THE CONDORDEC-61 A JOURNAL OF AVIAN BIOLOGY LIBRARY

Volume 95 Number 4 November 1993

The Condor 95:769-78 I 0 The Cooper Ornithological Society 1993

VOCAL REPERTOIRE OF THE LONG-TAILED AND ITS RELATION TO MALE-MALE COOPERATION’

JILL M. TRAINER Department of Biology, Universityof Northern Iowa, Cedar Falls, IA 50614

DAVID B. MCDONALD ArchboldBiological Station, P.O. Box 2057, Lake Placid, FL 33852

Abstract. We examined the vocal repertoire of lek-mating Long-tailed (Chi- roxiphia linearis, Pipridae) in Monteverde, Costa Rica. Males in this genusare unusualin performing a cooperativecourtship display, including duet songsand coordinateddual-male dance displays. Males give at least 13 distinct vocalizations, several of which occur in clear behavioral contexts. By observing the behavioral context and the sequencein which calls were given, we found that the most frequent calls occurred during three types of activity: song bouts, dance, and noncourtship interactions. The responsesof males to playback of six vocalizations indicated that the calls function as much in mediating cooperative inter- actions as in expressingmale-male agonism. The evolution of the large vocal repertoire in Long-tailed Manakins may be associatedwith their unique social system based on long- term, cooperative relationshipsamong males. Key words: Vocalization;call function; Long-tailed Manakin; linearis; so- ciality; cooperation,lek.

Resumen. Estudiamosel repertorio vocal de1Saltarin Toledo (Chiroxiphia linearis, Pi- pridae) en Monteverde, Costa Rica. Los machos de este genera se comportan muy parti- cularesen relacibn el cortejo cooperative, incluyendo cancionesa duo y danzascoordinadas de parejasde machos. Los machos emiten al menos 13 vocalizacionesdistintas, muchasde ellas con un context0 claro con respeto al comportamiento. Se encontrb, mediante obser- vacion de1 context0 de comportamiento y las secuenciasde sus vocalizaciones, que las llamadas m&s frecuentesse emiten durante tres tipos de actividades:el canto, el baile, y la interaction no-relacionada al cortejo. Las respuestasde 10s machos a seis vocalizaciones pre-grabadassugieren una funcion de interaction cooperativa tanto coma para expresar agresionentre machos. La evolution de las vocalizacionesdiversas en Chiroxiphia linearis puede ser asociadacon el sistema social unico fundado en relaciones cooperativasa largo plazo entre machos.

INTRODUCTION most in that males provide no re- Although much information exists on the use sourcesvaluable to breeding females other than and function of vocalizations in territorial song- gametes (Bradbury 1981). In addition, lek spe- , little such information exists about pas- cies lack persistent pairbonds and do not defend serines with other social systems. Here we de- exclusive, resource-basedterritories. Long-tailed scribethe diverse vocal repertoire of a lek-mating Manakins are especially unusual in that groups , the Long-tailed Manakin (Chiroxiphia of males form complex networks of cooperative lineuris). Lek social systemsdiffer from those of alliances (McDonald 1989a). This cooperation involves a joint courtship display that consists of highly coordinated singing and dancing per- I Received 11 March 1993. Accepted 8 July 1993. formed by two males (Snow 1977, Foster 1981,

[7591‘ 170 JILL M. TRAINER AND DAVID B. MCDONALD

McDonald 1989b). Our results suggestthat the playing males tolerate the presence of other af- functions of Long-tailed Manakin calls reflect the filiates, and overt aggressionamong males is rare. importance of cooperative as well as aggressive In order to attract females to the perch-zone, interactions among males, in contrast to what is two male partners cooperate in performing the observed in territorial species. coordinated song and dance displays. The court- Chiroxiphia manakins are unusual among their ship display usually is performed by the alpha suboscinerelatives in possessinga large and var- and beta, although occasionally other combina- ied repertoire of vocalizations. Several speciesof tions of affiliates may display (McDonald 1989b). manakins have mechanical pops and snaps in Males begin by broadcastinglong bouts of unison the acoustic repertoires, but generally, manakin duet song, or toledos, from the lower branches vocalizations are few in number and simple in of a tree crown (subcanopy). If a female arrives, structure (Prum 1990b). The Long-tailed Mana- both males and the female descendto the dance kin’s repertoire of 13 distinct calls is rich, even perch, a horizontal branch or vine within 1.5 m when compared to the call repertoires of song- of the ground. The males begin a series of 20- birds (exclusive of primary songvariants). A ma- 100 backward leapfrog hops alternating with as jor factor contributing to the evolution of such much as 60 set of a special butterfyjlight char- a diverse repertoire may be the complex network acterized by deep wing beats. A dance bout may of long-term relationships among males. include as many as 10 setsof hops and butterfly Here we describe the functions of Long-tailed flight. Dances leading to copulation include a Manakin vocalizations, and examine the com- solo butterflyflight by the alpha male. At times, plexity of the vocal repertoire in the context of two or more males, often including predefinitive and the social structure. We em- males, gather on a dance perch in the absenceof ployed three approaches in our study. (1) We a female and perform a variation of the hops observed the social and behavioral context of (popcorns),in which males jump straight up and each call. (2) We analyzed the sequencein which do not leapfrog over the partner(s). calls were given. These approachesallowed us to identify three general contexts in which the ma- METHODS jority of calls occurred. (3) We recorded the re- The study area, in Monteverde, Costa Rica sponsesof manakins to experimental playback (1 0”18’N, 84”48’W), is 80 ha of premontane trop- of taped calls in the field. ical moist forest (Holdridge 1966) at an elevation of 1,300 m. McDonald uniquely color-banded NATURAL HISTORY AND DISPLAY 270 manakins between 198 1 and 1987. The sex A Long-tailed Manakin lek consists of a core of individuals with green was deter- partnership of an alpha and a beta male and as mined by subsequent observations of behavior. many as 11 male affiliates (McDonald 1989b). Vocalizations were recorded with a Sony WM These males regularly attend a traditional display D6C cassetterecorder and a Sennheiser ME 80 area called a perch-zone or lek arena, which con- directional microphone. In addition, behavior tains one or two major perches for the dance sequenceswere recordedin the field using a Tandy display. Perch-zonesare separatedby 75-300 m, 102 personal computer programmed as an event and each is occupied by a different alpha male. recorder. Sonagramswere made using a Kay Ele- From a particular perch-zone, one can some- metrics Model 5500 Sonagraph and grey scale times hear the singingof manakins in other zones, printer. A synoptic tape of the Long-tailed Man- but cannot see them. akin vocal repertoire is on file at the Library of Plumage maturation occurs over a four-year Natural Sounds at the Cornell Laboratory of Or- period with distinct, age-specific, predefinitive nithology (Accession #49097). stages(McDonald 1989b, 1993a). Affiliates of a The terms “call” and “vocalization” are used perch-zone may include both definitive and pre- interchangeably to denote an individual utter- definitive males. Affiliations among males de- ance of sound. “Song” refers only to the toledo velop over their lifetimes of at least 13 years call, which is given in unison as a duet by singing (McDonald 1993b) and result in the establish- partners. “Call type” refers to one of the 13 ment of a dominance hierarchy. The alpha male acoustically distinguishable kinds of manakin is highest ranking, and with very rare exceptions calls. The term “dance bout” refers to a display only he mates with female visitors. Pairs of dis- sequenceincluding one or more of the following VOCAL REPERTOIRE OF LONG-TAILED MANAKIN 111 elements: a variable number of sets of leapfrog Chatfield 1971). Therefore we excluded transi- dance hops, dual-male butterflydisplay, and solo tions between identical call types when calculat- display by the alpha male. A “dance bout” is, ing the expected values. Expected values were therefore, essentially synonymous with a single calculated by iteration (Goodman 1968, Trainer continuous visit to the dance perch by one or 1988) becausesimply using row and column to- more females, because males almost invariably tals is not valid for an incomplete matrix. display throughout female visits. Comparing the observed and expected fre- quenciesofeach call type transition revealedthose

BEHAVIORAL CONTEXT pairs of call types that tended to occur or not to occur together in sequence.To obtain a consis- Ad libitum observations of vocal behavior and tent criterion for identifying major departures responses of male and female manakins were from expected values, we collapsed the contin- made at perch-zones.In addition, observerscon- gency table around each matrix cell, combining ducted scheduled two-hour observations from frequenciesfrom other call types to form 42 two- blinds 8-12 m from display perches between by-two matrices (Slater 1973). We tested the re- 06:00-l 5:O0. During the scheduled observation sulting chi-square values at the 0.05 level. The periods, an observer recorded the number of to- chi-square analyses of the two-by-two matrices ledo and teeamoo calls occurring in each 5-min were not intended as significance tests; rather, block, and the number of female and male man- they indicate those transitions that contribute akins present in the perch-zone during the first most to the nonrandom sequence of call types min of each 5-min block. The details of this be- overall. havioral sampling protocol were described in McDonald (1989a). PLAYBACK EXPERIMENT The vocal responsesto each of five manakin call SEQUENCE ANALYSIS types were tested by broadcasting recorded calls Thirty real-time samples of vocalizations were through an Aiwa SC-A2 speaker. Each test con- recorded on cassettetapes. The samples were of sistedof broadcastingone type of call and writing different durations and togethertotaled 285 min. down all vocalizations heard during a playback The transitions between calls were transcribed in period. The frequencies of occurrence of re- the sequence heard, regardlessof the identities sponse vocalizations were compared to their of the callers. Thus, transitions represented the baseline frequenciesduring a preplayback period call sequenceof all males together rather than of in which nothing was broadcast. All calls except individuals. Two vocalizations separatedby more toledoswere broadcast from the ground near the than 30 set were not scored as a transition. dance perch. For these calls, nyanyownh, teea- We tested whether the occurrence of a given moo, weet, and wheeoo,a five-minute preplay- call type dependedon the immediately preceding back period was followed immediately by 5 min type using chi-square contingency analysis as de- ofplayback at the rate of 1 call/min. Toledosongs scribedin Trainer (1988). Transitions among the were broadcast from a speaker suspendedfrom seven most frequent call types were analyzed, a tree 6 m above the ground at a rate of 24 songs/ excluding thosecalls normally given on the dance min. For the toledos, the duration of the pre- perch. Four infrequent calls (doodoodoo,toodle- playback and playback periodswere each 10 min. loo, chitter, and squawk) were excluded because At a given perch, we tested three different call sample sizes were insufficient for chi-square types on the same day with 10 min between ex- analysis. periments. Each type of call was broadcast at 10 Because we were interested in how different different perches. For each of the six most fre- call types were arranged in sequence,our analysis quent vocal responses,the numbers of calls dur- concerned only those transitions between unlike ing the preplayback and playback periods were call types. Like many birds, Long-tailed Mana- compared using one-tailed Wilcoxon matched- kins tend to repeat most calls severaltimes before pairs signed-rank tests. Becausethis resulted in switching to a different call type. Such call rep- 26 simultaneous tests,we used a sequential Bon- etitions are extremely numerous and eclipse the ferroni technique (Rice 1989) to adjust the sig- importance of transitions between unlike call nificance level. Individual comparisons were types in a contingency analysis (Lemon and considered significantly different if P I 0.002, 712 JILL M. TRAINER AND DAVID B. MCDONALD

TABLE 1. Descriptionsand usesof Long-tailedManakin call types.The quantitiesin parenthesesare the numberof occurrencesof the call in the recordedsamples.

Call Ampli- Frequencyof Call recipient tude occurrence Function of call During-song-bout toledo female loud abundant(1119) Attract females wit partner,female soft abundant(75 1) Synchronizesinging of partners,pre- cedessocial interaction with affiliate or female owng partner,female soft infrequent(42) Stimulatefemale to move to dance perch Dance nyanyownh female loud abundant Accompanycourtship dance display buzz-weent subordinatepartner loud infrequent Signaldominance Noncourtship teeamoo partner loud intermittent(294) Attract partner weet male loud intermittent(143) Mild distress Other waanh male soft intermittent(4 18) Closeproximity contact call doodoodoo male soft infrequent(13) Uncertain toodleloo male soft infrequent(12) Uncertain wheeoo strangemale, predator loud intermittent(114) Mobbing call chitter dominant male soft infrequent(4) Signalsubmissiveness squawk predator loud infrequent Distressin mist net

yielding a maximum, simultaneous significance ly persistent singers; over entire seasons,mean level of 5%. toledo rates by the most active teams exceeded 300/hr (McDonald 1989a). During a single 2 hr RESULTS scheduledobservation period one pair delivered BEHAVIORAL CONTEXT 1,9 19 songs. On a calm day toledos could be Most of the call types (Table 1) were given ex- heard from a distance of up to 250 m. clusively by male manakins. Rarely, banded fe- Rarely, we heard a toledosung by a single male, males were observed to give wheeoocalls as de- and these usually occurred during a bout of nor- scribed below. Both sexes gave the mist net mal duet songs.No bout of more than eight con- distresscall, squawk,although femaleswere more secutive solo toledoswas noted during more than likely to vocalize than males. The names of the 3,000 hours of scheduled observation. Also in- calls were intended to be onomatopoeic. frequent were toledosperformed by three males Toledo-a loud song, almost always sung as a togetherand McDonald observed only two cases duet by two males perchedapproximately 10 cm of four males calling togetherin synchrony. Man- apart on a branch (Fig. lb). Usually the alpha akins sang toledos from the forest subcanopy, and beta males sang together as partners, but usually from the lower branches of a tree crown, sometimes other combinations of males affili- 5-15 m from the ground. The long, loud bouts ated with a perch-zone performed the duet. The of song appeared to attract females to the perch- song contribution of each male consisted of a zone (McDonald 1989a). When females ap- continuous loud, pure tone of alternately falling proachedthe dance perch,males switchedto dual- and rising pitch approximately 0.60 set in du- male dance displays on the perch. ration. The two males’ songs overlapped with An interesting feature of courtship display in one male beginning to sing approximately 0.10 this speciesis the lack of any diurnal pattern of set after his partner. For a sample of 1,583 bouts activity in singing or dancing (McDonald 1989a). of toledosby one set ofpartners the mean number Many other species of lekking birds, including of toledo songsper bout was 48.6 delivered at a other species of manakins are noted for pro- rate of 15.6 songs/min. Manakins were extreme- nounced morning and afternoon peaksin activity VOCAL REPERTOIRE OF LONG-TAILED MANAIUN 773

4-

C

2 -

0 : I I 1 1 2 3

Time (seconds)

FIGURE 1. During-song-bout Vocalizations of Long-tailed Manakins. (a) wits, (b) toledo,(c) owng.

(e.g., Snow 1962, Lill 1976, Bradbury et al. 1989, males on the dance perch (Fig. 2a). The nyuny- Atwood et al. 199 1). ownh accompanied each hop in the leapfrog por- Wit-a brief, soft note, 0.10 set in duration tion of the dance display. A sample of 95 dance (Fig. la). Partners alternated giving wits in a rap- bouts that progressedfrom the dual to the solo id seriesof 1O-25 calls. Wits were usually deliv- stage contained a mean of 3.9 sets of leapfrog ered by two partners perched close together in hops (range = l-l 1, SD = 2.6) interspersedwith the subcanopy immediately before beginning a dual-male butterfly displays. The mean number bout of toledos.In addition, slower seriesof wits of leapfrogsper set was 34.8 (range = 2-l 39, SD were associated with the owng call, and some- = 12.0). The silent, dual-male butterfly displays times preceded dual-male display for a female, lasted a mean of 26.3 set (range = 2-75, SD = or occurred when a female was near the dance 12.2) during which males flew in radial flights perch and only one high-ranking male was pres- within 5-20 m from the focusat the dance perch. ent at the perch-zone, as describedfurther below Buzz-weent-a loud call associated with the for the owng call. leapfrog portion of the dance display (Fig. 2b). Owng-a brief, soft vocalization with an in- After one or more series of leapfrog hops, the terrogative sounding quality given intermittently dominant male delivered buzz-weent from be- from the subcanopy(Fig. lc). Often singing males tween the female and his partner, along the axis interjected owng calls during bouts of toledos. of the dance perch. He directed the call toward Sometimes singing would become erratic, with the partner, while facing away from the female. frequent interruptions containing long bouts of The male giving the call then left the perch while wits and owngs. This erratic singing may have his partner remained on the perch with the fe- been associatedwith the arrival of a female that male. Directed buzz- weentsalso occurredduring had not yet descendedto the dance perch. Such dances in which several males but no females females would not necessarily be visible to ob- were present. Buzz-weents with no other males servers sitting in a blind, but observers some- present were given by males of all ages. times recordedthe appearanceof a female on the In 20 of 129 dance bouts with suitable data dance perch after erratic song bouts. In addition between 1984 and 1988, the subordinate male to this context, a lone male would give bouts of left the perch area after buzz- weentsthat followed wits and owngswhen a female was near the dance the final or final two sets of leapfrog hops. The perch, but no partner was present to perform the dominant male then began solo butterfly flight. dual-male display. Such bouts of wits and owngs In 33 of the 129 dance bouts, however, three or were often followed by solo butterfly display by more buzz- weentsoccurred without subsequent the calling male. departure of the beta male. Furthermore, in 76 Nyunyownh-a loud, nasal call given by both of the dance bouts, the lower-ranking male JILL M. TRAINER AND DAVID B. MCDONALD

8

a b

6 I

4-

2-

1 2 3 buzz weent

0, I I 1 7 1 2 3

Time (seconds) FIGURE 2. DanceVocalizations. (a) threenyanyownh calls, (b) buzz-weent.

droppedout of the dual-male portion of the dance ly 20 set, usually by high-ranking males when display without any buzz- weentcalls being given. no other high-ranking male was in the immediate The buzz-weent call, therefore, was neither nec- vicinity. Of 182 teeamoosfor which the identity essary nor sufficient to produce the transition of the caller was known, 139 (76%) were by the from dual-male to solo male dance display. alpha male, while the beta was absent. Often Buzz- weentswere given only when the dancers following a teeamoocall, a partner would arrive, had disparate dominance ranks; males directed perch next to the caller, exchange wit calls and the call toward other males to whom they were begin to duet with him. Data from the scheduled clearly dominant, and definitive males directed observation periods provided evidence that teea- calls toward predefinitive males. This call was moos were used to attract a partner to sing a not heard between males of nearly equal or toledoduet. Before toledobouts, males tended to equivocal dominance status.For 179 buzz-weents give teeamooswhen they were alone in the perch- where both the caller and the recipient could be zone, and tended not to do so when their partners identified, in every case, the male who gave the were already present (G = 28.52; df = 1; P = buzz-weent was known, by other criteria, to be 0.001) (Table 2). Often a dance partner would dominant. In 160 of the casesthe male giving teeumoo while the alpha male performed solo the call was an alpha male, and in 21 casesthe butterflyflight following a dual-male dance dis- call was given by a definitive male toward a pre- play for a female. definitive male. Weet-a brief, loud, sharp call given inter- Teeamoo-a loud call, approximately 1.O set mittently from the subcanopy (Fig. 2a). Weet in duration, consistingof two pure tones, the first calls were associated with situations involving descending in pitch to approximately the fre- mild distress.For example, malesgave weetswhen quency of the second (Fig. 3b). Teeamoos were a partner failed to respond to repeated teeamoo given intermittently at intervals of approximate- calls. Weets were sometimes given when a hu- VOCAL REPERTOIRE OF LONG-TAILED MANAIUN 17.5

a II Jf, .‘ I-- .0 ’ Gl ’

Time (seconds)

FIGURE 3. Noncourtship Vocalizations of Long-tailed Manakins. (a) three weet calls, (b) two teeumoocalls.

man observer disturbed a male manakin in the variety of calls consisting of soft tones of alter- perch-zone. nately falling and rising pitch, given infrequently Wuanh-a soft, nasal, descending call given in the subcanopy (e.g., the toodlelooof Fig. 4e). intermittently from the subcanopy (Fig. 4a). Like doodoodoo,these calls were usually given Waanh was used in a variety of situations, in- by males who were not engaged in singing or cluding during song bouts and during noncourt- dancing, but a more specific context could not ship interactions. Often waanh wasgiven by males be ascertained. Many of the calls markedly re- when they were neither singing nor dancing. For sembled portions of the toledo or teeamoo calls example, this call was given by assemblagesof but were never given in unison and were always predefinitive males who were not displaying, or of considerably lower amplitude. by a third definitive male in the presenceof two displaying males. TABLE 2. Number of song bouts preceded or not Doodoodoo-a call consisting of three or four precededby teeamoo callswhen one or two males were present before the song bout. soft tones of the same pitch, given in the sub- canopy (Fig. 4d). It was usually given by a male Song bouts Song bouts who was not engagedin singing or dancing. Be- preceded by not preceded teexnow by teeamoo cause this call was given infrequently, a more specific context was not detected. Human imi- One male presentb 41 9 tation of the call often served to elicit close but Two males presentb 15 35 silent approach of males. dAt least one teeamoo call occurred in the five-minute sample block priceding the start of the song bout. Toodleloo,tuhweeko, federico and variants-a h During the sample minute preceding the start of the song bout. 176 JILL M. TRAINER AND DAVID B. MCDONALD

I I I 1 2 3

Time (seconds)

FIGURE 4. Other Vocalizations. (a)~ I waanh. Cbj\ I chitter, (c) sevenwheeoo calls, (d) one doodoodoocall, (e) one toodleloocall.

Wheeoo-a brief, moderately loud call (Fig. model snake, but not of the manakin mounts, 4c) given during mobbing choruses directed at the other speciesremained in the vicinity of the potential predatorssuch as the Mottled Owl (Cic- threatening stimulus and joined the chorus. caba virgata) and experimentally presentedsnakes Wheeoo choruses almost invariably involved made of modeling clay (McDonald 1993a). three or more male Long-tailed Manakins, and Wheeoo choruses also occurred during experi- resulted in calling rates of as many as 100 wheeoo mental presentation of taxidermically mounted calls per minute. specimens of male manakins. Such chorusesof- Chitter-a soft trill given infrequently in the ten persistedfor several minutes and attracted a subcanopy by a lower-ranking male to a domi- variety of other species including Stripe-tailed nant male (Fig. 4b). Almost invariably, the call Hummingbirds (Eupherusa eximia), Golden- was given by a predefinitive male as a response crowned Warblers (Basileuterus culicivorus), and to chasesor other agonistic behaviors directed Orange-billed Nightingale-Thrushes (Catharus at them by a definitive male. aurantiirostris). During presentations of the Squawk- a squawking or screechingcall given VOCAL REPERTOIRE OF LONG-TAILED MANAKIN 171

TABLE 3. Call transition matrix: frequency of occurrenceand expected values (in parentheses).

Preceding Followmg call type call type Toledo Wit OtVflg T&UI?lOO W&?l Waanh Wheeoo

Toledo 26 (28.7) 15 (5.4)* 4 (28.3)* 12 (14.5) 72 (53.7)* 8 (7.1) wit 64 (43.5)* 25 (7.3)* 29 (40.6)* 30 (20.8)* 40 (77.1)* 11 (9.5) Owng 6 (7.5) 26 (7.1)* 2 (7.0)* 1 (3.6) 5 (13.3)* 0 (1.7) Teeamoo 5 (41.6)* 33 (39.5) 0 (7.2)* - 26 (20.0) 117 (73.8)* 10 (9.3) Weet 9 (21.8)* 28 (20.7) 0 (3.7)* 45 (20.4)* 23 (38.7)* 5 (4.8) Waanh 129 (94.7)* 65 (89.8)* 2 (15.7)* 108 (88.5)* 34 (45.4)* 17 (20.4) Wheeoo 6 (9.8) 17 (9.3)* 0 (1.7) 6 (9.2) 6 (4.7) 17 (17.4) -

*Transitions for which the departurein the observedfrom the expectedfrequencies exceeded a thresholdcriterion (chi-square, P 5 0.05, df = 1). by individuals handled in the mist nets. Females DISCUSSION tended to give the call more frequently than did Analysis ofbehavioral context and call sequences males. of the most frequent calls revealed that call types tended to occur in one of three general contexts: SEQUENCEANALYSIS during song bouts, during dancing, and during Table 3 showsthe call transitions for which the noncourtship interactions. Two vocalizations observed frequencies showed a major departure (waanh and wheeoo),occurred during both song from the expected frequencies. Transitions be- bouts and noncourtship interactions. One other tween positively and negatively associated call (weet) showed only a weak tendency to occur in types are summarized in Table 4. This analysis noncourtship situations. Detailed observations revealed that calls other than dance calls oc- made during scheduled sample periods allowed curred in two general contexts: during bouts of us to make inferencesabout the functions of sev- toledosong, and during interactions among males eral of the calls. that did not involve courtship of a female. Dur- ing-song-bout calls were positively associated DURING-SONG-BOUT VOCALIZATIONS with toledo songsand with one another. These Toledo-The long, continuous bouts of toledos, call types include toledo, wit, and owng. Non- and their occurrence in the absence of females courtship calls, teeamoo and weet, were posi- clearly indicates that the duet song functions as tively associatedwith one another, and usually advertisement to attract females to the dance negatively associatedwith during-song-bout calls perch. McDonald (1989b) showed that success (except weet tended to follow wit). Waanh oc- in attracting females is correlated with singing curred in both contexts and was positively as- sociated with toledo and teeamoo. Wheeoo was TABLE 4. Call transitionsthat occurredmore often associatedonly with wit. (positive call associations)or less often (negative call associations)than expectedby chance. The double ar- PLAYBACK EXPERIMENT row indicates significanceof both transitions in which a call type either precedes or follows the other type. Manakins did not respond to playback of the five Calls positively associatedwith toledo songsappear in call types by singing toledo bouts; song occurred bold face, calls negatively associatedwith toledo songs in only 19 out of 50 playback periods. For three appear in italics, and calls not associatedwith toledo songsappear in normal type. of the call types, the number of non-zero obser- vations of toledoswas insufficient to conduct a Positive call associations Negatwe call associatmns significance test (Table 5). Wit increased signif- icantly during playback of nyanyownh, teeamoo, toledo+ owng toledo- teeamoo and week Wheeoo increased during playback of wit + toledo wit - teeamoo wit - owng owng++ teeamoo nyanyownh, and wheeoo. Teeamoo and weet in- toledo++ waanh weet- toledo creasedduring playback of several calls, although teeamooH waanh weet- owng these increaseswere not significant. Wheeoowas weet+ teeamoo weet- waanh the only playback call to stimulate like calls in wit + weet wit - waanh wheeoo- wit owng++ waanh response. 178 JILL M. TRAINER AND DAVID B. MCDONALD

TABLE 5. Mean number of callsheard during the preplaybackand playbackperiods.

Vocal response Toledo Wit Teeamoo Weet Waanh Wheeoo VOCd stimulus PR Play PR Play PR Play PR Play Pre Play PR Play Toledo 13.9 15.6a 14.2 5.8 5.1 8.0 8.1 3.1 6.4 7.8 2.8 1.1 Nyanyownh 16.3 4.3 4.2 26.8* 4.1 6.4 3.0 8.9 1.4 4.2 0.8 11.0* Teeamoo 0.5 3.Q 12.6 41.0* 5.1 8.2 2.4 1.6 3.0 5.8 1.0 1.3” Weet 0.0 4.2a 3.3 57.7* 1.0 10.1 3.2 6.1 4.1 4.7 1.3 0.3 Wheeoo 1.2 9.4 6.9 12.6 6.4 5.6 1.4 5.2 3.1 5.1 0.6 11.3*

* Wilcoxon matched-pairs signed-rank test (P c 0.002, n = 10). 1The number of non-zero samples was insufficient to conduct the significance test persistence.Furthermore, females apparently re- noncourtship interactions. It was most often as- spond to differencesin the quality of duets; the sociated with the teeamoo call. Weet probably rate at which females visit a team of males is servesto indicate mild distressin responseto an correlated with the degree to which the sound unwelcome or unexpected stimulus. For exam- frequencies of their songs match one another ple, weet increased during playback of four out (Trainer and McDonald, unpubl. observ.). of five calls, although these increases were not Wit -Virtually all toledo bouts were preceded significant. by a series of wit calls given alternately by the Waanh-This soft call is usedboth during song two male partners Apparently these calls func- and noncourtship activities, for communication tion to synchronize the behavior of partners just among males in close proximity. It may be used before a bout of toledoduets. It also functions in as a contact call to indicate a male’s location conjunction with owng as described below. when he is not otherwise vocalizing. Owng-This call probably signals a male’s ex- Doodoodooand toodleloo- Becauseof the in- citement in the presenceof a female. Males give frequent occurrenceof doodoodooand toodleloo, owngs when they are prepared to dance for a the specific behavioral context, sequence asso- female but the partner is absent, or when a female ciations, and use in responseto playback could is present but has not moved to the dance perch. not be determined. The occurrence of wit and In combination with soft, slower seriesof wits it wheeooincreased after playback of doodoodoo, seems to act to incite females to move to the but the interpretation of this result is unclear. dance perch for solo male dance displays. Wheeoo-This call appears to function as a mobbing call usedby both male and female man- DANCE VOCALIZATIONS akins in the presence of predators. In addition, Nyanyownh-This call is the vocal component wheeoois given by males affiliated with the perch- of the leapfrog portion of the dance display. It zone when a strangemale appears,and may serve probably functions to enhance the display and to attract other resident males to investigate. stimulate the female. Although it is a loud vo- Consistent with a mobbing function, this was the calization, it does not appear to function in at- only call that stimulated like calls in responseto tracting males or females to the dance display. playback. Buzz-weent-This call signals the superior Chitter-This call is a submissive signal to an dominance status of the male to his dance part- older or more dominant male. ner. Buzz- weentoccurs only in the ritualized con- Squawk-This is a distresscall, probably used text of leapfrog or popcorn dance displays, and when a manakin is attacked by a predator, as does not occur in conflict situations that arise well as in mist nets. infrequently in other contexts. The results of the playback experiment show that manakins do not respond in a typical ter- NONCOURTSHIP VOCALIZATIONS ritorial manner to playback of several vocaliza- Teeamoo -A loud vocalization, teeamoo is tions. Territorial speciesusually respond to play- clearly used to attract a singing partner, usually back of conspecific song by approaching and when he is out of visual contact. engaging in countersinging; in some cases they Weet- This loud vocalization occurs during may attack a mount (Petrinovich and Patterson songbouts as well as in the context of male-male, 198 1, Ratcliffe and Grant 198.5, Weary et al. VOCAL REPERTOIRE OF LONG-TAILED MANAKIN 719

1987, Stoddard et al. 1988). Manakins did not ties in which different song types occur in dif- approach closely, nor respond by singing toledos. ferent behavioral contexts have complete rep- Nor did they countervocalize with the tape by ertoiresof functionally distinct vocalizations that responding with the same call type as the stim- are comparable in size to those mentioned above ulus. Only wheeoo stimulated like calls in re- (Smith et al. 1978, Trainer 1987). The large call sponse,but this probably occurredbecause man- repertoires of the Long-tailed Manakin may be akins were stimulated to mob, not becausethey related to the unique nature of male-male inter- were countersinging with the tape. actions, which, unlike the above territorial spe- The use of wheeoo in response to playback cies, are characterized by joint display and co- corroborates other evidence that this call serves operation in the context of intense sexualselection a mobbing function. When McDonald (1993a) characteristic of systems. presented taxidermic mounts of male manakins The size of the Long-tailed Manakin’s vocal in perch-zones, males responded with mobbing repertoire is even more striking when compared behavior and wheeoo calls (though only two with that of other manakins. The revised, mono- physical attacks). McDonald interpreted the phyletic manakin family, Pipridae, contains 40 mobbing of mounts as a responseto strangemales species(Prum 1990a, 1992). The behavior of 13 that might pose a threat to the established affil- of these specieshas been described sufficiently iations of males at a perch-zone. In the playback to estimate the size of the vocal repertoire. At experiment, only the dance vocalization, nyuny- least 12 speciesalso produce mechanical sounds ownh, and wheeoo elicited mobbing calls. The with special wing feathers (Prum 1990b). The sound of dancing might signal the presence of size of the complete acousticrepertoires of man- strange or noncooperative males attempting to akins varies between 1 and 8 (Snow 196 1, 1963a; court females, and stimulate residents to mob. Skutch 1969; Lilll974,1976; Schwartzand Snow The sound of mobbing may serve to recruit other 1978; Snow and Snow 1985; Robbins 1983; Prum lek affiliates to mobbing choruses. 1985; Prum and Johnson 1987; Prum 1990b). Other playback responses illustrate the im- Some of the repertoire sizes may be underesti- portanceof cooperativerelationships among male mates because some specieswere observed for affiliates. Playback of toledo, teeamoo and weet, brief periods or only during the courtship dance. calls that normally occur among perch affiliates, Nevertheless,further study of thesespecies seems failed to elicit mobbing calls. Wits were strongly unlikely to reveal any repertoire size as large as stimulated by three calls, and teeamoo showed that of Chiroxiphia. increases, though not significant, to four calls. If call diversity were purely a consequenceof Our contextual observations of these two vocal- joint, nonterritorial display, then we would ex- izations clearly show that they mediate cooper- pect large repertoires of acoustic signals in other ative interactions among males, including the members of the Pipridae that perform joint dis- joint song.This suggeststhat male manakins gen- plays. A review of the literature, however, does erally respond to playback by soliciting inter- not support this prediction. Besidesthe four spe- action with an affiliated male. cies in the Chiroxiphia, joint displays by The Long-tailed Manakin repertoire of 13 call two males have been observed in eight species: types is unusually diverse compared to those of pyrocephalus,M. regulus (Sick other speciesfor which complete repertoireshave 1967), Masks chrysopterus(Prum and Johnson been described. For example, three suboscine 1987) aureola (Snow 1963a), P. coronata speciesstudied by Smith (1967, 1969, 1971) and (Skutch 1969), P. fusciicauda (Robbins 1983) P. five oscine speciesstudied by Moynihan (1962a, filicauda (Schwartz and Snow 1978) P. serena 1962b, 1963, 1966) have repertoires of 6-8 call (Prum 1985). These specieswould be expected types.Some songbirdspecies develop diversesong to have large repertoiresof acousticsignals. Rep- repertoires composed of several versions of the ertoire sizes of these species ranged from ap- primary song (Kroodsma 1982, Kroodsma and proximately four to eight, although further study Canady 1985, Derrickson 1987). However, in may reveal additional acoustic signals. Appar- species with extremely large song repertoires, ently, joint display alone is not sufficient to ac- acoustically distinct song types seem to play count for large repertoires. The behavior of these functionally similar roles in communication manakins differs in several ways from that of (Verner 1976, Smith and Reid 1979). Even spe- Chiroxiphia, and this may account for their lack 180 JILL M. TRAINER AND DAVID B. MCDONALD of large repertoires. First, only Chiroxiphia man- Further studv of the relationship between vocal akins have obligate joint displays (Snow 1963b, behavior and social organization in manakins is 197 1; Foster 1977, 198 1). In the other manakins, warranted. solitary displays for females occur on a regular basis and may be more frequent than joint dis- ACKNOWLEDGMENTS plays. The occurrence of joint displays in these This project was funded by NSF grant BNS-8814038 other manakins ranges from very infrequent to to JMT, a JessieSmith Noyes Foundation grant from the Organization for Tropical Studies, The Frank M. frequent but appearsnever to be obligatory. Sec- Chapman fund of the American Museum of Natural ond, only Chiroxiphia manakins have both joint History and the American PhilosophicalSociety. DBM advertisement and joint dance. In Pipra fascii- is grateful to the Harry Frank Guggenheim Founda- cauda, two males may display to attract a female, tion, the Organization for Tropical Studies and the but subsequently only the dominant male courts National GeographicSociety for support.Joe and Jean Stuckey graciouslypermitted the use of their farm as her (Robbins 1985). In the remaining species, a study site. Members of the Monteverde community males display individually to attract females, have helped in many ways. Personnel in the offices of usually on dispersed territories, but two males Vida Silvestrewere remarkably courteous,prompt and may perform a joint dance display. None of these helpful in securingnecessary permits. We would like to thank Albert0 Hemandez for help translating the specieshas a joint cooperative vocalization, as abstract,and Mercedes S. Foster and Richard 0. Prum does Chiroxiphia. Finally, the cooperative rather for comments on the manuscript. than aggressivefunction ofjoint display has been demonstrated only in Chiroxiphia (McDonald LITERATURE CITED 1989a, 1989b). To be cooperative, a joint display ATWOOD,J. L., V. L. FITZ, AND J. E. BAMESBERGER. must involve investment by more than one male 1991. Temporal patterns of singing activity at in attracting or exciting females during courtship leks of the White-bellied Emerald. Wilson Bull. (Prum and Johnson 1987). Whether joint dis- 103:373-386. plays are cooperative in this sense is unknown BRADBURY,J. W. 1981. The evolution of leks, p. 138- 169. In R. D. Alexander and D. W. Tinkle teds.], for most manakin species (Snow 1963b, Sick Natural selection and social behavior: recent re- 1967, Prum and Johnson 1987). A cooperative searchand new theorv. Chiron, New York. function of display is questionable in P. coronata BRADBURY,J. W., S. L.-VEHRENCAMP,AND R. M. and P. serenabecause joint displays may be more GIBSON.1989. Dispersionof displayingmale Sage prevalent during the nonbreeding season or in Grouse. 1. Patterns of temporal variation. Behav. Ecol. Sociolbiol. 24:1-14. the absence of females than during courtship DERRICKSON.K. C. 1987. Yearlv and situational (Skutch 1969, Prum 1985). Prum (1985) sug- changesin the estimate of repertoire size in North- gestedthat P. serenajoint displays are probably em Mockingbirds (Mimes polyglottos).Auk 104: aggressiverather than cooperative because they 198-207. FOSTER,M. S. 1977. Odd couples in manakins: a are performed by two males that occupy different study of social organization and cooperative territories, and they usually occur during coun- breedingin Chiroxiphialinearis. Amer. Natur. 11: tersinging when no females are present. 84.5-853. We suggestthat the diversity of the Long-tailed FOSTER,M. S. 1981. Cooperative behavior and social Manakin vocal repertoire is related to the com- organizationof the Swallow-tailed Manakin. (Chi- roxiphia caudata).Behav. Ecol. Sociobiol. 9: 161- plex, nonterritorial, cooperative social system. 171. Each male maintains a network of relationships GOODMAN,L. A. 1968. The analysis of cross-classi- with several other males that persistsfor a num- fied data: independence,quasi-independence, and ber of years (McDonald 1989a). These long-term, interactionsin contingencytables with or without missing entries. J. Am. Statist. Assoc. 63: 1091- cooperative relationships may favor the evolu- 1131. tion of diverse communication signals. At least HOLDRIDGE,L. 1966. The life zone system. Adan- 10 ofthe Long-tailed Manakin call types function sonia 6: 199-203. in communication among males; they are di- KROODSMA,D. E. 1982. Song repertoires:problems in their definition and use, p. 125-146. In D. E. rectedat males or occur in the absenceof females. Kroodsma and E. H. Miller Ieds.], Acoustic com- It may be that excluding males from a territory munication in birds. vol. 2. Academic Press.New does not require as many functionally distinct York. KROODSMA,D. E., ANDR. A. CANADY. 1985. Differ- call types as doescooperation. Among manakins, encesin repertoire size, singingbehavior, and as- joint display in the absence of cooperation ap- sociatedneuroanatomy among Marsh Wren pop- pears not to be associatedwith large repertoires. ulations have a eenetic basis. Auk 102:439-446. VOCAL REPERTOIRE OF LONG-TAILED MANAKIN 781

LEMON,R. E., AND C. CHATL~ELD.197 1. Organisation tailed Manakin (Pipra fasciicauda). Condor 87: of song in Cardinals. Anim. Behav. 19:1-17. 449-456. LILL, A. 1974. Sexual behavior of the lek-forming SCHWARTZ,P., ANDD. W. SNOW. 1978. Display and White-bearded Manakin (M. trinitatus related behavior of the Wire-tailed Manakin. Liv- Hartert). Z. Tierpsychol. 36: l-36. ing 17:51-78. LILL, A. 1976. Lek behavior in the Golden-headed SICK, H. 1967. Courtship behavior in manakins (Pi- Manakin, Pipra erythrocephala, in Trinidad (West pridae). J. Omith. 10&269-302. Indies). Z. Tierpsychol. Adv. Ethol. 18. SKUTCH.A. F. 1969. Life histories of Central Amer- MCDONALD,D. B. 1989a. Cooperation under sexual ican birds. III. Pacific Coast Avifauna 35: l-580. selection: age-graded changesin a lekking bird. SLATER,P.J.B. 1973. Describing sequencesof behav- Amer. Natur. 134:709-730. ior.u. 131-153.ZnP.B.G.BatesandP.H.Klonfer MCDONALD,D. B. 1989b. Correlatesof male mating [eds:],Perspectives in ethology.Plenum Press,New successin a lekking bird with male-male cooper- York. ation. Anim. Behav. 37:1007-1022. SMITH, D. G., AND F. A. REID. 1979. Roles of the MCDONALD,D. B. 1993a. Delayed plumage matu- songrepertoire in Red-winged Blackbirds.Behav. ration and orderly queues for status: a manakin Ecol. Sociobiol. 5~279-291. mannequin experiment. Ethology 94:3145. SMITH, W. J. 1967. Displays of the Vermilion Fly- MCDONALD, D. B. 1993b. Demoarauhic- _ conse- catcher (Pyrocephalus rebinus). Condor 69:601- quences of sexual selection in the Long-tailed 605. Manakin. Behav. Ecol. 4:297-309. SMITH,W. J. 1969. DisplaysofSayornisphoebe(Aves MOYNIHAN,M. 1962a. Display patterns of tropical Tyrannidae). Behaviour 33:283-322. American “nine-primaried” . I. Chlo- SMITH,W. J. 1971. Behavioral characteristicsof ser- rospingus. Auk 79~310-344. pophaginine tyrannids. Condor 73:259-286. MOYNIHAN,M. 1962b. Display patterns of tropical SMITH,W. J., J. PAWLUKIEWICZ,AND S. T. SMITH. 1978. American “nine-primaried” songbirds. II. Some Kinds of activities correlatedwith singingpatterns speciesof Ramphocelus. Auk 79:655-689. of the Yellow-throated Vireo. Anim. Behav. 26: MOYNIHAN, M. 1963. Display patterns of tropical 862-884. American “nine-primaried” songbirds. III. The SNOW,D. W. 1961. The displays of the manakins Green-backed Sparrow. Auk 80: 116-144. Pipra pipra and virescens. Ibis 103: MOYNIHAN, M. 1966. Display patterns of tropical 110-113. American “nine-primaried” songbirds. IV. The SNOW,D. W. 1962. A field study of the Black and Yellow-rumped Tanager. Smithsonian Miscella- White Manakin, Manacus manacus, in Trinidad. neous Collection 140(5):l-34. Zoologica 47:65-104. PETRINOVICH,L., AND T. L. PATTERSON.198 1. The SNOW,D. W. 1963a. The display of the Orange-head- responsesof White-crowned Sparrowsto songsof ed Manakin. Condor 65:44-48. different dialects and subspecies.Z. Tierpsychol. SNOW,D. W. 1963b. The display of the Blue-backed 57:1-14. Manakin, Chiroxiphia pareola, in Tobago, W. I. PRUM,R. 0.. 1985. Observationsofthe White-front- Zoologica 48:167-176. ed Manakin (Pipra serena) in Suriname. Auk 102: SNOW,D. W. 1971. Social organization of the Blue- 384-387. backed Manakin. Wilson Bull. 83:35-38. PRUM, R. 0. 1990a. A test of the monophyly of the SNOW,D. W. 1977. Duetting and other synchronised manakins (Pipridae) and of the (Cotin- displays of the Blue-backed Manakins, p. 239- gidae)based on morphology.Oct. Pap. Mus. Zool. 251. In B. Stonehouseand C. M. Perrins [eds.], Univ. Michigan 723: 144. Evolutionary ecology.University Park Press,Bal- PRUM,R. 0. 1990b. Phylogeneticanalysis of the evo- timore. lution of display behavior in the neotropical man- SNOW,B. K., AND D. W. SNOW. 1985. Display and akins (Aves: Pipridae). Ethology 84:202-23 1. related behavior of male Pin-tailed Manakins. PRUM,R. 0. 1992. Syringealmorphology, phylogeny, Wilson Bull. 97~273-282. and evolution of the neotropical manakins (Aves: STODDARD,P. D., M. D. BEECHER,AND M. S. WILLIS. Pipridae). Amer. Museum Notitates 3043: l-65. 1988. Responseof territorial male SongSparrows PRUM. R. 0.. AND A. E. JOHNSON. 1987. Disnlav to song types and variations. Behav. Ecol. Socio- behavior, foragingecology, and systematicsof the biol. 22:125-130. Golden-winged Manakin (Mask chvysopterus). TRAINER,J. M. 1987. Behavioral associationsof song Wilson Bull. 99:521-539. types during aggressiveinteractions among male RATCLIFFE,L., AND P. GRANT. 1985. Speciesrecog- Yellow-rumped Caciques.Condor 89:73 l-738. nition in Darwin’s finches (Geospiza Gould). III. TRAINER,J. M. 1988. Singing organization during Male responsesto playbackof different songtypes, aggressiveinteractions among male Yellow-rump- dialects, and heterospecificsongs. Anim. Behav. ed Caciques.Condor 90:68 l-688. 331290-307. VERNER,J. 1976. Complex song repertoires of male RICE,W. R. 1989. Analyzing tablesofstatistical tests. Long-billed Marsh Wrens in easternWashington. Evolution 431223-225. Living Bird 14:263-300. ROBBINS,M. B. 1983. The display repertoire of the WEARY.D. M.. R. E. LEMON.AND E. M. DATE. 1987. Band-tailed Manakin (Pipra fasciicauda). Wilson Neighbour-strangerdiscrimination by songin the Bull. 95:321-342. Veery, a specieswith songrepertoires. Can. J. Zool. ROBB~NS,M. B. 1985. Socialorganization ofthe Band- 65:1206-1209.