seasons would provide Further study can suggestanswers to these insight into the effects of temporal resource questions. availability on territorial plasticity.Behavioral Are floodplainforests in the Amazon a sink plasticity,as well, may occur if M. fortis can- or a sourcehabitat for obligate ant-following not affordits conspecificintolerance because of birds? Floodplain habitat constitutesonly 3% larger territory size in the dry season. to 4% of the Amazon basin,yet the ubiquitous A secondquestion, of largerscope, is how ter- systemof rivers running like veins through- ritorialitycontributes to the stabilityof tropical out the Amazon means that no region is far bird populations.In the obligateant-followers, from floodplain (Goulding 1993). For obligate data presented here are consistentwith the ant-followers, floodplain may offer higher hypothesisthat nonterritorialbirds have larger food-resource density, but at the potential populationfluctuations (both positive and neg- costof higher seasonalvariability. Floodplain ative)than territorialspecies. Most tropicalbird variability and ensuing changesin resource work has concentratedon territorialspecies; an availability may cause alterationsin relative investigationof this questionwill profit from abundancesof bird speciesin differentyears, collectionof long-termpopulation data on vari- as was seen in the five study species.In good ous tropicalguilds that includeboth territorial years,juveniles may disperseinto terrafirrnefor- and nonterritorialspecies. est. Floodplainmay act as a sourcefor the terra My proposedhypothesis that nest-siteavail- firme, which shouldhold lower populationsof ability is a key limiting factorfor populations insectivores because of lower moisture and thus of P. nigrornaculataneeds to be tested.Nest-site lower arthropod abundance (Williams 1941, limitation may act to depresspopulations of Janzen and Schoener 1968, Janzen 1973, Wolda this dominantspecies below the environment's 1978,Levings and Windsor1982). However, bad carryingcapacity based on food-resourceavail- years for some speciesin the floodplainmay ability. Becausethey use palm sheaths,the translateinto the terrafirrneacting as a source,as subordinatespecies R. rnelanostictaand G. sal- individualsimmigrate into the floodplainand vini are likely lessconstrained in nestsubstrate raisepopulations of thosespecies back to carry- and would be able to utilize some of the space ing capacity.Long-term studies of dispersalof at antswarmsbeyond what the P. nigrornaculatajuveniles from adjoininghabitats, in conjunc- populationcan occupy. Nest-site limitation may tion with measuresof resourcedensity in both alsoexplain another phenomenon specific to P. habitats,are necessaryto answerthis question. nigromaculata--cooperativebreeding. If nesting Source-sinkdynamics are well establishedin stumpsare so limiting that there are few benefits temperateregions, where fragmentation,pre- to youngerbirds in leavingthe natal territory to dation, and nest parasitismclearly define what initiate pairing and nesting,that would imply areas act as sources versus sinks (Donovan et a potential causeof the evolutionof coopera- al. 1995, Robinsonet al. 1995, Thompsonet al. tive breedingin the species.An experimentally 2002).In the contiguouslowland rainforestof expandedselection of nest stumpsmay allow theAmazon, the naturaldynamics of floodplain for populationincrease in P. nigrornaculata,over ecosystemsand the greater stability of terra 58 ORNITHOLOGICAL MONOGRAPHS NO. 55 firmeforests may dictatethat sourcesand sinks ecologicaldata for the two army ant species,E. vary year to year, as populationsrise and fall burchelliand L. praedator--datawere not previ- becauseof fluctuationsin rainfall levels,prey ously available for Amazonian populations. availability,competitive pressures, predation, Army antsand obligateant-following birds are and ensuingsurvival and recruitment. amongthe first groupsof organismsto go lo- cally extinctin fragmentedforest (Harper 1987, CONSERVATION APPLICATIONS 1989; Bierregaardand Lovejoy 1989; Stouffer and Bierregaard1995; Bierregaard and Stouffer The natural world is rapidly being de- 1997). Presenceor absenceof thesespecies in graded and fragmentedby human activities. fragmentedor threatenedforest may be an ex- My research,done in a protected area of the cellentindicator of forestdegradation or health Amazon, provides crucial baseline data on (Thiollay 1992, Canaday 1997).Both the army population dynamics and densities, survival antsand obligateant-following birds are poten- rates,and home-rangerequirements of obligate tial indicatorspecies for use in rapid ecological ant-followingbirds. It also provides baseline assessments of threatened areas. ACKNOWLEDGMENTS

I thank my dissertationcommittee mem- who helped in many ways. Funding for this bers-John Faaborg,Rex Cocroft, Bette Loiselle, study was provided by a Teagle Foundation John Terborgh,and Frank ThompsonIII--for Scholarship,a GraduateAssistance in Areas their interest and insight, and for invaluable of National Need (GAANN) Fellowshipfrom suggestionsthat improvedthis manuscript. The the U.S. Departmentof Education,two Trans work presentedhere would have been impos- World Airlines Environmental Scholarships, sible without the competentassistance of the an Organizationof American StatesRegular following field assistants:Luz Jimenez,Lindy TrainingProgram (PRA) GraduateFellowship, McBride, Gwylim Blackburn, Fred Werner, and travelgrants from the Divisionof Biological Daniel Huaman, C. E. Timothy Paine, David Sciencesat the Universityof Missouriand from Wilamowski, Rich Pagen, Dan Hogan, Kevin the North AmericanOrnithological Conference. Svara, Edith Suazo, Stuart Hillman, Silvia ! thank the Instituto Nacional de Recursos CastroDelgado, Stacy Small, James DeStaebler, Naturales(INRENA) of Peru for permissionto Mike Stake, Wendy Jess, Kristin Wornson, live and work at CochaCashu Biological Station Windy Davis, and Nadia Castro Izaguirre. between 1997 and 2002. I would like to acknowl- Gordon Burleigh and Tara Robinsoncarried edgea friendfrom CochaCashu, the late Francis out sex determinationof monomorphicbird Bossuyt,whose dedicationto his work in the species. Discussions with Mercedes Foster, field awed and inspiredme. I would alsolike to Katie Dugger,Jeff Rouder, and BrettSandercock thankthe Matsigenkapeople for their respectof substantially improved the manuscript. theno-hunting boundaries around Cocha Cashu, Specialthanks to my husband,Stuart Hillman, which make it sucha specialplace.

59

LITERATURE CITED

ABRAMS,P. 1983.The theoryof limiting similarity. CHARNOV,E. L. 1976.Optimal foraging:The mar- AnnualReview of Ecologyand Systematics14: ginal value theorem. Theoretical Population 359-376. Biology9:129-136. BATES,H. W. 1862. The Naturalist on the River CHESSER,R. t. 1995. Comparativediets of obli- Amazons.John Murray, London. gate ant-followingbirds at a site in northern BATES,J. M., S. J. HACKETT,AND J. M. GOERCK.1999. Bolivia.Biotropica 27:382-390. High levels of mitochondrialDNA differen- Copy,M. L. 1966.A generaltheory of clutchsize. tiation in two lineagesof antbirds(Drymophila Evolution 20:174-184. and Hypocnemis).Auk 116:1093-1106. COEHLO,C. A. S., C. B. Uvo, AND t. AMBRIZZI. BIERREGAARD,R. O., JR., AND T. E. LOVEJOY.1989. 2002.Exploring the impactsof tropicalPacific Effectsof forestfragmentation on Amazonian SSTon the precipitationpatterns over South understorybird communities.Acta AmazSnica America during ENSO periods. Theoretical 19:215-241. and Applied Climatology71:185-197. BIERREGAARD,R. O., JR.,T. E. LOVEJOY,C. GASCON, CONDIT• R., S. P. HUBBELL, AND R. B. FOSTER. AND R. MESQUITA, EDS. 2002. Lessons from 1996.Changes in tree speciesabundance in a Amazonia:The Ecologyand Conservationof Neotropicalforest over eightyears: Impact of a FragmentedForest. Yale University Press, climatechange. Journal of TropicalEcology 12: New Haven, Connecticut. 231-256. BIERREGAARD,R. O., JR., AND P. C. STOUFFER.1997. CONNELL,J, H. 1961.The influenceof interspecific Understorybirds and dynamichabitat mosa- competitionand other factors on the distri- ics in Amazonian rainforests.Pages 138-155 in bution of the barnacle Chthamalus stellatus. TropicalForest Remnants: Ecology, Management, Ecology42:710-723. and Conservationof FragmentedCmmmunities CONNELL,J, H. 1978.Diversity in tropicalrain for- (W. F. Lauranceand R. O. Bierregaard,Jr., Eds.). ests and coral reefs. Science 199:1302-1310. Universityof ChicagoPress, Chicago. CONNELL,J. H., ANDE. ORIAS.1964. The ecologi- BOWDEN,D.C. 1993. A Simple Technique for cal regulation of speciesdiversity. American Estimating Population Size. Department of Naturalist 98:399-414. Statistics, Colorado State University, Fort DELHOYO, J., A. ELLIOTT,AND D. A. CHRISTIE,EDS. Collins. 2004. Handbook of the Birds of the World, vol. BRIGHTSMITHD. J. 2000. Use of arboreal termi- 8. Lynx Edicions,Barcelona, Spain. taria by nestingbirds in the PeruvianAmazon. DIAMOND,J. M. 1975. Assembly of speciescom- Condor 102:529-538. munities. Pages 342-444 in Ecology and BROWN,J. H. 1975. Geographicalecology of des- Evolutionof Communities(M. L. Cody and J. ert rodents. Pages 315-341 in Ecology and M. Diamond, Eds.).Belknap Press of Harvard Evolution of Communities(M. L. Cody and J. UniversityPress, Cambridge, Massachusetts. M. Diamond, Eds.).Harvard University Press, DOBZHANSKY,t. 1950. Evolution in the tropics. Cambridge,Massachusetts. American Scientist 38:209-221. BROWN,J.L. 1964.The evolutionof diversityin avian DONOVAN,t. M., R. H. LIMBERSON,F. R. THOMPSON territorialsystems. Wilson Bulletin 76:160-169. III, ,•NDJ. FAABORG.1995. Modeling the effects BROWN,J. $. 1989a. Coexistenceon a seasonalre- of habitat fragmentationon sourceand sink source. American Naturalist 133:168-182. demography of Neotropical migrant birds. BROWN,J. $. 1989b. Desert rodent community ConservationBiology 9:1396-1407. structure: A test of four mechanisms of coexis- DURANT,$. M. 1998.Competition refuges and coex- tence.Ecological Monographs 59:1-20. istence:An examplefrom Serengeticarnivores. BURNHAM, K. P., AND D. R. ANDERSON. 2002. Journalof Animal Ecology67:370-386. Model Selection and Multimodel Inference: EMMONS,L. H. 1990. Neotropical Rainforest A PracticalInformation-Theoretic Approach. Mammals: A Field Guide. University of Springer-Verlag,New York. ChicagoPress, Chicago. CADENA,C. D., G. A. LONDOl•O,AND J. L. PARgA. FAABORG,J. 1982. Avian population fluctuations 2000. Nesting recordsof five antbird species during drought conditionsin Puerto Rico. from the Colombian Amazon. Wilson Bulletin Wilson Bulletin 94:20-30. 112:313-317. FEDOROV,A. A. 1966.The structureof the tropical CANADAY,C. 1997.Loss of insectivorousbirds along rain forestand speciationin the humid tropics. a gradient of human impact in Amazonia. Journalof Ecology54:1-11. BiologicalConservation 77:63-77. FEINSINGER,P. 1976. Organizationof a tropical

61 62 ORNITHOLOGICAL MONOGRAPHS NO. 55

guild of nectarivorous birds. Ecological biogeographicrelationships in the Neotropical Monographs 46:257-291. genus Gymnopithys(Formicariidae). Wilson FISCHER,A. G. 1960. Latitudinal variation in or- Bulletin 105:301-315. ganicdiversity. Evolution 14:64-81. HACKETT, S. J., AND K. V. ROSENBERG.1990. FITZPATRICK,J. W. 1985. Form, foraging behavior, Comparisonof phenotypicand genetic dif- and adaptive radiation in the Tyrannidae. ferentiation in South American antwrens OrnithologicalMonographs 36: 447-470. (Formicariidae).Auk 107:473489. FOGD•N,M.P. L. 1972.The seasonalityand popu- HARPER,L. H. 1987. The conservation of ant-fol- lation dynamicsof equatorial forest birds in lowing birds in small Amazonianforest frag- Sarawak. Ibis 114:307-343. ments. Ph.D dissertation,State University of FORSYTH,A., ANDK. MIYATA.1984. Tropical Nature. New York,Albany. CharlesScribner's Sons, Macmillan Publishing, HARPER,L. H. 1989.The persistenceof ant-follow- New York. ing birds in smallAmazonian forest fragments. FP,•NKS,N. R. 1980.The evolutionaryecology of Acta Amaz6nica 19:249-263. the army ant Ecitonburchelli on BarroColorado HIMES,J. G. 2003.Intra- and interspecificcompeti- Island, Panama.Ph.D. dissertation,University tion among the water snakesNerodia sipedon of Leeds,Leeds, United Kingdom. and Nerodiarhombifer. Journal of Herpetology FRANKS,N. R. 1982a. A new method for census- 37:126-131. ing animalpopulations: The numberof Eciton HOLDRIDGE,L. R. 1967.Life Zone Ecology.Centro burchelliarmy ant colonieson Barro Colorado de CienciasTropicales, San Jos6, Costa Rica. Island, Panama.Oecologia 52:266-268. H6LLDOBL•R,B., AND E. O. WILSON. 1990. The FRANKS,N. R. 1982b. Ecology and population Ants. Harvard University Press,Cambridge, regulation in the army ant Eciton burchelli. Massachusetts. Pages389-395 in The Ecologyof a Tropical HOLMGREN,M., M. SCHEFFER,E. EZCURRA,J. R. Forest (E.G. Leigh, Jr.,A. S. Rand, and D. M. GUTIERREZ,AND G. M. J. MOHREN.2001. E1 Windsor, Eds.). SmithsonianInstitution Press, Nifio effects on the dynamics of terrestrial Washington,D.C. ecosystems.Trends in Ecologyand Evolution FRETWELL,S. D., AND H. L. LUCAS.1970. On territo- 16:89-94. rial behaviour and other factors influencing HOOGE,P. N., AND B. EICHENLAUB.1997. Animal habitat distribution in birds. Acta Biotheoretica movement extension to ARCVIEW, version 19:16-36. 1.1. Alaska ScienceCenter-Biological Science GISTON,K. J., ED. 1996.Biodiversity: A Biologyof Office, U.S. Geological Survey, Anchorage, Number and Difference. Blackwell Sdentific Alaska. Publications, Oxford. HUSTON,g. A. 1979.A generalhypothesis of spe- GAuss, G. F. 1934. The Struggle for Existence. ciesdiversity. American Naturalist 113:81-101. Williamsand Wilkins, Baltimore, Maryland. HUTCHINSON,G. E. 1957. Concluding remarks. GENTRY,A. H., ED. 1990. Four Neotropical Population Studies: Animal Ecology and Rainforests. Yale University Press, New Demography.Cold Spring Harbor Symposia Haven, Connecticut. on QuantitativeBiology 22:415427. GOUrDING,M. 1993. Flooded forests of the Amazon. HUTCHINSON,G. E. 1959.Homage to SantaRosalia Scientific American 266:114-120. or why are there so many kinds of animals? GRANT, P. g. 1985. Climatic fluctuations on the American Naturalist 93:145-159. Galapagos Islands and their influence on ISLER, M. L., P. R. ISLER, AND g. M. WHITNEY. Darwin's finches.Ornithological Monographs 1997. Biogeographyand systematicsof the 36:471-483. Thamnophiluspunctatus (Thamnophilidae) GREENBERG,g., AND J. GP,•DWOHr.1986. Constant complex. Ornithological Monographs 48: densityand stableterritoriality in sometropi- 355-381. cal insectivorousbirds. Oecologia69:618-625. ISLER,M. L., P. R. ISLER,AND g. M. WHITNEY.1998. GREENBERG, g., AND J. GRADWOHL. 1997. Use of vocalizationsto establishspecies lim- Territoriality, adult survival, and dispersal its in antbirds(Thamnophilidae). Auk 115: in the Checker-throated Antwren in Panama. 577-590. Journalof Avian Biology28:103-110. ISLER,M. L., P. R. ISLER,AND B. M. WHITNEY. 1999. GRIFFITHS,R., M. C. DOUBLE,K. ORRt AND g. J. G. Species limits in antbirds (Passeriformes: DAWSON. 1998. A DNA test to sex most birds. Thamnophilidae):The Myrmotherulasurina- MolecularEcology 7:1071-1075. mensiscomplex. Auk 116:83-96. GRINNELL,J. 1922. The trend of avianpopulations Iv•s, A. R. 1995.Measuring competition in a spa- in California. Science 56:671-676. tially heterogeneousenvironment. American HACKETT, S. J. 1993. Phylogenetic and Naturalist 146:911-936. OBLIGATE ANT-FOLLOWING BIRDS IN PERU 63

JANZEN,D. H. 1967. Why mountain passesare takesguts (and more) to eat fruit: Lessonsfrom higherin the tropics.American Naturalist 101: aviannutritional ecology. Auk 118:819-831. 233-249. LEVINGS,S.C., ANDD. M. WINDSOR.1982. Seasonal JANZEN,D. H. 1973. Sweep samples of tropical and annual variationin litter arthropodpopu- foliage insects:Effects of seasons,vegetation lations. Pages355-387 in The Ecology of a types, elevation, time of day, and insularity. TropicalForest (E •. G. Leigh,Jr., A. S. Rand,and Ecology54:687-708. D. M. Windsor, Eds.). Smithsonian Institution JANZEN, D. H., AND t. W. SCHOENER.1968. Press,Washington, D.C. Differencesin insectabundance and diversity LOISELLE,B. A., AND W. G. HOPEES.1985. Nest betweenwetter and drier sitesduring a tropi- predation in insular and mainland lowland cal dry season.Ecology 49:96-110. rainforest in Panama. Condor 85:93-95. JULLIEN,M., ANDJ.-M. THIOLLAY.1998. Multi-spe- MACARTHUR,R. H. 1958. Populationecology of ciesterritoriality and dynamicsin Neotropical some warblers of northeastern coniferous for- forest understorey bird flocks. Journal of ests.Ecology 39:599-619. Animal Ecology67:227-252. MACARTHUR,R. H. 1969. Patterns of communities KARR, J. R. 1971. Structure of avian communi- in thetropics. Biological Journal of theLinnean ties in selected Panama and Illinois habitats. Society1:19-30. EcologicalMonographs 41:207-233. MACARTHUR,R. H. 1972. GeographicalEcology: KARR,J. R. 1989.Birds. Pages 401416 in Tropical Patternsin the Distributionof Species.Harper Rain ForestEcosystems (H. Leith and M. J. A. and Row, New York. Werger, Eds.). Elsevier Science,Amsterdam, MACARTHUR R. H., AND R. LEVINS. 1964. The Netherlands. Competition, habitat selection,and charac- KARR, J. R., AND K. E. FREEMARK.1983. Habitat ter displacementin a patchy environment. selection and environmental gradients: Proceedingsof the National Academy of Dynamicsin the "stable"tropics. Ecology 64: Sciences USA 51:1207-1210. 1481-1494. MACARTHUR,R. H., ANDR. LEVINS.1967. The limit- KARR,J. R., D. W. SCHEMSICE,AND N. V. L. BROr,AW. ing similarity, convergenceand divergenceof 1982.Temporal variation in the understorybird coexistingspecies. American Naturalist 101: communityof a tropicalforest. Pages 441453 in 377-385. The Ecologyof a TropicalForest (E.G. Leigh,Jr., MARANTZ, C. A., A. ALEIXO, L. R. BEVIER, AND M. A. S. Rand, D. M. Windsor, Eds.). Smithsonian A. PATTEN.2003. Family Dendrocolaptidae InstitutionPress, Washington, D.C. (Woodcreepers).Pages 358447 in Handbook KRICHER,J. 1997.A NeotropicalCompanion, 2nd of the Birds of the World, vol. 8. (J. del Hoyo, ed. PrincetonUniversity Press, Princeton, New A. Elliott, and D. A. Christie, Eds.). Lynx Jersey. Edicions,Barcelona, Spain. LACK,D. 1944.Symposium on the ecologyof close- MARENGO,J. A. 1992.Interannual variability of sur- ly allied species.Journal of Animal Ecology13: face climate in the Amazon basin. International 176-177. Journalof Climatology 12:853-863. LEBRETON,J.-D., K. E BURNHAM,J. CLOBERT,AND MARRA,P. P., AND J. V. REMSEN,JR. 1997. Insight into D. R. ANDERSON.1992. Modeling survivaland the maintenanceof high speciesdiversity in testing biologicalhypotheses using marked the Neotropics:Habitat selectionand foraging animals:A unifiedapproach with casestudies. behaviorin understorybirds of tropical and EcologicalMonographs 62:67-118. temperateforests. Ornithological Monographs LEIGH,E.G., JR.1982. Introduction: The significance 48:445483. of populationfluctuations. Pages 435440 in The MAY,R. 1973. Stability and Complexity in Model Ecologyof a TropicalForest (E.G. Leigh,Jr., A. Ecosystems. Princeton University Press, S. Rand, and D. M. Windsor, Eds.). Smithsonian Princeton,New Jersey. InstitutionPress, Washington, D.C. McNAIR, J. N. 1982. Optimal giving up times LEIGH,E.G., JR. 1999. Tropical Forest Ecology. and the marginal value theorem. American Oxford UniversityPress, New York. Naturalist 119:511-529. LEIGH, E.G., JR., AND D. M. WINDSOR. 1982. MORTON, E. 5., AND B. J. M. STUTCHBURY.2000. Forest productionand regulation of primary Demographyand reproductivesuccess in the consumerson Barro Colorado Island. Pages dusky antbird,a sedentarytropical . 111-122 in The Ecologyof a TropicalForest Journalof Field Ornithology71:493-500. (E. G. Leigh, Jr., A. S. Rand, and D. M. MUNN,C. A. 1985.Permanent canopy and under- Windsor, Eds.). SmithsonianInstitution Press, story flocks in Amazonia: Speciescomposi- Washington,D.C. tion and population density. Ornithological LEVEY,D. J., AND C. MARTINEZDEL RIO. 2001. It Monographs36:683-712. 64 ORNITHOLOGICAL MONOGRAPHS NO. 55

MUNN,C. A., ANDJ. TERBORGH.1979. Multi-species army ants. University of Kansas Scientific territoriality in Neotropical foraging flocks. Bulletin 44:281-465. Condor 81:338-347. RICKLEFS, R. E., AND D. $CHLUTER, EDS. 1993. NAMS,V. O. 2000.Program Locate II. Pacer,Truro, SpeciesDiversity in EcologicalCommunities: Nova Scotia, Canada. [Online.] Available at Historical and Geographical Perspectives. www. nsac.ns.ca/envsci/staff/vnams/Locate.htm Universityof ChicagoPress, Chicago. NEWTON,I. 1998. Population Limitation in Birds. RIDGELY,R. $., ANDP. J. GREENFIELD.2001. The Birds Academic Press, New York. of Ecuador,vol. 2. Cornell University Press, NORTH AMERICAN BANDING COUNCIL. 2001a. The Ithaca, New York. Instructor'sGuide to Training PasserineBird ROBINSON,S.K., ANDJ. TERBORGH. 1995. Interspecific Banders in North America. [Online.] North aggressionand habitatselection by Amazonian American Banding Council, Point Reyes birds.Journal of Animal Ecology64:1-11. Station,California. Available at nabanding.net/ ROBINSON, S. K., F. R. THOMPSON IILT. M. nabanding/ DONOVAN,D. R. WHITEHEAD,AND J. FAABORG. NORTH AMERICAN BANDING COUNCIL. 200lb. The 1995. Regionalforest fragmentationand the North AmericanBanders' Manual for Banding nesting successof migratory birds. Science PasseLinesand Near (Excluding 267:1987-1990. Hummingbirds and Owls). [Online.] North ROBINSON,T. R. 2000.Factors affecting natal disper- American Banding Council, Point Reyes sal by SongWrens (Cyphorhinus phaeocephalus): Station,California. Available at nabanding.net/ Ecologicalconstraints and demography.Ph.D. nabanding/ dissertation,University of Illinois, Urbana- NORTH AMERICAN BANDING COUNCIL. 2001c. The Champaign. North American BandeLs' Study Guide. ROBINSON,W. D. 1999. Long-term changesin the [Online.] North American Banding Council, avifauna of Barro Colorado Island, Panama, Point Reyes Station, California. Available at a tropicalforest isolate. Conservation Biology nabanding.net/nabanding/ 13:85-97. ORIANS,G. H. 1969.The number of bird speciesin ROBINSON,W. D., J. D. BRAWN,AND S. K. ROBINSON. sometopical forests. Ecology 50:783-801. 2000a. Forest bird community structure in PEARSON,D. L., ANDJ. A. DERIl.1986. Seasonal patterns centralPanama: Influence of spatial scaleand in lowlandforest floor arthropodabundance in biogeography.Ecological Monographs 70: southeasternPeru. Biotropica 18:244-256. 209-235. PEZZI, L. P., AND I. F. A. CAVALCANTI.2001. The rela- ROBINSON,W. D., T. R. ROBINSON,S. K. ROBINSON, tive importanceof ENSO and tropicalAtlantic ANDJ. D. BRAW•.2000b. Nesting successof seasurface temperature anomalies for seasonal understory forest birds in central Panama. precipitationover South America: A numerical Journalof Avian Biology31:151-1 64. study.Climate Dynamics17:205-212. ROPER,J. J. 1996. Nest predation and its impor- PIANKA,E. g. 1966. Latitudinal gradients in spe- tance for a Neotropical bird, the Western ciesdiversity: A review of concepts.American Slaty Antshrike (Thamnophilusatrinucha): Naturalist 100:33-46. The problem, experiments,and simulations. PIANKA,E. g. 1981.Competition and nichetheory. Ph.D. dissertation,University of Pennsylvania, Pages167-196 in TheoreticalEcology (R. M. Philadelphia. May, Ed.). Sinauer Associates,Sunderland, ROSENBERG,K. V. 1997. Ecologyof dead-leaf for- Massachusetts. aging specialistsand their contribution to PIERPONT,N. 1986. Interspecific aggression Amazonian bird diversity. Ornithological and the ecology of woodcreepers (Aves: Monographs48:673-700. Dendrocolaptidae). Ph.D. dissertation, ROSENZWEIG,M. L. 1995. Species Diversity in PrincetonUniversity, Princeton, New Jersey. Spaceand Time. CambridgeUniversity Press, PULLIAM, H. R., AND B. J. DANIELSON. 1991. Cambridge,United Kingdom. Sources, sinks, and habitat selection: A land- SAS INSTITUTE.1997. SAS Software. SAS Institute, scape perspectiveon population dynamics. Cary,North Carolina. American Naturalist 137:S50-S66. SCHNEIRLA,t. C. 1971. Army Ants: A Study in RAIM, A. 1978. A radio transmitter attachment for Social Organization. W. H. Freeman, San small passerines.Bird-Banding 49:326-332. Francisco, Califomia. RAPPOLE,J. H., AND g. R. TIPTON.1991. New har- SCHOENER,T. W. 1968.Sizes of feedingterritories ness design for attachmentof radio trans- amongbirds. Ecology49:123-141. mitters to small passerines.Journal of Field SCHOENER,T. W. 1974.Resource partitioning in eco- Ornithology62:335-337. logicalcommunities. Science 185:27-39. RETTENMEYER,C. W. 1963. Behavioral studies of SEAMAN,D. E., J. J. MILLSPAUGH,B. J. KEERNOHAN, OBLIGATE ANT-FOLLOWING BIRDS IN PERU 65

G. C. BRUNDIGE,K. J. RAEDEKE,AND R. A. SUTHERLAND,W. J., AND G. A. PARKER.1985. GITZEN.1999. Effectsof sample size on ker- Distribution of unequal competitors.Pages nel home rangeestimates. Journal of Wildlife 255-274 in Behavioural Ecology: Ecological Management63:739-747. Consequencesof AdaptiveBehaviour (R. M. SEAMAN,D. E., AND R. A. POWELL.1996. An evalu- Siblyand R. H. Smith,Eds.). Blackwell Scientific ation of the accuracyof kernel density esti- Publications,Oxford, United Kingdom. mators for home range analysis.Ecology 77: SWARTZ,M. B. 2001. Bivouac checking, a novel 2075-2085. behavior distinguishingobligate from oppor- SHELDON,F. H., AND D. W. WINKLER. 1999. Nest tunisticspecies of army-ant-followingbirds. architectureand avian systematics.Auk 116: Condor 103:629-632. 8754377. TERBORGH,J. 1980. Vertical stratification of SHERRY,T. W. 1984.Comparative dietary ecology of a Neotropical forest bird community. sympatric,insectivorous Neotropical flycatch- Pages 1005-1012 in Acta XVII Congressus ers (Tyrannidae).Ecological Monographs 54: InternationalisOrnithologici (R. N6hring,Ed.). 313-338. DeutscheOrnithologen-Gesellschaft, Berlin. SIEVINg,K. E. 1992.Nest predationand differential TERBORGH,J. 1983. Five New World Primates: A insular extinctionamong selectedforest birds Study in Comparative Ecology. Princeton of centralPanama. Ecology 73:2310-2328. UniversityPress, Princeton, New Jersey. SILMAN, M. R., J. W. TEREORCH,AND R. A. KILTIE. TEREORC}•,J. 1990. An overview of research at 2003.Population regulation of a dominantrain CochaCashu BiologicalStation. Pages 48-59 foresttree by a major seedpredator. Ecology in Four NeotropicalForests (A. H. Gentry, 84:431-438. Ed.). Yale University Press, New Haven, SKUTCH,A. F. 1985.Clutch size, nesting success, and Connecticut. predationon nestsof Neotropicalbirds, revis- TERBORGH,J. W., J. W. FITZPATRICK,AND L. EMMONS. ited.Ornithological Monographs 36:575-594. 1984. Annotated checklist of bird and mammal SKUTCH, A. F. 1996. Antbirds and Ovenbirds. speciesof Cocha Cashu BiologicalStation, University of TexasPress, Austin. Manu National Park,Peru. Fieldiana (Zoology) SMYTHE,N. 1974. Biologicalmonitoring data-- 21:1-29. Insects.Pages 70-115 in 1973 Environmental TERBORGH,J., S. K. ROBINSON,T. A. PARKERIII, C. A. Monitoringand BaselineData (R. W. Rubinoff, MUNN, AND N. PIERPONT.1990. Structure and Ed.). Smithsonian Institution Environmental organizationof an Amazonianforest bird com- ScienceProgram, Washington, D.C. munity.Ecological Monographs 60:213-238. SNEDECOR,G. W., ANDW. G. COCHRAN.1989. Statistical THIOLLA¾,J.-M. 1992.Influence of selectivelogging Methods.Iowa StateUniversity Press, Ames. on bird speciesdiversity in a Guiananrain for- SPRUCE,R. 1908. Notes of a Botanist on the Amazon est.Conservation Biology 6:47-63. and Andes. MacMillan, London. THOMPSON,F. R., III 1994. Temporaland spatial STILES,F. G. 1992.Effects of a severedrought on the patternsof breedingBrown-headed Cowbirds populationbiology of a tropicalhummingbird. in the midwestern United States. Auk 111 Ecology73:1375-1390. 979-990. STOUFFER,P. C., AND R. O. BIERREGAARD,JR. 1993. THOMPSON, F. R., III, T. M. DONOVAN, R. M Spatial and temporal abundancepatterns of DEGRAAF,J. FAABORG,AND S. K. ROBINSON.2002 Ruddy Quail-Doves near Manaus, Brazil. A multi-scaleperspective of the effectsof for- Condor 95:896-903. est fragmentationon birds in easternforests STOUFFER,P. C., AND R. O. BIERREGAARD.1995. Use Studiesin Avian Biology25:8-19. of Amazonianforest fragments by understory VANDERMEER,J., I. G. DE LA CERDA,D. BOUCHER,I insectivorousbirds. Ecology76:2429-2445. PERFECTO,AND J. Ruiz. 2000. Hurricane distur- STOUFFER,P. C., AND R. O. BIERREGAARD.1996. banceand tropicaltree diversity.Science 290 Seasonal variation in hummingbird 788-791. abundance in a fragmented Amazonian VON HUMEOLDT, AND A. BONELAND. 1814--1829 landscape.Ararajuba (Journal of the Brazilian PersonalNarrative of Travelsto theEquinoctial OrnithologicalSociety) 4:9-14. Regionsof the New Continent, During the STUTCHBURY,B. J. M., AND E.S. MORTON. 2001. Years 1799-1804 [Title of first English edi- Behavioral Ecology of Tropical Birds. tion, translatedby H. M. Williams].Longman, Academic Press,New York. Hurst, Rees, Orme, and Brown, London. STYRSKY,J. N. 2003. Life-history evolution and WALLACE,A. R. 1889. A Narrative of Travels on populationdynamics of a Neotropicalforest the Amazon and Rio Negro. Ward, Lock and bird (Hylophylaxnaevioides). Ph.D. dissertation, Company,London. Universityof Illinois,Urbana-Champaign. WHITE, G. C. 1996. NOREMARK: Population 66 ORNITHOLOGICAL MONOGRAPHS NO. 55

estimation from mark-resighting surveys. two forms of White-chinnedWoodcreepers WildlifeSociety Bulletin 24:50-52. (Dendrocinclamerula, Dendrocolaptidae)in WHITE, G. C. 2002. Discussion comments on: The Amazonia. PapelsAvulsos de Zoologia, Silo use of auxiliaryvariables in capture-recapture Paulo 33:27-66. modelling.An overview.Journal of Applied WILLIS, E. O. 1979. Ecologia e comportamento Statistics 29:103-106. da mfie-de-ta6ca(Phlegopsis nigromaculata, WHITE,G. C., ANDK. P. BURNHAM.1999. Program Formicariidae).Revista Brasileira de Biologia MARK: Survival estimation from popula- 39:117-159. tions of marked animals. Bird Study 46 WILLIS, E. O. 1982a. Ground-cuckoos (Aves, (Supplement):120-138. Cuculidae) as army ant followers. Revista WIENS,J. A. 1977.On competitionand variableen- Brasileirade Biologia42:753-756. vironments. American Scientist 65:590-597. WILLIS, E. O. 1982b. The behavior of Black-headed WILKINSON, F. A., AND U. R. SMITH. 1997. The first Antbirds( ruJ•frons) (Formicariidae). nestrecords of the SootyAntbird (Myrmeciza RevistaBrasileira de Biologia42:233-247. fortis)with noteson eggsand nestlingdevelop- WILLIS, E. O. 1982c. The behavior of Scale-backed ment. Wilson Bulletin 109:319-324. Antbirds. Wilson Bulletin 94:447-462. WILLIAMS,E. C., JR.1941. An ecologicalstudy of the WILLIS, E. O. 1984. The behavior of Black-banded floor fauna of the Panama rain forest. Bulletin Woodcreepers (Dendrocolaptespicumnus). of the ChicagoAcademy of Science6:63-124. Condor 84:272-285. WILLIS, E. O. 1966a. Interspecific competition WILLIS,E. O. 1985.Myrmeciza and relatedantbirds and the foraging behavior of Plain-brown (Aves, Formicariidae)as army ant followers. Woodcreepers,Dendrocindafuliginosa. Ecology RevistaBrasileira de Zoologia2:433-442. 47:667-672. WILLIS, E. O., and Y. ONIKI. 1978. Birds and WILLIS,E. O. 1966b.The role of migrant birds at army ants. Annual Review of Ecology and swarmsof army ants.Living Bird 5:187-231. Systematics9:243-263. WILLIS, E. O. 1967. The Behavior of Bicolored WILLIS, E. O., AND Y. ONIKI. 1992. As aves e as for- Antbirds.University of CaliforniaPublications migas de correigfio.Boletin Museo Paraense in Zoology,no. 79. Emilio Goeldi 8:122-150. WILLIS, E. O. 1968. Studies of the behavior of WILLSON, S. K. 2000. First nest record of the Lunulated and Salvin's antbirds. Condor 70: White-throated Antbird (Gymnopithyssal- 128-148. vini) and detailednest recordsof the Hairy- WILLIS,E. O. 1969.On the behaviorof five species crested Antbird (Rhegmatorhinamelanosticta). of Rhegmatorhina,ant-following antbirds of the OrnitologiaNeotropical 11:353-357. Amazon basin. Wilson Bulletin 81:363-395. WOLDA, H. 1978. Seasonal fluctuations in rain- WILLIS, E. O. 1972. The behavior of Plain-brown fall, food, and abundanceof tropicalinsects. Woodcreepers,Dendrocincla fuliginosa. Wilson Journalof Animal Ecology47:369-381. Bulletin 84:377-420. WRIGHT,S. J., C. CARRASCO,O. CALDERON,AND S. WILLIS, E. O. 1973. The behavior of Ocellated PATON.1999. The E1 Nifio Southern Oscillation, Antbirds. Smithsonian Contributions to variable fruit production, and famine in a Zoology,no. 144. tropicalforest. Ecology 80:1632-1647. WILLIS, E. O. 1974. Populations and local ZIMMER,K. J., AND M. L. ISLER.2003. Family extinctions of birds on Barro Colorado Island, Thamnophilidae(Typical Antbirds). Pages 448- Panama.Ecological Monographs 44:153-169. 531 in Handbook of the Birds of the World, vol. WILLIS,E. O. 1976.Seasonal changes in the inver- 8. (J. del Hoyo, A. Elliott, and D. A. Christie, tebrate litter fauna on Barro Colorado Island, Eds.).Lynx Edicions, Barcelona, Spain. Panama. Revista Brasileira de Biologia 36: ZYSKOWSKI,K., ANDR. O. PRUM.1999. Phylogenetic 643-657. analysisof thenest architecture of Neotropical WILLIS, E. O. 1978. Behavior and ecology of ovenbirds(Furnariidae). Auk 116:891-911. APPENDIX

67 BIRDERS' EXCHANGE THE BIRDS OF NORTH AMERICA LIFE HISTORIES FOR THE 21 ST CENTURY A. POOLE AND F. GILL, Editors

Individual profiles now available!

In 1992the AmericanOrnithologists' Union, in parmershipwith the Academyof NaturalSciences of Philadelphia,undertook the publication of speciesprofiles for eachof the morethan 700 specieswhich breed in the United Statesand Canada. These illustrated reviews provide the most comprehensivesummaries of the currentknowledge of eachspecies, with rangemaps and an extensivelist of references.

ButeoBooks is pleasedto offerindividual species accounts for $7.50 each. All 716 accounts are listed on our website in taxonomic order (rearrangedto conformwith the SeventhEdition of theA.O.U. Check-list). Singlesmay be orderedby mail, phone,fax, or e-mail. Shippingand handlingis $4 for thefirst prof'de and $1 for eachadditional profile to a maximumcharge of $10per order. •

Profilesrange from 12 to 48 pagesin length, and measure 8 1/2" x 11". The information containedin eachprofile includes breeding, nesting,habitat, food and feeding, range, sound,and conservation, plus much more. They providethe mostcomprehensive, up-to-datedata for eachspecies covered. Theseaccounts are an indispensablepart of yourreference library. Contactus to orderyours today.

Visitour websitefor more information: www.buteobooks. com 3130 LaurelRoad; Shipman,VA 22971; USA

allen•buteobooks.com orders:800-722-2460 phone:434-263-4842 fax:434-263-4842 ZOOLOGY & ORNITHOLOGY

SPECIMEN CABINETS

PERMANENT For over forty yea name museums, most to protect tt

To learn more or any of our www. lanescience. ORNITHOLOGICAL MONOGRAPHS (Continued from back cover)

No. 38. An Analysisof Physical,Physiological, and OpticalAspects of Avian Colorationwith Emphasison Wood-Warblers. E. H. Burtt,Jr. x + 122pp. 1986.$15.00 ($12.50). No. 39. TheLingual Apparatus of theAfrican Grey Parrot, Psittacus erithacus Linne (Aves: Psittacidae):Description and TheoreticalMechanical Analysis. D.G. Homberger.xii + 236pp. 1986.$30.00. No. 40. Patternsand Evolutionary Significance of GeographicVariation in theSchistacea Group of theFox Sparrow (Passerella iliaca). R. M. Zink. viii + 119pp. 1986.$15.00. No. 41. Hindlimb Myology and Evolutionof Old World SuboscinePasserine Birds (Acanthisittidae,Pittidae, Philepittidae, Eurylaimidae). R. J. Raikow. viii + 81 pp. 1987. $15.00. No. 42. Speciationand Geographic Variation in Black-tailedGnatcatchers. J. L. Atwood.vii + 74 pp. 1988.$10.00. No. 43. A DistributionalSurvey of theBirds of theMexican State of Oaxaca.L. C. Binford.viii + 418 pp. 1989.$20.00. No. 44. RecentAdvances in theStudy of NeogeneFossil Birds: I. TheBirds of theLate Miocene- EarlyPliocene Big Sandy Formation, Mohave County, Arizona (K. J. Bichart);II. Fossil Birdsof theSan Diego Formation, Late Pliocene, Blancan, San Diego County, California (R. M. Chandler).vi + 161pp. 1990.$20.00. Nos. 45 & 46. Descriptionsof Thirty-twoNew Speciesof Birdsfrom theHawaiian Islands: Part I. Non-Passeriformes(S.L. Olsonand H. F. James),88 pp.; PartII. Passeriformes(H. F. Jamesand S. L. Olson),88 pp. 1991.Bound together (not availableseparately). $25.00($22.50). No. 47. Parent-OffspringConflict and Its Resolutionin theEuropean Starling. E. Litovichand H. W. Power. 71 pp. 1992.$15.00 ($12.00). No. 48. Studiesin NeotropicalOrnithology Honoring Ted Parker. J. v. RemsenJr., Ed. xiv + 918pp. 1997.$49.95 ($39.95). No. 49. Avian ReproductiveTactics: Female and Male Perspectives.P. G. Parker and N. T. Burley,Eds. v + 195pp. 1998.$20.00 ($16.00). No. 50. AvianCommunity, Climate, and Sea-Level Changes in thePlio-Pleistocene of the Florida Peninsula.S. D. Emslie.iii + 113pp. 1998.$20.00 ($16.00). No. 51. A Descriptiveand PhylogeneticAnalysis of PlumulaceousFeather Characters in Charadriiformes.C. J. Dove. iii + 163pp. 2000.$19.95 ($15.96). No. 52. Ornithologyof Sabah:History, Gazetteer, Annotated Checklist, and Bibliography. F. H. Sheldon,R. G. Moyle,and J.Kennard. vi + 285pp. 2001.$25.00 ($22.50). No. 53. Evolution of Flightlessnessin Rails (Gruiformes:Rallidae): Phylogenetic, Ecomorphological,and OntogeneticPerspectives. B.C. Livezey. x + 654 pp. 2003. $10.00 ($9.00). No. 54. PopulationDynamics of theCalifornia Spotted Owl (Strixoccidentalis occidentalis): A Meta-Analysis.Alan B. Franklin,R. J. Guti6rrez,James D. Nichols,Mark E. Seamans,Gary C. White,Guthrie S. Zimmerman,James E. Hines,Thomas E. Munton,William S. LaHaye,Jennifer A. Blakesley,George N. Steger,Barry R. Noon,Daniel W. H. Shaw,John J. Keane, Trent L. McDonald,and SusanBritting. viii + 54 pp. 2004.$10.00 ($9.00).

Order from: Buteo Books, 3130 Laurel Road, Shipman, VA 22971, 1-800-722-2460;E-mail [email protected];or www.buteobooks.com. Prices in parenthesesare for AOU members.

For a completelist of OrnithologicalMonographs including both in-print and out-of-print books, please visitthe AmericanOrnithologists' Union website at www.aou.org. Supplementto The Auk 121:2

ORNITHOLOGICAL MONOGRAPHS

No. 3. TheBirds of Kentucky. R. M. Mengel. 1965.$25.00. No. 6. Adaptationsfor Locomotionand Feedingin the Anhinga and the Double-crested Cormorant. O. T. Owre. 1967. $10.00. No. 7. A DistributionalSurvey of theBirds of Honduras. B. L. Monroe)Jr. 1968.$25.00. No. 10. TheBehavior of SpottedAntbirds. E. O. Willis. 1972.$10.00. No. 11. Behavior,Mimetic Songsand Song Dialects,and Relationshipsof the Parasitic Indigobirds(Vidua) of Africa. R. B. Payne.1973. $10.00. No. 12. Intra-islandVariation in theMascarene White-eye Zosterops borbonica. F. B. Gill. 1973. $10.00. No. 13. EvolutionaryTrends in theNeotropical Ovenbirds and Woodhewers. A. Feduccia.1973. $10.00. No. 14. A Symposiumon theHouse Sparrow (Passer domesticus) and European Tree Sparrow (P. montanus)in NorthAmerica. S.C. Kendeigh,Ed. 1973.$10.00. No. 15. FunctionalAnatomy and Adaptive Evolution of theFeeding Apparatus in theHawaiian HoneycreeperGenus Loxops (Drepanididae). L. P. Richards.and W. J. Bock.1973. $1o.oo. No. 16. TheRed-tailed Tropicbird on Kure Atoll. R. R. Fleet.1974. $6.00. No. 17. ComparativeBehavior of the American Avocet and the Black-neckedStilt (Recurvirostridae). R. B. Hamilton. 1975. $10.00. No. 18. BreedingBiology and Behavior of theOldsquaw (Clangula hyemalis L.). R. M. Alison. 1975. $6.00. No. 19. Bird Populationsof AspenForests in WesternNorth America.J. A.D. Flack. 1976. $10.00. No. 21. Social Organizationand Behaviorof the Acorn Woodpeckerin Central Coastal California.M. H. MacRobertsand B. R. MacRoberts.1976. $10.00. No. 22. Maintenance Behavior and Communication in the Brown Pelican. R. W. Schreiber. 1977. $6.00. No. 23. SpeciesRelationships in the Avian Genus Aimophila. L. L. Wolf. 1977.$12.00. No. 24. LandBird Communitiesof GrandBahama Island: The Structure and Dynamicsof an Avifauna.J. T. Emlen.1977. $10.00. No. 25. Systematicsof SmallerAsian Night BirdsBased on Voice. J. T. Marshall.1978. $10.00. No. 26. Ecologyand Behaviorof the PrairieWarbler, Dendroica discolor. V. Nolan, Jr. 1978. $45.00. No. 27. Ecologyand Evolution of Lek Mating Behaviorin theLong-tailed Hermit Hummingbird. F. G. Stiles and L. L. Wolf. 1979. $10.00. No. 28. The ForagingBehavior of Mountain Bluebirdswith Emphasison SexualForaging Differences.H. W. Power.1980. $10.00. No. 29. The Molt of ScrubJays and Blue Jays in Florida. G. T. Bancroft and G. E. Woolfenden. 1982. $10.00. No. 30. AvianIncubation: Egg Temperature, Nest Humidity, and Behavioral Thermoregulation in a Hot Environment. G. S. Grant. 1982. $10.00. No. 31. TheNative Forest Birds of Guam.J. M. Jenkins.1983. $15.00. No. 32. TheMarine Ecology of Birds in theRoss Sea, Antarctica. D. G. Ainley,E. F. O'Connor and R. F. Boekelheide.x + 97 pp. 1984.$15.00. No. 33. SexualSelection, Lek and Arena Behavior, and Sexual Size Dimorphism in Birds.R. B. Payne.viii + 52 pp. 1984.$15.00. No. 34. Pattern,Mechanism, and Adaptive Significance of Territorialityin HerringGulls (Larus argentatus).J. Burger.xii + 92 pp. 1984.$12.50.

(Continued on inside back cover)