Division of Gastroenterology, Department of Internal Medicine, The University of Texas Medical Branch, Galveston, Texas 77555; and Section of Birds, Carnegie Museum of Natural History, Pittsburgh, Pennsylvania 15213, USA The Dendrocolaptidae and Furnariidae, two groups of Central and South American suboscine passetines, have always been considered closely related. With 52 species, the woodcreepers occur from northern Mex- ico to central Argentina in forest, forest~edge, and open-woodland habitats, where they nest in holes or cavities. Their body lengths range from 13.5 to 37 cm, and their mass from 12 to 120 g; the plumage is gen- erally dull rufous brown, often streaked or spotted. The most distinctive external features of dendroco- laptids are: usually large, laterally compressed, often decurved bills used as foraging probes on tree trunks; and their rigid, spiny-tipped rectrices used as props against those trunks. Their ecology, behavior, and superficially homogenous morphology reflect the scansorial habits of woodcreepers. The much larger group of ovenbirds (218 species) occurs from central Mexico to Patagonia and the Falkland Islands. The Furnariidae have radiated into an astonishing diver- sity of niches, having adapted to arboreal and terres- trial habitats from deep forest to open desert, and from coastal sand dunes and cliffs to above treeline in the Andes. Accordingly, they also are widely vari- able in external morphology and in the sites and types of their closed nests. Furnariids generally are smaller than dendrocolaptids, ranging from 10 to 28 cm long and with a mass of 9 to 46 g. Their plumage is more varied than that of woodcreepers, but they are still primarily dark brown, olive, or rufous above and lighter below. A few of the treerunners (Margarornis "assemblage"; Rudge and Raikow 1992) have some- what spiny tails. Morphologically the two groups have similar trach- eophone syringes with two pairs of intrinsic syringeal muscles (Ames 1971); their front toes are variously joined at the base (Ridgway 1911, Vaurie 1980); and they have a single pair of notches in the posterior sternum (with a few exceptions and individual vari- ability; Heimerdinger and Ames 1976). DNA-DNA hybridization studies also indicate a close relation- ship of the two groups (Sibley and Ahlquist 1990). The taxonomic ranking of woodcreepers and oven- birds has never been settled, with some authors (e.g. Sibley and Ahlquist 1990, Raikow 1994) regarding them as subfamilies of a larger family, Furnariidae. Most traditional classifications, however, have listed them as separate families--the first two (although either family may appear first) of the suborder Ty- ranni, superfamily Furnarioidea (e.g. Peters 1951, Morony et al. 1975, AOU 1983). The Dendrocolaptidae have been separated from the Furnariidae on the basis of additional syringeal features, particularly the horns on the processi vocales of woodcreepers that are lack- ing in all ovenbirds except Geositta. More tradition- ally, the separate familial status has rested on the different shapes of the nares--holorhinal (rounded) in dendrocolaptids, pseudoschizorhinal (slitlike) in furnariids--as well as on several characters in the feet and on plumage patterns, especially wing stripes. Feduccia (1973) also found differences in the electro- phoretic patterns of their hemoglobin, but as with all the characters used thus far to divide these two groups, there were a few exceptions or intermediate forms. Perhaps it is primarily because no single character has been found that unequivocally separated the woodcreepers and ovenbirds that their taxonomic status is still debated. Materials and methods.--The technique used has been described previously (Clench 1970, 1985). Skins were removed from 34 specimens freshly collected or pre- served in alcohol, pinned flat, dried, and the under- side examined under a 10-40 x binocular dissecting microscope. The series included 12 specimens of Den- drocolaptidae (11 species, 10 of 13 genera): Dendro- cincla anabatina, D. tyrannina, Deconychura longicauda, Sittasomus griseicapillus, Glyphorhynchus spirurus (2), Drymornis bridgesii, Xiphocolaptes major, Dendrocolaptes certhia, Xiphorhynchus guttatus, Lepidocolaptes affinis, and Campylorhamphus trochilirostris. The series of Furnari- idae included 22 specimens (21 species, 16 of 34 gen- era, 3 subfamilies): Geositta cunicularia, Chilia melanura, and Furnarius rufus in the Furnariinae; Synallaxis al- bescens, S. cinnamomea, S. erythrothorax, Certhiaxis cin- namomea, C. (Cranioleuca) sulfurifera, Phacellodomus ruf- ifrons, Phleocryptes melanops, and Anumbius anumbi in the Synallaxinae; and Margarornis (Premnoplex) brun- nescens, M. rubiginosus, Lochmias nematura, Pseudoseisura lophotes, Philydor rufus, Automolus leucophthalmus, Scler- urus mexicanus, Xenops (Heliobletus) contaminatus, X. minutus (2), and Pygarrhichas albogularis in the Phily- dorinae. The taxonomy employed here is that of Mor- ony et al. (1975), who followed Vaurie (1971) for the Furnariidae. Genera recognized by Peters (1951) but synonymized by Vaurie (1980) are noted above by parentheses. Included in my sample were two specimens of a species in each family in order to check for significant individual variation; none was found. Similarly, two species in the dendrocolaptid genus Dendrocincla and two species of the furnariids Certhiaxis, Margarornis, and Xenops and three of Synallaxis were examined for significant intrageneric variation; none was found. The reliability of using a single specimen to establish the pterylographic pattern and relative number of feathers of a genus was confirmed in these families as it has been in other passerines (Clench 1970, 1985, 1992). Also included in the sample were representatives of the genera that have been found by others to be atypical of their respective families: the woodcreepers Dendrocincla, Deconychura, Sittasomus, and Glyphorhyn- chus, and the ovenbirds Geositta, Automolus, Xenops, Pygarrhichas, Sclerurus, and two of the ovenbird/tree- runner assemblage (Margarornis, Premnoplex; Rudge and Raikow 1992). I studied the pattern of the main body tracts--the pteryla spinalis (spinal tract) that extends down the dorsal surface from the base of the skull to the uro- pygial gland, and the pteryla ventralis (ventral tract), a paired, mirror-image, wishbone-shaped band of feathers on either side of the breast and abdomen. These tracts have proved to exhibit the most impor- tant variation in passetines. Terminology follows that used in Clench (1985). Statistically significant differ- ences in numbers of feathers were determined by the $tudent's t-test. Results.--In the pteryla spinalis, both families had the general pattern of a solidly feathered rhombic pars dorsalis (i.e. there were no spaces, gaps, or ap- teria). The partes interscapularis and pelvica were narrow and undistinguished. All specimens had a gap between the posterior border of the pteryla capitalis and the pars interscapularis at the base of the neck; this gap has been present in all the suboscines (Old or New World) that I have studied, but in no oscines. The spinal tract of all the woodcreeper specimens,  PARS INTERSCAPULARIS o o,O o O,o,O PRS OORSAUS ,d ? ?' ?, ,o, o, o, o?,o o %%%0 ? o. oxoxo o o : PARS PELVICA Dendrocolaptes certhia PTERYLA SPINALIS Fig. 1. Diagrammatic representation of pteryla spinalis of Dendrocolaptes certhia. Individual feathers of pars dorsalis indicated by circles, those of last row of pars interscapularis as squares, and first row of pars pelvica as triangles. exemplified by Dendrocolaptes certhia (Fig. 1), was rel- atively narrow and comparatively sparsely feathered for passerines of this large body size. The mean num- ber of feathers in the pars dorsalis (the tract's rhom- bic-shaped central portion) was 100.7 + SE of 3.8 (range 92-112) in eight rows. There were 92 feathers in the pars dorsalis of Dendrocolaptes certhia, which has a mass of about 80 g. In contrast, all the generally smaller furnariids studied had a significantly more heavily feathered pars dorsalis, with a mean of 143.0 + 5.9 feathers in 8 to 10 rows (P < 0.001). Automolus leucophthalmus (Fig. 2) was typical in pattern. The family sample ranged from 128 to 164 feathers in the pars dorsalis, except for 112 in the tiny species Xenops minutus (body mass of only 9 g). In the pteryla ventralis, the two groups had the same difference in degree of feathering, but there was an even more striking difference in pattern. In the flank region where the two parts of the tract divide, woodcreepers showed a simple splitting of the more anterior chevron-shaped rows at the central feather  PARS INTERSCAPULARIS dd  o  PARS PELVICA Automolus leucophthalmus PTERYLA SPINALIS Fig. 2. Diagrammatic representation of pteryla spinalis of Automolus leucophthalmus. Conventions as in Figure 1. and a continuation of the pars pectoralis as single- armed lateral rows (e.g. Dendrocolaptes certhia; Fig. 3). In my experience, this pattern is unique within the Passeriformes, and all the woodcreepers examined had it--with no exceptions or intermediate condi- tions. In contrast, the pars pectoralis in the Furnariidae (e.g. Automolus leucophthalmus; Fig. 4) was formed by two-armed chevron-shaped rows and ended in an oblique border with the lateral arms shorter than the roedial. Without exception, all ovenbirds examined had this pattern, which is also seen in other New World suboscines such as members of the Formicar- iidae, and in all oscines. Discussion.--Before my study, the pterylosis of the Dendrocolaptidae and Furnariidae was almost un- known. In the only previous description of the pter- ylography of these families, Nitzsch (1867) did not analyze the tracts in detail, but noted that Dendroco- laptes had an elongated rounded rhombic pars dor- salis, and the posterior end of the flank was widely separated from the pars abdominalis. He also studied four genera of ovenbirds--"Synallaxis" (=Leptasthen- ura) setaria, Ochetorhynchus (=Upucerthia) ruficaudus, "Opetiorhynchus rupestris" (=Cinclodes patagonicus), and PARS PECTORALIS Dendrocolaptes certhia PTERYLA VENTRALIS Diaġrammatic representation of left flank Fig. 3. region of pteryla ventralis of Dendrocolaptes certhia. Central row feathers indicated as O. several species of Furnarius, including F. rufus and F. figulus. He noted correctly that they had a solid rhom- bic pars dorsalis and typically passerine pterylosis. Feduccia (1973) considered Dendrocincla, Decony- chura, Sittasomus, and Glyphorhynchus to be an "inter- mediate" group of woodcreepers, with a mosaic of ovenbirdlike characters, particularly in the wing stripe and in skull, hindlimb, and sternal characters. Gly- phorhynchus also showed a hemoglobin pattern iden- tical to that of ovenbirds. Sibley and Ahlquist (1990) studied three of the four genera in question (Dendro- cincla, Sittasomus, and Glyphorhynchus) along with five others and concluded a close relationship within the woodcreepers, and between that group and the ov- enbirds. Raikow (1994) found the woodcreepers as a whole to be monophyletic. The pterylographic pattern of the four intermediate genera was perfectly uniform and typical of the other dendrocolaptids examined. In all respects the feather counts were also identical or very similar. For ex- ample, the two smaller species, Sittasomus griseicapillus and Glyphorhynchus spirurus, had 92 and 96 feathers, respectively, in the pteryla spinalis, pars dorsalis. Dendrocincla anabatina, D. tyrannina, and Deconychura longicauda had 112, 112, and 100, respectively (family œ = 100.7 + 3.8). Ames (1971) found that Geositta, an otherwise typ- ical furnariid, had large horns on the processi vocales of the syrinx, typical of dendrocolaptids. In his dis- cussion of the anomaly, Feduccia (1973:47) noted "it is difficult to understand what similar function would be demanded of this lark-like open country bird, that would be similar to the deep-forest inhabiting wood- hewers." Geositta has remained a member of the ov- PARS ABDOMINALIS i RS PECTORALIS Automolus leucophthalmus PTERYLA VENTRALIS Fig. 4. Diagrammatic representation of left flank region of pteryla ventralis of Automolus leucophthal- mus. Central row feathers indicated as O. enbirds in all recent classifications, and its pterylosis agrees with this designation. The syringeal oddity remains to be explained. Based primarily on comparative osteology, Feduc- cia (1973) proposed that the species in the Margarornis assemblage, despite their scansorial habits and mak- ing some use of the tail as a prop against tree trunks, were not dendrocolaptids (contra Sclater 1890) and should be retained in the Furnariidae. Depending on which taxonomist one follows, this assemblage has six or eight species in four genera; Vaurie (1971, 1980) included them all in Margarornis. For his study, Fed- uccia (1973) had specimens of Margarornis rubiginosus and Premnoplex brunnescens, but not of Premnornis (gut- tuligera) or Roraimia (adusta). With material of all four genera, and emphasizing myological characters of the fore- and hindlimbs, Rudge and Raikow (1992) con- cluded that the treerunners were a monophyletic group of ovenbirds. The pterylosis also supports Mar- garornis as a furnariid. Feduccia (1973) noted that Automolus was another intermediate form, similar to Dendrocincla and related woodcreepers. In addition, he remarked that Xenops was similar to dendrocolaptids in its tibiotarsus and recommended that the scansorial genus Pygarrhichas, which Sclater (1890) had considered a woodcreeper, be retained in the Furnariidae. The pterylosis of all three is in perfect agreement with the other ovenbirds I examined. In summary, the body pterylosis of the Dendro- colaptidae and Furnariidae is entirely consistent within their respective groups. Most importantly, the Dendrocolaptidae have a unique configuration in the ventral tract that distinguishes them from all other Passerines. In contrast, the ventral pattern of the Fur- nariidae is similar to that seen in other New World suboscines and all oscines. Woodcreeper and oven- bird genera that have intermediate or conflicting as- pects in other morphological characters show a clear pterylographic pattern that can be used to place any of the genera studied in one group or the other. Pter- ylosis always has proved in my other studies of pas- setines to be an evolutionarily conservative character (Clench 1970, 1985). Therefore, because there is a striking pattern difference in the ventral tract and relatively strong differences in the degree of feath- ering, and because a body of other morphological data supports the traditional separation of the two groups, I suggest that they should continue to be regarded as separate families. Acknowledgments.--Kenneth C. Parkes, Philip S. Humphrey, and I collected some of the specimens specifically for this study; the specimens are now part of the alcohol-preserved or flatskin material at Car- negie Museum of Natural History or Peabody Mu- seum of Natural History, Yale University. The ma- jority of specimens came from alcohol collections at: Carnegie Museum of Natural History; Peabody Mu- seum of Natural History, Yale University; and the National Museum of Natural History, Smithsonian Institution. I sincerely thank all the curatorial friends who provided material for this very long-term study. LITERATURE CITED AMERICAN ORNITHOLOGISTS' UNION. 1983. Check-list of North American birds, 6th ed. Am. Ornithol. Union, Washington, D.C. AMES, P.L. 1971. The morphology of the syrinx of passefine birds. Peabody Mus. Nat. Hist., Yale Univ. Bull. 37. CLENCH, M.H. 1970. Variability in body pterylosis, with special reference to the genus Passer. Auk 87:650-691. CLENCH, M.H. 1985. Body pterylosis of Atrichornis, Menura, the "corvid assemblage," and other pos- sibly related passetines (Aves: Passeriformes). Records Australian Mus. 37:115-142. CLENCH, M.H. 1992. Pterylography of birds-of-par- adise and the systematic position of Macgregor's Bird-of-Paradise (Macgregoria pulchra). Auk 190: 923-928. FEDUCCIA, A. 1973. Evolutionary trends in the Neo- tropical ovenbirds and woodhewers. Ornithol. Monogr. 13. HEIMERDINGER, M. A., AND P. L. AMES. 1976. Varia- tion in the sternal notches of suboscine passeri- form birds. Peabody Mus. Nat. Hist., Yale Univ. Postilia 105. MORONY, J. J., JR., W. J. BOCK, AND J. FARRAND, JR. 1975. Reference list of the birds of the world. American Museum of Natural History, New York. NITZSCH, C. L. 1867. Pterylography (P. L. Sclater, Ed.). Ray Society, London. PETERS, J. L. 1951. Check-list of birds of the world, vol. 7. Museum of Comparative Zoology, Cam- bridge, Massachusetts. RIKOw. R.J. 1994. A phylogeny of the woodcree- pers (Dendrocolaptidae). Auk 111:104-114. RIDGWAY, R. 1911. The birds of North and Middle America, part 5. U.S. Natl. Mus. Bull. 50. RUDGE, D. W., AND R. J. RAIKOw. 1992. The phylo- genetic relationships of the Margarornis assem- blage (Furnariidae). Condor 94:760-766. SCLATER, P.L. 1890. Catalogue of birds in the British Museum, vol. 15. British Museum (Natural His- tory), London. SIBLEY, C. G., AND J. E. AHLQUIST. 1990. Phylogeny and classification of birds. A study in molecular evolution. Yale Univ. Press, New Haven, Con- necticut. VAURIE, C. 1971. Classification of the ovenbirds (Fur- nariidae). H. F. and G. Witherby, London. VAURIE, C. 1980. Taxonomy and geographical dis- tribution of the Furnariidae (Aves, Passeri- formes). Am. Mus. Nat. Hist. Bull. 166:1-357. Received 22 June 1994, accepted 31 August 1994.