A Review of Duetting, Sociality and Speciation in Some African Barbets (Capitonidae)

Condor 85323-332 0 The Cooper Omithologlcal Society 1983

A REVIEW OF DUETTING, SOCIALITY AND SPECIATION IN SOME AFRICAN BARBETS (CAPITONIDAE)

LESTER L. SHORT

AND JENNIFER F. M. HORNE

ABSTRACT.-Our East African studies have focused on the sociality, duetting and intra- and interspecific systematicsof barbets (Capitonidae). The Afrotropical barbets include 14 duetting, social speciesand 22 non-duetters, both social and non-social. In this review of them, comparisons are broadly conducted, but cen- tered about the ground barbets (Trachyphonus) and some speciesof Lybius. The sexually dichromatic Red and Yellow Barbet (T. erythrocephalus)occurs in pairs and in social groups; duetting is by the pair, or mainly by the primary pair of a group. Within groups, secondary adults and young may join in part of the duet but their roles are limited by the dominance of the primary pair. The sexual songs are not very different. Subspeciesof erythrocephalusdiffer little morphologically except in color tone and size. Darnaud’s Barbet (T. darnaudii) is sexually monomorphic with racesmorphologically somewhat divergent. Sexual duetting roles are distinctive, and in groups the primary pair actively prevents duetting of secondaryadults or young birds. Removal of a duetter leads to instant replacement from within the group, another bird assuming the duet role of the lost bird. In the Red-faced Barbet (Lybius rubr$zcies), the Black-collared Barbet (L. torquatus) and the Black-billed Barbet (L. guifsobalito) the sexes are alike, pairs and social groups occur, and duetting is by the primary pair, with distinct sexual duet roles, following a greeting ceremony in which other individuals may take part. Secondary “pairs” occur in some groups; these may duet when apart from the primary pair, but not in the presenceof the latter. Allopatric torquatus and guifsobalitoshow no or modest geographicvariation and have similar though distinctive duets. Lybius rubrifaciesshows no morphological geographicvariation, and marginally overlaps torquatus;its duet markedly differs from that of torquatus. The duet songs, whatever the degree of genetic control, are species-specific. Only paired duetting birds hold territories. Systematic comparison and studiesof sociality and duetting in Afrotropical barbets suggestthat the nature of their social systemsand their duets, where these exist, has affected their speciation, and vice versa.

Duet-singing in birds can be intricately timed the genetic basesand experiential background and complex (Thorpe 1972, 1973). Duets are that go into precisely timed duetting songs(Ty- usually between paired birds and may be pre- roller 1974). It seemslogical that the evolution cisely alternating (antiphonal), or in unison of speciesin avian groups that are social and (polyphonic). Unison-singing may simply be regularly duet has been affected by their so- the simultaneous singing of two birds at their ciality and duetting habits. More particularly, own individual rhythms with overlapping the ability of individuals of one population to songs, or more precise singing together in a duet appropriately-and hence to pair with in- coordinating rhythm. Additionally, the songs dividuals of another population-ought to be of the duetters may be identical, they may dif- of importance in relation to speciation. Gen- fer somewhat, or may be strikingly different. eralizations must be tempered, however, by These categoriesof duetting are modified here the fact that each avian species has its own from those of Thorpe (1973). unique distribution, ecology, history, and par- Although the relative importance of the dif- ticular relations with congeneric, distantly re- ferent functions of duetting is arguable (Wick- lated, and even unrelated sympatric species. ler 1976) the complexity and coordination of We have studied various Afrotropical bar- many duets clearly shows close interplay be- bets in the field, devoting attention especially tween the duetters. Studies of the development to their sociality, vocal habits and related be- of duets in young barbets also indicate the havior. Barbets generally are aggressive,om- complex nature of duets. Such works suggest nivorous or fiugivorous, hole-nestingand hole- 13231 324 LESTER L. SHORT AND JENNIFER F. M. HORNE

TABLE 1. Duetting features of barbets

Stactolaemaolivacea 0* 0 pitch mainly occasionally Lybius melanopterus 0 0 0 Lybius leucocephalus A 0 0 + & Lybius vieilloti + 0 pitch + 0 Lybius rubrifacies + f +++ 0 + Lybius torquatus + 0 ++ 0 + Lybius guifsobalito + 0 ++ 0 + Trachyphonusmargarita&s + 0 + 0 + Trachyphonuserythrocephalus + 0 + 0 + Trachyphonusdarnaudii + 0 +++ 0 +

* 0 mdlcates absence, + prescncc:in the third column the plusesmdlcate the degree of difference III form (+, some d~lTerence, to +++. vev marked deference). Parentheses mean that the data are not concluswe or certain for all subspecies.

roosting, sedentarypiciform birds. The species Tape recordingswere made by Horne using we have investigated in some detail that form a Stellavox SP-7 recorder, 72-cm parabolic re- the subject of this article are: the Green Barbet flector, and Schoeps CMC-45 microphone, (Stactolaema olivacea, Short and Horne with playback conductedby Short usinga Sony 1980b); the White-headed Barbet (Lybius feu- cassette recorder. Audiospectrographic anal- cocephalus,two subspecies;unpubl. data); the ysis involved use of a Kay Elemetrics Sona- Black-billed Barbet (L. guifsobalito; unpubl. Graph 6061B to produce about 2,500 sono- data); the Black-collared Barbet (L. torquatus; grams of voices of the barbets discussed.The Short and Horne 1982); the Red and Yellow various analyses took place in the American Barbet (Trachyphonus erythrocephalus;Short Museum of Natural History. and Horne 1980a, in press); and Darnaud’s Barbet (T. darnaudii, three subspecies;Short DUET FUNCTION, SEASONALITY, and Horne 1980a, in press). Fewer details are AND THE DUETTERS available from our less extensive (Short and Duetting barbetsare sedentaryand they clearly Horne, unpubl. data) studies of the Brown- employ the duet in territorial defense and breasted Barbet (L. melanopterus), the Yel- maintenance.At a distance,only the loud notes low-billed Barbet (T. purpuratus) and the of the duet can be heard, and thus they can be Crested Barbet (T. vaillantii). We also have effective in those functions. Only at rangesup been supplied by C. Chappuis with tape re- to 20 or 30 m are the pre- and post-duet notes cordings of duets of the Vieillot’s Barbet (L. and the softer duet notes audible to humans, vieilloti; Payne and Skinner 1970) and the Yel- and presumably to the duettersthemselves and low-breasted Barbet (T. margaritatus; Short to any members of their group closely asso- and Horne, unpubl. data). The duetting fea- ciated with them. This suggeststhat these soft- tures of thesebarbets are summarized in Table er notes function in pair relations and pair 1. Our field researchesmainly have been in maintenance. Kenya for several months of each of the years Duets are uttered to some extent the year- 1976-1982; the Red-faced Barbet (L. rubri- round, although lessfrequently when the birds facies) was investigated in Rwanda during Jan- are not breeding. Duetters, as well as other uary 1982. members of a barbet group, even including Ecological, distributional, morphological, subadults, have enlarged gonads all year (go- social and vocal information on the African nads over 2.5 mm in diameter, pers. observ.). barbets mentioned above forms a background Individual barbets of duetting speciesdo not for consideration of the interrelationship singalone; when one bird rarely does so under among their sociality, form and complexity of exceptional circumstances,it singsonly its par- singing,and evolution, particularly speciation. ticular duet song,and that haltingly. Thus, the In this paper we treat Lybius and Trachy- singingof these barbets is virtually entirely in phonus in special detail, as their species are the form of duets, and in most speciesthese widespread and vary interspecifically and in- are uttered only by the primary pair. Dar- traspecificallyin the parametersjust noted. The naud’s Barbets who are members of a group ecologyand sociality of duetting and non-duet- other than the duetting pair are prevented by ting barbets are compared. Within this frame- aggressiveactions of that pair from duetting work we explore the possibleeffects of specia- (Short and Horne 1980a). In groups of Lybius tion on the occurrence and complexity of rubrifacies and L. torquatus, sub-pairs (pairs duetting in the African barbets. other than the primary pair) occasionallyduet. BARBET SOCIALITY 325

However, they do so only in the absenceof the “moderate” response,reacting either to play- primary pair, which if within hearing, usually back or to actual duets of a related duetting approach the sub-pair, causing them to cease species by some approaching and mainly duetting. Singlebarbets in the wild do not utter counter-duetting (seeLybius rubrifaciesand L. both duet parts. The duets sungby the pair are torquatus, below). So far in our researches, crucial for the establishment and holding of a however, very close,instant and consistentap- territory. Hence, a barbet is likely to maintain proaches to a playback duet (i.e., a “strong” a territory, and to breed, only if it finds a duet response), followed by search behavior and partner of the opposite sex and the two manage counter-duetting mark only conspecific inter- to synchronize their duet. actions. For example, this occurs among sub- The non-duetting members of a group are species of Trachyphonus darnaudii (see be- exposed to frequent duets of the primary pair, low). Red-fronted (Pogoniulus pusillus) and and to some extent to those of distantly duet- Yellow-fronted (P. chrysoconus)tinkerbirds ting adjacent pairs. This exposure may influ- also react in this manner to playback of each ence the form of duet song and the ability to other’s two main calls, (not duets or songsas pair with a duet partner later in life. As a pair such). In fact, we regularly used the calls of the duet, they may be surrounded by the often allopatric tinkerbird to elicit immediate, ag- conspicuous,associated birds of a group (the gressiveresponses within the range of its coun- duetters’ presencestressed, in the casesof Tra- terpart. Although these reactions are indistin- chyphonusmargarita&s and T. erythrocepha- guishablefrom those of conspecifics,we do not lus, by loud rattling calls of incoming group consider the specific status of these non-duet- members who may actually join in singingpart ters as established. of the duet). The group may reinforce the duetters’ ability to hold or add to the size of a COMPARISON OF DUETTING territory. Our meager data suggestthat groups, AFRICAN BARBETS and especiallylarger groups,of barbets occupy Differentiation and speciation (see, e.g., Mayr and maintain larger, ecologically more favor- and Short 1970) can be compared in 14 duet- able territories than do pairs. ting and 22 non-duetting speciesof the 36 (of In our barbet studies we frequently used 42) Afrotropical barbets whose vocal habits playback techniques, despite their problems are known (Table 2). Among the non-duetters, (different seasonal schedules of birds tested, 7 are monotypic, 11 are simply polytypic, 4 duplication of social situation, etc.). We have have megasubspecies(i.e., are strongly poly- encountered diverse responses to playback, typic, see Amadon and Short 1976) and 12 which can be categorized hierarchically as fol- are involved in five superspecies.In contrast, lows: many barbetsand woodpeckerscall, sing, of 14 duetters, 2 are monotypic, 6 are simply or even, after some time, duet as a response polytypic, 6 have megasubspecies,and only 3 to playback of calls or duets of unrelated or are involved in two superspecies.These data distantly related species.Part of these “weak” suggestthat: (1) duetters show considerably responsesmay be due to the quality and gen- more differentiation, especially strong differ- eral resemblance of their voices, but part also entiation, than non-duetters; and, (2) specia- is likely to relate to competition for nesting tion may be completed more rapidly in duet- and roosting cavities. Thus, playback of sev- ters (i.e., duetters tend to be more polytypic eral duets of Lybius guifsobalitomight elicit a and have more megasubspeciesbut fewer al- “nyah” call or two from a Spotted-flanked Bar- lospecies). bet (Tricholaema lachrymosa)and duetsplayed Despite their appeal, such generalizations back of Trachyphonuserythrocephalus some- have complications. For example, genera such times got such a response from the African as Gymnobucco (non-duetters), Pogoniulus Black-throated Barbet (Tricholaema melano- (non-duetters), Lybius (many duetters) and cephala). Also, the calls of Cardinal Wood- Truchyphonus (all duetters) differ greatly in peckers(Dendropicos fuscescens) often evoked their morphology and ecology. Approaching a responseby T. melanocephala,and the loud, these data ecologically for duetters and non- ringing calls (songs)of the Nubian Woodpeck- duetters (Table 2) one finds that 14 of 22 non- er (Campethera nubica) might stimulate Tra- duetters are forest barbets, whereas only 3 of chyphonus erythrocephalus to duet (Short 14 duetters are forest birds. Within Trachy- 1982). None of these responsesinvolve more phonus the sole forest speciesis the only weak than a vocal utterance; the respondant usually (infrequent) duetter, but largely non-forest Ly- does not approach the caller unless the latter biushas both duettersand non-duetters.Hence, is near its roosting or nestingcavity. At a some- ecological and behavioral factors, as well as what different level, as discussedbelow, inter- phylogenetic factors are likely to influence specifically territorial duetters may show a duetting and speciation (but see Discussion). 326 LESTER L. SHORT AND JENNIFER F. M. HORNE

TABLE 2. Duetting, sexual dimorphism and systematicsof Afrotropical barbets.

SEWally SCWdly Simply Superspecies SptTk monomorphx dimorphic Monotypic polytypic relations

Gymnobuccocalvus 0 X 0 X 0 0 G. bonapartei X 0 : X 0 G. peli X 0 X 0 0 G. sladeni ri; X 0 X 0 0 ; Stactolaemaanchietae 0 X 0 0 X 0 S. whytii 0 X 0 0 0 X 3 S. leucotis 0 X 0 0 X 0 0 S. olivacea X X 0 0 0 X 0 Buccanodonduchaillui 0 X 0 X 0 0 Pogoniulusscolopaceus i. X 0 0 X 0 0 P. simplex 0 X 0 X 0 0 P. coryphaeus 0 X 0 0 X 0 3. P. leucomystax 0 X 0 X 0 0 x7 P. “makawai” 7 ? 7 X 0 9 7 P. bilineatus X 0 0 0 x 0 P. subsulphureus (:I X 0 0 X 0 0 P. atrojlavus c;, X 0 X 0 0 P. chrysoconus X 0 X 0 P. pusillus 0 X 0 i 0 X ; Tricholaemahirsuta 0 0 X 0 0 X 0 T. melanocephala Xi X 0 0 0 X 0 T. lachrymosa Xk 0 X 0 X 0 0 T. diademata X 0 0 X 0 T. yrontata‘ ” i X 0 X 0 0 T. leucomelaina 0 X 0 0 X 0 3 Lybius undatus ? X 0 0 0 X 0 L. vieilloti Xi X 0 0 X 0 L. leucocephalus Xi X 0 0 0 X L. chaplini ? X 0 X 0 0 ;? L. rubrifacies X X 0 X 0 0 L. gutjsobalito X X X 0 i L. torquatus X X : 0 0 X i L. melanopterus 0 X 0 X 0 0 0 L. minor 3 X 0 0 0 X 0 L. bidentatus Xi 0 Xi 0 X 0 3.7 L. dubius 3 0 Xi X 0 0 7 L. rolleti 7 0 Xi X 0 0 3. Trachyphonuspurpuratus Xi X 0 0 0 X 0 T. vailtantii X X 0 0 X 0 T. margaritatus X 0 X X 0 T. erythrocephalus X 0 X i X 0 ; T. darnaudii X X 0 0 0 X 0

The nature of duetting itself renders some- the “groups” are families. The non-duetter cat- what simplistic a dichotomous classification egory also presentscomplications. We believe of duetters and non-duetters. Some duetters it significant that the non-duetters include the actually achieve duets infrequently (e.g., Stac- extremes of the more solitary, less social bar- tolaema olivacea, Trachyphonuspurpuratus). bets (e.g., speciesof Pogoniulus),as well as the Others, such as Lybius leucocephalusvary ra- most socialof Afrotropical barbets(e.g., species cially in their “duetting” (Short and Horne, of Gymnobucco, Stactolaema, Lybius mela- unpubl. data). The more specialized duetters nopterus,pers. observ.). Thus, the duetters in- have different duet roles (i.e., the sexes may clude neither the least, nor the most social bar- have separate songs), and their helpers may bets. Where there is intrageneric variation in singin the duet. Social groupswithin the complex duetters may include up to eight individ- duetting, duetters often are closerelatives with uals, although in all these speciesthe majority allopatric ranges. In contrast, sympatric con- of suchgroups probably consistsof pairs. That genersoften are more distantly related and the is, social groups are common but do not rep- sympatry involves a duetter and a non-duetter resent the usual situation, except, of course, (e.g., Lybius torquatus,a duetter, and L. mela- immediately after fledging of the young when nopterus,a non-duetter). BARBET SOCIALITY 327

THE GROUND BARBETS, GENUS darnaudii have two distinctive duet roles, a TRACHYPHONUS phrase or set of the song of one duetter pre- Among the ground barbets (7’rachyphonus), cisely fitting within the several-note set of the duets of the forest Yellow-billed Barbet are other duetter. In all races of darnaudii, groups uncommon, and appear from limited data to are smaller than in its two relatives. Subor- be associatedwith the roosting or nesting cav- dinates rarely sing or duet with adults. The ity (Short and Horne, unpubl. data). The subspeciesof T. darnaudii also are morpho- woodland CrestedBarbet duetsmore frequent- logically distinctive. Those races (T. d. dar- ly, especiallyin responseto playback of its calls naudii, T. d. bohmi) that are sympatric with (Rwanda, 1982 experiments, unpubl. data). In large T. erythrocephalusare smaller than are each of these two speciesthe sexes are alike, the others, and the two subspeciesdiffer vo- and so are their duet notes, except for some cally from one another only in the temporal variation in pitch. Morphological geographic pattern of their duets (Short and Horne, in variation in the Yellow-billed Barbet is slight, press:Fig. 4). The southern race, T. d. emini, except that the West African forest isolate T. is distinctively colored and slightly larger than purpuratus gojinii is morphologically distinc- darnaudii and bohmi; it partly overlaps with tive and is sometimes treated as a separate much larger, and ecologically divergent (more species. Known vocalizations of gofinii (C. arboreal, tree-nesting) T. vaillantii. In its West Chappuis, copy tapes) differ only in some mi- Kenyan-West Tanzanian range, T. d. usam- nor details from those of the other forms, but biro is known to meet a congener, T. eryth- the bird’s duet is not yet known. Geographic rocephalusat only two points (we have ex- variation in T. vaillantii is very slight indeed plored much of the eastern fringe of the range (Short and Horne, unpubl. data). of usambiro in Kenya, and examined all rel- The two large ground barbets, T. margari- evant literature). Apparently erythrocephalus tutus and T. erythrocephalus,are parapatric, dispersescasually into areasoutside its normal closelyrelated specieswhose exact distribution range, as indicated by sporadic occurrencesin and interactions where they meet in Somalia Nairobi suburbs, and one Serengeti record and Ethiopia are unknown. In both these (Schmidl 1982). Not distinctive in color (it is species,the sexesdiffer in head pattern (males no more marked than is T. d. bohmi), usam- with black crown patch, lacking in females). biro is considerably larger in size. We consider Both nest and roost in holes excavated in it significant that the race of T. darnaudii that stream-beds or termite mounds. The coordi- is most like T. erythrocephalusin body size nated polyphonic duetsare very similar in these and in size of bill is the race allopatric with T. two species(pers. observ.; Short and Horne, erythrocephalus,i.e., it may be showing “re- in press:Fig. l), the sexeshaving slightly dif- lease” in the absenceof a larger congener. By ferent notes. Matching sexes,the notes of the similar logic, the simpler, divergent duet form two taxa parallel one another in form and pitch. may be due to “release” from possible con- Geographical variation in morphology of both fusion of duet song with that of a congener speciesis relatively slight, featuring size shifts (Short and Horne, in press). Unlike all other and brightnessof color (Short and Horne, un- congeners,but exactly like the other three races publ. data). We lack recordingsof the duets of of T. darnaudii, usambiro excavates a nesting T. erythrocephalusand of T. margaritatusfrom hole straightdown into the earth on flat terrain. Ethiopia and Somalia, where their rangesmeet. The duet of Trachyphonusdarnaudii usam- Kenyan erythrocephalus,however, singsduets biro is composed of two distinct songs.One of very like those of Chadian margaritatus. Their theseclosely resemblesduet songsof T. d. boh- songsare so similar that unless these species mi, T. d. darnaudii and T. d. emini (Short and vocally diverge where they meet, there exists Horne 1980a; in press:Figs. 3-6); the other is a reasonablepossibility of their interbreeding. a structurally simpler, grating rattle overlap- Trachyphonus darnaudii is smaller than ping the partner’s duet-note sets about as do either T. erythrocephalusor T. margaritatus. the two duet forms of the other races.Playback It overlaps broadly in range with erythro- experiments in the field indicate strong reac- cephalusbut not with margaritatus. In contrast tion (close approach to the person playing the to the latter two species,the sexesare alike in tape, counter-duetting and following the sound darnaudii. Its two duet song-forms and asso- consistently) of usambiroto bohmi duets, and ciated visual displays differ from those of the vice versa. The visual displays of duetting other two species (Wickler and Uhrig 1969, usambiro analyzed from motion pictures are Fernald 1973, Kalas 1973). However, the two identical to those of the other races. Experi- duet song types of darnaudii are not always mental removal of one duetter of a pair in sexually specific (Short and Horne 1980a, in bohmi and in usambiro resulted very soon press; see below). All four subspeciesof T. (within 3-5 min) in replacement of the missing 328 LESTER L. SHORT AND JENNIFER F. M. HORNE

TABLE 3. Comparison of duets of three speciesof Lybiu~.~

Duration- Duration- pre-duet post-duet Duration- greeting Duration- greeting Number ;sfez full duet CW3llO”y duet CW3llO”y of duet sets Species (s) (s) (9 6) sets (per second) rubrifacies(n = 5) 18.47 5.49 6.55 6.49 12.5 0.52 torquatus (n = 29) 9.04 3.47 5.57 0 10.8 0.51 guifobalito (n = 7) 8.65 2.78 7.55 0 16.1 0.47

aThe n mdicates the number of full, clear duets analyzed, a duet set includes a single pair of coordinated notes of the two duetters.

duetter by a subordinate member of the group, COMPARISON OF SOME SPECIES OF LYBZUS who in all casessang the appropriate duet role. Two, and as many as three, consecutive re- The species of Lybius we have studied offer movals in five experiments resulted in appro- useful information concerning duetting and priate, quick replacement from within each systematics (Tables 1, 2). Three well-studied group (the groupshad four-seven individuals). antiphonal duetters are L. rubrzjacies,L. guif- We also obtained males that sangthe duet part sobalitoand L. torquatus(Tables 1,3). Of these, usually ascribed to the female, as well as fe- rubrzjaciesinhabits a very small area of Rwan- males that sangthe supposedmale role (Short da, northwestern Tanzania, and southwestern and Horne, in press). Hence the vocal duet Uganda; guifsobalitohas a larger but not very roles are not sexually absolute, but vary. The extensive range in Uganda, Ethiopia and west- supposedlysexual, visual display roles can vary ernmost Kenya; and torquatus is widely dis- as well, independently of the vocal roles (Short tributed from Angola, Zaire, Rwanda and and Horne, motion pictures; in press). Also, Kenya to South Africa. In terms of geographic usambiro varies greatly in duet form (more so morphological variation, rubrifaciesis mono- than in bohmi and darnaudii), such that some typic, and we also consider guifsobalitomono- duets contain a “grating” song whose formed typic (Short and Horne, unpubl. data). In con- notes are nearly identical to those of the other trast, torquatus is morphologically variable, song (Short and Horne, in press:Fig. 6). with one major variant (the subspecieszom- Our data for T. darnaudii show that geo- bae) distinctive, if only by virtue of replace- graphic variation in duet form coincides gen- ment of red color on its head by black and erally with morphological divergence (there is white. The other morphological variation in also some local variation within races, espe- torquatusis chiefly in intensity of color and in cially in usambiro;Short and Horne, in press). size. Lybius rubrifaciesespecially, and proba- The intraracial variation and strong playback bly L. guifsobalito,are not apt to vary greatly reactions of darnaudii, bohmi and usambiroto in their vocal repertory, although each has been each other’s duets suggestthat some structural studied at only a relatively few sites. Despite shifts in the duet do not preclude reaction to morphological geographicvariation in torqua- it as if to members of the same population. tus, its duet does not vary greatly (personal Whether or not effective interbreeding can oc- experience in Rwanda, Kenya, Malawi; Payne cur between a young, potential duetter of the and Skinner 1970: 174, South Africa). Kenyan vocally most divergent race, usambiro, and an birds tend to have higher-pitched notes than appropriate partner of the other races is not larger, more southern birds, and Rwandan known. We are presently engaged in seeking duets show longer type B notes (Short and points of contact between usambiro and boh- Horne 1982). The three speciesare closely re- mi. On the basisofits duet form, Wickler (1973) lated, rubrifacieshaving been allied especially considered T. d. usambiro as a speciesdistinct with guijkobalito(White 1965). All are sexually from T. darnaudii. However, we concur (Short monomorphic, and sing antiphonal duets in and Horne, in press) with Payne (197 1:136) pairs with two distinct (least so in rubrifacies) that such separationis unwarranted. The latter duet roles (songs),but their associatedvisual view is supported by: the lack of strong color displays generally are similar. Some differ- differences of usambiro from other races of T. encesare that rubrifaciesbill-snaps distinctly, darnaudii; the apparently identical nesting unlike the others, and rubrifaciesand torquatus habits and visual displays of usambiro and use more wing and flight displays than does those of the other races; resemblance of one guifsobalito. duet role (song)to the duet songsof the others; When studying the systematics of related and playback responsesof the races to each species,one should consider interspecific con- other’s duets. tacts and possibleinteractions that may affect BARBET SOCIALITY 329

0' 0.5 Xl 1.5 210 - 81 ’ D /‘ I‘ , 8i

TIME IN SECONDS FIGURE 1. A, B, parts of pre-duet, C, start of duet and D part of post-duet of Red-faced Barbet. E, start, and F, end of duet of Black-collared Barbet. G,’ start, and H, end of duet of Black-billed Barbet. I, part of pre-duet of Black- collared Barbet. J, part of pre-duet of Black-billed Barbet. K, three duet sets from middle of duet of Red-faced Barbet. L, three duet setsfrom middle of duet of Black-collaredBarbet. M, three duet setsfrom middle of duet of Black-billed Barbet. All sonogramsmade with wide-band filter.

their morphology and behavior (Short 1969, It is tempting to view the extreme duet form 1973). Lybius guifiobalito is entirely allopatric of rubrzjkciesas possibly related to its sym- with L. torquatusand L. rubrifacies,but rub- patry with torquatus.However, in the area of rifaciesmeets torquatusand its pairs and groups Rwanda where both are present, each utters hold interspecificterritories againstthat species greeting ceremonies with or without duets in in eastern Rwanda. Comparing the duet forms responseto greeting ceremonies and duets of of these three species(Fig. 1, see also Table 3) the other, and to playback of the other’s duets reveals structural differences among them, (both also respond to gufiobalito duets played guifsobalito and torquatus having somewhat to them). These playback responsesinvolve similar duet songs(both roles) and a pre-duet counter-duetting or greeting ceremonies ut- greeting ceremony. The duet songs of rubri- tered at a distance, with no close approach or facies differ markedly from those of the other overflying of the playback recordist, or of the two, and this speciesplaces its duet within a birds of the other species.This moderate re- framework of pre- and post-duet notes (greet- sponseto a congener’s duet does not preclude ing ceremony notes).‘Not shown on the figure the functioning of the precise details of duet are the characteristicpreliminary bill-snaps of structure to prevent mixed matings. Barbets rubrzjkcies,which are lacking in the others. generally tend to call or even duet when they Within all three species,the birds of a group hear the distant duets or calls of other sym- other than the primary pair participate, if they patric barbets, whether or not they are closely are present, in the greeting ceremonies (pre- related, as noted above. The notes of the greet- and post-duet), but only the primary pair duets ing ceremony (pre- and post-duet)calls are very in the vast majority (90%) of cases. much alike in the three speciesof Lybius (Fig. 330 LESTER L. SHORT AND JENNIFER F. M. HORNE

1). The ecology of these speciesis similar, and stricted to wetter habitat, and is not numerous. it may be mutually advantageousfor them not Lybius rubrifacies(as is its dry woodland hab- to overlap territorially. Of course, the re- itat) is more widespread in eastern Rwanda, sponsesmay be learned, in order to function with a larger population than that of torquatus, interspecifically. All of these birds duet con- and it occurs beyond the range of torquatus, spicuously from tree-tops, dead trees, tops of considerablyto the north and west of their area bushes,etc. Any group of one of these species of contact in southern Akagera Park. gathering conspicuouslyand uttering notes of Placingthe data on thesethree Lybius species a greetingceremony with associatedvisual dis- in context (see Table 1): Lybius melanopterus, plays is bound to attract the attention, and which is highly social, does not duet and has elicit reaction, of either territorial conspecific no song as such, individually uttering inter- individuals or sympatric barbet species who active and noisy, loud location notes; L. vieil- may see or hear the calling, displaying birds. loti sings simultaneous polyphonic “duets” The form and timing of the duet notes and sets with a “pre-duet” greeting ceremony, but does of rubrifaciesand torquatus are so dissimilar not actually duet in a coordinated fashion; and that their interbreeding is unlikely. It is con- L. leucocephalusvaries racially, but tends to ceivable, but unlikely that the vocally more have raucous greeting ceremonies, frequently similar but equally morphologically different calling near potential nesting cavities (Table guifsobalito and torquatus might interbreed, 1). Races of L. leucocephalusshow strong vo- were they to meet. cal differences,for apparently populationsfrom We must examine the very precise nature of Cameroun (L. 1. adamauae; C. Chappuis’s the duets, the structure and timing of their tapes) east to western Kenya (nominate L. 1. notes, and differences between the speciesin leucocephalus, unpubl. 1982 studies) show these parameters in order to consider the ef- clear, formed polyphonic duets, the songssung fects of vocal variations on the evolution of in unison. These differ from the unison-calls the individual species.The duet notes of the of Kenyan L. 1.albicaudus and L. 1.senex much Red-faced Barbet have a very different struc- more than do duets of Trachyphonusdarnau- ture from thoseof the Black-collaredand Black- dii usambiro from those of T. d. bohmi and T. billed barbets (Fig. 1). Lybius rubrifacies(Ta- d. darnaudii. Preliminary playback experi- ble 3) also has a long pre-duet, almost as long ments show at least approach responsesand as its duet, in addition to a long post-duet that unison-calling of senex to the duets of nomi- is wholly lacking in the others. Lybius torqua- nate leucocephalus. tus has a short pre-duet and its duet is short with few notes. Lybius gutfiobalito has an even DISCUSSION shorter pre-duet (resembling that of torquatus) Undifferentiated polyphonic, unison-sung and its duet is longer, faster, and containsmore duets and duet roles intuitively seem lesscom- notes than do those of the others. Although plex-requiring less coordination of the call- the duet notes of torquatusand guifobalito are er-than the precise,antiphonal duets, and co- similar, the low-pitched duet note (type A note, ordinated polyphonic duetswith different songs Short and Horne 1982) of torquatusis double and roles of the duetting barbets. Long asso- (versus single in guifsobalito),and the notes of ciation of secondarybirds with a primary duet- both guifobalito duet parts have distinct, noisy ting pair provides an intimate arrangement for pre-note elements generally lacking in the the “fine-tuning” of the duet songsthat they clearer songsof torquatus (Fig. 1). will eventually sing. Much remains to be Effects of interspecific interactions on the learned about the experiential overlay of the duetting of barbetsare likely. Strongly duetting genetic background for duetting. Certainly we speciesof Lybius are largely or entirely allo- have heard no resemblancesof particular duets patric with one another, but often are broadly of one speciesto those of sympatric congeners, sympatric with: (1) non-duetters (e.g., L. tor- as of T. darnaudii and T. erythrocephalus,so quatuswith L. melanopterus,and with L. mi- there are somewhat narrow limits of modifi- nor); (2) simultaneouspolyphonic duetters(e.g., ability. Data are needed regarding the sexual L. guifsobalitowith L. leucocephalus,although nature of the barbets’ duet roles, and the con- the latter often is ecologically isolated from its ditions under which the roles may be switched. congeners);or (3) rarely duetting species(e.g., Such “switching” is possiblein shrikes (Lani- L. guifsobalito and L. rubrifacies with L. bi- arius, Thorpe 1972). Elsewhere, however, we dentatus).Even though duetting L. rubrijkcies have stressed the difficulties of comparing and L. torquatusmeet in Rwanda, they are but duetting in so phylogenetically divergent and narrowly parapatric, their contacts being very behaviorally different birds as barbets and limited. Lybius torquatus there is at the ex- songbirds(Short and Horne 1982). treme northwestern point of its range, is re- However sexualthe duet roles may be, duet- BARBET SOCIALITY 331 ting in the speciesof Trachyphonus and LyJbius patric Trachyphonus darnaudii and T. eryth- that we have discussedalways involves but rocephalus have different duets, whereas para- two duet roles; the two may be distinctly dif- patric, closely related T. erythrocephalus and ferent or essentiallysimilar (in a few cases,see T. margaritatus have almost indistinguishable below). The precision with which duet sets fit duets. Lybius torquatus meets and interacts together, especially in such speciesas T. dar- stronglywith L. rubrtjacies; thesespecies differ naudii (Short and Horne 1980a; in press:Fig. markedly in duets as well as in coloration. 2), suggests a strong requirement for the However, related L. guifobalito, which is al- matching of appropriate duet roles, however lopatric with both torquatus and rubrijkies, they may develop through the experience of has a duet resembling that of torquatus and the birds within a family or broader group. A coloration approaching that of rubrijacies. barbet leaving its group should have to either Other sympatric duetters with very divergent match its song with that of a potential mate, duets are congeneric L. guifsobalito with L. or find a potential mate who can match its duet bidentatus and unrelated Tricholaema lach- songappropriately, with or without some prac- rymosa, and Lybius vieilloti with Trachypho- tice. Further, not only must the duet songs nus margaritatus. In a senseeven more diver- match, but the intricate regulation and syn- gent in “songs” are sympatric L. torquatus, a chronization of the songs, shown for L. tor- duetter, with L. melanopterus, a non-duetter, qua&s (Short and Horne 1982:36-37; also and Stactolaema olivacea, a duetter, with non- Greenewalt, Horne and Short, unpubl. data), duetting S. leucotis. These examples show cir- and apparently important in the other coor- cumstantial evidence but correlation never- dinated duetters discussed, must be estab- thelessbetween the occurrenceand divergence lished and maintained. If these requirements of duetting and speciation. are not met, a barbet that leaves its group is We earlier cautioned against broad system- apt to be unsuccessfulin mating, and in hold- atic generalizing from data involving duetters ing a territory. Lone singersare uncommon to and non-duetters becauseof ecologicalparam- rare in duetting barbets (unpubl. data), and eters (forest or non-forest habits) that seem appear to represent attempts of subordinate related to duetting. Some problems can be birds to find a mate. Thus, there would seem overcome by comparisonsof congeneric,sym- to be severeconstraints placed upon the degree patric (hence under similar ecologicalregimes) of divergence in duet songswithin the species, species.The allopatric occurrence of the taxa although the exact nature of such constraints having similar duets, and the sympatric oc- and the magnitude of the allowable divergence currenceof congenershaving very diverse duets that does not preclude mixed matings remain (or no duets, e.g., L. melanopterus with com- to be established. plex duetting L. torquatus) strongly support As to how distinctive the duets must be in the view that speciatingtaxa (megasubspecies order to prevent interbreeding of speciating to allospecies) undergo major shifts in their taxa, we need more information on the de- duetting during late stagesof speciation (the velopment of duets, for there is some variation stages involving reinforcement of isolating in notesof the two duet songsin barbetshaving mechanisms, competitive interactions and distinct song-forms.The notes of the two songs range overlap). Enhancing this view is a com- composingthe duet resemble one another very parison of speciesfor which duetting data are closelyin some duetsof Lybius torquatus (Short available, usingnon-duetting Gymnobucco and and Horne 1982: Plate 5d), and of Trachy- Pogoniulus and duetting Lybius and Trachy- phonus darnaudii usambiro (Short and Horne, phonus (Table 2). Gymnobucco showsno me- in press:Fig. 6). The partner’s songsin duets gasubspecies,and two species form a super- of Trachyphonus erythrocephalus and T. mar- speciesamong its four species.Ony two of nine garita&s in each case also show close resem- Pogoniulus have megasubspecies.although five blance in form. If an adult bird could indeed speciesseem involved in two superspecies.Two alter its duet role, it might more readily accept of the six duetters, but not the non-duetter a duet partner singing a song that was some- (melanopterus) of Lybius have megasubspe- what to markedly different. The development ties, and two of five (duetting) Trachyphonus and modifiability of visual display elements of show megasubspecies.The megasubspecies duetting also have to be considered. with divergent songs(duets) occur only in the The comparison of museum specimens duetters among Lybius and Trachyphonus. (Short, unpubl. data) suggeststhat Lybius leu- Investigations are needed: (1) of the duets cocephalus albicaudus and L. I. senex inter- of populations or individuals morphologically breed over much of southernKenya and north- intermediate between vocally and morpholog- ern Tanzania, despite the distinctive unison ically distinct taxa such as Lybius leucoceph- calling of each; neither form singsduets. Sym- alus leucocephalus and L. 1. senex, and Tra- 332 LESTER L. SHORT AND JENNIFER F. M. HORNE chyphonus darnaudii bohmi and T. d. American birds. A contribution to comparative sys- usambiro; and (2) of the duets of the vocally tematics. Nuttall Ornithol. Club Publ. 9. PAYNE,R. B. 1971. Duetting and chorus singing in Af- similar allospecies T. margaritatus and T. rican birds. Ostrich, Suppl. 9: 125-l 46. erythrocephalusfrom the regions of Ethiopia PAYNE, R. B., AND N. J. SKINNER. 1970. Temporal pat- and Somalia where they are likely to meet. terns of duetting in African barbets. Ibis 112:173- The more complex antiphonal and coordi- 183. nated polyphonic singingduetters with strong SCHMIDL,D. 1982. The birds of the SerengetiNational Park Tanzania. British Ornithologists’ Union Check- behavioral correlates of the duetting (suppres- list 5. sion of singingof subordinategroup members) SHORT,L. L. 1969. Taxonomic aspectsof avian hybrid- seem to have diverged vocally when historical ization. Auk 86:84-105. factors fostered geographic isolation and thus SHORT,L. L. 1973. Hybridization, taxonomy and avian evolution. Ann. Mb. Bot. Card. 591447-453. morphological differentiation. Sympatric, re- SHORT.L. L. 1982. Woodpeckers of the world. Mono- lated duetters appear to have mutually influ- graph 4. Delaware MLseum of Natural History, enced the nature of each other’s duets and the Greenville, DE. course of their evolution (i.e., of their ecology SHORT.L. L.. AND J. F. M. HORNE. 1980a. Ground bar- and morphology aswell astheir vocalizations), bets of East Africa. Living Bird 18:179-l 86. SHORT, L. L., AND J. F. M. HORNE. 1980b. Vocal and as in the likely caseof Trachyphonusdarnaudii other behaviourofthe Green Barbetin Kenva. Ostrich and T. erythrocephalus.These effects seem to 51:219-229. be greatestduring the later stagesof speciation. SHORT,L. L., AND J. F. M. HORNE. 1982. Vocal and Barbets of duetting groups seem either to spe- other behaviourof Kenvan Black-collaredBarbets Lv- bius torquatus. Ibis 124127-43. ciate fully and rapidly, or to remain differen- SHORT,L. L., AND J. F. M. HORNE. In press. Aspects of tiated at the level of megasubspecies.Further duetting in ground barbets. Proc. V Pan-African Or- data on additional speciesshould lead to firm- nithological Congress. er conclusionsabout the relationship between THORPE,R. H. 1972. Duetting and antiphonal song in duetting and speciation in the African barbets. birds. Behaviour Suppl. 18. THORPE,R. H. 1973. Duet-singing birds. Sci. Am. 229: 70-79. ACKNOWLEDGMENTS TYROLLER,G. 1974. Duett-Entwicklung bei handauf- The Leonard C. Sanford Fund and the Gerald and May gezogenenBartvijgeln (Trachyphonus darnaudii em- Ellen Ritter-Eugene Eisenmann Fund of the American &i).% Tierpsychol. 35:102-i07. Museum of Natural History supportedShort ’s field work. WHITE. C. M. N. 1965. A revised check list of African G. R. Cunningham van Someren was helpful in Kenya, non-passerinebirds. Govt. Printer, Lusaka, Zambia. and J.-P. Vande wegheassisted us in Rwanda. This review WICKLER,W. 1973. Artunterschiede im Duettgesang is based upon a symposiumlecture presented at the XVIII zwischen Trachyphonus darnaudii usambiro und den International Ornithological Congress.We are grateful to anderen Unterarten von T. darnaudii. J. Ornithol. J. David Ligon and Robert Payne for suggestionsbenefit- 114:123-128. ting the manuscript, and to C. Chappuis for use of his WICKLER,W. 1976. Duetting song in birds: biological recordings. significanceof stationaryand non-stationary process- es. J. Theor. Biol. 61:493-497. LITERATURE CITED WICKLER,W., AND D. UHRIG. 1969. Bettelrufe, Ant- wortszeit und Rassenunterschiedeim Begriissungs- AMADON,D., AND L. L. SHORT. 1976. Treatment of sub- duett des SchmuckbartvogelsTrachyphonus darnau- speciesapproaching species status. Syst. Zoo1 25: 16l- dii. 2. Tierpsychol. 26165l-66 1. 167. FERNALD,R. D. 1973. A group of Barbets. II. Quanti- tative measures.Z. Tierpsychol. 33:341-35 1. American Museum of Natural History, New York, New KALAS, K. 1973. A group of Barbets.I. Some ethological York 10024. Address of second author: P.O. Box 24622, observations.Z. Tierpsychol. 33:335-340. Karen, Nairobi, Kenya. Received 17 September 1982. Fi- MAYR, E., AND L. L. SHORT. 1970. Speciestaxa of North nal acceptance31 March 1983.