Novitatesamerican MUSEUM PUBLISHED by the AMERICAN MUSEUM of NATURAL HISTORY CENTRAL PARK WEST at 79TH STREET, NEW YORK, N.Y

NovitatesAMERICAN MUSEUM PUBLISHED BY THE AMERICAN MUSEUM OF NATURAL HISTORY CENTRAL PARK WEST AT 79TH STREET, NEW YORK, N.Y. 10024 Number 3043, 65 pp., 26 figures, 2 tables May 7, 1992

Syringeal Morphology, Phylogeny, and Evolution of the Neotropical Manakins (Aves: Pipridae)

RICHARD 0. PRUM1

CONTENTS Abstract ...... 2 Resumen ...... 3 Introduction ...... 3 Acknowledgments ...... 5 Historical Review of Piprid Classification ...... 7 Methods ...... 8 Syringeal Morphology and Terminology ...... 8 Supporting Elements ...... 8 Membranes ...... 9 Musculature ...... 10 Innervation ...... 11 Syringeal Morphology of the Piprids ...... 12 Corapipo ...... 12 Masius ...... 13 Ilicura ...... 14 Manacus ...... 15 Chiroxiphia ...... 16 Antilophia ...... 18 Machaeropterus ...... 19 Chloropipo ...... 21 Xenopipo ...... 23 Heterocercus ...... 24 Pipra ...... 24 Syringeal Character Analysis ...... 3 1 A Elements (1-19) ...... 31 B Elements (20-39) ...... 35

1 Chapman Fellow, Department ofOrnithology, American Museum ofNatural History. Present Address: Museum of Natural History and Department of Systematics and Ecology, University of Kansas, Lawrence, KS 66045.

Accessory Cartilages (40-42) ...... 37 Syringeal Shape (43, 44) ...... 37 Extrinsic Musculature (45-47) ...... 37 Intrinsic Musculature (48-56) ...... 38 Innervation (57, 58) ...... 41 Membranes (59) ...... 41 Phylogenetic Analysis ...... 41 Alternative Phylogenetic Hypotheses for the Piprids ...... 44 Syringeal Variation, Function, and Evolution ...... 48 Previous Descriptions of Piprid Syringes ...... 48 Intraspecific Variation ...... 49 Evolution of Interspecific Variation ...... 51 Functional Consequences of Syringeal Variation ...... 52 The Two-Voice Model and the Suboscine Syrinx ...... 54 Vocal Learning, Lek Behavior, and Syringeal Evolution ...... 56 Phylogenetic Classification of the Piprids ...... 57 References ...... 60 Appendix ...... 64

ABSTRACT The syringeal morphology of37 ofthe 40 species Intraspecific and interspecific variations in pip- ofthe monophyletic manakin family (Aves: Pipri- rid syringeal morphology are reviewed, and pos- dae) is described. Observations were made of sible correlations between syringeal function and cleared and double stained, and iodine stained structural variation are discussed. Piprids appar- specimens. Twenty-five distinct syringeal mor- ently lack independent innervation ofthe two sides phologies were identified within the family, en- of the syrinx, a hypothetical prerequisite for con- compassing variation in supporting elements, trol of two independent syringeal sound sources. membranes, musculature, and innervation. This A general model is proposed to explain the dif- extensive interspecific variation indicates that the ferences in the diversity of syringeal morphology syringeal evolution in the piprids has been dynam- in the suboscine and oscine passerines. In oscines, ic. songs are largely learned, whereas in suboscines, A phylogenetic analysis ofpiprid syringeal mor- available evidence indicates that vocalizations are phology is presented. Syringeal variation was cod- innate. Natural or sexual selection on innate vo- ed as 59 characters and polarized by outgroup calizations should contribute to correlated mor- comparison to cotingas (Cotingidae), tyrant fly- phological differentiation, whereas selection on catchers (Tyrannidae), and other suboscine pas- learned vocalizations should result in cultural evo- serines. Parsimony analysis using PAUP identified lution without morphological change. The predic- two equal length, maximally parsimonious phy- tions of the model are congruent with the differ- logenetic hypotheses (length = 76; Consistency In- ences in syringeal diversity of these two main dex = 0.82). A strict consensus tree based on these clades. output trees includes 23 resolved clades, and is A phylogenetic classification of the manakins identical to one of the two shortest trees. The that reflects the syringeal hypothesis ofphylogeny monophyly of eight piprid genera and several re- is presented. Four new tribes are recognized within cently recognized species groups is supported. the family. The four species of the paraphyletic However, the traditional piprid genera Chloropipo genus Chloropipo are combined with Xenopipo and Pipra are paraphyletic and polyphyletic, re- atronitens in a single monophyletic genus to which spectively. The monophyly of Heterocercus and the senior genus group name Xenopipo is applied. the Pipra serena species group could not be sup- The polyphyletic genus Pipra is split into three ported by syringeal characters. monophyletic genera: Pipra Linnaeus including the The syringeal hypothesis of phylogeny is com- aureola and erythrocephala clades; Dixiphia Rei- pared to previous classifications ofthe piprids and chenbach including pipra; and Lepidothrix Bona- to recent biochemical hypotheses of phylogeny. parte including the eight species ofthe serena spe- The relative confidence of the clades in the syrin- cies group. geal hypothesis is discussed. 1992 PRUM: MANAKIN SYRINGES AND PHYLOGENY 3

RESUMEN Se describe la morfologia de las siringes de 37 cifica de la morfologia de la siringe de los pipridos, de las 40 especies de todos los 11 generos de la y se discuten las correlaciones posibles entre la familia de saltarines (Aves: Pipridae). Se obser- funcion de la siringe y la variacion de la estructura. varon ejemplares aclarado y teiiidos, y teniidos de Los pipridos aparentemente carecen de la iner- yodo. Venticinco morfologias de siringe distintas vacion independente de los dos lados de la siringe, se identifican en la familia, abarcando variaciones que es un requisito hipotetico para controlar dos en los elementos de apoyo, las membranas, los origenes independentes del sonido en la siringe. musculos, y la innervacion. Esta variacion inter- Se propone un modelo general para explicar las especifica extensiva indica que la evolucion de las diferencias en la diversidad de la siringe de los siringes en los saltarines ha sido dinamica. paserinos suboscinos y oscinos. En los oscinos, las Se presenta un analisis filogenetico de la mor- vocalizaciones son generalmente aprendidas, fologia de la siringe de los saltarines. La variacion mientras que en los suboscinos, la evidencia dis- de las siringes fue registrada como 59 caracteres y ponible indica que las vocalizaciones son innatas. polarizada por compariciones con los cotingas La selecion natural y sexual de las vocalizaciones (Cotingidae), atrapamoscas (Tyrannidae), y otros innatas deben contribuir a la diferenciacion cor- suboscinos. Un analisis parsimomnco usando PAUP relacionada, mientras que la selecion sobre vo- identifico dos hipoteses filogeneticas de igual lon- calizaciones apprendidas debe producir la evoluci- gitud y de maxima parsimonia (longitud = 76; on cultural sin cambio morfologico. Estas Indice de consistencia = 0.82). Un arbol de con- predicciones del modelo son congruentes con las senso exacto basado en estos arboles incluye 23 differencias en la diversidad de la siringe de esto clados resueltos y es identico a uno de los dos dos clados mayores. arboles mas corto. Se apoyan la monofilia de ocho Se propone una clasificacion filogenetica de la generos de piprides y algunos de los grupos de familia de los saltarines, que representa la hipo- especies reconocidos recientemente. Los dos ge- tesis filogenetico de la siringe. Se reconocen cuatro neros tradicionales de pipridos, Chloropipo y Pip- tribus nuevas en la familia. Las cuatros especies ra, son parafiletico o polifiletico, respectivamente. del genero parafil6tico Chloropipo se combinan con La monofilia de Heterocercus y el grupo de espe- Xenopipo atronitens en un genero monofiletico que cies Pipra serena no pueda apoyarse por caracteres tiene el nombre mayor Xenopipo. El genero poli- de la siringe. filetico de Pipra se divide en tres generos monofile- La hipotesis filogenetico de la siringe se compara ticos: Pipra Linnaeus incluyendo los grupos de es- con las clasificaciones previas de los saltarines y pecies aureola y erythrocephala; Dixiphia Rei- con hipoteses filogen6ticos de caracteres bioquimi- chenbach incluyendo pipra; and Lepidothrix cos. Se discute la confianza relativa de los clados Bonaparte incluyendo las ocho especies del grupo de la hipotesis filogenetico de las siringes. de especies serena. Se revisa la variacion intraespecifica y interespe-

INTRODUCTION The manakins (Pipridae) are a small family investigation ofpiprid syringeal morphology of Neotropical passerine birds. They are was initiated to provide a phylogenetic members of the suboscine superfamily Ty- framework for comparative investigations of rannoidea that also includes the cotingas (Co- the evolution of lek display behavior of the tingidae) and tyrant flycatchers (Tyrannidae) family (Prum, 1990a). (Snow, 1975, 1979; Prum, 1990b). The man- The syrinx-the avian vocal organ-is akins are of particular interest because most composed of specialized skeletal supporting behaviorally known species breed in polygy- elements, membranes, and musculature ofthe nous, lek/arena systems. Many species are lower respiratory tract. Syringeal morphol- strikingly sexually dimorphic in plumage and ogy of passerines has been traditionally an perform acrobatic courtship displays (Snow, important character system in the recogni- 1963; Sick, 1967; Prum, 1990a). Traditional tion ofthe major division ofpasserine birds- analyses of the evolution of piprid display the oscines and suboscines (Muller, 1847, behavior have been limited by the lack of a 1878; Wunderlich, 1886; Haecker, 1900; Set- corroborated hypothesis ofphylogeny for the terwall, 1901; Koditz, 1925). However, the group (Sick, 1959, 1967; Snow, 1963). This diversity of suboscine syringes was not well 4 AMERICAN MUSEUM NOVITATES NO. 3043 understood until Ames (1971) provided a tory (FMNH); University ofKansas Museum comprehensive overview of their syringeal of Natural History (KU); Louisiana State morphology. Ames documented an enor- University Museum of Zoology (LSUMZ); mous syringeal diversity in the Tyrannoidea, Museu Paraense Emilio Goeldi (MPEG); including a small sample of piprids. More Museum of Vertebrate Zoology, University recently, syringeal characters have been used of California at Berkeley (MVZ); Museum of in phylogenetic analyses at both higher and Zoology of the University of Sao Paulo lower levels within tyrannoids (Lanyon, (MZUSP); Royal Ontario Museum (ROM); 1984a, 1984b, 1985, 1986, 1988a, 1988b, University of Michigan Museum of Zoology 1988c; McKitrick, 1985; Prum and Lanyon, (UMMZ); National Museum ofNatural His- 1989; Lanyon and Lanyon, 1989; Prum, tory (USNM); Yale Peabody Museum of 1990b). Natural History (YPM); and Zoologia, Uni- Traditionally, the Pipridae included 55 versidade do Brasilia (ZUB). Specimens re- species in 17 genera (Hellmayr, 1929; Snow, ferred to in the figure captions and the text 1975, 1979). The borders of the three large are identified by the abbreviations given tyrannoid families have long been regarded above. A complete list of specimens of pi- as artificial and unsatisfactory. A recent phy- prids examined is given in the Appendix. logenetic analysis of higher-level relationships in the Tyrannoidea based on morphology indicates that the traditional Pipridae ACKNOWLEDGMENTS is not monophyletic (Prum, 1990b). Six tra- This research was conducted as part ofmy ditional piprid genera, Schiffornis, Piprites, doctoral thesis at the University ofMichigan. Sapayoa, Neopelma, Tyranneutes, and Neo- I would like to thank the members of my pipo, share derived syringeal characters with doctoral committee: Mary C. McKitrick, other nonpiprid tyrannoids (Prum and Lan- Robert B. Payne, William L. Fink, Brian A. yon, 1989; Prum, 1990b). However, a unique Hazlett, and W. Herb Wagner for their help- derived syringeal character supports a clade ful comments during the research and on the including 40 species in the other 11 tradi- manuscript. I would especially like to thank tional piprid genera, Pipra, Manacus, Ma- Wesley E. Lanyon, Emeritus Lamont Cura- chaeropterus, Chiroxiphia, Antilophia, Chlo- tor, Department of Ornithology, American ropipo, Xenopipo, Heterocercus, Corapipo, Museum of Natural History for introducing Ilicura, and Masius (Prum, 1990b). This re- me to suboscine syringes, and for allowing stricted Pipridae is the subject of this inves- me to study cleared and stained syringeal tigation. specimens he prepared as part ofhis research In this paper, I report my observations of on tyrannids. His instruction and encourage- the syringeal morphology ofall available pip- ment made this research possible. George rid species and perform a phylogenetic analy- Barrowclough, Steve Goodman, Geoff Hill, sis of the family based on syringeal charac- Rebecca Irwin, and Michael Nachman also ters. I then discuss the implications of these gave helpful advice and stimulating discus- findings for models ofsyringeal function and sions on the research. Peter Ames gave en- evolution, and for the phylogeny and classi- couraging suggestions early in the project. W. fication of the piprids. I have used these sy- E. Lanyon and R. L. Zusi provided useful ringeal data and the syringeal hypothesis of comments on the final version of the manu- phylogeny for the piprids elsewhere in a de- script. The Resumen was proofread by Pa- tailed analysis ofthe evolution ofpolygynous tricia Escalante Pliego. Syringeal illustrations display behavior in the family (Prum, 1 990a). were prepared by Susan D. Kielb (figs. 2 left, The investigation is based on anatomical 3,6,7,9, 10, 11, 13, 14, 15, 16, 18 left, 19, specimens borrowed from the American Mu- 20, 22, and 23) and David A. Rini (figs. 2 seum of Natural History (AMNH); British right, 4, 5, 8, 12, 17, 18 right, and 21). Museum (Natural History) (BMNH); Car- I acknowledge the following curators and negie Museum of Natural History (CMNH); colleagues for lending me specimens in their Coleccion Ornitologia Phelps, Caracas, Ven- collections: J. C. Barlow, G. F. Barrowclough, ezuela (COP); Field Museum ofNatural His- P. J. K. Burton, R. Cavalcanti, G. Cowles, J. 1 992 PRUM: MANAKIN SYRINGES AND PHYLOGENY 5

W. Fitzpatrick, N. K. Johnson, R. F. John- ochloris (= Laniisoma), Heteropelma (= ston, S. M. Lanyon, W. E. Lanyon, M. Len- Schiffornis turdinus), Schiffornis, Neopelma, tino, M. A. Marini, M. C. McKitrick, F. C. and Heterocercus. Sclater used male plumage Novaes, D. C. Oren, J. V. Remsen, E. H. coloration and tail and bill shape to distin- Stickney, D. F. Stotz, P. E. Vanzolini, D. S. guish these genera. Wood, D. E. Willard, and R. L. Zusi. In his first revision of Pipra, Hellmayr The research was funded by a Research (1906) defined the genus and species by Partnership Grant with M. C. McKitrick, a plumage traits. He removed Corapipo from Dissertation Improvement Grant, and a one- Pipra, but left the two species of Tyranneutes term Predoctoral Fellowship from the Rack- in the genus. ham Graduate School of the University of Ridgway (1907) removed some species Michigan. I received additional support for from larger genera and recognized Corapipo, research and fieldwork from the University Tyranneutes, Allocotopterus (to include Ma- ofMichigan Department ofBiology; the Her- chaeropterus deliciosus), and Chiriprion (to bert and Betty Carnes Award ofthe American include Chiroxiphia pareola, C. lanceolata, Ornithologists' Union; the Sigma-Xi Foun- and C. linearis). He also identified senior syn- dation; and a Graduate Student Research in onyms for several genera recognized by Scla- the Collections Grant from the American ter, changing Metopia to Antilophia, Helicura Museum ofNatural History sponsored by W. to Ilicura, and Chiromachaeris to Manacus. E. Lanyon. During the final preparation of Hellmayr (1910) provided detailed diag- the manuscript, I received additional support noses of the genera of Pipridae, and placed as a Chapman Fellow from the Frank M. Chiroprion back in Chiroxiphia. He suggest- Chapman Memorial Fund ofthe Department ed that Masius and Antilophia, and Corapipo of Ornithology, American Museum of Natu- and Manacus might be each other's closest ral History. relatives, based on male plumage similarities. To all these people and institutions, I give Later, Hellmayr (1929) presented a formal my sincere thanks. classification ofall species ofpiprids in which he recognized the senior name Teleonema for Cirrhipipra. HISTORICAL REVIEW OF These early authors did not explicitly jus- PIPRID CLASSIFICATION tify the order of taxa in their classification. I have previously reviewed the history of In Sclater's and Ridgway's classifications, the familial limits of the Pipridae (Prum, taxonomic order corresponds to the position 1990b). Here, I summarize previous classi- in the diagnostic keys, which were apparently fications with an emphasis on hypotheses of constructed for convenience. However, com- interrelationship for the members of the 11 ments in the text or footnotes of all three genera in the restricted monophyletic Pipri- authors suggest that proximity of taxa was dae (Prum, 1990b): Chloropipo, Xenopipo, also meant to reflect evolutionary relation- Antilophia, Heterocercus, Machaeropterus, ship. Manacus, Corapipo, Ilicura, Masius, Chi- Sibley (1957) remarked that the piprid gen- roxiphia, and Pipra. era were based only on male plumage from Sclater (1888) provided the first complete Hellmayr's (1910) diagnosis, and that a ge- classification and keys of species and genera neric revision of the family based on other of the Pipridae. He divided the family into characters was necessary. But in the most re- two subfamilies: the Piprinae including the cent classification of the Pipridae, Snow 14 genera Piprites, Chloropipo, Xenopipo, (1975, 1979) followed Hellmayr's generic Ceratopipra (= Pipra cornuta), Cirrhipipra (= limits, with only two minor exceptions, be- Pipra filicauda), Metopia (= Antilophia), cause ofthe lack of additional systematic in- Masius, Metopothrix, Pipra (including Cora- vestigation on the group. Snow returned Al- pipo and Tyranneutes), Neopipo, Machaerop- locotopterus deliciosus to Machaeropterus, and terus, Chiroxiphia, Helicura (= Ilicura), and Teleonemafilicauda to Pipra. He also placed Chiromachaeris (= Manacus); and the Ptilo- the genera ofpiprids in a sequence to proceed chlorinae which included the five genera Ptil- from "least" to "most" specialized in breed- 6 AMERICAN MUSEUM NOVITATES NO. 3043 ing behavior and sexual dimorphism, stress- aureola, erythrocephala, and serena species ing the lack of information for many genera. groups identified by Haffer (1970, 1974), ex- Snow (1975) made brief comments on inter- cept that he excluded cornuta from the Pipra relationships among piprid genera. He placed erythrocephala group. Snow placed the Cen- Chloropipo and Xenopipo near the beginning tral American and Pacific South American of the sequence as unspecialized manakins forms ofManacus as subspecies ofManacus with other problematic genera that were sub- manacus, also following Haffer's (1967) rec- sequently removed from the family. He sug- ommendation. Elsewhere in the family, Snow gested that the plumage of Antilophia indi- also recognized species groups that included cates affinities with Pipra, but placed it near the members of Corapipo, Heterocercus, and the beginning of the sequence because of its Chiroxiphia, and Machaeropterus regulus and apparent lack of elaborate displays (Sick, pyrocephalus. In Chloropipo, Snow recog- 1959, 1967). Snow considered Heterocercus, nized a species group including flavicapilla, Ilicura, and Masius to be isolated genera holochlora, and uniformis. He questioned without obvious affinities to other manakins, whether C. unicolor was related to other but apparently based on plumage, he placed Chloropipo species because of differences in the latter two genera late in the sequence of overall plumage and external morphology, but typical or "specialized" piprids. Based on de- in the absence ofany behavioral information, tails ofbehavior and plumage, Snow also sug- he hesitated to erect a new monotypic genus gested that Machaeropterus and Corapipo for unicolor. might be closely related to Manacus, and that S. M. Lanyon (1985) analyzed allozyme Chiroxiphia is closest to Pipra. variation in the tyrannoids including seven Hypotheses of interspecific relationships species ofpiprids. In a variety ofdistance and within manakin genera have been presented cladistic analyses of the data with Sapayoa recently in the form ofproposed superspecies as the root or outgroup, the allozyme data or species groups. Haffer (1967, 1970, 1974) supported two main piprid clades: one in- identified several species groups of piprids, cluding Machaeropterus regulus, Manacus based on plumage similarity and patterns of manacus, and Pipra pipra, and another in- geographic distribution (allopatry or para- cluding Chiroxiphia pareola, Masius chry- patry). He (1967) recommended that all forms sopterus, Corapipo leucorrhoa altera, and of Manacus be placed as semispecies in the Chloropipo holochlora. He also placed Neo- single superspecies Manacus manacus, citing pelma and Tyranneutes as the sister group to his observations of limited hybridization be- either of these piprid clades. tween populations ofManacus in Colombia, Prum and Johnson (1987) presented a cla- and the similarity of the courtship displays distic analysis of male courtship display el- ofthe various geographic forms ofManacus. ements supporting a clade with Ilicura as the In the genus Pipra, Haffer (1970) recognized sister group to Masius and Corapipo. Prum three superspecies including all species ex- (1988) did a cladistic analysis of the Pipra cept Pipra pipra: the Pipra aureola species serena species group based on plumage char- group including aureola, fasciicauda, and fil- acters in an investigation of patterns of bio- icauda (previously placed in the monotypic geographic vicariance in Neotropical forest genus Teleonema); the Pipra erythrocephala birds. Cracraft (1988) hypothesized that Pip- species group including erythrocephala, ra cornuta is the sister group to Pipra erythro- rubrocapilla, chloromeros, mentalis, and cor- cephala and P. rubrocapilla, based on a cla- nuta; and the Pipra serena species group in- distic analysis of plumage characters. cluding serena, coronata, isidorei, coeruleo- Sibley and Ahlquist (1985, 1990) presented capilla, nattereri, vilasboasi, and iris. Haffer a phylogeny ofthe tyrannoids based on DNA- (1970, 1974) hypothesized that the geograph- DNA hybridization distances including some ic patterns of species replacement in these piprid taxa, but the hypothesis was not highly groups were the result of primary isolation resolved, included no reciprocal hybridiza- in forest refugia and secondary expansion tions, and provided little information about during climatic cycles in the Pleistocene. piprid phylogeny. Sibley and Monroe (1990) Snow (1975, 1979) recognized the Pipra presented a classification of the piprids in a 1 992 PRUM: MANAKIN SYRINGES AND PHYLOGENY 7 subfamily Piprinae of an expanded Tyran- 18 right, and 21, no stippling and small stip- nidae. Besides the change in the family-level pling are used to portray ossified and carti- status ofthe group, this classification was not laginous elements. based on any DNA-DNA hybridization ev- Monophyly of the piprids was accepted idence. Sibley and Monroe reversed the order based on their possession of dorsally fused of Snow (1979) to conform to Hellmayr B 1-2 syringeal elements (Prum, 1990b). The (1929), but accepted Snow's changes of ge- hypothesized sister group to the piprids is the neric limits. Sibley and Monroe (1990) recog- cotingid clade, and their sister group is the nized the Pipra aureola species group ofHaf- tyrannids (Prum, 1990b). For this investi- fer (1974) and Snow (1975, 1979), but they gation the problematic tyrannoids that may suggested new limits for the Pipra erythro- be related to either cotingids or tyrannids were cephala species group which included only assumed to belong to the cotingids (Prum, erythrocephala, rubrocapilla, and mentalis. 1990b). Further, they removed Pipra coeruleocapilla Morphological variation in syringes ofpip- and isidorei from the Pipra serena species rids was coded as 56 binary characters and group, and placed them in their own coeru- three unordered multistate characters and leocapilla group. At the specific level, they polarized by outgroup comparison to other split Manacus into four species (manacus, tyrannoids. Unknown character states were vitellinus, aurantiacus, and candei), and Co- coded as (?). Hypothesized transition series rapipo leucorrhoa into leucorrhoa and altera. were coded as a series of binary characters. Each character description proceeds from the METHODS distribution and description of the derived state to the distribution and description of Observations were made of 264 syringeal the primitive state, and ends with a statement specimens of 37 species in all 11 genera of of the character polarity. Complex character piprids using a Wild M5A dissecting micro- descriptions include additional discussion. scope. Sixty-nine specimens were cleared and The distributions of hypothesized derived double stained with alcian blue and alizarin characters are summarized in table 2. crimson to distinguish cartilage and bone Throughout the paper, specific characters are (Dingerkus and Uhler, 1977) by Dr. W. E. referred to by the character number in paren- Lanyon ofthe American Museum ofNatural theses. Alternative derived states ofthe three History. These specimens exhibited the complex, unordered characters (13, 54, and shape, relative position, and composition of 57) are referred to by .1 or .2 following the the syringeal supporting elements. One-hun- character number. dred-ninety-five syringes from 33 species in The most parsimonious phylogenetic reso- 11 genera were removed from spirit speci- lutions of the syringeal character data were mens using the method of Ames (1971) and identified using the PAUP computer pro- Cannell (1988). Observations were made of gram, version 3 (Swofford, 1989), on a Mac- these uncleared specimens using reversible intosh computer using the branch-and-bound iodine stain of Bock and Shear (1972) to aid algorithm, condensed zero branch lengths, in the description of musculature and inner- and the consensus tree option on the unor- vation. Most specimens were sexed by ob- dered character set. For simplicity, the mor- servations of the gonads. A list of all cleared phologically identical taxa were combined and double-stained and iodine-stained piprid into single OTUs in the PAUP analysis. Some syringeal specimens observed is presented in apparent autapomorphies in the results were the Appendix. used to support the monophyly of polytypic Illustrations of the syringeal material were terminal taxa. prepared using a Wild M5A dissecting scope Throughout, I follow the taxonomy ofSnow and a camera lucida. In figures 2 left, 3, 6, 7, (1979) with a few exceptions. I recognize three 9, 10, 11, 13, 14, 15, 16, 18 left, 19, 20, 22, species in the genus Manacus: M. manacus, and 23, small and large stippling are used to M. vitellinus, and M. candei. Also, I recognize illustrate ossified and cartilaginous elements, the two differentiated allopatric geographic respectively. In figures 2 right, 4, 5, 8, 12, 17, forms formerly placed in Pipra serena as sep- 8 AMERICAN MUSEUM NOVITATES NO. 3043 arate species: Pipra serena and Pipra suav- leading to difficulties in identifying homolo- issima (Prum, in prep.). For convenience, I gous elements among species. A tracheal refer to several species groups of Pipra as drum, or tympanum, is not consistently pres- defined by Haffer (1970, 1974) and Snow ent in suboscine species, and cannot be used (1975): the Pipra serena species group in- as a landmark in assessing homology. cluding serena, suavissima, coronata, isidor- Since these two traditional systems for ei, coeruleocapilla, nattereri, vilasboasi, and naming syringeal supporting elements are in- iris; the Pipra aureola species group including appropriate for suboscines, Ames (1971) de- aureola, fasciicauda, and filicauda; and the vised an alternative system based on the ob- Pipra erythrocephala species group including servation that tracheal and bronchial cornuta, mentalis, chloromeros, erythroceph- supporting elements typically differ in com- ala, and rubrocapilla. position, i.e., the degree of ossification. He proposed naming the cranial, mostly trache- SYRINGEAL MORPHOLOGY AND al, series of supporting rings "A elements," TERMINOLOGY and the caudal, mostly bronchial series "B elements." In Ames's system, each ringlike The tracheobronchial syrinx of piprids is supporting element is numbered in sequence composed of four morphological compo- in the A or B series, beginning with the first nents: supporting elements, membranes, A and B elements near the tracheobronchial muscles, and nerves. Latin synonyms ofmor- junction, and continuing cranially and cau- phological structures are from Baumel et al. dally away from the tracheobronchial junc- (1979). Basic morphology and terminology tion. The difficulty of this system is in estab- of each of these character systems are out- lishing homology among the first supporting lined below. elements in the A and B series despite variation in morphology of elements and their SUPPORTING ELEMENTS relative position to the tracheobronchial The syringeal supporting elements include junction. Ames (1971) differentiated the first tracheal and bronchial rings, and other non- A and B elements by differences in cross- ringlike supporting structures called acces- sectional shape (flat vs. D-shaped), consis- sory cartilages. King (1979, 1989) has pro- tency (transparent and stiff vs. opaque, posed naming syringeal supporting rings spongy, and flexible), and orientation ofcon- based on their position relative to the bifur- cavity (caudal vs. cranial). The first two cri- cation of the airway. However, a simplistic teria are indirect assessments of the degree nomenclature based solely on the tracheal/ ofossification, and the last criterion is highly bronchial dichotomy proposed by King (1979, variable among taxa and may be very difficult 1989) results in the assignment of different to assess once the syrinx has been removed names to homologous elements within and from the body cavity. Further, I have ob- among species that vary in the relative po- served significant variation among piprids in sition of the tracheobronchial junction. Al- the degree of ossification of the first putative though such a nomenclature may be conve- A and B elements, making it problematic to nient in functional analyses, this system would apply Ames's criteria. be an impediment to genuine evolutionary In this investigation, I use the supporting and comparative investigations. element terminology proposed by Ames In oscines, the names of the ringlike sup- (1971), but I employ a different set ofcriteria porting elements are traditionally based on for establishing homology among the initial relative position to the tracheal tympanum A and B elements (in decreasing order ofim- or "drum" which is composed ofthree or four portance): (1) special similarities in shape, (2) fused, ossified tracheal rings immediately relative position of syringeal muscle inser- cranial to the tracheobronchial junction tions, and (3) composition (ossified vs. car- (Ames, 1971). This drum is uniformly pres- tilaginous). Within piprids, homologies are ent in oscine birds (Ames, 1971). However, most easily established by the relative posi- in suboscines the morphology of the sup- tion ofan element to the B1-2 elements, which porting elements is much more variable, are dorsally fused by a short cartilaginous bar 1992 PRUM: MANAKIN SYRINGES AND PHYLOGENY 9

(Prum, 1990b) and distinctly widened and fused or unfused to other elements dorsally angled at their ventral ends. In piprids, pu- or ventrally. tative homologs established by the first cri- In all tyrannids and the six genera of pip- terion are always consistent with hypotheses ridlike tyrannoids, the syrinx also includes ofhomology supported by the second criteri- non-ringlike accessory cartilages in the me- on-relative position to the insertion of the dial tympaniform membranes, referred to as intrinsic syringeal musculature. In only a few internal or medial cartilages (Lanyon, 1984a, critical instances, application ofthese criteria 1985, 1986, 1988a, 1988c; McKitrick, 1985; leads to different homology assignments than Prum, 1990b), but these structures are absent reported by Ames (1971) (discussed below). in all true piprid genera described here. Two At higher levels within Tyrannoidea, estab- novel accessory cartilages occur in the pip- lishment of supporting-element homologies rids, and are evolutionarily independent of is complicated by variation in the position those in tyrannids. The paired medial bron- of the insertions of syringeal musculature chial cartilage bars of Ilicura, Masius, and (such as the insertion of M. tracheolateralis Corapipo are dorsoventrally oriented at the on the A1/Bl membrane; Prum, 1990b). In cranial margin of the medial tympaniform these higher-level comparisons, it is neces- membrane. Second, Chiroxiphia and Antil- sary to invoke the third criterion of compo- lophia have a unique sheet ofcartilage which sition to identify putative homologs. forms the craniomedial walls of the bronchi In describing the variation in shape of the and the ventral surface of the trachea at the syringeal supporting elements, I use termi- tracheobronchial junction. Some individuals nology similar to that ofAmes (1971). In this in a number of piprid species have disklike scheme, supporting elements are character- accessory cartilages on the ventral surface of ized as: the syrinx at the tracheobronchial junction which are irregularly distributed among and complete-element forms a closed ring, within species. They appear to be anomalies incomplete-element "open" dorsally, ven- associated with the development ofthe trach- trally, or medially in the form of a partial eobronchial junction or the ventral insertion ring, ofthe syringeal muscles. They were observed single-element composed of only a single irregularly in specimens of several species: ring, Machaeropterus regulus (4), Chloropipo un- double-element divided into a pair of left iformis (2), Xenopipo atronitens (I), Pipra and right elements which may themselves suavissima (1), Pipra mentalis (2), Pipra chlo- be incomplete or complete. romeros (1), and Pipra erythrocephala (1). Ex- amples are illustrated in figures 14, 15, and Ames (1971) referred to incomplete double 24, but they are not mentioned further in the elements as "divided" and complete double taxonomic descriptions below. elements as "double." In addition, supporting elements may vary in composition (either cartilaginous or ossified) and in the extent of MEMBRANES ossification (either dorsally, ventrally, or me- As the trachea passes caudally into the tho- dially, in double bronchial elements). Lastly, racic cavity, it enters the interclavicular air elements may be fused to one another en- sac (Saccus clavicularis) which surrounds the tirely, along their dorsal or ventral margins, syrinx. The membranes of the syrinx are the or by a bar of cartilage or bone. connective tissue that comprises the "walls" The pessulus is a dorsoventrally oriented of the respiratory tract and separate it from bar ofcartilage or bone which forms the cau- the interclavicular air sac. The most promi- domedial margin of the tracheobronchial nent syringeal membranes in most passerines junction. Although the pessulus is probably are the medial or internal tympaniform not a single homologous element in all sub- membranes (Membrana medialis). They form oscine birds, it is a ubiquitous accessory sy- the craniomedial surfaces ofthe bronchi and ringeal supporting element in piprids. The are generally considered to be the sources of pessulus may be cartilaginous or ossified, and sound generation in the tracheobronchial 10 AMERICAN MUSEUM NOVITATES NO. 3043 passerine syrinx (Greenewalt, 1968; Stein, insert on syringeal supporting elements, and 1968; Gaunt, 1983; Gaunt and Gaunt, 1985; intrinsic muscles that originate and insert on Brackenbury, 1989; Suthers, 1990). A pair of the syringeal supporting elements (e.g., Ames, membranes on the lateral surfaces between 1971). These categories, however, are fre- the supporting elements in some oscines and quently artificial since few syringeal muscles some nonpasserines are sometimes called the classified as "intrinsic" are composed exclu- external or lateral tympaniform membranes sively of fibers originating on syringeal sup- (Membrana lateralis). They have been hy- porting elements. pothesized to be important in sound produc- Alternatively, muscles may also be classi- tion in oscines (Chamberlain et al., 1968), fied as either tracheal or syringeal, depending but none of the lateral interannular mem- on their origins, and each ofthese classes can branes in piprids is drumlike or "tympani- then be distinguished into extrinsic and in- form," so they are not likely to be sound trinsic subgroups, depending on the site of sources during vocalization. In oscines, the their insertion (King, 1989). This classifica- lateral tympaniform membrane is located be- tion produces needless additional elaboration tween Al and 2, and not between Al and B 1, of essentially arbitrary distinctions by re- as indicated by Ames (1971: 88). quiring recognition ofthe limits ofthe syrinx In many birds, the medial walls of the relative to musculature origins and inser- bronchi are connected by a transverse mem- tions. Here, I use extrinsic and intrinsic in brane called the interbronchial ligament (lig- the former, traditional sense. mentum interbronchialis) or bronchidesmus. The two extrinsic syringeal muscles in pas- In piprids and most other tyrannoids, the in- serines are M. tracheolateralis and M. ster- terbronchial ligament is present as a thin sheet notrachealis. In piprids and other tyrannoids, of connective tissue between the caudoven- M. tracheolateralis originates on the cricoid tral surface ofthe medial tympaniform mem- cartilage of the larynx and continues caudad brane and the inner, dorsal surface of the on the lateral surfaces of the trachea. In the interclavicular air sac on either side of the absence of intrinsic syringeal muscles, M. esophagus. The syrinx is also connected to tracheolateralis inserts on the A1-2 elements the dorsal surface of the interclavicular air near the tracheobronchial junction. M. ster- sac by thin membranous strands. In all pip- notrachealis originates on the inner surface rids, a pair of membranous strands runs be- of the craniolateral process of the sternum tween the fused dorsal ends of the B 1-2 el- and inserts on M. tracheolateralis along the ements and the interclavicular air sac on either ventral or lateral surface of the trachea, be- side of the esophagus. In many specimens, tween the tracheobronchial junction and the these strands contain fibers from the vagus tracheal anastomosis with the interclavicular nerve (see Innervation), but they are also air sac. present in specimens that apparently lack The intrinsic syringeal muscles ofpasserine these vagus nerve fibers. In cotingas and oth- birds are composed ofthe differentiated cau- er tyrannoids, these strands are similar in po- dal fibers ofM. tracheolateralis (Ames, 1971). sition, but the B 1-2 elements are not fused. In piprids with intrinsic musculature, M. With one exception, I observed little varia- tracheolateralis inserts completely or par- tion in the syringeal membranes of piprids, tially on some A element(s) cranial to the so these structures are not included in the tracheobronchial junction and immediately syringeal descriptions below. However, a gives rise to a belly of partially independent novel patch offibrous tissue is present on the muscle fibers that continue caudad to insert medial tympaniform membrane in Pipra cor- on other, more caudal A elements. Many nuta, and it is described under that species. variations on this general plan occur in piprids, and in only a few instances are these MUSCULATURE muscles completely "intrinsic"-i.e., entirely independent of M. tracheolateralis fibers. Syringeal musculature has been tradition- Typically, the deep fibers of M. tracheolater- ally divided into two categories-extrinsic alis and the intrinsic muscles form insertions muscles that originate outside the syrinx and and origins on the A elements, while the su- 1992 PRUM: MANAKIN SYRINGES AND PHYLOGENY I1I perficial fibers are continuous between the two muscles. However, the partially intrinsic syringeal muscles in piprids are as independent or differentiated from M. tracheolater- 9, alis as the intrinsic syringeal muscles in other tyrannoids or in oscines. li Within the guidelines of the existing literature, many of these intrinsic syringeal muscles deserve different names. But intrin- Ip sic muscles have apparently evolved inde- %. pendently multiple times within the family 11 _ I a and have a variety offorms. In certain species they may be present in males and absent in Fig. 1. Lateral view of the plexus of the glos- females and immature males. Using the sopharyngeal (IX), vagus (X), and hypoglosso- names applied to intrinsic syringeal muscles cervical (XII + Cl) cranial nerves of Pipra coro- of oscines and other tyrannoids would mis- nata (UMMZ 225064). The syrinx is innervated takenly imply homology among these evo- by the tracheosyringeal branch of the hypoglos- lutionarily independent muscles. Rather than socervical nerve. Unlike the formation in other obscure the literature with an additional set birds, the hypoglossocervical nerve does not ex- of names for these muscular novelties, I will change fibers with the vagus or the esophageal refrain from formal names, and merely refer branch ofthe glossopharyngeal nerve. See text for further discussion. Abbreviations: c first cervical to them descriptively as left or right, dorsal nerve; e esophageal branch of glossopharyngeal or ventral intrinsic muscles. cranial nerve; g glossopharyngeal (IX) cranial nerve; h hypoglossal (XII) cranial nerve; hc hy- INNERVATION poglossocervical nerve formed by the union ofthe hypoglossal cranial nerve and the first cervical As in oscines and most other birds, the nerve; 1i lingual branch of the glossopharyngeal syrinx of piprids and other tyrannoids is cranial nerve; 11 laryngeolingual branch ofthe hy- innervated by the left and right tracheosyrin- poglossocervical nerve; lp lamyngeopharyngeal geal branches (Ramus syringealis) of the hy- branch of the glossopharyngeal cranial nerve; t poglossocervical nerves (N. hypoglossocer- tracheosyringeal branch ofthe hypoglossocervical vicalis), which are formed by the union of nerve; v vagus (X) cranial nerve; vg vagus-glos- the two roots ofthe hypoglossal (XII) cranial sopharyngeal nerve connection. Scale bars in all nerve and the first cervical nerve (fig. 1) (Ko- figures equal 1 mm. ditz, 1925; Breazile and Yasuda, 1979; Bub- ieni-Waluszewska, 1981). To my knowledge, tracheosyringeal nerves veer caudoventrad to the cranial nerves and the syringeal inner- form a prominent X-shaped chiasma on the vation of tyrannoids have not been previ- ventral midline of the trachea. The position ously described. My observations of a single of this ventral anastomosis varies between specimen ofPipra coronata indicate that, un- A20 and 35. The left and right nerves then like oscines and Old World suboscines, the continue craniolaterally from the chiasma, hypoglossocervical nerve in this piprid does return to the lateral midlines of the trachea, not exchange any fibers with either the vagus and innervate M. tracheolateralis, M. ster- (X) or the glossopharyngeal (IX) cranial notrachealis, and intrinsic syringeal muscles nerves before dividing into the laryngeolin- by small branches. This pattern is apparently gual and tracheosyringeal branches (fig. 1). derived in piprids, but has degraded in Chlo- From their origin in the laryngeal region, ropipo and Xenopipo. the left and right tracheosyringeal nerves of In piprids and cotingids, a small branch of piprids continue caudad on the lateral sur- the vagus nerve that originates in the distal faces ofthe trachea imbedded in the fibers of vagal ganglion innervates the trachea and M. tracheolateralis. Typically in piprids, as bronchi. In piprids, these nerves usually at- M. tracheolateralis widens ventrally to cover tach to the dorsal ends ofthe B 1 elements in the entire ventral surface of the trachea, the membranous strands of connective tissue 12 AMERICAN MUSEUM NOVITATES NO. 3043

geal musculature is described beginning with M. tracheolateralis and any intrinsic muscles present, and ending with M. sternotrachealis. Lastly, the position ofthe major components of syringeal innervation is described, along with any differences from the typical pattern. In Pipra cornuta, a variation in the medial tympaniform membrane is described. The syringes of some species are described in reference to differences from another similar Fig. 2. Syringeal supporting elements of Cor- form, but each abbreviated description is only apipo gutturalis (AMNH 2256). Left dorsal view, one reference away from a complete one. right ventral view. Abbreviations: Al Al supporting element; A2 A2 supporting element; Bi Bl supporting element; P pessulus; M medial car- Corapipo gutturalis (N = 2) tilage. Supporting elements (fig. 2). A1-5 are double and medially incomplete, and A6-7 are single and dorsally incomplete. Al is broad which extend from these elements to the dor- and thick; it is ossified for the ventrolateral sal wall ofthe interclavicular air sac on either half, and cartilaginous for the dorsal halfand side ofthe In some an esophagus. specimens, the ventral end. The ossified portion is en- additional branch innervates the dorsal sur- larged or swollen, and tubular in shape. A2- face ofthe trachea, following the arterial sup- 7 are narrower and thinner than Al, and are ply to meet the trachea between A8 and 12. fully ossified except for the ventral ends of Autonomic vagal nerve fibers are probably A2-4. A8 and above are single, complete, not involved in voluntary control of vocal- fully ossified, and unfused. In lateral view, ization but in the control ofmucous secretion in the airway. K6ditz (1925) found similar the syrinx conspicuously widens ventrallyjust vagal innervation ofthe syrinx in oscines and caudal to the tracheobronchial junction, as a Old World result of the prominent ventral lengthening suboscines, but Nottebohm and of the double, medially incomplete A and B Nottebohm (1976) were unable to locate any elements. A narrow, ossified pessulus is fused vagal syringeal innervation in several os- ventrally to A6, and is dorsally free or fused cines. However, these fibers are very small to half of A7 and A8. A1-2 are connected and may have been overlooked. In piprids, I dorsally to partially ossified bars of cartilage did not find any branches of the esophageal that run dorsoventrally along the cranial edge branch of the glossopharyngeal nerve which ofthe medial tympaniform membrane to ap- innervate the M. tracheolateralis, M. ster- proach but not fuse to the ventral end ofA3. notrachealis, or the intrinsic syringeal mus- These medial cartilage bars are fully ossified, cles, as reported in some oscine and nonpas- except for their dorsal tips. B 1 is broad, thick, seine birds (K6ditz, 1925; Breazile and Yasuda, 1979). and completely ossified except for its dorsal and ventral ends. The ossified portion of B 1 SYRINGEAL is enlarged or swollen and tubular in shape MORPHOLOGY OF THE like A1. All other B's are completely carti- PIPRIDS laginous. B 1-2 are fused by a short cartilagi- In each ofthe following descriptions ofthe nous bar just before their dorsal ends, pro- syringeal morphology of piprid species, the ducing small dorsal extensions beyond the supporting elements are described first, pro- fusion from both elements. The ventral end ceeding from A elements to the pessulus, oth- of B 1 is a thin cartilaginous bar that extends er accessory cartilages, and the B elements. caudodorsad to the ventral end of Al. All piprid B elements are double and medi- Muscles. Apparently as in Corapipo leu- ally incomplete, and will not be described as corrhoa (fig. 4), but only cleared and stained such below. Unless otherwise indicated, B specimens were observed. elements are cartilaginous. Next, the syrin- Innervation. Not observed. 1992 PRUM: MANAKIN SYRINGES AND PHYLOGENY 13

Corapipo leucorrhoa (N = 13) Supporting elements. Like Corapipo gut- teralis (fig. 2) except as follows: The double A4-6 elements are incomplete medially, but their ossified ventral ends and cartilaginous dorsal ends extend medially to give substantial support to the medial walls of the bronchi. The pessulus is fused dorsally to A8 and ventrally to A7. The medial cartilage bars are thinner and completely cartilaginous. The ventral widening of the syrinx and bronchi caudal to the tracheobronchial junction is even more pronounced than in C. gutturalis, Fig. 3. Syringeal supporting elements ofMas- though ossified portions of Al and B1 are ius chrysopterus (CM 1 161). Left dorsal view, right less extensive and less swollen. ventral view. Abbreviations: Al Al supporting Muscles (fig. 4). M. tracheolateralis is a element; M medial cartilages. robust, well developed muscle; it converges on the ventral midline between A30 and 35, and continues caudally to divide on the ven- Innervation. Typical piprid pattern. Ven- tral midline into left and right halves around tral anastomosis at A20-25. A1 0. It diverges toward the lateral surfaces ofthe trachea. The left and right halves insert Masius chrysopterus (N = 9) partially on the ventrolateral and lateral sur- Supporting elements (fig. 3). The ventral faces ofA6-8. The partial insertion gives rise widening ofthe syrinx cranial to the tracheo- to dorsal and ventral pairs ofoblique intrinsic bronchial junction is similar to that of Cora- musculature. Each is distinct in fiber direc- pipo, but not as prominent. A1-4 are double tion but both are partially continuous with and medially incomplete. Al is a broad, wide M. tracheolateralis fibers. The ventral pair of element, which is ossified except for the dor- intrinsic muscles originates on A5-7 on the sal quarter and the ventral end. A2-4 are fully ventral midline and on A4-5 or A2-4 on the ossified except for their cartilaginous dorsal lateral midline. Ventral fibers run directly tips and the ventral tips of A2-3. The carti- caudad and the lateral fibers pass obliquely laginous dorsal tips ofA1-2 are weakly fused caudoventrad to insert as a fleshy mass of in one specimen. AS is single and complete fibers on the cartilaginous ventral end of Al. (ventrally incomplete in one specimen). A6 The dorsal intrinsic musculature originates and subsequent A's are single, complete,-fully laterally at A4-5 and its fibers become con- ossified, and unfused. The dorsal ends ofA2- tinuous with those of the M. tracheolateralis 3 are fused to a dorsoventrally oriented pair dorsally. The fibers run caudodorsad forming of narrow medial cartilages which form the a large belly to insert fleshily on the cartilag- cranial margin of the medial tympaniform inous dorsal end of A2. M. sternotrachealis membrane. The wide ossified pessulus is fused is moderately well developed and inserts dorsally to A5-6 or A6-7 and ventrally to broadly on the lateral and ventral surfaces of A5 or A6. B 1 is ossified heavily for its ventral M. tracheolateralis at Al 5-20. Most fibers quarter, and is fused to the cartilaginous B2 run parallel with the M. tracheolateralis and dorsally by a short bar. The ventral end of are assimilated rapidly into it. The ventral- Bl is prominently hooked dorsomedially to most fibers fan out across the ventral surface rest closely to Al. The ventral end of B2 is ofthe trachea to insert on the ventral midline a thin, pointed dorsomedial hook. All sub- and form a conspicuous chevron-shaped in- sequent B elements are cartilaginous and sim- sertion with the fibers of M. sternotrachealis ply shaped. of the other side. A few dorsal fibers may Muscles. As in Corapipo leucorrhoa (fig. 4) insert directly on the tracheal A elements on except as follows. The ventral widening of the dorsal edge of M. tracheolateralis. the syrinx is less extreme and, correspond- 14 AMERICAN MUSEUM NOVITATES NO. 3043

Fig. 4. Ventrolateral view of syringeal muscles: left Corapipo leucorrhoa (LSUMZ 108683); center Ilicura militaris (FMNH 107028); right Manacus manacus (LSUMZ 112834). Abbreviations: I intrinsic syringeal muscles; S M. sternotrachealis; T M. tracheolateralis. ingly, the oblique angle ofthe intrinsic muscle cartilage lies medial to the dorsal ends ofAl fibers is less exaggerated. M. tracheolateralis and A2, and is connected to them; it runs is weakly differentiated on the ventral mid- cranioventrad along the craniodorsal margin line near A1O, but the two sides do not di- of the medial tympaniform membrane and verge laterally as in Corapipo. Rather, M. ends adjacent to the pessulus in the tracheo- tracheolateralis inserts continuously along the bronchial junction. All B elements are car- lateral and ventral surfaces ofA6-7. The in- tilaginous and thin. B1-2 are dorsally fused sertion ofM. sternotrachealis is the same ex- by a short bar; the dorsal ends of B1 extend cept that it occurs between Al 0 and 15. slightly beyond the short bar. The ventral Innervation. Typical piprid pattern, but ends ofB1-2 are thin and pointed, and close- more well developed and visible than in most ly nested to the caudal edge of A1. Subse- other piprids. Ventral anastomosis near A30. quent B's are unspecialized in shape. Each side sends a separate branch to inner- Muscles (fig. 4). M. tracheolateralis con- vate M. sternotrachealis and the intrinsic verges on the ventral midline above A35, and musculature. continues caudally without division to insert on the ventral and lateral surfaces of A6-7. Ilicura militaris (N = 5) A few of the dorsal fibers continue caudally to insert with the intrinsic muscles on the Supporting elements (fig. 5). Al is double, dorsal ends ofAl. A pair ofintrinsic muscles medially incomplete, and completely carti- originates on the ventral surface of A4 and laginous. A2-A6 or A7 are double, medially the ventrolateral surfaces of A5-6, and con- incomplete, and fully ossified except for the tinues obliquely caudodorsad to insert on Al dorsal ends of A2 and the ventral ends of from its dorsal end to the ventrolateral sec- A2-4. A7 or A8 and above are single, com- tion. M. sternotrachealis is thin and moder- plete, fully ossified, and unfused. The pes- ately developed, and inserts simply on the sulus is ossified, free dorsally, and possibly lateral surface of M. tracheolateralis at Al 3- fused ventrally to A5, A6, or A7. A bar of 15 (A10-12 on right side of one specimen). 1992 PRUM: MANAKIN SYRINGES AND PHYLOGENY 15

Fig. 5. Syringeal supporting elements. Left II- icura militaris (FMNH 102023), dorsal view; right Pipra coeruleocapilla (FMNH 291664), dorsal Fig. 6. Syringeal supporting elements ofMan- view. Abbreviations: Al Al supporting element; acus candei (AMNH 6654). Left dorsal view, right M medial cartilages. ventral view. Abbreviation: Al + A2 fused Al and A2 supporting elements. The musculature of males is well developed whereas female musculature is thin and trally to A5. All B elements are completely weakly developed. cartilaginous. B 1-2 are dorsally fused by a Innervation. Typical piprid pattern. Ven- short cartilaginous bar. The ventral ends of tral anastomosis near A25. B1-3 are angled abruptly craniad into small, squarish extensions at right angles to the cra- Manacus manacus (N = 13), M. vitellinus nial margins of the elements. B5-10 or B5- (N = 3), and M. candei (N = 1) 12 are each broader and flatter in cross sec- Supporting elements (fig. 6). A1-2 are dou- tion than B 1-3, and are fused in a continuous ble, medially incomplete, and ossified except cartilaginous lattice formed by their widened for their cartilaginous dorsal tips. They are dorsal ends. This reinforced lattice provides completely fused except for their rounded additional support to the medial bronchial ventral ends which indicate the line offusion walls and the medial tympaniform mem- between the elements. Both Al and 2 are branes. distinctly rough or grainy in texture unlike Muscles (fig. 4). M. tracheolateralis con- all other A's. Al sometimes has a prominent verges on the ventral midline at A30 to cover ventrally flared shelf on its caudal margin. the entire ventral surface ofthe trachea. The A3-4 are double, medially complete, and os- muscle continues caudad but is weakly united sified except for their medial portions. In one along the ventral midline. It differentiates specimen ofM. vitellinus and the single speci- completely into left and right halves on the men ofM. candei, A3 is largely fused to Al- ventral midline at A12. These halves do not 2 in the same manner. The cartilaginous me- diverge laterally from one another, but con- dial sections of A3-4 and the dorsal ends of tinue caudad as adjacent muscle masses. Both Al-2 are fused by short cartilaginous con- sides insert by a thin sheet of connective tis- nections. The next one to six A elements (A5 sue on A3 at the ventral midline and spiral alone to A5-10) are single, complete, un- caudodorsad to the ventrolateral surfaces of fused, and fully ossified except for a short A2, and the lateral and dorsolateral surface section at the dorsal midline. These dorsal of Al. The connective tissue sheet is contin- cartilaginous sections are connected by a con- uous between the left and right sides across tinuous caudocranial dorsal bar of cartilage. the ventral midline, even though the fibers Cranial to the last element in this dorsally of the left and right muscles are completely cart laginous series, all A's are single, com- distinct. In males, M. tracheolateralis is a very plete, unfused, and fully ossified. The carti- robust and thick muscle with a prominent, laginous pessulus is fused dorsally and ven- but completely extrinsic belly caudal to the 16 AMERICAN MUSEUM NOVITATES NO). 3043

Fig. 7. Syringeal supporting elements: Chiroxiphia caudata (AMNH 2447), left dorsal view, center ventral view; Antilophia galeata (BMNH 1968.62.205), right dorsal view. Abbreviations: Al Al supporting element; M medial cartilage. lateral division at A12. In females, the muscle of the Al elements alone, which are double, is the same in form but thin and weakly de- medially incomplete, and cartilaginous. Al veloped. M. sternotrachealis is a moderately is wide and indented craniad on the caudo- well developed muscle and inserts on the dor- ventral margin. All other A elements are fully solateral surface ofM. tracheolateralis at Al 1- ossified. The second A series is composed of 13. Its fibers are incorporated rapidly into M. the next two to five A elements which are tracheolateralis. double, medially incomplete, unfused, and Innervation. Usually the typical piprid pat- narrower than Al. The third series is com- tern is present. The ventral anastomosis is posed of two or three A elements cranial to near A30, and large branches innervate M. the second series which are single, ventrally tracheolateralis and M. sternotrachealis cau- fused, and dorsally incomplete. The fourth dally. In one specimen, the nerve continues series is composed of between six and nine caudad on the ventral midline as a single fiber A elements which are fused together to form for 1 mm before dividing again into left and a prominent tracheal drum. The caudal one right branches. or two of the elements in the drum are typically completely fused dorsally and only partially fused ventrally, whereas the cranial el- Chiroxiphia linearis (N = 7), C. lanceolata ements in the series are entirely fused. The (N = 1), C. pareola (N = 10), and C. caudata fifth series of A elements is made of the rest (N = 7) ofthe tracheal A's which are single, complete, Supporting elements (fig. 7). The bronchi unfused, and fully ossified. The A elements are prominently flared laterally and ventrally that comprise each ofthese series in the four so that the medial surfaces are exposed in species of Chiroxiphia are presented in table dorsal view. The trachea becomes increas- 1. The dorsal and ventral ends of the second ingly widened toward the tracheobronchial series of A elements, the dorsal ends of the junction caudal to a series of fused single A third A series, and the caudoventral margin elements that form a tracheal drum. The of the drum are connected by a sheet of car- structure of A elements can be described as tilage that forms the craniomedial surfaces of five series ofdistinct element types. The spe- the bronchi and the dorsal surface of the cific A elements that comprise each series tracheobronchial junction. The caudal mar- vary slightly among and within species (table gin of the cartilage sheet is indented craniad 1). In all Chiroxiphia, the first series is made between its connections to the dorsal and 1 992 PRUM: MANAKIN SYRINGES AND PHYLOGENY 17

TABLE 1 Elements Comprising the Five A Element Series in Chiroxiphia and Antilophia A element seriesa Species I II III IV V Chiroxiphia linearis 1 2-5 5-8 8-15 14-16+ Chiroxiphia lanceolata 1 2-3 4-5 6-13 14+ Chiroxiphia pareola 1 2-4 5-8 7-14 14-15+ Chiroxiphia caudata 1 2-5 5-8 8-17 14-18+ Antilophia galeata 1 2-5 6-7 8-14 15+ a The five A element series are described in the text under the supporting elements of Chiroxiphia and Antilophia. The range of elements found in each A series in each species is given. ventral ends of the first A elements, and it trachea. The connective tissue is continuous forms the cranial margin of the medial tym- between the left and right sides across the paniform membrane. Al is fused dorsally to ventral midline. In pareola, the muscle at- A2 and the caudolateral corner of the acces- tenuates in the same manner and inserts by sory cartilage sheet by a bar of cartilage. Al a connective tissue sheet on A4 on the ventral and B 1 are also fused by a thin cartilaginous midline, but the sheet extends caudad to in- bridge which has a small square projection sert on the ventral end of A2 just lateral to on its craniolateral edge between the two el- the ventral midline on both sides, instead of ements. All B elements are cartilaginous. B 1- spiraling caudally along the entire lateral sur- 2 are dorsally fused by a narrow cartilage face as in linearis and caudata. The dorsal strip. The ventral ends of B-1-3 are widened portion of the intrinsic musculature is com- and roundly angled dorsomedially. The ven- posed of fibers that are continuous with M. tral end of B1 is closely nested next to the tracheolateralis and the ventral intrinsic ventral end of Al. muscle, and of other fibers that originate di- Muscles (fig. 8). M. tracheolateralis widens rectly on the A elements at the cranial margin ventrally to cover the entire ventral surface of the drum. These dorsal intrinsic fibers of the trachea above A35-40, and continues originate in a broad fan on the most cranial caudally in a thin, weakly developed sheet of elements of the drum at the dorsal edge of fibers to insert partially on the ventral and M. tracheolateralis and on the dorsolateral lateral surfaces ofthe A elements at the crani- surfaces of the next three or four drum ele- al margin ofthe drum (A13-17). The intrinsic ments. The dorsal intrinsic musculature is musculature originates on the cranial portion continuous laterally with the ventral intrinsic ofthe drum. It is largely continuous with the fibers, but forms a distinct intrinsic belly on fibers of M. tracheolateralis, and is weakly the dorsolateral surface of the trachea. The differentiated into dorsal and ventral por- deep fibers insert with the ventral intrinsic tions from the origin. The ventral portion of belly by a connective tissue sheet on the dor- the intrinsic muscle forms a large, prominent sal ends ofA2, but the more superficial fibers belly of fibers on the ventral and lateral sur- insert by a short, narrow strip of connective faces of the trachea. The left and right sides tissue to the dorsal end of Al. M. sternotra- are differentiated laterally on the ventral mid- chealis is a thin and weakly developed muscle line from origin to insertion, but they run that inserts on the dorsal edge of M. trache- parallel and adjacent to one another. This olateralis just cranial to its partial insertion mass of muscle fibers attenuates rapidly at at the cranial margin of the drum and to the A5 on the ventral midline and A2 laterally, origin ofthe dorsal intrinsic musculature. Fi- and each side inserts by a sheet ofconnective bers of M. sternotrachealis insert directly on tissue that is attached to A4-5 at the ventral the dorsalmost M. tracheolateralis fibers and midline and spirals caudolaterally to insert on the tracheal A elements. The musculature on A3 and A2 on the lateral surface of the of females is similar in form to that of males 18 AMERICAN MUSEUM NOVITATES NO. 3043

Fig. 8. Synngeal muscles of Chiroxiphia caudata (FMNH 107323): left dorsolateral view, right ventrolateral view. Abbreviations: I intrinsic syringeal muscles; S M. sternotrachealis; T M. tracheolateralis. but less well developed. The musculature of medially incomplete, cartilaginous), A2-5 lanceolata is the same in general form, but (double, medially incomplete, ossified), A6- no uncleared specimens were observed to 7 (single, dorsallyincomplete), A8-14 (single, document the details of the ventral intrinsic fused into a drum), and A 15 and above (sin- muscular insertion. gle, complete, unfused). As in Chiroxiphia, Innervation. All but four specimens (one Al and all B's are cartilaginous, and all other each of linearis and pareola, and two of cau- A's are ossified. The syrinx ofAntilophia dif- data) differ from typical piprid morphology. fers from that of Chiroxiphia in that the el- The left and right tracheal nerves converge ements ofthe second A series are wider than on the ventral midline at A20-30, and con- those of Chiroxiphia and more similar in tinue caudad as a single large fiber along the shape to Al. Also, the cartilaginous dorsal ventral midline for 0.8 mm to more than 6.0 bridge between Al and B1 is shorter and lacks mm, before dividing once again into left and the square projection on its craniolateral right branches that innervate the intrinsic margin found in Chiroxiphia. muscles and M. sternotrachealis ofeach side. Muscles. Generally as in Chiroxiphia (fig. The three other specimens have the typical 8). The ventral intrinsic muscles originate on piprid pattern with a brief anastomosis. A14 and insert spirally on AS to A2; the dorsal intrinsic muscles are continuous with the Antilophia galeata (N = 6) ventral intrinsic fibers laterally, but they originate independently on the dorsal surface of Supporting elements (fig. 7). Very similar A12-14 and insert on the dorsal end of Al to those ofChiroxiphia. The A elements form by a narrow tendon. M. sternotrachealis is five series with the same characteristics in- more well developed than in Chiroxiphia, and cluding an extensive tracheal drum and me- inserts on the lateral surfaces of M. tracheo- dial bronchial cartilage sheet. The five series lateralis and the trachea at A15-17. of A elements are composed of Al (double, Innervation. Four specimens have the dis- 1 992 PRUM: MANAKIN SYRINGES AND PHYLOGENY 19 tinct Chiroxiphia pattern in which left and right tracheosyringeal nerves unite into a single large fiber before innervating the intrinsic muscles and M. sternotrachealis. One specimen has the typical piprid pattern ofa simple nerve anastomosis.

Machaeropterus regulus (N = 8) Supporting elements (fig. 9). A1-2 are double and medially incomplete. Al is arched cranially at the lateral midline and is cartilaginous except for a small ossified oval near Fig. 9. Syringeal supporting elements of Ma- the ventral end of the element (about a sixth chaeropterus regulus (LSUMZ 85891). Left dorsal to an eighth ofthe entire element). A2 is fully view, right ventral view. Abbreviation: Al Al ossified except for its cartilaginous dorsal and supporting element. ventral tips. A3-4 are double, medially complete, and generally ossified except for the medial cartilaginous portions; one side ofA4 and partially inserts on A5 to produce in- is continuous dorsally with the pessulus in trinsic muscles. In one male, the left and right all specimens. A5-6 are single and ossified, sides partially insert on A5, and give rise to but dorsally cartilaginous or incomplete. A7 partially intrinsic muscles that are mostly and subsequent A elements are single, com- continuous with M. tracheolateralis fibers, but plete, and fully ossified. The pessulus is fused have some fibers originating on A4-5. In the ventrally to A5 and is attached dorsally to other male specimen, each side of M. tra- the dorsal tips of A4-6 or A5-6, and to the cheolateralis inserts on the lateral surface of ossified dorsal portion ofA7. All B elements A5. An intrinsic muscle originates on the dor- are entirely cartilaginous. B 1-2 are fused dorsal margin of the M. tracheolateralis on A6- sally by a bar ofcartilage. In most specimens 9 and its fibers pass obliquely caudoventrad (N = 6 of 8), B3 is also dorsally fused to B 1- over the insertion of M. tracheolateralis to 2 in the same manner. The ventral ends of insert on the ventral end of Al. A few fibers B 1-3 are widened slightly and roundly turned are continuous with the dorsalmost fibers of dorsad. The ventral end of B2 is a unique, M. tracheolateralis at A9, and some ventral symmetrical spoon shape, and the distortion fibers of this muscle originate on A5 at the ofB3 is slight. All subsequent B elements are ventral edge of M. tracheolateralis. Another thin and simple in shape. completely intrinsic muscle originates on the Muscles (fig. 12). There is substantial vari- dorsolateral surface ofA5 and the lateral sur- ation in the degree ofintrinsic differentiation face of A4, and its fibers pass dorsoventrad of the caudal portions of M. tracheolateralis. to insert on the third ofAI just dorsal to the The left side of M. tracheolateralis is much lateral midline. These two intrinsic muscles more developed than the right, and it con- are similar in fiber direction to the ventral tributes all ofthe fibers on the ventral surface and dorsal extrinsic portions of M. tracheo- ofthe trachea. In both sexes the ventral sheet lateralis in females, but have different origins of fibers divides on the ventral midline at at least partially independent of this muscle. A0, and the left and right halves diverge The left M. sternotrachealis inserts on the left toward the lateral surfaces of the trachea. In ventrolateral and lateral surfaces of M. tra- females, each side divides into dorsal and cheolateralis at Al 0-15, and the right side ventral extrinsic portions on the lateral mid- inserts on the ventrolateral surfaces of M. line at A2-4; the dorsal branch inserts just tracheolateralis and directly on the lateral lateral to the dorsal end ofA1, and the ventral surfaces of the tracheal A elements at A10- portion passes obliquely caudoventrad to in- 15. In females, M. sternotrachealis is weakly sert on the ventral end of Al. In males, the developed and ribbonlike, but in males it is M. tracheolateralis is more well developed a well developed muscle. The particularly 20 AMERICAN MUSEUM NOVITATES NO. 3043

Fig. 10. Syringeal supporting elements ofMa- chaeropterus deliciosus (AMNH 8713). Left dorsal view, right ventral view. Abbreviation: Al + A2 Fig. 1 1. Syringeal supporting elements of fused Al and A2 supporting elements. Chloropipo holochlora (KU 65552). Left dorsal view, right ventral view. Abbreviation: Al Al supporting element. large left side contributes a large volume of fibers to the mass of M. tracheolateralis. Innervation. Typical piprid pattern. Ven- A8-12, because of the reduced fibers of the tral anastomosis at A20-25. A separate right M. tracheolateralis. branch innervates M. sternotrachealis. Innervation. Typical piprid pattern as in M. regulus. Machaeropterus pyrocephalus (N = 11) Supporting elements. As in M. regulus (fig. Machaeropterus deliciosus (N = 6) 9) except as follows. A1-2 are fused for the Supporting elements (fig. 10). Al-2 are short ventral section where Al is ossified. double, medially incomplete, and fused to- There are no dorsally cartilaginous tracheal gether for the ventral two-thirds and at the A elements. The lateral portion of Al is not dorsal ends. Al is ossified for the ventral half arched craniad. B3 is not dorsally fused to (except for the tip) and cartilaginous for the B1-2. The ventral ends of B1-3 are slightly dorsal half. A2 is completely ossified except widened, paddle shaped, and angled cranio- for the dorsal tips. The cartilaginous ventral dorsad. All three are more distorted in shape tip ofA1 projects caudad from the caudoven- than in regulus. tral extremity ofthe fused A1-2 element. A3 Muscles. The left side of M. tracheolater- or A3-4 are double, complete, and generally alis is much more developed than the right. ossified but medially cartilaginous. A5 or A4- The two sides converge on the ventral mid- 5 are single, complete, and fully ossified. In line above A25, and the left side contributes one specimen, A4-5 are dorsally fused to one most of the fibers that continue caudad on another. A6 and subsequent A's are single, the ventral surface of the trachea. This sheet complete, fully ossified, and unfused. A car- of fibers divides weakly on the ventral mid- tilaginous pessulus is fused dorsally and ven- line at A8-10 into separate left and right trally to A4. All B elements are completely halves that continue caudad to diverge com- cartilaginous. B1-2 are dorsally fused by a pletely at A2 or A3. All fibers insertjust later- short cartilaginous bar. The ventral ends of al to the dorsal and ventral ends of Al. The B 1-3 are widened, paddle-shaped, angled muscle may be developed into strong bellies craniad, and nested together. The cranial edge caudally, but is weakly differentiated laterally of these paddle-shaped elements is addition- in three specimens. The left M. sternotra- ally distorted by a caudal concavity. All other chealis inserts on the lateral surface ofthe M. B elements are simple in shape. tracheolateralis between A8 and 12, and is Muscles. M. tracheolateralis is thin and very well developed. The right M. sternotra- weakly developed in both males and females. chealis inserts on the ventral surface of M. It covers the ventral surface of the trachea tracheolateralis and on the lateral surface of caudal from A30, but in all specimens the 1 992 PRUM: MANAKIN SYRINGES AND PHYLOGENY 21

Fig. 12. Ventrolateral view of syringeal muscles: left Machaeropterus regulus (LSUMZ 114486), center Chloropipo holochlora (UMMZ 225050), right Pipra coeruleocapilla (USNM 512291). Abbre- viations: I intrinsic syringeal muscles; S M. stemotrachealis; T M. tracheolateralis. right side is much less developed or almost sively fused along their dorsal margins. A thin, absent, and the left side contributes most of ossified pessulus is fused dorsally and ven- the fibers to the sheet of M. tracheolateralis. trally to A5. All B elements are completely The thin sheet of muscle continues caudad cartilaginous. B1-2 are robust and dorsally and divides into left and right halves on the fused by a short cartilaginous bar. The ventral ventral midline at A7-8, before inserting on ends of B1-3 are widened, angled dorsome- the ventral half of Al. M. sternotrachealis is dially, and closely nested together. Subse- a thin, ribbonlike muscle, and it inserts sim- quent B elements are fused and unspecialized ply on the lateral portions ofthe M. tracheo- in shape. The shape ofthe ventral ends of B3 lateralis at A-1012. varies within the sample between nominate Innervation. Typical piprid pattern. Nerve holochlora from Amazonian Ecuador and fi- fibers minute. Ventral anastomosis at A20. tae from the Pacific slope of Ecuador. In nominate holochlora, the ventral end of B3 is smaller, rounded caudally, and more Chloropipo holochlora (N = 11) strongly angled than in litae, which resembles Supporting elements (fig. 11). A1-2 are other piprids more closely. An additional ab- double, medially incomplete, and completely errant cleared-and-stained specimen from ossified. The cartilaginous extensions of the Amazonian Ecuador differs markedly from dorsal ends of A2 connect to A3. A3-4 are the rest of the sample (fig. 1 1). A3-4 are not double, complete, and fully ossified, and their straight medially, and the left A2 is cartilagi- medial portions are straight instead ofround- nously complete medially. No tracheal A ed. A5 and subsequent elements are single, elements are fused, and the ventral ends of complete, and fully ossified. These single el- B3 are different in shape from those of all ements are fused dorsally by a wide ossified other specimens. This specimen may either bar between elements. This partial drum is indicate significant intraspecific variation, or made up of A5-6 (N = 4), A5-7 (N = 4), or be the result of hybridization or misidenti- A5-8 (N = 2). In one ofthese last specimens, fication, but requires further investigation. the elements composing the drum are exten- Muscles (fig. 12). The left and right Mm. 22 AMERICAN MUSEUM NOVITATES NO. 3043

plete. A5-7 are fused dorsally. The dorsal portion ofA5 is chevron-shaped and extends medially to form the thin ossified pessulus which is fused ventrally to A5. Al-7 are 6 i~~~~Al closely fitted next to one another; the caudal edges of the double A elements, especially A1-2, are larger in diameter than the cranial edges, so that the bronchi widen laterally, giving the syrinx a unique inverted funnel shape. The B elements are very thin and the Fig. 13. Syringeal supporting elements of shapes ofthe ventral ends ofB 1-3 are unique. Chloropipo unicolor (LSUMZ 89472). Left dorsal B 1-2 are dorsally fused by a short bar ofcar- view, right ventral view. Abbreviation: Al Al tilage, but all other B's are unfused. B1 is supporting element. larger in diameter than A1, resulting in a conspicuous gap between the elements. The ven- tracheolaterales widen to cover the ventral tral half to one-third ofB1 is ossified, and its surface of the trachea between A30 and 40. thin ventral end curves dorsomedially to ap- The fibers continue caudad until a slight con- proach the ventral end of A1. The ventral striction in muscle mass and a partial inser- end of B2 angles dorsomediad, and is lightly tion on A8-9, where the muscle gives rise to ossified. The extreme ventral tip of B3 left and right partially intrinsic muscles. The abruptly angles craniad 900 to nest close to left and right bellies divide into dorsal and B2. All subsequent B's are unspecialized in ventral intrinsic muscles on the lateral mid- shape. line between A4 and 8. The dorsal portions Muscles. Similar to those of Chloropipo are composed of lateral fibers that are con- holochlora (fig. 12). M. tracheolateralis cov- tinuous with M. tracheolateralis and, in some ers the ventral surface of the trachea above specimens, additional, completely indepen- A30, and continues in an undivided sheet to dent fibers that originate on the lateral and insert partially on the ventral and lateral sur- dorsolateral surfaces on A4-8. The dorsal and faces of A6-7. These fibers are continuous ventral intrinsic muscles insert on the dorsal with the intrinsic musculature that originates and ventral ends of Al. M. sternotrachealis as a laterally differentiated pair at A6. The is a well-developed muscle that inserts di- intrinsic muscles continue caudad and divide rectly on the lateral fibers ofM. tracheolater- into dorsal and ventral bellies on the lateral alis between A14 and 17. Female syringeal midline at A4. These four muscles insert musculature is the same in general form as broadly on the ventral and dorsal thirds of in males but less developed, and the dorsal Al. M stemotrachealis is well developed and intrinsic muscle is completely continuous with inserts on the lateral and ventrolateral fibers the fibers of the M. tracheolateralis. of the M. tracheolateralis between A17 and Innervation. The tracheosyringeal nerve fi- 22. A few fibers insert directly on A17-22 at bers are generally thin or degenerate. The the dorsal edge of M. tracheolateralis. In fe- small left and right fibers wander irregularly males, the musculature is very thin and un- across the ventral surface of the trachea and developed, but in males it is well developed exchange small secondary fibers at various and robust. points between AIO and 30. Small caudal Innervation. As in Chloropipo holochlora. fibers innervate the dorsal intrinsic musculature. Chloropipo uniformis (N = 7) Supporting Chloropipo unicolor (N = 7) elements (fig. 14). Al-2 are double, medially incomplete, and completely Supporting elements (fig. 13). All A ele- ossified. The caudal margin ofAI tapers nar- ments are fully ossified. A1-2 are double, and rowly toward the ventral end ofthe element. medially incomplete. A3-4 are double and A3 is double, complete, and fully ossified. A4 complete. A5 and above are single and com- is single but ventrally incomplete in at least 1 992 PRUM: MANAKIN SYRINGES AND PHYLOGENY 23

Fig. 14. Syringeal supporting elements of Chloropipo uniformis (AMNH 7680). Left dorsal Fig. 15. Syringeal supporting elements ofXen- view, right ventral view. Abbreviation: Al Al opipo atronitens (AMNH 8083). Left dorsal view, supporting element. right ventral view. Abbreviation: Al Al supporting element. one specimen. The obliquely aligned left and right halves of A4 are fused dorsally in a ture is well developed in males and thin in prominent chevron shape, which extends females. medially to form the pessulus. A5 and sub- Innervation. Nerve fibers not visible in most sequent A's are single, complete, and com- specimens. One male specimen has the typ- pletely ossified. A5 has an ossified caudal pro- ical piprid pattern with the ventral anasto- jection on the dorsal midline that extends mosis at A25. caudad into the area just craniad to the margins of the chevron-shaped A4 element. All Xenopipo atronitens (N = 7) B 1 elements are completely cartilaginous. Supporting elements (fig. 15). As in Chlo- B1-2 are dorsally fused by a broad cartilag- ropipo uniformis except as follows: A4 ven- inous bar; all other B's are unfused. The ventrally incomplete in at least one specimen. tral end ofB 1 is slightly widened and roundly A5-6 (N = 4), A4-5 (N = 1), or A5-7 (N = curved craniomediad. The ventral end of B2 1) are fused dorsally by an ossified bar. A3 is widened asymmetrically. The caudal mar- is medially incomplete on one side in one gin is straight, but the cranial margin flares specimen. The ventral end of B1 abruptly craniad giving the B2 element a lopsided turns dorsomediad with a distinct protruding spatulate ventral end. B3 and subsequent B's corner on the ventral extreme ofthe element. are unspecialized in shape. The center of the ventral end of B1 is ossi- Muscles. Similar to those of Chloropipo fied. B1 is more abruptly angled than in C. holochlora (fig. 12). M. tracheolateralis is a uniformis. B2-3 are asymmetrically spatulate moderately developed and uniform muscle like B2 of C. uniformis. that expands to cover the ventral surface of Muscles. Similar to those of Chloropipo the trachea near A35. The fibers continue holochlora (fig. 12). Musculature is well de- caudad and divide into left and right halves veloped and robust in the male specimens at A8-10. The left and right halves insert observed. It is similar to C. uniformis except discontinuously on the lateral and ventrolat- as follows. M. tracheolateralis does not di- eral surfaces of A5-6, and then give rise to verge ventrally before its insertion on A6 and the intrinsic musculature. These partially in- the origin of the intrinsic musculature. The trinsic muscles divide on the lateral midlines intrinsic muscles are differentiated laterally at A2 to insert on the dorsal and ventral ends from their origin; they are more weakly dif- of Al. M. sternotrachealis is moderately de- ferentiated dorsoventrally than in uniformis veloped, and inserts on the lateral surface of or unicolor, and divide at A2 to insert on the M. tracheolateralis from A10-14. A few dor- ventral and dorsal ends of Al. M. sternotra- sal fibers insert directly on A10-13 at the chealis is well developed and inserts on the dorsal edge of M. tracheolateralis. Muscula- lateral and ventrolateral fibers ofM. tracheo- 24 AMERICAN MUSEUM NOVITATES NO. 3043

widened and angled dorsomedially at their ventral ends. B4 and subsequent elements are unfused and unspecialized in shape. Muscles. M. tracheolateralis covers the entire ventral surface above A30 in a well de-

+ veloped, undivided sheet, and it inserts continuously on the ventral and lateral surfaces of the cranial margin of the drum at A7-8. The left and right intrinsic muscles are laterally differentiated on the ventral midline from their origin just caudal to this partial Fig. 16. Syringeal supporting elements ofHet- insertion. They form prominent bellies of fi- erocercusflavivertex (AMNH 15204). Left dorsal bers and insert along the entire surface ofAI view, right ventral view. Abbreviation: Al + A2 without dorsoventral differentiation. M. ster- fused Al and A2 supporting elements. notrachealis is a robust muscle that inserts on the ventrolateral and lateral surfaces of M. tracheolateralis at Al 1-14. The ventral lateralis between A10 and 15. The ventral- fibers fan out cranioventrad across the ven- most fibers insert on the ventral midline in tral surface of the trachea to meet the fibers a chevron pattern. of the other side on the ventral midline. A Innervation. Not well developed and dif- few ofthe dorsal fibers also insert on the tra- ficult to see in most specimens. Main fibers cheal A elements at the dorsal margin of M. anastomose on the ventral midline at A25, tracheolateralis. The musculature is less mas- but secondary fibers on both sides continue sive in female specimens but the intrinsic directly caudad connecting the main fibers muscles are still completely developed. above and below the anastomosis. Innervation. Typical piprid pattern. Ven- tral anastomosis at Al 8-25. Heterocercusflavivertex (N = 7) and H. linteatus (N = 3) Pipra coronata (N = 11), P. isidorei (N = 1), Supporting elements (fig. 16). Al is double, P. coeruleocapilla (N = 7), P. nattereri medially incomplete, ossified for the ventral (N = 8), and P. iris (N = 6) two-thirds, and cartilaginous for the dorsal Supporting elements (fig. 5). A1-2 are dou- third. A2 is double, complete, and mostly ble and medially incomplete. Al is complete- ossified but medially cartilaginous. The car- ly cartilaginous, and A2 is ossified with cartilaginous medial portions ofA2 are variable tilaginous dorsal and ventral tips. A3-4 are in extent. A1-2 are fused for the ventral two- double, complete, and ossified except for the thirds and at their dorsal ends. A3-4 are dou- medially cartilaginous sections. A3-4 are ble, complete, and fully ossified; their medial partially ossified medially in iris and most portions are quite straight rather than round- nattereri. The cartilaginous medial portions ed. A5-7 are single, complete, ossified, and of A3-4 are partially fused in isidorei, iris, fused into a partial drum (A5-8 in one fla- nattereri, and two of six specimens of coro- vivertex and one linteatus). The fusion of the nata. The cartilaginous ends of A2 extend elements at the dorsal midline is complete, medially and fuse to the medial portions of and the entire dorsal margins ofthe elements A3 in coeruleocapilla. A5 and subsequent el- are partially fused, but the presence of an ements are single, complete, and usually en- ossified dorsal bar is still detectable. A8 and tirely ossified. In two specimens ofcoronata, above are single, fully ossified, and unfused. A5 is dorsally cartilaginous, and in one spec- The pessulus is ossified, wide dorsally, and imen A5-6 are dorsally cartilaginous. All fused dorsally and ventrally to A5. All B el- subsequent A elements are single, complete, ements are cartilaginous. B1-2 are dorsally and entirely ossified. A thin cartilaginous pes- fused by a short cartilaginous bar. B 1 is slight- sulus is fused dorsally and ventrally to A5 ly widened at its ventral end, and B2-3 are (fused dorsomedially to one A4 in one spec- 1992 PRUM: MANAKIN SYRINGES AND PHYLOGENY 25 imen of coronata). In three specimens of iris and most nattereri, the pessulus is partially or mostly ossified. All B elements are entirely cartilaginous. B 1 and B2 are fused dorsally by a short cartilaginous bar. The ventral ends ofB1-3 are widened, paddle-shaped, and angled craniad. B3 is the most exaggerated in shape. The ventral ends of B1-3 of P. coronata are less specialized in shape than the other species. Subsequent B elements are thin and unspecialized in shape. Muscles (fig. 12). M. tracheolateralis widens ventrally to cover the entire ventral surface of the trachea by A25, and continues Fig. 17. Syringeal supporting elements ofPip- caudally in a thin sheet of fibers to divide ra suavissima. Left dorsal view (AMNH 816768), into left and right halves at A7-10. The left right ventral view (AMNH 9366). Abbreviation: and right extrinsic muscles generally insert as Al Al supporting element. a single continuous sheet of fibers from the ventrolateral to dorsolateral surfaces of A1. In three of six coronata and two of seven inous ventral and dorsal tips. A3-4 are dou- coeruleocapilla, M. tracheolateralis divides ble, complete, and entirely ossified except for on the lateral midline at A2-4 into two weak- cartilaginous medial sections, which are fused ly differentiated dorsal and ventral groups of together. A5 and above are single and com- fibers which insert on the dorsolateral and plete. The A5-8 or A5-12 are ossified except ventrolateral surfaces of A1. This dorsolat- for a small cartilaginous portion on the dorsal eral differentiation is produced by the un- midline. Subsequent A's are entirely ossified. derdevelopment of the lateral fibers rather The cartilaginous pessulus is fused ventrally than by the development of separate dorsal to A5, and dorsally it is closely associated and ventral groups of fibers. There are no with but not fused to the cartilaginous dorsal prominent bellies in the M. tracheolateralis portion of A5. All B elements are entirely before insertion. M. sternotrachealis is mod- cartilaginous. B 1-2 are fused dorsally by a erately well developed and inserts directly on short cartilaginous bar. The ventral ends of the lateral and ventrolateral surfaces of M. B 1-3 are widened and spatulate, and angled tracheolateralis. The insertion varies in ex- craniodorsad to nest close to one another and tent from A9-15 to A12-18. A few of the to the tapered ventral end ofAl. The ventral dorsalmost fibers of M. sternotrachealis usu- ends of B 1-2 are slightly widened, but B3 is ally insert directly on the A elements at the the most distorted. All subsequent B ele- dorsolateral margin of the M. tracheolater- ments are thin and unspecialized. alis. In two of five coronata, the left M. ster- Muscles. Generally weakly developed. M. notrachealis is more strongly developed than tracheolateralis divides laterally on the ven- the right, but in general there is limited asym- tral midline at A7. The insertion of M. tra- metry in syringeal musculature in these spe- cheolateralis differs in the three uncleared cies. specimens. In one adult male, it inserts on Innervation. Typical piprid pattern. Ven- the lateral and ventral surface of A1; in an tral anastomosis between A25 and 30. immature plumage male, M. tracheolateralis is weakly differentiated into ventral and dor- Pipra suavissima (N = 6) sal bundles to insert on the dorsal and ventral Supporting elements (fig. 17). A1-2 are ends of Al. In another adult male, M. tra- double and medially incomplete. Al is car- cheolateralis inserts on the lateral and ventral tilaginous and arched craniad along its lateral surfaces of A2, and a portion of the dorsal- portions. Its caudoventral margin tapers most fibers extend to insert on the dorsal end craniad. A2 is ossified except for its cartilag- of the cartilaginous Al. M. sternotrachealis 26 AMERICAN MUSEUM NOVITATES NO. 3043

complete ventrally, and fully ossified except for its dorsal tips which are fused to the cranial end ofthe medial bronchial cartilaginous plate or to the cartilaginous pessulus. A7-14 or 15 are single, complete, and fully ossified except for a short cartilaginous portion on the dorsal midline. All subsequent A elements are single, complete, and entirely ossified. The cartilaginous pessulus is fused ventrally to A6, and is attached dorsally to the dorsal tips ofA7 and to the medial, bronchial cartilaginous plate. All B elements are Fig. 18. Syringeal supporting elements ofPip- completely cartilaginous. The slightly wid- ra serena (ROM 127643). Left dorsal view, right ened ventral ends of B1 are slightly angled ventral view. Abbreviation: Al Al supporting el- craniad. B2-3 elements are very wide ven- ement. trally and thin dorsally. The difference between the broad ventral ends and narrow dorsal ends of these elements produces a strong is thin and inserts on the lateral surfaces of distortion in the shape of the bronchi. B 1-2 M. tracheolateralis at Al 2-15. are dorsally fused together in an acute angle Innervation. Typical piprid pattern with and not by a short cartilaginous bar. The more ventral anastomosis at A25. distorted, spatulate ventral ends of B2-3 are acutely angled craniad to nest close to one Pipra serena (N = 2) another and to B 1 and Al. Subsequent B el- Supporting elements (fig. 18). Pipra serena ements are thin, simple, and not distorted in is similar in general form to P. suavissima, shape. but much larger and distorted in shape. The Muscles. These observations are based only diameter of the syrinx ofsuavissima at Al is on cleared-and-stained specimens in which 2.5-2.6 mm, and at A10, the diameter is 1.2- the musculature is only partially preserved. 1.3 mm. In nominate serena, the widths at Muscles are not very well developed, appar- Al and AlO are 3.8 or 3.9 mm and 2.6 or ently similar to those of P. suavissima. The 2.7 mm, respectively. In the two specimens ventral division of M. tracheolateralis is not of serena (one male and one female), the en- preserved. The muscle appears to insert on tire trachea and syrinx are twisted and dor- A2 and dorsally on Al as in one specimen soventrally compressed into a distorted oval ofsuavissima. M. sternotrachealis is thin and shape, and the tracheobronchial junction is inserts on the lateral sides of the trachea at displaced craniad by an extensive series of A14-15. double, complete bronchial A elements. Al Innervation. Not observed. is double, medially incomplete, and cartilaginous. It is very broad in its dorsal half and = consistently tapers toward the ventral end. Pipra pipra (N 17) A2 is double, medially complete, and ossified Supporting elements (fig. 19). Al is double, for its ventrolateral two-thirds. It is con- medially incomplete, and ossified for its ven- nected medially to A3 by a thin cartilaginous tral half. The dorsal and ventral ends of Al strip, but it is not truly complete medially. are wider than the lateral portions, resulting A3-5 are double, complete, and ossified ex- in a prominent arch-shaped indentation along cept for the medial third. The cartilaginous the caudal margin ofthe element. A2 is dou- medial portions ofA2-5 are connected by an ble and mostly ossified, but its medial bron- oval plate ofcartilage that is supported in the chial portions may be incomplete (N = 5), middle of medial wall of each bronchus by medially complete and cartilaginous (N = 4), the dorsal and ventral ends of A2-5. This or medially complete and ossified (N = 6). plate ofcartilage is buckled or depressed into A1-2 are fused for the ventral two-thirds of the lumen of each bronchus. A6 is single, their length. A3-4 (N = 12) or A3-5 (N = 3) 1 992 PRUM: MANAKIN SYRINGES AND PHYLOGENY 27

Fig. 19. Syringeal supporting elements of Pip- Fig. 20. Syringeal supporting elements ofPip- ra pipra (AMNH 9358). Left dorsal view, right ra fasciicauda (AMNH 2301). Left dorsal view, ventral view. Abbreviation: Al Al supporting el- right ventral view. Abbreviation: Al + A2 fused ement. Al and A2 supporting elements. are double, complete, and fully ossified. Sub- notrachealis is a broad, well developed sequent A elements are single, complete, and muscle, and it inserts on the lateral surfaces ossified. The left and right halves of the first of M. tracheolateralis from Al 1-17. The single element are oblique to the tracheal A dorsal fibers insert on the A elements at the elements, giving this element a prominent dorsal edge of M. tracheolateralis, while the chevron shape. In most specimens (N = 10), ventral fibers fan out obliquely over the ven- no tracheal A elements are fused dorsally, but tral surface of the trachea. Musculature in in five specimens the first two single A ele- females is very thin and undeveloped but the ments are dorsally fused by an ossified bar. same as in males in general form. In all three cleared-and-stained specimens, Innervation. Typical piprid pattern in most which are dorsally unfused, a free floating specimens. Ventral anastomosis at A20. In piece of cartilage is located between the first one specimen, the small lateral fibers of the and second single elements in the position of tracheal nerves exchange several tiny diago- the ossified dorsal bar in other specimens. A nal fibers across the ventral surface of the very narrow pessulus is fused dorsally and trachea between Al 5 and 25. ventrally to the first single element, A5 or A6. All the B elements are cartilaginous. B 1- Pipra aureola = 3), P. (N = 2 are a (N fasciicauda dorsally fused by cartilaginous bar. 5), and P. filicauda (N = 8) The ventral ends of B1-3 are widened and roundly curved craniomedially. Subsequent Supporting elements (fig. 20). All A ele- B's are unspecialized in shape. ments are completely ossified. A1-2 are dou- Muscles. M. tracheolateralis widens to ble and completely fused. A2 is complete. Al cover the ventral surfaces of the trachea be- is medially incomplete and somewhat tween A30-35. It continues caudally as a sin- straighter dorsally than A2, and its dorsal gle sheet of fibers, and differentiates into left ends form prominent rounded projections at and right halves between A8 and 10. The left the caudodorsal corners ofthe fused element. and right sides continue caudad adjacent to The caudomedial edge of A2 is sculpted in a one another on the ventral midline before distinctive double-arched curve. A3 is dou- separating laterally at A2-4. Each side di- ble, complete, and unfused. A4-5 are single vides into dorsal and ventral groups of fibers and completely fused into a short tracheal on the lateral midline at A2-4. The ventral drum. The thin ossified pessulus is fused dor- and dorsal fibers insert on the ventral end sally and ventrally to A4. A6 and subsequent and ventrolateral surface ofA1. These caudal elements are single, complete, ossified, and portions ofM. tracheolateralis are complete- unfused. All B elements are cartilaginous, ly extrinsic and differentiate without inter- thin, and fragile, and the bronchi are very mediate insertions or constrictions. M. ster- weakly supported. The dorsal ends of B1-2 28 AMERICAN MUSEUM NOVITATES NO. 3043

Fig. 21. Ventrolateral view of syringeal muscles: left Pipra filicauda (UMMZ 225067), right Pipra rubrocapilla (LSUMZ 114486). Abbreviations: I intrinsic syringeal muscles; S M. sternotrachealis; T M. tracheolateralis. run directly into one another and are fused the caudal edge ofAl exposed in ventral and in an arrowhead shape, instead of by a per- lateral view. M. sternotrachealis is a short but pendicular cartilaginous bar. These fused very robust muscle. The thickbundle offibers dorsal ends ofA1-2 are located very close to tapers rapidly to insert by a sheet of connec- the protruding dorsal end ofAl. The ventral tive tissue on the lateral surface of M. tra- end of B2 is widened, angled craniad under- cheolateralis at A12-15. Few M. sternotra- neath the ventral end of B 1, and fused to the chealis fibers are continuous with the M. medial side ofthat element. The blunt ventral tracheolateralis. In females, M. tracheolater- end ofB3 is turned abruptly craniad in a right alis is less well developed and massive. One angle. Subsequent B elements are thin and female specimen has weakly developed but unspecialized in shape. completely differentiated intrinsic muscles, Muscles (fig. 21). In males, M. tracheolat- and another has no intrinsic muscles. M. ster- eralis is a robust, well-developed muscle that notrachealis is thinner but the same in the widens ventrally to converge on the ventral form of insertion. midline between A30 and 35. It continues Innervation. Typical piprid pattern, with a caudally in single well-developed mass with- ventral anastomosis at A26. The main nerve out lateral differentiation until a partial in- fibers innervate the massive Mm. sternotra- sertion on A4-5. This insertion gives rise to cheales and smaller branches innervate the a pair of partial intrinsic muscles that are intrinsic musculature. differentiated on the ventral midline from their at A4-5. mus- origin The left and right Pipra mentalis (N = 7) cles form prominent bellies that are adjacent on the ventral midline, and insert along the Supporting elements (fig. 22). Al-2 are lateral and ventrolateral surfaces ofthe com- double, fully ossified, and completely fused. bined Al-2 elements. The intrinsic muscles Al is medially incomplete, and A2 is com- insert on the middle of the fused element, plete, except for a thin dorsomedial suture. approximately where A1-2 are fused, leaving The caudoventral and caudolateral margins l1992 PRUM: MANAKIN SYRINGES AND PHYLOGENY 29 of this fused element are flared outward and at right angles to one another, forming squared supports for the laterally spread bronchi. A3 (N = 2) or A3-4 (N = 5) are double, complete, and completely ossified. The medial portions ofthese complete double elements are straight Al + A2 instead of rounded. The next 2-4 cranial elements (A4-8) are single, complete, fully ossified, and fused extensively at their dorsal margins. All subsequent elements are single, complete, ossified, and unfused. The ventral ends of B1-3 are distinctly shaped into a prominent "fish hook" profile; they broaden Fig. 22. Syringeal supporting elements ofPip- ventrolaterally, turn acutely dorsomediad, ra mentalis (LSUMZ 95070). Left dorsal view, and gradually narrow toward the ventral tip. right ventral view. Abbreviation: Al + A2 fused They are closely nested together at the ventral Al and A2 supporting elements. margin of A1-2. The ventral end of Bl is ossified ventral to the bend in the element. Because of their widened ventral ends, the Pipra cornuta (N = 10) dorsal portions ofB 1-2 approach one anoth- Supporting elements. Similar to those of er at an acute angle and are directly fused Pipra mentalis (fig. 22) except as follows: Al- instead ofbeing connected by a separate car- 2 are completely fused and squared on the tilaginous bar. B4 and subsequent elements caudoventral margin, but the elements are are not specialized in shape. more robust in shape and the lateral flare of Muscles. In adult males, M. tracheolater- the bronchi is reduced. A2 is medially in- alis converges ventrally to cover the entire complete. A3 is double, complete, and fully ventral surface of the trachea, and continues ossified. The dorsomedial corner of A3 is caudad in a single, undivided sheet of fibers broadened caudally to nest next to the cranio- to insert partially on the ventral and lateral dorsal end of the medially incomplete A2. surfaces of the cranial margin of the drum A4-6 are fused extensively along their dorsal (A6-8). A pair ofintrinsic muscles originates margins into a partial drum. The ventral end caudal to this insertion and is laterally dif- of B1 is similar in shape but is thinner, and ferentiated. The left and right intrinsic mus- less acutely angled (about 900). The ventral cles are strongly developed into prominent end ofB I is ossified beyond the dorsomedial bellies that insert on the entire length of Al. bend. The ventral ends of B2-3 are smaller, In some specimens, M. sternotrachealis orig- cartilaginous, and not acutely angled. inates from a pair of fleshy raphes on the Muscles. In males, M. tracheolateralis is medial surface ofthe craniolateral process of well developed, covering the ventral surface the sternum. The two origins fuse immedi- ofthe trachea above A30. The left side of M. ately to form a single well-developed muscle. tracheolateralis is more well developed, pro- M. sternotrachealis inserts on the lateral sides ducing a pronounced asymmetrical bulge in of M. tracheolateralis at Al 5-18. Dorsally a musculature left of the ventral midline. On few fibers insert on the A elements at the the lateral surfaces, M. tracheolateralis in- dorsal margin of M. tracheolateralis, and serts on A6 and gives rise to a pair ofintrinsic ventrally, a few fibers fan out across the ven- muscles; however, the ventral fibers continue tral surface of the trachea. In females, M. caudad without any intermediate insertion to tracheolateralis is a thin, undeveloped sheet insert on the ventral ends of Al. These ven- offibers. It does not insert on the cranial edge tral fibers are not laterally differentiated until of the drum, but it divides on the ventral immediately prior to their insertion on Al. midline at A6-8, and inserts on Al without The well-developed left and right intrinsic any prominent intrinsic bellies. muscles insert on the lateral and ventrolateral Innervation. Typical piprid pattern. Ven- surfaces ofthe syrinx, whereas the caudal ex- tral anastomosis between A30 and 35. trinsic fibers of M. tracheolateralis cover the 30 AMERICAN MUSEUM NOVITATES NO. 3043

sified. B4 and subsequent elements are not specialized in shape. Muscles. Similar to those ofP. rubrocapilla (fig. 21). M. tracheolateralis inserts in a con- + tinuous sheet on A8-9 laterally and on A5- 6 at the ventral midline. The insertion is chevron shaped. The left and right intrinsic muscles originatejust caudal to this insertion, the ventral fibers originate on A4, and the lateral fibers on A8-9. A few of the most Fig. 23. Syringeal supporting elements ofPip- on dorsal to ra erythrocephala (AMNH 8081). Left dorsal view, dorsal fibers originate A8-9 just right ventral view. Abbreviation: Al + A2 fused the inserting fibers of M. tracheolateralis. M. Al and A2 supporting elements. sternotrachealis has a single origin, and inserts on the lateral and ventrolateral surfaces of M. tracheolateralis between A14 and 19. In females, the musculature is thin and un- ventral eighth ofthe syringeal circumference. developed as in female mentalis, but the Musculature in female specimens is weakly change in fiber direction before insertion of developed, and is similar generally in shape the left and right halves ofM. tracheolateralis to that offemale Pipra mentalis. M. tracheo- on Al reflects the chevron-shaped origin lateralis is not differentiated into intrinsic found in male intrinsic musculature. muscles. Its fibers extend caudally to insert Innervation. Typical piprid pattern. Ven- on A1 without any significant partial inser- tral anastomosis at A30. tion on the drum. M. sternotrachealis has a single origin, and inserts on the lateral surface Pipra rubrocapilla (N = 9) of M. tracheolateralis at Al 3-16. Innervation. Typical piprid pattern. Ven- Supporting elements. As in Pipra chloro- tral anastomosis near A30. The main fibers meros except as follows: A2 is complete but divide into a pair ofprominent smaller fibers medially cartilaginous. The dorsomedial su- caudal to the insertions of Mm. sternotra- ture between the medial and dorsal portions cheales. They continue caudad to innervate of A2 is still visible. A5-7 (N = 4) or A5-8 the intrinsic muscles, and innervate the Mm. (N = 4) are extensively fused dorsally into a sternotracheales from their medial surface. partial drum. Ventral ends of both Bl and B2 are ossified. Pipra chloromeros (N = 10) Muscles (fig. 21). As in Pipra chloromeros. M. tracheolateralis inserts and intrinsic mus- Supporting elements. Generally similar to culature originates on A7-8, at the cranial those of P. mentalis (fig. 22). A1-2 are dou- margin of the partial drum. M. sternotra- ble, ossified, and completely fused. Al is me- chealis inserts on the lateral and ventrolateral dially incomplete. A2 is complete and has a surface of M. tracheolateralis at A12-16. distinct dorsomedial suture. A3-4 are dou- Innervation. Typical piprid pattern. Ven- ble, complete, and completely ossified. A5- tral anastomosis at A25. 6 (N = 1 1) or A5-7 (N = 2) are single, ossified, and extensively fused at the dorsal margins. Pipra erythrocephala (N = I 1) A7 or A8 and subsequent elements are single, complete, ossified, and unfused. An ossified Supporting elements (fig. 23). As in Pipra pessulus is formed by the dorsomedial por- chloromeros except as follows: The dorsal tion of A5 and is fused ventrally to A5. The ends of Al are cartilaginous. A2 is complete ventral ends ofB1-3 are widened and angled but partially cartilaginous medially. In most dorsomediad into "fish hook" shapes, which specimens the dorsomedial suture is observ- are more exaggerated than in P. mentalis. The able. The bronchi are more laterally flared, bronchi are more laterally flared than in men- caudally exposing the medial walls of the talis. The ventral half or third of B 1 is ossi- bronchi. A5-6 are fused only by a narrow bar fied, and the ventral ends of B2 are also os- on the dorsal midline. Subsequent A ele- 1992 PRUM: MANAKIN SYRINGES AND PHYLOGENY 31 ments are unfused. B 1 is very robust; because confidently whether they are derived or prim- of the lateral spread of the bronchi, the itive for this character, and they are coded as rounded lateral portion of B1 elements arch unknown (?). craniad between their ventral and dorsal ends. 2. Double, complete, fully ossified A3-5 el- Muscles. As in Pipra chloromeros and rub- ements. In Chloropipo, Xenopipo, Heterocer- rocapilla (fig. 21), except as follows: M. ster- cus, Pipra pipra, the Pipra aureola species notrachealis inserts on the M. tracheolateralis group, and the Pipra erythrocephala species at A15-19. group, one or more of the double, complete Innervation. Typical piprid pattern. Ven- A elements are fully ossified. The number of tral anastomosis at A30. Innervates both the fully ossified, double A elements varies among intrinsic muscle and the M. sternotrachealis. these taxa from one to three. In the Pipra serena species group, Machaeropterus, and SYRINGEAL CHARACTER ANALYSIS Manacus, the double complete A elements The characters are organized into sub- are medially cartilaginous. In syringeal de- headings A Elements (1-19), B Elements (20- velopment of tyrannoids, formation of car- 39), Accessory Cartilages (40-42), Syringeal tilaginous supporting elements precedes os- Shape (43-44), Extrinsic Musculature (45- sification (Ames, 1971). The medially ossified, 47), Intrinsic Musculature (48-56), Inner- double A elements are apparently a terminal vation (57-5 8), and Membranes (59). In each addition to the ontogeny ofthe complete but character description, derived state and its medially cartilaginous double A's found in distribution are followed by the primitive other piprids. Completely ossified double A state and its distribution in ta- elements are not present in any cotingids and (summarized only in a few tyrannids in which they are ble 2). See Methods for details. hypothesized as independently derived. The A ELEMENTS completely ossified double A's are hypothesized to be derived by ontogenetic criteria 1. Double, complete A elements. In all pip- (Fink, 1982) and by outgroup comparison. rids except Masius, Corapipo, and Ilicura, 3. Double, complete A2 elements. In Het- one or two A elements are double and com- erocercus, Pipra pipra, the Pipra aureola spe- plete, providing support to the medial bron- cies group, and the Pipra erythrocephala spe- chial walls. Three double, complete A ele- cies group (except P. cornuta), the A2 elements ments are present in Pipra serena and a few are double and complete. This morphology individuals ofPipra pipra. Double, complete is absent in other piprids, cotingids, and al- A elements are found in some cotingids (Pip- most all tyrannids, and is here hypothesized reola, Ampelioides, Porphyrolaema, Cotinga, to be derived. Conioptilon, and Procnias) and in some tyr- 4-5. Double, complete, medially ossified A2 annid clades. In the tyrannid clades, medially elements. In the Pipra aureola species group, complete A elements have been hypothesized and in P. mentalis and P. chloromeros, the to be independently derived (W. E. Lanyon, A2 elements are completely ossified medi- 1984a, 1985, 1986, 1988a, 1988c). Since both ally. In P. mentalis and P. chloromeros, the ingroup states are present in the immediate medial portion of A2 connects to the dorsal outgroup to piprids, the cotingids, it is also and lateral portion in a conspicuous dorso- necessary to refer to more distant outgroups. medial suture. In P. erythrocephala and P. Based on outgroup comparison to fumarioids rubrocapilla, the medial portions of A2 are and Old World suboscines and to the oscines, cartilaginous, but the suture is still present, medially complete, double A elements are indicating that in these taxa such elements derived within tyrannoids. These medially have become secondarily cartilaginous. In complete, double A elements in piprids are Pipra pipra and Heterocercus, the complete, hypothesized here to be derived indepen- double A2's are variably ossified and not as dently ofthose in cotingids. The piprids Chi- robust or consistent as in the Pipra species roxiphia and Antilophia are so specialized in above. These medially ossified A2's are ab- the morphology of their tracheobronchial sent in other piprids, cotingids, and most tyr- junction that it is not possible to determine annids, and are hypothesized here to be de- 32 AMERICAN MUSEUM NOVITATES .NO. 3043

TABLE 2 Distribution in the Piprids of the Derived States of 59 Syringeal Morphology Characters (Characters are coded: 0 primitive state; 1 derived state; 2 alternative, unordered derived state. Character descriptions and polarity assessments are given in the text.) Characters Species 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 2122 23 24 25 Corapipo gutturalis 0 0 0 0 0 0 0 Q 0 0 0 1 0 0 0 0 0 O 0 1 0 0 0 0-0 Corapipo leucorrhoa 0 0 0 0 0 0 0 0 0 O 0 1 0 0 0 0 0 O 0 1 0 0 0 0 0 Masius chrysopterus 0 0 0 0 0 0 0 0 0 O 0 1 0 1 0 0 0 O 0 1 0 0 0 0 0 Ilicura militaris 0 0 0 0 0 0 0 0 0 0 O 0 1 0 0 O 0 0 0 1 0 0 0 0 0 Machaeropterus deliciosus 1 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 O 0 1 0 0 o0 00 Machaeropterus regulus 1 0 0 0 0 0 0 0 0 0 0 0 2 0 0 0 O 0 1 1 0 0 0 0 0 Machaeropterus pyrocephalus 1 0 0 0 0 0 0 0 0 0 0 0 2 0 0 00 0 0 1 0 0 0 0 0 Manacus manacus 1 0 0 0 0 1 0 0 00 0 0 0 0 0 0 00 11 0 0 1 0 0 Manacus vitellinus 1 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 1 1 -00 1 0 .0 Manacus candei 1 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 O 0 1 1 0 0 1 0 0 Chiroxiphia linearis ? 0 0 0 0 0 0 0 0 0 0 01 0 0 0 1 0 1 1 0 0 0 1 1 Chiroxiphia lanceolata ? 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 1 0 1 1 0 0 0 1 -1 Chiroxiphia pareola ? 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 1 0 1 1 0 0 0 1 1 Chiroxiphia caudata ? 0 0 0 0 0 0 0 0 0 00 0 10 0 0 1 0 1i1 0 0 0 1 1 Antilophia galeata ? 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 1 0 1 1 0 0 0 1 0 Chloropipo holochlora 1 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 O 0 1 0 0 0 0 0 Chloropipo unicolor 1 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 O 0 1 0 0 0 0 0 Chloropipo uniformis 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 1 0 1 0 0 0 Xenopipo atronitens 1 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 1 0 1 0 1 0 0 0 Heterocercus linteatus 1 1 1 0 0 0 1 0 0 0 0 0 0 0 1 0 0 O 0 1 0 0 0 0 0 Heterocercusflavivertex 1 0 0 0 1 0 0 0 0 0 0 0 1 0 0 O 0 1 0 0 0 00 Pipra pipra 1 1 0 0 0 1 0 0 0 1 0 0 0 1 0 0 O 0 1 0 0 0 0 0 Pipra coronata 1 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 O 0 1 0 0 0 0 0 Pipra isidorei 1 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 1 O 0 0 0 0 Pipra coeruleocapilla 1 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 O 0 1 0 0 0 00 Pipra nattereri 1 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 1 0 0 0 0 0 Pipra iris 1 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 O 0 1 0 0 0 0 0 Pipra suavissima 1 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 1 1 0 0 0 0 0 Pipra serena 1 0 0 0 0 0 0 0 0 0 0 0 1 0 0 O 00011 0 0 0 0 0 Pipra aureola 1l1 1 0 0 1 1 0 1 0 0 0 0 1 1 0 0 0 1 1 0 0 0 0 Pipra fasciicauda 1 1 1 1 0 0 1 1 0 1 0 0 0 0 1 1 0 0 0 1 1 0 0 0 0 Pipra filicarda 1l1l11 0 0 1 1 0 1 0 0 0 0 1 1 0 0 0 1 1 0 0 0 0 Pipra cornuta 1 0 0 0 0 1 1 1 0 0 0 0 0 1 1 0 O 0 1 0 0 0 0 0 Pipra mentalis l1 11 0 0 1 1 1 0 0 0 0 0 1 1 0 0 0 1 0 0 0 0 0 Pipra chloromeros 1 1 1 1 0 0 1 1 1 0 0 0 0 0 1 1 0 0 0 1 0 0 0 0 0 Pipra rubrocapilla 1l1l1 1 0 1 1 1 0 0 0 0 0 1 1 0 0 0 1 0 0 0 0 0 Pipra erythrocephala 1 1 1 1 1 0 1 1 1 0 0 0 0 0 1 0 0 O 0 1 0 0 000 Outgroups 0 0 0 0 0 0 0 0 0 a0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 rived in the Pipra aureola and P. tain their distinct characters; the lines of fu- erythrocephala species groups (4). The car- sion between the elements and their shapes tilaginous medial portions of A2, which re- are easily discernible. The outer surfaces of tain the medial suture, are hypothesized to Al-2 are also distinctly rough in texture. In be secondarily derived in P. erythrocephala Machaeropterus deliciosus, Heterocercus, and and P. rubrocapilla (5). Pipra pipra, Al-2 are fused for the ventral 6-8. A1-2 partially or completely fused. In half to two-thirds of their length and at their Manacus, A1-2 and sometimes A1-3 are ex- dorsal ends into single composite elements. tensively fused together, but the elements re- In the Pipra aureola and P. erythrocephala 1992 PRUM: MANAKIN SYRINGES AND PHYLOGENY 33 ,

TABLE 2-(Continued)

Characters 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 0 1 0 0 0 0 0 0 O 0 1 0 1 0 1 1 0 0 0 0 1 0 0 1 1 0 0 0 0 0 0 1 0 0 0 1 0 00O0'0 0 0 1 0 1 0 1 0 0 0 0 0 1 0 0 1 1 0 0 0 0 0 0 1 0 0 0 1 0 0 0 0 0 0 0 0 1 0 0 0 1 0 0 0 0 0 0 O 0 1 1 0 0 0 0 0 0 1 0 0 0 1 1 0 00 0 0 0 0 0 0 0 0 1 0 0 0 0 0 O 0 1 0 1 0 0 0 0 0 0 1 0 0 0 1 0 0 1 0 0 0 0 Q 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 O 0 1 0 0 0 1 0 0 1 Q 00 0 0 0 0 0 0 0 0 0 0 0 1 1 0 0 0 0 1 0 0 0 0 0 1 0 0 0 1 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0. 0 1 0 0 0 0 0 0 0 0 0 O 0 1 0 0 0 1 0 0 1 1 0 0 0 0 000 0 0 0 0 0 0 0 0 0 0 0 O 0 1 0 0 0 0 1 0 0 0 1 0 0 1 1 0 0 0 0 0 0 0 0 0 0 0O0 0 0 0 0 0 0 O 0 1 0 0 0 0 1 0 0 0 1 0 0 1 1 0 0.0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 O 0 1 0 0 0 0 1 0 0 0 1 0 0 1 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 O 0 1 1 0 0 0 1 0 0 0 1 O 0 1 0 0 0 000 0 0 0 O 0 1 0 0 0 0 0 0 0 O 0 1 1 0 0 0 1 0 0 0 1 0 0 1 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 O 0 1 1 0 0 0 1 0 0 0 1 0 0 1 0 0 0 0 0 0 0 0 0 0 0 1 0 00 0 0 0 0 O 0 1 1 0 0 0 1 0 0 0 1 0 0 1 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 O 0 1 1 0 0 0 1 0 0 0 1 0 0 1 0 0 0 Q 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 2 0 0 2 0 0 0 1 0 0 1 0 1 0 0 0 0 1 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 2 0 0 2 0 0 0 1 0 0-0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 2 0 0 2 0 0 0 1 0 0 1 0 0 0 0 1 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 2 0 0 2 0 0 0 1 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 O 0 1 0 0 1 0 0 0 1 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 O 0 1 00 1 0 0 0 1 0 0 1 0 0 0 0 0 000 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 O 0 1 0 0 0 1 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 O 0 1 0 0 0 1 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 "10 0 0 1 0 0 1 0 0 0 0 0 0.0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 O 0 1 0 0 0 1 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 O 0 1 0 0 0 1 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 O 0 1 0 0 0 1 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 O 0 1 0 0 0 1 0 0 1 0 0 0 QO00 0 0-0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 O 0 1 0 0 0 1 0 1 1 1 0 0-0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 1 0 0 1 1 0 0 1 0 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 1 0 0 1 1 0 0 1 0 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 1 0 0 1 1 0 0 1 0 0 100 0 0 0 1 0 0 0(0 00 0 0 0 0 0 0 0 0 O 0 1 1 0 1 0 1 0 1 0 0 1 0 0 1 1 0 0 1 0 0 0 0 0O0 0 0 0 0 0 0 0 0 O 0 1 0 0 1 0 0 0 1 0 0 1'0 0 1 1 0 0 1 0 1 0 0 0 0 0 0 0 0 0 0 0 0 O 0 1 0 1 1 0 0 0 1 0 0 1 O 0 1 1 0 0 1 0 1 0 0 0 0 0 0 0 0 0 0 0 0 O 0 1 0 1 1 0 0 1 1 0 0 1 O 0 1 1 0 0 1 0 1 0 0 0 0 0 0 0 0 0 0 0 0 O 0 1 0 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 groups, the A1-2 elements are completely hypothesized as independently derived (6). fused forming a single composite element The partially and completely fused states are without apparent distinctions between the el- hypothesized to be derived in the following ements. Fused double A elements are absent transition series: (7) A1-2 partially fused into in other piprids, cotingids, and most tyran- a single composite element, derived in Ma- nids, and are-hypothesized to be derived in chaeropterus deliciosus, Heterocercus, Pipra piprids. The A1-2 fusion ofManacus differs pipra, the Pipra aureola, and P. erythroceph- markedly in character and detail from the ala species groups; (8) A1-2 completely fused fused A1-2 ofthese other piprids and is here into a single, composite element; derived in 34 AMERICAN MUSEUM NOVITATES NO. 3043 the Pipra aureola and P. erythrocephala spe- terus, Xenopipo atronitens, Chloropipo cies groups. holochlora, Chloropipo unicolor, Pipra eryth- 9. Caudoventral and caudolateral margins rocephala, and some individuals ofMachaer- of Al flared and square. In the Pipra eryth- opterus deliciosus, and Pipra pipra. In Mas- rocephala species group, the caudoventral and ius, this dorsal bar is a direct, cranial extension caudolateral margins of the Al elements are ofthe pessulus which is fused dorsally to the flared outward and at right angles to one an- first two dorsally complete elements, whereas other, forming squared supports for the latin these other taxa, the dorsal bar is not a erally spread bronchi. This morphology is continuation ofthe pessulus. Single, tracheal unique in tyrannoids and is hypothesized to A elements can also be fused extensively along be derived. their dorsal margins (as in some Chloropipo 10. Double arch-shaped indentation on cau- holochlora and Heterocercus), or completely domedial edge of A2. In the Pipra aureola fused dorsally (as in the Pipra erythrocephala species group, the caudomedial edge ofA2 is species group except erythrocephala), or, they "indented" or "incised" in a double arch can be fused entirely along their dorsal and shape or double s-curve. This morphology is ventral margins (as in the Pipra aureola spe- unique in tyrannoids and hypothesized as de- cies group). Fused tracheal elements are not rived. present primitively in cotingids, and are ab- 11. Arch-shaped caudolateral edge of Al. sent in tyrannids. The fused tracheal drum In Pipra pipra, the caudolateral edge of Al of Chiroxiphia and Antilophia is composed is indented or arched craniad (i.e., the ele- of a different series of A elements and is dif- ment thins at the lateral midline). This unique ferent in many other details; it is here hy- morphology is hypothesized to be derived. pothesized to be independently derived (see 12. Al element broad and expanded. In 17). The coding of the variation in this de- Masius and Corapipo, Al is a robust, broad rived character in piprids is made more com- element. The mode of ossification produces plex by several factors, including the different a pronounced bulge or inflation of the ele- types or degrees of fusion, intraspecific vari- ment. This morphology is unique in tyran- ation in fusion, and apparent reversals in de- noids and is hypothesized to be derived. gree of fusion. However, morphological de- 13. Completely or largely cartilaginous Al. tails indicate that these tracheal fusions In Ilicura, Chiroxiphia, Antilophia, and the evolved in a transition series from lack of Pipra serena species group, the Al element fusion, to fusion by a dorsal bar, to extensive is completely cartilaginous. In Machaerop- or nearly complete dorsal fusion. In species terus regulus and M. pyrocephalus, the Al with extensive dorsal fusion, the most com- element is almost completely cartilaginous plete fusion occurs at the dorsal midline and except for a conspicuous ossified oval in the a remnant ofa dorsal bar is observable. Like- center of the ventral quarter of the element. wise, in species with extensive or complete In almost all other tyrannoids, the Al ele- fusion, occasional anomalies or interruptions ments are completely ossified. Both the com- in fusion occur between the ventral or lateral pletely cartilaginous and the mostly cartilag- margins of elements, but not at the dorsal inous state of Al are derived in piprids, but margins. In individuals of Pipra pipra that the state found in Machaeropterus regulus and lack dorsal fusion, independent (but nor- pyrocephalus may be intermediate to the mally fused) floating pieces of cartilage are completely cartilaginous Al or it may be in- present between the elements in the position dependently derived. The completely carti- of the dorsal bar, indicating that these indi- laginous (13.1) and mostly cartilaginous (13.2) viduals are anomalous, incompletely fused states of the Al element will be coded here specimens within a derived taxon. These tra- as unordered derived states of a single com- cheal fusion characters will be coded as the plex character. following derived characters (the latter two 14-16. Fused, single A elements. In some form a transition series): piprids, a pair or series of single, tracheal A (14) A5-6 or A6-7 fused dorsally by a cra- elements is fused. These elements are fused nial extension of the pessulus; derived in by an ossified dorsal bar in Masius chrysop- Masius. 1992 PRUM: MANAKIN SYRINGES AND PHYLOGENY 35

(15) Partial or extensive dorsal fusion ofA gle, tracheal A elements are dorsally incom- elements by a dorsal bar; derived in Chlo- plete or dorsally cartilaginous. The number ropipo holochlora, C. unicolor, Xenopipo, of elements and the series vary among taxa: Heterocercus, and Pipra (except the Pipra se- Manacus (A5-10 or 12), P. serena (A6-1 1), rena species group). P. suavissima (A6-8), Machaeropterus regu- (16) Dorsal and lateral fusion or complete lus (A5-6), Chiroxiphia (A4-5 to A6-8), and fusion; derived in a transition series from (15) Antilophia (A6-7). In two specimens of P. in the Pipra aureola species group and the coronata, A5 is single and dorsally cartilag- Pipra erythrocephala species group (except inous. It is difficult to distinguish dorsally erythrocephala). incomplete and dorsally cartilaginous A el- 17. Extensive tracheal drum. In Chiroxi- ements within or among species because the phia and Antilophia, an extensive tracheal cartilaginous dorsal area may appear to be drum is formed by the partial or complete either an organized continuation of the ele- fusion of a series of six to nine single A elements or a generalized strip of cartilage re- ments between A6 and 17. In most individ- lated to the pessulus or accessory cartilages. uals, the cranialmost elements are entirely However, dorsally cartilaginous tracheal A fused and difficult to differentiate, whereas elements are not found in any other piprids the caudal elements of the drum are usually or cotingids. In tyrannids, they have been only fused dorsally. The caudodorsal margin hypothesized to be independently derived in of the drum often has a short caudal projec- tody-tyrants (Lanyon, 1988c). Dorsally car- tion between the most cranial, dorsally in- tilaginous, single A elements in piprids are complete pair of A elements. This morphol- here hypothesized to be derived. The one ogy is unique in tyrannoids and is here variable taxon, Pipra coronata, is coded as hypothesized as derived. primitive or absent for this character because 18. Chevron-shaped A4 element at tracheo- only a small minority of the specimens have bronchial junction. In Chloropipo uniformis this trait. and Xenopipo atronitens, the left and right halves of the single A4 element are oriented parallel to the double, bronchial A's and B ELEMENTS oblique to the more cranial, single, tracheal 20. B1-2 dorsally fused. In all piprids, B1- A's. The left and right halves are dorsally 2 are dorsally fused by a short cartilaginous fused in a conspicuous chevron shape. The bar. This morphology is unique in suboscines caudodorsal end ofthis element extends me- and has been hypothesized to be a synapo- diad and forms the pessulus. In a few C. uni- morphy of the family (Prum, 1 990b). formis and X. atronitens, the single A4 ele- 21. B1-2 fused directly. In Pipra aureola, ment is ventrally incomplete. In both C. P. fasciicauda, and P. filicauda, the B1-2 el- uniformis and Xenopipo atronitens, the cau- ements are directly fused at their dorsal ends dodorsal end of A5 extends caudad into the in an acute angle, or arrowhead shape, and space cranial to the chevron-shaped A4. A not by a distinct cartilaginous bar. This unique few other piprids have obliquely oriented, morphology is hypothesized to be derived fused halves ofthe first single A element (e.g., secondarily from (20). Pipra pipra), but in none ofthese species are 22. B1-2 broadly fused dorsally. In Chlo- these elements ventrally incomplete, strongly ropipo uniformis and Xenopipo atronitens, the distorted, or a major contribution to the pes- dorsal ends of B1-2 are broadly fused by a sulus. Also, in none of these does the next, wide cartilaginous connection. This mor- cranial element extend caudally in a similar phology is unique among tyrannoids and is manner. This unique morphology in C. uni- hypothesized to be derived secondarily from formis and Xenopipo atronitens is hypothe- (20). sized to be derived. 23. Caudal series of B elements dorsally 19. Dorsally incomplete or cartilaginous fused. In Manacus, a series of caudal B ele- tracheal A elements. In Manacus, Pipra sere- ments (B5-10 to B5-12) are broadened and na, P. suavissima, Machaeropterus regulus, flattened, and their widened dorsal ends are Chiroxiphia, and Antilophia, a series of sin- fused to one another in a lattice. This mor- 36 AMERICAN MUSEUM NOVITATES NO. 3043 phology is unique in tyrannoids and hypoth- (31) Ventral ends ofB3 reduced to a short, esized as derived. blunt process angled craniad at 900; derived 24-25. Al dorsally fused to Bl-2. In Chi- and present in Manacus and the Pipra au- roxiphia and Antilophia, Al is dorsally fused reola species group. to Bl-2 by a thin strip of cartilage. In Chi- (32) B1-3 elements very thin and ventral roxiphia, the strip is longer than in Antilophia ends of B 1-3 angled dorsad, as in other pip- and it has a distinct square projection on its rids, but distinctly thinner and spindly in craniolateral edge which is absent in Antilo- shape; derived in Chloropipo unicolor. phia. This fusion is absent in all other piprids (33-34) In all of the Pipra erythrocephala and all other tyrannoids, and is here hypoth- species group except cornuta, ventral ends of esized to be derived in Chiroxiphia and An- B 1-3 are very broad, acutely angled dorsad, tilophia (24). The presence of the longer car- and pointed at the tip in a unique "fish hook" tilaginous Al-B1 bridge with the square shape. These specialized ventral ends are craniolateral projection in Chiroxiphia is hy- nested closely together and next to the cranio- pothesized to be secondarily derived in that ventral margin of the laterally and ventrally genus (25). flared Al (see 9). In P. cornuta, the B1 ele- 26. Lateral section of Bi arched craniad. ment is prominently crooked in a right angle In P. erythrocephala, the lateral portion of that is less exaggerated than the others in this the double, medially incomplete B1 element species group. The B2-3 elements ofcornuta is arched craniad. As a result, the B 1-2 meet are not acutely angled dorsad or "fish hook" dorsally at an acute angle, and are fused di- shaped, but are more similar to the less spe- rectly to one another, instead of being fused cialized shapes present in Chloropipo holo- by a perpendicular cartilaginous bar. This chlora, the Pipra serena species group, and unique, derived morphology is associated Machaeropterus deliciosus, which are prob- with the extreme lateral flare of the bronchi ably primitive in the group. The ventral ends in this species, and is hypothesized to be de- ofB 1-3 in the Pipra erythrocephala group are rived. hypothesized to be derived in an ordered 27-35. Specialized ventral ends of Bl-3. transition series: (33) ventral end of B1 wid- The ventral ends of B1-2 of all piprids are ened and angled into a large hook shape, de- widened, paddle shaped, or angled in some rived in the Pipra erythrocephala group; (34) specialized manner. The B3 of the majority ventral ends ofB 1 further distorted and ven- ofpiprids is also specialized in a similar man- tral ends ofB2-3 widened and angled acutely ner, and the three elements are typically nest- into "fish hook" shapes, derived in Pipra ed together next to the caudoventral margin erythrocephala, rubrocapilla, chloromeros, of A1. No similarly shaped elements occur and mentalis. in any ofthe outgroups examined, making it (35) In Chloropipo uniformis and Xenopipo difficult to polarize the variation in these atronitens, ventral ends of B1-2 asymmet- structures within the group. Some restricted rically widened. The caudal margin is straight, variations, however, appear to be derived but the cranial margin veers craniad at the within piprids because of their highly dis- ventral end of the element producing a lop- tinctive shapes. This set of morphologies is sided, spatulate, ventral end to the element. coded here as the following binary characters This morphology is unique among piprids in a complex transition series: and is hypothesized to be derived. (27) Ventral ends ofB 1-2 widened and an- 36-39. Bi or Bl-2 elements partially os- gled dorsad or craniad; derived in all piprids. sified. In Masius, Chloropipo unicolor, Xeno- (28) Specialized ventral ends of B 1-2 re- pipo, and the Pipra erythrocephala species duced but still present; derived in Ilicura. group, the ventral quarters or ventral ends of (29) Ventral end of B2 angled craniad un- the B 1 elements are ossified. In Corapipo, the derneath B1 and fused to the ventromedial broad B1 element is almost completely os- surface of B 1; derived in the Pipra aureola sified. The ossified portions of B1 elements species group. in Masius and Corapipo are enlarged or swol- (30) B3 widened and angled craniad with len in a unique manner. In Pipra rubrocapilla, B1-2; derived in all piprids except Ilicura, P. erythrocephala, P. chlaromeros, and Chlo- Masius, Corapipo, and Chloropipo uniformis. ropipo unicolor, the ventral ends of the B2 1992 PRUM: MANAKIN SYRINGES AND PHYLOGENY 37 elements are also ossified. In all other piprids piprids and is here hypothesized to be de- and most other tyrannoids, the B elements rived. Similar cartilage sheets are present in are completely cartilaginous. In some cotin- Neopelma, Tyranneutes, Tityra, Lipaugus gids and tyrannids, the central portion ofthe (sensu stricto; Prum, 1990b), and Calypto- more caudal B elements are ossified, but these mena (Eurylaimidae; R. 0. Prum, in prep.), morphologies are independent of those of but these structures are hypothesized to be piprids. The partially ossified B elements in independently evolved. piprids are hypothesized to be derived in the following transition series of binary charac- SYRINGEAL SHAPE ters: 43. Trachea and syrinx greatly enlarged (36) Ventral ends of B1 elements ossified and distorted in shape. The syrinx of Pipra and enlarged; derived in Masius and Cora- serena is similar to that ofP. suavissima but pipo. the trachea and bronchi are greatly enlarged (37) Ventral ends of Bl elements ossified and twisted. Associated with this increase in but not enlarged; derived in Chloropipo uni- size are a number of changes or distortions color, Xenopipo atronitens, and Pipra eryth- of primitive morphological details. In suav- rocephala species group. issima and other members ofthe Pipra serena (38) B 1 almost completely ossified; derived species group, there are small cartilaginous in Corapipo. connections between the cartilaginous medial (39) Partially ossified ventral ends of B2 sections ofthe double, complete A elements. elements; derived in Pipra erythrocephala, P. In serena, an expanded oval plate ofcartilage rubrocapilla, and P. chloromeros. is supported in the medial bronchial wall by the cartilaginous medial sections of these ACCESSORY CARTILAGES double A's. The number ofdorsally cartilaginous single A's increases from three to six 40-41. Medial bronchial cartilage bars. In in suavissima to nine or ten in serena. The Masius, Corapipo, and Ilicura, a pair of ac- shape of the B1-3 elements is also tremen- cessory cartilaginous bars is fused to the dor- dously distorted by increase in the diameter sal ends of Al-2 and continues ventrad or of the bronchi. This suite of derived char- cranioventrad toward the tracheobronchial acters is clearly an autapomorphy of serena. junction and the pessulus. These paired me- Its presence in serena implies that the reor- dial bronchial cartilages form the cranial or ganization of primitively present morpho- craniodorsal margin of the medial tympani- logical details the result form membrane. These structures are unique is of a single, radical in tyrannoids and hypothesized to be derived change in syringeal ontogeny in this species. in these three genera (40). In Corapipo gut- 44. Tracheobronchial junction flared into turalis, the medial bronchial cartilages are en- funnel shape. In Chloropipo unicolor, the edg- larged and ossified. This morphology is hy- es of A1-7 are closely fitted next to one an- pothesized to be additionally derived in this other; the caudal margins of these elements species (41). (especially A1-2) are larger in diameter than 42. Accessory cartilaginous sheet. In Chi- the cranial edges, so that the entire syrinx roxiphia and Antilophia, a sheet of cartilage flares laterally in an inverted funnel shape. forms the caudoventral surface ofthe trachea This morphology is unique among tyran- and the ventral and medial walls ofthe bron- noids and is here hypothesized to be derived. chi at the tracheobronchialjunction. The sheet is connected to the dorsal ends ofall dorsally incomplete elements (from A1-6 to A1-8), EXTRINSIC MUSCULATURE and is also connected to the ventral ends of 45. Mm. tracheolaterales highly asymmet- A1-2. The tracheal portion of the cartilagi- rical. In Machaeropterus, the left and right nous sheet is bowed outward or cylindrically Mm. tracheolaterales are asymmetrically de- rounded, and its caudal margin in both bron- veloped. The right side is underdeveloped; chi is indented craniad where it supports the the left side contributes almost all ofthe fibers cranial edge of the medial tympaniform to the ventral sheet of muscle that is usually membrane. This morphology is unique in formed by the union ofthe left and right sides. 38 AMERICAN MUSEUM NOVITATES NO. 3043

The right M. sternotrachealis partially inserts are no morphological details to support the onto the lateral surface of the tracheal A el- hypothesis that the entire class of"intrinsic" ements because the fibers of the right M. musculature novelties in piprids are homol- tracheolateralis are underdeveloped or miss- ogous. Variation in the intrinsic syringeal ing. Many other piprids have limited musculature of piprids is difficult to polarize muscular asymmetry associated with the because the immediate outgroup, the cotin- asymmetrical position of the trachea to the gids, generally lacks intrinsic syringeal mus- mid-sagittal plane, but in no other piprids is cles, and the more distant outgroups, the ty- the asymmetry as extreme or consistent rannids and furnarioids, have very different among individuals as in Machaeropterus. This syringeal muscles that are unlikely to be ho- condition is not found in any cotingids or mologous with those ofpiprids (Ames, 1971; tyrannids and is here hypothesized to be de- Prum, 1 990b). A combination of inferences rived. from outgroup comparison, differences in the 46. Ventral division of M. tracheolateralis intrinsic muscles of piprids, and limited ev- cranial to insertion. In Corapipo, Machaerop- idence from ontogeny is used to polarize vari- terus regulus, and Chloropipo uniformis, the ation in piprid intrinsic musculature. continuous sheet of ventral muscle fibers Several groups of piprids lack intrinsic sy- formed by the left and right Mm. tracheo- ringeal muscles. The Pipra serena species laterales divides on the ventral midline at group, Machaeropterus deliciosus, and M. AlO into differentiated left and right halves pyrocephalus have Mm. tracheolaterales that which veer laterad before insertion and dif- converge on the ventral midline, continue ferentiation into intrinsic musculature. In caudad as a single sheet, and then differen- most piprids with intrinsic muscles, the ven- tiate into right and left halves on the ventral tral sheet does not differentiate laterally at all midline to insert directly on Al. Most spec- or until immediately before insertion. In imens of the Pipra serena species group and Manacus, Chiroxiphia, and Antilophia, M. Machaeropterus deliciosus show no dorso- tracheolateralis is differentiated into left and ventral differentiation in the left and right right halves between A12 and 17, but these halves and insert continuously on Al. But in halves continue caudad as adjacent bellies M. pyrocephalus and a few specimens of the and do not divide and veer laterad to form Pipra serena species group, the left and right separate bodies of fibers on the lateral sur- sides are differentiated weakly into dorsal and faces ofthe trachea. The unique morphology ventral bundles of fibers that insert on the of Corapipo, Machaeropterus regulus, and dorsal and ventral ends ofA1. In Pipra pipra, Chloropipo uniformis is hypothesized to be the Mm. tracheolaterales are strongly differ- derived. entiated into four (left and right, dorsal and 47. M. sternotrachealis inserts by thin ventral) bellies that do not have any inter- membrane. In the Pipra aureola species group, mediate insertion before A1, and are thus M. sternotrachealis is a well-developed and completely extrinsic. In Manacus, M. tra- robust muscle that narrows abruptly before cheolateralis divides on the ventral midline insertion by a thin sheet of connective tissue into two large bellies of completely extrinsic to the lateral surface of M. tracheolateralis. musculature. None of the fibers of M. sternotrachealis is All other piprids have some form of in- continuous with M. tracheolateralis, as they trinsic musculature. In Corapipo, Masius, and are in all other piprids and other tyrannoids. Ilicura, there are two distinct forms ofoblique This unique morphology is hypothesized as intrinsic musculature that are clearly derived, derived. and share an additional derived detail with one another which indicates that the forms INTRINSIC MUSCULATURE are historically related. Chiroxiphia and An- tilophia have distinctive, derived intrinsic The intrinsic syringeal muscles of piprids muscles that share a derived form ofinsertion are very diverse and have arisen a number with the extrinsic muscles of Manacus. In ofindependent times within the clade. There Machaeropterus regulus males, the caudo- 1992 PRUM: MANAKIN SYRINGES AND PHYLOGENY 39 dorsal and caudoventral fibers ofM. tracheo- sert directly on the ventral ends of A1. The lateralis have developed into variable, par- syringes of P. cornuta females lack intrinsic tially intrinsic muscles which are absent in muscles and resemble those offemales ofoth- females. Each ofthese forms ofintrinsic mus- er species in the erythrocephala species group. culature is quite distinct and is hypothesized It remains problematic to determine the here to have an independent origin. polarity among the two pairs ofintrinsic mus- The two main remaining forms ofintrinsic cles in Chloropipo and Xenopipo, and the sin- musculature are certainly derived within pip- gle pair in Heterocercus, and the Pipra au- rids, but it is difficult to determine whether reola and Pipra erythrocephala species groups. both are independently derived or whether Given the information available, they are one state is primitive to the other in a com- treated as unordered alternative states of a plex transition series. In Chloropipo and Xe- complex character. The additional variations nopipo, M. tracheolateralis inserts on the tra- found within subsets of the Pipra erythro- chea at A6-9, and gives rise to left and right cephala species group are hypothesized to be intrinsic muscles, which themselves differ- derived as well. entiate into dorsal and ventral bellies at A2- The variation in piprid syringeal muscu- 4, to insert on the dorsal and ventral ends of lature will be coded as the following series of Al. In Heterocercus, the Pipra aureola spe- binary characters and as a single, unordered, cies group, and males of the Pipra erythro- complex character. cephala species group, M. tracheolateralis in- 48. Single pair of ventrolaterally oblique serts on the lateral and ventral surfaces of intrinsic muscles. In Ilicura, M. tracheolater- A4-9, and gives rise to a laterally differen- alis inserts continuously on the ventral and tiated pair ofintrinsic muscles that insert on lateral surfaces ofA6; a pair ofintrinsic mus- the entire surface of Al, without any dorso- cles originates on the ventral surface of A4 ventral differentiation. Among these latter and on the lateral surface ofA5. These fibers species there is additional variation in the pass obliquely caudodorsad to insert on the insertion ofMm. tracheolaterales and the or- dorsal and lateral surfaces ofA1. This oblique igin ofthe intrinsic muscles. In Heterocercus, pair ofintrinsic muscles is unique among pip- the Pipra aureola species group, P. mentalis, rids and cotingids, and is hypothesized to be and P. cornuta, the insertion and origins of derived. These muscles are only superficially these muscles are straight along the tracheal similar to the independently derived Mm. circumference and perpendicular to the tra- obliquus ventrales oftyrannids (Ames, 1971; cheal axis. In Pipra erythrocephala, P. rub- Prum, personal observ.). rocapilla, and P. chloromeros, M. tracheolat- 49. Two highly differentiated, oblique pairs eralis inserts on A8-9 on the lateral surfaces of intrinsic muscles. In Corapipo and Masius, ofthe trachea and on A5 on the ventral mid- M. tracheolateralis inserts on the ventral and line, in an oblique, chevron pattern. The in- lateral surfaces (Masius) or lateral surfaces trinsic muscles originate immediately caudal (Corapipo) of A6-8, and is partially contin- to this insertion in a similar chevron pattern. uous with two pairs ofintrinsic muscles. The Females, immature males, and a fledgling of ventral pair originates on the ventral surfaces the Pipra erythrocephala species group, and of A5-7 and on the lateral surfaces of A2-4 one of two female specimens of the aureola or A4-5; these fibers run caudad or caudo- species group completely lack intrinsic mus- ventrad to insert as a fleshy mass on the ven- culature; M. tracheolateralis does not insert tral end of A1. The dorsal pair of intrinsic on any tracheal elements before differenti- muscles originates on the lateral surface of ating laterally on the ventral midline and in- A4-5; these fibers run caudodorsad to insert serting on the entire Al. on the dorsal end ofAl in a large fleshy belly. An additional unique variation is present This morphology is unique among piprids in Pipra cornuta males; the paired, lateral and is hypothesized to be derived. intrinsic muscles are restricted to the lateral 50. Dorsal fibers of M. tracheolateralis and ventrolateral surfaces of the syrinx, and continuous with dorsal intrinsic musculature. the ventral fibers of Mm. tracheolaterales in- In Ilicura, Corapipo, and Masius, the dor- 40 AMERICAN MUSEUM NOVITATES NO. 3043

salmost fibers of M. tracheolateralis do not Al (Manacus). In Chiroxiphia pareola, the insert on the tracheal A elements, but con- connective tissue inserts on A4 at the ventral tinue parallel along the dorsal margin of the midline and extends caudad immediately lat- intrinsic muscle fibers to insert with them on eral to the ventral midline on both sides to the dorsal end of Al in a characteristic, ste- insert on A2 from its ventral end. This feature reotyped manner. In other piprids, M. tra- is a synapomorphy ofthe polytypic biological cheolateralis fibers are generally continuous species pareola which is not coded in this with intrinsic musculature, but they are not character analysis. The dorsal intrinsic mus- restricted to a well-defined group of dorsal cles of Chiroxiphia and Antilophia have ad- fibers. The morphology in Corapipo, Masius, ditional independent origins and insertions and Ilicura is unique among piprids and co- (see 53). The form of ventral differentiation tingids and is hypothesized to be derived. and the complex insertion of the ventral in- 51. Oblique, partially intrinsic ventral mus- trinsic muscle in these three genera are unique cles. In Machaeropterus regulus, M. tracheo- in tyrannoids and are hypothesized to be de- lateralis diverges ventrally at A10, and forms rived. two separate bundles of fibers on the lateral 53. Dorsal intrinsic muscle inserts by ten- surfaces of the trachea (see 46). In females, don on dorsal end of Al. In Chiroxiphia and each side divides into dorsal and ventral por- Antilophia, the dorsal portion ofthe intrinsic tions at A2-4 which insert on the dorsal and musculature originates on the dorsolateral ventral ends of Al. The ventral portion in surface of the A elements at the cranial mar- particular has a distinct, oblique lateroven- gin ofthe tracheal drum and inserts partially tral fiber direction. In males, M. tracheola- by a narrow strip of tendon on the extreme teralis inserts on A5 and gives rise to two dorsal end ofAl. This morphology is unique variable, partially intrinsic dorsal and ventral among tyrannoids and is here hypothesized muscles that insert on the dorsal and ventral to be derived. ends of A1. These muscles have the same 54. One or two pairs of laterally differen- fiber direction as those of females. Latero- tiated intrinsic muscles. In Heterocercus, and ventrally oblique intrinsic muscles are found P. aureola and P. erythrocephala species only in Corapipo and Masius, but these mus- groups, M. tracheolateralis inserts on the cra- cles are different in form, sexual dimorphism, nial margin of the fused tracheal drum (A5- and degree of development from the mus- 9) and gives rise to a single pair of laterally culature in M. regulus. The morphology in differentiated intrinsic muscles that insert on M. regulus is hypothesized to be indepen- Al. In Chloropipo and Xenopipo, M. tracheo- dently derived. lateralis inserts on A5-7 and gives rise to four 52. Intrinsic or extrinsic muscles insert by laterally and dorsoventrally differentiated spiral sheet of connective tissue. In Chiroxi- bellies of intrinsic muscles that insert on the phia and Antilophia, M. tracheolateralis in- dorsal and ventral ends of A1. In all these serts partially on A13-17 and gives rise to a genera, the intrinsic muscle fibers are cau- well-developed, partially intrinsic belly of fi- docranially oriented and not oblique. These bers that is cryptically divided from its origin morphologies are not found in any other pip- at the ventral midline into adjacent left and rids or other tyrannoids and are hypothesized right groups of fibers. In Manacus, M. tra- here to be derived. However, it cannot be cheolateralis is differentiated on the ventral determined which state is primitive relative midline into left and right portions of fibers to the other or whether they are indepen- at A 10 that continue caudad as adjacent parts dently derived. They will be coded here as ofa single well-developed belly ofcompletely unordered, derived states: (54.1) a single pair extrinsic musculature. In all three genera, the of nonoblique intrinsic muscles, present in left and right sides insert by a single, laterally Heterocercus, Pipra aureola species group, and continuous sheet of connective tissue on a Pipra erythrocephala species group; (54.2) two series of elements beginning with A3 (Man- pairs ofnonoblique intrinsic muscles, present acus) or A4-5 (Chiroxiphia and Antilophia) in Chloropipo and Xenopipo. on the ventral midline and spiraling caudo- 55. Extrinsic Mm. tracheolaterales fibers dorsad to the ventrolateral and lateral sur- insert on Al ventral to lateral pair of intrinsic faces of A2 (Chiroxiphia and Antilophia) or muscles. In male Pipra cornuta, the pair of 1992 PRUM: MANAKIN SYRINGES AND PHYLOGENY 41 intrinsic muscles is restricted to the lateral is hypothesized to be derived as well (57.2). and ventrolateral surfaces of the syrinx. The Both states are scored as unordered alterna- ventral extrinsic fibers of Mm. tracheolater- tive derived states of a single character. ales continue caudad to insert directly on the An additional variation in the pattern of ventral ends of Al. Mm. tracheolaterales re- syringeal innervation was not coded as an main laterally undifferentiated until imme- independent character because it was not diately before insertion. This morphology is completely fixed within any species. In most unique among piprids and other tyrannoids, specimens of every species ofManacus, Chi- and is hypothesized to be derived. roxiphia, and Antilophia, the tracheosyrin- 56. Chevron-shaped ventral insertion of M. geal nerves combined into a single fiber on tracheolateralis. In Pipra erythrocephala, P. the ventral surfaces ofthe trachea for several rubrocapilla, and P. chloromeros, M. tracheo- millimeters before splitting into left and right lateralis inserts on the lateral surface of A9 branches to innverate the syringeal muscles. and on the ventrolateral surfaces of the next In the few remaining specimens, the typical caudal series of elements in a spiral to A5 at X-shaped chiasma was present. Although in- the ventral midline. The left and right sides traspecific variation in this novelty makes it combine to form a V-shaped or chevron- difficult to code formally, it is a unique vari- shaped insertion in ventral view. The left and ation among tyrannoids and may consitute right intrinsic muscles originate immediately strong evidence for monophyly ofthese three caudal to the insertion ofM. tracheolateralis. genera. In all other piprids with intrinsic muscles, the 58. Main branches of tracheosyringeal insertion of M. tracheolateralis and the ori- nerves innervate M. sternotrachealis. In the gins of intrinsic muscles are straight across Pipra aureola species group, the main the trachea. The chevron-shaped M. tracheo- branches of the tracheosyringeal nerves veer lateralis insertion and intrinsic muscle origin laterad from the ventral chiasma and pass in Pipra erythrocephala, P. rubrocapilla, and directly into the medial surface of Mm. ster- P. chloromeros are unique in tyrannoids, and notracheales. The nerves innervate the cau- hypothesized to be derived. dal fibers of M. tracheolateralis and the intrinsic muscles by smaller secondary INNERVATION branches. In other piprids, M. sternotrachealis is innervated by small branches of the tra- 57. X-shaped ventral anastomosis of the cheosyringeal nerves. The morphology in the tracheosyringeal nerves. In all piprids except Pipra aureola species group is unique in tyr- Chloropipo and Xenopipo, the tracheosyrin- annoids and is here hypothesized to be de- geal nerves anastomose on the ventral mid- rived. line ofthe trachea between A20 and 35 before innervating the M. sternotrachealis, M. MEMBRANES tracheolateralis, and the intrinsic musculature. In other tyrannoids including the cotin- 59. Fibrous mass on medial tympaniform gids, Neopelma, Tyranneutes, and the tyran- membrane. In Pipra cornuta, there is a thick, nids, the tracheosyringeal nerves pass down round fibrous mass in the center ofthe medial the lateral surfaces of the trachea sometimes tympaniform membrane, medial to the B 1- exchanging smaller secondary nerve fibers. In 4 elements. The interbronchial ligament aris- Chloropipo and Xenopipo, the nerve fibers are es just caudal to this fibrous mass. These poorly developed and difficult to observe. structures occur sporadically in the syringes Most frequently, the nerves appear as a fine of other tyrannoids, but are hypothesized to network of tiny fibers across the ventral sur- be independently evolved. The novelties face ofthe trachea, but a few specimens with found in Pipra cornuta are unique among pip- observable nerves in C. uniformis and X. rids and are hypothesized here to be derived. atronitens have suggestions ofpiprid ventral anastomosis. The ventral anastomosis ofthe PHYLOGENETIC ANALYSIS tracheosyringeal nerves is here hypothesized The 59 derived syringeal character states to be derived in piprids (57.1). The alterna- supported two different maximally parsi- tive state found in Chloropipo and Xenopipo monious phylogenetic hypotheses with zero 42 AMERICAN MUSEUM NOVITATES NO. 3043 branch lengths collapsed. Each had a length The rest ofthe piprids belonged to a mono- of 76 and a consistency index of 0.82. The phyletic group made up of three clades with strict consensus tree based on these two trees unresolved interrelationships. The first clade included 22 resolved clades and was identical was the genus Machaeropterus, within which to one ofthe two shortest trees (fig. 24). Both deliciosus was supported as the sister group trees included a single large trichotomy about to regulus and pyrocephalus. which none of the characters were informa- The second clade included Manacus, Chi- tive. The two shortest trees differed only in roxiphia, Antilophia, and the Pipra serena the resolution of the relationships of a single species group. In this clade, Manacus was the species: Chloropipo unicolor. At least 25 dis- sister group to Chiroxiphia and Antilophia. tinct combinations of syringeal characters Interrelationships of species within Chiroxi- were identified among the 40 species of pip- phia and Manacus were not resolved. The rids. Thirteen lineages were diagnosed by sy- members of the Pipra serena species group ringeal autapomorphies. Twelve of the 59 form an unresolved, paraphyletic set of lin- characters were autapomorphies ofpresently eages related to the Manacus-Chiroxiphia- recognized biological species (11, 14, 26, 28, Antilophia clade. In both ofthe shortest trees, 32, 41, 43, 44, 48, 51, 55, 59). With these Pipra serena and P. suavissima are most autapomorphies removed, the shortest phy- closely related to these three genera but are logenetic hypotheses had a length of 64 and not a diagnosable clade. Pipra coronata and a consistency index of 0.77. the other members of the species group were The monophyly ofa number oftraditional an undiagnosable set of lineages at a basal genera and recently recognized species groups position in the clade. was supported by one or more syringeal syn- The third clade within this large assem- apomorphies (fig. 24). These corroborated blage of piprids included Chloropipo, Xeno- groups include Corapipo, Manacus, Ma- pipo, Pipra pipra, Heterocercus, the Pipra au- chaeropterus, the Machaeropterus regulus- reola clade, and the Pipra erythrocephala pyrocephalus clade, Chiroxiphia, the Pipra clade. The monophyly of Chloropipo plus aureola clade (including aureola, fasciicauda, Xenopipo was supported; and Xenopipo and filicauda), and the Pipra erythrocephala atronitens was the sister group to Chloropipo clade (including erythrocephala, rubrocapilla, uniformis. The relationships of C. unicolor to chloromeros, mentalis, and cornuta). The fol- this clade were not completely resolved. The lowing monotypic genera and species were two most parsimonious syringeal trees sup- also diagnosed by syringeal autapomorphies: ported either unicolor as the sister group to Masius (14), Ilicura (13.1, 28, 48), Corapipo uniformis and atronitens, or an unresolved gutturalis (4 1), Machaeropterus deliciosus (7), trichotomy of holochlora, unicolor, and uni- M. regulus (51), Chloropipo unicolor (32, 44), formis-atronitens clade. C. uniformis, Pipra pipra (11), Pipra cornuta The sister group to Chloropipo and Xeno- (55, 59), Pipra erythrocephala (- 16, 26), and pipo included Heterocercus and the rest ofthe Pipra serena (43). However, the monophyly genus Pipra. Within this clade, Pipra pipra ofPipra and ofChloropipo was not supported was the sister group to Heterocercus, the Pip- in the consensus tree or any of the six sup- ra aureola clade, and the Pipra erythroceph- ported resolutions. Neither Heterocercus nor ala clade. The monophyly of the Heterocer- the Pipra serena species group was supported cus was not supported by any syringeal as monophyletic by any derived syringeal synapomorphy, so the three species of Het- characters. erocercus were the undiagnosable primitive The first, basal clade in the piprid syringeal sister taxa to the Pipra aureola and erythro- consensus tree was composed of the three cephala clades. The relationships among the genera Corapipo, Masius, and Ilicura. This species of Heterocercus or the Pipra aureola clade is supported as the sister group to the clade were unresolved because no informa- rest of the piprids. Within this clade, Ilicura tive syringeal variations were identified. militaris was the sister group to Masius chry- However, the syringeal characters supported sopterus and Corapipo, and C. eucorrhoa and a single resolution of the phylogeny of the C. gutturalis were sister groups. five species in the Pipra erythrocephala clade: 1 992 PRUM: MANAKIN SYRINGES AND PHYLOGENY 43 Corapipo gutturalis Corapipo leucorrhoa Masius chrysopterus Ilicura militaris Machaeropterus deliciosus Machaeropterus regulus Machaeropterus pyrocephalus Manacus Chiroxiphia Antilophia galeata - Pipra serena - Pipra suavissama _ Pipra coronata species Chloropipo holochlora Chloropipo unicolor Chloropipo uniformis Xenopipo atronitens Pipra pipra

Pipra aureola Pipra filicauda Pipra fasciicauda Pipra cornuta Pipra mentalis I Pipra chloromeros - Pipra rubrocapilla Pipra erythrocephala Fig. 24. One of two maximally parsimonious phylogenetic hypotheses based on 59 syringeal characters for the piprids. Each input tree had a length of76 and a consistency index of0.81 . This phylogenetic hypothesis is same as the strict consensus tree of the two maximally parsimonious trees. The other shortest tree differed only in the placement of Chloropipo unicolor; C. unicolor is the sister group to Chloropipo uniformis and Xenopipo atronitens in the other tree (see fig. 25). cornuta as the sister group to the other four In none of the maximally parsimonious species; mentalis as the sister group to the resolutions of the data were Pipra pipra or remaining three; and chloromeros as the sis- the Pipra serena species group most closely ter group to erythrocephala and rubrocapilla. related to each other or to the remainder of Pipra erythrocephala and P. cornuta were both the genus Pipra. distinguished by syringeal autapomorphies. The syringes of Chloropipo flavicapilla, 44 AMERICAN MUSEUM NOVITATES NO. 3043

Heterocercus aurantiivertex, and Pipra vilas- pattern of partial ossification of the Al ele- boasi were unavailable for examination ments. This result confirms Snow's (1975, (Wood et al., 1982), but the relationships of 1979) placement of deliciosus back in Ma- these species to other piprids are discussed chaeropterus and the monophyly of Snow's below. regulus-pyrocephalus species group. The next large clade in this assemblage in- ALTERNATIVE PHYLOGENETIC cludes Chiroxiphia, Antilophia, Manacus, and HYPOTHESES FOR THE PIPRIDS the Pipra serena species group. Within this clade, the Chiroxiphia-Antilophia clade is A phylogenetic analysis of piprid syringeal conclusively supported by a suite of four de- characters supports a well-resolved phylo- rived characters (17, 24, 42, 53). Snow (1975) genetic hypothesis for the 40 species of the hypothesized that these two genera may each family. These results corroborate cladistical- be related to Pipra, based on behavior and ly some traditional piprid taxa that have not plumage, respectively, but they have not been been diagnosed before, and support many previously associated with one another. novel phylogenetic hypotheses among man- The Chiroxiphia-Antilophia clade shares a akin genera. Parsimonious optimizations of unique, derived musculature character with the evolution of the syringeal characters are Manacus (52). Most specimens ofthese three shown on the resolved syringeal hypothesis of phylogeny in figure 25. Here, the relative genera also have a unique, probably derived strength ofsupport ofthe various piprid clades form of syringeal innervation in which the is discussed, and the syringeal hypothesis of left and right tracheosyringeal nerves com- phylogeny is compared to previous phylo- bine into a single large fiber (described in 57). genetic hypotheses and classifications of the This novelty was not used as a character be- family. cause a few individuals of all genera had the The monophyly of the Ilicura-Masius- general piprid X-shaped pattern. However, Corapipo clade is supported by two synapo- this unique character provides further cor- morphies (40, 50). The Masius-Corapipo roboration of the Chiroxiphia-Antilophia- clade is diagnosed by three syringeal synapo- Manacus clade. The monophyly ofManacus morphies (12, 36, 49) including details ofboth is supported by a unique, dorsal fusion of a supporting elements and musculature. The caudal series of B elements (23) and several monophyly of Corapipo was supported also other synapomorphies (6, -13.1, 31). The by two characters (38, 46). This resolved hy- monophyly of Chiroxiphia is supported by a pothesis for the phylogeny ofthe four species single derived character (25). This novel hy- in these three genera is exactly congruent with pothesis for the interrelationships of these the previous hypothesis for their interrela- three genera is congruent with the large body tionships based on a cladistic analysis ofmale size and with plumage and soft part colors of courtship display elements (Prum and John- females in these genera. son, 1987). Plumage similarities that lead The monophyly ofthe Pipra serena species Snow (1975) and Hellmayr (1910) to hy- group was not supported in any of the most pothesize close relationships between Cora- parsimonious resolutions. Although the sy- pipo and Manacus, and between Masius and ringes of these species are identifiable and Antilophia are apparently convergent. generally similar to one another, these sim- The large clade composed ofthe rest ofthe ilarities are not arguably derived. All mem- piprids is supported by two derived syringeal bers of the group have completely cartilagi- characters: (1) complete, double A elements, nous Al elements (13.1) that are shared with and (30) specialized ventral ends of the B3 the Chiroxiphia-Antilophia clade and Ilicura. elements. Within this assemblage, there are Pipra serena and P. suavissima share the de- three main clades with unresolved relation- rived dorsally cartilaginous, single A ele- ships. The first includes only the genus Ma- ments with Manacus, Chiroxiphia, Antilo- chaeropterus whose monophyly was sup- phia, and Machaeropterus regulus (19). ported by a single syringeal character (45). The syringeal morphology of the Pipra se- The monophyly of the M. regulus-pyroce- rena species group is generally similar but not phalus clade was supported by (13.2) a unique arguably derived within the family, so no sy- 1992 PRUM: MANAKIN SYRINGES AND PHYLOGENY 45

Corapipo gutturalis Corapipo leucorrhoa 50 144 40, 1 Masius l13.1, 28B, 48 chrysoptenus llicura militaris 45 | Machaeropterus deliciosus l ~~~~~19,46, 51 13.2 Machaeropterus regulus Machaeropterus pyrocephalus

l ~~~~~~~~~~~~52|2 Manacus 20, 27, 57.1 l Chiroxiphia 19 17, 24, Antilophia galeata

[113.1 | 43 - Pipra serena Pipra suavissama Pipra coronata species 1,30 1 54.2,57.2 32,44 Chloropipo holochlora I Chloropipo unicolor 3737-15, -30, 37, 46 Chloropipo uniformis 18, 22,35 Xenopipo atronitens 2,15 l P4ra pipra Heterocercus

10,21, 29,31,47, 58 Pipra aureola 54.1 | Pipra tilicauda Pipra fasciicauda ~~~~3,-4,55,59 4, 8, 16 | Pipra cornuta Pipra mentalis 9,33,37 Pipra chloromeros 34l 39, T + rPipra rubrocapilla 5 Pipra erythrocephala -16, 26 Fig. 25. Hypotheses for the evolution of the 59 syringeal characters within one of the maximally parsimonious phylogenetic hypotheses for the piprids. The alternative derived states of the three unordered, multistate characters (13, 54, and 57) are indicated as decimals following the character number (#.1, #.2). Hypothesized reversals are indicated by a minus sign preceding the character number. All character optimizations are unambiguous except for characters 4 and 37. Both ofthese are hypothesized to have evolved once and to be lost secondarily once, but it would be equally parsimonious to hypothesize two independent origins. Character 4 is hypothesized to a synapomorphy of the Pipra aureola-erythrocephala clade with a reversal in Pipra cornuta, but it may also be independently derived in the Pipra aureola clade and the erythrocephala clade, excluding cornuta. Character 37 is hypothesized to be a synapomorphy of Chloropipo unicolor, C. uniformis, and Xenopipo atronitens, with a reversal in uniformis; but it could be hypothesized to be independently evolved in unicolor and atronitens, as in the other maximally parsimonious hypothesis based on these data (fig. 24). 46 AMERICAN MUSEUM NOVITATES NO. 3043 ringeal characters support the monophyly of ossified ventral end of the B1 elements -as the group. Haffer (1970) recognized the spe- either a synapomorphy of a clade including cies group based on overall plumage similar- unicolor, atronitens, and uniformis with a ity. Subsequently, a cladistic analysis of subsequent reversal in uniformis, or as two plumage characters was used to support the independent derivations in unicolor and monophyly ofthe Pipra serena species group atronitens. The hypothesis of a single origin (Prum, 1988). However, this analysis was fully resolves the relationships of these spe- confounded by the inappropriate use ofPipra cies while the independent origin hypothesis pipra and other Pipra species as an outgroup, produces an unresolved trichotomy with hol- since the syringeal characters analyzed here ochlora, unicolor, and the uniformis-atroni- indicate that Pipra pipra is not closely related tens clade. to the Pipra serena group. Furthermore, ad- The syrinx of Chloropipo flavicapilla was ditional errors were made in the analysis not available for observation in this study, (Prum, 1988). The iris color of Pipra serena because no spirit specimens of the species and suavissima was coded as white, and the were available (Wood et al., 1982). However, iris color of nattereri, iris, and vilasboasi was derived plumage characters support a clade coded as dark, when the opposite is the case. includingflavicapilla, unicolor, uniformis, and Furthermore, an alternative equally parsi- atronitens, resolving the ambiguity in the sy- monious resolution of the plumage charac- ringeal data (Prum, in prep.). ters would place exquisita as more closely Snow (1975) and Wolters (1977) hypoth- related to nattereri-vilasboasi-iris than to cor- esized that Chloropipo flavicapilla, C. hol- onata (J. Haffer, personal commun.). The ochlora, and C. uniformis comprise a species monophyly ofthe Pipra serena species group group or a separate subgenus from C. uni- may ultimately be supported by several de- color. However, the syringeal characters rived traits that are found only in members strongly support uniformis as the sister group of this species group but have also been sub- to Xenopipo atronitens. Both atronitens and sequently lost in other members, such as the unicolorhave been separated at various times white or blue rump patch. Derived plumage from other Chloropipo species based on black and iris color traits still support the serena- male plumage, which is probably indepen- suavissima clade and the nattereri-vilasboasi- dently derived in these two species within the iris clade within the species group, but other group (Prum, in prep.). The Chloropipoflavi- aspects ofthe phylogeny ofthe species group capilla-holochlora-uniformis species group is are unresolved. As in syringeal morphology, a paraphyletic assemblage composed of the plumage of the Pipra serena species group sexually monomorphic or mostly green shows a uniformity and similarity which is plumaged members of this clade. not certain to be derived, but the overall ev- The genus Pipra is currently recognized to idence implies that the assemblage is highly include 16 species, many ofwhich have most- likely to be monophyletic. No syringeal ev- ly black male plumage with distinct patches idence was found to refute or corroborate Sib- of bright red, yellow, white, or blue. The sy- ley and Monroe's (1990) placement of the ringeal characters of these species indicate two Andean species, coeruleocapilla and isi- that the genus is polyphyletic and composed dorei, in a separate species group. of three main clades. The Pipra aureola and Two syringeal synapomorphies support a Pipra erythrocephala clades share three de- clade including Chloropipo and Xenopipo rived, detailed syringeal traits (4, 8, 16), and (54.2, 57.2). Within this group, a clade in- constitute a well-supported monophyletic cluding Chioropipo uniformis and Xenopipo group. A single syringeal muscle character in- atronitens is diagnosed by three derived sy- dicates that they are most closely related to ringeal characters (18, 22, 35). The syringeal Heterocercus: (54.1) presence of a single pair characters do not resolve the relationships of intrinsic muscles. Two derived syringeal among the other species in this group: C. hol- characters support Pipra pipra as the sister ochlora and C. unicolor. It is equally parsi- group to these three clades (3, 7). The seven monious to explain the single derived char- species of the Pipra serena species group are acter shared by unicolor and atronitens- (37) not closely related to this clade. The tradi- 1 992 PRUM: MANAKIN SYRINGES AND PHYLOGENY 47 tional plumage characters used to define the species constitute a superspecies and this was genus Pipra are apparently homoplasious and accepted by the American Ornithologists' phylogenetically uninformative. Union (1983), but Haffer did not resolve their The genus Heterocercus is not diagnosed cladistic interrelationships. Snow (1979) rec- by any syringeal characters, and anatomical ognized the erythrocephala species group ex- specimens of one of the three species in the cluding cornuta, and a single syringeal synap- genus, aurantiivertex, were unavailable omorphy also supports the monophyly ofthat (Wood et al., 1982). However, all three spe- assemblage. Based on plumage, chloromeros cies of Heterocercus share three derived is perhaps most similar to mentalis, but the plumage characters that unambiguously sup- clade including chloromeros, erythrocephala, port the monophyly of the genus: (a) silky, and rubrocapilla is well supported by two sy- elongate, erectile white throat feathers; (b) a ringeal synapomorphies. Cracraft (1988) hy- graduated tail with the outer rectrices short- pothesized that Pipra cornuta is the sister est; and (c) a narrow, central crown patch of group to erythrocephala and rubrocapilla red, yellow, or orange. The first two charac- based on plumage characters, but a large ters are unique among piprids and all other number of syringeal characters support the tyrannoids. This derived crown patch char- placement of cornuta as the sister group to acter is unique among piprids but has been all other species in the clade. Sibley and Mon- independently derived in some tyrannids. roe (1990) recognized a species group includ- The syringeal data indicate that the plum- ing mentalis, erythrocephala, and rubrocap- age and size traits, and notions about unde- illa only, but the syringeal characters strongly veloped display behavior that were tradi- indicate that this assemblage is not mono- tionally used to align Heterocercus with phyletic. Pipra chloromeros shares (39) Schiffornis are homoplasious or inaccurate. ventrally ossified B2 elements, and (56) chev- Parkes (1961) redescribed the unique type ron-shaped origin of intrinsic muscles with specimen ofPipra anomala Todd as a hybrid erythrocephala and rubrocapilla that are both between Heterocercus linteatus and Pipra au- absent in mentalis. reola. He remarked that the hybrid provided The various lineages in the syringeal hy- evidence for a closer relationship between the pothesis of phylogeny have different degrees two genera than implied by Hellmayr's (1929) of relative confidence. The relationships placement of Heterocercus as distant from among Corapipo, Masius, and Ilicura are well Pipra and nearest to Schiffornis. These syrin- supported, and congruence with a previous geal data indicate that the two species in- behavioral hypothesis of phylogeny of these volved in this hybridization are even more genera increases the confidence in this hy- closely related than Parkes (1961) hypothe- pothesis (Prum and Johnson, 1987). Sister- sized. group relationships between Chiroxiphia and The Pipra aureola clade, previously rec- Antilophia, the Pipra aureola and Pipra ognized as a species group by Haffer (1970) erythrocephala clades, Machaeropterus reg- and Snow (1975, 1979), is diagnosed by six ulus and M. pyrocephalus, and Chloropipo syringeal synapomorphies (10,21,29, 31,47, uniformis and Xenopipo atronitens are also 58). No syringeal variation among the three well supported by two or more derived char- species in this clade was observed. In con- acters. However, other higher level clades are trast, each of the five species of the Pipra less well supported or dependent on addi- erythrocephala clade have a distinct set of tional, possibly overly simplified, hypotheses derived syringeal characters that completely of character evolution. Homoplasy or mis- resolve their interrelationships. The entire takes in polarization ofsome characters, such clade is supported by three derived traits (9, as (1) complete double A elements, or (30) 33, 37), and each branch within is supported specialized ventral ends of the B3 elements, by additional synapomorphies: (34) derived could eliminate support for some major pip- in all but cornuta, (39, 56) derived in all but rid clades. Further, no characters resolve a mentalis and cornuta, and (5) derived in major trichotomy in the hypothesis, and any erythrocephala and rubrocapilla alone. error in characters supporting the monophyly Haffer (1970) hypothesized that these five of these three clades would yield additional 48 AMERICAN MUSEUM NOVITATESNNO. 3043 instability in the higher-level phylogeny of gruence conflict with other Delta T50H values the family. Although this is the first compre- reported: e.g., Pipra erythrocephala to P. hensive morphological investigation of pip- mentalis, 2.2; P. erythrocephala to P. filicau- rid phylogeny, it is not a complete solution da, 2.7. The former is shorter than the dis- to the phylogenetic interrelationships of the tance to Heterocercus, and the latter is larger family, and should be augmented with com- than other distances reported to species of parative analyses of other character systems. Chiroxiphia, Manacus, Machaeropterus, Two phylogenies including some piprids Chloropipo, and the Pipra serena species based on biochemical data have been pre- group. The available DNA-DNA hybridiza- sented. S. M. Lanyon (1985) did an analysis tion distances contradict the monophyly of of allozyme variation of tyrannoids that in- most traditional piprid genera and species cluded seven piprid species. He identified 14 groups, many ofwhich are strongly supported allozymes that varied among these taxa and by syringeal synapomorphies. Larger com- supported two main clades, which included plete matrices with reciprocal hybridizations Machaeropterus regulus, Manacus manacus, are required before these inconsistencies in and Pipra pipra; and Chiroxiphia pareola, DNA-DNA hybridization distances can be Chloropipo holochlora, Masius chrysopterus, evaluated. and Corapipo leucorrhoa. The distance trees I used these syringeal data and an earlier also placed Neopelma and Tyranneutes as the version of the syringeal hypothesis of phy- sister group to either of these two groups of logeny (Prum, 1989) in an analysis of the piprids, and Schiffornis as the sister group to evolution of piprid display behavior (Prum, all other piprids. None of the derived syrin- 1 990a). Phylogenetic analysis ofthe separate geal characters described here are congruent syringeal and behavioral data sets, the com- with either of the two groups supported by bined data sets, and superimposition of the allozyme variation. Furthermore, a cladistic behavioral characters on the syringeal con- analysis of other morphological characters sensus tree revealed a high degree of congru- indicates that Schiffornis, Neopelma, and Ty- ence between the two data sets, and dem- ranneutes are most closely related to non- onstrated the tradeoffs among the alternative piprid tyrannoids (Prum, 1 990b). The lack of phylogenetic tests of behavioral hypotheses congruence between the syringeal and allo- of homology. Since that publication, I ob- zyme data may be the result ofextensive ho- served additional piprid specimens which led moplasy in either data set, but this cannot be me to reevaluate some characters and to evaluated appropriately until a more com- change the coding of others. For example, plete molecular analysis of piprids is per- additional specimens of Machaeropterus py- formed. rocephalus led me to reexamine this species; Sibley and Ahlquist (1985, 1990) have pyrocephalus turned out to lack dorsally car- published two analyses of DNA-DNA hy- tilaginous A elements which I had errone- bridization distances for piprids. Pipra eryth- ously coded as present (Prum, 1989). This rocephala was used as a driver in compari- change led to additional resolution of the sons to 13 other piprid species. Piprids were phylogeny of the group, including unambig- also used as tracers in comparisons to Pip- uous support of the monophyly of Machae- reola arcuata, Schiffornis turdinus, and other ropterus. However, these changes did not suboscines, but no reciprocal hybridizations substantially affect the results of the behav- were performed. Values of Delta T50H for ioral analysis presented in Prum, 1990a. comparisons between Pipra erythrocephala and other piprids ranged from 2.1 (to Het- erocercusflavivertex) to 3.2 (to Masius chry- SYRINGEAL VARIATION, FUNCTION, sopterus). These values seem to imply some AND EVOLUTION congruence with the syringeal hypothesis of phylogeny, which places the Pipra erythro- PREVious DESCRIPTIONS OF cephala clade near Heterocercus, and Masius PIPRID SYRINGES in a small clade that is the sister group to the Previous descriptions ofthe syringeal mor- rest of the piprids. But these points of con- phology ofpiprids come from three authors. 1 992 PRUM: MANAKIN SYRINGES AND PHYLOGENY 49

Muller (1847, 1878) described the syringes ments are spatulate, as are B 1-3 ofalmost all of Pipra pipra, P. erythrocephala, Chiroxi- other piprids, indicating that these elements phia pareola, and Manacus manacus. Lowe are B1-3 and that the Al in these genera is (1942) made additional observations of cartilaginous. Manacus vitellinus. Ames (1971) observed a Hypotheses of homology among support- total of 10 individuals from 8 species -Pipra ing elements established by these conserved erythrocephala, Pipra mentalis, Chiroxiphia special similarities in shape are consistent caudata, Chiroxiphia lanceolata, Ilicura mil- with other details in shape and in the position itaris, Corapipo gutturalis, Manacus vitelli- of intrinsic muscle insertions. nus, and Manacus candei. Ames (1971) also described the syringes ofSchiffornis and Pip- rites as piprids, but these genera are not mem- INTRASPECIFIC VARLATION bers of the monophyletic assemblage of pip- Intraspecific variation in syringeal mor- rids analyzed here (Prum and Lanyon, 1989; phology may be geographic, sexual, ontoge- Prum, 1990b). netic, or random individual variation (Ames, Previous descriptions of piprid syringeal 1971). The substantial samples ofindividuals morphology are largely congruent with these of some species make it possible to access findings, but they are limited in detail be- some aspects of intraspecific syringeal vari- cause the stains and preparations employed ation in piprids. Most piprids show limited here were not used. Most apparent differences individual variation in syringeal supporting from Ames's (1971) descriptions result from elements, and somewhat more variation in differences in the initial identification of ho- syringeal musculature. Although no sexual mologous supporting elements. Ames's ho- dimorphism was observed in syringeal sup- mology criteria were based largely on indirect porting elements, significant sexual dimor- observation ofossification, and this typically phism is present in syringeal musculature of led to the identification of the first ossified some species. double element as A1. The criteria utilized Syringeal supporting elements have a ste- here emphasize special similarities in shape reotypical shape within a species. Intraspe- of elements over their composition, and as a cific variations are more likely to occur in the result completely cartilaginous Al elements extent of fusion or ossification of supporting were identified in Ilicura, Chiroxiphia, An- elements. Pipra pipra, Machaeropterus reg- tilophia, and the Pipra serena species group. ulus, Manacus manacus, the Chiroxiphia In Ames's descriptions of three species from species, and Chloropipo holochlora have some the two former genera, the cartilaginous Al's variation in the fusion ofelements in the ob- were identified as B l's. For example, the in- served samples. In Chiroxiphia, the number trinsic muscles of Ilicura militaris were de- ofelements incorporated in the tracheal drum scribed by Ames as inserting on B 1, but sim- varied by one or two within each species (ta- ilarities in the dorsal fusion of B 1-2 to other ble 1). Specimens of Manacus vary in the piprids favor the hypothesis described here- degree offusion ofthe double A elements. In that these muscles insert on a cartilaginous all specimens, A 1-2 are completely fused, but Al. in one individual of M. vitellinus and one of Likewise, the cartilaginous Al of Chiroxi- M. candei, A1-3 are fused. Specimens ofPip- phia was identified by Ames as B 1. This ho- ra pipra vary in the presence ofdorsal fusion mology assignment is a little more difficult, of single, tracheal A elements. In most Ma- because the first three double, cartilaginous chaeropterus regulus (N = 6 of8), the dorsally elements (either A1-B2, or B 1-3) in Chiroxi- fused B 1-2 are additionally fused to B3. phia and Antilophia are dorsally fused by The most variable sample observed was of strips of cartilage. The former situation is Chloropipo holochlora. It varied in the degree found nowhere else in piprids, and the latter of fusion of the first single, tracheal A ele- is found only in some specimens ofMachae- ments (A5-7 or A5-8) ranging from fusion ropterus regulus. However, in Chiroxiphia and by an ossified dorsal bar, to fusion along their Antilophia the ventral ends of the second dorsal and ventral margins into a nearly com- through fourth double, cartilaginous ele- plete drum. One anomalous specimen lacked 50 AMERICAN MUSEUM NOVITATES NO. 3043 dorsal fusion entirely (fig. 11). There is also mature males. In males of the Pipra eryth- some variation in degree ofossification ofthe rocephala clade, M. tracheolateralis inserts medial section of A2, the first double, me- on the cranial margin of the tracheal drum; dially complete element. This variation does right and left intrinsic muscles originate just not appear to be geographically based; some caudal to that insertion, and insert them- extremes in syringeal structure come from the selves on Al. Both M. tracheolateralis and same geographic regions. the intrinsic muscles are strongly developed Another type of intraspecific variation is into large bellies ofmuscle fibers. In females, found in Pipra pipra. In 3 of 15 specimens, fledglings, and immature males (sex deter- an additional double, complete A element is mined by observations of the gonads), M. present. This type of variation is interesting tracheolateralis continues caudad in a thin in that it changes the relative position of the sheet of fibers to insert directly on the A1 tracheobronchial junction to the syringeal elements after separating on the ventral mid- supporting elements, but does not result in line at A4-8. The muscle has no traces of any significant change in the overall form of intermediate insertions before Al or devel- the syrinx. oped bellies of muscle fibers. The Pipra au- Ames (1971) described occasional asym- reola clade is apparently intermediate; one metrical, "extra" supporting elements in ty- female specimen completely lacks intrinsic rannoids, and these do occur in low frequency muscles and another has weak but completely within piprids. These additional rings or half differentiated intrinsic musculature. rings are developmental accidents, and usu- Other piprids differ from the Pipra eryth- ally do not change the shape of the syrinx in rocephala and Pipra aureola group in that any significant fashion. females and young males have syringeal mus- There is limited individual variation and culature that is identical in form to adult males significant sexual variation in syringeal mus- but is significantly less developed in mass. culature of most species. In the Pipra serena This is true ofIlicura, Corapipo, Masius, An- species group, there is some minor variation tilophia, and Chiroxiphia. in the extent ofthe insertion ofM. tracheolat- The sexual dimorphism and ontogeny of eralis on Al. In most specimens, M. tracheo- M. tracheolateralis and the intrinsic muscu- lateralis inserts continuously along the lateral lature in the P. erythrocephala clade support surface of Al1, whereas in a few, M. trach- Ames's (1971: 138-140) hypothesis that pas- eolateralis fibers are absent on the lateral serine intrinsic syringeal musculature is on- midline. This variation appears to be more togenetically derived from undifferentiated related to underdevelopment of the muscle M. tracheolateralis fibers. However, it con- in these specimens rather than a well-marked tradicts Ames's (1971: 95-96) generalization dorsoventral differentiation. In three speci- that the ontogeny of syringeal musculature is mens ofPipra suavissima, the M. tracheolat- complete by the time of fledging. eralis inserts on Al, A2, or both. Many individuals differ in the degree of In general, most females have less robust symmetry ofthe syringeal musculature. This and well developed musculature than males, muscular asymmetry is associated with the although the musculature of males and fe- consistent asymmetry of the trachea, which males is identical in general form. In Chlo- is situated on the right side of the neck and ropipo holochlora, the intrinsic muscles are twists clockwise as it proceeds caudad into well differentiated in males but only partially the interclavicular air sac. Because of this intrinsic in females. In Machaeropterus reg- twist, the left M. sternotrachealis has to travel ulus, the two males in the small sample farther from its origin on the craniolateral showed significant differences in the inde- process of the sternum to its insertion on the pendence or intrinsic distinctness of the trachea. The left M. tracheolateralis is also oblique caudal fibers of M. tracheolateralis, more exposed ventrally than the right and whereas females showed no indication of in- may be less constricted in development. For trinsic musculature. In the most extreme case, the most part, if any asymmetry is present, intrinsic syringeal muscles are found only in the left M. tracheolateralis and M. sternotra- 1992 PRUM: MANAKIN SYRINGES AND PHYLOGENY 51 chealis contribute more to the total mass of Some syringeal apomorphies have evolved extrinsic musculature. Usually, intrinsic independently many times within this hy- musculature is not noticeably asymmetrical. pothesis of phylogeny for the family. For ex- Asymmetry of M. tracheolateralis is example, intrinsic syringeal muscles have aris- treme in the three species ofMachaeropterus en independently at least four times within (character 45). This extreme asymmetry does the piprids, and have radiated into eight dis- not appear to be correlated only with the gen- tinct forms. Within the Manacus-Chiroxi- eral underdevelopment ofthe syringeal mus- phia-Antilophia clade, dorsal and ventral in- cles in this genus. For comparison, the species trinsic muscles have evolved subsequent to in the Pipra serena group also have poorly the evolution of a novel extrinsic muscle in- developed Mm. tracheolaterales, but they do sertion. Oblique intrinsic musculature not have exaggerated asymmetry. evolved a single time in the Ilicura-Masius- Corapipo clade, and subsequently diversified into the distinct single pair and double pair EVOLUTION OF INTERSPECIFIC of intrinsic muscles found in these genera. VARIATION The ventral oblique pair present in Ilicura is The syringeal morphology of piprids has similar in position and fiber direction to the undergone an extensive radiation in both M. obliquus ventralis present in many tyran- supporting element structure and muscula- nids. Some supporting element novelties have ture. In this regard, the piprids are similar to also evolved multiple times within the pip- other tyrannoids (Ames, 1971; Lanyon, rids, such as tracheal fusion of A elements 1984a, 1985, 1986, 1988a, 1988b, 1988c; (14-17), and ventral ossification of B ele- Prum and Lanyon, 1989) and differ from the ments (36-39). oscine passerines, which are characterized by A striking example of apparently rapid sy- a complex but generally stereotyped syrinx ringeal evolution comes from two sister spe- (Ames, 1971, 1975, 1987; Warner, 1972). cies, Pipra serena and P. suavissima. These However, the extent ofthe interspecific vari- two differentiated, allopatric geographic forms ation in piprids is extreme even within in the are currently placed as subspecies of Pipra generally diverse suboscine groups. Ofthe 40 serena (Snow, 1979), but I recognize them species in the family, 25 distinct combina- here as separate species. Pipra serena is found tions of syringeal novelties are identifiable. in Suriname, French Guiana, Amapa and Only a few oscine families are known to have Para Provinces, Brasil. Pipra suavissima is diagnosable syringeal characters (Ames, 1971, found in the western Guiana Highlands re- 1975, 1987; Warner, 1972; B. D. Cutler in gion of southern Venezuela, Guyana, and Baptista and Trail, 1988). northern Amazonas and Roraima Provinces, This extensive variation in syringeal mor- Brasil. The males of these two populations phology in piprids is used here to reconstruct differ in the distribution and tone of yellow the phylogeny of the family, but it is of ad- in the plumage, and the females are similar ditional interest as an example of syringeal in plumage but differ in the tone of blue on diversification in birds. It may be preferable the crown. The two forms were first com- to make inferences about the process of sy- bined as subspecies of Pipra serena by Hell- ringeal evolution in the context of an inde- mayr (1910, 1929). Derived plumage simi- pendent hypothesis of phylogeny based on larities indicate that the two forms are sister some other data set, but this syringeal data taxa (Prum, 1988), but differences in plum- set is the only complete hypothesis available. age, song, and syringeal morphology between However, the patterns of variation are so these two allopatric forms indicate that they complex that any phylogenetic hypothesis are separate species with distinct evolution- would indicate that the piprid syrinx has un- ary histories (Prum, in prep.). dergone significant changes in supporting el- The syrinx of P. suavissima is similar to ement shape, composition and fusion, form those of other members ofthe P. serena spe- and development ofmusculature, and inner- cies group, but it is slightly larger and has a vation patterns. series of dorsally cartilaginous single A ele- 52 AMERICAN MUSEUM NOVITATES NO. 3043 ments (fig. 17). The syrinx of P. serena has inx. The other potential sound sources -the all the typical characters of other members lateral membranes and the semilunar mem- of the species group and the specific features brane-are reduced or absent in piprids. of P. suavissima, but is tremendously en- Attempts to correlate syringeal structure larged in size, distorted in shape, and char- with vocal ability have been limited by the acterized by a suite of unique morphological extreme complexity of the problem. Ames details (fig. 18). This suite of features of the (1971) summarized the general trend of sy- syrinx ofPipra serena (43) appears to be cor- ringeal and vocal complexity in passerines. related with its increase in size and the re- He observed that although the simplest sy- sulting exaggeration and distortion of prim- ringes do produce simple songs, strong cor- itive structures. These rearrangements may relations are lacking between syringeal and have been caused by a single change or mu- vocal complexity. Ames concluded that ad- tation in the developmental program of the vances in neural control may be more im- syrinx. portant in the evolution of vocal complexity The doubling in size and tremendous dis- within the passerines. Gaunt (1983) hypoth- tortion in shape of the syrinx of serena have esized that complex syringeal musculature is evolved since common ancestry with suav- necessary for the evolution of vocal com- issima. These changes have been relatively plexity, defined as rapid frequency and am- more rapid and extensive than the change in plitude modulations. He hypothesized that plumage coloration. The absolute age ofthese antagonistic pairs of intrinsic syringeal mus- sister species is unknown, but it is apparent cles are necessary to perform fine adjust- that serena has undergone a rapid accelera- ments ofmembrane tensions and relative po- tion in the rate of syringeal evolution by the sition ofsyringeal supporting elements. Gaunt acquisition of a suite of major apomorphies. found a rough correlation indicating that sy- This extent of syringeal variation between ringeal muscle complexity was necessary but sister species is unprecedented in other birds, not sufficient to explain the distribution of and is especially striking given the degree of avian vocal complexity. No other significant overall plumage similarity. While it is diffi- patterns of relationship between syringeal cult to compare rates of morphological evo- structure and vocal ability have been docu- lution, such a rapid rate of diversification is mented. probably rare for avian morphological sys- All observed variations in supporting ele- tems. ments and musculature ofpiprid syringes can This case raises the possibility that the evo- be hypothesized to have direct or indirect lution of syringeal diversity in the piprids affects on vocalization through the control of proceeded through the accumulation of apo- the tension, relative position, support, or size morphies in rapid "revolutions" in syringeal of the medial tympaniform membranes. But organization (Gould and Eldredge, 1977). the actual functional consequences of these structural variations cannot be evaluated or interpreted in terms of present theories of FUNCTIONAL CONSEQUENCES OF avian vocalization. SYRINGEAL VARLATION Direct electromyographic observations of The consensus among functional mor- the extrinsic muscles ofsome species ofbirds phologists is that the medial tympaniform indicate that they are important in vocal membranes are the main sound sources in modulation (Brackenbury, 1989), but no di- the tracheobronchial passerine syrinx rect observations of intrinsic muscle activity (Greenewalt, 1968; Gaunt, 1983; Gaunt and during vocalization have been made. The in- Gaunt, 1985; Brackenbury, 1989; Suthers, sertion of M. tracheolateralis or intrinsic 1990). In the absence of direct observations musculature on the double A elements prob- of the medial tympaniform membrane dur- ably functions to alter the position of these ing vocalization, these observations ofpiprid elements and control the acoustical proper- syringeal morphology confirm that the me- ties of the medial tympaniform membranes dial tympaniform membrane is the only fea- they support. However, this apparent func- sible vibratile sound source in the piprid syr- tional association does not explain the vari- 1992 PRUM: MANAKIN SYRINGES AND PHYLOGENY 53 ety in dorsoventral differentiation, insertion, other piprids, few ofthese M. sternotrachealis and fiber direction of M. tracheolateralis or fibers are continuous with M. tracheolater- the intrinsic muscles of piprids. alis. Such an insertion would potentially have Piprid intrinsic musculature may insert on very different action on the syrinx during vo- the cartilaginous dorsal and ventral ends of calization. Al (in the oblique muscles of Corapipo and The composition and shape of syringeal Masius), on the entire length of a completely supporting elements may also have signifi- ossified Al (in Heterocercus, and the Pipra cant effects on vocal control and modulation. aureola and Pipra erythrocephala clades), on Supporting elements should vary significant- the entire cartilaginous Al (in Ilicura mili- ly in rigidity and mobility depending on taris), or in a complex insertion on a number whether they are cartilaginous or ossified, in- of cartilaginous and ossified A elements (in dependent or fused. The variations in the Chiroxiphia and Antilophia). Intrinsic mus- shape, degree of ossification, and fusion of cles also vary in the degree of dorsoventral supporting elements in piprids present a wide differentiation. In some piprids, the left and array of potential acoustic consequences. right intrinsic muscles are not divided or dif- Independent association ofsome muscular ferentiated into dorsal and ventral bundles of and supporting element novelties may indi- fibers (e.g., Pipra aureola and Pipra eryth- cate that these structures function together in rocephala clades), and in others dorsoventral vocalization. In two phylogenetically inde- differentiation is complete as in the oblique pendent instances, M. tracheolateralis inserts intrinsic muscles of Corapipo and Masius. and the intrinsic muscles originate near the The simpler extrinsic musculature ofsome cranial margin of a drum of fused tracheal piprids also varies in dorsolateral differenti- elements (e.g., Chiroxiphia and Antilophia; ation. In Pipra pipra, M. tracheolateralis is most Chloropipo, Xenopipo, Heterocercus, the laterally and dorsoventrally differentiated into Pipra aureola and Pipra erythrocephala four parts that insert on the dorsal and ventral clades). A fused drum between the origin and ends of A1, but these differentiated caudal insertion of an intrinsic muscle may serve to parts are completely continuous with M. minimize tracheal compression and maxi- tracheolateralis and lack any intermediate mize displacement of the caudal A elements tracheal insertion or belly. In the Pipra serena during contraction ofthat muscle. These gen- species group, M. tracheolateralis inserts era are similar to oscines, in which complex broadly along Al, but in some individuals of intrinsic muscles originate on the cranial several species, the fibers are weak along the margin of a fused tracheal drum. In contrast, lateral midline producing a partial dorsolat- well-differentiated oblique intrinsic muscles eral differentiation. appear in the absence of fused tracheal ele- M. sternotrachealis has been hypothesized ments in Corapipo, Masius, and Ilicura. to act antagonistically to M. tracheolater- However, these intrinsic muscles act oblique- alis by pulling the syrinx caudad and less- ly on the tracheal rings, potentially avoiding ening tensions on the medial tympaniform tracheal compression and eliminating the ne- membrane (Brackenbury, 1989). In most cessity of any tracheal fusion. There are ad- piprids, Mm. sternotracheales are robust and ditional exceptions to these trends. Chloro- strongly developed, especially in males. They pipo uniformis has intrinsic muscles but has insert broadly on the lateral and ventral sur- secondarily lost any tracheal fusion. Some faces of Mm. tracheolaterales. The fibers specimens ofMachaeropterus deliciosus have are typically helically coiled between origin the first two single, tracheal A elements fused and insertion. The one major variation in dorsally into a partial drum, but this species M. tracheolateralis in piprids is present in the lacks any intrinsic syringeal musculature. Pipra aureola clade. In these three species, Piprids also vary in the degree and form Mm. tracheolaterales are short and very ro- of support ofthe medial tympaniform mem- bust, and the fibers are straight and not coiled. brane. In all piprids, B 1-2 are dorsally fused, The muscle attenuates sharply before insert- giving additional support to the craniodorsal ing on the lateral surface of M. tracheolater- margin of the medial tympaniform mem- alis by a thin ligament, and unlike those of brane. In the Pipra aureola clade and Ilicura, 54 AMERICAN MUSEUM NOVITATES NO. 3043 the B elements are quite thin and fragile, and sources. They identified many examples of support for the medial tympaniform mem- simultaneous, harmonically unrelated notes brane is minimal. In contrast, in Manacus, a in sonagrams of vocalizations of many os- caudal series of B elements (typically B5-10) cines and some nonoscine birds. Greenewalt is widened and fused dorsally into an exten- (1968: 60-61) concluded that the ability to sive lattice that gives substantial support to vocalize using two independent sound sources each bronchus and to the dorsal edge of the could be found in members of"almost every medial tympaniform membrane. In Corapi- family for which representative recordings are po, Masius, and Ilicura, the cranial or cranio- available." Greenewalt's (1968: 61) two- dorsal margin of the medial tympaniform sound-source model "start[s] with the prem- membrane is supported by an accessory bar ise that birds can ... activate their two acous- of cartilage, subsequently ossified in C. gut- tical sources (the two medial tympaniform turalis. In Chiroxiphia and Antilophia, the membranes) separately and independently, ventral wall of the trachea and the ventro- that this ability occurs in all groups of birds, medial walls of the bronchi at the tracheo- and that the ability to modulate frequency or bronchial junction are formed by an acces- amplitude or both is available for either sound sory cartilage sheet that also supports the source." The morphological prerequisites for cranial margin of the medial tympaniform two-voice sound production are 1) two dis- membrane. These structural novelties ap- crete potentially vibratile membranes, 2) lat- parently support or reinforce the medial tym- erally independent syringeal musculature, and paniform membrane, but their functional 3) independent neural control over the two roles remain enigmatic. sides of the syrinx. Gaunt (1983) has discussed the problem of Greenewalt's conclusions on the unifor- the exaggeration ofthe stereotypy ofthe "typ- mity and ubiquity of the two-voice ability ical" syrinx in many discussions ofavian vo- in birds underestimated both their morpho- cal mechanisms. The syringeal diversity in logical and vocal diversity. Variations in the piprids provides an opportunity to take syringeal morphology provide interesting op- advantage of syringeal variation, and inves- portunities to test comparatively the mor- tigate the relationship between form and phological assumptions ofGreenewalt's two- function in vocalization. A rigorous compar- sound-source model. ative approach to test hypotheses about the The oscines meet the three morphological evolution of syringeal morphology and func- criteria of the model. They have a pessulus tion would be to document the phylogenetic that divides the two medial tympaniform patterns in morphological and functional di- membranes, and they have independent in- versity, and test hypotheses for the sequence nervation ofthe intrinsic musculature ofboth of morphological and functional transfor- sides of the syrinx. Evidence in support of mations based on their hierarchical distri- the two-voice model in oscines has come from bution (Lauder, 1981; Schaefer and Lauder, several sources. Most recently, Suthers (1990) 1986). Parallel transformational sequences has gathered direct evidence for the two-voice from independent clades would support or model in oscines, by recording air pressure refute generalizations about the evolution of in both bronchi during singing. Nottebohm the system. Advances in the phylogenetic re- (1971, 1972, 1976) investigated the neuro- lationships oftyrannoids will permit detailed logical basis of lateralization and control of investigations ofsyringeal morphology, func- the two syringeal sound sources in three tion, and evolution that are not possible oscines (Serinus serinus, Fringilla coelebs, and through gross correlations. Zonotrichia leucophrys), and concluded that selective denervation ofone side ofthe oscine syrinx can result in degradation ofsome notes THE Two-VOICE MODEL AND THE in the repertoire with no effect on the form SUBOSCINE SYRINX of other notes. Greenewalt (1968) and Stein (1968) docu- Nottebohm and Nottebohm (1976) did a mented that the syrinx of many birds con- similar investigation of neural control of vo- tains two acoustically independent sound calizations in the Orange-winged Parrot 1 992 PRUM: MANAKIN SYRINGES AND PHYLOGENY 55

(Amazona amazonica). Parrots have paired, Elsewhere in suboscines, the pittas (Pitti- vibratile lateral tympaniform membranes, dae) and broadbills (Eurylaimidae) have in- and tracheal nerves that anastomose into a dependent innervation ofthe two sides ofthe single fiber on the ventral surface of the tra- syrinx, but some species lack a pessulus di- chea before dividing again into the left and viding the medial tympaniform membranes right branches that innervate the intrinsic and most lack intrinsic muscles (K6ditz, 1925; muscles. Nottebohm and Nottebohm (1976) Ames, 1971; Prum, in prep.). The furnarioids showed that the left and right tracheal nerves have paired dorsal and ventral tracheal mem- contribute equally to normal vocal control in branes and a single undivided medial tym- Amazona; selective denervation of one side paniform membrane (Ames, 1971; Prum, ofthe syrinx led to general degradation of all personal observ.), but the innervation of the vocalizations, and not to selective loss of a furnarioid syrinx has not been described. The set of independent vocal phrases, as in os- morphological prerequisites ofthe two-voice cines. Nottebohm and Nottebohm (1976) also model are absent in some of these species, failed to detect simultaneous, harmonically leading to the prediction that they should be unrelated notes in a preliminary investiga- unable to produce two independent sounds. tion of parrot vocalizations. Previous analyses of suboscine vocaliza- Variation in the syringeal morphology of tions have not been thorough enough to doc- the piprids and other suboscines provides ad- ument accurately the systematic distribution ditional opportunities to test the two-voice oftwo-voice ability within this diverse clade. model. In piprids, the medial tympaniform Stein (1968) did not include any suboscines membranes are separated laterally by a pes- in his analysis. Greenewalt (1968) apparently sulus, but the tracheal nerves anastomose on analyzed some North American tyrannids and the ventral midline cranial to the innervation other suboscines, but found evidence of two of the caudal portion of M. tracheolateralis, independent sound sources only in the cotin- M. sternotrachealis, or the intrinsic muscles. gid Procnias nudicollis, the Bare-throated Histological or physiological investigations Bellbird. Like other cotingas, Procnias meets are necessary to confirm that the tracheosy- all three morphological criteria for two-voice ringeal nerves of piprids are not laterally in- production: independent medial tympani- dependent despite this anastomosis. How- form membranes, laterally independent ever, additional evidence of the absence of syringeal innervation, and laterally indepen- neural lateralization in piprids comes from dent muscles. But Procnias also has well de- Chiroxiphia, Antilophia, and a few specimens veloped intrinsic muscles which are absent of Manacus. In these taxa, the tracheosyrin- in nearly all other cotingas (Prum, personal geal nerves continue caudad as a single fiber observ.). No information is available on two- for 0.8-6.0 mm after the ventral anastomosis voice vocal ability in other tyrannoids, fur- before dividing again into left and right fibers narioids, or Old World suboscines. that innervate the syringeal musculature. This Additional surveys of suboscine vocaliza- pattern is similar to that ofpsittacids; Greene- tions could provide an important compara- walt's model predicts that piprids should be tive test of Greenewalt's two-voice model. unable to produce two, acoustically indepen- This test is complicated by the possibility dent sounds. that Greenewalt's morphological criteria may In other tyrannoids, a pessulus is also pres- be necessary but not sufficient to explain the ent but the tracheosyringeal nerves do not evolution of the behavior. Some tyrannids anastomose ventrally. In most cotingids, the appear to "fit the bill" but apparently lack nerves remain on the lateral surfaces of the two-voice control, indicating a possible ad- trachea to innervate the muscles of the two ditional role for neural control, as hypothe- sides. In many tyrannids, the tracheosyrin- sized by Ames (1971) and Gaunt (1983). Fur- geal nerves remain separated on the left and thermore, laterally differentiated intrinsic right sides of the trachea, but they exchange muscles may be an additional prerequisite for secondary, horizontal fibers across the ven- two-voice ability, as in Procnias. A phylo- tral surface ofthe trachea before innervating genetic test of Greenewalt's model would the intrinsic muscles. compare the phylogenetic distribution ofthese 56 AMERICAN MUSEUM NOVITATES NO. 3043 morphological traits and two-voice ability, Sexual selection based on female preference and examine the sequence of functional for heritable variation in male traits should change in independent lineages. Despite re- produce more rapid evolution of male traits cent confirmation of two-voice model in os- than natural selection (West-Eberhard, 1983; cines (Suthers, 1990), comparative tests would Lande, 1981). If the intensity of sexual se- evaluate the generality of the model. lection on song varies in related clades in which vocalizations are innate, then this model further predicts that syringeal diversification will be more rapid and more ex- VOCAL LEARNING, LEK BEHAVIOR, tensive in clades that have undergone signif- AND SYRINGEAL EVOLUTION icantly higher levels of sexual selection. It is The striking syringeal diversity of piprids difficult to quantify syringeal variation in a and other suboscines requires some general clade or compare unique morphological nov- explanation. The oscine passerines are more elties of separate clades, but it would be in- than four times as diverse as the suboscines teresting to pursue this possibility with com- in numbers of biological species (Bock and parative tests of syringeal diversity in clades Farrand, 1980), yet the oscine syrinx is com- which differ in breeding system and therefore paratively stereotyped. There are few known in the intensity of sexual selection. variations in oscine supporting elements and An intriguing preliminary association be- musculature even though the oscine syrinx is tween syringeal morphology and breeding be- muscularly more complex (Ames, 1971, 1975, havior is also apparent in this investigation. 1987; Warner, 1972). This difference in Many male piprids produce mechanical wing diversity of syringeal morphology between noises as part of their courtship displays; oscines and suboscines is associated with however, one exploded lek species, Machae- another major difference in their vocal be- ropterus deliciosus, uses mechanical wing havior. Oscines generally learn their songs noises exclusively as territorial advertise- (Kroodsma, 1982), whereas the limited avail- ment calls (Willis, 1966; Prum, personal able evidence indicates that suboscine vo- observ.); male vocalizations are simple and calizations are innate (Kroodsma, 1982, infrequent (Prum, personal observ.). Ma- 1984). Ifthere is any correlation between sy- chaeropterus deliciosus differs strikingly from ringeal structure and function, then selection all other piprids, including its congeners M. on innate vocalizations will produce corre- regulus and pyrocephalus, in that the syrin- lated evolution in syringeal structure. In con- geal musculature ofboth adult males and fe- trast, selection on learned vocalizations will males is largely undeveloped. Further, all not lead to evolution of sound-producing or- other known manakins, including Machae- gans but to social evolution of culturally ropterus regulus (Prum, personal observ.), vo- transmitted behaviors. Syringeal morpholo- calize frequently on their display territories. gy should be less subject to selection when This unique secondary loss of developed sy- interindividual vocal variation is the result ringeal musculature in M. deliciosus is asso- of learning. This general model is congruent ciated with the evolution of obligate me- with the differences in syringeal diversity of chanical wing noise production and the the oscine and suboscine passerines, and could reduction of male vocalization. The corre- be further tested with comparisons of syrin- lation of these two unique factors indicates geal diversity in closely related clades that that replacement of vocal behavior by other differ in vocal learning ability. means of acoustic communication can lead There are several potential sources of se- to a degeneration of syringeal morphology lection on vocalizations. For example, vo- and reduction in vocal ability. calizations could be selected for efficiency of The extraordinary behavior of M. delicio- acoustic transmission (Wiley and Richards, sus has few, if any, parallels in other birds 1982). Vocalizations that are part ofbreeding that might permit us to test this intriguing behavior could also be subject to sexual se- hypothesis. Likewise, the differences between lection (Payne, 1983; West-Eberhard, 1983). Pipra serena and P. suavissima described 1992 PRUM: MANAKIN SYRINGES AND PHYLOGENY 57 above may be a unique example ofespecially polyphyletic traditional generic groups are rapid syringeal evolution. But both findings changed, and four new tribes are named. In underscore the conclusion that syringeal evo- figure 26, I present the syringeal hypothesis lution in piprids has been very dynamic. Fu- ofphylogeny that incorporates the taxonomic ture models of syringeal evolution and func- changes which are described fully below. tion must take into account the diversity and The genera Chloropipo and Xenopipo form plasticity of the syrinx in suboscine birds. a clade, but Chloropipo is paraphyletic. Chlo- Detailed, comparative investigations of sy- ropipo uniformis shares several unique syrin- ringeal morphology and vocalizations in pip- geal synapomorphies with Xenopipo atroni- rids and other suboscines should lead to tens. Additional syringeal and plumage advances in our understanding of the rela- characters indicate that Chloropipo unicolor tionship between syringeal evolution, vocal and flavicapilla are the sister group to this complexity, and avian natural history. clade (Prum, in prep.). The paraphyletic genus Chloropipo should be combined with Xenopipo in a single monophyletic genus. The PHYLOGENETIC CLASSIFICATION name Xenopipo Cabanis 1847 has priority over Chloropipo Cabanis and Heine 1859, so OF THE PIPRIDS all five species in this clade are placed in the The currently accepted classification ofthe genus Xenopipo in the proposed classifica- Pipridae (Snow, 1975, 1979) is based almost tion. entirely on Hellmayr's (1910, 1929) diag- The genus Pipra is polyphyletic. Hetero- noses of piprid genera using male plumage cercus is more closely related to the Pipra characters. Snow (1975, 1979) subsequently aureola and erythrocephala clades than is lumped some monotypic genera and rear- Pipra pipra. Although the syringeal charac- ranged the order of the genera to reflect the ters do not completely resolve the relation- degree ofdevelopment ofpolygynous display ships of the serena group, they strongly in- behavior within the family. Although the sy- dicate that the serena group is unrelated to ringeal hypothesis of phylogeny may be al- other species currently placed in Pipra, and tered by future investigations of other mor- more closely related to Manacus, Chiroxi- phological or molecular character systems, a phia, and Antilophia. The monophyly of the phylogenetic classification of the piprids is serena species group is accepted here based proposed here as an advance over the current, on plumage characters discussed above. The eclectic classification that is based solely on three clades currently placed in the polyphy- overall plumage and behavioral similarity letic genus Pipra-the aureola-erythroceph- (Snow, 1975, 1979). ala clade, the species pipra, and the serena The proposed phylogenetic classification group- should each be recognized as separate reflects the syringeal hypothesis ofphylogeny genera. (figs. 24, 25) by recognizing monophyletic The type of the genus Pipra Linnaeus is taxa. To simplify the number of named Parus aureola Linnaeus (by subsequent des- groups, I use a phylogenetic ranking conven- ignation; Gray, 1840: 33). The Pipra aureola- tion by which subcategories ofequal rank are erythrocephala clade should retain this genus each the sister group to the remaining taxa name. Within this restricted genus Pipra, I ofthe same level within that higher category recognize two subgenera: the subgenus Pipra (Raikow, 1985). Areas of ambiguity in the with the type species aureola Linnaeus to in- consensus tree are indicated by the placement clude aureola,filicauda, andfasciicauda; and of a taxon as sedis mutabilis (s.m.) within a the subgenus Ceratopipra with the type spe- monophyletic higher taxon; a taxon labeled cies cornuta Spix to include cornuta, men- sedis mutabilis has unresolved phylogenetic talis, chloromeros, rubrocapilla, and eryth- relationships to the other taxa of that level rocephala. Ceratopipra is an available junior in the same higher category (Raikow, 1985). synonym ofPipra, and its type species is Pip- In order to conform to these phylogenetic ra cornuta Spix. classification criteria, several paraphyletic or The generic names Dixiphia Reichenbach 58 AMERICAN MUSEUM NOVITATES NO. 3043

1850, Lepidothrix Bonaparte 1854, and Da- clades is a novel hypothesis that is well cor- syncetopa Bonaparte 1854 are available for roborated here by syringeal synapomorphies pipra and the serena species group. Dixiphia and elsewhere by derived behavioral traits has priority over the other two names, which (Prum, 1990a). both date from the same publication (Bona- Traditional systematists may see this ap- parte, 1854). The type species of Dixiphia is plication of genus group names as overly re- leucocilla Linnaeus (= pipra Linnaeus), so strictive or narrow. For example, Sibley Dixiphia should be applied to the species pip- (1957) maintained that the piprids were ra. The type species of Lepidothrix is cyan- "oversplit" at the generic level because ofthe ocapilla Hahn (= coronata; type by subse- reliance on male secondary sexual plumage quent designation, Gray, 1855). The type characters for generic diagnoses. However, species of Dasyncetopa is serena (by mono- 33 of the 40 species in the family were orig- typy). Both coronata and serena are in the inally described as members ofthe genus Pip- serena clade, so either name is available. Pri- ra and were eventually moved into seven oth- ority among available names which date from er genera (Hellmayr, 1910, 1929; Snow, the same publication should be set by the first 1979). The history ofthe classification ofthe reviser (International Code ofZoological No- piprids has been the gradual restriction ofthe menclature, 1985). Lepidothrix is here rec- genus Pipra and the recognition of smaller ommended to have priority over Dasynce- distinct genera. Most of these genera have topa because its etymology is more been corroborated as monophyletic in this appropriate, and it appears first in the orig- investigation. This proposed classification inal publication (Bonaparte, 1854). Lepi- extends the dismantling of the polyphyletic dothrix is apparently derived from the Latin genus Pipra to include the last taxa in the lepidus (pretty) and the Greek thrix (hair). I genus whose generic relationships have never could not determine an obvious etymology been critically evaluated. Recognition of a for Dasyncetopa, but it may be derived from restricted, monophyletic Pipra, a monotypic the Latin dasys (hairy) and the Greek ceo Dixiphia, and novel, monophyletic Lepidoth- (split) and topos (place). In the proposed phy- rix will reflect more accurately the phyloge- logenetic classification, the serena clade is netic history of these species and expedite placed in Lepidothrix. explicit evolutionary analyses based on tax- Alternatively, it would also be possible to onomy. recognize a reconstituted genus Pipra to in- Four new tribes ofpiprid genera are created clude the aureola-erythrocephala clade, Het- to reflect the higher-level structure in the sy- erocercus, and the speciespipra, but this would ringeal hypothesis of phylogeny. Each tribe eliminate the informative monophyletic ge- name is based on the oldest generic name in nus Heterocercus from the classification. that tribe. The tribe Ilicurini is created to Likewise, it would be possible to recognize a include Ilicura, Masius, and Corapipo. It is more restricted genus Pipra to include only placed first in the sequence to indicate that the three species of the aureola clade, and to it is the sister group to all other piprids. The place the five species of the erythrocephala tribe Machaeropterini includes Machaerop- clade in separate genus, for which the name terus only and is labeled sedis mutabilis to Ceratopipra Bonaparte is available. I have indicate that it has unresolved relationships chosen the proposed limits to the genus Pipra to the two remaining, monophyletic piprid to emphasize the monophyly of these eight tribes. The tribe Manacini includes Lepi- species. The monophyly of the aureola and dothrix sedis mutabilis, Manacus, Antilo- erythrocephala clades themselves is more phia, and Chiroxiphia. The last tribe Piprini widely appreciated because of their recogni- contains Xenopipo, Dixiphia, Heterocercus, tion as species groups in recent classifications and Pipra. Within tribes, the order ofgenera and biogeographic analyses (Haffer, 1974; follows the ranking convention, so that Xeno- Snow, 1975, 1979). However, the monophyly pipo is the sister group to the rest of the Pi- of the combined aureola and erythrocephala prini, etc. 1 992 PRUM: MANAKIN SYRINGES AND PHYLOGENY 59

llicura militaris Ilicurini Masius chrysopterus Corapipo leucorrhoa Corapipo gutturalis Machaeropterini, s.m. Machaeropterus deliciosus Machaeropterus regulus Machaeropterus pyrocephalus Lepidothrix species Lepidothrix suavissima Manacini Lepidothrix serena Manacus Antilophia galeata Chiroxiphia Xenopipo holochlora Xenopipo unicolor Xenopipo uniformis Xenopipo atronitens Piprini Dixiphia pipra Heterocercus Pipra aureola subgn. Pipra Pipra filicauda Pipra fasciicauda Pipra cornuta Pipra mentalis subgn. Ceratopipra Pipra s Pipra mnlochioromeros Pipra rubrocapilla Pipra erythrocephala Fig. 26. A phylogenetic hypothesis for the piprids incorporating taxonomic changes of the proposed phylogenetic classification of the family. The four new tribes are: Ilicurini; Machaeropterini, s.m.; Ma- nacini; and Piprini. The genera Chloropipo and Xenopipo are combined into the single senior generic group name Xenopipo. The polyphyletic genus Pipra is split. The aureola and erythrocephala clades remain in Pipra, in the separate subgenera Pipra and Ceratopipra. The species pipra is placed in the monotypic genus Dixiphia. The serena species group is placed in the available generic name Lepidothrix. See text for details and classification. 60 AMERICAN MUSEUM NOVITATES NO. 3043

Family PIPRIDAE Genus Heterocercus Sclater Heterocercus aurantiivertex Sclater and Tribe Ilicurini, new; type-Ilicura Reichen- Salvin, s.m. bach Heterocercusflavivertex Pelzeln, s.m. Genus Ilicura Reichenbach Heterocercus linteatus (Strickland), Ilicura militaris (Shaw and Nodder) s.m. Genus Masius Bonaparte Genus Pipra Linnaeus Masius chrysopterus (Lafresnaye) Subgenus Pipra, new; type-Pipra au- Genus Corapipo Bonaparte reola (Linnaeus) Corapipo leucorrhoa (Sclater) Pipra aureola (Linnaeus), s.m. Corapipo gutturalis (Linnaeus) Piprafasciicauda Hellmayr, s.m. Tribe Machaeropterini, new, s.m.; type- Piprafilicauda Spix, s.m. Machaeropterus Bonaparte Subgenus Ceratopipra, new; type-Pipra Genus Machaeropterus Bonaparte cornuta Spix Machaeropterus deliciosus (Sclater) Pipra cornuta Spix Machaeropterus regulus (Hahn) Pipra mentalis Sclater Machaeropterus pyrocephalus (Scla- Pipra chloromeros Tschudi ter) Pipra rubrocapilla Temminck Tribe Manacini, new; type-Manacus Bris- Pipra erythrocephala (Linnaeus) son Genus Lepidothrix Bonaparte, s.m. Lepidothrix coronata (Spix), s.m. REFERENCES Lepidothrix isidorei (Sclater), s.m. Lepidothrix coeruleocapilla (Tschudi), American Ornithologists' Union s.m. 1983. Check-list ofNorth American birds,6th Lepidothrix nattereri (Sclater), s.m. ed. Lawrence, KS: Allen Press. Lepidothrix vilasboasi (Sick), s.m. Ames, P. L. 1971. The morphology of the syrinx in pas- Lepidothrix iris (Schinz), s.m. serine birds. Peabody Mus. Nat. Hist. Lepidothrix suavissima (Salvin and Bull. 37: 194 pp. Godman) 1975. The application of syringeal morphol- Lepidothrix serena (Linnaeus) ogy to the classification ofthe Old World Genus Manacus Brisson insect eaters (Muscicapidae). Bonn. Manacus manacus (Linnaeus), s.m. Zool. Beitr. 26: 107-134. Manacus vitellinus (Gould), s.m. 1987. The unusual syrinx morphology of the Manacus candei (Parzudaki), s.m. Australian treecreepers Climacteris. Genus Antilophia Reichenbach Emu 87: 192-195. Antilophia galeata (Lichtenstein) Baptista, L. F., and P. W. Trail 1988. On the origin of Darwin's finches. Auk Genus Chiroxiphia Cabanis 105: 663-671. Chiroxiphia caudata (Shaw and Nod- Baumel, J. J., A. S. King, A. M. Lucas, J. E. Brea- der), s.m. zile, and H. E. Evans (eds.) Chiroxiphia linearis (Bonaparte), s.m. 1979. Nomina anatomica avium. New York: Chiroxiphia lanceolata (Wagler), s.m. Academic Press. Chiroxiphia pareola (Linnaeus), s.m. Bock, W. J., and J. Farrand, Jr. Tribe Piprini, new; type-Pipra Linnaeus 1980. The number of species and genera of Genus Xenopipo Cabanis recent birds: a contribution to compar- Xenopipo holochlora (Sclater) ative systematics. Am. Mus. Novitates Xenopipo unicolor (Taczanowski), s.m. 2703: 29 pp. Bock, W. J., and C. R. Shear Xenopipoflavicapilla (Sclater), s.m. 1972. A staining method for gross dissection Xenopipo uniformis (Salvin and God- of vertebrate muscles. Anat. Anz. 130: man) 222-227. Xenopipo atronitens Cabanis Bonaparte, C. L. Genus Dixiphia Reichenbach 1854. Conspectus volucrum anisodactylo- Dixiphia pipra (Linnaeus) rum. Ateneo Italiano 2 (11): 16 pp. 1992 PRUM: MANAKIN SYRINGES AND PHYLOGENY 61

Brackenbury, J. H. Greenewalt, C. H. 1989. Functions of the syrinx and the control 1968. Bird song: acoustics and physiology. of sound production. In A. S. King and Washington: Smithsonian Inst. Press. J. McLelland (eds.), Form and function Haecker, V. in birds 4: 193-220. New York: Aca- 1900. Der gesang der Vogel. Jena: Gustav demic Press. Fisher. Breazile, J. E., and M. Yasuda Haffer, J. 1979. Systema Nerviosum peripheriale. In J. 1967. Speciation in Colombian forest birds J. Baumel, A. S. King, A. M. Lucas, J. west ofthe Andes. Am. Mus. Novitates E. Breazile, and H. E. Evans (eds.). 2294: 57 pp. Nomina anatomica avium, pp. 473-503. 1970. Art-Entstehung bei einigen Waldvogeln New York: Academic Press. Amazoniens. J. Ornithol. 111: 285-331. Bubiefn-Waluszewska, A. 1974. Avian speciation in tropical South 1981. The cranial nerves. In A. S. King and America. Publ. Nuttall Club 14: 390 pp. J. McLelland (eds.), Form and function Hellmayr, C. E. in birds 2: 385-438. New York: Aca- 1906. A revision of the species of the genus demic Press. Pipra. Ibis, ser. 8, 6: 1-46. Cannell, P. A. 1910. Fam. Pipridae. In P. Wytsman (ed.), 1988. Techniques for study of avian syringes. Genera avium 9: 31 pp. Brussels: V. Wilson Bull. 100: 289-293. Verteneuil and A. Desmet. Chamberlain, D. R., W. B. Gross, G. W. Cornwell, 1927. Catalogue ofbirds ofthe Americas, part and H. S. Mosby 5. Tyrannidae. Publ. Field Mus. Nat. 1968. Syringeal anatomy of the Common Hist. 266: 517 pp. Crow. Auk 85: 244-252. 1929. Catalogue ofbirds ofthe Americas, part Cracraft, J. 6. Oxyruncidae-Pipridae-Cotingidae- 1988. Deep-history biogeography: retrieving Rupicolidae-Phytotomidae. Publ. Field the historical pattern of evolving con- Mus. Nat. Hist. 266: 258 pp. tinental biotas. Syst. Zool. 37: 221-236. International Code of Zoological Nomenclature Dingerkus, G., and L. D. Uhler 1985. International code ofzoological nomen- 1977. Enzyme clearing of alcian blue stained clature, 3rd. ed. Berkeley, CA: Univ. whole small vertebrates for demonstra- California Press. tion ofcartilage. Stain Technol. 52: 229- King, A. S. 232. 1979. Systema respiratorium. In J. J. Baumel, Fink, W. L. A. S. King, A. M. Lucas, J. E. Breazile, 1982. The conceptual relationship between and H. E. Evans (eds.), Nomina ana- ontogeny and phylogeny. Paleobiology tomica avium, pp. 227-265. New York: 8: 254-264. Academic Press. Gaunt, A. S. 1989. Functional anatomy ofthe syrinx. In A. 1983. An hypothesis concerning the relation- S. King and J. McLelland (eds.), Form ship of syringeal structure to vocal abil- and function in birds 4: 105-192. New ities. Auk 100: 853-862. York: Academic Press. Gaunt, A. S., and S. L. Gaunt Koditz, W. 1985. Syringeal structure and avian phona- 1925. Uber die Syrinx einiger Clamatores und tion. In R. F. Johnston (ed.), Current auslandischer Oscines. Z. Wiss. Zool. ornithol. 2: 213-245. New York: Ple- 126: 70-144. num Press. Kroodsma, D. E. Gould, S. J., and N. Eldredge 1982. Learning and the ontogeny ofsound sig- 1977. Punctuated equilibria: tempo and mode nals in birds. In D. E. Kroodsma and ofevolution reconsidered. Paleobiology E. H. Miller (eds.), Acoustic commu- 3: 115-151. nication in birds 2: 1-23. New York: Gray, G. R. Academic Press. 1840. A list ofthe genera ofbirds. London: R. 1984. Songs of the Alder Flycatcher (Empi- and J. Taylor, 80 pp. donax alnorum) and Willow Flycatcher 1855. Catalogue of the genera and subgenera (Empidonax traillii) are innate. Auk 10 1: of birds contained in the British Mu- 13-24. seum. London: Trustees British Muse- Lande, R. um, 192 pp. 1981. Models ofspeciation by sexual selection 62 AMERICAN MUSEUM NOVITATES NO. 3043

on polygenic traits. Proc. Nat. Acad. Sci. schen Verschiedenheiten der Stimmor- 78: 3721-3725. gane der Passerinen. Berlin: Konnig. Lanyon, S. M. Akad. Wiss. 1985. Molecular perspective on higher-level 1878. On certain variations in the vocal or- relationships in the Tyrannoidea (Aves). gans of the Passeres that have hitherto Syst. Zool. 34: 404-418. escaped notice, F. J. Bell (transl.). Lon- Lanyon, S. M., and W. E. Lanyon don: MacMillan. 1989. The systematic position of the Plant- Nottebohm, F. cutters, Phytotoma. Auk 106: 422-432. 1971. Neural lateralization ofvocal control in a passerine bird. I. Song. J. Exp. Zool. Lanyon, W. E. 177: 229-262. 1984a. A phylogeny of the kingbirds and their 1972. Neural lateralization ofvocal control in allies. Am. Mus. Novitates 2797: 28 pp. a passerine bird. II. Subsong, calls, and 1984b. The systematic position of the Cocos a theory ofvocal learning. J. Exp. Zool. Flycatcher. Condor 86: 42-47. 179: 32-49. 1985. A phylogeny ofthe myiarchine flycatch- 1976. Phonation in the Orange-winged Am- ers. In P. A. Buckley, M. S. Foster, E. S. Morton, R. S. Ridgley, and F. G. azon Parrot, Amazona amazonica. J. Buckley (eds.), Neotropical ornithology. Comp. Physiol. 108: 157-170. Ornithol. Monogr. 36: 361-380. Nottebohm, F., and M. E. Nottebohm 1986. A phylogeny of the thirty-three genera 1976. Left hypoglossal dominance in the con- ofthe Empidonax assemblage of tyrant trol of Canary and White-crowned flycatchers. Am. Mus. Novitates 2846: Sparrow song. J. Comp. Physiol. 108: 64 pp. 171-192. 1988a. A phylogeny ofthe thirty-two genera in Parkes, K. C. the Elaenia assemblage ofthe tyrant fly- 1961. Intergeneric hybrids in the family Pipri- catchers. Am. Mus. Novitates 2914: 57 dae. Condor 63: 345-350. PP. Payne, R. B. 1988b. The phylogenetic affinities of the fly- 1983. Bird songs, sexual selection, and female catcher genera Myiobius Darwin and mating strategies. In S. K. Wasser (ed.), Terenotriccus Ridgway. Am. Mus. Social behavior of female vertebrates, Novitates 2915: 11 pp. pp. 55-90. New York: Academic Press. 1988c. A phylogeny of the flatbill and tody- Prum, R. 0. tyrant assemblage of tyrant flycatchers. 1988. Historical relationships among avian Am. Mus. Novitates 2923: 41 pp. forest areas of endemism in the Neo- Lauder, G. V. tropics. In H. Ouellet (ed.), Acta Int. 1981. Form and function: structural analysis Ornithol. Congr. 19: 2562-2572. Otta- in evolutionary morphology. Paleo- wa: Nat. Mus. Nat. Sci. biology 7: 430-442. 1989. Phylogenetic analysis of the morphological and behavioral evolution of the Lowe, P. R. neotropical manakins (Aves: Pipridae). 1942. The anatomy of Gould's Manakin Ph.D. diss., Univ. Michigan, Ann Ar- (Manacus vitellinus) in relation to its bor. display, with special reference to an un- 1990a. Phylogenetic analysis of the evolution described pterylar tract and the attach- of display behavior in the neotropical ments of the Flexor carpi ulnaris and manakins (Aves: Pipridae). Ethology 84: Flexor sublimis digitorum muscles to 202-231. the secondary wing feathers. Ibis, 14th 1990b. A test of the monophyly of the mana- ser. 14, 6: 50-83. kins (Pipridae) and of the cotingas (Co- Maddison, W. P., M. J. Donoghue, and D. R. tingidae) based on morphology. Occas. Maddison Pap. Mus. Zool. Univ. Michigan. 723: 1984. Outgroup analysis and parsimony. Syst. 44 pp. Zool. 33: 83-103. Prum, R. O., and A. E. Johnson McKitrick, M. C. 1987. Display behavior, foraging ecology, and 1985. Monophyly of the Tyrannidae (Aves): systematics ofthe Golden-winged Man- comparison of morphology and DNA. akin (Masius chrysopterus). Wilson Bull. Syst. Zool. 34: 34-45. 87: 521-539. Muller, J. P. Prum, R. O., and W. E. Lanyon 1847. Ueber die bisher unbekannten Typi- 1989. Monophyly and phylogeny ofthe Schif- 1992 PRUM: MANAKIN SYRINGES AND PHYLOGENY 63

fornis group (Tyrannoidea). Condor 91: Stein, R. C. 444-461. 1968. Modulation in bird sounds. Auk 85: Raikow, R. J. 229-243. 1985. Problems in avian classification. In R. Suthers, R. A. F. Johnston (ed.), Current ornithol. 2: 1990. Contributions to birdsong from the left 187-212. New York: Plenum Press. and right sides of the intact syrinx. Na- Ridgway, R. ture 347: 473-477. 1907. The birds of North and Middle Amer- Swofford, D. L. ica, part 4. Bull. U.S. Natl. Mus. 50: 973 1989. Phylogenetic Analysis Using Parsimony PP. (PAUP) Computer Program. Version 3. Schaefer, S. A., and G. V. Lauder Champaign, IL: Illinois Nat. Hist. Sur- 1986. Historical transformation of functional vey. design: evolutionary morphology of Traylor, M. A., Jr. feeding mechanisms in loricarioid cat- 1977. A classification ofthe tyrant flycatchers. fishes. Syst. Zool. 35: 489-508. Bull. Mus. Comp. Zool. 148(4): 129- Sclater, P. L. 148. 1888. Catalogue ofthe birds in the British Mu- 1979. Tyrannidae. In M. A. Traylor, Jr. (ed.), seum, vol. 14. London: Trustees British Check-list of birds of the world 8: 1- Museum, 494 pp. 228. Cambridge, MA: Museum Com- Setterwall, C. G. parative Zoology, Harvard Univ. 1901. Studier ofver Syrinx hos Polymoda Pas- Warner, R. W. seres. Lund, Sweden: Akad. Afhandl. 1972. The anatomy of the syrinx in passerine Sibley, C. G. birds. J. Zool. 168: 381-393. 1957. The evolutionary and taxonomic sig- Warter, S. L. nificance ofsexual dimorphism and hy- 1965. The cranial osteology ofthe New World bridization in birds. Condor 59: 166- Tyrannoidea and its taxonomic impli- 191. cations. Ph.D. diss., Louisiana State Sibley, C. G., and J. E. Ahlquist Univ., Ann Arbor, Michigan: Univer- 1985. Phylogeny and classification ofthe New sity Microfilms (order no. 66-761), 161 World Suboscine passerines (Passeri- PP. formes: Oligomyodi: Tyrannides). In P. West-Eberhard, M. J. A. Buckley, M. S. Foster, E. S. Morton, 1983. Sexual selection, social competition, and R. S. Ridgley, and F. G. Buckley (eds.), speciation. Q. Rev. Biol. 52: 155-183. Neotropical Ornithology, Ornithol. Wiley, E. 0. Monogr. 36: 396-430. 1981. Phylogenetics. New York: Wiley. 1990. Phylogeny and classification of birds. Wiley, H. R., and D. G. Richards New Haven, CT: Yale Univ. Press. 1982. Adaptations for acoustic communica- Sibley, C. G., and B. L. Monroe tion in birds: sound transmission and 1990. Distribution and taxonomy of birds of signal detection. In D. E. Kroodsma and the world. New Haven, CT: Yale Univ. E. H. Miller (eds.), Acoustic commu- Press. nication in birds 1: 131-181. New York: Sibley, C. G., S. M. Lanyon, and J. E. Ahlquist Academic Press. 1985. The relationships ofthe sharpbill (Oxy- runcus cristatus). Condor 86: 48-52. Willis, E. 0. Sick, H. 1966. Notes on a display and nest ofthe Club- 1959. Die Balz der Schumckvogel (Pipridae). winged Manakin. Auk 83: 475-476. J. Ornithol. 100: 269-302. Wolters, H. E. 1967. Courtship behavior in manakins (Pipri- 1977. Die Vogelarten der Erde. III. Hamburg: dae): a review. Living Bird 6: 5-22. Paul Parey. Snow, D. W. Wood, D. S., R. L. Zusi, and M. A. Jenkinson 1963. The evolution of manakin displays. 1982. World inventory of avian spirit speci- Proc. Int. Ornithol. Congr. 13: 553-561. mens. Norman, OK: American Orni- 1975. The classification ofthe manakins. Bull. thologists' Union. B. 0. C. 95: 20-27. Wunderlich, L. 1979. Tityrinae, Pipridae, Cotingidae. In M. 1886. Beitrage sur vergleichen Anatomie und A. Traylor, Jr. (ed.), Check-list of birds Entwicklungsgeschichte desunteren of the world 8: 229-308. Cambridge, Kehlkopfs der V6gel. Kaiserlich Leo- MA: Museum Comparative Zoology, poldinischen-Carolinischen Deutschen Harvard Univ. Akad. Naturf. Nova Acta 18. 64 AMERICAN MUSEUM NOVITATES NO. 3043

APPENDIX FMNH 289998. IS: FMNH 290401,,322567; LSUMZ 79589, 131842, 131844, 131845; List of piprid syringeal specimens ob- MPEG 4080, 5860, 5861, 5863. served. Syringeal specimens were either Machaeropterus deliciosus- C&S: AMNH cleared and double stained (C&S) for carti- 8713. IS: BM A/1986-3-1; UMMZ 225054, lage and bone (Dingerkus and Uhler, 1977) 225055, 225056, 225057. or stained with reversible iodine stain (IS) for Chloropipo holochlora -C&S: LSUMZ observation ofmuscle fibers (Bock and Shear, 112837; KU 60693, 65552. IS: UMMZ 1972). Specimens were borrowed from var- 225045, 225046, 225047, 225048, 225049, ious institutions, and abbreviations are given 225050, 225051, 226601. in the introduction. Uncataloged specimens Chloropipoflavicapilla -no specimens ob- are identified by the original collector's num- served. ber in parentheses. Chloropipo uniformis-C&S: AMNH 7680. IS: AMNH 10378, 10379, 10407; USNM Corapipo leucorrhoa -C&S: LSUMZ 504508, 504509, 504510. 102429, 108441. IS: LSUMZ 104779, Chloropipo unicolor-C&S: LSUMZ 108683; MVZ 4801, 4802, 4804; USNM 71544, 89472. IS: LSUMZ 70637, 71541, 510718, 510719; UMMZ 226607, 226608, 71543, 89471, 89474. 226609, 226610. Xenopipo atronitens-C&S: AMNH 8083, Corapipo gutturalis-C&S: AMNH 2256, 8152. IS: AMNH 18118; MPEG 4099, 4619, BM 1968.46.17. 5849, 5850. Masius chrysopterus-C&S: CM 1161, Heterocercus flavivertex-C&S: AMNH LSUMZ 117121. IS: BM A/1968-3-12; 9840, 15203, 15204; USNM 504517. IS: LSUMZ 83866, 89470; UMMZ 225059, AMNH 9374; COP ML-1983, ML-1987. 226604, 226605, 226606. Heterocercus linteatus -C&S: FMNH Ilicura militaris-C&S: FMNH 107023, 330655. IS: MPEG 4129, 5836. 107029. IS: FMNH 107028,107030; MZUSP Heterocercus aurantiivertex -no speci- uncat. mens observed. Manacus candei-C&S: AMNH 6654. Pipra coronata-C&S: AMNH 2257, 9838, Manacus vitellinus-C&S: AMNH 8084; 15194, 15199; LSUMZ 102412, 102419. IS: LSUMZ 108418, 108423. UMMZ 225063, 225064, 225065, 225066, Manacus manacus-C&S: AMNH 7707, 226602. 7710, 8147. IS: LSUMZ 35355, 112833, Pipra isidorei-C&S: LSUMZ 118033. 112834, 114482, 114483; ROM 107498, Pipra coeruleocapilla -C&S: FMNH 113119, 127649; USNM 515118, 515119. 291664. IS: LSUMZ 70636, 71540; USNM Chiroxiphia linearis-C&S: AMNH 3671, 512022, 512071, 512291, 512295. 3672. IS: USNM 541043, 541044, 541045, Pipra nattereri -IS: FMNH 333842; 541046, 541048. MPEG 1979, 1980, 1981, 1983, 3739, 3740, Chiroxiphia lanceolata-C&S: CM 373. CG-1 105. Chiroxiphia pareola -C&S: AMNH 8080, Pipra vilasboasi-no specimens observed. LSUMZ 95546. IS: LSUMZ 95548, 123335, Pipra iris-C&S: AMNH 9892. IS: AMNH 131837; MPEG 5701; MZUSP uncat.; 17691; MPEG 4590, 4784, 5838, 5840. UMMZ 225060, 225061, 225062. Pipra suavissima-C&S: AMNH 9366, Chiroxiphia caudata-C&S: AMNH 2447, 9368, 816768. IS: BM A/1968-46-19; AMNH 2525, 2526. IS: FMNH 107322, 107323, 10326, 10327. 107334. Pipra serena-C&S: ROM 127643, Antilophia galeata-C&S: BM 1968.66.- 127657. 205. IS: MPEG 7296, 7297; ZUB 88008, Pipra pipra-C&S: AMNH 2259, 8145, 88015, MAM-1988. 9358. IS: CM 1395, 1442, 1446; LSUMZ Machaeropterus regulus-C&S: LSUMZ 118022, 118027, 118028, 118030; ROM 85980, 85981, 115836, KU 66684. IS: 107489, 107637, 112440; UMMZ 226613; LSUMZ 111081, 114486, 120584; UMMZ USNM 515095, 515100, 515101. 225052. Pipra aureola-C&S: CM 1280. IS: AMNH Machaeropterus pyrocephalus - C&S: 17689; USNM 515055. 1992 PRUM: MANAKIN SYRINGES AND PHYLOGENY 65

Piprafasciicauda -C&S: AMNH 2301. IS: Pipra rubrocapilla-C&S: AMNH 2517, LSUMZ 35361, 72966, 91517, 123381. 2520; BM 1933.2.4.1. IS: AMNH 17690; Pipra filicauda -C&S: LSUMZ 83807, LSUMZ 114472, 131823, 131828, 131830, 115617; KU 66660. IS: UMMZ 225067, 131831. 225068, 225069, 225070, 226603. Pipra chloromeros-C&S: LSUMZ 102394, Pipra mentalis-C&S: LSUMZ 95070; 102395. IS: FMNH 322838, 322840, 322844; UMMZ 226611. IS: UMMZ 226612; USNM LSUMZ 75568, 79587, 103303, 103304, 510741, 510743, 510746, 510747. 107647. Pipra erythrocephala -C&S: AMNH 8081, Pipra cornuta-IS: AMNH 10484, 10485, 9839, 15178; LSUMZ 108388, 108390, 10486, 10487, 10488, 10489, 10490, 10491; 110345, (SWC-1412). IS: BM A/1970-34-1; USNM 504506, 504507. UMMZ 225071, 225073, 225074, 225075.

Recent issues of the Novitates may be purchased from the Museum. Lists of back issues of the Novitates, Bulletin, and Anthropological Papers published during the last five years are available free of charge. Address orders to: American Museum of Natural History Library, Department D, Central Park West at 79th St., New York, N.Y. 10024.

THIS PUBLICATION IS PRINTED ON ACID-FREE PAPER.